Complexity Economics: Building a New Approach to Ancient Economic History [1st ed.] 9783030478971, 9783030478988

Table of contents :
Front Matter ....Pages i-xix
Introduction: Finding a New Approach to Ancient Proxy Data (Koenraad Verboven)....Pages 1-18
Front Matter ....Pages 19-19
Playing by Whose Rules? Institutional Resilience, Conflict and Change in the Roman Economy (Koenraad Verboven)....Pages 21-51
Networks as Proxies: A Relational Approach Towards Economic Complexity in the Roman Period (Johannes Preiser-Kapeller)....Pages 53-103
Evaluating the Potential of Computational Modelling for Informing Debates on Roman Economic Integration (Tom Brughmans)....Pages 105-123
Visualising Roman Institutional Environments for Exchange as a Complex System (Merav Haklai)....Pages 125-159
Front Matter ....Pages 161-161
Social Complexity and Complexity Economics: Studying Socio-economic Systems at Düzen Tepe and Sagalassos (SW Turkey) (Dries Daems)....Pages 163-201
A Method for Estimating Roman Population Sizes from Urban Survey Contexts: An Application in Central Adriatic Italy (Dimitri Van Limbergen, Frank Vermeulen)....Pages 203-249
Complexity and Urban Hierarchy of Ancient Urbanism: The Cities of Roman Asia Minor (Rinse Willet)....Pages 251-294
Front Matter ....Pages 295-295
Disease Proxies and the Diagnosis of the Late Antonine Economy (Colin P. Elliott)....Pages 297-326
Measuring and Comparing Economic Interaction Based on the Paths and Speed of Infections: The Case Study of the Spread of the Justinianic Plague and Black Death (Lars Börner, Battista Severgnini)....Pages 327-356
Back Matter ....Pages 357-363

Citation preview

PALGRAVE STUDIES IN ANCIENT ECONOMIES

Complexity Economics Building a New Approach to Ancient Economic History Edited by Koenraad Verboven

Palgrave Studies in Ancient Economies Series Editors Paul Erdkamp Vrije Universiteit Brussel Brussels, Belgium Ken Hirth Pennsylvania State University University Park, PA, USA Claire Holleran University of Exeter Exeter, Devon, UK Michael Jursa University of Vienna Vienna, Austria J. G. Manning Yale University New Haven, CT, USA Osmund Bopearachchi Institute of East Asian Studies University of California, Berkeley Berkeley, CA, USA

This series provides a unique dedicated forum for ancient economic historians to publish studies that make use of current theories, models, concepts, and approaches drawn from the social sciences and the discipline of economics, as well as studies that use an explicitly comparative methodology. Such theoretical and comparative approaches to the ancient economy promotes the incorporation of the ancient world into studies of economic history more broadly, ending the tradition of viewing antiquity as something separate or ‘other’. The series not only focuses on the ancient Mediterranean world, but also includes studies of ancient China, India, and the Americas pre-1500. This encourages scholars working in different regions and cultures to explore connections and comparisons between economic systems and processes, opening up dialogue and encouraging new approaches to ancient economies. More information about this series at http://www.palgrave.com/gp/series/15723

Koenraad Verboven Editor

Complexity Economics Building a New Approach to Ancient Economic History

Editor Koenraad Verboven Department of History Ghent University Ghent, Belgium

Palgrave Studies in Ancient Economies ISBN 978-3-030-47897-1    ISBN 978-3-030-47898-8 (eBook) https://doi.org/10.1007/978-3-030-47898-8 © The Editor(s) (if applicable) and The Author(s), under exclusive licence to Springer Nature Switzerland AG 2021 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the ­publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and ­institutional affiliations. This Palgrave Macmillan imprint is published by the registered company Springer Nature Switzerland AG. The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland

Contents

1 Introduction: Finding a New Approach to Ancient Proxy Data  1 Koenraad Verboven Part I Theoretical Frameworks and Methodologies  19 2 Playing by Whose Rules? Institutional Resilience, Conflict and Change in the Roman Economy 21 Koenraad Verboven 3 Networks as Proxies: A Relational Approach Towards Economic Complexity in the Roman Period 53 Johannes Preiser-Kapeller 4 Evaluating the Potential of Computational Modelling for Informing Debates on Roman Economic Integration105 Tom Brughmans 5 Visualising Roman Institutional Environments for Exchange as a Complex System125 Merav Haklai v

vi 

Contents

Part II Urban Systems 161 6 Social Complexity and Complexity Economics: Studying Socio-economic Systems at Düzen Tepe and Sagalassos (SW Turkey)163 Dries Daems 7 A Method for Estimating Roman Population Sizes from Urban Survey Contexts: An Application in Central Adriatic Italy203 Dimitri Van Limbergen and Frank Vermeulen 8 Complexity and Urban Hierarchy of Ancient Urbanism: The Cities of Roman Asia Minor251 Rinse Willet Part III Epidemics 295 9 Disease Proxies and the Diagnosis of the Late Antonine Economy297 Colin P. Elliott 10 Measuring and Comparing Economic Interaction Based on the Paths and Speed of Infections: The Case Study of the Spread of the Justinianic Plague and Black Death327 Lars Börner and Battista Severgnini Index357

Notes on Contributors

Lars  Börner is Professor of Economics at Martin-Luther-University of Halle-Wittenberg, Germany, and a senior research fellow at Data Analytics for Finance and Macro (DAFM), King’s Business School, King’s College London, UK. Tom  Brughmans is an associate professor at the Centre for Urban Network Evolutions (UrbNet) and Classical Archaeology, School of Culture and Society, Aarhus University, Denmark. Dries Daems  is a post-doctoral researcher of the Sagalassos Archaeological Research Project at the University of Leuven, Belgium. Colin P. Elliott  is an assistant professor of History at Indiana University. Merav Haklai  is a lecturer in the Department of General History at the Ben-Gurion University of the Negev, Israel. Johannes Preiser-Kapeller  is a senior research associate at the Institute for Medieval Research, Division of Byzantine Research, of the Austrian Academy of Sciences, and lecturer at the Institute for Byzantine and Modern Greek Studies, University of Vienna, Austria. Battista  Severgnini is an associate professor in the Department of Economics at the Copenhagen Business School, Denmark. Dimitri Van Limbergen  is a senior post-doctoral fellow at the Flanders Research Foundation, Ghent University, Belgium.

vii

viii 

NOTES ON CONTRIBUTORS

Koenraad Verboven  is Professor of Ancient History in the Department of History at Ghent University, Belgium. Frank Vermeulen  is Professor of Classical Archaeology in the Department of Archaeology at Ghent University, Belgium. Rinse Willet  is a post-doctoral researcher and intensive survey coordinator at the Sagalassos Archaeological Research Project, University of Leuven, Belgium.

List of Figures

Fig. 3.1

Fig. 3.2

Fig. 3.3

Fig. 3.4

Fig. 3.5

Fig. 3.6 Fig. 3.7

Nodes in the network model of riverine transport in period I (first to fifth century CE) in the Po plain sized according to their betweenness centrality (data: L. Werther, map: J. PreiserKapeller, 2015) 79 Nodes in the network model of riverine transport in period II (sixth to ninth century CE) in the Po plain sized according to their betweenness centrality (data: L. Werther, map: J. PreiserKapeller, 2015) 80 Nodes in the network model of riverine transport in period I (first to fifth century CE) in the Po plain sized according to their closeness centrality (data: L. Werther, map: J. PreiserKapeller, 2015) 81 Nodes in the network model of riverine transport in period II (sixth to ninth century CE) in the Po plain sized according to their closeness centrality (data: L. Werther, map: J. PreiserKapeller, 2015) 82 The matrices for the network model of riverine transport in period I (first to fifth century CE, left) and period II (sixth to ninth century CE, right) (data. L. Werther, graphs: J. PreiserKapeller, 2015) 82 User interface of the “ORBIS Stanford Geospatial Network Model of the Roman World” (screen shot from: http://orbis. stanford.edu/)83 ORBIS Stanford Geospatial Network Model of the Roman World—visualisation of the nodes (= places) sized according to their degree-centrality (analysis and map J. Preiser-Kapeller, 2015)83 ix

x 

List of Figures

Fig. 3.8

Fig. 3.9

Fig. 3.10

Fig. 3.11

Fig. 3.12

Fig. 3.13

Fig. 3.14

Fig. 3.15

Fig. 3.16

ORBIS Stanford Geospatial Network Model of the Roman World—visualisation of the nodes (= places) sized according to their betweenness-­centrality (analysis and map J. PreiserKapeller, 2015) 84 ORBIS Stanford Geospatial Network Model of the Roman World—visualisation of the nodes (= places) coloured according to their closeness-­centrality (colour scale from red/ low centrality to green/high centrality; analysis and map J. Preiser-Kapeller, 2015) (color online) 85 ORBIS Stanford Geospatial Network Model of the Roman World—identification of clusters (red) and sub-clusters (green) with the help of the Newman-algorithm (analysis and map J. Preiser Kapeller, 2015) (color online) 86 ORBIS Stanford Geospatial Network Model of the Roman World—visualisation of routes with a “cost” of maximum one day’s journey between two places (analysis and map J. PreiserKapeller, 2015) 87 Two-mode network of places (red nodes) and commodities (green nodes) exported from or imported to them as narrated in the “Periplus of the Erythraean Sea” (data: E. H. Seland, http://bora.uib.no/handle/1956/11470; visualisation: J. Preiser-Kapeller, 2015) (color online) 88 One-mode network of commodities due to their common export from or import to places as narrated in the “Periplus of the Erythraean Sea”; nodes sized according to their degreecentrality (data, visualisation and analysis as above) 89 One-mode network of places due to their common export from or import of commodities as narrated in the “Periplus of the Erythraean Sea”; nodes sized according to their degreecentrality (data, visualisation and analysis as above) 90 One-mode network of places due to their common export from or import of commodities as narrated in the “Periplus of the Erythraean Sea” visualised on a geographical map (the links indicate ties of similarity due to the exchange of the same goods, not direct ties of interaction); nodes sized according to their betweenness-centrality (data, visualisation and analysis as above)91 One-mode network of places due to their common export from or import of commodities as narrated in the “Periplus of the Erythraean Sea” visualised on a geographical map; identification of seven clusters of nodes (of different size) with the help of the Newman-algorithm (data, visualisation and analysis as above) 92

  List of Figures 

xi

Fig. 3.17 Frequency distribution of degree values of nodes in the network model of potters from Roman potter shops (of terra sigillata) of Rheinzabern (Tabernae, ca. 150–270 CE) due to the co-occurrence of commonly used hallmarks (data: MEES (2002); graph and analysis: J. Preiser-Kapeller, 2015) 93 Fig. 3.18 The network model of potters from Roman potter shops (of terra sigillata) of Rheinzabern (Tabernae, ca. 150–270 CE) due to the co-occurrence of commonly used hallmarks; nodes are arranged in the eight groups of potters identified by Mees (data: MEES (2002); network modelling and graph: J. Preiser-­ Kapeller, 2015) 94 Fig. 4.1 Relationship between distance from Rome and distance discount of six documented grain prices (author’s own reconstruction of figure 2.2 in Temin, The Roman Market Economy, 43) 110 Fig. 4.2 Abstract example of network distance. Nodes represent traders, lines represent social network edges, and large circles represent sites/settlements. When trader A successfully buys an item from trader B, the “distance from trader” to the production site is five whilst the “distance from site” to the production site is two 117 Fig. 4.3 Example results from one experiment with the following variable settings (variables in brackets): integration (proportion-inter-site-links) = 0.002; reliability information (local-knowledge) = 1; inter-market transport cost (transportcost) = 0.01. Results for “distance from trader” shown on the left, results for “distance from site” shown on the right. These results illustrate some of the general trends across all simulation results: high variability of prices overall, generally lower price in production site, generally a correlation of price with distance although mean prices are hiding interesting variability119 Fig. 5.1 The Planar Maximally Filtered Graph for stocks traded in the US equity markets (1995–1998), Tumminello et al., “A Tool for Filtering Information in Complex Systems,” 10,423, Fig. 2. (Copyright (2005) National Academy of Sciences, U.S.A.)134 Fig. 5.2a A systemic structure of obesity, foresight. (Vandenbroeck, Goossens, and Clemens, Tackling Obesities, 76, Fig. 12) (for an interactive map, with separate view of the nine groups see http://www.shiftn.com/obesity/Full-Map.html)135

xii 

List of Figures

Fig. 5.2b A systemic structure of obesity, linkages between psychology and food environment (Vandenbroeck, Goossens, and Clemens. Tackling obesities: “Map 15—Full generic map: linkages between the psychology and food environment areas—Linkages between these two areas are predictably dense, with 14 arrows going from psychology to the food environment and nine arrows in the other direction: (*) Key tail variables in the psychology area are education, media availability, socio-cultural valuation of food, perceived lack of time, stress and food literacy. They link into demand-side factors such as demand for health and the social pressure to consume. But supply-side variables are also triggered: the food industry’s business model is grafted onto what people want. (*) Tail variables in the food environment area are dispersed, with one arrow only leaving each of the variables. They include food exposure, food abundance, social pressure to consume, and industry’s desire to maximise volume. They drive three variables in the psychology area: exposure to food advertising, perceived lack of time, and psychological ambivalence” 136 Fig. 5.3 Systems Biology of Human Aging. (Furber, Systems Biology of Human Aging, available at http://www.LegendaryPharma. com/chartbg.html; appearing also in Lima, Visual Complexity, online electronic database: http://www.visualcomplexity.com/ vc/project_details.cfm?id=521&index=40&domain=Biology)138 Fig. 5.4 Taxation flow in the Roman empire inspired by Hopkins’ “Taxes and Trade” model. A revised version of Davies, “Linear and Nonlinear Flow Models,” 140, Fig. 6.6. The (here added) cursive lines represent how money “trickles-­down” from the frontiers back to the middle-zone via economic transactions 140 Fig. 5.5 Flow chart of resource movement; inspired by Davies, “Linear and Nonlinear Flow Models,” 150, Fig. 6.14: Flow chart of resource movement, model 6: Modified to incorporate bandwidths, motors, gates, and reservoirs 142 Fig. 5.6 Roman legal institutions for using money in private exchange 146 Fig. 5.7 Causal loop diagram for legal institutions from three legal systems operating in the Roman empire 153 Fig. 7.1 Map of the study area with the indication of all town sites mentioned in the text. (Map by D. Van Limbergen) 207 Fig. 7.2 First phase of the town of Potentia, with indication of the initial forum square (F) and the temple for Jupiter (T). (Map by PVS team) 215

  List of Figures 

Fig. 7.3

xiii

Map of Potentia, based on the integration of survey and excavation data. (Map by PVS team) 216 Fig. 7.4 Detailed plan of the monumental central area of Potentia, based mainly on the survey results by the PVS. (Map by PVS team)219 Fig. 7.5 Map of the town site of Trea, based on the survey results by the PVS. (Map by PVS team) 223 Fig. 7.6 Detailed plan of the monumental central area of Trea, based on the survey results by the PVS. (Map by PVS team) 224 Fig. 7.7 Reconstruction of the eastern residential quarter of Tifernum Mataurense, based on aerial photography. (Catani and Monacchi, Tifernum Mataurense, 245, fig. 63) 230 Fig. 7.8 Map of Sentinum based on the integration of excavation and geophysical survey data. (After Medri, “Materiali”, 213, fig. 3.1.12)231 Fig. 7.9 Map of the town site of Ostra, with the indication of a residential quarter individuated through aerial photography (B). (After Boschi and Silani, “Aerofotografia e geofisica”, 79, fig. 7) 232 Fig. 7.10 Plan of the domus in the area “La Fenice” at Senigallia. (After Salvini, Area archeologica, 22) 233 Fig. 7.11 Reconstruction of the Republican domus under the Imperial temple complex of Urbs Salvia. (After Montali, “Considerazioni”, 132, fig. 16) 234 Fig. 7.12 The Imperial Domus dei Coiedii at Suasa, with indications of the earlier Casa ad atrio and the Casa del primo stile. (After Campagnoli, “Fasi edilizie”, 320, fig. 1) 235 Fig. 7.13 Plan of the Imperial domus at Tifernum Mataurense. (Tornatore, “Domus con mosaici”, 883, fig. 2) 237 Fig. 8.1 Map displaying the towns and cities with official selfgovernment (n = 446 and 13 possible; note that only 428 are located and plotted on the map) 278 Fig. 8.2 Heat map of the self-governing cities and communities in the late second to early third centuries CE; the radius used for this heat map was 15 km 279 Fig. 8.3 The official self-governing cities with a 15 km radius buffer drawn around them 279 Fig. 8.4 Number of cities as listed by Broughton, Roman Asia Minor in Hellenistic/late republican Anatolia 280 Fig. 8.5 Number of cities as listed by Broughton, Roman Asia Minor in Flavian-­Severan times and the third century CE in Anatolia 280

xiv 

List of Figures

Fig. 8.6 Fig. 8.7 Fig. 8.8 Fig. 8.9 Fig. 8.10 Fig. 8.11 Fig. 8.12 Fig. 8.13

Fig. 8.14 Fig. 8.15

Fig. 8.16

Fig. 8.17

Fig. 8.18

Map of all theatres located in Asia Minor (n = 143 + 11 possible theatres) 281 Map of the datable theatres up to c. 30 BCE (n = 59 + 1 place with multiple theatres) 281 Map of the datable theatres up to c. 300 CE (n = 95 + 6 places with multiple theatres + 9 places with expanded seating areas) 282 Map of all baths in Asia Minor (n = 84 + 2 possible) 282 Map of all the datable baths until c. 100 CE (n = 23 plus 7 places with multiple bath-houses) 283 Map of all the datable baths until c. 300 CE (n = 44 plus 24 places with multiple bath-houses) 283 Map displaying the sizes of all measurable cities and settlements in Roman Asia Minor (n = 224)284 Kolossai from the air; note the semi-circular recess of the theatre on the east of the central hill. The city probably extended north towards the river, where the necropolis was located284 Rank-size plot for the settlements of Asia Minor (n = 168) using area in hectares 285 Rank-size plot on logarithmic scales for the settlements of Asia Minor (n = 168) using area in hectare. The dashed grey line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fit 285 Rank-size plot on logarithmic scales for the settlements of Asia Minor (n = 168) together with modelled sizes of the unmeasured official cities (n = 300), both using area in hectare. The dashed grey line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fitting trend line following a power-law 286 Rank-size plot on logarithmic scales for the self-governing settlements of the Roman province of Asia (n = 55) using area in hectare. The grey line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fitting trend line following a power law 286 Rank-size plot on logarithmic scales for the self-governing settlements of the Roman province of Lycia et Pamphylia, Pisidia (n = 46), using area in hectare. The grey dashed line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fitting trend line following a power law. This province incorporated large parts of the region of Pisidia in the second century CE 287

  List of Figures 

xv

Fig. 8.19 Rank-size plot on logarithmic scales for the self-governing settlements of the Roman province of Cilicia (n = 18) using area in hectare. The grey dashed line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fitting trend line following a power law 287 Fig. 8.20 Rank-size plot on logarithmic scales for the self-governing settlements of Maiandros River valley (n = 11) using area in hectare. The grey dashed line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fitting trend line following a power law 288 Fig. 9.1 Private wheat prices (exclusive of uncertain dates/prices/ commodities) from Roman Egypt, 100–270 CE. (Sources: Rathbone, “Prices and Price Formation in Roman Egypt,” 217–233; Rathbone and Von Reden, “Mediterranean Grain Prices,” Table A8.12. I have only used prices in which both the year and the price was confirmed with absolute certainty. An odd state price of 6 denarii per artaba in 246 CE was ultimately left out because it is unclear whether the nominal parity between denarii and tetradrachmai held at this time— hence, the price is uncertain) 304 Fig. 9.2 Nominal wheat prices (drachmai/artaba) to 275 CE (Harper, “People, Plagues, and Prices” 816) 305 Fig. 9.3 State prices of wheat in Roman Egypt, 100–270 CE. (Sources: Rathbone, “Prices and Price Formation,” 217–233; Rathbone and Von Reden, “Mediterranean Grain Prices,” Table A8.12 I have chosen to use a third-order polynomial as, with such a limited amount of data, I wanted to use as few constants as possible)307 Fig. 9.4 Two measures of Roman legionary pay (160–211 CE). (Sources: Fineness figures through 192/3 CE come from Butcher and Ponting, “The Beginning of the End”; Elliott, “The Acceptance and Value of Roman Silver Coinage”. Figures from 193/4 through 211 are from Gitler and Ponting, The Silver Coinage of Septimius Severus and His Family, 55–57. Weights are those of Duncan-Jones, Money and Government, 227; Walker, The Metrology of the Roman Silver Coinage, 3–18) 317 Fig. 10.1 The spread of the Black Death 333 Fig. 10.2 The spread of the Justinianic Plague (Red = 541; Green = 542; Blue = 543; Orange = 544–547) 339

List of Graphs

Graph 7.1 Diagram of shares public and private architecture at Potentia Graph 7.2 Diagram of shares public and private architecture at Trea

221 226

xvii

List of Tables

Table 3.1 Table 3.2 Table 3.3

Table 6.1 Table 6.2 Table 6.3 Table 6.4 Table 7.1 Table 7.2 Table 7.3 Table 8.1

Newman-cluster numbers and regions included in these clusters 66 Comparison of network measures for the network model of riverine transport in period I (first to fifth century CE) and period II (sixth to ninth century CE) 81 The network model of potters from Roman potter shops (of terra sigillata) of Rheinzabern (Tabernae, ca. 150–270 CE) due to the co-occurrence of commonly used hallmarks; structural quantitative measure for the network models of the eight groups of potters identified by Mees 93 Coding of intensity measures of development 172 Summary of the number of types per functional group, in the two different periods 183 Parameters of socio-economic complexity 187 Causal factors and mechanisms of complexity development at Düzen Tepe and Sagalassos with indication of relative intensity of each process 193 Estimated number of residents per house size 239 House sizes in central Adriatic Italy ranked according to consecutive 200 m2 bands, with the exception of the >1000 m2 category 240 House sizes in central Adriatic Italy ranked according to adjusted 200 m2 bands 241 Number of located official cities and communities (n = 446 and 13 possible) using the provincial division of 117 CE and the geographic regional division 261

xix

CHAPTER 1

Introduction: Finding a New Approach to Ancient Proxy Data Koenraad Verboven

1   What This Book Is About This book is about a big problem in economic history research: how to study economic development in societies that lack archival records or other written sources suitable for the familiar cliometric analyses used by economic historians. Ancient economic historians have been acutely aware of this challenge for many decades. Since the 1980s it led them to look for theoretical models to bridge the gap between their shaky empirical data and macro-level realities. In the early 2000s New Institutional Economics combined with neo-Malthusian models was hailed as the new paradigm that allowed tying together the data into a meaningful explanatory narrative of growth and development. Despite growing criticism1 and the rising popularity of climate historians,2 NIE remains the dominant framework used by ancient economic historians today. 1 2

 For example, Verboven, “The Knights Who Say NIE”.  See for instance Harper, The Fate of Rome.

K. Verboven (*) Department of History, Ghent University, Ghent, Belgium e-mail: [email protected] © The Author(s) 2021 K. Verboven (ed.), Complexity Economics, Palgrave Studies in Ancient Economies, https://doi.org/10.1007/978-3-030-47898-8_1

1

2 

K. VERBOVEN

Over the same past decades, however, economic archaeology also boomed. The close collaboration between archaeologists and physical anthropologists, climate scientists, and natural scientists in general led to huge advances in abilities to identify, collect, and interpret material remains as proxies for economic interactions and performance. Ancient economic historians enthusiastically embraced the sets of proxy data produced by archaeologists but interpreting them proved excessively hard. The datasets derive from the material record but the patterns they display are heavily determined by archaeological classifications and methodologies unfamiliar to historians. Archaeologists in turn struggle to factor in human governance, structured by institutions that are rooted in shared cultural beliefs and transmitted verbally and symbolically in ways that are invisible in the material record. The aim of this book is to stimulate the necessary collaboration between economic archaeologists and historians to overcome these difficulties by offering a theoretical and methodological framework to evaluate and integrate archaeological proxy data and data modelling in economic history research. For reasons I explain below, this is not possible by relying only on the now familiar frameworks of mainstream economics—whether neoclassical, Keynesian, neo-Malthusian, or neo-institutional—even though they continue to provide useful insights in other respects. Instead we explore and advocate the paradigm of complexity economics, developed in the 1980s and 1990s as a complement and an alternative to mainstream equilibrium economics. I will provide a more elaborate account of complexity economics below. Suffice now to say that it studies societies and economies as “complex adaptive systems” consisting of components and agents (human and non-human) that act upon and respond to real and (in the case of humans) anticipated events. These actions produce emergent patterns that in turn feedback into the agents’ behaviour. This paradigm, we argue, provides an appealing theoretical framework to interpret archaeological proxy data because it does not interpret these a priori as a reflection of the outcomes of linear processes of supply, demand, and distribution, determined by production factors, institutions and so on. Instead the data are treated as reflecting the outcomes of non-linear network dynamics. The advantage for historians and archaeologists is that methodologically the framework calls for network analyses and agent-based modelling and thereby opens up for analyses source data that provide only anecdotal evidence or intuitive impressions when interpreted in the light of mainstream equilibrium economics.

1  INTRODUCTION: FINDING A NEW APPROACH TO ANCIENT PROXY DATA 

3

The rest of this introduction discusses more in detail the problems involved in the interpretation of proxy data and the importance of models and theories to address them. After briefly situating the role of models in ancient history and archaeology, I discuss the current limitations of proxy data, survey previously advocated theoretical models, and discuss the advantages of “complexity economics”. This book is situated in the tradition of reflective works on the use of theories and models in ancient economic history.3 Classicists, historians as well as archaeologists, are often wary of theories and models. To reject them in the case of economic history, however, implies accepting the assumption that all the information we need to build a reliable image of ancient economies and to explain economic developments is locked in the scanty empirical data that are available to us, and that common sense suffices to unlock this information and to bridge any gaps that might be left. This is highly improbable given the state of our sources and the imperfections of human intellect. One may not always like them but models are a necessary evil and theories are what tie them together. That being said, however, we need to resist the temptation to cherry-pick the models we like and pretend to make sense of the data we have. The social scientists who usually make these models rarely have historical questions in mind and are generally unfamiliar with the messiness of historical or archaeological sources and the technical difficulties involved in interpreting them. It is important, therefore, that we reflect on the models and theories we use. They all have limitations and pitfalls that we need to avoid. That, however, is equally true for archaeological proxy data, on which historians have much less reflected. Interdisciplinary approaches in archaeology the past decades have massively produced new proxy data on ancient economies. Soil and climate data reveal productivity levels. Ice-core samples show air pollution caused by mining. Skeletal remains reveal differences in diets and health status. Isotopes in dental enamel show mobility of human populations and livestock. Newly identified parts and remains document the wide spread of advanced technical devices and hydraulic technology. Sedimentary budgets of rivers reveal changing agricultural methods. The list continues to grow. New digital tools make it possible to 3  See for instance Verboven and Erdkamp, Structure and Performance in the Roman Economy; Manning and Morris, The Ancient Economy; Scheidel and Von Reden, The Ancient Economy; Remesal Rodríguez, Revilla Calvo, and Bermúdez Lorenzo, Cuantificar las economías antiguas; for an earlier example see Finley, Ancient History.

4 

K. VERBOVEN

store these data in (big) digital data collections suitable for rapid information retrieval or for feeding into visualisation or modelling software. Integrating the results in economic history research has profoundly changed the way we look at and think about ancient economies. But the process of collecting, classifying, processing, and visualising data is not in itself revealing the dynamics underneath that caused the patterns we find in the data. It is not telling us how economic processes and outcomes are connected to broader societal structures and dynamics— social, political, or cultural. Millions of sherds recorded in a database, processed through graphs, plotted on maps, will not tell what was in the minds of their makers, the obstacles they faced, and how they overcame them. What does it mean if a dataset shows a diachronic shift from wheat to barley? Does the increased mining activity documented in the ice-core samples signal high economic performance, or does it merely show that the Roman state devoted massive resources to exploit silver and gold deposits in much the same predatory way as the Spanish crown did in South America more than a millennium later?4 Are the disease burdened bones from Pompeii showing us how unhealthy urban populations were, or are they indicating efficient coping strategies for diseases that would otherwise have killed their victims? What is the outcome of all this in terms of living standards and well-being? And whose living standards and well-­ being are we talking of? It is hard to establish how patterns in disparate and discontinuous datasets are related to each other. What is the connection, for instance, between patterns in amphorae finds and output of wine in production areas, given that barrels and wineskins have left little or no traces in the archaeological record? It is even harder to establish how they relate to patterns in non-material historical reality. Can we link urbanisation patterns to institutional changes? How many micro-level studies do we need to make statements about trends at meso- and macro-levels? Uncertainty increases every time we posit connections between datasets or we generalise from sample sets. Nevertheless, these questions need to be addressed. Economic history is not just about documenting change. It is about understanding and explaining it. Why was the Roman Empire so good at producing and distributing not just the basic necessities of life, but so much more that constitutes “a good life” or “civilisation”: high-­ quality tools, materials, and know-how to build comfortable houses, high-­ calorie, protein-rich food, good shoes, basic health-care, art, and 4

 Scheidel, “In Search of Roman Economic Growth”, 50.

1  INTRODUCTION: FINDING A NEW APPROACH TO ANCIENT PROXY DATA 

5

entertainment? Was it markets? Administered trade? Efficient institutions? Who benefitted from Rome’s economic success? Only elites and city dwellers or rural populations as well? And why did it stop? If Roman levels of technology and logistics were so high, why did they fall back instead of break through? Roman economic performance seems to have reached sixteenth- or maybe early seventeenth-century European levels, but these were still primitively low when compared to modern economies. The Great Divergence at that time was only visible yet with hindsight. Quantification and digital data processing become meaningful only when the data provide reliable proxies to measure or express levels of production, distribution, changes in supply and demand, and all the rest we believe determines peoples’ welfare and well-being. There is no reliable method to convert ancient proxy data into the economic indicators used in economics and economic history.5 Attempts to guestimate economic indicators so far have used theoretical models and comparative evidence to create “matrices of possibilities” (in the words of Keith Hopkins6) that would accommodate the sparse textual evidence we have. Given the kind and quality of data available in the material record, it is unlikely that it will ever be possible wholly to dispense with this “controlled conjecture” approach. Nevertheless, the impressive amount of new empirical data documenting economic phenomena and the analytical firepower provided by new software tools cannot be ignored. The potential they offer for data modelling is huge. One of the greatest challenges today, therefore, is to find ways to bridge the gap between the theoretical/comparative approach, and the new processing capabilities we now have for a greatly increased body of empirical data. We need a methodology that connects processed empirical data to the assumptions derived from theoretical models and comparisons, allowing us to corroborate or reject these assumptions. But where shall find our assumptions? The Finleyan model of the socially embedded status-driven economy, inspired by Max Weber and Karl Polanyi, was intellectually stimulating and elegant in its logic but failed to explain the growth in material output and energy consumption, the increasing diversification of cash crops, industry and services, the growing levels of skilled labour, and capacities to co-ordinate these in larger units, as witnessed by the quality and quantity of output. In the  Verboven, “Ancient Cliometrics”.  Hopkins, Conquerors and Slaves, 19–20; Hopkins, “Taxes and Trade”, 43; see the discussion in Verboven, “Ancient Cliometrics”, 347–50. 5 6

6 

K. VERBOVEN

1980s Keith Hopkins advocated a Keynesian “taxes and trade” model. In this model the different locations where taxes were levied and where they were spent (primarily Rome and the frontier provinces) fuelled trade because “tax-payers had to sell goods in those distant parts in order to earn the money with which to pay next year’s taxes”. This need gave rise to “a complex network of trade. … Local trade fed into medium-range and long-range trade and vice versa.”7 Hopkins was the first to use quantified archaeological (shipwrecks) and numismatic data (hoards and die-­ estimates) to substantiate his models. At that time only a fraction was available of what we have today and numerous pitfalls were not yet corrected. Patterson’s dating method for shipwrecks was not ideal and the distribution he established was heavily distorted by the change from amphorae to barrels as preferred containers for wine.8 Scholars since the 1990s have taken up Hopkins’ preference for mainstream economics. Neo-institutional and neo-Malthusian models seem well suited to explain both the quantitative growth that the Roman economy experienced (thanks to its favourable institutional framework and relative peace), and its limitations and subsequent contraction (due to Malthusian checks). Both, however, stay close to the basic neoclassical tenet that economic systems tend towards a state of general equilibrium in which supply and demand are overall matched. This equilibrium depends on exogenous factors, such as particular climates favouring specific crops, particular property regimes supporting private or public investment, particular demographic regimes or cultural preferences shaping demand, and so on. Real life economies, of course, never achieve equilibrium; supply and demand change continuously under the influence of external forces— drastic ones such as wars or harvest failures, but also common ones as changing consumer tastes or erroneous decisions by producers. Nevertheless, the assumption that the drive towards equilibrium determines how an economic system plays out in real life (if left alone) has been at the heart of mainstream economics since Léon Walras published his Elements of Pure Economics (1874). From that perspective, the perfect “Walrasian” market serves as a yardstick to analyse real economies: what forces prevent equilibrium from emerging; what forces disturb equilibria 7  Hopkins, “Models, Ships and Staples”, 85; see also Hopkins, “Taxes and Trade”; Hopkins, “Rome, Taxes, Rents and Trade”; Hopkins, “Rents, Taxes, Trade and the City of Rome”. 8  Wilson, “Approaches to Quantifying Roman Trade”.

1  INTRODUCTION: FINDING A NEW APPROACH TO ANCIENT PROXY DATA 

7

that may have emerged?9 The paradigm has achieved a lot and proved useful to explain short-term economic developments in stable situations, when exogenous shocks are absent or mild enough to be absorbed. The evident downside of the approach, however, is that it starts from the question “why doesn’t reality conform to the model?” There is a clear tension between the equilibrium-drive inherent in the logic of walrassian markets, and the imperfections, contingencies, and uncertainties of life; the accidents, misunderstandings, ignorance, lack of information, miscalculations, and change of minds and hearts of real people.10 As Douglass North put it: “The rationality assumption of neoclassical economics assumes that the players know what is in their self-interest and act accordingly. Ten millennia of human economic history says that is a wildly erroneous assumption.”11 Not surprisingly, using Walrasian logic to explain economic history—to retrodict rather than to predict—has rarely been successful. Comparative advantage, to give only one example, may be a useful concept to explain why it made sense that Egypt exported grain and Italy wine. It is not very useful to explain why this eventuality materialised only from the first century BCE onwards and remained heavily state-regulated throughout the rest of Roman history. Part of the difficulty is that from the viewpoint of general equilibrium theory history is always a succession of exogenous shocks. This makes it hard for any economy to “discover” its equilibrium in the real world.12 Equilibrium economics deals with this by identifying the shocks and changing externalities, and establishing how these prevent or change the equilibrium predicted by economic theory. For instance, text-book ­wisdom predicts that pandemics increase the price of labour (now scarce) and decrease the price of land (now abundant). We assume that long-term strong and stable polities (such as the Roman Empire) support and are supported by resilient economic systems, capable of absorbing minor local 9  Cf. Arthur, Complexity and the Economy, 3: “[E]conomics early in its history took a simpler approach, one more amenable to mathematical analysis. It asked not how agents’ behaviors would react to the aggregate patterns these created, but what behaviors (actions, strategies, expectations) would be upheld by—would be consistent with—the aggregate patterns these caused. It asked in other words what patterns would call for no changes in microbehavior, and would therefore be in stasis, or equilibrium” (emphasis by the author). 10  On the advantages of equilibrium economics and its limits see Arthur, 3–4. 11  North, “Some Fundamental Puzzles in Economic History/Development”, 237. 12  The terminology is revealing. It suggests that equilibrium is really “out there”. It is merely hard to find because life keeps getting in the way.

8 

K. VERBOVEN

shocks (such as crop failures) and returning to a viable equilibrium after major shocks (such as the Antonine Plague or the third-century political crisis). But why would text-book economics be a reliable guide? Claiming that real economic systems tend towards theoretical states of equilibrium is an assumption derived from nineteenth-century mathematical theory, it is not based on empirical observation. Mathematical elegance is not enough to support the assumption that reality conforms to theory. As in physics, theoretical predictions that logically follow from mathematical models need confirmation from reliable observations before they can be accepted as proof that the model is correct. The logic of Walrasian economics may be undeniable, but its predictions have time and again been proven wrong because “it is stuck with an unworkable paradigm—applying to an unstable world concepts derived from the assumption of stability”.13 While equilibrium models are by definition static, social reality is dynamic. Empirical data from real-life economic systems show that, like all other social systems, they are dynamic and, therefore, inherently unstable. Change is not an externality, but a prevalent characteristic of economic and social systems. Equilibrium is not the natural (end) state of an economic system. It is a short-lived exception; a passing phase. As New Institutional Economics began its shift away from the neoclassical model (inspired by one of its founding fathers, Douglass North) and Behavioural Economics pinpointed flaws in neoclassical assumptions, another economic model was developed in the 1980s and 1990s. Complexity Economics embraces the view that disequilibrium is an inherent feature of economic systems and aims to study the dynamics of real economic systems.14 Physicists and mathematicians were involved from the start in developing its theory and methodology, trying to address the gap they perceived between the neoclassical model premised on mathematical logic and a reality that begged for a different kind of maths. Complexity Economics assumes that economic systems exist in reality as dynamic networks and can be studied as such. According to Complexity Economics, economic systems are inherently dynamic and unstable but not chaotic. They are “complex adaptive systems”, non-linear in their development, unpredictable in their behaviour, yet self-organised in the sense that they contain nested (complex adaptive) subsystems that co-­ evolve within the larger system. They consist of independent agents that  Freeman and Carchedi, Marx and Non-Equilibrium Economics, viii.  Beinhocker, The Origin of Wealth.

13 14

1  INTRODUCTION: FINDING A NEW APPROACH TO ANCIENT PROXY DATA 

9

interact with each other and adjust their actions and strategies to the perceived outcome of previous actions. Economic systems are themselves subsystems of larger social and ecological complex adaptive systems. Complex societies can be defined as complex adaptive social systems characterised by the interaction of many diverse interacting components (individuals, groups, organisations, etc.) and by elaborate diversified rule sets, some of which support societal subsystems, such as the economic system.15 A variety of techniques can be used to study the emergent patterns of complex adaptive systems. One of these, familiar to historians, is Network Analysis to visualise both the structure and behaviour of systems. Another, more familiar to archaeologists, is using the framework of thermodynamics to identify the controlling variables and thresholds of a society’s Socio-­ ecological system (SES). Yet another familiar to archaeologists, is agent-based modelling, whereby the results derived from computer simulations are compared to observed real-world patterns to verify or falsify assumptions regarding the cause of these patterns. Contrary to neoclassical economics, complexity economics analyses data as the outcome of network dynamics and the logic of change, rather than as deviations or confirmations of theoretical supply and demand curves. The questions then relate to the structure of these networks, the regularities in their behaviour and their degree of inter-connectedness. The advantage for ancient economic historians (and more generally economic historians working on preindustrial economies with poorly quantifiable data) is that it opens up for analyses source-data that are relevant to study economic development but are poorly suited for neoclassical analyses. Thus, for economic historians, complexity economics complements the more established (or familiar) neoclassical approaches.

2   A Brief Survey of the Chapters While some of the contributions in this volume are more empirical than others the emphasis is not on the concrete historical cases they discuss or the conclusions they reach but on theories and methodological questions. How can we improve the use of theories and models in ancient economic history rather than reflect on the intrinsic merits of this or that theory? Chronologically we cover broadly all of classical and late antiquity, although clearly some chapters have a more limited scope than others.  For a more elaborate discussion see the chapters by Verboven.

15

10 

K. VERBOVEN

The first part focuses on theoretical frameworks and methodologies. The second and third offer case studies, respectively, on urbanisation and epidemics. The first chapter, by Verboven, explores ways to connect the new-institutional framework (popular among ancient historians) to that of system theory (popular among archaeologists).16 It addresses the major problem of non-designed institutions—social norms and conventions—in New Institutional Economics. While scholars agree they are extremely important, they are very hard to model. A common approach, therefore, is to take them as unchanging externalities. I argue that this is not a valid approach when dealing with economic development in multi-cultural empires because incongruence between social norms and legal norms greatly increases the costs of enforcement and maintaining social order. Social systems theory offers a way to conceptualise institutional change at this level. Social system theory and complexity economics provide a theoretical framework to think about ancient realities. They offer models—simplified versions of reality that lay bare the essence of things and the relations between them. For theories to be useful, however, we need to confront them with reality. In historical research that means reliable empirical information about past realities. We need ways to transform raw historical source data into concepts that the models understand. But how can we control this process? And how can we avoid the well-known pitfalls of theoretical approaches: you only find what you are looking for, and you always find what you are looking for? Without a clear methodology the framework of complexity economics is powerless. Johannes Preiser-­ Kapeller and Tom Brughmans in Chaps. 3 and 4 respectively discuss two such methodologies: network analysis and agent-based modelling. Network analysis is the core methodology “par excellence” to study real-life complex adaptive systems, including social and economic systems. It aims to capture patterns that emerge from the interaction between the “nodes” (individuals, households, families, groups, sites, etc.). Such emergence is the key characteristic of a complex system, the properties of which cannot be reduced to the characteristics of the component parts. Preiser-­ Kapeller argues that “[n]etwork models can serve as proxies to estimate the range, density and complexity of past economic life”. An additional advantage is that the level of abstraction achieved through network 16  The chapter is useful also to links this book to a previous book in the same research programme: Verboven and Erdkamp, Structure and Performance in the Roman Economy.

1  INTRODUCTION: FINDING A NEW APPROACH TO ANCIENT PROXY DATA 

11

analysis is conducive to cross-cultural and -temporal comparisons, allowing us to capture trajectories of economic development. He illustrates his point with a study of the riverine transport network from Roman antiquity to the Early Middle Ages showing the reduction of organisational, economic, and infrastructural complexity from the Roman to the post-Roman period. Network analysis can be useful also to test hypotheses regarding the Roman Empire’s macro-economic “nature”. Preiser-Kapeller next uses the data from the ORBIS Stanford Geospatial Network Model to show how the empire was structured in twenty-five regional and supraregional clusters of higher internal connectivity, mostly thanks to their maritime or riverine connections. The cohesion of the empire depended on the links between these clusters. A network analysis of the Indian Ocean trade based on the Periplus, in turn, reveals that while the Egyptian harbours were important, the integration of the system as a whole depended on other intermediary nodes and circuits in the Arabian Peninsula, India, the Persian Gulf and Bay of Bengal. Network analysis allows us to see through the evident Roman bias of the author. The last case study presented by Preiser-Kapeller uses the Rheinzabern pottery stamps collected by Allard Mees. It shows different forms of internal organisation or cooperation, and the diffusion of technical skills. Network analysis is data driven. It is built from the ground up using available raw data. Gaps can be filled in to some extent; patterns may become visible even when pieces are missing. But there are limits to what is possible. Sometimes, network analysis is not an option. Computational (agent-based) modelling attempts to work around this problem. It is the second major research methodology of complexity economics. Computational modelling creates simulations of reality. Virtual agents are created with predefined preferences, and made to interact in environments affecting the agents’ predefined choice sets, for instance by the distribution of material and/or immaterial resources, and events impacting the outcome of choices. The simulation is then run and the outcomes are observed. By changing variables modellers can study the effects different variables have on outcomes. The closer the match between computer-­ generated outcomes and real data, the higher the statistical likelihood that the set of chosen variables and their values also conforms to reality. Agentbased modelling has become popular in archaeology the past decades. Tom Brughmans in Chap. 4 here argues how this methodology is useful also to analyse problems in Roman economic history. He illustrates this with a simulation of the correlation of tableware prices and distance from

12 

K. VERBOVEN

their place of production. The objective is to study economic integration based on the common assumption of economists that well integrated competitive markets produce a strong correlation between the price of goods in a location and the distance to the production region of that good. The simulation is meant to overcome the well-known problem that reliable data to evaluate the degree and mechanisms of integration are simply not available for the Roman economy. It does so by translating the components of competing hypotheses—in this case of Peter Temin and Peter Fibiger Bang17—into variables driving the simulation and subsequently comparing the outcomes with archaeological distribution maps. Brughmans first designed his model, MERCURY, to represent and test the availability and reliability of commercial information on Roman tableware and then expanded it to include also the distance away from its production place. The virtual agents in the model are traders who are interconnected in a social network through which commercial information and products move. The model generates simulated distributions of tableware, which can then be compared to the data of excavated and published sherds in the ICRATES database. When variables were set to represent conditions of low integration, the simulated distributions produced by the model were incompatible with the observed real distribution data. This suggests a high likelihood that markets for tableware were well-integrated. Temin’s hypothesis, however, of a single integrated Mediterraneanwide market economy, is not confirmed. Prices do rise with “distance”, but exponentially—not merely as a result of increasing transport costs. The last contribution in the first part of the volume deals with visualisations. These play an important role in network analysis and complexity economics. Graphs, matrices, and flowcharts not only illustrate or support ex ante intellectual conclusions. They serve as cognitive tools to better understand relationships between concepts or data that have no visual dimension. They are created to express pre-defined relations between concepts—often generated by computers from databases—and provide data for further analyses. Merav Haklai in Chap. 5 discusses this role of visualisations and offers a complexity-oriented model to analyse the institutional environment for the use of money in the Roman world. Based on this visualisation-driven analysis she argues that changes in legal institutions governing the use of various forms of money were not (only) the result of conscious redefinitions by Roman jurists. Rather, the potential for change  Temin, The Roman Market Economy; Bang, The Roman Bazaar.

17

1  INTRODUCTION: FINDING A NEW APPROACH TO ANCIENT PROXY DATA 

13

was created as an emergent property of the interrelations between existing arrangements within the Roman legal system and with non-Roman legal systems. Jurists then used this potential to formulate new remedies to problems that transcended the possibilities of existing remedies. Thus, the potential for institutional changes was generated by the dynamic interplay of material and intellectual interactions between money-users and jurists. The second part of the book comprises three contributions on urban systems. The first, by Dries Daems in Chap. 6 applies complexity analysis on pottery finds as proxy data to study the increase in social, economic, and political complexity at late Achaemenid and early Hellenistic Sagalassos and Düzen Tepe, two village settlements located 1.8 km from each other. Both communities were very similar in terms of socio-economic complexity until around 200 BCE. Yet, Düzen Tepe was abandoned in the second century BCE at the time when Sagalassos began its development into a major regional centre. Both sites were centres of pottery production. The author looks at resource procurement, material production, and distribution, showing how the recorded pottery finds reflect a much wider transformation from the second century BCE onwards towards increased complexity driven by changing structures of supply and demand, capital investment, institutionalisation, division of labour, technological development, and property rights. Each of these causal factors can be traced in the pottery data. They show that Düzen Tepe was and remained an inward-­ oriented village until its disappearance in the second century BCE. Sagalassos, on the other hand, developed further into an urban regional centre. This was made possible through loopback effects that greatly increased the initially modest and historically contingent opportunities for non-subsistence activities created by changing political conditions in Sagalassos. Dimitri Van Limbergen and Frank Vermeulen in Chap. 7 aim to establish new and more reliable parameters to estimate population figures of ancient cities that are not or only partially excavated. On the basis of a case study of Potentia and Trea, using remote sensing techniques, they propose new realistic figures for the total space taken by roads, urban infrastructure, and public architecture and look for realistic estimates of inhabitants per hectare of residential areas depending on the type of housing. The authors established for both towns that the residential area occupied respectively about 60 and 50 per cent of the town, with large domus being the dominant type of residential buildings. Based on comparative evidence from other sites the authors subdivide domus into bands

14 

K. VERBOVEN

depending on size and suggest occupation ratios. This provides new estimates for population densities in the range of 75–125 persons per hectare, considerably below the commonly used density figures of 120–150 persons per hectare, based on Pompeii and Herculaneum combined with comparative figures from medieval and early modern times. The authors suggest these low-density figures reflect the nature of small towns as service centres rather than population centres. Rinse Willet in Chap. 8 argues that “[c]ities and their formation are good examples of how the actions of a number of individual agents produce structures that display complexity and behave as complex systems”. Empirical studies of modern cities and urban networks reveal scalar properties (or self-similarity)—similar patterns and functions emerge at different scales; for instance, neighbourhoods have shrines or chapels, larger units churches or temples, cities cathedrals or temples servicing a wider area, and so on. The emergence of these properties is less studied and understood for premodern urban systems. Willet argues that cities in Roman Asia Minor were part of wider urban networks. We can study these as complex adaptive systems by using archaeological proxy data, such as (but not limited to) surface sizes and public buildings. The uneven geographic distribution of cities cannot be explained by topography and environment alone. The increase in the number of cities in the Roman period and their geographic spread reflect institutional changes betraying increased division of labour, social hierarchy, and inequality—signs of increasing complexity in the overall urban system. Public buildings provide more information. Willet focuses on theatres and bath-houses. Both required considerable investments; the latter, in addition, substantial infrastructure and organisation to ensure water and fuel supply—both signifying greater societal and economic complexity. The data show a dramatic increase of numbers and sizes during the Roman period. Rank-size analysis, using surface areas, shows a convex pattern, which may reflect a less than perfect systemic integration. The province of Asia, however, shows markedly less convexity, indicating a greater, although still imperfect, integration. We cannot yet make out whether this integration was (also) the result of economic integration or (merely) of political factors, but it seems to be path-dependent and related to the urbanisation process during Roman times. Roman administration and economic commercialisation increased the interdependence of cities. Hierarchy in the urban system was an emergent property of this development.

1  INTRODUCTION: FINDING A NEW APPROACH TO ANCIENT PROXY DATA 

15

The last part of this volume consists of two contributions, respectively on the Antonine and the Justinianic Plague. Colin Elliott in Chap. 9 notes that the debate on the (potential) effects of the Antonine Plague is inspired by equilibrium economics, with demographic contraction leading to a loss of labour supply (in turn creating a upward pressure on wages), but an excess supply of land and money. The roughly doubling of price levels between the 160s and 190s was thus easily explained as a result of the Antonine Plague. Elliot, however, notes that it is impossible to construct statistically significant trend lines from the limited dataset of private wheat prices we have. Thus, the hypothesis that there was a sudden inflation around 160/190 caused by the Plague cannot be confirmed statistically. In stark contrast to this the data for (non-market) state prices show a statistically significant trend line of stable prices until at least 216 CE. Rather than having a stabilising effect on markets, Elliot argues that such a state policy would have disturbed the market for private wheat sales, thus not stimulating equilibrium prices, but preventing them. If the rise of private price levels was induced by population decline due to the Antonine Plague, moreover, neo-Malthusian logic would predict a return to “normal” equilibrium conditions afterwards. The (presumed) fact that price levels did not return to first- and second-century levels would then imply that second-century prices reflect a disequilibrium due presumably to the effects of Egypt’s role as Rome’s main supplier of wheat. Conspicuously, real wages (contrary to nominal wages) remained stable before and after the Antonine Plague. Military wages show a similar pattern for the second century and the Severan era. Elliott concludes that “[e]ither the plague was not as severe or the Roman economy was not as integrated as we have been led to believe; or perhaps these are both true”. One of the problems with comparative economic history so far is not only the scarcity of quantitative data from antiquity and the early Middle Ages—economic archaeology is rapidly changing that—but particularly the dissimilarity of available data. For late medieval and early modern Europe various archives provide useful proxy data to reconstruct various economic indicators. These data are lacking for the ancient world and the quantitative data we do have are impossible to translate into the economic indicators used in the cliometric history of Middle Ages and early modern period.18 Lars Boerner and Battista Severgnini in Chap. 10 test the potential of using the spread of respectively the Justinianic Plague and the  Verboven, “Ancient Cliometrics”.

18

16 

K. VERBOVEN

Black Death to measure and compare economic interaction. The authors showed in a previous article that the spread of the Black Death along trading routes reflects institutional determinants of trade, such as political borders or religion, as well as the intensity of trade and long-term growth dynamics. In this chapter they test the same methodology for the Justinianic Plague and investigate whether this allows the creation of a comparative perspective. Their results show a correlation with the Stanford ORBIS dataset, indicating (not surprisingly) the role of geography and transportation technology. The pattern of the spread of both epidemics is similar, but the speed is much slower in the case of the Justinianic Plague. The authors interpret this as an indication of relatively weaker economic interactions and lower levels of economic activity in Justinian’s Empire than in the Mediterranean and Europe during the Commercial Revolution—possibly because the former presented a phase of decline or incomplete recovery, the latter one of growth. Summarising the message of this book we can say that equilibrium economics is probably not the most useful theoretical framework to discuss economic developments in the Roman world, or to integrate archaeological and historical source data in a comprehensive analysis. There are not enough indications that the Roman economy was sufficiently integrated through markets to warrant an equilibrium-based approach. Complexity economics, using network analyses and agent-based modelling, allows us to analyse the Roman economy as a system integrated through multidimensional interactions—including but not limited to market relations. Rather than positing for analytical purposes the assumption that the system’s characteristics conformed to an intrinsic drive towards equilibrium, this alternative approach interprets these characteristics as emergent properties of underlying real interactions that left traces both in the material record and in textual sources. Acknowledgements  Last but no least it is my pleasure here to thank our supporters, collaborators, and funding bodies without whom this book would never have been possible. The work is part of the 2017–2021 research programme of the international network “Structural Determinants of Economic Performance in the Roman World”, generously funded by the Flanders Research Foundation (FWO). Six of the contributions19 originate from discussions at an international workshop 19  By Verboven, Preiser-Kapeller, Brughmans, Van Limbergen and Vermeulen, Willet and Elliot.

1  INTRODUCTION: FINDING A NEW APPROACH TO ANCIENT PROXY DATA 

17

held in Sagalassos, Turkey, in 2015 funded by this network and by the Sagalassos Archaeological Research Project. Special thanks are due to Jeroen Poblome, director of the Sagalassos Project for making this possible.

Bibliography Arthur, W. Brian. Complexity and the Economy, 2015. Bang, Peter Fibiger. The Roman Bazaar: A Comparative Study of Trade and Markets in a Tributary Empire. Cambridge; New York: Cambridge University Press, 2008. Beinhocker, Eric D. The Origin of Wealth: Evolution, Complexity, and the Radical Remaking of Economics. Boston, Mass: Harvard Business School Press, 2006. Finley, M.  I. Ancient History: Evidence and Models. New  York, N.Y., U.S.A.: Viking, 1986. Freeman, Alan, and Guglielmo Carchedi. Marx and Non-Equilibrium Economics. Cheltenham, UK; Brookfield, US: Edward Elgar, 1996. Harper, Kyle. The Fate of Rome: Climate, Disease, and the End of an Empire. Princeton: Princeton University Press, 2017. Hopkins, Keith. Conquerors and Slaves. Cambridge; New  York: Cambridge University Press, 1978. Hopkins, Keith. ‘Models, Ships and Staples’. In Trade and Famine in Classical Antiquity, edited by Peter Garnsey and Charles Richard Whittacker, 84–109. Cambridge Philological Society Supplementary Volume No. 8. Cambridge, 1983. Hopkins, Keith. ‘Rents, Taxes, Trade and the City of Rome’. In Mercati permanenti e mercati periodici nel mondo romano. Atti degli Incontri capresi di storia dell’economia antica: (Capri 13–15 ottobre 1997), edited by Elio Lo Cascio, 253–67. Bari: Edipuglia, 2000. Hopkins, Keith. ‘Rome, Taxes, Rents and Trade’. Kodai 6 (1995/1996): 41–75. Hopkins, Keith. ‘Taxes and Trade in the Roman Empire (200 B.C.–A.D. 400)’. Journal of Roman Studies 70 (1980): 101–25. Manning, Joseph Gilbert, and Ian Morris, eds. The Ancient Economy: Evidence and Models. Stanford, Calif.: Stanford University Press, 2005. North, Douglass C. ‘Some Fundamental Puzzles in Economic History/ Development’. In The Economy as an Evolving Complex System II, edited by W.  Brian Arthur, Steven N Durlauf, and David A Lane, 223–37. Reading: Addison-Wesley, Advanced Book Program, 1997. Remesal Rodríguez, José, Victor Revilla Calvo, and Juan Manuel Bermúdez Lorenzo, eds. Cuantificar las economías antiguas: problemas y métodos = Quantifying Ancient Economies: Problems and Methodologies. Collecció Instrumenta 60. Barcelona: Universitat de Barcelona. Edicions, 2018.

18 

K. VERBOVEN

Scheidel, Walter. ‘In Search of Roman Economic Growth’. Journal of Roman Archaeology 22, no. 1 (2009): 46–70. Scheidel, Walter, and Sitta Von Reden, eds. The Ancient Economy. New  York: Routledge, 2002. Temin, Peter. The Roman Market Economy. The Princeton Economic History of the Western World. Princeton, N.J.: Princeton University Press, 2013. Verboven, Koenraad. ‘The Knights Who Say NIE.  Can Neo-Institutional Economics Live up to Its Expectation in Ancient History Research?’ In Structure and Performance in the Roman Economy. Models, Methods and Case Studies, edited by Koenraad Verboven and Paul Erdkamp, 33–57. Brussels: Latomus, 2015. Verboven, Koenraad. ‘Ancient Cliometrics and Archaeological Proxy-Data. Between the Devil and the Deep Blue Sea’. In Cuantificar las economías antiguas. Problemas y métodos. Quantifying Ancient Economies. Problems and Methodologies, edited by José Remesal Rodríguez, Víctor Revilla Calvo, and Manuel Bermúdez Lorenzo, 345–71. Collecció Instrumenta 60. Barcelona: Universitat de Barcelona. Edicions, 2018. Verboven, Koenraad, and Paul Erdkamp, eds. Structure and Performance in the Roman Economy. Models, Methods and Case Studies. Collection Latomus 350. Brussels: Latomus, 2015. Wilson, Andrew. ‘Approaches to Quantifying Roman Trade’. In Quantifying the Roman Economy: Methods and Problems, 213–49. Oxford Roman Economy Series. Oxford: Oxford University Press, 2009.

PART I

Theoretical Frameworks and Methodologies

CHAPTER 2

Playing by Whose Rules? Institutional Resilience, Conflict and Change in the Roman Economy Koenraad Verboven

1   Framing the Problem In terms of the amount and variety of production and energy capture the Roman economic system—encompassing roughly a quarter of the world’s population—was one of the most successful in pre-industrial history. It was matched (probably) by Song China (960–1279) but structurally surpassed only when the combination of science and technology raised the reach of human achievements forever. Through its connections with the outside world—Arabia, Eastern Africa, India, and indirectly China—Rome shaped the first economic world system. Not everyone in the system benefited. Inequality was rampant—with a small “one-percenter” elite, maybe 10 per cent middling groups and a great mass of people hovering slightly

K. Verboven (*) Department of History, Ghent University, Gent, Belgium e-mail: [email protected] © The Author(s) 2021 K. Verboven (ed.), Complexity Economics, Palgrave Studies in Ancient Economies, https://doi.org/10.1007/978-3-030-47898-8_2

21

22 

K. VERBOVEN

above subsistence, like the undergrowth in a forest or occasionally like a dangerous undertow.1 Scholars usually interpret this success as the outcome of three factors: favourable institutional arrangements (including the Pax Romana) that lowered transaction costs; favourable climatic conditions; and technological innovations such as building concrete or water-powered mills and wheels. Clearly, there were links and cross-overs between these three, but for practical reasons I will address only the first factor: institutions. Rather than look at the effect of institutions on transaction costs, however, as many have done the past twenty-five years or so, I will focus on the relation between institutions and social networks. My aim is to present a framework derived from theories in social sciences to analyse deep institutional change in the Roman empire—not changes in the visible formal institutions, but in the invisible rules that produce the patterns of social life. The first and longest part of this chapter will tie together models in a useful framework. The second part applies these insights to the Roman empire.

2   Institutions and Institutional Change Institutions (laws, conventions, customs, etc.) are the man-made “rules of the game” that structure social interaction.2 They regulate the bestowal of social roles, determine which choices and actions are permitted and which results are acceptable. Property rights and tax regimes prescribe how control over resources is distributed, law codes and customs to what use they may legitimately be put. Different institutional arrangements explain why agents with similar personalities, driven by similar psychological needs and ambitions, facing similar challenges, make different choices and deploy different strategies with different outcomes. Different incentive structures laid out by institutions in different societies incite aggressive young males to become warriors, pirates, or football hooligans; alfa-types to become 1  Scheidel and Friesen, “The Size of the Economy and the Distribution of Income in the Roman Empire”, 1, on inequality as a structural feature of historical states and empire see Scheidel, The Great Leveler; compare also Piketty, Le capital au XXIe siècle; Milanovic, Lindert, and Williamson, “Pre-Industrial Inequality*”; Milanović, Global Inequality. On the size and development of the Roman economy see also Harper, The Fate of Rome; but see also the critical remarks by Haldon et al., “Plagues, Climate Change, and the End of an Empire”. 2  See Verboven 2015 for a more detailed discussion of New Institutional Economics and its use in ancient economic history.

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

23

chiefs, generals, captains of industry, or gang leaders; cunning minds to grow into Ulysses, political brokers, used car dealers, or stock traders. Thus, a society’s institutional make-up determines its overall performance, whether in terms of political control, military capacity, tribute extraction or economic production and distribution. But institutions come in many guises. The most common distinction is that between formal (or designed) institutions and informal (or implicit or organic) ones. The former are laws, decrees, statutes or procedures designed by agents or organisations endowed with the authority to formulate and impose them. They are based on perceived interests (material and immaterial) which they are intended to protect. Informal institutions are social norms, conventions, and customs that pose constraints on agents’ behaviour without having been consciously thought out.3 Both formal and informal institutions require monitoring and enforcement. Self-monitoring (based on moral principles or a desire to fit in) plays a role, but rule defection becomes attractive when compliance is perceived as not being in one’s personal or group interest. All societies, therefore, need ways to detect and punish cheaters and free-riders. Formal institutions are enforced following (institutionalised) procedures by special (institutionally) designated agents, using resources that are usually (but not always) given to them to perform that specific task. Informal institutions are enforced through emotional sanctions (shame, guilt, remorse, etc.) and, more importantly, social sanctions, ranging from gossip to social ostracism and lynching. Monitoring and enforcement of informal institutions is relatively effective and low-cost in close-knit face-to-face communities, where interdependence is high and reliable information is easily and swiftly obtained. It is difficult to achieve, however, in larger complex societies where information is often unreliable and agents form cliques to defend particularistic interests contrary to the interests of others. Monitoring and enforcement costs of designed institutions on the other hand are inhibitive unless they can rely on support from implicit social institutions, based on strong cultural beliefs about norms and values— including values regarding the legitimacy of the enforcement apparatus. Without a minimum of congruence between implicit social norms and

3  Alesina and Giuliano, “Culture and Institutions”. Propose “(formal) institutions” and “culture” as “more appropriate and less confusing” labels (902).

24 

K. VERBOVEN

designed institutions, therefore, the monitoring and enforcement costs of formal institutions imposes a drain on resources.4 In economic history, institutional arrangements have been used mostly to explain economic performance—as an explanans rather than an explanandum. Merely positing favourable institutions to explain performance, however, does not solve the problem. It shifts the question: where do these institutions come from? How do they change? The Roman empire is a school example of institutional change on a massive scale. The long-term stability of the empire and its economic success indicate that it succeeded in overcoming incongruities between formal and informal institutions. But how did it do this? By defining institutional change as an explanandum we move the debate to one of the central problems in new institutional sociology and economics: how to explain and model institutional change. One influential strand in social sciences posits that durable institutional change is the result of evolutionary selection at group levels. Driven by competition over scarce resources, groups (households, families, clans, tribes, cities, states, firms, etc.) with more efficient institutions drive out groups that are constrained by less efficient ones.5 The crux, however, is how to define efficiency. Early New Institutional Economics took a market-­centred view on the problem. Market competition weeds out inefficient organisations. Therefore, nations with more efficient market institutions were believed to outperform nations with less efficient ones. The former achieve stronger economic growth, becoming richer and more powerful. Ultimately nations with inefficient markets are either absorbed, subdued, or forced to adapt.6 Thus (supposedly) early modern England and the United Provinces outperformed the Spanish empire and France because they had more efficient institutions, which allowed their ruling elites to mobilise more financial and military resources and better co-­ ordinate military action. By the late eighteenth century England had become the leading European colonial and industrial power thanks to its liberal institutions and efficient capital market. This early teleological market-­centred view, however, has long been criticised and although it 4  Nee, “Norms and Networks in Economic and Organizational Performance”; Elster, “Social Norms and Economic Theory”; Verboven, “Cité et réciprocité. Le rôle des croyances culturelles dans l’économie romaine”. 5  Bowles, Microeconomics. Bowles and Gintis, A Cooperative Species, 121. 6  For a more fundamental(ist) view on markets as the natural outcome of evolution see Hayek, The Fatal Conceit.

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

25

continues to be cited in debates on the modernisation of Europe, it fails to convince as an explanatory model for real historical developments. Institutions that are inefficient from a market-economic viewpoint can remain stable for centuries because they are efficient to maintain a status quo in resource distribution that disproportionately favours elites who control the formation and enforcement of institutions. Polities that develop institutions that favour market competition may lose out when forced to compete, economically or militarily, with polities that do not, for instance because the latter establish effective monopolies or because the institutional make-up of the non-market polity is more effective in mobilising and co-ordinating military power, or even more mundanely because the other polity is bigger and/or geographically and ecologically more favourably situated.7 In a more general sense, however, efficiency and competition are still widely accepted as the keys to unravel institutional change. Efficiency to deal with crises (robustness, resilience) determines the survival of any institutional system. Military, political, or economic competition realigns resource distribution in favour of more efficient organisations—whether inside a complex society or between societies. Without competition, organisations that benefit from the way in which resources are (re)distributed, have no incentive to change the rules by which the social, economic, or political game is played. Market competition is part of this story. For instance, bearing Occam’s razor in mind, the rise of villa estates in second-­ century BCE Italy is best explained by assuming that this was the most efficient mode of production to maximise the estate owners’ income. Since they largely belonged to the same social elites from which political and military elites were drawn, formal institutional changes that facilitated villa management are best explained as driven by market demands experienced by the estate owners. Similarly, legal innovations, such as the actiones adiecticiae qualitatis, which allowed the use of slave managers to run a business, are best explained as attempts to solve agency problems for social elites who wished to invest resources in market oriented businesses. 7  See Ogilvie, “Whatever Is, Is Right?”, and there for more references; see for instance the debate on the guilds between Greif, Institutions; Dessi and Ogilvie, “The Political Economy of Merchant Guilds”; Ogilvie, “Whatever Is, Is Right?”; Ogilvie, “Guilds, Efficiency, and Social Capital”; Ogilvie, “Rehabilitating the Guilds”; Ogilvie, Institutions and European Trade; Epstein, “Craft Guilds in the Pre-Modern Economy”; Gelderblom, Cities of Commerce; Ogilvie, The European Guilds. For the now more popular state-centred approach, see Johnson and Koyama, “States and Economic Growth”.

26 

K. VERBOVEN

Clearly, however, Rome did not conquer the Seleucid empire or Gaul because its economy was more efficient, but because its army was. The demand for cheap labour in second-century BCE Italy may have drawn enslaved war captives, but these were not captured because the slave-mode of production in Italy was more efficient than traditional agriculture in the conquered territories. Caesar did not win power because the market favoured him, but because he was a better and luckier military commander. Rome was a huge metropolitan market, but Augustus well understood that market incentives alone would not suffice to secure the capital’s needs. The imperial food supply system—the anonna—was designed as a mix of direct control, compulsory services and subsidised trade. But if markets are merely part of the story and if institutional change is neither only driven by or necessarily directed towards better market institutions, why did institutional arrangements in the Roman empire give so much scope to free enterprise and market exchanges compared to other empires such as Han China? How important was market competition in institutional change in the Roman empire? Designed institutions (laws, statutes, procedures, etc.) are relatively easy to change in principle. In practice, however, conflicts of interests and entanglement with other institutions can obstruct even the most essential changes—especially in societies where resources are distributed unequally in favour of elites who monopolise or control the right to create or change formal institutions and determine their enforcement. Conflicts or competition over the distribution of resources—induced by exogenous factors (such as wars, epidemics, or resource depletion) or endogenous tensions (such as elite competition or social conflicts)—are the main reason why formal institutions are adapted. In the long(er) run changing cultural beliefs as well may lead to changes in designed institutions. Same-sex marriages are an example in our time. The right for mothers to inherit from their children ab intestate is a Roman example. Most are only minor changes (for instance Nero’s senatorial decree of 61 CE laying down new rules for sealing legal documents),8 but some introduce major changes. The Lex Aebutia, for instance—which generalised the formulary procedure in litigation making the edicts of praetor urbanus and the aedilis

8  Suet., Ner. 17; Paulus, Sent. 5,25,6; cf. Meyer, Legitimacy and Law in the Roman World, 165–68.

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

27

curulis the main sources of law—provides an example of a major change in Roman legislation in the second century BCE.9 But economic performance is not just determined by a society’s formal institutional framework. Cultural beliefs and values profoundly impact economically relevant choices and behaviour.10 They constitute informal rules that are much harder to change. While figures with moral authority, such as priests or charismatic leaders, have some influence, the fact that social institutions are ingrained in agents’ minds and habits and are subject to informal social sanctions makes them elusive to formal authorities. Many studies by economists and economic historians work around this problem by assuming that informal institutions change so slowly that in modelling or explaining institutional change they can simply be considered as important but static externalities.11 This is a defensible option for short- and mid-term studies in homogenous and stable societies, but it is not when we are dealing with societies experiencing phases of instability or rapid formal institutional change: exogenous shocks (as for instance violence or (civil) wars) and formal institutional changes have been shown to affect values and cultural beliefs.12 Modelling cultural beliefs as static externalities is certainly not an option in the case of long-lasting multilingual and multicultural empires, such as the Roman, formed through violent conquests, going through phases of regional conflicts and internal struggles. Not only are we then confronted with inevitable incongruities between formal institutions imposed by the victor and social institutions in force among the vanquished, but the legitimacy itself of enforcement institutions imposed by formal authorities becomes an issue. If a population refuses to interiorise a new rule imperial authorities need to invest more resources in rule enforcement, which adversely affects the system’s overall performance.

9  Gai., Inst. IV, 30; Gell., NA XVI, 10; Jolowicz and Nicholas, A Historical Introduction to the Study of Roman Law, 218–25; Schiller, Roman Law, Mechanisms of Development, 405. 10  See Alesina and Giuliano, “Culture and Institutions” for an overview. 11  Williamson, “The New Institutional Economic: Taking Stock, Looking Ahead”, 596; Eggertsen, “A Note on the Economics of Institutions”, 13. On the long persistence of cultural traits see for instance Putnam, Leonardi, and Nanetti, Making Democracy Work. For attempts to model the impact of institutional change on cultural beliefs and values see Alesina and Giuliano, “Culture and Institutions”. 12  See Alesina and Giuliano, “Culture and Institutions”, who stress the interaction between culture and formal institutions, rather than any one-sided causality.

28 

K. VERBOVEN

This poses a challenge if we want to understand long-term economic development in historical empires. It is not enough to identify the formal institutions that obstructed or stimulated production and distribution. We need to identify the social institutions and cultural beliefs underneath. Most of all, since multicultural/linguistic empires by definition contain different cultural traditions, religious beliefs, and local identities, we need to understand how such empires succeed in setting in motion changes in social rule systems guided by different cultural beliefs—ultimately changing these beliefs themselves. In other words: we need to understand “deep” institutional change. Attempts to model deep institutional change have mostly been based on game-theory.13 Game-theoretical models assume that social institutions are the unintended results of interactions between self-interested rational agents trying to maximise their own. Social institutions then emerge endogenously as Nash equilibria—that is as states of the game where no player can improve her returns by unilaterally changing strategy. The players interiorise the patterns they perceive that result from the strategies chosen and thereby create perceptions of reality and behavioural rules that govern the decisions they make in pursuit of maximal pay-offs. Because different individual players interiorise the same patterns individual beliefs and rules become shared beliefs and rules. The game is a Prisoner’s dilemma, but free-riding is discouraged because the game is (infinitely) repetitive. Cheaters may win one round, but are punished the next, making rule compliance is a more effective strategy than cheating, which feeds back into the game reinforcing the patterns perceived by the players. Change comes only because external events affect the distribution of resources with which the game is played. This forces (some) players to change their strategy, which in turn creates new patterns.14 13  Greif’s work is the best-known example among historians; Greif, Institutions; Greif, “The Maghribi Traders”; more specifically (but less known/cited by historians) Greif, “Cultural Beliefs and the Organization of Society”; Greif, “Commitment, Coercion, and Markets: The Nature and Dynamics of Institutions Supporting Exchange”. 14  Note that the overall result may not be Pareto-optimal; individual and overall pay-offs might be higher if players could find a way to collectively change their strategies. Because social institutions emerge as Nash equilibria, however, no player has an apparent interest to change her strategy. As a result, the whole group (community) of players can become lockedin in a non-Pareto-optimal game. For a discussion see Greif, Institutions; Wegerich, “Institutional Change: A Theoretical Approach”; the notion underlies already Hayek’s thinking about cultural evolution, but Hayek did not recognise possible non-Pareto-optimal outcomes cf. Hayek, “Notes on the Evolution of Systems of Rules of Conduct”.

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

29

There are several problems, however, with this approach. The repeated game effect against free-riders has empirically been shown to work only with small numbers of players (usually two), endowed with perfect monitoring capacity, under predictable “laboratory” conditions, and for games that are not nested in super-games. Game theoretical models that endogenously produce cooperative strategies, with institutions as Nash equilibria providing incentives to co-operate, also have been shown to work well only under laboratory conditions. Such conditions rarely if ever occur in real life.15 The model assumes, moreover, that players who have the same position in the game (for instance, “only labour, no capital”) have the same unchanging perceptions of reality and the same unchanging preferences (“if I prefer to hire myself out today rather than till rented land, I will do so tomorrow as well”). This, again, is an unrealistic assumption. The predicted Nash equilibria may be “real” given the rules of the game and the initial conditions, but the odds of the players “finding” and then reproducing them are infinitesimal. Social life is too complex and information too scarce and imperfect for rational agents (assuming for the sake of the argument that this is the dominant behavioural mode in homo sapiens) endowed with merely human cognition first to detect and then consistently to play out the predicted Nash equilibrium rules. Bowles and Gintis for that reason call them “evolutionary irrelevant Nash equilibria”: The problem with achieving a Nash equilibrium is that individuals may have heterogeneous and incompatible beliefs concerning how other players will behave, and indeed what other players believe concerning one’s own behavior. Therefore, individuals may choose best responses to strategies that the other players in fact are not playing, resulting in game play that is far from any Nash equilibrium.16

Social institutions, therefore, cannot be explained as emerging from the patterns produced by the interactions of self-interested agents. On the contrary, cooperative equilibria in games require “correlating devices” that instruct the players on their best strategy and punish players that defect. In game-modelling social reality, social norms can be conceptualised as correlating devices that help to maintain economic/social/political stability. These, however, are not (re)produced because they are the 15  Hechter, “The Emergence of Co-Operative Social Institutions”; Vromen, Economic Evolution, 171–81. 16  Bowles and Gintis, A Cooperative Species, 79–92 (88 for the citation).

30 

K. VERBOVEN

outcome of a strategy; they are culturally inherited from the past as collective social preferences.17 Social institutions are prescriptive cultural beliefs underpinned by cultural values (social, moral, religious, etc.) and beliefs concerning the nature of reality. Rule compliance or defection (depending on the perceived reality) are themselves cultural values. Enforcement of social norms, as well, is guided by value judgments. In punishing cheaters and free-riders instrumental considerations are secondary to the innate human propensity towards “altruistic punishment”—costly punitive action by agents who have no apparent material interest in punishing cheaters, but do so because they feel morally compelled to defend values in which they believe.18 Core cultural beliefs are highly resilient to change, but they are not static. Norm deviation and transgression by individuals or collective agents, whether because of errors of judgment or deliberate cheating, are inevitably part of social realities.19 Based on personal experiences agents develop individual beliefs that can reinforce but also contradict and weaken cultural beliefs. The problem is complicated further because correlated equilibria require signals that allow players to choose their best strategy; but signals can be misunderstood, distorted, or consciously manipulated. So the robustness of any presumed cooperative equilibrium that has been made possible on the basis of a social norm set is vulnerable to incomplete and unreliable information, giving room to cheaters and free-riders. The non-determinism of social institutions, however, is not merely a side effect of messy individual agency. Social preferences and values are interrelated and many of them are hierarchically ranked. Honesty for instance, may be greatly valued but so is life and the former is usually weaker than the latter. This ranking differs not only between individuals, but is also (sub-)culture and context-dependent. Cultural beliefs always need to be interpreted to fit specific situations; the same situation may be interpreted differently depending on the (sub-)cultural beliefs and habits, as well as on the individual experiences, personalities, and tastes of agents. The results feed back into the perceptions of the agents. Cultural beliefs

 See also in this sense Hodgson, “The Evolution of Institutions”, 114.  Bowles and Gintis, A Cooperative Species; see also Verboven, “Like Bait on a Hook. Ethics, Etics and Emics of Gift-Exchange in the Roman World”; Verboven, “The Knights Who Say NIE”. 19  Bowles and Gintis, A Cooperative Species, 89–91. Cf. Beja, “Imperfect Equilibrium”. 17 18

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

31

are inherited, transmitted, and interpreted, but they undergo changes in the process as historical contingency claims its due. Institutional change is mostly an incremental process, although sometimes—as in the case of violent military occupations, revolutions or cataclysmic events—it can be spectacularly sudden and disruptive. This incremental process is strongly subject to path dependence.20 This clearly narrows the scope for institutional change and innovation. Path dependence is reinforced because organisations that benefit from an existing institutional set-up resist changes. The main problem of institutional change, however, and another major cause of path dependence, is that institutions come in packages.21 This is true of nearly all institutions—formal and informal. Roman legal regulations concerning agency, for instance, were intimately connected with the use of slave and freedman managers and with the structure of households in which the head of the house—the pater familias—automatically acquired everything that any of his children or slaves acquired. The “adiectician” actions which regulated the contractual liability of masters and patres were designed for slaves and children in potestate. Some of them were later extended to free persons but they were never replaced by a single comprehensive system regulating agency. The old basic principle of indirect representation remained in force: an independent (sui iuris) person could not legally bind or be bound by the actions of third persons. As I have argued elsewhere the survival of this basic principle reflected the dominant type of business organisation: familiae (slave/freedman workers and agents) surrounded by trust networks based on amicitia and patronage and legally supported by mandatum, societas, and negotiorum gestio.22 Institutional bundles, however, are not limited to rules that govern material interests. Social institutions are enmeshed in social rule systems in which they are linked to rule sets that play a role as signalling devices to express social status, identity and integration: do not pick your nose (in public), dress to the occasion, participate in religious events (and make sure you are seen to do so), bathe and make sure you do not smell (or the opposite if you are a medieval monk), and so on. These rules do not appear directly to affect material interests, but they play an important role in 20  The “constraints on the choice set in the present that are derived from historical experiences of the past” (North, Understanding the Process of Economic Change, 52). 21  Ogilvie, “Whatever Is, Is Right?”, 668; Verboven, “The Knights Who Say NIE”, 37. 22  Verboven, The Economy of Friends.

32 

K. VERBOVEN

structuring social life because they signal who are the ins and the outs, the upper-, middle- and lower-class, whom to trust, whom to follow.23 Social rule systems are usually based on immaterial moral or religious values and world views that profoundly affect the regulation of resource distributions. Gender biases for instance are a major handicap for economic development in many parts of the world even today. Gender bias also affected Roman women, but to a much smaller extent than Greek women. Roman law gave women a much greater control over their property and their labour. Did that favourably affect production? The entanglement of economic institutions—such as property rights or contract enforcement institutions—with social rules props up the robustness of a particular rule system and increases the resilience of institutions against change. To understand institutional change, therefore, we need ways to analyse the effects of social rule entanglement and agency on a level of complexity that exceeds the possibilities of game theory. System theory provides a promising approach to this problem.

3   Social System Theory Systems are defined by their structure, their behaviour, and their interconnectivity. The behaviour and structure of a system is governed by its rule system, that is the structured set of all rules that apply to that system. No rule operates in isolation from other rules. Systems are further defined also by their openness. Open systems exchange matter and energy, closed systems only energy and isolated systems neither matter or energy. Social systems are made up of the interconnections between persons; they are structured, they show systemic behaviour, and are governed by social rule systems, of which institutions are a part. “Matter” are people, “energy” are their interactions. Open social systems allow persons freely to migrate and interact with persons in other systems, closed systems allow cross-­ border exchanges (for instance trade contacts, but also military conflicts) but no migration, isolated systems allow neither.24 23  The distinction between informal institutions as governing interests and social rules more generally is not consistently made in neo-institutionalist literature; both are social norms; I am here following the example of Nee and Swedberg, “Economic Sociology and New Institutional Economics”. 24  The literature on social system theory is huge. Useful starting points are Burns and Flam, The Shaping of Social Organization; Burns and Dietz, “Cultural Evolution: Social Rule Systems, Selection, and Human Agency”; Machado and Burns, “Complex Social

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

33

The world today forms an isolated social super-system (our “global village”), with myriads of nested systems at different levels, with various degrees of closure, but none of which are isolated. At village and town levels social systems are mostly open. National states reduce the openness of their national social system, but with the exception perhaps of North Korea, some migration is allowed. Inside the European union national social systems are open. This, however, is not a modern phenomenon. Throughout history small(er) social systems (families, clans, villages) cluster into larger systems (tribes, poleis, nations, empires) with various degrees of closure and various levels of exchange. Each region, each town, or civitas in the Roman world formed an open system, nested in the Imperial super-system (see also Preiser-Kapeller in this volume on the concept of nesting). The empire’s frontiers were never hermetically closed. Some migration— forced or voluntary—always occurred, but not to a significant extent, that is not to the extent that it changed the system’s behaviour. For all intents and purposes the empire was a closed system. It was not, however, an isolated system. Via the eastern trade, the empire was connected even to other super-systems, such as Han China or India. Complex systems are made up of components that differ from each other. As these components interact the system displays emergent properties—behaviour that cannot be reduced to the system’s constituent parts. When the components of a complex system are endowed with agency and react to the system’s behaviour, we get complex adaptive systems. In these systems, the constituent parts adapt in response to changes in the system and thereby change the behaviour of the system. All social systems are complex adaptive systems, but there are degrees of complexity. A settlement consisting of only peasant households has complexity in so far as households differ, for instance in terms of size, gender composition, age of members, and so on, but this society is less complex than a town with households practising different professions, having different levels of income, wealth, slaves, administrations, and so on. Smaller (complex) systems can be nested in larger super-systems that define the environment in which a sub-system exists and with which it interacts. For instance a low complexity peasant village (composed of similar households) may be part of a larger more complex society, with Organization”; Burns and Machado, “Social Rule System Theory. Universal Interaction Grammars”.

34 

K. VERBOVEN

harbour towns, agro-towns, political capitals, military settlements, and so on. Social systems are governed by social rule systems.25 Social rules constrain the agency of individuals and organisations in a social system. These constraints enable co-ordination of action between individual agents. This is because social rules materialise only by being performed by agents who interpret them to fit specific situational contexts.26 The loopback effect is that agents who share rule sets recognise which rules might apply in a given situation. Even though they cannot be sure how other agents will perform these rules, they can anticipate the constraints on other agents to comply to the rules in question. The specific situation may allow also to anticipate the risk of rule defection, for instance because it is in the self-­ interest of the other agent(s) to defect and enforcement is not possible. All social rule systems are (sub-)culture specific. However, even within the same (sub-)cultural system, social rules are played out differently according to the specific role (itself a set of rules) an agent is playing in a specific situation. Social roles themselves differ according to gender, age, status, nationality, and other variables that make up a person’s social identity. The “playing out” of social rules according to social roles, implies that constraints are never absolute; they depend on the agent’s perception of the situation in which she is playing and the role she is playing in it.27 Rule compliance, however is more than a question of recognising roles and interpreting situations. An agent can choose openly or in secret to ignore, resist or deviate from rules that she should play out in a given situation given her role in it. She can do so either because of personal interests (the “me-myself-and-I-rule”)28 or because she perceives rule compliance as damaging to values or persons she holds dear. The mirror image, of course, is that agents can choose to comply to rules even when they perceive this as against their personal interest because of overriding values. More often than not situations and roles invoke different rule sets that may come into conflict. A person, for instance, can be cast in the role of a 25  Burns and Dietz, “Cultural Evolution: Social Rule Systems, Selection, and Human Agency”. 26  “Action situation” in the IAD framework, cf. McGinnis and Ostrom, “Social-Ecological System Framework”, 29. 27  Cf. the concept “action situation” and the “Institutional Analysis and Development” (IAD) framework McGinnis and Ostrom, “Social-Ecological System Framework”. 28  On “strong reciprocity” and “true altruism”, see Verboven, “Like Bait on a Hook. Ethics, Etics and Emics of Gift-Exchange in the Roman World”.

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

35

soldier, which requires him to obey orders. But on active duty he may find himself in situations where the rules that apply to his role as a soldier, come into conflict with the rules that follow from his other cultural beliefs, for instance not to kill civilians. Social institutions and the rule sets to which they are connected are subject to change because they need to be played out in specific situations to take effect. Societies differ in the degree of autonomy that agents have to take up roles (self-casting) and interpret them. Open societies (in Popper’s sense of the term) allow agents a large freedom of choice, constrained only for instance by the resources they control. Closed societies place constraints on agents’ autonomy of role choice. No society is fully closed, none is fully open; degrees of autonomy may differ depending on age, gender, lineage, and so on. The more open a society is, the more room it has for social rule changes. Social rule theory shows us how institutional resilience or change is not just determined by the centrality of an institution to the rule system of which it is a part, but rather by how and which social roles are distributed over a population and in which situations agents play out their role. For instance, when an authorised agent imposes a newly designed (formal) rule, enforced by designated agents (police, army, secret service, law courts, etc.), this changes the situational contexts in which other persons and organisations play out their social rules according to the social roles they are cast in. If the new formal rule complements an existing social rule (set),29 the risk of rule defection diminishes because the cheater now risks formal sanctions in addition to the emotional (shame, guilt, etc.) and social sanctions already in place and is more likely to be found out (by the police/ secret service/etc.). If, on the other hand, a newly designed rule contradicts an existing social rule (set) different things may happen. Persons and organisations may adapt by interiorising the new rule, that is by creating an informal rule that matches the newly designed formal rule. They can do so for instance because that is in their best interest or because they genuinely believe that the new rule is an improvement. If a sufficient number of agents do so and start playing out the new rule their interactions lead to the emergence of a new social rule that matches the new designed rule. The process is based on the agents’ intersubjective experiences (others 29  Cf. Aoki, Toward a Comparative Institutional Analysis, 10 for the view that only social institutions are real institutions.

36 

K. VERBOVEN

play by the rule too) reinforced through symbolic interaction (mainly but not exclusively linguistic).30 Alternatively, however, agents can reject the new rule and adapt their behaviour to dodge it or actively sabotage it in order to protect the social rule set they already have and which they prefer. This greatly raises the cost of enforcement for the new rule up to the point that it may fail. The risk of defection from the old rule set, however, is also likely to increase as some agents will presumably choose to stop playing by the old rules—whether out of self-interest or because they genuinely endorse the new rule. The extent of this risk depends on the social sanctions enforcing the old rule set and the costs attached to imposing these social sanctions (gossip is cheap, lynching potentially costly). Rule systems change also in response to changing environments that affect the outcome of rule compliance or defection in specific situations. Outcomes matter for any purposeful action. So agents constantly reinterpret and rearrange rule sets to apply to situations they encounter. Rule sets not only change through this process; their relative importance to alternative rule sets also changes. The new social rules that emerge from this process, or their changed position vis-à-vis other rules, can then push for changes to existing designed rules or the creation of new ones. Outcomes affect also role casting of agents. Imagine an iron age Gallic chief with retainers tied to him in a gift-exchange relation—giving them protection and allowing them to work the fields he owns or controls in exchange for part of the harvest and military assistance. Under Roman dominion military assistance becomes irrelevant, but money becomes an important asset to maintain social position. Share-cropping retainers can then be recast into the role of rent paying tenants, chiefs into that of large landowners and possibly suppliers to the Roman army. Both processes—rule set modification and role casting modification— are subject to “natural” selection. Agents that successfully play out a modified rule set and successfully assume a new role outcompete agents that fail to do so. Some gain power, status, wealth, and authority, all of which increase the symbolic capital attached to their new role. Others, through accommodation and submission to their new role and the rule sets attached to it, may merely succeed in surviving, while those who reject the rule and role are killed or starve to death.

30  On the impact of formal institutions on culture, see Alesina and Giuliano, “Culture and Institutions”, 921–28.

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

37

In this view social institutional change on the systemic level emerges from adaptations on the level of the system’s constituent parts, that is from social agents who may choose to ignore, reinterpret or contradict existing rules or formulate new rules to cope with unexpected situations, or who may on the contrary re-emphasise and re-enforce existing rules which they feel express important values that are under threat in the new situation. Such self-organisation is an essential characteristic of agents’ behaviour in complex adaptive systems. Agents choose and deploy strategies in response to the reality they perceive. As all psychological and social processes, this process is not pre-determined but neither is it random. Because agents differ from each other structurally in terms of the social roles ascribed to them, the resources they control, and the situations they find themselves in, based on their ascribed role in society, there is a statistical probability31 that agents with similar social roles, in similar situations governed by similar rule sets, make similar organisational choices. This probability increases if the agents also exchange information regarding their choices and experiences. The same principle is operative when agents design organisations for any functional purpose—social, religious (a cult association), economic (firms), political (parties), or mixed (a guild combining social, religious, and economic functions). The results are self-­ similar structures within a social system, which (since the system is a complex adaptive system) feed-back into the agents’ adaptive behaviour, which then again influences their self-organisation and so on.32 Throughout the social system self-similarity allows agents in functionally different positions, to share common experiences, because the situations they experience are produced, controlled or influenced under structural constraints with similar features, for instance that of social hierarchy or valuation in terms of money. The same is true, of course, for the processes by which social rules, including institutions, are regenerated, reinterpreted, modified, or created. All of these have a high statistical probability of showing similar features if they are generated by agents acting out similar roles under similar structural constraints. 31  Only statistical: agents also differ on the basis of non-structural features, such as their intelligence, personality, the random situations they encounter, or unpredictable (stochastic) events such as diseases or accidents. 32  On self-similarity Abbott, Chaos of Disciplines, 157–231 (who distinguishes various mechanisms that produce self-similarity in social structure: design, fission, fractionation, ossification).

38 

K. VERBOVEN

Self-similarity is a typical feature of complex adaptive systems and very likely the main reason why a social system does not fall apart. Not only are agents who share similar experiences more likely to show solidarity, but self-similar rule sets facilitate cooperation between agents and their organisations.33 Thus, complex adaptive systems are engines of non-disruptive endogenous rule change.

4   Impact of Empire We now have some of the basic building blocks to understand institutional change in successful empires, such as the Roman. Sheer military power financed by tribute extraction (imperial elites) and elite collaboration motivated by rent extraction (patrimonial elites) certainly played a part in keeping the empire together. While these may explain the endurance of the empire, however, they cannot explain its economic success. The following pages focus mainly on Gaul and Britain but they document patterns of change which I suggest can be found also elsewhere. 4.1  Changing Designed Institutions The first wave of changes pertained only to formal institutions. The conquered nations and tribes had to pay tribute in kind and/or in money; some had to provide recruits for the auxiliary forces or labour for clearing land or other army projects. Legal disputes that involved Roman citizens or parties from different polities had to be brought before the governor. The empire, however, lacked the resources to enforce more than a bare minimum of newly imposed institutions. The resources that were available were needed for maintaining order, protecting Roman citizens (mainly negotiatores on whom the army depended for supplies) and allies, suppressing local resistance, and levying taxes. The legitimacy of Roman enforcement institutions in the eyes of the indigenous populations, moreover, remained dubious. Local enforcement structures therefore had to be left intact. We know very little about what these looked like, but we do know that the execution of formal enforcement was left in the hands of local elites. How many of the elite families in the Augustan era—when Gaul was definitively settled—were new is impossible to make out, but the authority of local elites was still largely based on tribal traditions.  Abbott, 183–85.

33

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

39

Local resistance never disappeared. While most of it remained limited to brigandage and anachoresis”, in some cases it led to major revolts. Tacitus’ comments on the revolt led by Florus and Sacrovir among the Treveri and Haedui under Tiberius in 21 CE illustrate the hatred against Roman negotiatores and tax collectors; their “continuation of tribute payments, the heavy burden of usury, the cruelty and arrogance of governors”.34 4.2  Changing Social Rule Sets The archaeological, epigraphic, and literary record, however, indicates that social rules complexes were gradually shifting. The best proxy we have for these changes and their economic effects are coins. Iron age coinage in Gaul long remained primarily a social currency, used to “to create, maintain, or sever relations between people rather than to purchase things” by serving as an object in gift-exchange or ritual practices.35 From the middle of the second century BCE onwards, however, small silver coins (dubbed quinarii by numismatists for their resemblance to Roman quinarii) and cast coins in potin36 became increasingly common among tribes that also developed oppida—proto-urban centres which besides having defensive purposes served also as centres for trade and crafts. After Caesar’s conquest small struck bronze coinage replaced the older potin coins. Presumably in the oppida the quinarii and the low-value potin and small bronze coins were already used in transactions that were to some extent disembedded from social relations between seller and buyer. This would reflect an early evolution of transactions towards market exchange, although it is impossible to say whether and to what extent these markets were supported by formal institutions. Strikingly, the silver quinarii seem to have been strongly personalised currencies. Many of them bear the personal names of powerful chiefs who ordered their production and presumably distributed them among their followers. During Caesar’s Gallic wars (and after) the scale on which quinarii were minted greatly increased as they were used to pay or reward allies on both sides. The intended original function of the potin coinages remains a mystery. They may have 34  Tacitus, Ann. 3,40: de continuatione tributorum, gravitate faenoris, saevitia ac superbia praesidentium. 35  Graeber, Debt, 158. 36  A high tin-content bronze alloy with a silvery appearance.

40 

K. VERBOVEN

served ritual and symbolic purposes at first, but that does not exclude them acquiring monetary functions in low value transactions afterwards. The gold coinages appear to reflect general tribal identities in their iconography, although the themes and topics they depict are hardly distinctive: warriors, horses, sun-wheels, “fantastic beasts”, and so on. Their volatile weight and purity, however, suggest that minting authorities were fragmented also in the case of gold coinage. In addition the intrinsic value of gold coins was too high to serve in day-to-day transactions. Most likely, therefore, gold coinages never ceased being primarily social currencies. The social disembedding of silver and potin/bronze coinages remained a phenomenon linked to the oppida. On the whole the numismatic evidence correlates well with Caesar’s description of the role played by powerful nobles and warrior-chiefs in Gallic society.37 Desocialised, impersonal monetary exchanges became more common from around 15 BCE onwards with the massive concentration of military units along the Rhine. These entailed payments to personnel (mainly soldiers) and taxation but also market exchanges—first to army suppliers, secondarily to local craftsmen and elites.38 Clearly, indigenous communities did not change overnight. The process can be followed over a century in Belgica and Germania Inferior. Under the Julio-Claudian emperors coins appear mainly in military contexts or in contexts derived from these, such as depositions.39 From the Flavians onwards, however, the spread of Roman coins of all denominations becomes much more homogenised, indicating that the rule “sell for money/buy for money” had become immersed and that money had become a social institution to regulate impersonal exchanges for which coins (but potentially also bank and private accounts and IOUs) served as instruments.40 37  Wigg-Wolf, “The Function of the Last Celtic Coinages in Northern Gaul”; Delestrée, “Les pouvoirs émetteurs en Gaule, des origines à l’époque augustéenne”; Creighton, Coins and Power in Late Iron Age Britain, 14–15; Roymans, Ethnic Identity and Imperial Power the Batavians in the Early Roman Empire, 11–12 and passim; Roymans and Aarts, “Coin Use in a Dynamic Frontier Region. Late Iron Age Coinages in the Lower Rhine Area”; for a more “commerce-oriented” interpretation on Icenian coinage now see Talbot, Made for Trade. Aarts, “Monetization and Army Recruitment in the Dutch River Area”. 38  Aarts, “Monetization and Army Recruitment in the Dutch River Area”. 39  Roymans and Aarts, “Coin Use in a Dynamic Frontier Region. Late Iron Age Coinages in the Lower Rhine Area”. 40  Verboven, “Currency, Bullion and Accounts. Monetary Modes in the Roman World”.

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

41

As material witnesses of monetised transactions, coins are merely the tip of the iceberg. The Vindolanda Tablets provide a further glimpse of how monetisation rolled out on the frontier of Roman and indigenous interactions. An older example is the famous tablet of the so-called sale of a Frisian ox (emptio bovis frisicae) from Tolsum—in fact an acknowledgement of debt drawn up in the context of auxiliary troops in 29 CE. The money was lent by a woman called Iulia Secunda through a slave. The name is very common among Romanised Gauls, but in view of the early date of the tablet, Iulia Secunda probably belonged to a Romanised aristocratic family that received Roman citizenship from Caesar or Augustus.41 Although slightly different from the Pompeian loan and debt tablets, the Tolsum tablet testifies to how Roman legal practises seeped in. It implicitly confirms the legitimacy of Roman enforcement procedures in the eyes of some (but certainly not all) locals. The use of a slave agent, as well, is typically Roman and relies on the Roman law of slavery. As we saw above, institutions are embedded in social rule complexes that are not limited to material interests but regulate signalling devices expressing social identities, roles, and relations. Tacitus’ description of Agricola’s treatment of British elites nicely illustrates shifts in such practices: In order that a population scattered and uncivilised, and proportionately ready for war, might be habituated by comfort to peace and quiet, he would exhort individuals, assist communities, to erect temples, market places, houses: he praised the energetic, rebuked the indolent, and the rivalry for his compliments took the place of coercion. Moreover he began to train the sons of the chieftains in a liberal education, and to give a preference to the native talents of the Briton as against the trained abilities of the Gaul. As a result, the nation which used to reject the Latin language began to aspire to rhetoric: further, the wearing of our dress became a distinction, and the toga came into fashion, and little by little the Britons went astray into alluring vices: to the promenade, the bath, the well-appointed dinner table. The simple natives gave the name of “culture” to this factor of their slavery.42 41   Bowman, Tomlin, and Worp, “Emptio Bovis Frisica: The “Frisian Ox Sale” Reconsidered”. 42  Tac., Agr. 21: Sequens hiems saluberrimis consiliis absumpta. Namque ut homines dispersi ac rudes eoque in bella faciles quieti et otio per voluptates adsuescerent, hortari privatim, adiuvare publice, ut templa fora domos extruerent, laudando promptos, castigando segnis: ita honoris aemulatio pro necessitate erat. Iam vero principum filios liberalibus artibus erudire, et ingenia Britannorum studiis Gallorum anteferre, ut qui modo linguam Romanam abnue-

42 

K. VERBOVEN

Cultural identities tend to be very persistent but are not immune to change. Roymans, for instance, showed how Batavians still cherished their cultural identity early in the third century CE. By the early second century, however, this tribal identity had become fundamentally different from what it had been under Augustus. Over the course of the second century the “Roman” municipium Batavorum identity was added on to old tribal identities until eventually the municipium identity was the only local identity that remained.43 In the east, Greek civic traditions, broadly similar to Roman traditions, had little difficulty finding a modus vivendi with Roman social rule complexes. Late Iron Age society in Gaul was more different but was relatively open to change because it gave considerable scope for individual leaders to form warrior bands (comitatus) held together by gift-exchange relations. The comitatus potentially transcended local communities and were easily integrated as auxiliaries in the Roman army.44 Judaean society, by contrast, was much more strict in how agents were role casted and how roles should be played out. Jewish cultural and political leaders willing to support the cause of Rome, had difficulty controlling their people. Popular resistance rooted in strong cultural traditions and upheld by religious convictions produced its own natural leaders who played out Jewish identity against Greek and Roman identities. This led to the Wars of 66–73  CE and 132–135 CE (the Bar Kochba revolt). Similar tensions in the Jewish communities in the former Ptolemaean territories (Egypt, Cyrenaica, and Cyprus), whose traditional autonomy had been confirmed by Caesar and Augustus, led to the Kitos war (115–117 CE). The challenge of securing Roman control was considerable. Roads, supply lines, storage facilities, river port installations, and so on needed to be built and maintained in operation. Both the installations and the associated personnel needed to be protected. In addition, the property rights of veterans and (important) allies had to be guaranteed. Traces of Roman centuriation in the landscape show that Roman authorities took an interest in the matter. But restructuration of the landscape went beyond securing military logistics. Tacitus’ description of Agricola’s policy in Britain bant, eloquentiam concupiscerent. Inde etiam habitus nostri honor et frequens toga; paulatimque discessum ad delenimenta vitiorum, porticus et balinea et conviviorum elegantiam. Idque apud imperitos humanitas vocabatur, cum pars servitutis esset. (translation M. Hutton (Loeb)). 43  Roymans, Ethnic Identity and Imperial Power the Batavians in the Early Roman Empire. 44  Roymans, 22.

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

43

draws attention to the restructuration of indigenous social environments. Agricola is said to have persuaded British nobles to compete for status through building town houses, temples and fora, that is through evergetism and residential conspicuous consumption (see above). Cassius Dio alludes to a similar policy in Germania Magna before the Teutoburger disaster in 9 CE, whereby (what he calls) poleis were organised and German tribes began to “set up markets and peaceful assemblies”.45 Structural changes affect the situational context in which roles are played out, they create new social roles and modify existing ones. Along the Rhine frontier a dense concentration of troops and military settlements formed a strongly roman environment to which indigenous auxiliary were attached. In the hinterland smaller castella served as nuclei of Roman presence. Local tribes were reorganised into civitates. Each received a caput civitatis (often a place of some importance already in pre-­ roman times) that served logistic purposes. The capita civitatum soon became important central places for elite competition and status affirmation. Nevertheless, traditional social structures remained intact and in most cases old tribal identities and traditions were not abandoned. Gallic and Briton nobles traditionally had clientes whom they could mobilise to display their status position, but the new rules for status affirmation required different relations. In many cases the land on which religious and civic monuments were built was public. So negotiation with (local) public authorities was required before a building project could commence. Labour from retainers (clientes) may have been used to some extent, but the skills required called for specialists. Imported products, such as wine, had long played a key part of status related consumption, but the chiefs could now no longer depend on political power either to mobilise the required resources to buy these products (for instance by exchanging slaves for wine), or to demand that merchants gave them part of their wares as tribute. Money was now a necessity and could only be acquired by either serving in the Roman auxilia or supplying the Roman army directly or through Roman negotiatores. This municipalisation and monetisation process gradually changed the formal institutions governing status positions and formal leadership. Around the middle of the first century (but possible decades earlier) the Batavian civitas was presided by a summus magistratus. Roymans no doubt 45  Cass. Dio LVI,18,2: πόλεις συνῳκίζοντο … καὶ ἀγορὰς ἐνόμιζον συνόδους τε εἰρηνικὰς ἐποιοῦντο.

44 

K. VERBOVEN

correctly interprets this office as the adaptation and latinisation of a traditional tribal office.46 The process continued until the civitas of the Batavians received the status of municipium (latinum), probably in the early second century. This implied a Roman style local senate and elected offices. Former magistrates and council members received Roman citizenship. No doubt by this time the principles of Roman contract law and enforcement through judicial magistrates were common. We should not romanticise these changes. They went in tandem with merciless oppression. Property rights were violated and redefined along with traditional rules of allegiance and social support. New obligations were imposed that had to be fulfilled in money. Dues received in money could only be spent in faraway towns. Goods were requisitioned in return for promises of payment of doubtful validity. Negotiatores exercised alien contractual rights that escaped local institutions because the negotiatores were mostly Roman citizens or closely linked to provincial and military authorities. And so forth. A substantial part of the indigenous population undoubtedly became marginalised or was simply murdered outright. Yet at the same time potential leaders were coaxed into a new way of life that assured their social position. In order to succeed in maintaining this position, however, elite families had to change their strategies. They had to adapt to the new realities. Roman authorities were wise enough to create possibilities via the auxiliary units and via the Council of the Gauls. The auxiliary units provided opportunities for social mobility to non-elite local populations and positions of power and prestige to local elites. But only the highest members of each civitas were admitted to the Consilium Galliarum. The restructuration of social systems under Roman rule shows self-­ similar features emerging throughout social organisations. Not all of these were unique to Roman culture, but they were all compatible with it and they served as the structuring principles of an increasingly complex social order. Money became the dominant unit of value for material goods and obligations and the core institution regulating transactions.47 Skills in various domains (military, bureaucratic, crafts, etc.) supported an extensive 46  CIL XIII, 8771; cf. Roymans, Ethnic Identity and Imperial Power. The Batavians in the Early Roman Empire, 64, 200–202. 47  See Verboven, “Currency, Bullion and Accounts. Monetary Modes in the Roman World”. For this phenomenon of “deep monetisation”, see Lucassen, “Deep Monetisation”. For a similar development in the late medieval and early modern Netherlands.

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

45

division of labour. Public generosity (evergetism) replaced personalised gift exchange as the basis for acquiring legitimacy. Latin became the common language for public communication; Roman status criteria—such as citizenship, membership of the ordines (senatorius, equester, decurionum), slavery, freedmanship, freeborn status, military/veteran/civic status, and so on—came to define social positions and hierarchy. Local identities and differences remained important, but they were redefined in ways that mirrored Roman imperial realities. Restructuration of social and material environments affects situational contexts, which in turn affect how roles and rules are played out. The impact of Roman rule, however, did not remain limited to structural changes. Its most immediate and profound impact was on the connectivity of social systems, which in turn profoundly affected both structural and situational contexts as well. The connectivity structures of the pre-Roman Mediterranean that the Roman empire inherited had long been very dense. This was particularly (but not exclusively) true in the eastern Mediterranean, which had formed an interdependent economic trade system since at least the late fourth century BCE.48 The impact of the Roman empire was particularly pronounced on the inland connections in the territories bordering on the western Mediterranean. Existing “nodes” and “edges” were taken over from pre-Roman networks: oppida, religious centres, settlements, emporia and so forth, but their content was redefined to fit the restructured territories: oppida became capitals of civitates, and smaller settlements became relay posts to supply the frontier armies. New centres were added: castella, river ports, vici. Existing roads were enlarged and improved, new roads were built. The number of alternative routes was deliberately increased, their capacity enhanced, and connection speeds increased. The overall effect was that the intensity of exchanges also increased and connections became more resilient. In the early Roman period small-­ world effects at supra-local levels depended on a small number of elite families bridging indigenous networks to official Roman networks. The “betweenness centrality” of these families was high, which gave them power and influence, but the network’s overall robustness was low. Over time, thanks to the auxilia, but also to increased monetisation and the business it attracted, the number of bridges increased until eventually 48  For an impressive study of the role of connectivity of the Mediterranean see Horden and Purcell, The Corrupting Sea.

46 

K. VERBOVEN

indigenous and Roman official networks largely merged. Roman and non-­ roman negotiatores had travelled north before, but not only were they now more numerous, they had also different demands and expectations— backed up by Roman enforcement structures if necessary—to which locals who wished to do business had to adapt. The prime effect of denser and more robust networks is that information flows faster and becomes more reliable; hence the outcome of actions and the reactions of other agents can be anticipated more reliably. In addition, the increased density of networks combined with the restructuration of social environments and the patterning of self-similarity that emerged from it, changed the choice sets of non-Roman individuals and collective agents. This affected how agents played out their roles based on the social rule sets they had previously interiorised but also on the anticipation of the outcomes of their actions and the resulting reactions of other agents. “Playing-out-roles-and-rules”, however, was not random; it was premised on the position of agents within the social system and the structure and connectivity that characterised it. As we saw above non-­ random changes in the anticipation by agents spur social rule changes, including institutional changes, because agents perceive that effective action depends on aligning themselves with new intersubjective expectations. Interiorising these patterns of behaviour potentially results in the emergence of new social rules.

5   Conclusion Incongruities between culture-specific social norms, conventions, and beliefs regarding the nature of the world on the one hand and formal rules imposed by political elites and authorities on the other force the latter to spend valuable resources to maintain order and enforce formal rules. To understand economic development in the Roman empire, therefore, we need to understand how deep institutional change—affecting the beliefs and social habits of conquered peoples—took place. I have argued that social rule system theory and complexity economics provide insights into this process. Social institutions rely on cultural beliefs that are subject to change as they are inherited, transmitted, and interpreted to fit specific situations. The combined effects of restructured social and political environments and increased connectivity changed pre-Roman social rule sets and the institutions governing economic exchange embedded in them. Rome allowed local elites to continue exercising power while setting them

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

47

on a path that brought them ever closer to the Roman social order. The process was not always peaceful. Resistance was brutally suppressed. The history of the Jewish people shows it was not always successful. By the second century CE, however, local societies and culture in most provinces had accommodated themselves to the imperial super-system by creating self-similar social structures, such as civic institutions and monetised transactions, and dense and robust translocal social networks. Local informal institutions—social norms and conventions—now seamlessly supported empire-wide formal ones.

Bibliography Aarts, Joris G. “Monetization and Army Recruitment in the Dutch River Area in the Early 1st Century AD”. In Kontinuität Und Diskontinuität, Germania Inferior Am Beginn Und Am Ende Der Römische Herrschaft, edited by Th. Grünewald and S. Seibel, 162–80. Berlin; New York, 2003. Abbott, Andrew Delano. Chaos of Disciplines. Chicago: University of Chicago Press, 2001. Alesina, Alberto, and Paola Giuliano. “Culture and Institutions”. Journal of Economic Literature 53 (December 2015): 898–944. https://doi. org/10.1257/jel.53.4.898. Aoki, Masahiko. Toward a Comparative Institutional Analysis. Cambridge, Mass.: MIT Press, 2001. Beja, Avraham. “Imperfect Equilibrium”. Games and Economic Behavior 4, no. 1 (January 1992): 18–36. https://doi.org/10.1016/0899-8256(92)90003-B. Bowles, Samuel. Microeconomics: Behavior, Institutions, and Evolution. New York; Princeton, N.J.: Russell Sage; Princeton University Press, 2004. Bowles, Samuel, and Herbert Gintis. A Cooperative Species: Human Reciprocity and Its Evolution. Princeton: Princeton University Press, 2011. Bowman, Alan K., R. S. O. Tomlin, and Klaas A. Worp. “Emptio Bovis Frisica: The “Frisian Ox Sale” Reconsidered”. Journal of Roman Studies 99 (2009): 156–70. Burns, Tom R., and Thomas Dietz. “Cultural Evolution: Social Rule Systems, Selection, and Human Agency”. International Sociology 26 (1992): 250–83. Burns, Tom R, and Helena Flam. The Shaping of Social Organization: Social Rule System Theory with Applications. London: Sage, 1990. Burns, Tom R., and Nora Machado. “Social Rule System Theory. Universal Interaction Grammars”. CIES e-Working Paper N.o 175/2014, 2014. Creighton, John. Coins and Power in Late Iron Age Britain. Cambridge UK; New  York: Cambridge University Press, 2000. https://doi.org/10.1017/ CBO9780511489587.

48 

K. VERBOVEN

Delestrée, Louis-Pol. “Les pouvoirs émetteurs en Gaule, des origines à l’époque Augustéenne”. Revue Belge de Numismatique et de Sigillographie 163 (2017): 1–26. Dessi, Roberta, and Sheilagh Ogilvie. “The Political Economy of Merchant Guilds: Commitment or Collusion?”, March 2004. http://idei.fr/doc/wp/2004/ political_guilds.pdf. Eggertsen, Thráinn. “A Note on the Economics of Institutions”. In Empirical Studies in Institutional Change, edited by Lee J. Alston and Douglass C. North, 6–24. Cambridge: Cambridge University Press, 1996. Elster, Jon. “Social Norms and Economic Theory”. The Journal of Economic Perspectives 3 (1989): 99–117. Epstein, Stephan R. “Craft Guilds in the Pre-Modern Economy: A Discussion”. The Economic History Review 61, no. 1 (2008): 155–74. https://doi. org/10.1111/j.1468-0289.2007.00411.x. Gelderblom, Oscar. Cities of Commerce: The Institutional Foundations of International Trade in the Low Countries, 1250–1650. Princeton University Press, 2013. Graeber, David. Debt: The First 5,000 Years. Brooklyn (NY): Melville House, 2011. Greif, Avner. “Commitment, Coercion, and Markets: The Nature and Dynamics of Institutions Supporting Exchange”. In Handbook of New Institutional Economics, edited by Claude Ménard and Mary M Shirley, 727–86. Berlin: Springer, 2008. Greif, Avner. “Cultural Beliefs and the Organization of Society: A Historical and Theoretical Reflection on Collectivist and Individualist Societies”. Journal of Political Economy 102, no. 5 (1 October 1994): 912–50. Greif, Avner. Institutions and the Path to the Modern Economy: Lessons from Medieval Trade. Cambridge; New York: Cambridge University Press, 2006. Greif, Avner. “The Maghribi Traders: A Reappraisal?” The Economic History Review 65, no. 2 (2012): 445–469. https://doi.org/10.1111/j.14680289.2012.00645.x. Haldon, John, Hugh Elton, Sabine R. Huebner, Adam Izdebski, Lee Mordechai, and Timothy P.  Newfield. “Plagues, Climate Change, and the End of an Empire: A Response to Kyle Harper’s The Fate of Rome (1): Climate”. History Compass 16, no. 12 (2018): e12508. https://doi.org/10.1111/hic3.12508. Harper, Kyle. The Fate of Rome: Climate, Disease, and the End of an Empire. Princeton: Princeton University Press, 2017. Hayek, Friedrich A. von. “Notes on the Evolution of Systems of Rules of Conduct”. In Studies in Philosophy, Politics and Economics, 66–81. London: Routledge & Kegan Paul, 1978. Hayek, Friedrich A. von. The Fatal Conceit: The Errors of Socialism. Chicago: University of Chicago Press, 1989.

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

49

Hechter, Michael. “The Emergence of Co-Operative Social Institutions”. In Social Institutions: Their Emergence, Maintenance and Effects, edited by Michael Hechter, Karl Dieter Opp, and Reinhard Wippler, 13–34. New  York: de Gruyter, 1990. Hodgson, Geoffrey M. “The Evolution of Institutions: An Agenda for Future Theoretical Research”. Constitutional Political Economy 13, no. 2 (June 2002): 111–27. Horden, Peregrine, and Nicholas Purcell. The Corrupting Sea: A Study of Mediterranean History. Oxford; Malden, Mass: Blackwell, 2000. Johnson, Noel D., and Mark Koyama. “States and Economic Growth: Capacity and Constraints”. Explorations in Economic History 64 (1 April 2017): 1–20. https://doi.org/10.1016/j.eeh.2016.11.002. Jolowicz, H.  F., and Barry Nicholas. A Historical Introduction to the Study of Roman Law. Cambridge: Cambridge University Press, 1972. Lucassen, Jan. “Deep Monetisation”. Tijdschrift voor Sociale en Economische Geschiedenis 11 (2014): 73–121. Machado, Nora, and Tom R.  Burns. “Complex Social Organization: Multiple Organizing Modes, Structural Incongruence, and Mechanisms of Integration”. Public Administration 76 (1998): 355–386. McGinnis, Michael, and Elinor Ostrom. “Social-Ecological System Framework: Initial Changes and Continuing Challenges”. Ecology and Society 19 (20 May 2014). https://doi.org/10.5751/ES-06387-190230. Meyer, Elizabeth A. Legitimacy and Law in the Roman World. Tabulae in Roman Belief and Practice. Cambridge, UK; New  York: Cambridge University Press, 2004. Milanović, Branko. Global Inequality: A New Approach for the Age of Globalization, 2016. Milanovic, Branko, Peter H. Lindert, and Jeffrey G. Williamson. “Pre-Industrial Inequality*”. The Economic Journal 121 (2011): 255–272. https://doi. org/10.1111/j.1468-0297.2010.02403.x. Nee, Victor. “Norms and Networks in Economic and Organizational Performance”. The American Economic Review 88, no. 2 (1998): 85–89. Nee, Victor, and Richard Swedberg. “Economic Sociology and New Institutional Economics”. In Handbook of New Institutional Economics, edited by Claude Ménard and Mary M. Shirley. Berlin: Springer, 2008. North, Douglass C. “Some Fundamental Puzzles in Economic History/ Development”. In The Economy as an Evolving Complex System II, edited by W.  Brian Arthur, Steven N Durlauf, and David A Lane, 223–37. Reading: Addison-Wesley, Advanced Book Program, 1997. North, Douglass C. Understanding the Process of Economic Change. Princeton, N.J.: Princeton University Press, 2005.

50 

K. VERBOVEN

Ogilvie, Sheilagh. “Guilds, Efficiency, and Social Capital: Evidence from German Proto-Industry”. The Economic History Review 57 (2004): 286–333. https:// doi.org/10.1111/j.1468-0289.2004.00279.x. Ogilvie, Sheilagh. Institutions and European Trade: Merchant Guilds, 1000–1800. Cambridge; New York: Cambridge University Press, 2011. Ogilvie, Sheilagh. “Rehabilitating the Guilds: A Reply”. The Economic History Review 61 (February 2008): 175–82. https://doi.org/10.1111/j.14680289.2007.00417.x. Ogilvie, Sheilagh. The European Guilds: An Economic Analysis. Princeton: Princeton University Press, 2019. Ogilvie, Sheilagh. “’Whatever Is, Is Right’? Economic Institutions in Pre-Industrial Europe”. The Economic History Review 60, no. 4 (1 November 2007): 649–84. https://doi.org/10.1111/j.1468-0289.2007.00408.x. Piketty, Thomas. Le capital au XXIe siècle. Paris: Éditions du Seuil, 2013. Putnam, Robert D, Robert Leonardi, and Raffaella Nanetti. Making Democracy Work: Civic Traditions in Modern Italy. Princeton, NJ: Princeton Univ. Press, 1993. Roymans, Nico. Ethnic Identity and Imperial Power. The Batavians in the Early Roman Empire. Amsterdam: Amsterdam University Press, 2004. Roymans, Nico, and Joris Gerardus Aarts. “Coin Use in a Dynamic Frontier Region. Late Iron Age Coinages in the Lower Rhine Area”. Journal of Archaeology in the Low Countries 1 (2009): 5–26. Scheidel, Walter. The Great Leveler: Violence and the History of Inequality from the Stone Age to the Twenty-First Century, 2016. Scheidel, Walter, and Steven J.  Friesen. “The Size of the Economy and the Distribution of Income in the Roman Empire”. Journal of Roman Studies 99 (2009): 61–91. Schiller, A. Arthur. Roman Law, Mechanisms of Development. Reprint 2010. Berlin, Boston: De Gruyter Mouton, 1978. https://doi.org/10.1515/ 9783110807196. Talbot, John. Made for Trade. A New View of Icenian Coinage. Oxford: Oxbow Books, 2018. Verboven, Koenraad. “Cité et réciprocité. Le rôle des croyances culturelles dans l’économie romaine”. Annales. Histoire, Sciences Sociales 67 (2012): 913–42. Verboven, Koenraad. “Currency, Bullion and Accounts. Monetary Modes in the Roman World”. Revue Belge de Numismatique et de Sigillographie 155 (2009): 91–124. Verboven, Koenraad. “Like Bait on a Hook. Ethics, Etics and Emics of Gift-­ Exchange in the Roman World”. In Gift Giving and the “embedded” Economy in the Ancient World, edited by Filippo Carlà and Maja Gori, 135–53. Heidelberg: Universitätsverslag Winter, 2014.

2  PLAYING BY WHOSE RULES? INSTITUTIONAL RESILIENCE, CONFLICT… 

51

Verboven, Koenraad. The Economy of Friends: Economic Aspects of « Amicitia » and Patronage in the Late Republic. Collection Latomus 269. Bruxelles: Latomus, 2002. Verboven, Koenraad. “The Knights Who Say NIE.  Can Neo-Institutional Economics Live up to Its Expectation in Ancient History Research  ?” In Structure and Performance in the Roman Economy. Models, Methods and Case Studies, edited by Koenraad Verboven and Paul Erdkamp, 33–57. Brussels: Latomus, 2015. Vromen, Jack J. Economic Evolution: An Inquiry into the Foundations of the New Institutional Economics. London: Routledge, 2007. Wegerich, Kai. “Institutional Change: A Theoretical Approach”. School of Oriental and African Studies (SOAS): Occasional Paper No 30, 2001. Wigg-Wolf, David G. “The Function of the Last Celtic Coinages in Northern Gaul”. In Coin Finds and Coin Use in the Roman World, edited by David G. Wigg-Wolf and Cathy E. King, 415–36. Berlin: Mann Verlag, 1993. Williamson, Oliver. “The New Institutional Economic: Taking Stock, Looking Ahead”. Journal of Economic Literature 38 (2000): 595–613.

CHAPTER 3

Networks as Proxies: A Relational Approach Towards Economic Complexity in the Roman Period Johannes Preiser-Kapeller 1   Systems, Complexity and Networks “Complexity economics” has become a prominent topic over the last three decades; as W. Brian Arthur, one of the pioneers of this field, explains: “complexity economic builds on the proposition that the economy is not This paper was prepared within the project “Harbours and landing places on the Balkan coasts of the Byzantine Empire (4th to 12th centuries)” (part of the SPP-1630 “Harbours from the Roman period to the Middle Ages”, funded by the Deutsche Forschungsgemeinschaft, 2013 to 2019); the project was undertaken at the Römisch-Germanisches Zentralmuseum (RGZM) in Mainz in cooperation with the Institute for Byzantine and Modern Greek Studies of the University of Vienna and the Institute for Medieval Research, Division of Byzantine Research, Austrian Academy of Sciences. Project leader was Prof. Falko Daim (RGZM). The paper was finished within the project “Moving Byzantium: Mobility, Microstructures and Personal Agency” (FWF Z288 Wittgensteinpreis), whose PI is Prof. Claudia Rapp (University of Vienna and Austrian Academy of Sciences). J. Preiser-Kapeller (*) Austrian Academy of Sciences, Vienna, Austria e-mail: [email protected] © The Author(s) 2021 K. Verboven (ed.), Complexity Economics, Palgrave Studies in Ancient Economies, https://doi.org/10.1007/978-3-030-47898-8_3

53

54 

J. PREISER-KAPELLER

necessarily in equilibrium: economic agents (firms, consumers, investors) constantly change their actions and strategies in response to the outcome they mutually create. (…) Complexity economics thus sees the economy in motion, perpetually ‘computing’ itself—perpetually constructing itself anew. Where equilibrium economics emphasizes order, determinacy, deduction, and stasis, complexity economics emphasizes contingency, indeterminacy, sense-making, and openness to change.”1 As Arthur is referring primarily to modern economic dynamics, one could ask how much place there was for “contingency, indeterminacy, sense-making, and openness to change” in pre-modern economic frameworks such as the Roman one. In any case, notions of “system”2 and “complexity”3 as well as “network” (whose study “for many scientists in the community (…) is synonymous with the study of complexity”4) are very much present in historical and even more in archaeological papers for some time now. While in many studies, these terms are used in a more “metaphorical” way or as novel conceptual framework for otherwise traditional narratives,5 a considerable number of scholars actually refer also to their formal and mathematical basis. Among these approaches we find:

1  Arthur, Complexity and the Economy, 1, see also 89–102 on “process and emergence in economy”. Cf. also Beaudreau, “On the Emergence and Evolution of Economic Complexity”. 2  For the notion of “system” with regard to economy, its “autopoiesis” and its relations to other “social systems” cf. Luhmann, Die Wirtschaft der Gesellschaft, esp. 43–90; Luhmann cites Maturana, “Autopoiesis,” for a definition of autopoiesis: “there are systems that are defined as unities of networks of productions of components that (1) recursively, through their interactions, generate and realize the network that produces them; and (2) constitute, in the space in which they exist, the boundaries of this network as components that participate in the realization of the network”. 3  As Arthur, Complexity and the Economy, 3, highlights, “complexity is not a theory but a movement in the sciences that studies how the interacting elements in a system create overall patterns, and how these overall patterns in turn cause the interacting elements to change or adapt”. Cf. also Mainzer, Thinking in Complexity; Bentley and Maschner, Complex Systems and Archaeology; Kohler, “Complex Systems and Archaeology”. For a good introduction for historians see Gaddis, The Landscape of History. 4  Johnson, Simply Complexity, 13. For an overview of network analytical approaches to economic complexity cf. Chatterjee and Chakrabarti, Econophysics; Jackson, Social and Economic Networks; Sinha et al., Econophysics, esp. 203–243; Easly and Kleinberg, Networks, Crowds, and Markets; Knoke, Economic Networks. 5  Cf. for instance Malkin, A Small Greek World, and most contributions in Malkin, Constantakopoulou, and Panagopoulou, Greek and Roman Networks.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

55

• Attempts to identify statistical “signatures of complexity” in quantitative data (e. g., distributions of settlements sizes within a region or of wealth among individuals in a larger group, respectively time series, esp. of prices, but also of proxy data), such as unequal distributions patterns (power law, logarithmic, etc.) or indicators for non-­ linear dynamics (Lyapunov-exponent, etc.).6 Furthermore, systems of equations are proposed in order to capture essential factors for these dynamics on the basis of the correspondence between patterns emerging from these models and observed data (a top-down approach).7 • Efforts to survey, map and analyse the connections and interactions between various elements (individuals, groups, settlements, polities, but also objects or semantic entities) documented in historical or archaeological evidence with the help of network models in the form of graphs with “nodes” and “ties”, also in their spatial and temporal dynamics. Again, statistical “signatures of complexity” (e.g. patterns of distribution of the number of links among nodes) are identified and models for their emergence in growing or changing networks are proposed (e.g. mechanisms of preferential attachment causing increasing inequality among nodes regarding their “centrality”) (see below for a further outline). • Experiments to capture the “bottom up” dynamics of complex systems emerging from the interaction of single elements with the help of agent-based models, acting on the basis of a set of (often relatively simple) rules within a simulated (spatial) environment over several time steps.8 Again, emerging statistical properties of 6  Cf. Brown, “Measuring Chaos“; Kantz and Schreiber, Nonlinear Time Series Analysis; Sinha et al., Econophysics, 83–129 (on distribution patterns). See the contribution by Dries Daems in this volume for such an approach. For a survey on studies producing proxy data for economic growth in ancient Greece cf. for instance Ober, “Wealthy Hellas”, and for Rome Wilson, “Quantifying Roman Economic Performance”; Bowman, “Quantifying the Roman Economy“; Jongman, “Re-constructing the Roman economy”. 7  Cf. Weidlich, Sociodynamics. For application on historical phenomena, see also Turchin, Historical Dynamics. 8  For the actual complexity of economic behaviour of individual agents, cf. Wilkinson, An Introduction to Behavioural Economics. For the ancient economy cf. Schefold, “The Applicability of Modern Economics”, who argues “that there are still good reasons to regard the economic rationality of the ancients as sufficiently different from ours to expect differences between mainstream economics and analytical or verbal approximations to the economics of antiquity in the description of economic processes”. See the contribution by Tom Brughmans in this volume for an agent-based modelling approach.

56 

J. PREISER-KAPELLER

such models are compared with observed data in order to determine their explanatory value.9 In several cases, these approaches are combined.10 Yet despite all mathematical and computer-based sophistication, like any other study of the past they depend on the density and quality of historical (mainly textual) or archaeological evidence. And while they may provide insights into processes and patterns otherwise “invisible” for a conventional analysis, any evaluation of their heuristic explanatory value relies on their check against “real” data. As I will demonstrate, network analysis provides tools to visualise and analyse the inherent complexity of various types of “real” historical and archaeological data and their combination (archaeological, geographical, textual) or even of a single piece of evidence, thus producing a wide range of “proxies for complexity”. 1.1  The Relational Approach: How to Model and Analyse Networks and to Measure Their Complexity In general, network theory assumes “not only that ties matter, but that they are organized in a significant way, that this or that (node) has an interesting position in terms of its ties”.11 One central aim of network analysis is the identification of structures of relations which emerge from the sum of interactions and connections between individuals, groups or sites and at the same time influence the scope of actions of everyone entangled in such relations. For this purpose, data on the categories, intensity, frequency and dynamics of interactions and relations between entities of interest is systematically collected in a way that allows for further mathematical analysis. This information is organised in the form of matrices (with rows and columns) and graphs (with nodes [vertices] and edges [links]). These are not only instruments of data collection and 9  Cf. now Madella and Rondelli, Simulating the Past. For a more general overview, cf. Sinha et al., Econophysics, 147–203 (with further literature). 10  See for instance several of the contributions in: Collar et al., Journal of Archaeological Method and Theory 22/1. 11   Lemercier, “Formale Methoden,” 22. Cf. also Brughmans, “Thinking Through Networks“; Collar et al., “Networks in Archaeology“, and the contributions in Knappett, Network-Analysis in Archaeology, Collar et al., Journal of Archaeological Method and Theory 22/1, and Leidwanger and Knappett, Maritime Networks. For basic ideas of network theory, see also Knoke, Economic Networks, 21–24.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

57

visualisation, but provide also the basis of further mathematical operations (on the basis of matrix algebra and graph theory).12 Once a quantifiable network model has been created, it allows for a structural analysis on three main levels13: • The level of single nodes; respective measures take into account the immediate “neighbourhood” of a node—such as “degree”, which measures the number of direct links of a node to other nodes14 or the relative centrality of a node within the entire network due to its position on many or few possible paths between nodes otherwise unconnected—the measure of “betweenness”, which can be interpreted as a potential for intermediation.15 A further important indicator of centrality is “closeness”, which measures the length of all paths between a node and all other nodes. The “closer” a node is, the lower is its total and average distance to all other nodes. Closeness can also be used as a measure of how fast it would take to spread resources or information from a node to all other nodes.16 • The level of groups of nodes, especially the identification of “clusters”, meaning the existence of groups of nodes more densely connected to each other than to the rest of the network; if all nodes within such a group are directly connected with each other, they are called “clique”; a measure of the degree to which nodes in a graph tend to cluster together is the “clustering coefficient” (with values between 0 and 1).17 In order to detect such cliques and clusters, the 12  Wasserman and Faust, Social Network Analysis; Prell, Social Network Analysis, 9–16; Barabási, Network Science, 42–67; Fuhse, Soziale Netzwerke, 41–57. 13  Collar et al., “Networks in Archaeology”, pp. 17–25, includes also a useful glossary of basic terms and concepts of network analysis as well as a well-balanced discussion of potential and pitfalls of network models in archaeology. For historical studies, the best discussion in this regard is Lemercier, “Formale Methoden”. See also Ruffini, Social Networks in Byzantine Egypt, 1–39. 14  Wasserman and Faust, Social Network Analysis, 178–183; de Nooy, Mrvar, and Batagelj, Exploratory Social Network Analysis, 63–64; Newman, Networks, 168–169; Prell, Social Network Analysis, 96–99; Sinha et al., Econophysics. 15  Burt, Brokerage and Closure; Wasserman and Faust, Social Network Analysis, 188–192; de Nooy, Mrvar, and Batagelj, Exploratory Social Network Analysis, 131–133; Newman, Networks, 185–193; Prell, Social Network Analysis, 103–107. 16  Wasserman and Faust, Social Network Analysis, 184–188; Prell, Social Network Analysis, 107–109. 17  Wasserman and Faust, Social Network Analysis, 254–257; Fuhse, Soziale Netzwerke, 74–78.

58 

J. PREISER-KAPELLER

inspection of a visualisation of a network can be already quite helpful; common visualisation tools arrange nodes more closely connected near to each other (“spring embedder” algorithms) and thus provide a good impression of such structures.18 For exact identification, there exist various algorithms of “group detection” (such as the ones developed by the physicist M. Newman, see below), which aim at an optimal “partition” of the network. It is of course also of interest to see, for instance, if the presence of nodes within such clusters can be related to specific qualitative attributes.19 A different approach is the concept of “structural equivalence” of nodes; here nodes are not attributed to the same block because they are connected to each other but because they have the same (or a very similar) structure of ties to other actors (thus, within a school network, one would encounter a block of “teachers” and one of “students”, between which similar structures of relations could be identified). Again, several tools of blockmodelling exist.20 • The level of the entire network: basic key figures are the number of nodes and links, the “network diameter” or maximum distance between two nodes (expressed as the number of links necessary to find a path from one to the other), and the average distance (or path length) between two nodes. A low average path length among nodes together with a high clustering coefficient can be connected to the model of a “small-world network”, in which most nodes are linked to each other via a relatively small number of edges.21 “Density” indicates the ratio of possible links actually present in a network: theoretically, all nodes in a network could be connected to each other (this would be a density of 1). A density of 0.1 indicates that 10 per cent of these possible links exist within a network; the higher 18  Cf. Krempel, Visualisierung komplexer Strukturen; Dorling, The Visualization of Spatial Social Structure. 19  de Nooy, Mrvar, and Batagelj, Exploratory Social Network Analysis, 66–77; Newman, Networks, 372–382; Prell, Social Network Analysis, 151–161; Kadushin, Understanding Social Networks, 46–49. 20  Wasserman and Faust, Social Network Analysis, 461–493; de Nooy, Mrvar, and Batagelj, Exploratory Social Network Analysis, 259–285; Prell, Social Network Analysis, 176–194. For an application on historical data cf. Padgett and Ansell, “Robust Action,” 1259–1319. 21  de Nooy, Mrvar, and Batagelj, Exploratory Social Network Analysis, 125–131; Prell, Social Network Analysis, 171–172. For the small-world model cf. Watts, Small Worlds; Sinha et al., Econophysics, 212–216. See also Malkin, A Small Greek World.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

59

the number of nodes, the higher of course the number of possible links. Thus, in general, density tends to decrease with the size of a network. Therefore, it only makes sense to compare the densities of networks of (almost) the same size. Density can be interpreted as an indicator for the network’s cohesion, but also for its complexity.22 Other measurements are based on the equal or unequal distribution of quantitative characteristics such as degree among nodes; a high “degree centralisation” indicates that many links are concentrated on a relatively small number of nodes.23 These distributions can be statistically analysed and visualised for all nodes (by counting the frequency of single degree values) and used for the comparison of networks; again, certain distribution patterns (most prominently, power laws) are interpreted as signatures of complexity (“scale free-networks”).24 Networks are of course dynamic: relationships may be established, maintained, modified or terminated; nodes appear in a network and disappear (also from our sources). Standard tools of network analysis (still) force us to integrate these changes into one more or less static model. The common solution to capture at least part of these dynamics is to use time slices—created by a researcher knowing the material on the basis of meaningful caesurae in the development of the research object—and to model different networks for each of these (see a simple example with two time slices in the following section).25 1.2  Example: A Riverine Transport Network from Roman Antiquity to the Early Middle Ages Within the project “Studies of inland harbours in the Frankish-German Empire as hubs for European communication networks (500–1250)”, Lukas Werther (University of Jena) has created a database of harbours and landing sites at rivers and lakes in Central Europe, France and Northern Italy from the Roman period to the year 1000, also integrating the recently  Prell, Social Network Analysis, 166–168; Kadushin, Understanding Social Networks, 29.  Prell, Social Network Analysis, 168–170. 24  Newman, Networks, 243–261; Sinha et  al., Econophysics, 216–220; Fuhse, Soziale Netzwerke, 103–105. 25  de Nooy, Mrvar, and Batagelj, Exploratory Social Network Analysis, 28–29; Batagelj et al., Understanding Large Temporal Networks and Spatial Networks. 22 23

60 

J. PREISER-KAPELLER

published catalogue of Christina Wawrzinek.26 For a another paper, we used this data for modelling and analysing riverine traffic networks and comparing their structure in the Roman (first to fifth century CE) and post-Roman (sixth to ninth century CE) period.27 In our model, harbour sites documented in historical or archaeological evidence serve as nodes and routes between them via river or lake navigation as links. Links in this model are both weighted (meaning that a quantity is attributed to them) and directed (a link leads from point A to point B, for instance). The aim is to integrate aspects of what Leif Isaksen has called transport friction into our calculations; otherwise, the actual costs of communication and exchange between sites, which would have influenced the frequency and strength of connections, would be ignored in network building. Links are weighted by using the inverted geographical distance between them; thus, a link would be the stronger the shorter the distance between two nodes (“distant decay” effect). Furthermore, directed links leading upstream (from point A to point B) are weighted with a third of the strength of links leading downstream (from point B to point A).28 We modelled two networks, one on the basis of data for the first to fifth century CE (period I) and one for the sixth to ninth century CE (period II) and determined the standard centrality measures on the level of individual nodes (degree and especially betweenness and closeness) and of the entire network. While a visualisation of the nodes (sized according to their relative centrality) on a geographical map illustrates continuities and changes with regard to the focal points of connectivity in these two models (see Appendix, Figs. 3.1–3.4), a comparison of quantitative measures indicates significant differences in range, connectivity and complexity between the two graphs (see Appendix, Table 3.2): the number of nodes is 25 per cent smaller in the period II network, the number of links is only  Cf. Wawrzinek, In portum navigare.  Preiser-Kapeller and Werther “Connecting Harbours”, which beyond the Po catchment also includes models for the upper Danube, the Rhine and the Rhone. 28  For the analysis of transport and traffic networks, cf. Rodrigue, Comtoi, and Slack, The Geography of Transport Systems, 307–317; Taaffe and Gauthier, Geography of Transportation, 100–158; Ducruet and Zaidi, “Maritime Constellations,” 151–168. For a more general approach, see Barthélemy, “Spatial Networks,” 1–101. For transport networks of the past see: Carter, “An Analysis of the Medieval Serbian Oecumene”; Pitts, “The Medieval River Trade Network”; Gorenflo and Bell, “Network Analysis,” 80–98; Isaksen, “The Application of Network Analysis”; Graßhoff and Mittenhuber, Untersuchungen zum Stadiasmos von Patara. See also Leidwanger et al., “A Manifesto for the Study of Ancient Mediterranean Maritime Networks”. For transport friction, cf. also van Lanen et al., “Best Travel Options”. 26 27

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

61

half the one of the period I model. Period II network’s (weighted) density is only two-thirds and its clustering coefficient half the size as for the period I model. Significant is also the difference of values for transitivity (which indicates the percentage of link pairs in the network where when node A is linked to node B and node B is linked to node C also node A is linked to node C) between period I (0.657) and period II (0.25). Another measure especially developed for transport networks (as planar graphs) is circuitry, measuring the share of the maximum number of cycles or circuits (= a finite, closed path in which the initial node of the linkage sequence coincides with the terminal node) actually present in a traffic network model. It thus indicates the existence of additional or alternative paths between nodes beyond those necessary to connect all nodes in the network and thereby indicates the network’s relative connectivity and complexity29; here the difference is even more significant (0.38 for the period I model and 0.10 for the period II model). The model for the riverine network in period II is thus not only spatially more confined but also less well connected and complex when compared with the model for period I (see Appendix, Fig. 3.5). This would correlate with assumptions on the relative reduction of organisational, economic and infrastructural complexity from the Roman to the post-Roman period.30

2   The Complex Network of the Roman Empire: A Macro-Perspective 2.1  “Complex” Debates on the Roman Economy It is safe to say that there does not exist consensus on core characteristics of complexity of the Roman imperial economy. One intense debate focuses on the degree of economic integration within the Roman Empire: was it an “enormous conglomeration of interdependent markets”, whose degree of economic integration resulted also in the interdependence of prices in different regions (Peter Temin) or do we have to assume that “connectivity and isolation were unevenly spread across” a highly fragmented 29  Rodrigue, Comtoi, and Slack, The Geography of Transport Systems, 310, 313, 315–316; Taaffe and Gauthier, Geography of Transportation, 104–105: the circuitry or alpha-index is calculated as share of the maximum number of circuits actually present in a graph. 30  For aspects of environmental change in this region, cf. Cremonini, Labate, and Curina, “The Late-Antiquity Environmental Crisis”.

62 

J. PREISER-KAPELLER

Mediterranean world with only some pockets of integrated markets (Paul Erdkamp, cf. also Peter Fibiger Bang)?31 Another discussion centres on the role and share of the imperial state in the economy: was Rome a “tributary empire”, whose transfers of goods for the army or the provision of the imperial capitals represented the predominant (or even only) sector of large-quantity trade? Did the empire´s demands and logistics at least very much determine orientation and scale of the axes of over-regional distribution and exchange both for the public and the private sector? Or is there an “overvaluation of the state-controlled economic sector” (Jean-Michel Carrié), leading to the “inadequate and somehow unrealistic idea that the imperial economy was controlled by a large redistributive system” (Bang).32 As Jean-Michel Carrié has outlined, for sure there is an “overrepresentation of the (state) sector in surviving documents”33; but for our question, these sources at least provide evidence for considerations on the (minimum) scale and degree of infrastructural and organisational complexity necessary to maintain the “particular flow of resources and population directed by the imperial center” on which its success and survival depended (what Sam White for the Ottoman case has called the “imperial ecology”).34 When Emperor Julian in 362 CE provided 420,000 modii of wheat from imperial estates around the cities of Chalkis, Epiphania and Hierapolis for the starving population of the megalopolis of Antioch (ca. 160  km on the road west of Hierapolis), we may assume (according to Michael Decker) that “Julian mobilized the produce of more than 26,250 iugera of land and the sweat of more than 2500 cultivators” in addition to 28,000 camels (with drivers, for the transport over land) for this supply “sufficient to feed approximately 262,500 adult males for a month or 4468 families for a year”. But we also learn that the first measure of the emperor had been the fixing of grain prices, which had provoked the 31  Temin, The Roman Market Economy; Erdkamp, The Grain Market in the Roman Empire; Bang, The Roman Bazaar. Cf. also Bowman, Alan, “Quantifying the Roman Economy,” 15–28, and Jongman, “Re-constructing the Roman economy”. 32  Cf. the overview of positions in: Carrié, “Were Late Roman and Byzantine Economies Market Economies,” 20–21; Bang, The Roman Bazaar, 68–69. See also now van Bavel, The Invisible Hand, for pre-modern “market economies”. 33  Carrié, “Were Late Roman and Byzantine Economies Market Economies,” 21. 34  For the concept of imperial ecology and an analysis of the circuits of the imperial metabolism centred on Constantinople in the Ottoman period, cf. White, The Climate of Rebellion, 16–51 (17 for the citation). On models of flow, cf. also Davies, “Linear and Nonlinear Flow Models”.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

63

major producers and dealers to “[hold] their grain back from the market” for lack of acceptable profits. Only then the imperial apparatus had to step in.35 In any case, this episode would qualify as sign of complexity of a system of various “economic agents (…) changing their actions and strategies in response to the outcome they mutually create”, showing “a complicated mix of ordered and disordered behaviour” as proposed by Arthur and other theoreticians.36 Furthermore, recent debates in economic history somehow vindicate the significance of the state for economic development—in many cases, state activity was “not a sufficient, (…) but necessary condition” for growth. Even warfare can be interpreted as important economic activity, and a military commander could be regarded as an economic agent like others.37 The collapse of the (Western) Roman Empire also provides argumenta ex negativo for the relevance of the Roman imperial framework for economic complexity and its trajectory in its absence. While there may be no consensus on the degree of market integration in the Roman economy, on the basis of archaeological evidence, it seems clear that one of its most remarkable features was the widespread diffusion of goods (as especially evident from pottery) “not only geographically (sometimes being transported over many hundreds of miles), but also socially (so that it reached not just the rich, but also the poor)”.38 In contrast, according to Bryan Ward-Perkins, the reduction of the lateral as well as vertical range of connectivity indicated the end of complexity so that “even in the few places, like Rome, where pottery imports and production remained exceptionally buoyant, the middle and lower markets for good-quality goods (…) had wholly disappeared”. The “dismembering of the Roman state, and the ending of centuries of security, were the crucial factors in destroying the

35  Decker, Tilling the Hateful Earth, 83–84 (with sources) and 257 (where Michael Decker makes a case “to rethink the nature of overland trade”, at least in the Roman Near East: “it was neither dominated by luxury goods, nor was it infrequent”). 36  Arthur, Complexity and the Economy, 15–16. 37  Vries, State, Economy and the Great Divergence; Rodger, “War as an Economic Activity”; Dincecco and Onorato, From Warfare to Wealth; Johnson and Koyama, Persecution and Toleration. For the Roman case see also the considerations in Löffl, Die römische Expansion, 313–475 (especially for the provinces in the Alps and at the Danube, also on the basis of archaeological and experimental data), and 476–486 (for the autonomy of Roman commanders as economic agents regarding logistics and administration, for instance). 38  Ward-Perkins, The Fall of Rome, 88.

64 

J. PREISER-KAPELLER

sophisticated economy of ancient times”.39 Such an interpretation of the end of the system, in turn, implies a considerable degree of interdependence in the centuries before, since otherwise its collapse would not have affected even relatively peripheral regions such as Roman Britain to such a dramatic degree. If interdependence had been small, an agglomeration of isolated, maybe self-sufficient, clusters or “small-worlds” of settlements and regions could have (re-)appeared after the disappearance of the overarching imperial framework, whose (maybe only slightly reduced) welfare depended mainly on their internal socio-economic dynamics as it would have done before. Clearly, however, this was not the case, and the fragments of the former system were alone less than their sum (as could be expected for a complex system).40 2.2  Modelling the Imperial Traffic System: The “ORBIS Stanford Geospatial Network Model of the Roman World” Can we capture aspects of the discussed integration and disintegration of the Roman economic system with the help of structural models? Currently the most exhaustive network model of sea and land routes of the Imperium Romanum is the “ORBIS Stanford Geospatial Network Model of the Roman World”, developed by Walter Scheidel and Elijah Meeks in order to estimate transport cost and spatial integration within the Roman Empire. ORBIS is based on a network of roads, river and sea routes (in total, 1104 links) between 678 nodes (places), weighted according to the cost of transport41 (see Appendix, Fig. 3.6). Since it is aiming at the entire empire’s traffic system, it is less detailed on the regional and local level than network models for smaller areas (such as the one presented above for the river Po). We have corrected this data (especially with regard to the 39  Ward-Perkins, The Fall of Rome, 106–107, 133. Cf. also McCormick, Origins of the European Economy, 778 and 782–783. On the distribution patterns of Roman pottery, see also Mees, Die Verbreitung von Terra Sigilatta. 40  For a similar interpretation of the effects of the Late Bronze Age “Collapse” in the Eastern Mediterranean see Cline, 1177 B.C., 164–170. In general on collapses in complexity, cf. Arthur, Complexity and the Economy, 144–157; Tainter, The Collapse of Complex Societies; Tainter, “Social Complexity and Sustainability”; Scheffer, Critical Transitions; Forman, Urban Ecology, 65–90; Haldon, The Empire That Would Not Die. 41  Scheidel et al., ORBIS v2. The data set was downloaded from: https://purl.stanford. edu/mn425tz9757 (Creative Commons Attribution 3.0 Unported License). For a similar model, see also Graham, “Networks, Agent-Based Models and the Antonine Itineraries”, 45–64.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

65

location of some places) and modified the network model so that the strength of the link between two nodes (places) is inversely proportionate to the “cost” of overcoming the distance between them as expressed by the travel time according to the calculations of the ORBIS-team42. Thus, the strength of links reflects the ease or difficulty of transport and mobility between two localities (see also above 1.2). In a first step, we analysed the spatial and statistical distribution of measures of centrality among the nodes of the network. The number and accumulated strength of links of a node (its weighted degree43) is of course high where many places are connected to each other via short distances and via the most convenient transport medium: the sea (such as in the Aegean). Statistically, the distribution of these degree values is very unequal, with a high number of nodes with relatively low degree centrality and a small number of “hubs” with high centrality values (see Appendix, Fig.  3.7). In contrast to this, as indicated above, betweenness measures the relative centrality of a node in the entire network based on its position on many (or few) potential shortest paths between nodes.44 In the ORBIS network, the hubs of maritime transport serve as most important integrators of the entire system in this regard; at the same, the statistical distribution of betweenness values is even more unequal than the one of degree centrality (see Appendix, Fig. 3.8). “Closeness” in turn measures the average length of all paths between a node and all other nodes in a network and indicates its overall centrality (or remoteness).45 Statistically, closeness-values are relatively equally distributed, but their spatial distribution demonstrates the decisive role of maritime connectivity via the Mediterranean for the cohesion of the entire network (see Appendix, Fig. 3.9). The ORBIS model, which is also characterised by a (relatively to its size) high value of circuitry (0.32, see above 1.2 for this measure), thus shows several signatures of complexity of large-scale networks. The ORBIS model can also be used to approach structural differentiation within the Roman traffic system; as outline above, networks are often structured in clusters: groups of nodes which are more densely and closely connected among each other than with the rest of the network. For the 42  In case of parallel links between nodes in the data set, the “cheapest” connection was selected. 43  Newman, Networks, 168–169. 44  Newman, Networks, 185–193. 45  Wasserman and Faust, Social Network Analysis, 184–188.

66 

J. PREISER-KAPELLER

identification, we use the algorithm for group detection developed by M.  Newman, which aims at an “optimal” partition of the network into clusters.46 Complex network are characterised by “nested clustering” across several levels of hierarchy; within clusters further sub-clusters can be detected, within which further cluster can be identified.47 With the help of the Newman algorithm we identify twenty-five regional resp. supra-regional clusters of higher internal connectivity within the ORBIS model (see Appendix, Fig. 3.10) (Table 3.1). Table 3.1  Newman-cluster numbers and regions included in these clusters Newman-cluster nr.

Regions

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25

Upper Danube, Eastern Alps North Syria, Northwest Mesopotamia, South Asia Minor Rhine area Middle Danube, North Balkans North and central Adriatic Central North Africa, East coast of Iberian Peninsula, Baleares South Iberian Peninsula, West North Africa Region around the Sea of Marmara, North Aegean Central and Northwest Aegean Central South Italy Palestine Britannia and Channel coast Rome, Latium and Campania Egypt Cyrenaica, Crete and South Peloponnese Southwest Asia Minor Cyprus and north coasts of Levant East North Africa, Sicily and Southwest of South Italy Black Sea and North of Asia Minor Etruria, Liguria, Corsica and Southeast of Gaul Gaul and North of the Iberian Peninsula South Adriatic and North Epirus Western plain of the river Po Northwest central Greece, North Peloponnese and Ionian Sea Central Aegean (micro-cluster)

 Newman, Networks, 372–382.  Barabási, Network Science, 331–338.

46 47

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

67

The majority of these clusters owe their connectivity again to either maritime connections (nr. 5, 6, 7, 8, 9, 12, 13, 15, 16, 17, 18, 19, 20, 21, 22, 24, 25) or riverine routes (nr. 1, 3, 4, 14, 23).48 In order to test the concept of nested clustering, we applied the Newman algorithm also on each of the twenty-five (supra-)regional clusters, resulting in the identification of between three and eight local or regional sub-clusters within each of the larger clusters (see Appendix, Fig. 3.10). This complex network model of localities and routes in the Roman Empire therefore across several spatial scales can be perceived as a system of nested clusters, down to the level of individual settlements and their hinterlands (see also Verboven’s contribution to this volume). In such a network, speed and cohesion of empire-wide connectivity depends on the trans-regional links between these clusters which structure the entire system. It is therefore also somehow located between the scenario of a fragmented Roman Mediterranean of Erdkamp or Bang and the model of a Roman Empire economically integrated by trade links (Temin, see above). But what happens, if these relatively cost-intensive, “fragile links between different people and different economies” across larger distances, as Ward-Perkins calls them,49 disappear? In order to target this question, we eliminated step by step all links from the model which would “cost” more than five, more than three, more than two and finally more than one day’s journey(s) (according to the calculations of the ORBIS-team) (see Appendix, Fig.  3.11). The result is an increasing fragmentation of the network in components of different size, partially along the fault lines between the clusters and sub-clusters, which we identified for the unmodified network model. But even if we eliminate the connections across longer distances, some larger, supra-regional clusters especially of maritime connectivity demonstrate remarkable robustness.50 In the model, in which all connections which “cost” more than one day’s journey are deleted, the largest still fully connected component is located in the Eastern Mediterranean between the Tyrrhenian Sea and the Levant, with its centre in the Aegean (see Appendix, Fig. 3.11). This would correspond to the 48  See also McCormick, Origins of the European Economy, 77–114, on the significance of riverine and maritime shipping. 49  Ward-Perkins, The Fall of Rome, 382. 50  Cf. McCormick, Origins of the European Economy, 565–569 (with map 19.2) on the “resilience” of certain sea routes in the seventh to ninth century CE.

68 

J. PREISER-KAPELLER

central regions and communication routes which remained under control of the (Eastern) Roman Empire after the loss of its eastern provinces to the Arabs in the seventh century, at the end of an actual process of increasing fragmentation of the (post)Roman world.51 Yet besides the resilience of maritime connectivity in regions of the Eastern Mediterranean (and an uninterrupted cohesion of the Egyptian cluster52), we observe a general disentanglement of large parts of the Roman traffic system, especially in the West of Europe, equally in the interior of the Balkans or also between the North and South coasts of the Mediterranean. The model is of course no perfect depiction of historical reality, but at best an appropriation towards certain structural parameters of the web of transport links within the Imperium Romanum. Nevertheless, we observe some remarkable parallels to actual historical processes of the fifth to seventh century CE (Chris Wickham for instance wrote about a partial “micro-­regionalisation“ of the “Mediterranean world-system” during this period),53 which hint at the impact of processes of integration respectively disentanglement especially due to the establishment and growth respectively the contraction of long distance connections.54 2.3  A Network of Places and Commodities on the Basis of One Piece of Textual Evidence The traffic system of the Roman Empire served as infrastructure for the mobility of humans, the transport of commodities and thus any form of market exchange based both on commercial and non-commercial activities (on the discussion of respective shares of these segments in the 51  Haldon, The Empire That Would Not Die; Brubaker and Haldon, Byzantium in the Iconoclast Era. See also now Arthur, Imperiale, and Muci, “Amphoras, Networks, and Byzantine Maritime Trade”. 52  A further remarkable parallel to the actual economic development, cf. Wickham, Framing the Early Middle Ages, 759–769. 53  Wickham, Framing the Early Middle Ages, 693–794, esp. 778–780, 792–794; Wickham, “The Mediterranean around 800,” 161–174; McCormick, Origins of the European Economy, 28–63. See also Ward-Perkins, The Fall of Rome. 54  The transport of larger amounts of commodities and numbers of people as common in the Roman imperial framework could not be compensated to a comparable amount which would have guaranteed the enduring cohesion of the Mediterranean system by new forms of mobility such as pilgrimage to the Holy Land or the transfer of relics which were continued during and after the crisis of Late Antiquity, cf. also McCormick, Origins of the European Economy, 270–277, 385–387.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

69

Roman economy, see above). One of the most interesting contemporaneous texts in this regard does not touch upon the Mediterranean centre of the empire, but on its foreign trade: the “Periplus of the Erythraean Sea”, a guide to trade and navigation in the Indian Ocean usually dated to the first century CE.55 Most recently, Eivind Heldaas Seland has used this text as basis for the modelling of various networks, demonstrating also the successful application of the method on a single (albeit also unique) piece of evidence for questions of Roman economic history; as he explains: “first, the text describes existing networks of people, places and commodities at the time of its composition. Second, the text allows us speculate on possible and potential linkages that are not definitively described. Third, the text itself can be approached as an inclusive macronetwork where words, for instance those describing places, relate to other words de-scribing products. It is this latter aspect of the textually conceived network that allows us to reconstruct former networks that were actually in existence or might well have been so.”56 Especially the last approach, via which Seland models a two-mode network of localities and of goods either exported from or imported to these places, provides a most interesting insight into the complexity of circuits of ancient exchange in the Indian Ocean. Based on the data set provided generously by Seland online,57 we were able to rebuild this network of thirtynine places and 112 commodities to apply further manipulations and analyses on it (see Appendix, Fig. 3.12). As Seland demonstrates, “the advantage of this network is that it allows us to look at supply/demand relationship in first-century Indian Ocean trade. While the narrative of the Periplus relates only what the author knew was traded in each port, the graph gives access to information on all the places where these products were available.”58 For further analysis, on the basis of this two-mode (or “affiliation”) network I modelled two one-mode-­networks: one of commodities, where two commodities are connected if they have at least one marketplace in common (see Appendix, Fig. 3.13), and one of marketplaces, where two localities are connected if they have at least one commodity in common (see Appendix, Fig.  3.14). Both networks are  Casson, The Periplus Maris Erythraei. Cf. also Parker, The Making of Roman India.  Seland, “The Periplus of the Erythraean Sea,” 194. 57  Permalink to dataset at: http://bora.uib.no/handle/1956/11470 (Dataset published under Creative Commons license 4.0: http://creativecommons.org/licenses/by/4.0). 58  Seland, “The Periplus of the Erythraean Sea,” 202. 55 56

70 

J. PREISER-KAPELLER

weighted networks, with the strength of links differing according to the number of common marketplaces or commodities. The network of commodities reflects the co-occurrence of items on the same marketplaces and illustrates their relative ubiquity or special positions within the web of exchanges (as narrated in the text). The network of marketplaces reflects the relative similarity of markets with regard to the range and variety of commodities traded there (and not links of direct exchange, as is a frequent misunderstanding especially if such affiliation networks, also often used to analyse the co-occurrence of artefact types on archaeological sites, are visualised on a geographical map, see Appendix, Fig. 3.15).59 The network of commodities consists of 112 nodes and 2135 links and is the most complex model (in terms of the number of connections) presented in this paper. The values for density (0.34), the clustering coefficient (0.79) and transitivity (0.71) suggest a relatively high ubiquity of commodities among places so that the maximum distance between two nodes is 2.24 (a small-world network of goods60). Still, an analysis of the actual distribution of weighted degree values among commodities shows a high inequality in the accumulated strength of ties of individual nodes (see Appendix, Fig. 3.13).61 The heavy weights of relative ubiquity with the highest degree values are grain, wine (Roman), money, tin and slaves. The first four of these goods are also the only ones which connect all the places with the highest degree values in the core of similarity among places. The highest betweenness values, on the contrast, are those of silverware, molochinon (from which cloth and garments were produced62), cotton garments, frankincense and precious stones; these more luxurious products co-occur with goods otherwise not to be found in the same circuits of distribution and serve as intermediaries between these circuits in the network (see Appendix, Fig. 3.13). 59  For this method and its application especially in archaeological network analysis cf. Sindbæk, “Broken Links and Black Boxes,” 71–94; Brughmans, “Thinking Through Networks”; Östborn and Gerding, “Network Analysis of Archaeological Data”; PreiserKapeller, “Thematic introduction”; Arthur, Imperiale, and Muci, “Amphoras, Networks, and Byzantine Maritime Trade”. For similarities between sites as basis for network modelling, cf. also Östborn and Gerding, “The Diffusion of Fired Bricks”. 60  See above on the notion of the “small-world network”. 61  On such distribution patterns in economics, cf. Sinha et al., Econophysics, 115–123. 62  Casson, The Periplus Maris Erythraei, 249, assumes that these were also cotton garments of especially high quality, but the debate is still open; cf. also Parker, The Making of Roman India, 157.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

71

Also the one-mode network of the thirty-nine places connected through ties of co-occurrence of commodities seems to be a densely interwoven small world with an average path length of two, a density of 0.35 and a clustering coefficient of 0.732 (see Appendix, Fig.  3.14). But nodes are again unequally integrated into this network; the accumulated weighted degree values of the top nine nodes (Myos Hormos, Berenike, Barygaza, Muziris and ex aequo Nelkynda, Kamara, Poduke and Sopatma) amount to 51 per cent of the total, with the Egyptian harbours of Myos Hormos and Berenike alone representing 16 per cent. Between these two places can be found also the strongest link of similarity (tie strength 64), while the next strongest tie (between Mundu and Mosyllon) amounts only to 11. The top five nodes in betweenness centrality, however, are located on either side of the pivotal Bab-el-Mandeb between East Africa and the Arab peninsula (Avalites, Muza) or in India (Barygaza, Ozene, Taprobane) (see Appendix, Fig.  3.15); only then, Myos Hormos shows up in the list. From the view of the entire network, the Egyptian harbours are important players, but also located at the Western periphery of the overall exchange system, whose integration depends on other intermediary nodes. Our analysis thus confirms the findings of Seland: the bias of the Periplus towards the perspective of traders coming from Roman Egypt and aiming at exchanging their products for those provided elsewhere makes itself clearly felt also in the network model. Yet also “Arabian, Indian, Persian Gulf and Bay of Bengal circuits” and the centrality of other nodes “become more visible” by such an “exercise”; network analysis helps to extract this implicit information, which is embedded in the text, but can be identified through reading only with difficulties.63 We can therefore approach in a different way the structural and commercial parameters under which Roman trade into the Indian Ocean was entangled with various regional and supra-regional circuits (this clustering also becomes visible if we apply the Newman algorithm on the network of places, see Appendix, Fig. 3.16), summing up to another complex commercial system beyond the Mare Internum of the Mediterranean.64  Seland, “The Periplus of the Erythraean Sea,” 204–205.  This structuring of a complex web of commodities and markets, inherent in the information stemming from only one source, shows similarities with results from the application of network theory to modern-day data on the combination of countries and products that they 63 64

72 

J. PREISER-KAPELLER

2.4  Micro-Perspectives and Qualitative Approaches of Network Analysis After inspecting network models of river ports, routes, commodities and marketplaces we may remember Brian’s statement from the beginning of this paper that complexity economics is based on the assumption that “economic agents (…) constantly change their actions and strategies in response to the outcome they mutually create” and ask: where are these agents? Of course, we assume that network structures, (changing) relative positions of nodes or distribution patterns emerge from the interplay of these agents (be it the emperor, a merchant, a craftsman, a peasant or the associations and organisations they form65). Based on amphorae stamps in the Monte Testaccio, for instance, Rubio Campillo and colleagues conclude that olive oil production was structured similarly as current firm-size distributions, since its distribution similarly followed a power law. This supports our assumption on emergent complex properties.66 To reflect on the actual social interactions behind this statistical pattern, it may be helpful to take into consideration Harrison White´s elaborate model for markets as networks of firms. He perceives “markets are tangible cliques of producers watching each other” (and less the consumers), creating an emerging “pecking order” of firms. This hierarchy becomes “taken for granted” in the form of a “self-reproducing role structure of relations among the producers”. Critics have observed that such dynamics are imaginable only for small markets with maybe a handful of producers; thus, this model could be more valid for

export by Ricardo Hausmann and César A. Hidalgo. On the basis of a two-mode network model respectively its transformation into a “network of relatedness between products”, defining a “product space of world economy”, they were able to detect subtle differences in the positions of countries regarding the ubiquity or diversity of their products, also correlating with their relative economic performance through time, cf. Hidalgo and Hausmann, “The Building Blocks of Economic Complexity”; Hausmann and Hidalgo, “The Network Structure of Economic Output,” 309–342; Sinha et al., Econophysics, 230–234. See also now the “popularised” version of these findings in: Hidalgo, Why Information Grows, 129–142. For a further development of this approach, see Caldarelli et  al. “A Network Analysis of Countries´ Export Flows“, and Cristelli, Tacchella, and Pietronero, “The Heterogeneous Dynamics of Economic Complexity”. 65  Cf. for instance Terpstra, Trading Communities. 66  Rubio Campillo et al., “Bayesian Analysis and Free Market Trade”.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

73

pre-modern than modern conditions.67 But while late medieval material allows us to survey, visualise and model the actual networks of interactions between economic agents,68 in most cases we lack comparably dense evidence for earlier periods of Mediterranean history. Again, therefore, we have to rely on proxies. Similar to Campillo’s work, Shawn Grahams analysis of the “network dynamics of the Tiber Valley Brick Industry” relies on the co-occurrence of stamps on various sites and highlights some structural dynamics of these networks (with changing degree distribution patterns reflecting different organisational patterns, for instance).69 Based on the data collected for the Roman potter shops (of terra sigillata) of Rheinzabern (Tabernae) by Allard W.  Mees, I prepared a similar network model for the connections between potters and potter groups based on the co-occurrence of commonly used identification marks on their widely distributed products. Here also, we observe a highly unequal distribution of (weighted) degree values for the network model across the entire period of activity of Rheinzabern wares (ca. 150–270 CE) (see Appendix, Fig. 3.17). At the same time, however, we also find differences in the density and structure of connectivity between the eight potter groups identified by Mees, whose activities started and ended at different points in time and targeted different sales areas (see Appendix, Fig. 3.18, Table 3.3).70 These differences could also reflect different forms of internal organisation or strategies of cooperation, with some groups depending on close interaction and stronger centralisation, while others were more loosely structured. The emergence of new production sites of terra sigillata in various provinces also illustrates the spread of skills and technology via networks between places and agents. Such phenomena of diffusion have been a particular focus of dynamic network modelling, especially regarding the dependence of diffusion patterns on underlying network  White, Markets from Networks; cf. also Knoke, Economic Networks, 60–64.  Padgett and McLean, “Organizational Invention and Elite Transformation,” 1463–1568; Burkhardt, Der hansische Bergenhandel; Apellániz, “Venetian Trading Networks,” 157–179; Preiser-Kapeller, “Liquid Frontiers”: Preiser-Kapeller, “The Maritime Mobility”. See Ruffini, Social Networks in Byzantine Egypt, for the application of network analysis on papyri from sixth-century Egypt. 69  Graham, “Networks, Agent-Based Models and the Antonine Itineraries”. Cf. also Graham and Weingart, “The Equifinality of Archaeological Networks”. 70  Cf. Mees, Organisationsformen römischer Töpfer-Manufakturen. 67 68

74 

J. PREISER-KAPELLER

structures. By inference, such patterns in turn provide clues on the density, structure and axes of interaction of networks, as Lars Boerner and Battista Severgnini, for instance, have demonstrated by using the spread of the Black Death as a proxy to measure economic interactions in fourteenth-century Europe. For the sixth to eighth century, diffusion patterns of the waves of the Justinianic Plague could provide similar insights even though, unfortunately, the density of evidence is much smaller71 (see the contribution by Boerner and Severgnini in this volume). In any case, both skills and germs depended on the mobility of individuals to spread, which leads back to the question of networks of economic agents and groups. In his recent book on “Trading Communities in the Roman World”, Taco T.  Terpstra argued that “information circulated within small but far-reaching groups defined by their members’ shared geographical origin; the loss of a member’s reputation or trading position within the group formed the instrument of contract enforcement”. As Terpstra himself makes clear, he borrowed heavily from the findings on medieval trading communities (or “diasporas”) such as the studies of Avner Greif.72 Eivind Heldaas Seland was able to propose some network models for the diffusion of and connections between trading communities (such as those from Palmyra) in the ancient Western Indian Ocean.73 But again, only the dense evidence for late medieval trade diasporas would allow for a more elaborate structural and quantitative analysis.74 These communities likewise highlight the significance of the overlap of various qualities and categories of ties—common geographic origin, ethnic or religious affiliation, kinship and economic exchange—for the emergence of more durable 71  Cf. Mees, Die Verbreitung von Terra Sigilatta, 185–122, and Easly and Kleinberg, Networks, Crowds, and Markets, 497–604 (for modes of diffusion on networks). Boerner and Severgnini, “Epidemic Trade”. For an analysis of the diffusion of religious ideas in the Roman Empire with the help of network theory, cf. Collar, Religious Networks in the Roman Empire. 72  Terpstra, Trading Communities, 2; Greif, Institutions and the Path to the Modern Economy. But see also Ogilvie, Institutions and European Trade, with some critical evaluation of Greif´s theses. 73  Seland, “Networks and Social Cohesion”; Seland, “Trade and Christianity,” pp. 72–86. For some considerations on the interplay between the diffusion of trading or religious diasporas and economic connectivity in that period cf. also Preiser-Kapeller, Jenseits von Rom und Karl dem Großen. 74  Cf. Apellániz, “Venetian Trading Networks”; Burkhardt, Der hansische Bergenhandel.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

75

forms of partnership and association; John F. Padgett has analysed the “Birth of Partnership Systems in Renaissance Florence” with the help of a model of “multiple-network ensembles”.75 The diffusion of technological skills or agricultural practices implies also connections not only between people, but also with and between plants, animals and objects. One could reflect on the socio-economic agency of the silk worm, reportedly smuggled by Christian monks from China to Byzantium in the time of Emperor Justinian I (527–565), or of the sugar cane, allegedly brought from India to China by Buddhist monks in the seventh century CE; they imported also the complex and intensive entanglements connected with the breeding and manufacture of these products, creating communities of practice as well as new networks between producers, traders and consumers across large distances.76 Theorists of Actor-Network-Theory (ANT) such as Bruno Latour postulate to regard humans and objects as equal actors within a network; he states: “anything that modifies a state of affairs by making a difference is an actor (…). Thus, the question to ask about any agent is simply the following: does it make a difference in the course of some other agent´s action or not?77 ANT has found some attention in archaeology, especially in two books by Carl Knappett and Ian Hodder.78 The latter emphasises the intensity of entanglements between humans and things: things depend on people when they are procured, manufactured, exchanged, used and discarded but in particular they depend on people to maintain them if they are to remain as people want them. Or they depend on humans to maintain the environments in which they thrive. Made things are not inert or isolated. Their connections with other things and their maintenance depend on humans. (…) this dependence of things on humans draws humans deeper into the orbit of things. Looking after things as they get depleted or fall apart or as they grow and reproduce trap humans into harder 75  Padgett and Powell, The Emergence of Organizations and Markets, 168–207, and 70–114 (for the wider theoretical framework on “autocatalytic networks”, inspired by findings from chemistry, for the emergence of organisations and markets). 76  For silk, see Muthesius, Byzantine Silk Weaving, 5–26, for sugar: Mazumdar, Sugar and Society in China; Ouerfelli, Le Sucre. In general, see also Roux, “Spreading of Innovative Technical Traits” (with further literature). 77  Latour, “Nonhumans”; cf. also Latour, Reassembling the Social. 78  Knappett, An Archaeology of Interaction; Hodder, Entangled.

76 

J. PREISER-KAPELLER

labor, greater social debts and duties, changes schedules and temporalities. (…) Humans have had increasingly to invest labor and new technologies to manage and sustain these things and have found themselves organized by them.79

Again, this hints at the actual degree of (inevitable) complexity and the number and scale of feedbacks inherent in most networks of economic agency we may encounter in the Roman or pre-modern history in general (already among the Neolithic farmers of Çatalhöyük, which Hodder uses as examples in his book).80 Latour pleads for a complete survey of all possible entanglements within a site and beyond across spatial and temporal scales with all actors and places necessary for a specific place to do something (“localising the global, globalising the local”).81 At the same time, we have to be aware of aspects of bias, selection, manipulation and fragmentariness inherent in all our pieces of evidence, be it an archaeological assemblage or a text. But we could understand all these phenomena as “narratives of entanglements”, which provide us with a certain perspective, a specific extract of the actual totality of entanglements (impossible to capture even for modern-day cases).82 Both Network Theory and Narratology lead us again to the possibility of quantitative and structural analysis—and in the case of “Quantitative Narrative Analysis” as developed by Roberto Franzosi, they are flowing together. Franzosi wrote: “Narrative texts are doubly relational. They depict both social relations and conceptual relations. (…) It is one thing to be able to say which (and perhaps how often) themes, concepts, actors appear in a text and another to be able to map the network of relations that give meaning to a text (or the social world).” Thus, he applied tools of quantitative network analysis to map the relations of violence between actors in the narratives of Italian newspapers during the rise of Fascism.83 The example of Seland’s analysis of the Periplus of the Erythrean Sea (see above), but also other recent  Hodder, Entangled, 85–87.  For the most interesting Chinese case, cf. also Leddertose, Ten Thousand Things. 81  Latour, Reassembling the Social, 173 and 200–202. Cf. also Preiser-Kapeller, “The Maritime Mobility of Individuals and Objects”. 82  Cf. also McGlade, “Simulation as Narrative”. For the possibility to use calculations on the input of energy and manpower into major building projects as proxies for the scale and complexity of economies and the framework of “energetics” for the Byzantine case see Pickett, “Beyond Churches”. 83  Franzosi, From Words to Numbers; Franzosi, Quantitative Narrative Analysis. 79 80

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

77

studies illustrate the potential of this approach for the analysis of historical texts.84 Targeting the inherent complexity in narratives may provide even the opportunity to use “big data” for an analysis of the relations between places, individuals and objects in the ancient Mediterranean. Especially the project “Pelagios: Enable Linked Ancient Geodata In Open Systems” (by Leif Isaksen, Elton Barker and Rainer Simon) has demonstrated the potential of linking data from large collections of texts and information on sites and artefacts85; the platform provides also tools to map the spatial distribution of this data and the connections between places on the basis of narrative co-reference (Pelagios Graph Explorer) and demonstrates their application on an increasing number of classical (but recently also medieval) texts.86 At the same time, modern language statistics techniques (such as Latent Semantic Analysis) focus on the analysis of such large text corpora; recent studies suggest that co-occurrences of named entities (city names and person names) in a big number of texts can be used to estimate the longitude and latitude of cities and the relations between places or between individuals.87 Further developments of these methods may allow scholars to fully exploit the potential of the mass of evidence which is already there to analyse the complex relational webs framing respectively emerging from the interplay between (economic) agents in the Roman world.

3   Conclusion Network theory aims at central aspects of (economic) complexity: the entanglements between agents and the structural patterns of relations and connections framing and emerging from their interactions. It allows for a modelling of networks across scales (socially—from the individual to the level of cities and polities, spatially—from one production site up to an 84  Cf. also Crespo Solana, Spatio-Temporal Narratives; Senturk, Narrative Social Structure; Fernández-Aceves, A Relational View. 85  http://pelagios-project.blogspot.co.at/. See also the related project “Google Ancient Places”: https://googleancientplaces.wordpress.com/. 86  See for instance Barker et al., “Writing Space, Living Space”, 229–247. 87  Cf. Louwerse and Zwaan, “Language Encodes Geographical Information”; Louwerse, Hutchinson, and Cai, “The Chinese Route Argument”; Hutchinson, Datla, and Louwerse, “Social Networks Are Encoded in Language”.

78 

J. PREISER-KAPELLER

entire empire, and temporally—from static snapshots for periods of different duration to series of time slices) and for an overlap of webs of ties of different qualities and categories, also stemming from different sources (“multiplex networks”88). Elaborate structural and quantitative analysis depends on a considerable density of evidence, but as demonstrated a single text may suffice to provide a glimpse at the inherent complexity of past economic interaction. Network models can serve as proxies to estimate the range, density and complexity of past economic life. Equally, the relational approach invites to a structural and quantitative comparison between periods, regions and the economic systems of polities and empires.89 An increasing number of proxies of this kind will allow us to capture the trajectories of economic complexity (beyond metaphors) from antiquity into the Middle Ages, whether they are characterised by evolutionary dynamics, self-reinforcing processes or phase transitions towards system states of augmented or significantly reduced complexity. Their creation as well as interpretation demands an even more intensive dialogue between humanities and sciences.90

 For an impressive example see Brughmans, Earl, and Keay, “Complex Networks in Archaeology”; see also Brughmans, Keay, and Earl, “Understanding Inter-settlement Visibility”. Cf. also Preiser-Kapeller, “Networks of border zones”. For the general values of a multiplex network approach see Szell, Lambiotte, and Thurner, “Multi-Relational Organization,” 13636: “Human societies can be regarded as large numbers of locally interacting agents, connected by a broad range of social and economic relationships. (…) Each type of relation spans a social network of its own. A systemic understanding of a whole society can only be achieved by understanding these individual networks and how they influence and co-construct each other. (…) A society is therefore characterized by the superposition of its constitutive socio-economic networks, all defined on the same set of nodes. This superposition is usually called multiplex, multi-relational or multivariate network.” 89  For a comparison between polities within the framework of complexity theory, cf. Preiser-Kapeller, “Calculating the Middle Ages”. 90  On this perspective, cf. also Boldizzoni, The Poverty of Clio. 88

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

79

Appendix

Fig. 3.1  Nodes in the network model of riverine transport in period I (first to fifth century CE) in the Po plain sized according to their betweenness centrality (data: L. Werther, map: J. Preiser-Kapeller, 2015)

80 

J. PREISER-KAPELLER

Fig. 3.2  Nodes in the network model of riverine transport in period II (sixth to ninth century CE) in the Po plain sized according to their betweenness centrality (data: L. Werther, map: J. Preiser-Kapeller, 2015)

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

81

Fig. 3.3  Nodes in the network model of riverine transport in period I (first to fifth century CE) in the Po plain sized according to their closeness centrality (data: L. Werther, map: J. Preiser-Kapeller, 2015)

Table 3.2  Comparison of network measures for the network model of riverine transport in period I (first to fifth century CE) and period II (sixth to ninth century CE) River network period I Number of nodes Number of edges Density Clustering coefficient Diameter Degree centralisation Betweenness centralisation Transitivity Circuitry (alpha-index)

22 36 0.156 0.505 13 0.091 0.435 0.657 0.385

Data: L. Werther, calculations: J. Preiser Kapeller

Po River network period II 17 19 0.136 0.265 25 0.070 0.436 0.250 0.103

82 

J. PREISER-KAPELLER

Fig. 3.4  Nodes in the network model of riverine transport in period II (sixth to ninth century CE) in the Po plain sized according to their closeness centrality (data: L. Werther, map: J. Preiser-Kapeller, 2015)

Fig. 3.5  The matrices for the network model of riverine transport in period I (first to fifth century CE, left) and period II (sixth to ninth century CE, right) (data. L. Werther, graphs: J. Preiser-Kapeller, 2015)

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

83

Fig. 3.6  User interface of the “ORBIS Stanford Geospatial Network Model of the Roman World” (screen shot from: http://orbis.stanford.edu/)

Fig. 3.7  ORBIS Stanford Geospatial Network Model of the Roman World— visualisation of the nodes (= places) sized according to their degree-centrality (analysis and map J. Preiser-Kapeller, 2015)

84 

J. PREISER-KAPELLER

Fig. 3.8  ORBIS Stanford Geospatial Network Model of the Roman World— visualisation of the nodes (= places) sized according to their betweenness-­centrality (analysis and map J. Preiser-Kapeller, 2015)

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

85

Fig. 3.9  ORBIS Stanford Geospatial Network Model of the Roman World— visualisation of the nodes (= places) coloured according to their closeness-­centrality (colour scale from red/low centrality to green/high centrality; analysis and map J. Preiser-Kapeller, 2015) (color online)

86 

J. PREISER-KAPELLER

Fig. 3.10  ORBIS Stanford Geospatial Network Model of the Roman World— identification of clusters (red) and sub-clusters (green) with the help of the Newman-algorithm (analysis and map J. Preiser Kapeller, 2015) (color online)

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

87

Fig. 3.11  ORBIS Stanford Geospatial Network Model of the Roman World— visualisation of routes with a “cost” of maximum one day’s journey between two places (analysis and map J. Preiser-Kapeller, 2015)

88 

J. PREISER-KAPELLER

Fig. 3.12  Two-mode network of places (red nodes) and commodities (green nodes) exported from or imported to them as narrated in the “Periplus of the Erythraean Sea” (data: E. H. Seland, http://bora.uib.no/handle/1956/11470; visualisation: J. Preiser-Kapeller, 2015) (color online)

Fig. 3.13  One-mode network of commodities due to their common export from or import to places as narrated in the “Periplus of the Erythraean Sea”; nodes sized according to their degree-centrality (data, visualisation and analysis as above)

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

89

90 

J. PREISER-KAPELLER

Fig. 3.14  One-mode network of places due to their common export from or import of commodities as narrated in the “Periplus of the Erythraean Sea”; nodes sized according to their degree-centrality (data, visualisation and analysis as above)

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

91

Fig. 3.15  One-mode network of places due to their common export from or import of commodities as narrated in the “Periplus of the Erythraean Sea” visualised on a geographical map (the links indicate ties of similarity due to the exchange of the same goods, not direct ties of interaction); nodes sized according to their betweenness-centrality (data, visualisation and analysis as above)

92 

J. PREISER-KAPELLER

Fig. 3.16  One-mode network of places due to their common export from or import of commodities as narrated in the “Periplus of the Erythraean Sea” visualised on a geographical map; identification of seven clusters of nodes (of different size) with the help of the Newman-algorithm (data, visualisation and analysis as above)

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

93

14 12

Frequency

10 8 6 4 2 0

-60

0

60

120

180

240

300

360 420 480 Degree values

540

600

660

720

780

840

900

Fig. 3.17  Frequency distribution of degree values of nodes in the network model of potters from Roman potter shops (of terra sigillata) of Rheinzabern (Tabernae, ca. 150–270 CE) due to the co-occurrence of commonly used hallmarks (data: MEES (2002); graph and analysis: J. Preiser-Kapeller, 2015) Table 3.3  The network model of potters from Roman potter shops (of terra sigillata) of Rheinzabern (Tabernae, ca. 150–270 CE) due to the co-occurrence of commonly used hallmarks; structural quantitative measure for the network models of the eight groups of potters identified by Mees Group 1 Group 2 Group 3 Group 4 Group 5 Group 6 Group 7 Group R Number of nodes Number of links Weighted link sum Density Weighted density Degree centralisation Betweenness Centralisation

17

8

4

21

11

3

6

17

135

25

6

185

54

3

10

70

2950

122

150

1339

374

8

28

171

0.993 0.381

0.893 0.189

1 0.625

0.881 0.193

0.982 0.34

1 0.667

0.667 0.373

0.515 0.126

0.243

0.385

0.35

0.18

0.263

0.625

0.46

0.176

0.527

0.482

1

0.16

0.622

0.5

0.4

0.178

Data: MEES (2002); network modelling and analysis: J. Preiser-Kapeller (2015)

94 

J. PREISER-KAPELLER

Fig. 3.18  The network model of potters from Roman potter shops (of terra sigillata) of Rheinzabern (Tabernae, ca. 150–270 CE) due to the co-occurrence of commonly used hallmarks; nodes are arranged in the eight groups of potters identified by Mees (data: MEES (2002); network modelling and graph: J. Preiser-­ Kapeller, 2015)

Bibliography Apellániz, Francisco, “Venetian Trading Networks in the Medieval Mediterranean.” Journal of Interdisciplinary History 44 (2013): 157–179. Arthur, W.  Brian, Complexity and the Economy. Oxford: Oxford University Press, 2015. Arthur, Paul, Marco Leo Imperiale, and Giuseppe Muci, “Amphoras, Networks, and Byzantine Maritime Trade.” In Maritime Networks in the Ancient Mediterranean World, ed. Justin Leidwanger and Carl Knappett, 219–237. Cambridge: Cambridge University Press, 2018. Bang, Peter Fibiger, The Roman Bazaar: A Comparative Study of Trade and Markets in a Tributary Empire. Cambridge: Cambridge University Press, 2008. Barabási, Albert-László, Network Science. Cambridge: Cambridge University Press, 2016. Barker, Elton, Stefan Bouzarovski, Leif Isaksen, and Chris Pelling, “Writing Space, Living Space. Time, Agency and Place Relations in Herodotus’s Histories,” in The Ideologies of Lived Space in Literary Texts, Ancient and Modern, ed. Jo Heirman and Jacqueline Klooster, 229–247. Ghent: Academia Press, 2013.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

95

Barthélemy, Marc, “Spatial Networks.” Physics Reports 499 (2011): 1–101. Batagelj, Vladimir, Patrick Doreian, Anuska Ferligoj, and Natasa Kejzar, Understanding Large Temporal Networks and Spatial Networks. Exploration, Pattern Searching, Visualization and Network Evolution. Chichester: Wiley, 2014. Beaudreau, Bernard C., “On the Emergence and Evolution of Economic Complexity.” Modern Economy 2 (2011): 266–278. Bentley, R.  Alexander, and Herbert, D.  G. Maschner, ed. Complex Systems and Archaeology. Salt Lake City: University of Utah Press, 2003. Boerner, Lars, and Battista Severgnini, “Epidemic Trade.” London School of Economics, Economic History Working Papers no. 212 (2014), http://www.lse. ac.uk/Economic-History/Assets/Documents/WorkingPapers/EconomicHistory/2014/WP212.pdf (accessed December 10, 2019). Boldizzoni, Francesco, The Poverty of Clio. Resurrecting Economic Theory. Princeton and Oxford: Princeton University Press, 2011. Bowman, Alan, “Quantifying the Roman Economy: Integration, Growth, Decline?” In Quantifying the Roman Economy: Methods and Problems, ed. Alan Bowman and Andrew Wilson, 3–83. Oxford: Oxford University Press, 2009. Brown, Thad A., “Measuring Chaos using the Laypunov-Exponent.” In Chaos Theory in the Social Sciences. Foundations and Applications, ed. L. Douglas Kiel and Euel Elliott, 53–66. Ann Arbor: University of Michigan Press, 1997. Brubaker, Leslie, and John Haldon, Byzantium in the Iconoclast Era c. 680–850: a History. Cambridge: Cambridge University Press, 2011. Brughmans, Tom, “Thinking Through Networks: A Review of Formal Network Methods in Archaeology.” Journal of Archaeological Method and Theory 20 (2012): 623–662. Brughmans, Tom, Graeme Earl, and Simon Keay, “Complex Networks in Archaeology: Urban Connectivity in Iron Age and Roman Southern Spain.” Leonardo 45 (2012): 280. Brughmans, Tom, Simon Keay, and Graeme Earl, “Understanding Inter-settlement Visibility in Iron Age and Roman Southern Spain with Exponential Random Graph Models for Visibility Networks.” Journal of Archaeological Method and Theory 22/1 (March 2015): 58–143. Burkhardt, Mike, Der hansische Bergenhandel im Spätmittelalter: Handel— Kaufleute—Netzwerke. Vienna and Cologne: Böhlau Köln, 2009. Burt, Ronald S., Brokerage and Closure: An Introduction to Social Capital. Oxford: Oxford University Press, 2005. Caldarelli, Guido, Matthieu Cristelli, Andrea Gabrielli, Luciano Pietronero, Antonio Scala, and Andrea Tacchella, “A Network Analysis of Countries´ Export Flows: firm Grounds for the Building Blocks of the Economy.” PLOS One 7/10, https://doi.org/10.1371/journal.pone.0047278 (accessed December 10, 2019).

96 

J. PREISER-KAPELLER

Carrié, Jean-Michel, “Were Late Roman and Byzantine Economies Market Economies? A Comparative Look at Historiography.” In Trade and Markets in Byzantium, ed. Cecile Morrisson, 13–26. Washington, DC: Harvard University Press, 2012. Carter, Francis W., “An Analysis of the Medieval Serbian Oecumene: A Theoretical Approach.” Geografiska Annaler. Series B, Human Geography 51 (1969): 39–56. Casson, Lionel, The Periplus Maris Erythraei: Text with Introduction, Translation, and Commentary. Princeton: Princeton University Press, 1989. Chatterjee, Arnab, and Bikas K.  Chakrabarti, eds. Econophysics of Markets and Business Networks. Heidelberg and New York: Springer, 2007. Cline, Eric H., 1177 B.C. The Year Civilization Collapsed. Princeton and Oxford: Princeton University Press, 2015. Collar, Anna, Religious Networks in the Roman Empire. The Spread of New Ideas. Cambridge: Cambridge University Press, 2013. Collar, Anna, Fiona Coward, Tom Brughmans, and Barbara J. Mills, “Networks in Archaeology: Phenomena, Abstraction, Representation.” Journal of Archaeological Method and Theory 22/1 (March 2015): 1–31. Cremonini, Stefano, Donato Labate, and Renata Curina, “The Late-Antiquity Environmental Crisis in Emilia Region (Po River Plain, Northern Italy): Geoarchaeological Evidence and Paleoclimatic Considerations.” Quaternary International 316 (2013): 162–178. Crespo Solana, Ana, ed. Spatio-Temporal Narratives: Historical GIS and the Study of Global Trading Networks (1500–1800). Cambridge: Cambridge Scholars Publishing, 2014. Cristelli, Matthieu, Andrea Tacchella, and Luciano Pietronero, “The Heterogeneous Dynamics of Economic Complexity.” PLOS One 10/2 (2015), http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0117174 (accessed December 10, 2019). Davies, John K., “Linear and Nonlinear Flow Models for Ancient Economies.” In The Ancient Economy. Evidence and Models, ed. Joseph G.  Manning and Ian Morris, 127–156. Stanford: Stanford University Press, 2005. de Nooy, Wouter, Andrej Mrvar, and Vladimir Batagelj, Exploratory Social Network Analysis with Pajek. Cambridge: Cambridge University Press, 2005. Decker, Michael, Tilling the Hateful Earth. Agricultural Production and Trade in the Late Antique East. Oxford: Oxford University Press, 2009. Dincecco, Mark, and Massimiliano Gaetano Onorato, From Warfare to Wealth. The Military Origins of Urban Prosperity in Europe. Cambridge: Cambridge University Press, 2017. Dorling, Danny, The Visualization of Spatial Social Structure. Chichester: Wiley, 2012.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

97

Ducruet, César, and Faraz Zaidi, “Maritime Constellations: A Complex Network Approach to Shipping and Ports.” Maritime Policy and Management 39 (2012): 151–168. Easly, David, and Jon Kleinberg, Networks, Crowds, and Markets. Reasoning About a Highly Connected World. Cambridge: Cambridge University Press, 2010. Erdkamp, Paul, The Grain Market in the Roman Empire: A Social, Political and Economic Study. Cambridge: Cambridge University Press, 2005. Fernández-Aceves, Hervin, A Relational View of the Norman Kingdom of Sicily and its Royal Court: The Social Space Constructed by ‘Hugo Falcandus’. Budapest: Central European University Master-Thesis, 2013 Forman, Richard T. T., Urban Ecology. Science of Cities. Cambridge: Cambridge University Press, 2014. Franzosi, Roberto, From Words to Numbers. Narrative, Data and Social Science. Cambridge: Cambridge University Press, 2004. Franzosi, Roberto, Quantitative Narrative Analysis. Los Angeles and London: SAGE Publications, 2010. Fuhse, Jan A., Soziale Netzwerke. Konzepte und Forschungsmethoden. Konstanz and Munich: UTB, 2016. Gaddis, John Lewis (2002), The Landscape of History. How Historians Map the Past. Oxford: Oxford University Press, 2002. Gorenflo, Larry J., and Thomas L. Bell, “Network Analysis and the Study of past Regional Organization.” In Ancient Road Networks and Settlement Hierarchies in the New World, ed. Charles D. Trombold, 80–98. Cambridge: Cambridge University Press, 1991. Graham, Shawn, Ex Figlinis: The Network Dynamics of the Tiber Valley Brick Industry in the Hinterland of Rome. Oxford: British Archaeological Reports, 2006a. Graham, Shawn (2006b), “Networks, Agent-Based Models and the Antonine Itineraries: Implications for Roman Archaeology.” Journal of Mediterranean Archaeology 19 (2006): 45–64. Graham, Shawn, and Scott Weingart, “The Equifinality of Archaeological Networks: an Agent-Based Exploratory Lab Approach.” Journal of Archaeological Method and Theory 22 (2015): 248–274. Graßhoff, Gerd, and Florian Mittenhuber, eds. Untersuchungen zum Stadiasmos von Patara. Modellierung und Analyse eines antiken geographischen Streckennetzes. Bern: Bern Studies, 2009. Greif, Avner, Institutions and the Path to the Modern Economy: Lessons from Medieval Trade. Cambridge: Cambridge University Press, 2006. Haldon, John F., The Empire That Would Not Die: The Paradox of Eastern Roman Survival, 640–740. Harvard: Harvard University Press, 2016. Hausmann, Ricardo, and César Hidalgo, “The Network Structure of Economic Output.” Journal of Economic Growth 16 (2011): 309–342.

98 

J. PREISER-KAPELLER

Hidalgo, César, Why Information Grows. The Evolution of Order, from Atoms to Economies. London: Basic Books, 2015. Hidalgo, César, and Ricardo Hausmann (2009), “The Building Blocks of Economic Complexity.” Proceedings of the National Academy of Sciences of the United States of America 106/26 (2009): 10570–10575. Hodder, Ian, Entangled. An Archaeology of the Relationships between Humans and Things. Malden and Oxford: Wiley-Blackwell, 2012. Hutchinson, Sterling, Vivek Datla, and Max M. Louwerse, “Social Networks Are Encoded in Language.” In Proceedings of the 34th Annual Conference of the Cognitive Science Society, 491–496. Austin, Texas, 2012, https://escholarship. org/uc/item/3d88c1j1 (accessed December 10, 2019). Isaksen, Leif, “The Application of Network Analysis to Ancient Transport Geography: A Case Study of Roman Baetica.” Digital Medievalist 4 (2008): https://journal.digitalmedievalist.org/articles/10.16995/dm.20/ (accessed December 10, 2019). Jackson, Matthew O., Social and Economic Networks. Princeton and Oxford: Princeton University Press, 2008. Johnson, Neil, Simply Complexity. A Clear Guide to Complexity Theory. London: Oneworld Publications, 2009. Johnson, Noel D., and Mark Koyama, Persecution and Toleration: The Long Road to Religious Freedom. Cambridge: Cambridge University Press, 2019. Jongman, Willem M., “Re-constructing the Roman economy.” In The Cambridge History of Capitalism. Volume 1. The Rise of Capitalism: From Ancient Origins to 1848, ed. Larry Neal and Jeffrey G.  Williamson, 75–100. Cambridge: Cambridge University Press, 2014. Kadushin, Charles, Understanding Social Networks. Theories, Concepts, and Findings. Oxford: Oxford University Press, 2012. Kantz, Holger, and Thomas Schreiber, Nonlinear Time Series Analysis. Cambridge: Cambridge University Press, 2004. Knappett, Carl, An Archaeology of Interaction. Network Perspectives on Material Culture and Society. Oxford: Oxford University Press, 2011. Knappett, Carl, ed. Network-Analysis in Archaeology. New Approaches to Regional Interaction. Oxford: Oxford University Press, 2013. Knoke, David, Economic Networks. Cambridge: Polity, 2012. Kohler, Timothy A., “Complex Systems and Archaeology.” In Archaeological Theory Today, ed. Ian Hodder, 93–123. Cambridge, 2012. Krempel, Lothar, Visualisierung komplexer Strukturen. Grundlagen der Darstellung mehrdimensionaler Netzwerke. Frankfurt and New York: Campus Verlag, 2005. Latour, Bruno, “Nonhumans.” In Patterned Ground. Entanglements of Nature and Culture, ed. Stephan Harrison, Steve Pile, and Nigel Thrift, 224–227. London: Reaktion Books, 2004.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

99

Latour, Bruno, Reassembling the Social. An Introduction to Actor-Network-Theory. Oxford: Oxford University Press, 2005. Leddertose, Lothar, Ten Thousand Things. Module and Mass Production in Chinese Art. Princeton: Princeton University Press, 2000. Leidwanger, Justin, et al., “A Manifesto for the Study of Ancient Mediterranean Maritime Networks.” Antiquity 88 (2014): http://journal.antiquity.ac.uk/ projgall/leidwanger342 (accessed December 10, 2019). Leidwanger, Justin and Carl Knappett, eds. Maritime Networks in the Ancient Mediterranean World. Cambridge: Cambridge University Press, 2018. Lemercier, Clair (2012), “Formale Methoden der Netzwerkanalyse in den Geschichtswissenschaften: Warum und Wie?” Österreichische Zeitschrift für Geschichtswissenschaften 23 (2012): 16–41. Löffl, Josef, Die römische Expansion. Berlin: Frank & Timme, 2011. Louwerse, Max M., and Rolf A.  Zwaan, “Language Encodes Geographical Information”. Cognitive Science 33 (2009): 51–73. Louwerse, Max M., Sterling Hutchinson, and Zhiqiang Cai, “The Chinese Route Argument: Predicting the Longitude and Latitude of Cities in China and the Middle East Using Statistical Linguistic Frequencies.” In Proceedings of the 34th Annual Conference of the Cognitive Science Society, 695–700. Austin, Texas, 2012: https://pdfs.semanticscholar.org/870f/4e3876dfa64f0756032badcf ee48e0b00d0a.pdf (accessed December 10, 2019). Luhmann, Niklas, Die Wirtschaft der Gesellschaft. Frankfurt am Main: Suhrkamp, 1994. Madella, Marco, and Bernardo Rondelli, eds. Simulating the Past: Exploring Change Through Computer Simulation in Archaeology. Special issue of Journal of Archaeological Method and Theory, 21/2 (2014). Mainzer, Klaus, Thinking in Complexity. The Computational Dynamics of Matter, Mind, and Mankind. Berlin / Heidelberg: Springer, 2007. Malkin, Irad, A Small Greek World. Networks in the Ancient Mediterranean. Oxford: Oxford University Press, 2011. Malkin, Irad, Christy Constantakopoulou, and Katerina Panagopoulou, eds. Greek and Roman Networks in the Mediterranean. London and New York: Taylor & Francis, 2009. Maturana, Humberto R., “Autopoiesis.” In Autopoiesis: a Theory of Living Organization, ed. Milan Zeleney, 45–79. New York: Elsevier Science, 1981. Mazumdar, Sucheta, Sugar and Society in China: Peasants, Technology, and the World Market. Harvard: Harvard University Press, 1998. McCormick, M., Origins of the European Economy. Communications and Commerce AD 300–900. Cambridge: Cambridge University Press, 2001. McGlade, James, “Simulation as Narrative: Contingency, Dialogics, and the Modeling Conundrum.” Journal of Archaeological Method and Theory 21 (2014): 288–305.

100 

J. PREISER-KAPELLER

Mees, Allard W., Organisationsformen römischer Töpfer-Manufakturen am Beispiel von Arezzo und Rheinzabern. 2 Vols., Mainz: Römisch-Germanisches Zentralmuseum, 2002. Mees, Allard W., Die Verbreitung von Terra Sigilatta aus den Manufakturen von Arezzo, Pisa, Lyon und La Graufesenque. Mainz: Schnell & Steiner, 2011. Muthesius, Anna, Byzantine Silk Weaving AD 400 to AD 1200. Vienna: Fassbaender Verlag, 1997. Newman, Mark, Networks. An Introduction. Oxford: Oxford University Press, 2010. Ober, Josiah, “Wealthy Hellas.” Journal of Economic Asymmetries 8 (2011): 1–38. Ogilvie, Sheilagh, Institutions and European Trade. Merchant Guilds, 1000–1800. Cambridge: Cambridge University Press, 2011. Östborn, Per, and Henrik Gerding, “Network Analysis of Archaeological Data: a systematic Approach.” Journal of Archaeological Science 46 (2014): 75–88. Östborn, Per, and Henrik Gerding, “The Diffusion of Fired Bricks in Hellenistic Europe: A Similarity Network Analysis.” Journal of Archaeological Method and Theory 22 (2015): 306–344. Ouerfelli, Mohamed, Le Sucre. Production, commercialisation et usages dans la Méditerranée médiévale. Leiden and Boston: Brill, 2008. Padgett, John F., and Christopher K. Ansell, “Robust Action and the Rise of the Medici, 1400–1434.” The American Journal of Sociology 98 (1993): 1259–1319. Padgett, John F., and Paul D.  McLean (2006), “Organizational Invention and Elite Transformation: The Birth of Partnership Systems in Renaissance Florence.” American Journal of Sociology 111 (2006): 1463–1568. Padgett, John F., and Walter W.  Powell, The Emergence of Organizations and Markets. Princeton and Oxford: Princeton University Press, 2012. Parker, Grant, The Making of Roman India. Cambridge: Cambridge University Press, 2008. Pickett, Jordan, “Beyond Churches: Energetics and Economies of Construction in the Byzantine World.” In Byzantine Archaeology in Method and Theory, eds. Kostis Kourelis and William Caraher. Cambridge: Cambridge University Press, 2021. Pitts, Forrest R., “The Medieval River Trade Network of Russia Revisited.” Social Networks 1 (1978): 285–292. Preiser-Kapeller, Johannes, “Networks of border zones—multiplex relations of power, religion and economy in South-eastern Europe, 1250–1453 CE”. In Proceedings of the 39th Annual Conference of Computer Applications and Quantitative Methods in Archaeology, “Revive the Past” (CAA) in Beijing, China, 381–393. Amsterdam: Pallas Publications, 2012. Preiser-Kapeller, Johannes, “Calculating the Middle Ages? The Project “Complexities and Networks in the Medieval Mediterranean and the Near East”.” Medieval Worlds 2 (2015a): 100–127.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

101

Preiser-Kapeller, Johannes, “Liquid Frontiers. a Relational Analysis of Maritime Asia Minor as Religious Contact Zone in the 13th–15th Century.” In: Islam and Christianity in Medieval Anatolia, ed. Andrew Peacock, Bruno De Nicola, and Sara Nur Yildiz, 117–146. Aldershot: Ashgate, 2015b. Preiser-Kapeller, Johannes, “Thematic introduction.” In Harbours and Maritime Networks as Complex Adaptive Systems, ed. Johannes Preiser-Kapeller and Falko Daim, 1–24. Mainz: Römisch-Germanisches Zentralmuseum, 2015c. Preiser-Kapeller, Johannes, “The Maritime Mobility of Individuals and Objects: Networks and Entanglements” In Harbours and Maritime Networks as Complex Adaptive Systems, ed. Johannes Preiser-Kapeller and Falko Daim, 119–140. Mainz: Römisch-Germanisches Zentralmuseum, 2015d. Preiser-Kapeller, Johannes, Jenseits von Rom und Karl dem Großen. Aspekte der globalen Verflechtung in der langen Spätantike, 300–800 n. Chr. Vienna: Mandelbaum Verlag, 2018. Preiser-Kapeller, Johannes and Lukas Werther, “Connecting Harbours. a Comparison of Traffic Networks Across Ancient and Medieval Europe.” In Harbours as objects of interdisciplinary research—Archaeology + History + Geoscience, ed. Claus von Carnap-Bornheim et  al., 7–31. Mainz: Römisch-­ Germanisches Zentralmuseum, 2018. Prell, Christina, Social Network Analysis. History, Theory and Methodology. Los Angeles and London: SAGE Publications, 2012. Rodger, Nicholas Andrew Martin, “War as an Economic Activity in the “Long” Eighteenth Century.” International Journal of Maritime History 22 (2010): 1–18. Rodrigue, Jean-Paul, Claude Comtoi, and Brian Slack, The Geography of Transport Systems. London and New York: Routledge, 2013. Roux, Valentine, “Spreading of Innovative Technical Traits and Cumulative Technical Evolution: Continuity or Discontinuity?” Journal of Archaeological Method and Theory 20 (2013): 312–330. Rubio-Campillo, Xavier, María Coto-Sarmiento, Jordi Pérez-Gonzalez, and José Remesal Rodríguez, “Bayesian Analysis and Free Market Trade within the Roman Empire.” Antiquity 91 (2017): 1241–1252. Ruffini, Giovanni, Social Networks in Byzantine Egypt. Cambridge: Cambridge University Press, 2008. Scheffer, Marten, Critical Transitions in Nature and Society. Princeton and Oxford: Princeton University Press, 2009. Schefold, Bertram, “The Applicability of Modern Economics to Forms of Capitalism in Antiquity: Some Theoretical Considerations and Textual Evidence.” The Journal of Economic Asymmetries 8 (2001): 131–163. Scheidel, Walter, Elijah Meek, Karl Grossner, and Noemi Alvarez, ORBIS v2: The Stanford Geospatial Network Model of the Roman World. Stanford, 2014: http://orbis.stanford.edu/ (accessed December 10, 2019).

102 

J. PREISER-KAPELLER

Seland, Eivind Heldaas, “Trade and Christianity in the Indian Ocean during Late Antiquity.” Journal of Late Antiquity 5 (2012): 72–86. Seland, Eivind Heldaas, “Networks and Social Cohesion in Ancient Indian Ocean Trade: Geography, Ethnicity, Religion.” Journal of Global History 8 (2013): 373–390. Seland, Eivind Heldaas, “The Periplus of the Erythraean Sea: A Network Approach.” Asian Review of World Histories 4 (2016): 191–205. Senturk, Recep, Narrative Social Structure: Anatomy of the Hadith Transmission Network, 610–1505. Stanford: Stanford University Press, 2005. Sindbæk, Søren M., “Broken Links and Black Boxes: Material Affiliations and Contextual Network Synthesis in the Viking world.” In Network-Analysis in Archaeology. New Approaches to Regional Interaction, ed. Carl Knappett, 71–94. Oxford: Oxford University Press, 2013. Sinha, Sitabhra, Arnab Chatterjee, Anirban Chakraborti, and Bikas K. Chakrabarti, Econophysics. Weinheim: Wiley-VCH, 2010. Szell, Michael, Renaud Lambiotte, and Stefan Thurner, “Multi-Relational Organization of Large-Scale Social Networks in an Online World.” Proceedings of the National Academy of Sciences 107 (2010): 13636–13641. Taaffe, Edward James, and Howard L.  Gauthier, Geography of Transportation. Englewood Cliffs, NJ: Pearson Education, 1973. Tainter, Joseph A., The Collapse of Complex Societies. Cambridge: Cambridge University Press, 1988. Tainter, Joseph A., “Social Complexity and Sustainability.” Ecological Complexity 3 (2006): 91–103. Temin, Peter, The Roman Market Economy. Princeton and Oxford: Princeton University Press, 2013. Terpstra, Taco, Trading Communities in the Roman World. A Micro-Economic and Institutional Perspective. Leiden and Boston: Brill, 2013. Thome, Helmut, Zeitreihenanalyse. Eine Einführung für Sozialwissenschaftler und Historiker. Munich and Vienna: De Gruyter Oldenbourg, 2005. Turchin, Peter, Historical Dynamics: Why States Rise and Fall. Princeton and Oxford: Princeton University Press, 2003. van Bavel, Bas, The Invisible Hand?: How Market Economies have Emerged and Declined Since AD 500. Oxford: Oxford University Press, 2016. van Lanen, Rowin J., Menne C. Kosian, Bert J. Groenewoudt, Theo Spek, and Esther Jansma, “Best Travel Options: Modelling Roman and Early-Medieval Routes in the Netherlands Using a Multi-Proxy Approach.” Journal of Archaeological Science: Reports 3 (2015): 144–159. Vries, Peer, State, Economy and the Great Divergence. Great Britain and China, 1680s–1850s. London: Bloomsbury Academic, 2015. Ward-Perkins, Brian, The Fall of Rome and the End of Civilization. Oxford: Oxford University Press, 2005.

3  NETWORKS AS PROXIES: A RELATIONAL APPROACH… 

103

Wasserman, Stanley, and Katherine Faust, Social Network Analysis: Methods and Applications, Structural Analysis in the Social Sciences. Cambridge: Cambridge University Press, 1994. Watts, Duncan J., Small Worlds. The Dynamics of Networks between Order and Randomness. Princeton and Oxford: Princeton University Press, 1999. Wawrzinek, Christina, In portum navigare. Römische Häfen an Flüssen und Seen. Berlin: De Gruyter, 2014. Weidlich, Wolfgang, Sociodynamics: A Systemic Approach to Mathematical Modelling in the Social Sciences. Mineola, New York: Dover Publications, 2003. White, Harrison C., Markets from Networks: Socioeconomic models of production. Princeton: Princeton University Press, 2002. White, Sam, The Climate of Rebellion in the Early Modern Ottoman Empire. Cambridge: Cambridge University Press, 2011. Wickham, Chris, “The Mediterranean around 800: On the Brink of the Second Trade Cycle.” Dumbarton Oaks Papers 58 (2004): 161–174. Wickham, Chris, Framing the Early Middle Ages. Europe and the Mediterranean, 400–800. Oxford: Oxford University Press, 2005. Wilkinson, Nick, An Introduction to Behavioural Economics. New York: Red Globe Press, 2008. Wilson, Andrew Ian, “Quantifying Roman Economic Performance by Means of Proxies: Pitfalls and Potential.” In Long-Term Quantification in Ancient Mediterranean History, ed. François de Callataÿ, 147–167. Bari: Edipuglia, 2014.

CHAPTER 4

Evaluating the Potential of Computational Modelling for Informing Debates on Roman Economic Integration Tom Brughmans

1   Introduction The study of the Roman economy is a thriving discipline in need of methodological innovations. A large number of complex theories exist that aim to explain aspects of the functioning and performance of the Roman economy.1 Yet their sheer complexity, often involving a large number of explanatory factors that are argued to affect each other, means that they cannot be compared or tested through the traditional and current practice of qualitative argumentation and comparison with selected small sets of written and material sources alone. This has led to complex theories often being debated as conflicting even though their proponents admit they 1  Scheidel, Morris, and Saller, The Cambridge Economic History of the Greco-Roman World; Scheidel, The Cambridge Companion to the Roman Economy.

T. Brughmans (*) Centre for Urban Network Evolutions (UrbNet) and Classical Archaeology, Aarhus University, Aarhus, Denmark e-mail: [email protected] © The Author(s) 2021 K. Verboven (ed.), Complexity Economics, Palgrave Studies in Ancient Economies, https://doi.org/10.1007/978-3-030-47898-8_4

105

106 

T. BRUGHMANS

must to some extent be overlapping and compatible.2 The inability to formally study this overlap hampers comparison and evaluation of hypotheses and the development of new theories on solid foundations. The existing approaches to the study of the Roman economy therefore need to be complemented with formal computational tools and approaches to enable comparison of complex theories.3 Most scholars of the Roman economy recognise this need but have so far addressed it only to a very limited extent.4 We need to know what data patterns we would expect to see if a certain hypothesis was true, what happens when we change the importance of an explanatory factor, how these expected patterns differ from those of other hypotheses, and how the available written and material data can be used in formal tests to determine the probability of each hypothesis. I will argue here that this need can be addressed by adding computational modelling to the list of tools used by scholars of the Roman economy. Complex theories of the functioning and performance of the Roman economy need to be broken down into the hypothesised mechanisms and explanatory factors they consist of. Computational modelling allows for the behaviour of these components to be studied in isolation, and enables the simulation of the data patterns one would expect to see as their outcomes. Only when individual mechanisms are understood on their own terms can they be combined into more complex models. I argue that findings from computational models can constructively contribute to ongoing debates of complex theories by providing reproducible expectations for suggested processes, and creating solid foundations for more data-driven contextual discussions of these theories. Moreover, they hold the potential of assessing the probability of hypotheses through testing with large archaeological datasets where possible and available. In this chapter I illustrate this approach by means of a case study simulating the correlation between prices of tableware and the distance from their place of production. Economists argue that a strong correlation  A good example and overview of this is offered by Wilson et al., “A Forum on Trade”.  For an extensive discussion of this need, see this recently published manifesto: Brughmans et al., “Formal Modelling Approaches to Complexity Science in Roman Studies”. 4  Notable exceptions are the “Finding the limits of the Limes” project (Verhagen, Joyce, and Groenhuijzen, Finding the Limits of the Limes), the EPNet project in Barcelona (http:// www.roman-ep.net/ (last accessed 2019-10-15)) and recent work by Graham and Weingart (“The Equifinality of Archaeological Networks”) and Scheidel (“The Shape of the Roman World: Modelling Imperial Connectivity”). 2 3

4  EVALUATING THE POTENTIAL OF COMPUTATIONAL MODELLING… 

107

between the price of a good at a place and the distance from that place to the production region of that good is a signature of a perfectly competitive market in which prices affect each other.5 So we would expect to see a correlation between price and distance if the Roman economy was an integrated system in which reliable information about the prices of goods current at different markets was available to rational commercial actors constrained by a constant transport cost. Unfortunately, insufficient data on contemporary prices of goods are available to prove or reject this assumption.6 For example, Temin argues this correlation can be observed in the six contemporary prices he collected,7 whereas Bransbourg collected twelve prices to argue the correlation does not hold for most of the Empire if we incorporate some of the uncertainty in price data.8 This, however, has not prevented scholars of the Roman economy from formulating multiple competing theories about the degree of economic integration under the Roman Empire. The case study presented in this chapter will illustrate how, in the absence of empirical data, computational modelling can constructively contribute to ongoing debates by enabling formal comparisons between the expected outcomes of competing hypotheses. For example, Scheidel uses the ORBIS formal model of the Roman transport system to simulate transport costs taking into account physical friction to transport routes, allowing him to state that simulated results of this model disagree with Temin’s claims.9 In light of the methodological aim of this chapter, it should be stressed that the model presented here does not reflect our current knowledge of tableware prices and distribution mechanisms but merely aims to illustrate how predictions for a hypothesis of market integration can be simulated and compared to the archaeological record.

2   Temin’s Roman Market Economy Peter Temin offers a workable theoretical model to explore correlations between price and distribution distance of goods as a hypothesis. Temin posits that government involvement in wheat trade was limited and that 5  Kessler and Temin, “Money and Prices in the Early Roman Empire”; Temin, The Roman Market Economy. 6  Against this objection see Temin, The Roman Market Economy, 49. 7  Temin, The Roman Market Economy, 41–42. 8  Bransbourg, “Rome and the Economic Integration of Empire”. 9  Scheidel, “Shape of the Roman World”.

108 

T. BRUGHMANS

private enterprises dominated trade.10 He believes that Roman markets were integrated and strongly interconnected: I argue that the economy of the early Roman Empire was primarily a market economy. The parts of this economy located far from each other were not tied together as tightly as markets often are today, but they still functioned as part of a comprehensive Mediterranean market. There are two reasons why this conclusion is important. First, it brings the description of the Roman economy as a whole into accord with the fragmentary evidence we have about individual market transactions. Second, this synthetic view provides a platform on which to investigate further questions about the origins and eventual demise of the Roman economy and about conditions for the formation and preservation of markets in general.11

Temin argues that concepts from economics offer useful insights into the ancient economy and enable the formal representation and testing of hypotheses. The concepts of supply and demand are tools for understanding price-setting mechanisms and exchanges of individual commodities or services; New Institutional Economics offers an approach that helps to evaluate the operation of markets by focusing on the role of institutions; the concept of comparative advantage is suggested as a way to understand the economic interactions of regions; expensive information is introduced to reflect the costs involved in maintaining long-distance communication—costs which were reduced by being part of economic and social institutions; and asymmetric information refers to one party to a transaction having access to more information than the other.12 These concepts should be used in simple models that are necessarily abstractions of a complex reality, where a good model is distinguished from a bad model because it fits the available data better.13 In describing his hypotheses using this conceptual framework, Temin does something very few other scholars of the Roman economy do: his key statement (that the Roman economy was a well-functioning integrated market where prices are determined by supply and demand) is considered a theory consisting of multiple phenomena in interaction that each need to be tested individually through a series of hypotheses, some expected  Temin, The Roman Market Economy, 32.  Temin, The Roman Market Economy, 4. 12  Temin, “The Contribution of Economics”; Temin, The Roman Market Economy, 1–24. 13  Temin, The Roman Market Economy, 5. 10 11

4  EVALUATING THE POTENTIAL OF COMPUTATIONAL MODELLING… 

109

outcomes of which are unambiguously described (e.g. the correlation of price and distance), and the tools to enable testing are provided. Temin suggests two ways in which this theory can be tested: 1. “we can infer from the existence of prices that market exchange more closely describes the interaction containing the prices than reciprocity or redistribution.” 2. “people will behave instrumentally in market exchanges”; so by looking at the incentives people have to continue their particular behaviour one can evaluate the existence of market exchange.14 Temin performs the first type of test for the late Republican and Early Imperial period wheat trade: “I use wheat prices … to test the proposition that many wheat markets across the Mediterranean were interconnected and interdependent”.15 A series of assumptions is proposed to identify what experiment is necessary in order to evaluate his proposition: • Private enterprises dominated wheat trade. • “If there had been a unified wheat market, the main market would have been in Rome.” This is where the largest supplies and demands came together and where the price for wheat would have been set.16 • “Under these circumstances, wheat outside of Rome would be valued by what it was worth in Rome.”17 • Therefore the price outside Rome equals the price in Rome minus the costs involved in transporting it to Rome. • One would therefore expect a correlation between the price of wheat at a certain market and the distance to Rome. Alternatively, in the absence of an integrated market there would be only local prices determined by local conditions. If markets were not integrated then we would not expect a relationship between location and price. Temin collects six prices from different markets around the Mediterranean and with different dates, which are argued to be all the known prices that can be used in this experiment. He performs a regression  Temin, The Roman Market Economy, 8.  Temin, The Roman Market Economy, 29. 16  Temin, The Roman Market Economy, 36. 17  Temin, The Roman Market Economy, 36. 14 15

110 

T. BRUGHMANS

Fig. 4.1  Relationship between distance from Rome and distance discount of six documented grain prices (author’s own reconstruction of figure 2.2 in Temin, The Roman Market Economy, 43)

analysis which shows a strong correlation between as-the-crow-flies distance from Rome and price, seemingly confirming his hypothesis that the Roman wheat market in the late Republic and early Empire was integrated (Fig.  4.1).18 “The regressions confirm with very high probability that there was a unified wheat market that extended from one end to the other of the Mediterranean Sea. Transport costs were roughly proportional to distance, and the effects of distance were larger than the idiosyncratic influences of particular markets and places.”19

3   Critiques of Temin’s Approach There are some issues with these conclusions, as Temin himself acknowledges.20 The results of the experiment are misleading because they offer less proof to Temin’s theory than he is willing to admit. From a  Temin, The Roman Market Economy, 29–52.  Temin, The Roman Market Economy, 46. 20  Temin, The Roman Market Economy, 48–52. 18 19

4  EVALUATING THE POTENTIAL OF COMPUTATIONAL MODELLING… 

111

descriptive statistics point of view it seems that there is a correlation between price and location. However, with only six data points available it is impossible to derive a meaningful correlation score and statistically prove or reject Temin’s null-hypothesis. Temin acknowledges the dataset is small but argues this “does not affect the validity of the test”.21 Nevertheless, the correlation between price and distance from Rome is a key argument in Temin’s defence of the hypothesis that the Roman wheat market was integrated. The limited data leaves this hypothesis unproven for the late Republic and early Empire, despite Temin’s claims to the contrary. As stated above, Temin’s conclusions were also challenged by Bransbourg’s data analysis using twelve prices,22 and by Scheidel’s simulation study of transport costs. Scheidel argued that the use of as-the-­crowflies distances is inappropriate in testing the degree of Roman market integration and that simulations with more realistic physical transport frictions disagree with Temin’s conclusions.23 Temin’s hypothesis disagrees with the findings of Paul Erdkamp, who has argued against a strong integration of the wheat market and attributes a bigger role to government involvement in wheat trade.24 It was Erdkamp’s hypothesis Temin set out to disprove with his experiment. It is therefore no surprise that Erdkamp produced an extensive review of the strengths and shortcomings of Temin’s approach. Most crucially, Erdkamp argues that it is unnecessary to downplay state involvement in an argument for the existence of a market economy, since both state and private enterprises can function side by side. Temin interprets evidence for distribution and transportation as trade, which Erdkamp disagrees with. Moreover, it is argued that the annona is not the only non-market supply channel, Erdkamp mentions in particular the large-landowners who bring part of their produce to Rome to support their urban households and clientele: “the private market’s contribution to provisioning the empire’s capital was much smaller than Temin implies”.25 Both Erdkamp and Temin agree that Temin’s models are abstract simplifications of a complex past phenomenon, and that the nature, quantity and critique of the available data largely determine their success. The  Temin, The Roman Market Economy, 49.  Bransbourg “Rome and the Economic Integration of Empire”. 23  Scheidel “Shape of the Roman World”. 24  Erdkamp, The Grain Market in the Roman Empire. 25  Erdkamp, “How Modern Was the Market Economy of the Roman World”. 21 22

112 

T. BRUGHMANS

available historical data are clearly insufficient to evaluate the degree and mechanisms of integration of the Roman economy. The modelling approach is nevertheless promising, although not necessarily in the way Temin used it. His approach can be considered rather macroeconomic and deterministic in nature, where prices are considered to be in equilibrium, the data pattern studied is linear and the particular interactions of individuals each with their own motives have no place. Temin constructed the model like this on purpose and it is clear that interesting insights can be gained from it. However, such a model will never succeed in capturing the interesting variation and distinctive traits of Roman society that fascinates scholars of the Roman economy. Erdkamp argues that what is needed to solve the debate surrounding the Roman economy is “a more balanced and nuanced approach, allowing for the shades of grey that characterize all historic reality”.26 I argue that computational modelling will have an important role to play in achieving this aim, by simulating expected data patterns for the grey zone between competing theories and allowing for the flexibility to test a wide range of social, institutional, technological and ecological mechanisms on different scales of analysis. The remainder of this chapter will introduce a computational model that simulates the prices of goods current on markets as determined by rational agents with differing availability of commercial information and allows for comparison with the distance away from their place of production. It was decided to first explore the distance away from a place of tableware production rather than from Rome or any other major redistributive centre, since the latter introduces more assumptions: that all tableware distributions should be considered to have “piggy-backed” on distributions of primary products, and that Rome was the main market where the largest supplies and demands for tableware would come together, in addition to the assumption that the correlation between price and distance away from this market is an indication of an integrated economy. These additional assumptions would therefore require more experimentation than could be performed in the context of this chapter.

26  Erdkamp “How Modern Was the Market Economy of the Roman World,” 231 (online, p. 7).

4  EVALUATING THE POTENTIAL OF COMPUTATIONAL MODELLING… 

113

4   MERCURY as a Base-Model This chapter aims to illustrate the potential of computational modelling for studying the behaviour of different explanatory factors in isolation before integrating them in more complex models to make valuable contributions to ongoing debates on complex theories. It will do this by demonstrating how an existing computational model that was used in previous research to explore a specific explanatory mechanism (the degree of availability of commercial information to traders)27 can be extended with a new explanatory mechanism: transport cost for each transaction between markets. This is only possible because the behaviour of the model and the effect of the variables representing the availability of commercial information have been studied and are well understood. The effects of the new transport cost mechanism can therefore be distinguished through experimentation with a more complex computational model. MERCURY, after the Roman god of commerce, is an agent-based network model designed to formally represent and test selected explanatory factors (availability and reliability of commercial information) in the complex hypotheses proposed in Bang’s The Roman Bazaar and Temin’s The Roman Market Economy.28 It simulates the distribution of tableware by traders located on different markets and connected in a social network that allows for the flow of goods and commercial information. In previous experiments, the resulting simulated tableware distributions were compared with the excavated and published tableware evidence contained in the ICRATES database.29 In this chapter I will provide a brief non-­ technical description of the functioning of the model and the key findings of previous experiments. A full technical description and previous results along with the tableware dataset used for testing has been published,30 and the code of the original model is available through the open ABM repository.31

27  Brughmans and Poblome, “Roman Bazaar or Market Economy”; Brughmans and Poblome, “MERCURY”. 28  Bang, The Roman Bazaar; Temin, The Roman Market Economy. 29  Bes et al., Inventory of Crafts and Trade in the Roman East; Bes and Poblome. “(Not) See the Wood for the Trees”. 30  Brughmans and Poblome, “Bazaar or Market Economy?”; Brughmans and Poblome, “MERCURY”. 31  Brughmans, “MERCURY: An ABM of Tableware Trade”.

114 

T. BRUGHMANS

MERCURY is an agent-based network model programmed in Netlogo32: individual software agents form the primary scale of analysis; they are programmed with simple rules of behaviour and are connected in a social network. The model represents the structure of social networks between traders that act as the channels for the flow of commercial information and goods. When the model is initialised a social network structure is created between traders who are distributed among different sites. Four of the sites are production centres of four different tablewares, and traders located at these sites obtain a number of items of this locally produced ware in each turn. At each time step traders will determine the local demand for tableware they want to satisfy and will estimate the price they believe an item of tableware is worth based on their knowledge of the supply and demand of the traders they are connected to. Every item of tableware is then put up for sale, and pairs of traders who are connected in the network can buy or sell an item. When an item is successfully traded, the buyer will decide to either sell it to a local consumer (in which case the item is taken out of the trade system and is deposited at that site), or to store it for redistribution in the following turn in case this promises a higher profit. Over time, this model therefore gives rise to distributions of four tablewares. To test the probability of differing degrees of market integration, as implemented in MERCURY, we evaluated under what conditions the model could give rise to tableware distribution patterns similar to those found in the ICRATES dataset of tableware in the Roman East.33 Specifically, we were interested to know under what conditions one ware would be far more widely distributed than other wares, as was the case for the distribution of Eastern Sigillata A as compared to Eastern Sigillata B, C, and D during the late Hellenistic to early Imperial period in the Eastern Mediterranean. Comparing simulated distributions under different conditions with the known distribution of these four eastern tablewares allowed us to draw the following conclusions: equal numbers of traders at production sites, low degrees of social network links between markets, and limited availability of reliable commercial information from one’s direct neighbours in the network do not reproduce the observed tableware distributions in ICRATES.  We concluded that the emphasis on limited  Wilensky, “NetLogo”.  Bes et al, “ICRATES”; Inventory of Crafts and Trade in the Roman East, http://icrates. arts.kuleuven.be/icrates/network-analysis/webpages/icrates_mainframe.html (accessed 15-10-2019). 32 33

4  EVALUATING THE POTENTIAL OF COMPUTATIONAL MODELLING… 

115

integration of markets in Bang’s theory is unlikely under the conditions imposed in the model.

5   Extending MERCURY with a Transport Costs Variable Since the behaviour and effects of the variables in MERCURY are well understood, we can extend the model with an additional transport cost variable to explore the correlation between the price of tableware at markets and the distance away from their place of production. The code of the new model is openly accessible on the Open ABM repository.34 The transport cost variable can be set in experiments to any value between 0 and 1, where 1 is the maximum price of an item of tableware in the model. The procedure to determine whether to add this transport cost to a transaction is as follows: when a seller wishes to sell an item to one of his contacts in the social network, these contacts (potential buyers) will determine the price they believe an item of tableware is worth (by drawing on the commercial information available to them), and will reduce this price by the amount of the transport costs if and only if they are located on a different market than the seller (i.e. if transport of the item from one market to another is required, then a constant transport cost will be applied to the transaction). After this extension with a transport cost procedure, MERCURY’s trade procedure works as follows in every time step of the simulation (the following is pseudo-code, see the deposited model for the actual model code35): FOR EVERY item of tableware SELECT at random one of the traders with an item (the seller) > IF there are NO network neighbours of the seller that have a positive demand OR are willing to stock items for redistribution (i.e. potential buyers)   ADD the item to the seller’s stock > IF there ARE potential buyers  |Potential buyers determine whether a transport cost applies to the transaction  Seller identifies the potential buyer that offers the highest profit where: Transaction price = buyer price + transport cost  Brughmans, “MERCURY Extension: Transport-Cost”.  Brughmans, “MERCURY Extension: Transport-Cost”.

34 35

116 

T. BRUGHMANS

> IF the seller can make a profit or break-even   Seller sells item to buyer  Buyer stores the item in stock for redistribution if this promises a higher profit   If not,   the buyer sells the item to a consumer who deposits it at their site > IF the seller CANNOT make a profit or break-even   ADD the item to the seller’s stock Adding this transport cost variable and exploring its behaviour requires testing it in combination with different settings of all other variables, since the new variable can affect the behaviour of the existing variables in the model. Moreover, the experimentation presented here will reduce the number of tableware production sites from four to just one, in order to explore the correlation between distance and price for a single product before tackling the effect of having multiple products and production sites. The following section presents preliminary results of a set of experiments designed to explore the behaviour of this new variable.

6   Experiment Design MERCURY does not currently include geographical distances given its focus on testing network structures and not spatial structures. Distance is nevertheless present in MERCURY in two ways: all traders are located at markets that are considered to have different spatial locations, and transactions between traders on different markets can only take place if they are connected by an edge in the social network. This means that distance in MERCURY is currently measured in steps or hops over the network and not in geographical units. In future work, MERCURY can be extended to consider spatial distance more explicitly, first using as-the-­crow-flies distance as used by Temin and secondly by incorporating more realistic transport distances that affect transport costs as used by Scheidel.36 Due to the limited methodological aims of this chapter I only present experiment results using this abstract representation of distance. I will restrict experimentation by considering shortest network distance: the minimum number of steps one needs to make in the social network in order to connect a  Scheidel, “Shape of the Roman World”; Temin, The Roman Market Economy.

36

4  EVALUATING THE POTENTIAL OF COMPUTATIONAL MODELLING… 

117

pair of traders, which is also the minimum number of transactions needed in order to move an item of tableware from one trader to another. In light of the hypothesis this case study aims to represent, the experiment focuses on exploring the price for an item of tableware at increasing network distances to the production site. This will be explored in two ways: 1. For each successful transaction between a pair of traders, the minimum network distance from the buyer to any other trader at the production site (hereafter referred to as “distance from trader”. Bold lines Fig. 4.2). 2. For each successful transaction between a pair of traders, the minimum number of inter-site edges between any trader at the buyer’s site and any other trader at the production site, that is the number of transactions on which a long-distance transport cost is levied (hereafter referred to as “distance from site”. Dotted lines Fig. 4.2). A series of experiments were performed to explore how the phenomenon of interest (correlations of price with distance) is affected by the different processes in the computational model (so-called explanatory variables). The key processes tested in this way are the degree of market integration, the proportion of commercial information traders can access from their direct neighbourhood, and the transport cost. Because there are a number of random elements in the model, it was necessary to run each experiment multiple times: each experiment was run 100 times with exactly the same settings for the explanatory variables but the random numbers used differed.

Fig. 4.2  Abstract example of network distance. Nodes represent traders, lines represent social network edges, and large circles represent sites/settlements. When trader A successfully buys an item from trader B, the “distance from trader” to the production site is five whilst the “distance from site” to the production site is two

118 

T. BRUGHMANS

7   Results The results show a high degree of variation in transaction prices due to the randomness in the model. In fact, many of the simulated prices are not normally distributed but focus on a single extremely low or extremely high price, or they are bimodal. In such cases summary statistics (like the mean value) offer a misleading impression of the results. This ability to produce a variation of price distributions is in itself a highly interesting result: agent-based models with elements of randomness can allow scholars to explore differences in the behaviour and opportunities of traders. The variation reveals that tableware was not traded for one price at each market, but rather that price depended on how informed the particular buyer and seller of each transaction are. In many cases there is a dominant price for a market, but since prices are not always normally distributed we cannot assume the mean price represents this market price for tableware. The following general trends are apparent in the experiments (Fig. 4.3): • In all experiments, prices at the production site are lower than prices at distances away from the production site, although there is great variation. • Increasing transport cost, pushes up the maximum transaction price, limits the overall number of transactions and limits the distribution distance of goods. • A higher availability of reliable knowledge of the supply and demand of one’s neighbours in the social network leads to less extreme and more normally distributed prices both at “distances from trader” and at “distances from site”. A trend of increasing cost with increasing distance can be observed, especially for “distances from trader” but the average increase in price with increasing distance is higher than the transport cost. • Increasing the proportion of links between sites (the degree of market integration) results in similar price distributions but higher numbers of transactions at distances away from the production site despite transport costs of 0.01 and 0.1, but this pattern is less clear with a high transport cost of 0.5. • When no commercial information other than a trader’s own supply and demand are known, all transaction prices at the production site will be 0.5 and transactions at a distance will be 1.

4  EVALUATING THE POTENTIAL OF COMPUTATIONAL MODELLING… 

119

Fig. 4.3  Example results from one experiment with the following variable settings (variables in brackets): integration (proportion-inter-site-links) = 0.002; reliability information (local-knowledge)  =  1; inter-market transport cost (transport-cost)  =  0.01. Results for “distance from trader” shown on the left, results for “distance from site” shown on the right. These results illustrate some of the general trends across all simulation results: high variability of prices overall, generally lower price in production site, generally a correlation of price with distance although mean prices are hiding interesting variability

120 

T. BRUGHMANS

8   Discussion and Conclusion The results of the experiments presented here reveal an intuitive general trend and a less-intuitive high degree of variation. The general trend in the results suggests that there is a correlation between distance to the production site and the price of a transaction: under all tested conditions, the price of an item of tableware at the production site was lower on average than its price at other sites, where the price in most cases increases gradually with increasing network distance from trader and from site. However, this increase seems to be higher than the transport cost in most cases. Moreover, other than describing this general trend, no correlation value can be derived due to a high degree of variation in the results of iterations of the same experiment. These results therefore only confirm Temin’s theory to a limited extent: there is a correlation between price and distance, but it is an exponential increase of price with distance, and the price increase does not equal the transport cost. Importantly, this case study only considered distance from place of tableware production and not distance from the main wheat market as in Temin’s case. Moreover, this conclusion is strictly limited to the few explanatory variables tested in this computational model, and should be verified with a much wider range of variables in future work, crucially including geographical distance. These results are nevertheless informative, since they suggest that some correlation exists even in scenarios where very little to no commercial information is available to traders; that a correlation between the price of a good and the distance to the place of production will occur even in the case of a weakly integrated economy, but with rational agents following the laws of supply and demand applied to extremely limited reliable commercial information. However, this trend was most pronounced in experiments with a high degree of economic integration (high availability of commercial information and high proportion of links between markets) and a low transport cost. These computational simulation results reveal a rich multiscalar picture: they illustrate that both intuitive general macro-economic trends and highly localised variability can be identified in theorised economic behaviour. The interactions of hundreds of agents and the differences in variable settings between experiments lead to results that can only be demonstrated using computational simulation modelling. Some results may seem intuitive, like the correlation between price and distance from production centre, but proving that our intuition is correct in the absence of good

4  EVALUATING THE POTENTIAL OF COMPUTATIONAL MODELLING… 

121

empirical data can only be done through formal simulation modelling. Moreover, the variation revealed in these experiment results are not intuitive and are equally valuable as the general intuitive trend in light of our aim to map the grey zone between extreme hypotheses. I therefore argue these results highlight the interesting variability and the importance of idiosyncrasies of markets, in a way that the more deterministic experiment performed by Temin did not. Moreover, using agentbased modelling to test a range of variable settings enables one to explore the grey zone between extreme scenarios argued for by Erdkamp.37 For example, we have simulated price-distance correlations for varying degrees of economic integration. Further interpretation of these results will be the subject of a future wider substantive study and lies outside of the scope of the current chapter which aims to make a methodological argument. This case study aimed to illustrate how computational modelling can constructively contribute to ongoing debates on the Roman economy, by enabling the formal representation of aspects of theories, and by providing expected data patterns based on hypothetical assumptions even in the absence of empirical data. The formal nature of this approach has the benefit of enabling reproduction of results as well as facilitate communication and comparability of concepts: the definition and functioning of concepts and processes included in the tested theory will need to be explicitly coded in the computational model in order to make it work. The ability to simulate expected data patterns will enable scholars of the Roman economy to explore predictions of aspects of their theories, which has the potential to lead to the development of new theories or refinement of existing theories. However, the case study was also aimed at illustrating how the tested theories and related experiments need to be explicitly definable and include a very limited number of explanatory factors. Clearly this computational modelling approach should never aim to represent complex theories of the Roman economy in their full complexity. The sheer number of explanatory variables involved would make it nearly impossible to distinguish which variable causes the patterns of research interest, like a correlation between price and distance or the strong difference in tableware distributions. The precise contribution of computational modelling should be looked for in its ability to limit this complexity purely for analytical purposes: to explore selected aspects of complex theories in isolation so that 37  Erdkamp “How Modern Was the Market Economy of the Roman World,” p.  231 (online, p. 7).

122 

T. BRUGHMANS

their functioning can be better understood, before re-contextualising them with a wider range of contributing factors as well as archaeological and historical sources. This chapter has therefore sought to advocate computational modelling, not as a tool to be used in isolation, but rather as a new addition to the Roman scholar’s toolbox that has the potential to make constructive contributions to Roman economy studies when integrated within archaeological and historical research contexts.

Bibliography Bang, Peter Fibiger. The Roman Bazaar, a Comparative Study of Trade and Markets in a Tributary Empire. Cambridge: Cambridge university press, 2008. Bes, Philip, and Jeroen Poblome. “(Not) See the Wood for the Trees? 19,000+ Sherds of Tablewares and What We Can Do with Them.” In Rei Cretariae Romanae Fautores Acta 40, 505–14. Bonn, 2008. Bes, Philip, Rinse Willet, Jeroen Poblome, and Tom Brughmans. “Inventory of Crafts and Trade in the Roman East (ICRATES): Database of Tableware”, 2018. https://doi.org/10.5284/1050900. Bransbourg, Gilles. “Rome and the Economic Integration of Empire.” ISAW Papers 3 (2012): http://doi.org/2333.1/280gb73f. Brughmans, Tom. “MERCURY Extension: Transport-Cost.” CoMSES Computational Model Library. Accessed 15-10-2019, 2018. https://www. comses.net/codebases/d67fd7ce-a6df-4d25-b10c-765b455b80f0/ releases/1.0.0/. Brughmans, Tom. “MERCURY: An ABM of Tableware Trade in the Roman East.” CoMSES Computational Model Library. Accessed 15-10-2019, 2015. https://www.comses.net/codebases/4347/releases/1.1.0/. Brughmans, Tom, and Jeroen Poblome. “Roman Bazaar or Market Economy? Explaining Tableware Distributions through Computational Modelling.” Antiquity 90, no. 350 (2016a): 393–408. https://doi.org/10.15184/ aqy.2016.35. Brughmans, Tom, and Jeroen Poblome. “MERCURY: An Agent-Based Model of Tableware Trade in the Roman East.” Journal of Artificial Societies and Social Simulation 19, no. 1 (2016b): http://jasss.soc.surrey.ac.uk/19/1/3.html. Brughmans, Tom, John William Hanson, Matthew J. Mandich, Iza Romanowska, Xavier Rubio-Campillo, Simon Carrignon, Stephen Collins-Elliott, Crawford Katherine, Daems Dries, Fulminante Francesca, Haas de Tymon, Kelly Paul, Moreno Escobar Maria del Carmen, Paliou Eleftheria, Ritondale Manuela. “Formal Modelling Approaches to Complexity Science in Roman Studies: A Manifesto.” Theoretical Roman Archaeology Journal 2 (2019): 1–19. https:// doi.org/10.16995/traj.367.

4  EVALUATING THE POTENTIAL OF COMPUTATIONAL MODELLING… 

123

Erdkamp, Paul. The Grain Market in the Roman Empire: A Social, Political and Economic Study. Cambridge University Press, 2005. Erdkamp, Paul. “How Modern Was the Market Economy of the Roman World?” Oeconomia. History, Methodology, Philosophy 3 (1 June 2014): 225–35. Accessed 15-10-2019, 2014. http://journals.openedition.org/oeconomia/399. Graham, Shawn and Scott Weingart. “The Equifinality of Archaeological Networks: An Agent Based Exploratory Lab Approach.” Journal of Archaeological Method and Theory 22 (2015): 248–74. https://doi.org/10.1007/s10816-014-9230-y. Kessler, David, and Peter Temin. “Money and Prices in the Early Roman Empire”. In The Monetary Systems of the Greeks and Romans, edited by William V. Harris, 137–59. Oxford: Oxford University Press, 2008. Scheidel, Walter. “The Shape of the Roman World: Modelling Imperial Connectivity.” Journal of Roman Archaeology 27 (2014): 7–32. Scheidel, Walter. The Cambridge Companion to the Roman Economy. Cambridge: Cambridge University Press, 2012. Scheidel, Walter, Ian Morris, and Richard P.  Saller. The Cambridge Economic History of the Greco-Roman World. Cambridge: Cambridge University Press, 2007. Temin, Peter. “The Contribution of Economics.” In The Cambridge Companion to the Roman Economy, edited by Walter Scheidel, 45–70. Cambridge: Cambridge University Press, 2012. Temin, Peter. The Roman Market Economy. Princeton: Princeton University Press, 2013. Verhagen, Philip, Jamie Joyce, and Mark R. Groenhuijzen. Finding the Limits of the Limes: Modelling Demography, Economy and Transport on the Edge of the Roman Empire. Springer, 2019. https://doi.org/10.1007/978-3-03004576-0_1. Wilensky, Uri. “NetLogo”, Center for Connected Learning and Computer-Based Modeling, Northwestern University, Evanston, IL. Accessed 15-10-2019, 1999. https://ccl.northwestern.edu/netlogo/. Wilson, Andrew, Morris Silver, Peter Fibiger Bang, Paul Erdkamp, and Neville Morley. “A Forum on Trade.” In The Cambridge Companion to the Roman Economy, edited by Walter Scheidel, 287–317. Cambridge: Cambridge University Press, 2012.

CHAPTER 5

Visualising Roman Institutional Environments for Exchange as a Complex System Merav Haklai

1   Introduction During the second half of the twentieth century New Institutional Economics (NIE) gained accelerated acceptability among economists. NIE puts institutions at the centre of economic inquiry, seeing them as “both informal constraints (sanctions, taboos, customs, traditions, and codes of conduct), and formal rules (constitutions, laws, property rights).”1 Following economics, since the 1990s ancient historians have willingly applied NIE terminology in their studies, constructing their work in terms of an inquiry into ancient institutional economic environments. This paper combines an institutional approach with a recently emerging approach in economics, that of complexity economics. The paper advocates the

1

 North, “Institutions,” 97.

M. Haklai (*) Ben-Gurion University of the Negev, Beer-Sheva, Israel e-mail: [email protected] © The Author(s) 2021 K. Verboven (ed.), Complexity Economics, Palgrave Studies in Ancient Economies, https://doi.org/10.1007/978-3-030-47898-8_5

125

126 

M. HAKLAI

advantages of using complexity economics and applies complexity-inspired visualisations to analyse Roman legal environments during the High Empire, specifically the institutional framework within which money operated economically in the private sphere. When looking at the legal environment of economic interaction in the Roman empire, one finds similar institutions used across a spectrum of economic activities. Traditional categories, such as sale, hire, etc., often interact with one another and overlap. Similar pecuniary devices appear in various types of transactions. For example, paying monetary arra—that is, earnest money given ahead as guarantee or first instalment—may be found in both emptio venditio contracts and locatio conductio ones; credit is relevant not only for lending operations but also for “Sale on Delivery” transactions; interest rates (faenus/usura) are referred to and used in a much wider spectrum of activities than just narrowly defined loans, and serve as both remuneration and penalising instrument. The same concepts and terminology were applied to describe the apparatus of transactions whether these were facilitated via money or conducted in kind (both arra and usura serve as good examples). Patterns of thought which directed sophisticated credit apparatus, and which can be detected in banking operations as well as in elite finance,2 can also be traced in agricultural management and in village-based economic environment, where credit facilitated transactions which were conceptualised in monetary units though often conducted in kind.3 All of these duplications, interconnections, and overlapping usages make neoclassical economic modelling techniques—with their aim to oversimplify reality, usually in order to stress certain aspects of it—less helpful to represent visually the historical evidence of the Roman world. Traditionally, models and their visual representations are perceived as aimed at providing clear, hence often simplified, explanations of complicated realities.4 With complexity economics the simplifying role of models is pushed aside. One departs the safe waters of traditional neoclassical economics, leaving behind its unrealistic assumptions, so 2  Andreau, La vie financière dans le monde romain; Andreau, Banking and Business in the Roman World; Rathbone and Temin, “Financial Intervention in first century AD Rome and Eighteenth Century England”. 3  Rathbone, Economic Rationalism and Rural Society in Third-Century A.D. Egypt; Kehoe, Investment, Profit, and Tenancy; Kehoe, Law and the Rural Economy in the Roman Empire. 4  As put by Snowdon and Vane, Modern Macroeconomics, 4: “Theories, by definition, are simplifications of reality. This must be so given the complexity of the real world.”

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

127

helpful for model construction; its straightforward diagrams, usually in two or three dimensions; and its easily read graphs, the reassuring totems of economics.5 A complexity approach stresses the ability of a system to allow new patterns of activity to be created by individuals who self-adjust their use of it, altering and widening its application in accord with their own needs and the continuously evolving system in which they operate; thereby allowing the system to become more than the sum of its components. The abstract complexity-oriented visualisations here proposed, are not projections of a database. Rather, they offer a conceptual framework, which allows a wider spectrum of possibilities to represent multi-linkages, repetition, emergence, and loop and quasi-loop formations. These are later translated back into a verbal, narrative format to improve our understanding of the historical evidence. Thus, graphical representation stresses the advantages of analysing economic phenomena as complex structures. Since the visual product of complexity-economic models is somewhat counterintuitive to mainstream economic linear model-building, in what follows I give some clarifications regarding complexity and economics, and discuss several examples of visualised complexity. Afterward some comments are made on previous visualisations of ancient economies, and then complexity-oriented visualisations are offered to analyse institutional aspects of the historical example of the Roman empire.

2   Complexity and Economics Developed in the natural and physical sciences, complexity, chaos, and network theories have become an integral part of contemporary intellectual landscape,6 extending their influence also to the social sciences.7 Complexity theories see the world as composed of many systems rather than one, which operate side by side in different levels of structure.8 However, it should be made clear that complicated is not necessarily 5  To use the metaphor of Leijonhufvud, “Life among the Econ”, who wrote a spoof anthropological article about the two castes within the tribe of the Econ, the Micro-caste and the Macro-caste, each with its own totem, which look remarkably similar to each other, that is the supply-demand and the LM-IS curves crossing one another in the point of equilibrium. 6  Weaver, “Science and Complexity”, who offers a good starting point. 7   Wible, “What is Complexity?”; Prasch, “Complexity and Economic Method”; Montgomery, “Complexity Theory”. 8  Nicolis and Prigogine, Exploring Complexity, 2–6.

128 

M. HAKLAI

complex. The fact that a system is comprised of many variables, or that it ­displays complications, does not by itself makes it a case of complexity. Rather, “the principal distinguishing feature of any complex system is that it is fundamentally irreducible—it is more than the sum of its component parts.”9 Complex systems are defined as self-organising systems, in which new patterns of activity are created by the entities operating within the system.10 They are dynamic, self-adjusting systems, which endure and evolve over time, hence, tend to be structured in a sustained non-­ equilibrium pattern of evolution.11 It is not enough that the elements of a system be interconnected in a non-simple manner; they need also display a capacity for self-adjustment. In complex systems it is not only the elements comprising a system but also the system as a whole, which is capable of dynamic self-organising diversity. It is the nonlinearity which defines the system’s development, and which encourages, in fact, determines, the system’s capacity for emergence. In economics, the complexity perspective is a relatively new approach that began in the 1980s and to a greater extent during the 1990s,12 promoted especially by the Santa Fe Institute (SFI).13 It incorporates approaches which began as heterodox economics14—for example evolutionary economics,15 path dependency,16 and dynamic economics17— within a broader complexity perspective.18 In economics the complexity perspective challenges some of the cornerstones of mainstream neoclassical economics. It rejects the latter’s dependency on equilibrium analysis,

 Van Der Leeuw and McGlade, “Archaeology and Non-linear Dynamics,” 14.  Nicolis and Prigogine, Exploring Complexity, 25–7; Wible, “What is complexity?,” 17–8. 11  Wible, “What is complexity?,” 18–9. 12  On features of complexity advocated by Austrian economics, see below. 13  http://www.santafe.edu. Anderson, Arrow, and Pines, The Economy as an Evolving Complex System; Arthur, Durlauf, and Lane, The Economy as an Evolving Complex System; Wible, “What is Complexity?”. 14  Labelled “The changing face of Economics” by Colander, Holt, and Rosser, The Changing Face of Economics. 15  For instance Lindgren, “Evolutionary Dynamics in Game-Theoretic Models”; Metcalfe and Foster, Evolution and Economic Complexity. 16  For example, North, “Some Fundamental Puzzles in Economic History/Development”; Prasch, “Complexity and Economic Method,” 220–2. 17  For example, Brock, “Asset Price Behavior in Complex Environments”; Kirman, “The Economy as an Interactive System”. 18  Arthur, Durlauf, and Lane, “Introduction”. 9

10

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

129

replacing it with multi-equilibria or disequilibrium analyses.19 It casts off the linear structure of mainstream economic model building, stressing the essential non-linearity of economic phenomena,20 and applying adaptive and dynamic nonlinear mathematics in economic model building.21 Rather than attributing the decision-making mechanism to rational expectations, complexity economics emphasises the induction process of expectations formation, arguing that agents keep adjusting their expectations to an ever evolving economic environment, and stressing the network-based process by which agents acquire information and make choices.22 Instead of considering an aggregate outcome to be a multiplication of the average agent, whose behaviour is perceived as representative of all other agents, complexity economics views an aggregate outcome as more than the sum of the factors which comprise it. That is, aggregate performance involves a behavioural change, which is an outcome of the significant interactions occurring between economic agents.23 Coming from outside the social sciences, Warren Weaver, one of the most influential twentieth-century mathematicians, in his now-classical article from 1948 titled “Science and complexity,” offered to view economic phenomena as “organised complexity”: “On what does the price of wheat depend? This too is a problem of organized complexity. A very substantial number of relevant variables is involved here, and they are all interrelated in a complicated, but nevertheless not in helter-skelter, fashion.”24 Weaver surely was not the first to think of economics in terms of complexity theory. Features of complexity have been advocated in economics prior to the 1980s within the Austrian tradition of economic thought,25 especially by the Austrian economist and philosopher Friedrich

19   Arthur, “Positive Feedbacks in the Economy”; Arthur, Durlauf, and Lane, “Introduction,” 4–6. 20  Arthur, Durlauf, and Lane, “Introduction,” 2–5. 21   Day, “Complex Economic Dynamics”, 14; Montgomery, “Complexity Theory”; Shapiro, “A Hitchhiker’s Guide to the Techniques of Adaptive Nonlinear Models”; Velupialli, “Non-Linear Dynamics, Complexity and Randomness”. 22  Arthur, Durlauf, and Lane, The Economy as an Evolving Complex System; Lane, “Is What Is Good for Each Best for All”; Prasch, “Complexity and Economic Method”. 23  Day, “Complex Economic Dynamics”; Arthur, Durlauf, and Lane, “Introduction,” 5–11; Krugman, “How the Economy Organises Itself in Space,” 241–3. 24  Weaver, “Science and Complexity,” 539. 25  Vaughn, “Hayek’s Implicit Economics”; Koppl, “Complexity and Austrian Economics”, who even attributes a proto-complexity approach to the founding father of the Austrian

130 

M. HAKLAI

von Hayek (1889–1992).26 Hayek held the formation of a market economy to be an outcome of “spontaneous order,” a self-organising and constantly evolving process in which structures are formed by the combination of self-interested individuals, each pursuing her or his own wants, without intentionally trying to achieve order.27 Consequently, exchange, production, and distribution are incorporated within an economic system in ways that no individual mind could ever plan.28 Thus, institutions—whether these refer to The Market, Law, Culture, or Language—are “the result of human actions, but not the execution of human design.”29 Hayek’s views did not achieve wide acceptance among mainstream economics of his time; but nowadays, with the rise of complexity economics, they have become an integral part of economic discussion.30 However, there are distinctions between Austrian economics and the new complexity approach in economics.31 Unlike the Austrian school, which supports a verbal, non-mathematical approach to economic analyses, most of the economists advocating complexity economics do not reject mathematics as a methodological tool.32 Far from it. They point out difficulties arising from the linear mathematics traditionally applied in mainstream neoclassical economics,33 and instead habitually apply

economic school, Carl Menger, seeing him as a predecessor of the “spontaneous order” idea; J. Barkley Rosser, “How complex are the Austrians?”. 26  Hayek, Individualism and Economic Order; id., “The Theory of Complex Phenomena”. On Hayek’s complexity ideas, see Chaumont-Chancelier, “Hayek’s Complexity”; Vaughn, “Hayek’s Implicit Economics”; Wible, “What is complexity?,” 19–23; Caldwell, Hayek’s challenge, 323–82; Gaus, “Hayek on the evolution of society and mind”, 232–58. 27   Vaughn, “Hayek’s Implicit Economics;” Montgomery, “Complexity Theory: An Austrian perspective,” 228–9; Rosser, “How complex are the Austrians?”. 28  Hayek, “The Theory of Complex Phenomena”; Wible, “What is complexity?,” 19–23; Montgomery, “Complexity Theory: An Austrian perspective”; Rosser, “How Complex are the Austrians?”. 29  Adam Ferguson, The History of Economic Thought (1767), quoted in Hayek, The Constitution of Liberty, 57. 30  For example, Krugman, The Self-Organizing Economy. 31  Contra, Koppl, “Complexity and Austrian Economics”. On Austrian vs. Complexity economics, see Montgomery, “Complexity Theory”; Rosser, “How Complex are the Austrians?”. 32  Montgomery, “Complexity Theory”. 33  Arthur, Durlauf, and Lane, “Introduction,” 3.

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

131

dynamic nonlinear mathematics as a method of modelling.34 The mathematics used for building these complexity economic models is often beyond the capacity of the average educated person and is intended for specialists. The kind of extensive and detailed sets of data casted into these models usually are beyond the reach of ancient historians. Yet, the visual outcome of these complexity economic models, as detailed below, can offer a methodological tool for analysing historical economic patterns of behaviour.

3   Visualising Complexity Visualisation, according to the Oxford English Dictionary is the process of forming a mental picture of something that is actually invisible. Yet, it is not only an internal construct of the mind; visualisation refers also to the graphical representation of concepts or data that have no visual dimension. As such, it becomes an external artefact which functions as a cognitive tool. By producing a visual image that condenses massive amounts of data or communicates a message, whether abstract or concrete, visualisation supports the comprehension of compound and sophisticated ideas, therefore, may assist in forming hypotheses.35 One of the oldest formats for displaying information graphically are tree-structured models.36 Tree Models are very helpful in displaying hierarchical and chronological connections, but have little, if any, flexibility for accommodating recurrence, duplication, repetition, reciprocity, or any other circulatory-natured phenomenon. In tree-structured representations the explanatory narrative is generally that of “one thing which leads to another” in the sense of a descriptive mechanical causality,37 with very little room for circular occurrences.38 Similar observations can be made regarding many of the standard graphs used today: bar charts, pie charts, 34  Arthur, Durlauf, and Lane, “Introduction,” 3–4; Prasch, “Complexity and Economic Method,” 216. There is another prominent distinction: while economists affiliated with the new complexity perspective tend to advocate market-intervention policies, Austrian economics usually strongly rejects government intervention. 35  Ware, Information Visualization, 2–4. 36  Lima, Visual Complexity, who offers a short history of Tree-Models, further elaborated in Lima, The Book of Trees. 37  Van der Leeuw and McGlade, “Archaeology and Non-Linear Dynamics,” 13. 38  For criticism on Tree Models, see Lima, Visual Complexity, 43–69.

132 

M. HAKLAI

scatter plots, line charts, and so on. All derive from nineteenth-century techniques for graphic representation, which fit the scientific paradigm of reduction,39 that is, “breaking nature down into the simplest possible elements and defining rules on how these elements interact.”40 By contrast, much of the new network- and complexity-oriented methods of visualisation that became ever more frequent in the last three decades possess better capacity for displaying circular or overlapping formations that characterise complex phenomena. These methods tend to produce complex graphical designs, which nestle on network theory. They are often based on computerisation, statistics, and huge amounts of data and can depict an astonishing range of interconnectedness expressed by an intense arrangement of nodes and links that produce compound, often condensed, visual formations.41 The new types of visualisation may imitate molecular constructions, either centralised or circular42; may depict flows, visually assimilating either waves43 or arcs44; or reproduce visual patterns which resemble those apparent in the natural sciences and the biological world, such as cell and bacteria formations.45 When it comes to visualisation of complexity-oriented economic analyses, in particular those which concern money and financial markets, as far as I am aware, econophysics represents the forefront of research.46 Econophysics is a newly emerging field,47 which applies mathematical methods used in physics, and especially statistical physics, for the analysis of socioeconomic complex systems.48 It aims at describing “socioeconomic systems as complex systems suggesting the unavoidable result of bringing together numerous components in a non-simple manner,”49 while avoiding a priori assumptions concerning agents commonly used in 39  On nineteenth-century scientific paradigm of thought as one which concerns “simple problems,” see Weaver, “Science and complexity.” 40  Manovich, “Introduction,” 13. 41  Lima, Visual Complexity, 97, 158, who offers a sort of a manual for possible schemata. 42  Ibid., 98–157, who gives examples for network-based visualisations. 43  Ibid., 188–91. 44  Ibid., 160–3. 45  Ibid., 164–6. 46  Econophysics often is concerned with financial markets; for criticism, see Casti, “Money is Funny”. 47  Jovanovic and Schinckus, “The Emergence of Econophysics”. 48   Schinckus, “Between Complexity of Modelling and Modelling of Complexity”; Jovanovic and Schinckus, “The Emergence of Econophysics.” 49  Schinckus, “Between Complexity of Modelling and Modelling of Complexity,” 3655.

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

133

mainstream economics, such as, rationality, personal utility function, and risk aversion.50 Instead, econophysics bases its methodology on empirical verification,51 with the main objective of describing past phenomena through models based on historical analyses of empirical data and speculating about probability of future events.52 Although econophysics rests on mathematics which is beyond the capacity of most historians (myself included), and uses a volume of detailed data which can only be dreamt of by ancient historians, still, its visual product may serve as a helpful methodological tool. For example, let us look at the graphical representation of one such econophysic model (Fig. 5.1), which offers a tool for filtering information in complex systems, and applies it in an analysis of the one-hundred most capitalised stocks traded in the US equity markets during the years 1995–1998. Again, neither the mathematics nor the database concerns us here. Rather, focus is on the graphical representation of the model; what its authors call the Planar Maximally Filtered Graph “obtained from the fully connected graph associated with the correlation coefficient matrix of the data.”53 The diagram displays cross-correlation computed by using daily returns of stocks indicated by their Ticker Symbol, which is an abbreviation used to identify publicly traded stocks. The correlation provides a similarity measure among the behaviour of different elements in the system,54 although the link lengths do not reflect the value of the similarity measure between vertices.55 The computation of cross-correlation is displayed on a topological planar, meaning one that can be drawn on a plane without edge-crossings. Such a type of visualisation has been labelled “Organic Rhizome,”56 relying on the philosophical concept of a rhizome developed by the French philosophers Gilles Deleuze and Félix Guattari.57  Ibid.   Schinckus, “Between Complexity of Modelling and Modelling of Complexity”; Jovanovic and Schinckus, “The Emergence of Econophysics,” sec. III.2, 19–20, in the online version. 52  With the stationary hypothesis according to which future data will be a statistical reflection of past data; Schinckus, “Between Complexity of Modelling and Modelling of Complexity,” 3656, 3660, 3661 n. 26. 53  Tumminello et al., “A Tool for Filtering Information in Complex Systems,” 10423. 54  Ibid., 10421. 55  Ibid., 10423. 56  Lima, Visual Complexity, 158, 192–5. 57  Deleuze and Guattari, A Thousand Plateaus, 21; cited in Lima, Visual Complexity, 44. 50 51

134 

M. HAKLAI

Fig. 5.1  The Planar Maximally Filtered Graph for stocks traded in the US equity markets (1995–1998), Tumminello et  al., “A Tool for Filtering Information in Complex Systems,” 10,423, Fig.  2. (Copyright (2005) National Academy of Sciences, U.S.A.)

Though the authors of the econophysic study cited above do not mention Deleuze’s and Guattari’s notion of a rhizome, the visualisation that their model produced displays precisely such a formation; as do many other complexity oriented visualisations. In fact, this type of visualisation does not require a detailed and extensive numerical database. It can also depict a theoretic, qualitative approach to the relations and interconnections of factors within a complex system. For example, let us consider the visual product of a research on obesity funded by the UK government, which offers a qualitative study for mapping the complex systemic structure of obesity (Figs. 5.2a and 5.2b).58 Unlike the stocks-equity-market econophysic model, which creates a graphic projection of detailed data  Vandenbroeck, Goossens, and Clemens, Tackling Obesities; Lima, Visual Complexity, 194.

58

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

135

Fig. 5.2a  A systemic structure of obesity, foresight. (Vandenbroeck, Goossens, and Clemens, Tackling Obesities, 76, Fig. 12) (for an interactive map, with separate view of the nine groups see http://www.shiftn.com/obesity/Full-Map.html)

(Fig. 5.1), the obesity model is less database oriented in the sense that it is not statistically driven. Its authors chose the visualised format of a causal loop diagram,59 which is a model based on system dynamics methodology developed for strategic decision making.60 Causal loop diagrams are not predictive models; their essential contribution is in summarising and communicating trends, relationships, and constraints that influence the behaviour of complex systems by displaying interdependencies within a set of relevant and causally linked variables, which can be scaled at various levels of aggregation (e.g., individual, group, society).61 The visual product of the obesity model includes nine groups of factors identified as contributing to the obesity epidemic, which are organised  Lima, Visual Complexity, 158, 194, who categorises this model as “organic rhizome.”  Vennix, Group Model Building, 3–7 (on the assumptions of system dynamics methodology), 52–60 (on casual loop models). 61  Vandenbroeck, Goossens, and Clemens, Tackling Obesities, 3–6. 59 60

136 

M. HAKLAI

Fig. 5.2b  A systemic structure of obesity, linkages between psychology and food environment (Vandenbroeck, Goossens, and Clemens. Tackling obesities: “Map 15—Full generic map: linkages between the psychology and food environment areas—Linkages between these two areas are predictably dense, with 14 arrows going from psychology to the food environment and nine arrows in the other direction: (*) Key tail variables in the psychology area are education, media availability, socio-cultural valuation of food, perceived lack of time, stress and food literacy. They link into demand-side factors such as demand for health and the social pressure to consume. But supply-side variables are also triggered: the food industry’s business model is grafted onto what people want. (*) Tail variables in the food environment area are dispersed, with one arrow only leaving each of the variables. They include food exposure, food abundance, social pressure to consume, and industry’s desire to maximise volume. They drive three variables in the psychology area: exposure to food advertising, perceived lack of time, and psychological ambivalence”

according to thematic clusters and displayed via a causal loop diagram. In causal loop models, the system’s elements are the variables and are represented by boxes, while the causal relationships between variables are represented by arrows. Any two variables may be connected by either linear

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

137

or circular causalities. In linear causalities, the variable at the tail of the arrow affects the variable at the point in either a positive manner, with both variables changing in the same direction (displayed in the obesity diagram with arrowed lines), or a negative one, with variables changing in opposite directions (displayed in the obesity diagram with lines that end with a square). Causalities may differ also in strength and the time delay to which they are subject. Circular causalities (from A to B to A) produce “feedback-loops” and are an important feature of causal loop modelling since they help to explain the dynamic behaviour of the system. Lastly, the essential aim of causal loop models is explanatory rather than predictive. It aims to gain insight concerning the underlying structure of a complex situation, with the system map showing how variables interrelate.62 Thus, for example, in the obesity model one may consider linkages between psychological elements and food environment (Fig. 5.2b), and the visualisation can be verbalised (as in the caption below Fig. 5.2b below). Whereas the visual products of the stocks-equity-market econophysic model (Fig. 5.1) and the obesity causal loop model (Figs. 5.2a and 5.2b) portray multiple nodes connected by flexible curves, which may trigger an association of multi-tentacle octopuses randomly interacting with one another, a different choice of graphics may create different associations, yet result in the same conceptual outcome. Let us look at a third example, which applies visual complexity in a network model for Systems Biology of Human Aging (Fig. 5.3). This representation is both extremely colourful and intensely detailed, whilst nevertheless creating an impression of a mechanical apparatus, probably due to the use of quadrangles and other polygons instead of curved lines. Yet, the difference is due to a choice of graphics, and is mostly aesthetic rather than essential. In many respects, the visual product of the Systems Biology of Human Aging Model is similar to that of the Obesity System Map (Figs. 5.2a and 5.2b). As regards to methodology, there is no essential difference; all visualisations (Figs. 5.1, 5.2a, 5.2b, and 5.3) use an abstract plane—whether topological planar (Fig.  5.1), or not (Figs.  5.2a, 5.2b, and 5.3)—to display rhizomatic patterns of complex non-centralised phenomena, where the system’s elements interrelate with one another in non-simple manners which allow emergence; with “emergence” here referring to the capacity of a fundamentally chaotic or complex system to produce patterns that are seemingly non-random and predictable in behaviour. Although the precise  Ibid., 4–6.

62

138 

M. HAKLAI

Fig. 5.3  Systems Biology of Human Aging. (Furber, Systems Biology of Human Aging, available at http://www.LegendaryPharma.com/chartbg.html; appearing also in Lima, Visual Complexity, online electronic database: http://www.visualcomplexity.com/vc/project_details.cfm?id=521&index=40&domain=Biology)

initial conditions that trigger individual patterns within the system cannot be identified or used to predict the details of an outcome, the system as a whole does produce some emergence patterns, which include recognisable formations that can be used to describe, and perhaps even to predict, the overall behaviour of the system.63 Keeping in mind these three examples (Figs. 5.1, 5.2a, 5.2b, and 5.3), let us turn to ancient history.

4   Visualising Ancient Economies Within the field of ancient economic history scholars have tended to prefer verbal descriptions rather than graphical ones. Yet, visualisation is an inherent, though at times implicit, component of their analyses. Historians often map trade routes, envision “economic space,”64 or portray an image  Holland, Emergence; but see Corning, “The re-emergence of emergence”.  For example, Finley, The Ancient Economy, 177–8, who speaks of a “large unified economic space.” 63 64

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

139

such as “a cellular economy”65; all of which are metaphors based on an “intuition that vision and artful images are an alternate and seemingly direct route to insight.”66 A survey of graphic modelling of Greek and Roman economies was offered more than a decade ago by John K. Davies, who pointed out that visualisations of ancient economies tend to concentrate on the sort of space which economic interactions occupy.67 Thus, models are often concerned with boundaries, both physical and conceptual, and are frequently aimed at tracing flows within a spatial structure, either a geographic or a mental one; flows which usually signify the exchange of goods and services. Keith Hopkins’ monumental “Taxes and Trade” model for taxation and coinage flows in the Roman empire remains one of the most insightful and influential examples of modelling of this sort.68 Davies displayed this model graphically with a simplified diagram of three circles, one within the other, symbolising geographic space divided into regions, which derive their significance from political spheres: “centre,” “middle zone,” and “periphery” or “frontier” (Fig. 5.4). This simplified representation highlights an important generalising understanding concerning the role of the Roman state in generating economic relationship(s) between different regions of the empire, as well as encouraging monetisation. Its influential contribution lies in its powerful simplicity: it displays in a nutshell complicated political, economic, and social dynamics in the Roman empire that existed and developed for decades. However, the simplified diagram also has its drawbacks, as it leaves an important part of the model visually unexpressed, hence, unexplained; namely, the process by which money so-to-­ speak “trickles down” from the frontiers back to the middle-zone,69 as well as the economic and monetary relations between and within provinces.70 65  Duncan-Jones, Structure and Scale in the Roman Economy, 44. Also, Hopkins, “Taxes and Trade,” 104, who mentions “cellular autarky of individual peasant farmers”; Davies, “Linear and Nonlinear Flow Models for Ancient Economies,” 152–3, who offers cellularstructured visualisation for ancient economies. 66  Ware, Information Visualization, xvii. 67  Davies, “Linear and Nonlinear Flow Models,” 130–41, 152, who gives a survey of the relevant literature. 68  Hopkins, “Taxes and Trade”. 69  What Davies, “Linear and Nonlinear Flow Models,” 139, calls “the “return” (real or symbolic, full or inadequate) from the centre and periphery to the middle zone.” 70  For example, Andreau, Banking and Business in the Roman World, 135.

140 

M. HAKLAI

Fig. 5.4  Taxation flow in the Roman empire inspired by Hopkins’ “Taxes and Trade” model. A revised version of Davies, “Linear and Nonlinear Flow Models,” 140, Fig.  6.6. The (here added) cursive lines represent how money “trickles-­ down” from the frontiers back to the middle-zone via economic transactions

How can one model a process? Twenty years ago, Davies answered this question by innovatively adopting a dynamic, complexity-oriented approach for modelling ancient economic conditions.71 Davies analysed the economies of large, socially complex classical Greek city-states, by constructing nonlinear flow models that assist in conceptualising both internal and external economic dynamics for a variety of fifth- and fourth-century BCE Greek polities. Non-linear dynamics has been advanced as a modelling methodology in archaeology since the 1990s.72 However, until recently it was hardly ever applied as a prime methodological tool by ancient economic historians.73 Davies chose to portray a “processual or topological space,”74 rather than a geographical one, to trace flows, including nonphysical ones, and the structure within which they occurred.75 His first attempt at such a visualisation was published in 1998 with regard to 71  Davies, “Ancient Economies: Models and Muddles”; Davies, “Linear and Nonlinear Flow Models”. 72  Van Der Leeuw and McGlade, Time, process, and structured transformation in archaeology. 73  For example, Seland, “Here, There and Everywhere.” 74  Davies, “Linear and Nonlinear Flow Models,” 142. 75  Ibid., 140–2.

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

141

classical Athens,76 and in 2005 the model was elaborated and expanded to include other formations of ancient Greek polities.77 These publications offer complex visualisations for “modelling the salient characteristics of the economic behaviour of antiquity when we have no quantitative data worth the name,” and advocate “modelling in terms of process and of the flow of resources and return on a single surface within topological space.”78 Exact replica of Davies’ flow charts cannot be provided here. Instead, I offer a thinner modified version of his model to give readers who cannot consult Davies’ publication a sense of the graphical outcome of his complex analysis (Fig. 5.5). The visual product of Davies’ analysis consists of intense diagrams composed of multi-directional connections and counter-linkages, in which conceptual formats are represented by four major schemata: First, bandwidth, which represents the relative sizes of flows. Second, motors, which indicated the “driving forces,” meaning the internal incentives that drive resources and exchange. Third, gates, which denote impediments to flows and exchange of resources, goods, and services (e.g., level of technology, social values, or constraints imposed by a polity). Fourth, reservoirs, that is “reserves,” the accumulation of resources by the polity and in private or cultic contexts.79 All of these are portrayed on an abstract plane to depict the mechanism by which economic interaction occurred. The visual products of Davies’ analysis display organised complexity, which aims at tracing how large numbers of variables are interconnected with and interdependent on one another.80 The difference between Davies’ visualisation (Fig. 5.5) and the three examples we have seen in the previous section (Figs. 5.1, 5.2a, 5.2b, and 5.3) is mainly aesthetic rather than essential. Where Davies chose to display various factors (nodes) and the relations between them (arrows) with different types of graphical schemata (circles, ellipses, and squares, for the former; strait, dotted, and thick lines, for the latter), the stocks-equity-­ market econophysic model (Fig. 5.1) and the obesity causal loop model (Figs. 5.2a and 5.2b) cited above rely more heavily on round lines and use a variety of colours, rather than schemata. These two graphical  Davies, “Ancient Economies: Models and Muddles.”  Davies, “Linear and Nonlinear Flow Models.” 78  Ibid., 155. 79  Ibid., 142–52. 80  On “organised complexity,” see Weaver, “Science and Complexity.” 76 77

142 

M. HAKLAI

Fig. 5.5  Flow chart of resource movement; inspired by Davies, “Linear and Nonlinear Flow Models,” 150, Fig.  6.14: Flow chart of resource movement, model 6: Modified to incorporate bandwidths, motors, gates, and reservoirs

elements—a multiplicity of colours, and a variety of schemata—are combined in the Systems Biology of Human Aging Network Model (Fig. 5.3). Again, regardless of graphical techniques, all of these visualisations display rhizomatic properties, namely, nonlinearity, decentralisation, diversity, and emergence.

5   Money in Roman Law: A Diagram of Complex Mechanism Large amounts of detailed information, which could provide a basis for statistically driven visualisations of economic phenomena, as the one created by the stocks-equity-market econophysic model (Fig.  5.1), do not exist for ancient economies in the same way as they do for modern ones. However, complexity-oriented visualisations depicting social phenomena

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

143

need not rely on large numerical databases (e.g. see Figs. 5.2a and 5.2b). Even though numerical data are perceived as the building blocks of socalled “good” economic models, they are not a requirement of economic modelling. In fact, basic economic models are seldom based on databases,81 although data can sometimes be poured into them. Such models are helpful not because they would accurately mirror reality; they are not a graphical projection of empirical information. Rather, they are helpful because they create conceptual tools with which economic and other social phenomena can be categorised, described, or explained. Once information, in the sense of factual data, is cast into models created not by the data but for the data, trends, tendencies, and patterns can appear that help to assess changes. These observations are valid also for complexity-­oriented models. As stressed by Davies,82 intense visualisations, which display organised complexity and portray non-simple relations on an abstract plane, provide a helpful tool for qualitative analyses of ancient economic phenomena. I here follow Davies’ lead and attempt to offer such a visualisation for analysing the historical example of the Roman institutional environment for using money in the private sphere. When wishing to create a complexity-­ oriented model depicting the Roman institutional environment for using money economically in the private sphere, the first difficulty one tackles, regardless of graphic choices, is to decide what shall be displayed as “nodes,” meaning the factors or elements which comprise a system; and what shall be displayed as “arrows,” meaning the causal relations, connections, and boundaries imposed on the system’s elements. This difficulty derives first and foremost from the Janus-faced nature of money, being both abstract and concrete. These two characteristics are inseparably intertwined, making money difficult to define, scrutinise, and model. Money is the ultimate quantifier of value. Its inherent abstraction allows for every form of wealth to be considered as money, as noted already by the third-­ century Roman jurist, Ulpian.83 However, more often than not, when Romans talked of pecunia (i.e. money) they had a physical manifestation of money in mind, particularly coinage.84 Yet, they habitually used a multiplicity of things that served as monetary instruments. Roman money was 81  For example, the supply-demand or LM-IS curves, crossing one another at the point of equilibrium. 82  Davies, “Linear and Nonlinear Flow Models.” 83  Ulp. Dig. 50.16.178.pr. (49 ad sab.). 84  For example, Paul. Dig. 18.1.1.1 (33 ad ed.); Ulp. Dig. 14.6.7.3 (29 ad ed.).

144 

M. HAKLAI

not solely coinage; metals and other goods could serve as monetary ­instruments, and obligations, too, sometimes qualified as money.85 Which should be displayed as factors (nodes), and which as flows (arrows)? Furthermore, assuming that it is decided that monetary instruments would be displayed as factors, and it is established which objects represent physical manifestations of money, how shall a model depict the difference between cash and credit? These questions, complicated as they may be, comprise just one level of difficulty. Second, there is need to decide which legal formations that dictate economic interaction should provide the system’s terminology. Modern institutional categorisation of exchange sometimes essentially differs from that of antiquity. Let us consider, for example, locatio conductio, the Roman contract for letting and hiring. This consensual contract86 created bilateral obligations,87 with one party receiving rights of usage and enjoyment (uti frui),88 while the other received remuneration (merces), be it rent or wages. The two parties were called locator and conductor. The locator was the party who “placed” something with the other party, while the conductor was the party who “took along” that which was made available to her or him.89 The position of the parties varied in accordance with the details of the transaction. When the contract was one of lease or hire, the locator was the possessor of property and placed it with the conductor, the lessee, who in return provided a payment of some sort (merces). When the contract was one of employment, the locator was the employee, who placed her or his work capacity at the disposal of the contractor, the employer, who in return provided remuneration. When locatio conductio was used for contracting a specific task or job,90 the party bound to do the job was the conductor, who “carried along” the belongings on which the task was to be performed, while the party commissioning the task and paying the merces was the locator (contrary to the two previous cases, where the conductor rendered the merces). Thus, locatio conductio followed internal dynamics  Harris, “A revisionist view of Roman money”; Bange, Kreditgeld in der römischen Antike.  That is, based on the joint agreement of the parties; Gai. Inst. 3.135; Epit. 2.9.13; Paul. Dig. 19.2.1 (34 ad ed.); Just. Inst. 3.22; 29.4. 87  Gai. Inst. 3.137. 88  For example, Ulp. Dig. 19.2.9.pr. (32 ad ed.); Paul. Dig. 19.2.24.4 (34 ad ed.); Afric. Dig. 19.2.33 (8 quaest.). 89  Kaser, Das römishce Privatrecht, 563. 90  For example, Labeo Dig. 19. 2. 60. 8 (5 post. a javol. Epit.); Pompon. Dig. 18. 1. 20 (9 ad sab.). 85 86

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

145

different from modern ones, and encompassed several modern transactions including lease, hire, and employment. Another example is that of loans. Under the general modern category of “loans” fall several Roman legal institutions: Pecunia credita means “money given by way of credit” and was created via a stipulatio, a Roman verbal formula of question and answer widely used to create a variety of agreements and obligations. Both the Roman contract of mutuum and that of commodatum created loan relations and were real contracts in the sense that they came into being only after the actual transfer of the object of the transaction between the parties. While commodatum was a “loan for use” where the very same thing lent had to be returned, mutuum was a “loan for consumption” where the object of the loan usually was consumed, so that the exact same thing given as a loan could not be returned and instead an equivalent was to be repaid. A third type of real contract which could create, in practice though surely not in theory, financial credit relations was depositum. Depositum was the Roman contract for safekeeping, where initially neither depositor nor depositary were to receive remuneration. However, the writings of Roman jurists indicate that this was not always so. Roman legal sources preserve rulings of leading jurists who allowed both the practice of “open deposits,” where a depositor gave explicit permission to use the deposit,91 and that of interest-bearing deposits;92 both of which came to be known in the scholarship as depositum irregulare.93 Thus, the question stands: Which of these, if any at all, should be displayed as the system’s factors (nodes)? And how should connections between all of these legal contracts and formulae be displayed? In what follows, I offer a complex visualisation that depicts the institutional environment created by Roman law for using money in private exchange activities (Fig. 5.6). It is a symbolic diagram placed on an abstract plane. The factors constructing the model (nodes) represent Roman legal contracts and forms for agreements that dictated economic interaction. Square boxes represent Roman contracts and other legal formats for creating obligations; for example, emptio venditio (sale), mutuum (“loan for  For example, Papin. Dig. 16. 3. 25. 1 (3 respon.); Paul. Dig. 47. 2. 21. 1 (40 ad sab.).  For example, Papin. Dig. 16. 3. 24 (9 quaest.); Paul. Dig. 16. 3. 26. 1 (4 respon.); Ulp. Dig. 16. 3. 7. 2–3 (30 ad ed.); 42. 5. 24. 2 (63 ad ed.), in both excerpts Ulpian rules that depositors who received interest on their deposits should be “considered to have renounced their deposits,” thereby indicating that this was, in fact, a common practice; Scaev. Dig. 32. 37. 5 (18 dig.); Cod Iust. 4. 34. 4 (Gordian, 238–244 CE). 93  Kaser, Das römische Privatrecht, 536. 91 92

Fig. 5.6  Roman legal institutions for using money in private exchange

146  M. HAKLAI

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

147

consumption”), stipulatio (a binding verbal agreement), and so on. Round-edges boxes represent generalising categorisations; for example, consensual contracts, real contracts, and credit (pecunia credita). Arrowed boxes represent formats or instruments which increase the connection, commitment, or obligation of the parties; for example, penalties (poenae), interest (usura/faenus), and earnest payments (arra). The arrows in the model are of two kinds: thinner lines represent causal sequence of events, and thicker lines represent flows, movement, or transport of things. Additionally, colours also have meaning as indicated in the “colour key,” representing: cash, credit (including records of obligations used as money), coinage, metals, perishable goods used as money, and things (goods and services) not conceptualised in monetary terms. The diagram is not a complete representation of all that is relevant to the Roman institutional environment of using money in private transactions. Other contracts, agreements, and legal and economic institutions may be added; for example, securities, societas (partnership), or mandatum (the power to commission or enforce execution). Furthermore, the “colour key” does not accurately display the complicated relation between the various physical manifestations of money on the one hand, and the essentially abstract division into cash and credit on the other hand. Thus, for example, coins and consumable goods can be used as physical manifestations of money in both cash and credit transactions. Yet, even in its current state this diagram has several advantages. It takes into account the abstract nature of money whilst accommodating a spectrum of its physical manifestations. It displays just how central and omnipresent credit was in Roman economic life. And, it demonstrates connections between Roman legal institutions, and how these were elaborated and developed by the actual workings of Roman legal practitioners. The binding legal validity of the learned opinions and judgments of Roman jurists, who gave their rulings in response to the economic reality around them, enabled the self-­ adjustment of the institutional environment and the emergence of new practices and procedures. In what follows several verbal explanations of the diagram are offered to elucidate these dynamics. The diagram shows the outcome of two famous controversies between the two schools of thought that existed in Roman law: The Proculians and the Sabinians. Since the Julio-Claudians, Roman juristic intellectual life was dictated by a division into these two schools of thought (scholae), their

148 

M. HAKLAI

legal rulings differing in a long list of controversies.94 Two of these are the famous pretium in numerata pecunia controversy regarding the pecuniary nature of a price (pretium) in contracts of sale (emptio venditio), and the related controversy regarding the nature of merces (rent/wages) in contracts of locatio conductio (lease/hire/employment). After decades of educated legal argumentation, the Proculian stance prevailed, which held that a transaction was to be classified as a “sale” (emptio venditio)—thereby making available the legal aid of the actiones empti et venditi—only when the price (pretium) was in numerata pecunia; otherwise, the transaction was to be classified under the Roman contract of “barter” (permutatio).95 The outcome of this school controversy is shown in the diagram (Fig. 5.6) in the form of a clear division between permutatio (barter) and emptio venditio (sale). The former is placed in the sphere of real contracts, and is coloured in grey to indicate things not conceptualised in monetary terms. The latter is placed in the sphere of consensual contracts, and its colours indicate that whether performed in coinage (purple), metals (yellow), or perishable goods (green), and whether conducted in cash (red) or credit (blue), it was conceptualised in monetary terms. Additionally, the thick arrowed-line from emptio venditio to permutatio indicates that any permanent transfer of goods (hence the arrow’s green colour) not conceptualised in money (hence its grey frame) no longer could qualify as emptio venditio, but fell under the category of permutatio. Regarding depositum, the diagram condenses the subtle shift in its use, which occurred during imperial time. As it is a contract, depositum is represented by a square box, which is placed within the sphere of a round-­ edged box, representing the generalising category of real contracts. Because in its very essence depositum created credit relations—with one party, the depositary, indebted to the other party, the depositor; namely, to return the deposit intact—this box is coloured in blue. Since depositum could concern coinage, metals, perishable goods, or things not conceptualised in monetary terms, the box is also marked with the colours purple, yellow, green, and grey, respectively. Metals, perishable goods, coinage, and other forms of cash given by way of depositum create credit; hence, the yellow, green, purple, and red lines representing the causal sequence of 94   Gaius discusses twenty-two such controversies, and some are mentioned also in Justinian’s Corpus Iuris Civilis; Leesen, Gaius Meets Cicero. 95  The main texts are Gai. Inst. 3. 141; Paul. Dig. 18. 1. 1. 1 (33 ad ed.); 19. 4. 1. pr. (32 ad ed.).

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

149

events between depositum and pecunia credita. The latter, pecunia credita, is represented by a round-edges box, as it is a generalising categorisation, rather than a type of contract or a legal format that enhanced connection between parties. This type of visualisation demonstrates how the complexity of the system encapsulates a potential for emergence. Officially, depositum was no loan, neither was its goal to establish commercial credit relations. However, the practical approach of Roman jurists enabled the legal institutional environment to self-adjust in accordance with the needs of its users. Thus, a willingness to accept the use of the actio depositi in cases that violated the old requirement of depositum—namely, the non-use of the deposit and the obligation to return the same objects deposited, probably by using it as an actio utilis—resulted in the so-called depositum irregulare, which in reality apparently could accommodate the existence of interest bearing deposits (in practice but not in principle). Thus, the implicit connection between depositum and pecunia credita, clearly shown in the diagram, enabled users to self-adjust the system’s components. The outcome was using depositum also in financial credit relations, which found it traces in legislation of Roman jurists, later to be titled depositum irregulare.96 Thus, the complexity of the system generated the emergence of new patterns of interaction, as explicitly portrayed in the diagram. Other obligations may augment a depositum via, for example, a stipulatio; the latter being a legal format that created obligations hence, it is displayed by a square box. The two institutions, depositum and stipulatio, are connected with lines representing causal sequence of events, coloured in either red or blue, representing cash and credit, respectively. A stipulatio could create commitments to pay penalties (poenae) or interest (usura/faenus) hence, the lines representing the causal sequence of events between stipulatio on the one hand, and poenae and usura on the other hand. The latter two are both instruments that enhance the connection between parties hence, are displayed by arrowed-boxes. As regards “penalties,” it is causally linked with stipulatio via purple, yellow, green, and red lines, representing the offending party’s obligation to render coinage, metals, perishable goods, or cash in general, respectively; all of which conceptualised in monetary terms. Since a penalty could take the form of paying interest, poenae is casually linked to usura with purple, blue, and red lines, representing coinage, credit, and cash, respectively. As regards  See notes 91–93 above.

96

150 

M. HAKLAI

“interest,” besides penalising interest, real practice made it relevant also for interest-bearing deposits (see above) hence, the blue thicker arrowed-­ line between depositum and usura, generally representing flows, movement, or transport of things. The purple thick arrowed-line from locatio conductio to depositum indicates that safekeeping in return for pecuniary remuneration was no longer conducted solely under locatio conductio,97 but could also potentially be accommodated under interest-bearing deposits. The thick arrow-lines coloured in blue (representing credit) and framed in red (representing cash), which connect depositum with both mutuum and pecunia credita, represent the interchangeability with which parties actually used these legal formats. Traces of this practice are found in juristic references to what later became known as depositum irregulare. In other words, these arrows manifest the flexibility with which parties could switch from loans to deposits and back, creating de facto interest-bearing deposits. Lastly, the yellow and purple thick arrowed-lines connecting depositum to pecunia credita represent the potential of interest-bearing deposits of durable pecuniary instruments, such as coinage or metals, to contribute to the supply of money by allowing a mechanism which might have created credit money.98 Further connections and relations, not detailed here verbally, are displayed in the diagram. For example, between emptio veditio (sale), locatio conductio (lease/hire), and arra (earnest payments); with the latter, arra, having the potential of becoming a cash payment considered as penalty in cases of non-fulfilment of the contract (therefore the red line connecting arra with poenae). Or, between commodatum and mutuum, which via stipulatio could include obligations for either interest payments or penalising interest, and as such could create pecunia credita. I believe that these explanations have demonstrated how the intense diagram presented in Fig.  5.6 reflects the diversity, interconnectedness, decentralisation, and nonlinearity that characterised the Roman legal environment and, in particular, the Romans’ use of money in private economic transactions.

97  No matter whether regarded as the so-called locatio conductio operis (contracting a task), or locatio conductio operarum (employing a guard). 98  For example, Harris, “A revisionist view of Roman money”; Bange, Kreditgeld in der römischen Antike.

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

151

6   Legal Institutions in a System of Multi-­traditions: A Causal Loop Diagram One flaw of the visualisation offered in Fig. 5.6 is that it concentrates on Roman law alone, as if it was a closed system. However, Roman law was just one of the legal traditions at work in the Roman empire. The empire’s inhabitants could, at least sometimes, choose between legal corpora in accordance with their wants and needs.99 Thus, it is not only that modern institutional categorisation of exchange essentially differs from Roman one. Within the Roman empire there were several active legal traditions, each following its own internal logic hence, resulting in a somewhat different institutionalised organisation of exchange. For example, we have already mentioned that under the general modern category of “loans” fell the Roman contracts of mutuum and commodatum, pecunia credita created via stipulatio, and to a lesser extent depositum (in practice, but probably not in principle). Other ancient legal systems within the empire had their own institutions for contracting loans. For example, in the Greek-­ Hellenistic tradition practiced in Roman Egypt and preserved on papyri, formulae that accommodated lending relations included daneion (δάνειον), ~σις), paratheke (παραθήκη), and contracts which came to be chresis (χρƞ known in the scholarship as “Sale on Delivery,” which de facto created loans to be repaid in kind. In Jewish legal texts written within the Roman regime one finds loans given by a creditor, mal’veh (‫)מלוה‬, to a debtor, loveh (‫)לווה‬, both having the same root, l-w-h (‫ה‬-‫ו‬-‫ ;)ל‬alongside “loans-for-­ use” given by a mash’iil (‫ )משאיל‬to a sho’el (‫)שואל‬, the root for both is sh-a-l (‫ל‬-‫א‬-‫)ש‬. The latter type of loan resembles the Roman contract of commodatum; however, Jewish legal classification places it under the category of Dinei Shomrim (‫)דיני שומרים‬, literally “Laws of Guardians” but practically meaning “Laws of Deposits.” When looking at the empire’s legal environment as a whole, which of these, if any at all, should be displayed as the system’s factors (nodes)? How connections between all of these legal formations should be displayed? The second visualisation I offer (Fig. 5.7) is a first attempt to visually model, though only partially, the institutional environment of economic interactions carried out under simultaneously coexisting legal systems. It takes its role-model from causal loop diagrams, as the one offered by the 99  For example, the Babatha archive; Oudshoorn, The Relationship between Roman and Local Law in the Babatha and Salome Komaise Archives.

152 

M. HAKLAI

Obesity System Map cited above (Figs. 5.2a and 5.2b), to show ways in which Roman legal institutions interacted with non-Roman legal formations operating under the Roman regime. The image is created on an abstract plane and topologically presents the institutionalised legal sphere within which economic interaction occurred. It shows institutional connections between structural entities of different legal traditions in operation under Roman rule; and it displays in a condensed manner the direction of various shifts, connections, assimilations, and influences, which could take place between these factors within the Roman regime. Three legal traditions present the factors in the diagram: Roman law is marked in yellow, Greek-Hellenistic law as practiced in Roman Egypt is marked in green, and Jewish law is marked  in light blue. Additionally, institutions and concepts relevant for modern categorisation of exchange are presented in the middle of the diagram in grey. Contracts and legal formats which belonged to the same categorisation within each legal tradition are marked by a dashed-circle holding them together. The grey arrows in the diagram present the direction of the connections and similarities between the various legal institutions. The black arrows show the conceptual bondages between the modern concepts which affect categorisation of exchange. And the blue arrows indicate connections between these modern concepts and ancient legal formations. Thus, for example, in this diagram depositum is connected to both the Greek paratheke (παραθήκη) and the Jewish shomer h·inam (‫חינם‬ ‫)שומר‬, indicating that these could be treated as parallel formats, in which credit is created by rendering a deposit for the sake of safekeeping. Paratheke is ~σις), as all three conconnected to both daneion (δάνειον) and chresis (χρƞ tracts could be used to create loans and credit relations. Accordingly, paratheke is also connected to mutuum, showing that the two institutions could be treated as parallels, that is, as formats for contracting monetary loans. The diagram also connects shomer h·inam and shomer sakhar (‫שומר‬ ‫שכר‬, literally “a paid guardian”), both categorised under the Jewish Laws of Guardians. While the former (shomer h·inam) is based on trust and creates credit relations, the latter (shomer sakhar) is the Jewish legal format for paid safekeeping. As shomer sakhar requires pecuniary remuneration, it is connected to the Roman locatio conductio via the so-called locatio conductio operis, which enabled contracting a service, in this case safekeeping, rendered for a pecuniary payment.

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

153

Fig. 5.7  Causal loop diagram for legal institutions from three legal systems operating in the Roman empire

Verbal explanations of the diagram (Fig. 5.7) could continue on, but I believe the point is made clear. This graphical representation shows how the various legal traditions which coexisted under Roman regime created a complex institutional framework for exchange. This complex system had the ability to allow new patterns of activity to be created by individuals who self-adjusted their use of it. The inhabitants of the Roman empire could sometimes shift their use of legal institutions within and between legal traditions. They altered its application in accordance with their needs, contributing to the continuously evolving system in which they operated, and in so doing allowing it to become more than the sum of its components. How these dynamics came to be expressed, expounded, and particularised after the Constitutio Antoniniana, calls for further detailed analyses to be performed in future research.

154 

M. HAKLAI

7   Conclusions This chapter promotes a complexity-oriented approach to model the use of money in economic transactions in the Roman empire as framed by the Roman legal institutional environment. It views economies as complex systems and stresses the advantages of using visualised complexity diagrams to depict structural and processual analyses of institutional change. It demonstrates the multidirectional dynamics of economic and legal interactions within the system of Roman law, as well as in the diverse and multicultural environment which the Roman empire was. The reciprocal exchange of ideas and legal formats—which was the outcome of the different legal traditions that coexisted within the same political, administrative, and economic sphere—was stirred by the desire of individuals to better pursuit their interests. The complexity of the economic environment within which the empire’s inhabitants operated was such that reciprocal movements of ideas and legal formats took place within and between legal traditions. Individuals self-adjusted their course of action in a dynamic reality. They used a range of available legal formats according to their wants and interests, thereby casting new meaning into existing institutions. Perhaps the most deductive conclusion of the complexity approach is that we cannot understand “growth,” “decline,” or “crisis” only in linear terms or as outcomes of well-defined phenomena, such as occupation, inflation, or changes in the currency. These explanations are just too simplistic; hence, they fail to explain the interconnections between phenomena as well as patterns of emergence. Changes occurred, of course, but these were driven by dynamics which cannot be described simply in linear terms of cause and effect. Interconnectedness, multi-linkages, repetition, emergence, and causal loop formations all are integral elements of the system without which its operation cannot be properly realised, conceptually described, or visually depicted.

Bibliography Anderson, Philip W., Kenneth J. Arrow, and David Pines, eds. The economy as an evolving complex system: the proceedings of the Evolutionary Paths of the Global Economy Workshop, held September, 1987  in Santa Fe, New Mexico. Redwood City, Ca.: Addison-Wesley Pub. Co., 1988.

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

155

Andreau, Jean. La vie financière dans le monde romain: les métiers de manieurs d’argent (IVe siècle av. J.-C.- IIIe siècle ap. J.-C.). Rome: Ecole Francaise de Roma, 1987. Andreau, Jean. Banking and Business in the Roman World. Cambridge: Cambridge University Press, 1999. Arthur, W. Brian. “Positive Feedbacks in the Economy.” Scientific American 262 (1990): 92–9. Arthur, W. Brian, Steven N. Durlauf, and David A. Lane, eds. The economy as an evolving complex system II (Santa Fe Institute, Studies in the Sciences of Complexity). Reading, Mass: Addison-Wesley, 1997a. Arthur, W. Brian, Steven N. Durlauf, and David A. Lane. “Introduction.” In The economy as an evolving complex system II (Santa Fe Institute, Studies in the Sciences of Complexity), edited by W. Brian Arthur, Steven N. Durlauf, and David A. Lane, 1–14. Reading, Mass: Addison-Wesley, 1997b. Bange, Matthias. Kreditgeld in der römischen Antike. Ursprünge, Entstehung, Übertragung und Verbreitung (Pharos Studien zur griechisch-römischen Antike, Band 33). Rahden/Westf: Verlag Mary Leidorf GmbH, 2014. Brock, William A. “Asset Price Behavior in Complex Environments.” In The economy as an evolving complex system II (Santa Fe Institute, Studies in the Sciences of Complexity), edited by W.  Brian Arthur, Steven N.  Durlauf, and David A. Lane, 385–423. Reading, Mass: Addison-Wesley, 1997. Caldwell, Bruce. Hayek’s challenge: An intellectual biography of F.A.  Hayek. Chicago: University of Chicago Press, 2004. Casti, John L. “Money is funny, or why finance is too complex for physics.” Complexity 8, no. 2 (2003): 14–8. Chaumont-Chancelier, Frédérique. “Hayek’s Complexity.” Journal des Economistes et des Etudes Humaines 9, no. 4 (December, 1999): 543–64. https://doi. org/10.1515/jeeh-1999-0405. Colander, David. ed. Complexity and the History of Economic Thought: Perspectives on the History of Economic Thought. Selected papers from the history of economics Society Conference, 1998. London: Routledge, 2000. Colander, David, Richard P. F. Holt, and J. Barkley Rosser. The changing face of economics: Conversations with Cutting Edge Economists. Ann Arbor: University of Michigan Press, 2004. (Reprinted 2007). Corning, Peter A. “The re-emergence of “emergence”: A venerable concept in search of a theory.” Complexity 7, no. 6 (2002): 18–30. Davies, John K. “Ancient Economies: Models and Muddles.” In Trade, Traders and the Ancient City, edited by Helen Parkins and Christopher John Smith, 221–52. London: Routledge, 1998. Davies, John K. “Linear and Nonlinear Flow Models for Ancient Economies.” In The Ancient Economy: Evidence and Models, edited by Joseph G. Manning and Ian Morris, 127–56. Stanford, California: Stanford University Press, 2005.

156 

M. HAKLAI

Day, R.  H. “Complex Economic Dynamics: Obvious in History, Generic in Theory, Elusive in Data.” Journal of Applied Econometrics 7 (Supplement Special Issue on Nonlinear Dynamics and Econometrics) (1992): 9–23. Deleuze, Gilles, and Felix Guattari. A Thousand Plateaus: Capitalism and Schizophrenia. Translation and Foreword by Brian Massumi. Minneapolis; London: University of Minnesota Press, 1987. Duncan-Jones, Richard. Structure and scale in the Roman economy. Cambridge: Cambridge University Press, 1990. Finley, Moses I. The Ancient Economy. Foreword by Ian Morris. Berkeley; London: University of California Press, 1985. Reprinted 1999. Furber, John D., Systems Biology of Human Aging—Network Model Wall Chart (Rev 11 March 2018), available at http://www.LegendaryPharma.com/ chartbg.html. Gaus, G.  F. “Hayek on the evolution of society and mind.” In The Cambridge companion to Hayek, edited by E.  Feser, 232–58. Cambridge: Cambridge University Press, 2006. Harris, William V. “A revisionist view of Roman money.” Journal of Roman Studies 96 (2006): 1–24. Hayek, Friedrich A. von. Individualism and Economic Order. Chicago: University of Chicago Press, 1948. Hayek, Friedrich A. von. The Constitution of Liberty. Chicago: University of Chicago Press, 1960. Hayek, Friedrich A. von. “The Theory of Complex Phenomena.” In Studies in Philosophy, Politics, and Economics, 22–42. London: Routledge & Kegan Paul, 1967. Holland, John H. Emergence: From Chaos to Order. Oxford: Oxford University Press, 1998. Hollander, David B. Money in the late Roman Republic. Leiden; Boston: Brill, 2007. Hopkins, Keith. “Taxes and Trade in the Roman Empire.” Journal of Roman Studies 70 (1980): 101–25. Jovanovic, Franck, and Christophe Schinckus. “The Emergence of Econophysics: A New Approach in Modern Financial Theory.” History of Political Economy 45, no.3 (2013): 443–74. https://doi.org/10.1215/00182702-2334758. Kaser, Max. Das römishce Privatrecht, Erster Abschnitt: Das altrömische, das vorklassische und klassische Recht. 2nd ed. München: C.  H. Beck’sche Verlagsbuchhandlung, 1971. Kehoe, Dennis P. Investment, profit, and tenancy: the jurists and the Roman agrarian economy. Ann Arbor: University of Michigan Press, 1997. Kehoe, Dennis P. Law and the rural economy in the Roman Empire. Ann Arbor: University of Michigan Press, 2007. Kirman, Alan P. “The Economy as an Interactive System.” In The economy as an evolving complex system II (Santa Fe Institute, Studies in the Sciences of

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

157

Complexity), edited by W. Brian Arthur, Steven N. Durlauf, and David A. Lane, 491–531. Reading, Mass: Addison-Wesley, 1997. Koppl, Roger. “Complexity and Austrian economics.” In Handbook of research on complexity, edited by J.  Barkley Rosser, 393–406. Cheltenham, UK: Edward Elgar Publishing, 2009. https://doi.org/10.4337/9781781952665.00026. Krugman, Paul R. The Self-Organizing Economy. Oxford: Blackwell, 1996. Krugman, Paul R. “How the Economy Organises Itself in Space: A Survey of the New Economic Geography.” In The economy as an evolving complex system II (Santa Fe Institute, Studies in the Sciences of Complexity), edited by W. Brian Arthur, Steven N.  Durlauf, and David A.  Lane, 239–62. Reading, Mass: Addison-Wesley, 1997. Lane, David A. “Is What Is Good for Each Best for All? Learning from Others in the Information Contagion Model.” In The economy as an evolving complex system II (Santa Fe Institute, Studies in the Sciences of Complexity), edited by W.  Brian Arthur, Steven N.  Durlauf, and David A.  Lane, 105–27. Reading, Mass: Addison-Wesley, 1997. Leesen, Tessa G.. Gaius Meets Cicero: Law and Rhetoric in the School Controversies. Leiden: Martinus Nijhoff Publishers, 2010. Leijonhufvud, Axel. “Life among the Econ.” In Information and Coordination: Essays in Macroeconomic Theory, 347–59. Oxford; New York: Oxford University Press, 1981. Lima, Manuel. Visual Complexity: Mapping Patterns of Information. New  York: Princeton Architectural Press, 2011. Lima, Manuel. The book of trees: visualizing branches of knowledge. New  York: Princeton Architectural Press, 2014. Lindgren, Kristian. “Evolutionary Dynamics in Game-Theoretic Models.” In The economy as an evolving complex system II (Santa Fe Institute, Studies in the Sciences of Complexity), edited by W. Brian Arthur, Steven N. Durlauf, and David A. Lane, 337–67. Reading, Mass: Addison-Wesley, 1997. Manovich, Lev. “Introduction.” In Visual Complexity: Mapping Patterns of Information, by Manuel Lima, 11–3. New  York: Princeton Architectural Press, 2011. Metcalfe, J.  Stanley, and John Foster. eds. Evolution and Economic Complexity. Cheltenham, UK: Edward Elgar Publishing, 2007. Montgomery, Michael R. “Complexity Theory: An Austrian perspective.” In Complexity and the History of Economic Thought: Perspectives on the History of Economic Thought. Selected papers from the history of economics Society Conference, 1998, edited by David Colander, 227–40. London: Routledge, 2000. Nicolis, Gregoier, and Iya Prigogine. Exploring complexity: An introduction. New York: W. H. Freeman, 1989. North, Douglass C. “Institutions.” Journal of Economic Perspectives 5, no. 1 (1991): 97–112.

158 

M. HAKLAI

North, Douglass C. “Some Fundamental Puzzles in Economic History/ Development.” In The economy as an evolving complex system II (Santa Fe Institute, Studies in the Sciences of Complexity), edited by W. Brian Arthur, Steven N.  Durlauf, and David A.  Lane, 223–37. Reading, Mass: Addison-­ Wesley, 1997. Oudshoorn, Jacobine G. The relationship between Roman and local law in the Babatha and Salome Komaise archives: general analysis and three case studies on law of succession, guardianship, and marriage. Leiden: Brill, 2007. Prasch, Robert E. “Complexity and economic method: an institutionalist perspective.” In Complexity and the History of Economic Thought: Perspectives on the History of Economic Thought. Selected papers from the history of economics Society Conference, 1998, edited by David Colander, 215–26. London: Routledge, 2000. Rathbone, Dominic. Economic rationalism and rural society in third-century A.D.  Egypt: The Heroninos archive and the Appianus estate. Cambridge: Cambridge University Press, 1991. Rathbone, Dominic, and Peter Temin. “Financial Intervention in first century AD Rome and eighteenth century England.” In Pistoi dia tèn technèn: bankers, loans and archives in the ancient world. Studies in honour of Raymond Bogaert (Stadia Hellenistica 44), edited by Koenraad Verboven, Katelijn Vandorpe, and Véronique Chankowski, 371–419. Leuven: Peeters, 2008. Rosser, J. Barkley. “How complex are the Austrians?” In What is so Austrian about Austrian Economics? (Advances in Austrian Economics, Vol. 14), edited by Roger. Koppl, Steven Horwitz, and Pierre Desrochers, 165–79. Bingley, UK: Emerald, 2010. Schinckus, Christophe. “Between complexity of modelling and modelling of complexity: An essay on econophysics.” Physica A: Statistical Mechanics and its Applications 392, no. 17 (2013): 3654–65. https://doi.org/10.1016/j. physa.2013.04.005. Seland, Eivind Heldaas. “Here, There and Everywhere. A Network Approach to Textile Trade in the Periplus Maris Erythraei.” In Textiles, Trade and Theories. From the Ancient Near East to the Mediterranean, edited by Kerstin Droß-­ Krüpe and Marie-Louise Nosch, 211–9. Münster: Ugarit Verlag, 2016. Shapiro, Arnold F. “A Hitchhiker’s guide to the techniques of adaptive nonlinear models.” Insurance: Mathematics and Economics 26, nos. 2–3 (2000): 119–32. Snowdon, Brian, and Howard R.  Vane. Modern Macroeconomics: Its Origins, Development and Current State. Cheltenham, UK; Northampton, MA, USA: Edward Elgar, 2005. Tumminello, Michelle, Tomaso Aste, Tiziana Di Matteo, and Rosario Mantegna. “A Tool for Filtering Information in Complex Systems.” Proceedings of the National Academy of Sciences of the United States of America 102 (30) (August, 2005): 10421–6. https://doi.org/10.1073/pnas.0500298102.

5  VISUALISING ROMAN INSTITUTIONAL ENVIRONMENTS FOR EXCHANGE… 

159

Van Der Leeuw, Sander, and James McGlade. eds. Time, process, and structured transformation in archaeology. London: Routledge, 1997a. Van Der Leeuw, Sander, and James McGlade. “Archaeology and non-linear dynamics—new approaches to long-term change.” In Time, process, and structured transformation in archaeology, edited by Sander Van Der Leeuw and James McGlade, 1–31. London: Routledge, 1997b. Vandenbroeck, Philippe, Jo Goossens, and Marshall Clemens. Tackling obesities: Future Choices—Building the Obesity System Map, London: Government Office for Science, Dept. of Innovation, Universities and Skills, 2007. http://www. bis.gov.uk/assets/foresight/docs/obesity/11.pdf. Vaughn, Karen I. “Hayek’s Implicit Economics: Rules and the Problem of Order.” Review of Austrian Economics 11 (1999): 129–44. Velupialli, K.  Vela. “Non-Linear Dynamics, Complexity and Randomness: Algorithmic Foundations.” Journal of Economic Surveys 25, no. 3 (2011): 547–68. Vennix, Jac A. M. Group model building: facilitating team learning using system dynamics. Chichester: Wiley, 1996. Ware, Colin. Information visualization: perception for design. 3rd ed. Waltham, Mass.: Morgan Kaufmann, 2013. Weaver, Warren. “Science and complexity.” American Scientist 36, no. 4 (October, 1948): 536–44. Wible, James. “What is complexity?” In Complexity and the History of Economic Thought: Perspectives on the History of Economic Thought. Selected papers from the history of economics Society Conference, 1998, edited by David Colander, 15–30. London: Routledge, 2000.

PART II

Urban Systems

CHAPTER 6

Social Complexity and Complexity Economics: Studying Socio-economic Systems at Düzen Tepe and Sagalassos (SW Turkey) Dries Daems

1   Introduction It is well known that archaeologists are generally not highly trained in mathematics and are often more inclined to integrate their analyses in narrative frameworks. While there is nothing inherently wrong with narrative-­ based research, a different approach is sometimes called for. In recent years, archaeologists have been increasingly urged to formalise their arguments systematically both to analyse data and test hypotheses.1 Mathematics can be useful in this respect although it is not always necessary. Even

1

 Smith, “How Can Archaeologists Make Better Arguments”.

D. Daems (*) University of Leuven, Leuven, Belgium e-mail: [email protected] © The Author(s) 2021 K. Verboven (ed.), Complexity Economics, Palgrave Studies in Ancient Economies, https://doi.org/10.1007/978-3-030-47898-8_6

163

164 

D. DAEMS

without mathematical expressions, formalisation has the advantage of making assumptions explicit and facilitating comparative analysis. One research field in which an explicitly formal approach is called for is the study of complex systems. Archaeology has been cautiously warming up to complexity approaches.2 So far, however, they have spread only gradually and, while highly promising, their applications have not yet lived up to their inherent potential. Early applications were generally concerned merely with the principles of complex systems and their epistemological relevance to study the past. Consequently, the conceptual framework of complex systems has too often been used only metaphorically within larger narratives aiming to describe overall patterns rather than explain the underlying dynamics that generated them. If the use of complex systems studies in archaeology is to move beyond the descriptive level, a more formal approach is needed. In this paper, I will demonstrate how the application of general causal factors and mechanisms of complexity development—as established in complex systems studies—can contribute to our understanding of socio-­ economic complexity in the past. I wish to make clear from the outset that in this paper we will be considering specifically the dynamics of social and economic complexity at the level of individual settlements and communities. The framework presented here, and the conceptual model in which it is grounded, will be applied to a case study of the late Achaemenid and early Hellenistic (fifth to second centuries BCE) communities at Sagalassos and Düzen Tepe in southwest Anatolia. I will focus mainly on the material culture of both communities, more specifically their pottery as this constitutes the most abundantly preserved category of material culture at both sites. The aim of this paper is to use observations on resource procurement, production processes, production output, and structures of exchange as proxies to identify or approximate causal factors contributing to the development of socio-economic complexity at this local scale. It has been observed that Sagalassos from the second century BCE onwards went through a phase of rapid social, economic, and political transformation.3 The process has been associated axiomatically with a concordant 2  Bentley and Maschner, Complex Systems and Archaeology; Kohler, “Complex Systems and Archaeology”. 3  Daems and Poblome, “The Pottery of Late Achaemenid Sagalassos; Poblome et  al., “How Did Sagalassos Come to Be”; Talloen and Poblome, “The 2014 and 2015 Control Excavations on and around the Upper Agora of Sagalassos.

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

165

increase in (social/economic/political) complexity. The present paper intends to clarify the underlying factors that were important for the development of this complexity, focusing in particular on its socio-­economic component.

2   A Framework of Socio-economic Complexity The framework for this paper is based on a conceptualisation of human societies as complex adaptive systems (CAS). These can be defined as large networks of interacting components with simple rules of operation, exhibiting dynamic emergent behaviour that cannot be reduced to the aggregate of characteristics of the component parts but is responsive to systems’ environment.4 Human societies develop as complex adaptive systems from the multitude of social interactions between the individual and collective agents (such as households) making up the system. Through the development of social practices performed through time and space these interactions give rise to processes of structuration, thereby creating social systems that exhibit complex emergent behaviour. This system behaviour in turn exerts positive and negative feedback on the behaviour of the agents that make up the system. Because the archaeological record is essentially a fragmentary reflection of the material end result of social practices performed in the past, we hold that it is ontologically suited to match this conceptual framework.5 In discussing human societies as complex adaptive systems (CAS), it is essential to define what exactly constitutes complexity in these systems and how it develops. Unfortunately, “complexity” is often used as a descriptive concept—its origins and development remaining something of a black box. It has been stated that “one of the hurdles in defining a theory of complexity, and with it, developing a fundamental, helpful approach is that there is no uniformity in the meaning of complexity”.6 The term can, for example, refer to the various aspects or subparts of a system, as well as to the magnitude and variety of the overall system. It is commonly associated with aspects such as intricate interdependencies among parts, non-­ linear behaviour, emergence, and self-organisation.7 The complexity of a  Holland, Hidden Order; Mitchell, Complexity.  Lucas, Understanding the Archaeological Record. 6  Sitte, “About the Predictability and Complexity of Complex Systems,” 25. 7  Mitchell, Complexity. 4 5

166 

D. DAEMS

system is often tied into the non-linear nature of its emergent behaviour— meaning that no direct linear relation can be drawn between system input and output. When different system components interact and mutually affect each other it can be difficult to see where system changes come from. This is why many complex systems interpretations, so far, remain descriptive rather than explanatory. It has been noted that different aspects or manifestations of complexity can exist, sometimes simultaneously within the same system, but none of them “is” complexity per se. Renate Sitte described five fundamental types of complexity: structural, functional, topological, algorithmic, and architectural.8 The two latter, architectural and algorithmic complexity, have seen few applications beyond very specific fields and are of limited use in the context of the present paper. I will focus here, therefore, on the first three. Structural complexity involves elements of dimensionality, networks, hierarchy, and levels depth/breadth. Functional complexity pertains to the differentiation between single or multifunctional components. Topological complexity refers to aspects such as connectivity, relation, number of relations, and direction of relations. For the sake of readability, I will subsume the different aspects of each type under a common denominator, respectively: dimensionality (for structural complexity), diversity (for functional complexity), and connectivity (for topological complexity). Dimensionality refers to the constituent components of the system, structured both vertically and horizontally. In general, the deeper the vertical nesting of various horizontal groups of components, the more complex the system becomes. Diversity, at its most basic level, pertains to the distribution of quantities over distinct classes.9 The term covers two different aspects; on the one hand “richness”, pertaining to the number of different categories within a sample, and on the other, “evenness”, referring to how quantities are distributed among these categories. Finally, connectivity is what makes complex systems truly tick. Complexity emerges only when a diverse set of components become interconnected, start to interact, and thus generate new information that drives further system dynamics. Increasing returns induced by connectivity therefore have a strong multiplier effect in system dynamics. These three aspects—dimensionality, diversity, and connectivity—can be considered as the mechanisms of complexity development. Social complexity can then be defined as the extent 8 9

 Sitte, “About the Predictability and Complexity of Complex Systems”, 25.  Leonard and Jones, Quantifying Diversity in Archaeology.

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

167

of differentiation among social units, integrated in coherently organised systems in both horizontal dimensions—as in various roles or social ­(sub) groups—and vertical dimensions, as in hierarchical concentration of decision-­making and power.10 Since the nineteenth century the prevalent neoclassical paradigm in economics posits that economic systems are inherently in equilibrium.11 For a long time, mainstream economic models hardly considered the dynamic workings of complex systems that exhibit far-from-equilibrium properties. Complexity theory and economic thinking eventually became connected during a workshop held at the Santa Fe Institute in 1987, which brought together economists, physicists, biologists, and computer scientists to work out a new framework for thinking about economic problems. In the aftermath, a new paradigm of “complexity economics” was developed that focused on contingency, change, and adaptation of agent strategies in response to commonly created outcomes.12 To use this outline of complexity economics as a starting point, we must consider how complexity develops specifically in socio-economic systems. A key emergent property of complex adaptive systems is their capacity for computation and transmission of “information” among its components—that is, inputs function as information telling the system components what to do, thereby affecting their behaviour.13 System changes occur when information input is received, interpreted according to internal rules, and transformed through behavioural mechanisms into a system output in the form of an (adapted) pattern of behaviour. Formalising this provides us with a model of input information (I); causal factors (X); mechanisms of complexity development, that is, dimensionality, diversity, and connectivity (M); and (socio-economic) system output (Y). The resultant Y can then feature as (part of) new input I, creating a recursive loop of system dynamics. Due to the non-linear nature of complex system dynamics, multiple causal factors and mechanisms can interact and co-evolve simultaneously, rendering any interpretation of the resultant system output essentially probabilistic.14 Still, simplified representations help us to make sense of the different components of system 10  Blanton, Stephen and Finsten, Ancient Mesoamerica; Feinman, “The Emergence of Social Complexity,” 36. 11  Beinhocker, The Origin of Wealth, 17. 12  Arthur, Complexity and the Economy, 1. 13  Holland, Complexity; Mitchell, Complexity. 14  Ragin, The Comparative Method, 24–25.

168 

D. DAEMS

dynamics and the nature of their interrelations. Identifying these mechanisms could then effectively open up the “black boxes” in our argumentation. Ideally it can be stated that the probability (P) of a factor X causing Y if, and only if, P(Y|X) > P(Y|x), with x being any other factor part of the overall system, within a set of understood ceteris paribus background conditions.15 Such an ideal structure is, of course, hard to get by in the reality of analysing archaeological data. This is why many archaeologists prefer a more ambiguous narrative framing of interpretation to this more “bare-boned” approach. Still, the advantage of clarity makes such a formal approach worthwhile, even only as a preliminary attempt for others to build on. The formal approach can be represented as: Y   X    X | I    M | X 



The angular brackets indicate that the conjunction of events is ordered from left to right. X can be considered as an element of a given system state developed out of a combination of I from prior system outcomes and external stimuli. Information is then evaluated according to a rule set derived from internalised practical knowledge and socialised behaviour in causal factor X and transformed into a new system response Y through a mechanism M. This formal model of the dynamics of social complexity provides us with a “problem-solving tool” to explain why socio-economic complexity develops”.16 The algorithmically formalised model17 postulates that various driving forces, or stimuli—both human or nature induced, and internal or external (for instance shifts in agricultural production, differentiation in harvest yields leading to social inequality, war/conflict, environmental change, etc.)—operate on the emergence and subsequent development of communities through a recursive loop of signal/problem detection, information-­processing, and problem-solving that results in either successful adaptation or failure of social organisation. This loop consists of a  Gerring, Social Science Methodology, 199.  Cioffi-Revilla, “A Canonical Theory of Origins and Development of Social Complexity”; Tainter, “Complexity, Problem Solving and Sustainable Societies”; Tainter, “Social Complexity and Sustainability”. 17  The PoliGen model was developed on the MASON (Multi-Agent Simulator of Networks and Neighbourhoods) platform, an open-source Java simulation toolkit developed as a collaboration between the Evolutionary Computation Laboratory and the Center for Social Complexity at George Mason University (http://cs.gmu.edu/~eclab/projects/mason/). 15 16

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

169

“fast process” of crisis and opportunistic decision-making through collective action, which feeds a “slow” process of socio-political development. The model is designed to start from a blank initial state of complete egalitarianism, to take into account the full extent of social complexity development.18 With every iteration of the recursive loop, various subsequent strategies and solutions become superimposed, eventually generating a costly “apparatus” consisting of multiple, partially overlapping, structures of administration, laws, and measures of socio-political organisation, but also of intricate sets of social norms and values, and various venues of communication between people, social groups, and central administration, all of which are costly to maintain. Every iteration of the loop, therefore, even if successful, requires more energy. In this sense, executing and maintaining older measures of socio-political development will often induce additional stimuli or challenges for the community, requiring ever more measures to be undertaken in an ever-flowing loop of complexity development. Complexity as a problem-solving tool for both external and internal disruptive events can therefore explain what seems like a “natural” tendency towards growing complexity in many social systems, whereas the infinitely more numerous potential pathways leading to failure of socio-political development and societal collapse (a potential state space associated with every subsequent step of the recursive loop) explain why only some societies ever developed a complex socio-political configuration, whereas many more did not. We must then consider what causal factors can be responsible for developing complexity within socio-economic systems. I will focus here on a limited number of variables which return frequently in economic literature: (1) supply and demand, (2) (human and physical) capital investment, (3) institutionalisation, (4) division of labour, (5) technological development, and (6) property rights. As we will see, these causal factors contain the inherent potential to increase socio-economic complexity through the aforementioned complexity mechanisms. For example, it has been noted that the development of new technologies often induces further technological innovation in response to the creation of new needs associated with the original innovation.19 As a result of such positive feedback loops, a new 18   Cioffi-Revilla, “A Canonical Theory of Origins and Development of Social Complexity,” 133. 19  Arthur, The Nature of Technology; Arthur, Complexity and the Economy.

170 

D. DAEMS

technology is not just a one-time disruption to the current system state, but rather a permanent ongoing generator of further technological innovations that induce still further technological development. However, for this loop to emerge, complexity mechanisms are needed to operate onto these causal factors, in this case diversity in functional needs. Before moving on to the case study, let us first discuss how to operationalise the approximation of complexity development in socio-economic systems through the framework outlined so far. Here, I will focus on approximating the intensity of the relevant causal factors contributing to social complexity development, applied through a comparison between Sagalassos and Düzen Tepe.

3   Methodology A rich body of literature exists on measuring complexity but it has proven difficult to construct a suitable and widely applicable method.20 A list of complexity measures compiled by Seth Lloyd discerns three main groups: difficulty of creation, difficulty of description, and degree of organisation.21 The first group measuring difficulties of creation is mainly related to human-made or engineered complex systems and therefore not very relevant for organically developing complexity in human systems. Many complexity measures from the second group (difficulty of description) come from the field of cybernetics.22 They are based on measures of communication information and system entropy in description length of a given system.23 While entropy measures of information description work great in theory, they are often cumbersome to calculate and therefore difficult to apply in practice.24 It is not my intention here to add onto such elaborate measures with a new technique, trying to improve on others in potency or elegance. Although conceptually attractive, the practical use of such measures has turned out to be rather limited. Instead, I will attempt to provide a very basic way of approximating the intensity of certain causal factors in developing socio-economic complexity through mechanisms as diversity, dimensionality, and connectivity. This approach is more closely  Page, Diversity and Complexity, 27.  Lloyd, “Measures of Complexity: A Nonexhaustive List”. 22  Castellani and Hafferty, Sociology and Complexity Science, 115. 23  One seminal work is Shannon’s entropy equation in information theory, see C. E. Shannon, “A Mathematical Theory of Communication”. 24  Page, Diversity and Complexity. 20 21

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

171

related to the third group of measurements by degrees of system organisation. Measurement, by definition, has a connotation of objectivity and precision. If a phenomenon can be measured meaningfully, then it can be compared to any other phenomenon that is meaningfully measurable by comparable units of measurement. One particular approach to complexity, however, uses subjective measures of development.25 In this view, the degree of complexity of a system depends on available frames of reference starting from a principle of “reference simplicity”. This makes sense as a given system can only be considered complex insofar it can be compared to others that are perceived as simple. The equation goes: K  S   F    S  , D  SR  



where a subjective measure of system complexity K is a function (F) of inputs μ (size of the minimal description in a given context) and D (distance function).26 The proposed measure has the advantage of being able to compare just two cases, whereas more common comparative statistical methods used to measure distance between variables—such as cluster analysis—generally require a larger sample size to be effective.27 However, we cannot just conceptualise any distance of system change compared to a given input value. We must also make sure that any such distance is effectively contributing to system complexity. Any distance measure of social complexity must therefore be related to the mechanisms of system complexity outlined above: diversification, dimensionality, and connectivity. The present argument is an elaboration of an earlier paper where I proposed an (overly) simplified measure of complexity development based on the distance between two social systems, one reference system and a comparative system.28 Here, I intend to build on that approach. For each qualitative parameter of comparison, an evaluation is given for both systems. Next the intensity of development, that is, the distance, needed to get from the reference value to the comparison value, is approximated. In the  Efatmaneshnik and Ryan, “A General Framework for Measuring System Complexity”.  A distance function defines differences between pairs of types; see for example Weitzman, “On Diversity”. 27  No written rule exists but a general rule of thumb is 2m samples (where m = number of clustering variables). 28  Daems, “A Matter of Formalitie”. 25 26

172 

D. DAEMS

Table 6.1  Coding of intensity measures of development Nominal Range

Very low 0–0.2

Low 0.2–0.4

Moderate 0.4–0.6

High 0.6–0.8

Very high 0.8–1

previous paper, I used a ratio scale ranging from −3 to +3 to evaluate this distance. One strongly impeding factor in any attempt at an explicitly quantitative approach to archaeology, however, is that the archaeological data often do not allow a precise estimation of the extent and scope of a given process. This is why many archaeologists prefer to work with more ambiguous valuations such as “very low”, “low”, “moderate”, “high”, “very high”. Due to the nature of the archaeological record, such evaluations are probably unavoidable. Unfortunately, due to imprecise and uneven use of such denotations, sometimes even within the same publication, comparison is often difficult. In addition to the +/− system, therefore, I propose to ascribe a fixed numerical valuation ranging between 0 and 1 (Table  6.1) to all nominal evaluations.29 Using this fuzzy set of numerical values, we can clarify how different processes compare to one another through the consistent use of a measurement indicator. Subtracting for each parameter the numerical value of the reference system from the value of the comparative system then gives a value for the distance or intensity of this specific process. This intensity can provide an indication for the degree of potential generated by each causal factor for inducing further system complexity. By comparing intensities of development, we can determine which elements of the socio-economic systems at both communities contributed most to overall system complexity.

4   Results: Socio-economic Systems at Düzen Tepe and Sagalassos In this part I will present the results of a case study focusing on the earliest phases of habitation at Sagalassos and Düzen Tepe (southwest Anatolia). Both settlements were located at a distance of 1.8 km from each other, on the fringes of the Ağlasun river valley. The settlement at Düzen Tepe was studied by the Sagalassos Archaeological Research Project (then directed 29  For a similar methodological procedure, see Torvinen et al., “Transformation without Collapse”.

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

173

by Prof. Marc Waelkens, now under direction of Prof. Jeroen Poblome), through multidisciplinary surveying campaigns coordinated by Hannelore Vanhaverbeke in 2005 and 2006, followed by excavations between 2006 and 2011 coordinated by Hannelore Vanhaverbeke and Kim Vyncke.30 Excavations at Düzen Tepe revealed houses mainly built from organic material such as mudbrick, but with limestone fieldstones used for the foundations and lower parts of the walls. Structures were identified over an area of around 75 hectares, with a core settlement built-up area consisting of 200 structures extending over approximately 15 hectares. A fortification wall was found, starting from the north-eastern side and covering also the southern and south-western sides of the settlement. Towards the north and northwest, the settlement was protected by the steep slopes of Mount Zencirli. Based on preliminary ceramic evidence, coin finds, and radiocarbon dating,31 a maximum occupation date between the fifth and second century BCE was determined for Düzen Tepe, with recent material studies suggesting a core occupation period during the fourth and third centuries BCE.32 Multidisciplinary research has been conducted at Sagalassos ever since its (re)discovery in the 1980s during the Pisidia Survey Project (under direction of Prof. Stephen Mitchell) and the start of the excavations in 1989. This long research history has resulted in a significant understanding of the Roman imperial to early Byzantine phases of urban development at Sagalassos. For its earlier phases, comparatively less evidence is available. Recent material studies determined that the oldest body of ceramics found at the site are datable to the late Achaemenid and early Hellenistic period (fifth to third centuries BCE), but likely to be situated mainly from the fourth century BCE onwards, based on fabric and typological features.33 Unfortunately, due to stratigraphical superposition and often large-scale and invasive building operations of later phases, few architectural remains can be associated with these finds. From around 200 BCE onwards, the first monumental urban fabric was constructed at Sagalassos, which developed into a prominent regional urban hub in

 Vanhaverbeke et al., “Pisidian Culture”.  Ibid. 32  Poblome et al., “How Did Sagalassos Come to Be; Daems, Braekmans, and Poblome, “Late Achaemenid and Early Hellenistic Pisidian Material Culture from Düzen Tepe. 33  Daems and Poblome, “The Pottery of Late Achaemenid Sagalassos”. 30 31

174 

D. DAEMS

Hellenistic and Roman imperial times.34 Düzen Tepe, on the other hand, was abandoned during the second century BCE, roughly at the same point when developments at Sagalassos started to take off. This remarkably divergent process has puzzled archaeologists for some time now. Due to the complete lack of architectural remains, it is difficult to draw strong conclusions on the extent of the community at Sagalassos during its late Achaemenid and early Hellenistic phases. Still, the sizeable quantity of retrieved locally produced ceramics suggests the existence of a relatively extensive community.35 It has been noted that this pottery shows remarkable similarities to that of Düzen Tepe, both in typological spectrum and fabric composition, suggesting a similar socio-economic productive matrix.36 It was therefore suggested that both settlements were at this time very similar village communities, operating within the same overall societal framework.37 However, this means that if we are to uncover the underlying factors for the initial impetus of development at Sagalassos—as well as approximate the intensity of these factors for generating the necessary potential to sustain this development—we must compare with our evidence from Düzen Tepe for approximating its initial state. The reference point for our comparison will be Düzen Tepe—as a proxy by extension for the habitation phase during the late Achaemenid and early Hellenistic periods (fifth to third centuries BCE) at Sagalassos— to uncover the relevant causal factors as drivers of development. This reference point will then be contrasted with the subsequent system state, that is, the habitation phase during middle Hellenistic times (second century BCE) at Sagalassos, to determine the intensity of development. Again, any comparison of system dynamics in both periods of time can only be conducted under the assumption that both communities operated on a similar level of socio-economic complexity prior to 200 BCE. I will provide additional evidence for the validity of this assumption in the following sections of this paper. We will specifically look at three major components of the “chaîne opératoire” of pottery production and consumption as a proxy for 34  Poblome et  al., “How Did Sagalassos Come to Be”; Daems et  al., “The Hellenistic Ceramics of Sagalassos”. 35  Braekmans et al., “Reconstructing Regional Trajectories”; Daems and Poblome, “The Pottery of Late Achaemenid Sagalassos”. 36  Daems, Braekmans, and Poblome, “Late Achaemenid and Early Hellenistic Pisidian Material Culture from Düzen Tepe”; Daems and Poblome, “The Pottery of Late Achaemenid Sagalassos”. 37  Daems and Poblome, “Adaptive Cycles in Communities and Landscapes”.

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

175

the overall socio-economic complexity at both communities: resource procurement, material production, and distribution.38 Clearly, the fourth major domain, subsistence, and its importance as the economic basis of agricultural societies merit a full discussion in its own right. The three domains discussed here offer a window on economic practices and choices performed by members of the local community, embedded in the constraints and opportunities of their wider social, political, economic, and ecological framework. 4.1  Resource Procurement and Exploitation Numerous clay beds are present at the sites of both Sagalassos and Düzen Tepe as well as in the surrounding area, although with varying suitability for pottery production. Petrographic analysis of the pottery found throughout the wider research area39 has identified four regional ceramic production groups based on petrology and clay chemistry: (A) Burdur basin groups, (B) detrital clay groups from the Çanaklı and Ağlasun basin, (C) a mixed flysch-limestone group, and (D) an ophiolitic-volcanic group.40 The fine clays derived from the more distant Burdur plain are only marginally attested at Düzen Tepe41 and not at all at Sagalassos so far. The bulk of the late Achaemenid and early Hellenistic material found at Düzen Tepe and Sagalassos was made from clays derived from the sites themselves or from the immediate vicinity in various parts of the Ağlasun valley. The flysch-limestone fabric group was produced with clays derived from weathered bedrock found on the flanks of the mountain ranges surrounding the Ağlasun and Çeltikçi valleys.42 Clay quarrying has been attested in the central depression of what would become the Eastern Suburbium of Roman imperial Sagalassos, where core drilling indicated the development of a palaeosol horizon on top of a clay quarry phase that could be dated to

 Costin, “Craft Specialization”.  Here the research area of the current Sagalassos Archaeological Research Project, more or less coinciding with the territory controlled by Sagalassos in Roman imperial times. 40  Braekmans et al., “Reconstructing Regional Trajectories”. 41  Only eight diagnostic pieces were identified by the author, mainly related to a bowl functionality. 42  Neyt et al., “Long-Term Clay Raw Material Selection and Use in the Region of Classical/ Hellenistic to Early Byzantine Sagalassos”. 38 39

176 

D. DAEMS

370–200 BCE.43 This terminus ante quem for the quarrying activities suggests that these clays could have been in use in late Achaemenid and early Hellenistic times. Moreover, control excavations conducted at the Upper Agora of Sagalassos confirmed a large anomaly, previously identified through geophysical research, to be related to the fill of a large pit resulting from clay quarrying activities before the construction of a public square at this location.44 Although it cannot be proven conclusively at this point that these specific quarries were necessarily exploited for pottery production, it does seem plausible that at least part of the clay raw materials were used by potters, as ceramics attributed to this clay group represent the bulk of production of common wares and buff tablewares during late Achaemenid and early Hellenistic times. Pottery related to the ophiolitic-­ volcanic trace element group can be associated with the entire range of common wares found at Düzen Tepe. Specifically, the illite-rich ophiolite clay beds from the immediate vicinity of the settlement were used to produce the ceramics associated with this group.45 Interestingly, no tablewares seem to have been produced with these clays. The majority of tablewares at Düzen Tepe were produced from the flysch-limestone clays derived from the immediate vicinity of the site. A small portion of the tableware assemblage of Düzen Tepe, however, was made from detrital clays derived from the northwestern parts of the Çanaklı valley (located at a distance of four to five km from Düzen Tepe). As this relates to less than 1 per cent of the total amount of sherds found and studied at Düzen Tepe, exploitation of these clays can be considered as ephemeral compared to the majority of the local production. The potters at Düzen Tepe are thus presumed to have operated within a least-effort productive framework, where mainly those resources in the immediate vicinity of the settlement were targeted and exploited. At Sagalassos, largely the same picture emerges for the late Achaemenid and early Hellenistic periods, with a majority of the pottery material pointing towards the use of clays from the immediate vicinity of the site. This image starts to change towards end of the third century BCE, with the 43  Vermoere et  al., “Pollen Sequences from the City of Sagalassos”; more specifically: 2210±50 BP 14C date with 1σ confidence interval. 44  Talloen and Poblome, “The 2014 and 2015 Control Excavations on and around the Upper Agora of Sagalassos”. 45  Neyt et al., “Long-Term Clay Raw Material Selection and Use in the Region of Classical/ Hellenistic to Early Byzantine Sagalassos,” 1301–02; Braekmans et  al., “Reconstructing Regional Trajectories”.

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

177

development of a fine tableware fabric, which can be seen as the precursor of the local production of Sagalassos Red Slip Ware (SRSW) in Roman imperial times.46 Petrographic analysis conducted by the Center for Archaeological Sciences (University of Leuven), on some late Hellenistic sherds, indicated two provenance areas for the clay raw materials of this fabric.47 Besides local clay beds found at the site or its immediate environment, a component of this production also made use of greenish detrital clays originally accumulated as part of a sequence of lake deposits derived from the northwestern parts of the nearby Çanaklı valley (located at seven to eight km from Sagalassos). The associated tableware fragments from a body of ceramics found in control excavations at the Upper Agora, dated to the later third to early second centuries BCE, are produced almost exclusively in this well-levigated fabric.48 At this time, the systematic occurrence of pottery produced with these more distant clays is symptomatic for more consistent and controlled strategies of resource procurement and clay preparation for the production of the higher-end spectrum of finer tableware.49 This could be an indication for a more developed and extended raw material economy. It remains unclear, for now, whether the systematic exploitation of these more distant clays is only a sign of the increased catchment area upon which Sagalassos depended, or whether this development was matched by a genuine territorial increase in a political sense as well. The first clear indication for the establishment of a political territory can be found in the writings of Livy, who describes the expeditions of the Roman general and consul Gnaeus Manlius Vulso as he crossed large parts of southwestern Anatolia in the aftermath of the battle of Magnesia (190 BCE) to move against the Galatians and passed the territory of Sagalassos. The marshlands where Manlius Vulso is said to have approached the borders of the territory of Sagalassos50 can only have corresponded to the area

46  Poblome et al., “The Concept of a Pottery Production Centre; Degryse and Poblome, “Clays for Mass Production of Table and Common Wares, Amphorae and Architectural Ceramics at Sagalassos”. 47  Poblome et  al., “The Concept of a Pottery Production Centre”; Neyt et  al., “LongTerm Clay Raw Material Selection and Use in the Region of Classical/Hellenistic to Early Byzantine Sagalassos”; Braekmans et al., “Reconstructing Regional Trajectories”. 48  Daems et al., “The Hellenistic Ceramics of Sagalassos”. 49  Poblome et al., “The Concept of a Pottery Production Centre”. 50  Liv. 38. 15; Plb. 21. 36.

178 

D. DAEMS

immediately to the southwest of Lake Burdur, near modern Düğer.51 This would suggest that by 189  BCE, the territory of Sagalassos already extended all the way up to this point, including large parts of the fertile Burdur plain. Unfortunately, we have few indications of how and when the territory of Sagalassos was extended prior to this point in time. Recent material studies on the pottery material found during intensive surveys indicated that the majority of the material datable to the fourth and third centuries BCE found at numerous locations in the central parts of the Ağlasun valley could be linked to fabrics produced at Sagalassos. Düzen Tepe-related fabrics were only marginally present on a few locations closest to the site. This might suggest that, at least for the central parts of the Ağlasun valley, the majority of these lands were at this time mainly associated with Sagalassos rather than with Düzen Tepe. It is suspected that Düzen Tepe was mainly reliant on this western part of the valley. It can therefore be suggested that Sagalassos and Düzen Tepe relied mostly on the catchments immediately surrounding these sites—respectively the central parts of the Ağlasun valley and the valley of Yeşilbaşköy—for its subsistence and resource exploitation.52 The addition of (parts of) the fertile Burdur plain to the territory of Sagalassos would then have entailed a massive territorial increase unlike anything either settlement had seen before. Clearly, the exploitation of the energy potential derived from this far more extensive environment could have created the necessary base for an increasingly more potent hub of social dynamics and developments at Sagalassos from mid-Hellenistic times onwards. 4.2  Production Processes and Output The urban transformation occurring at Sagalassos around 200 BCE—possibly following an earlier socio-political phase of transformation53—not only impacted the built-up fabric of the town but is also associated with a profound change in material culture and production processes. Local production activity was attested at Düzen Tepe through the partial excavation of a workshop containing the remains of a dismantled kiln, likely related to pottery production. From this updraught kiln only the circular floor,  Waelkens and Loots, Sagalassos V.  Cleymans, Daems, and Broothaerts, “Sustaining People”. 53  Daems, “Building Communities”. 51 52

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

179

about one meter in diameter, consisting of a layer of fired clay was preserved.54 No stratigraphic association could be ascertained between the kiln and nearby structural remains. Strangely, the opening of the kiln is oriented towards the closest southwestern wall of the nearby structure, limiting the available space to operate the kiln to less than two meters, although it is hard to assess to what extent this would have actually impeded the activities of the artisans working the kiln. It is also possible that the structure was not yet present at the time the kiln was in use, or that this orientation was constructed intentionally for reasons unknown, perhaps related to ventilation and airflow.55 To what extent this structure was functionally linked to the production activities, or whether for example a combination with a domestic function can be supposed, is hard to assess. At Sagalassos, the remains of a similar kiln were discovered during excavations underneath the Roman Odeon. Pottery found in fill layers inside this dismantled kiln were dated to the end of the third century and early second century BCE.56 Given that the kiln had already been constructed, used, and abandoned, the existence of pottery production facilities at this location can be assumed to date back already to the third century BCE. As in Düzen Tepe, the structure likely consisted of a basic updraught kiln structure. Geophysical research revealed a number of anomalies in the vicinity of the excavated kiln. While so far no excavations have taken place at these locations, these anomalies can likely be related to other pottery kilns. If so, it might be suggested that already from the third century BCE onwards, this area was reserved for pottery production as a potters’ quarter.57 Geomagnetic surveys at Düzen Tepe yielded a number of magnetic anomalies throughout the settlement which might be linked to the presence of burnt clay.58 Whereas the presence of other kilns cannot be excluded, some of these anomalies are probably too small to be linked to remains of (pottery) kilns. Trace element analysis of approximately 100 soil samples collected from across the site moreover seems to suggest a connection with metalworking activities, possibly ore smelting.59 It can therefore not be excluded that certain of these anomalies were connected  Waelkens et al., “The 2010 Excavation and Restorations Campaigns”.  Vyncke, “Düzen Tepe,” 163. 56  Poblome et al., “How Did Sagalassos Come to Be”, 180–83. 57  Poblome et al., 177. 58  Waelkens et al., “The 2010 Excavation and Restorations Campaigns”, 177–90. 59  Vyncke et al., “The Metal Production at Düzen Tepe”. 54 55

180 

D. DAEMS

with metallurgy processes. Given their location strewn between domestic structures throughout the settlement, we do not have the same indications to suggest the presence of a distinct, spatially delineated area for craft activities at Düzen Tepe, as we have for mid-Hellenistic Sagalassos. Such a reserved area for production facilities, with multiple workshops operating simultaneously, would allow a markedly increased production output at Sagalassos from the late third to early second centuries BCE onwards. Full-time production activities, as for agricultural activities, were a priori impossible in this area, where climatic circumstances characterised by long, very cold winters with much snow and short dry summers60 would not have allowed year-round production, implying that seasonal production must have been the norm. Shifts between agricultural and production activities throughout the year are therefore quite likely. Production processes were presumably carried out by a small number of artisans, as the majority of population at Düzen Tepe consisted of farmers or herders who were mainly preoccupied with subsistence strategies, operating in a smallholder system.61 More important than trying to delineate time investment exactly in one or the other, however, is to consider to what extent people were economically dependent on either agriculture or artisanal production for their subsistence. This degree of dependence can be surmised from the degree of production specialisation and radius of distribution of the resultant production output. For late Achaemenid times, only a limited amount of material has been retrieved from Sagalassos. Although almost no stratigraphically secure contexts from the late Achaemenid period have been identified (except for a few contexts associated with a terrace wall in the eastern parts of the site), a small number of fragments have been found either in excavations as associated residual material in later contexts or as surface material during intensive urban surveys. Due to the nature of the find contexts, it is often quite difficult to securely date this material. Only a few fragments could be assigned unequivocally to the late Achaemenid period (late fifth to fourth centuries BCE), mainly based on properties of fabric and slip. The majority of this material is more generally considered late Achaemenid to early Hellenistic (fourth to third centuries BCE) in date.62 Most of these fragments are related to a jar or a vessel with storage or cooking  Paulissen et al., “The Physical Environment at Sagalassos,” 231.  Daems, and Broothaerts, “Sustaining People”. 62  Daems and Poblome, “The Pottery of Late Achaemenid Sagalassos”. 60 61

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

181

functionality, with only few attestations of tableware. The overall nature of this material, both in typological variation and in technical features such as slip and fabric use, appears to be quite similar to that of contemporary Düzen Tepe. The far larger amounts of material found there allow a more extensive analysis to be made, beyond the more descriptive work for the contemporary pottery of Sagalassos. The pottery of Düzen Tepe was characterised by low product standardisation, resulting in a high degree of variability in vessel dimensions, even within individual types.63 For example, the rim diameter of Achaemenid bowls found at Düzen Tepe64 ranged between twelve and twenty-four centimetres, with an average of eighteen centimetres. Almost no specific wares can be associated uniquely with a specific fabric, nor with specific parts of the overall functional assemblage. Most fabrics cover large parts of the full typological assemblage, although a few exceptions of more specialised production such as the black-glazed pottery and cookware do exist. Instead, we have identified only a relatively small number of types within a basic spectrum of forms that re-occurred throughout different fabrics, stressing the generic nature of the material. High variability in fabric compositions, vessel dimensions, fabric-function associations, and a generally low degree of standardisation together suggest that little specialisation can be found throughout the different steps of the productive process. This suggests that the artisans at Düzen Tepe generally invested little additional labour efforts towards producing specific and specialised goods, preferring instead to supply a generic product line. These production strategies were not geared towards wider structures of exchange but mainly aimed at fulfilling the basic needs of the local community. This is corroborated in the observed distribution patterns of this pottery material (see next part). It can therefore be suggested that the general nature of these production processes and the resultant material culture would best fit a more village-like nature of settlement. Artisanal production at Düzen Tepe was therefore likely conducted in addition to agricultural activities, which constituted the bulk of investment in time and labour. Insofar we can draw any strong comparisons from the more

63  Daems, Braekmans, and Poblome, “Late Achaemenid and Early Hellenistic Pisidian Material Culture from Düzen Tepe”. 64  Out of the total 97 identified pieces, a sample of 18 fragments was usable for measurements as for these sufficient parts of the rim have been preserved.

182 

D. DAEMS

limited amounts of late Achaemenid to early Hellenistic material at Sagalassos, both bodies of pottery show strong similarities.65 Along with the observed changes in production infrastructure at Sagalassos from middle Hellenistic times onwards, we also see marked changes in the resultant output of material culture dated to this period of the late third to early second centuries BCE. The pottery material associated with the pottery kiln found underneath the Odeon and a number of contexts from control excavations conducted at the Upper Agora66 have yielded a coherent body of material indicating marked developments compared to the earlier material at Düzen Tepe and Sagalassos. Whereas previously, almost the full typological range was covered by multiple fabrics, from this point onwards, a more defined typological division between tablewares and coarse wares can be observed. This is a clear indication of stronger functionally specific associations between fabric and end product. Moreover, we see that for the production of tableware, the potters of mid-­ Hellenistic Sagalassos increasingly started to employ the finer, well-­ levigated clays from the northwestern parts of the Çanaklı valley.67 This is indicative for the development of a more extensive raw material economy at the time. Coarse wares from mid-Hellenistic Sagalassos show the same range of poorly sorted inclusions. Compared to earlier times, however, these occur in notably lower quantities and are generally smaller and more rounded. Pores as well became smaller and less elongated. As a result, these Hellenistic coarse ware fabrics have a relatively more fine-grained overall texture. These changes may be linked to more extensive preparations during the productive process. Additional preparation of clays and inclusion material enhances plasticity, producing better shapeable clay pastes and allowing more precision and refinements to be applied to the objects being produced. By forming a more regular and uniform base material, its properties become more predictable, controllable, and suitable during forming and firing in (large-scale) production processes.68 Additional preparation

 Daems and Poblome, “The Pottery of Late Achaemenid Sagalassos”.  Talloen and Poblome, “The 2014 and 2015 Control Excavations on and around the Upper Agora of Sagalassos”. 67  Braekmans et al., “Reconstructing Regional Trajectories”; Daems et al., “The Hellenistic Ceramics of Sagalassos”; Poblome et al., “The Concept of a Pottery Production Centre”; Poblome, “The Potters of Ancient Sagalassos Revisited”. 68  Orton and Hughes, Pottery in Archaeology, 125. 65 66

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

183

measures performed during the production process are therefore an essential step for a more extensive and standardised production output. When looking at intended functionality of these objects, it is no surprise that both at Düzen Tepe and at Sagalassos, the full spectrum of domestic activities related to day-to-day use of pottery is present in the observed assemblage. We need to go a step further and see whether we can trace differences in variation within each functional header. We could for example look at the number of types identified for each of the functional groups, under the assumption that two different types within the same type group might be interpreted as indications for consumer choice. In this sense, the nature of the objects being produced hinges on prevalent patterns of consumption (in part) determined by the socio-economic roles available to the community.69 Looking at the major components of household functional assemblages—consumption, serving, storage, and cooking—a more diversified spectrum of shapes with an increasing number of specifically designed forms is produced in mid-Hellenistic Sagalassos, especially for the tablewares (as summarised in Table 6.2). For example, whereas at Düzen Tepe most open tableware forms ranged between bowls and dishes of variable sizes, with only the so-called Achaemenid bowl attested as a clear type of drinking cup, at mid-­ Hellenistic Sagalassos two additional types of drinking cups were identified in the form of mastoid cups and hemispherical cups. In general, typological variety at mid-Hellenistic Sagalassos was equal or higher in Table 6.2  Summary of the number of types per functional group, in the two different periods Functional category

Functional group

Düzen Tepe

Sagalassos

Consumption

Cups Bowls Dishes Jars Open containers Pithoi Jars Cooking vessels

1 4 7 7 3 3 5 4 34

4 4 8 11 3 3 5 7 45

Serving Storage Cooking Total

 Costin, “Craft Specialization”.

69

184 

D. DAEMS

every functional group compared to Düzen Tepe and early Sagalassos. Whether or not the noted typological differentiation is solely a reflection of distinct choices made by consumers or whether other factors were at play as well can at this point not conclusively be answered. We can however at the very least conclude that potential for choice diversity was higher in Sagalassos compared to Düzen Tepe. 4.3  Structures of Exchange Some of the elements discussed so far regarding the procurement of raw materials and the organisation of the production process can be seen as indicative examples for the predominantly locally oriented community at Düzen Tepe.70 This general conclusion is also corroborated by observed pottery distribution patterns. It is interesting to note that the distribution of pottery produced at Düzen Tepe is mostly limited to the site itself, while surveys in the adjoining Ağlasun valley system (although only with partial coverage of the valley lands surrounding the site71) show them to be only marginally present and even there only at those locations closest to the site and decreasing sharply as the distance from the site increases.72 Although import is attested occasionally at Düzen Tepe, it constitutes only a minor part of the total pottery assemblage and is mainly associated with specific vessel types such as Achaemenid bowls. In a recent study of 623 diagnostic sherds, 5 out of 97 identified fragments of Achaemenid bowls could be linked to import. On the total body of material under study, about 2 per cent is considered to have been imported. Contacts with the outside world did exist, as can be deduced from a handful of coins from Erythraea, Magnesia, and Selge found at Düzen Tepe, but it remains difficult to assess the nature and scale of these contacts.73 Additionally, the large denominations of these silver coins suggest that they were not used in day-to-day transactions or trade.74 The limited attestations of glass objects in the excavations at Düzen Tepe75 suggest these were imported  Daems and Poblome, “Adaptive Cycles in Communities and Landscapes”.  Parts of the Yesilbaşköy valley were surveyed in the 2019 fieldwork campaign. Results are being processed but preliminary analysis has indicated few clear links with the communities of Düzen Tepe and Sagalassos in material culture. 72  Braekmans et al., “Reconstructing Regional Trajectories”. 73  Vyncke, “Düzen Tepe”, 217–18. 74  Stroobants, “The Long-Term Monetization of Sagalassos”. 75  Only nineteen glass fragments were found during six years of excavation. 70 71

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

185

rather than locally produced, as was customary for this period of time. In general, the mechanisms of distribution at Düzen Tepe were mainly aimed at basic subsistence exchange within the settlement itself, with, safe for a few exceptions, little incentive or intent to move into larger-scale networks of exchange. At Sagalassos, a markedly different picture emerges from mid-­Hellenistic times onwards. Pottery from Sagalassos was at that time distributed throughout the entire Ağlasun valley and gradually spread towards neighbouring valley systems as well, especially from the middle of the second century BCE onwards.76 Fine tableware produced at Sagalassos was notably encountered in a range of settlements to the south, both within and outside the borders of its newly enlarged territory. Pottery imports found at Sagalassos also became more extensive, with a wider functional range attested, from drinking cups to containers, jars, unguentaria, and most notably also amphorae. It has been noted how amphorae are completely absent from Düzen Tepe, whereas these are attested, albeit in limited quantities, at Sagalassos from middle Hellenistic times onwards. The appearance of amphorae originating from Rhodos, Kos, and Chios around 200  BCE has been linked to participation in larger-scale exchange networks, associated with the initial phase of urban development at Sagalassos.77 At the same time, a new institutional fabric developed alongside and within this new urban matrix. Interestingly, the earliest material reflections of institutional development at Sagalassos can be situated in the socio-economic domain and appear to be intrinsically related to aspects of exchange. During the second century BCE, existing clay quarries in the settlement were filled to allow the construction of a first public square or agora, traditionally considered as the heart of social, political, religious, and commercial activities.78 The agora as a space for public exchange facilitated political and economic activity outside the closely knit social network of neighbourhood, friendship, and kinship ties. Moreover, the agora acted as a central hub for flows of goods, services, and money, both internally within the community and externally in connections with markets abroad.79 Development of an agora has also been specifically related to the political  Poblome et al., “How Did Sagalassos Come to Be,” 535.  Monsieur, Daems, and Poblome, “Hellenistic and Italic Amphorae from Sagalassos”. 78  Talloen and Poblome, “The 2014 and 2015 Control Excavations on and around the Upper Agora of Sagalassos”. 79  Davies, “Ancient Economies: Models and Muddles”. 76 77

186 

D. DAEMS

“coming of age” of urban communities or poleis.80 It has been noted that a form of political community may already have been in place in the third century BCE prior to the observed monumentalisation of institutions.81 The formalisation of social interaction expressed through the construction of such settings allowed the civic administration to facilitate governmental control over commercial exchange and financial transactions on the agora, for purposes of taxation, regulation, safety of transactions, and surveying weighting and measuring. All this in stark contrast with Düzen Tepe where except for a few communal endeavours such as the construction of a fortification wall and a (communal) bakery, few clear indications for institutionalisation beyond the household level have been found and none related to wider economic exchange and distribution.

5   Discussion: Approximating Socio-economic Complexity In this final part, I want to integrate the archaeological observations described in the previous part with the theoretical framework outlined at the beginning of this paper. The socio-economic dynamics underlying the remarkable developments at Sagalassos from the (late) third to early second centuries BCE onwards were part of a wider process of transformation. This has traditionally been subsumed (partially) under the notion of urbanisation, but it can actually be subdivided into distinct socio-­economic processes driven by developments that were induced by a number of causal factors. I compare properties of the socio-economic system at Düzen Tepe (fifth to third centuries BCE) with Sagalassos (third to second centuries BCE) through the intensity of development in a number of variables. A summary of these variables, outlined to various degrees already in the previous part as well as in the following discussion, can be found in Table 6.3. In this discussion, I focus on six crucial causal factors of socio-economic development: (1) structures of supply and demand, (2) capital investment, (3) institutionalisation, (4) division of labour, (5) technological development, and (6) property rights. This paper is not primarily concerned with the discussion whether developments in any one of these factors effectively entails economic growth, be it aggregate or per capita. Still, each of these

 Starr, Individual and Community.  Daems, “Building Communities”.

80 81

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

187

Table 6.3  Parameters of socio-economic complexity Domain

Parameter

Düzen Tepe

Sagalassos

Exploitation Exploitation Exploitation Production Production Production Production Production Production Production

Opportunity costs Catchment area Different resources Division of labour Specialisation level Temporal specialisation Technology level Tool use Infrastructure specialisation Standardisation in object dimensions Specialisation fabric Fabric composition Specialisation typology Assemblage diversity Typological diversity: Consumption Typological diversity: Serving Typological diversity: Storage Typological diversity: Cooking Distribution Import Institutionalisation

Low Low Moderate Low Low Low Moderate Low Moderate Low

Moderate Very high High Moderate Moderate Low Moderate Moderate High Moderate

+ + + + + 0 0 + + +

0.2 0.6 0.2 0.2 0.2 0 0 0.2 0.4 0.2

Low High Low High 12

Moderate Moderate Moderate High 16

+ − + 0 +

0.2 0.2 0.2 0 0.2

10 8 4 Low Low Low

13 8 7 Moderate Moderate High

+ 0 + + + +

0.2 0 0.2 0.2 0.2 0.4

Production Production Production Output Output Output Output Output Exchange Exchange Exchange

Trend Intensity

causal factors can at least provide the necessary potential for further socio-­ economic complexity development. From the evidence outlined above, we may now conclude that the community at Düzen Tepe relied mainly on its immediate vicinity within a locally oriented productive landscape (be it on the plateau itself or in the nearby valley of Yeşilbaşköy) to sustain its various activities, including resource procurement, production, but also raising livestock, farming, and other subsistence strategies.82 Isotopic analysis has, for example, indicated that livestock was primarily kept together in the immediate vicinity of the settlement.83 Likewise, production output was first and foremost intended 82  De Cupere et  al., “Animal Exploitation during the Classical/Hellenistic Period at Tepe Düzen”. 83  Fuller et al., “Isotopic Reconstruction of Human Diet and Animal Husbandry Practices During the Classical-Hellenistic, Imperial and Byzantine Periods at Sagalassos”.

188 

D. DAEMS

to supply the own community, with only limited involvement in wider exchange networks. The overall impression of Düzen Tepe is one of an inward-oriented village community. When taking the full “ecology of subsistence strategies” as a starting point for a complexity economics perspective,84 we can say that only a limited number of different strategies were available in such a village community, where the majority of population was mainly occupied with agricultural production aimed at household subsistence. The urban context developing from mid-­Hellenistic times onwards at Sagalassos, on the other hand, would have allowed a slightly more diversified ecology of strategies consisting of more opportunities beyond agriculture, with more people earning a living as craftsmen, traders, and so on. Increasing division of labour therefore results in increase of complexity through increased diversity of composition in socio-economic roles and professions. An important element here is the opportunity costs associated with non-subsistence activities, for example pottery production. Given the generally low degree of labour specialisation, only a limited number of artisans/potters would have been present in Düzen Tepe, with the majority of population rather involved in general subsistence activities. The bulk of potential opportunity costs would therefore not have been associated with the nature of labour per se, but rather with the conversion of agricultural lands for resource exploitation. For all locations with suitable raw material sources, an assessment is needed to be made whether to invest in resource exploitation or leave the land for agricultural conversion. If certain lands were to be targeted for exploitation of raw materials, these would no longer be available for agricultural production. This means that opportunity costs associated with this decision would be somewhat higher in a farmer community like Düzen Tepe—thus acting as a constraining factor for innovation—compared to the urban community at Sagalassos, where more possibilities might be available for people to generate their own income outside of the agricultural sector. It was recently calculated that both Düzen Tepe and Sagalassos in late Achaemenid times had access to sufficient land to sustain their subsistence activities.85 The exploitation of certain parcels of land for clay procurement would then have depended mainly on the availability of suitable clay beds and somewhat less on the need to choose between different strategies (subsistence or raw material  Arthur, Complexity and the Economy.  Cleymans, Daems, and Broothaerts, “Sustaining People”.

84 85

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

189

exploitation). Opportunity costs at this time would therefore have been rather low. Given the relatively higher degree of division of labour at Sagalassos from mid-Hellenistic times onwards, the potential opportunity costs would by default have increased, as relatively more possibilities for the populace to earn a living in non-subsistence activities would have presented themselves. This development might allow people to diversify their income portfolio, leading to more extended land ownership as well as allow long-term clay exploitation on specific land plots rather than an exclusive use for agricultural cultivation. Whereas the production infrastructure does not seem to have developed significantly between the fifth and second centuries BCE, as the same type of updraught kiln appears to have remained in use, certain technological innovations do seem to have been initiated. The systematic use of fine clays allowed better slip, and more refined finishing and shaping of the vessels to take place, resulting inter alia in more thin-walled pottery in Hellenistic times at Sagalassos compared to Düzen Tepe. Perhaps the main differences between both technological systems, however, pertain to differences in organisational structures. Intensification of production in antiquity was typically achieved by multiplying small-scale production units rather than enlarging existing facilities.86 The organisation of different workshops in a spatially distinct zone devoted to artisanal activity would then have allowed sufficient critical mass to induce a process of production and labour specialisation, generating an increasing return on investment. However, sufficient incentives needed to be present to intensify production beyond basic subsistence needs. If demand is not high enough, the average cost per unit will remain high because of fixed production costs for products reaching only a limited customer pool.87 To what extent division of labour was applied to different production units to offset the associated cost increase—for example contributing to a combined effort for resource exploitation and gathering as may perhaps be expected from the increasingly specialised use of Çanakli-based clay sources, rather than multiple individual efforts—remains unclear for now. The successful multiplication of production units through the establishment of a pottery production quarter observed at Sagalassos suggests that sufficient incentives of demand were at that time present or at least being  McCormick, Origins of the European Economy; Poblome, “Made in Sagalassos,” 349.  Acton, “Industry Structure and Income Opportunities for Households in Classical Athens,” 158. 86 87

190 

D. DAEMS

created in order to increase production output. Multiplication of production units then resulted in a positive feedback loop driving increased production output as long as demand continued to provide sufficient incentives. On a local scale, material culture generally operates within two different contexts of engagement: household and community.88 Accordingly, two different levels of economic contexts can be said to exist: domestic and political.89 In many pre-modern societies, domestic economies, characterised by a predominant focus on household subsistence and production, and inter-household reciprocity provided the economic base for a family-­ based social organisation.90 Political economy, on the other hand, constitutes an additional level where economic surpluses generated through material flows of goods are constricted and channelled through selective control measures and reinvested by social elites to create additional wealth in order to finance institutions of rule, construct status identity, and organise communal activities.91 Can the differences in socio-economic organisation between Düzen Tepe and Sagalassos and the increase in economic potential and system complexity be explained as the elaboration of the level of political economy, in addition to the continued existence of domestic economy? One way to try and trace the development of a political economy is through the emergence of institutions. An important explanatory factor for the increased economic potential of Sagalassos is undoubtedly its territorial expansion in mid-Hellenistic times, allowing a political and territorial claim over far more natural resources in function of their potential exploitation. A closely related advantage may have been that the extended territory could have allowed Sagalassos to reach a far larger potential customer pool. Unfortunately, we have only limited evidence regarding markets and other exchange structures on a local and regional scale in this period of time. Moreover, a long-term diachronic study on the material culture and settlement patterns of the Bereket valley, located in the southwestern part of the territory of Sagalassos, indicates that this area was structurally integrated only in Roman imperial times, and even then only

 Kohring, Odriozola, and Hurtado, “Materialising ‘Complex’ Social Relationships,” 107.  Earle and Kristiansen, Organizing Bronze Age Societies. 90  Vranić, “The Classical and Hellenistic Economy and the ‘Paleo-Balkan’ Hinterland,” 40. 91  Earle, Bronze Age Economics, 1. 88 89

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

191

incoherently so and for a relatively short period of time.92 We can therefore wonder whether the potential of this (assigned) expanded territory could have been efficiently exploited in Hellenistic times. Nevertheless, market diversity in general is an important element in the development of an economic system. Through the causal factor of supply and demand, diversity enters market exchange in three different ways: (1) diversity in what agents bring to buy and sell; (2) agents’ preferences for different goods; and (3) different adaptation to information, mainly in the form of prices.93 Although the exact structures of exchange are not known to us, some of their material reflections can be traced in the archaeological record. It has been noted how the appearance of amphorae in the archaeological record from Sagalassos from 200 BCE onwards suggests the initiation of participation in long-distance trade networks. Clearly, the shift from domestic to political economies resulted in a markedly different economic landscape even in  local communities. This need not necessarily mean that participation in such long-distance networks was a political or centrally driven process, but rather that people in the local community started to see and utilise a whole new range of possibilities to conduct their business. Such long-distance trade then contributes to economic development by increasing the effective size of markets reached by producers, enabling economies of scale and division of labour, as well as by enabling distributed and more complex manufacturing so that a wider range of goods may be produced in one place.94 It has also been noted how the range of pottery imports increased considerably at Sagalassos, compared to Düzen Tepe.95 Exchange in itself can be considered to have an important multiplier effect. Following the general non-zero-sum characteristics of communication and interaction,96 exchange has been argued to facilitate exploitation of diversity in the dynamics of supply and demand, as the sum of the marginal values of individual goods is greater after exchange has taken place than it was before.97 This general process has clear economic implications, as value is therefore not only created through production but also through the very act of exchanging goods.98  Kaptijn et al., “Societal Changes in the Hellenistic, Roman and Early Byzantine Periods”.  Page, Diversity and Complexity, 17. 94  Bowman and Wilson, Quantifying the Roman Economy, 30–31. 95  Daems et al., “The Hellenistic Ceramics of Sagalassos”. 96  Parsons, Social Systems and the Evolution of Action Theory. 97  Simmel, The Philosophy of Money. 98  Staubmann, “Self-Organization and the Economy”. 92 93

192 

D. DAEMS

The material configurations of trade and exchange generally only become archaeologically visible once they are institutionalised and social and political “rules” for economic exchange become fixed.99 One aspect of such institutionalisation entails the creation of permanent and fixed marketplaces to provide a formal setting and framework for these exchanges to take place.100 Interestingly, the agora, constructed in the second century BCE, is one of the oldest known instances at Sagalassos of such formalised material settings reflecting institutionalisation processes, testifying perhaps to the importance of commercial exchange in this community. It cannot be excluded that this phase of monumentalisation in stone at Sagalassos reflects the origin of a political community in an earlier phase of community formation during the third century CBE. Such formal settings reduce transaction cost related to information gathering since they bring together a large number of participating buyers and sellers (at least in periodic attendance, taking into account seasonality of production); hence they underwrite system development.101 Institutions can be considered a “petrification” of social practices.102 They provide a structural solution for frequently repeated actions, such as the exchange of goods, by reducing uncertainty and “noise” in communications by providing standardised structures of interaction.103 Of course, it has been recognised that institutionalisation does not necessarily enhance (continued) efficiency.104 Due to path-dependent structuration of its dynamics, institutions are costly to change and therefore tend to remain unchanged over longer periods of time. Whereas stability could at first offer suitable conditions for continued interactions to take place, it is prone to turn into rigidity when configurations remain unchanged—a characteristic feature of institutions—even if the circumstances within the rest of the system or its environment change. Processes of institutionalisation as well as increased specialisation are—among others—induced by the increase of internal and external connections within and between system components. At the same time, these trends often increase overall system rigidity to such a degree that the system may no longer be able to response adequately to disruptive events and break  Garraty, “Investigating Market Exchange in Ancient Societies,” 6.  Harris and Lewis, “Introduction”. 101  North, Institutions, Institutional Change and Economic Performance. 102  Turner, Human Institutions. 103  Fletcher, The Limits of Settlement Growth, 143–44. 104  Zuiderhoek, “Introduction,” 13–14. 99

100

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

193

down. Still, we suggest here that the system developments from late Achaemenid to mid-Hellenistic times sketched in this paper attest the transition towards increased institutionalisation and specialisation, generating additional potential and capital through the increased interconnection of system components, but with no indications that institutional rigidity had already started to set in. Throughout this paper I have sketched a number of developments in resource procurement, production, and exchange of pottery for the communities of Düzen Tepe and Sagalassos during late Achaemenid and Hellenistic times. The development of the latter into an urban community during the middle Hellenistic period is clearly reflected in each of these domains. Moreover, I have tried to indicate for each of these domains where the necessary capital, value, or potential might have been generated to sustain these developments. Table  6.4 summarises this argument by listing the most important causal factors for each of these parameters, Table 6.4  Causal factors and mechanisms of complexity development at Düzen Tepe and Sagalassos with indication of relative intensity of each process Parameter

Causal factor

Mechanism

Düzen Sagalassos Intensity Description Tepe

Resource Division of Diversity procurement labour

Low

Moderate

0.2

Resource Capital procurement

Connectivity

Very low

High

0.6

Production

Capital

Dimensionality Very low

Moderate

0.4

Production

Capital

Connectivity

Low

Moderate

0.2

Production

Supply and Diversity demand Supply and Connectivity demand Capital Connectivity

Low

Moderate

0.2

Low

High

0.4

Low

Moderate

0.2

High

0.6

Exchange Exchange Exchange

Institutions Dimensionality Very low

Opportunities generated by urbanisation Potentially exploitable territory Multiplication of production units Standardisation of production output Production output Potential customer pool Exchange networks Institutional development

194 

D. DAEMS

along with concordant mechanisms of complexity development, responsible for the observed development of increased potential/capital in the socio-economic system of Sagalassos. Of primordial importance were territorial increase and the associated access to additional resources and energy, an elaborated participation in exchange networks, an increased production output due to multiplication of production units, institutional development, and diversification of potential socio-economic roles through an increased division of labour. It should be remembered that the assigned intensity of development pertains only to a relative comparison between Düzen Tepe and Hellenistic Sagalassos. In the subsequent Roman imperial period, many of these factors would continue to develop on a hitherto unprecedented scale. Taking this development into account would of course strongly skew the intensity measures presented here for this earlier period. The purpose of this paper was not to present an absolute measure of complexity development, but rather to situate and interpret certain processes related to past socio-­ economic systems as observed from the archaeological record in one specific phase of societal transformation. Finally, I have left the matter of why development of the socio-­economic system and concordant system complexity occurred at Sagalassos but not at Düzen Tepe unanswered so far. Given the earlier assessment of complexity as a problem-solving tool, can the development of socio-economic complexity at Sagalassos perhaps be seen as a sign of successful adaptation to stimuli or disruptive events? It has been suggested elsewhere that due to the partial overlap of initial catchment areas of both communities, the development of Sagalassos into a full-fledged urban centre could have exceeded the carrying capacity of the local ecological system, resulting in a system configuration where two communities located so close to each other was no longer sustainable.105 More recent work seems to indicate however that this was not the case and that both communities would have had access to enough land to sustain necessary subsistence activities.106 Alternatively, the abandonment of Düzen Tepe and transformation of Sagalassos has been explained through the role of political decision-­ making processes. In an upcoming paper I elaborate on an earlier hypothesis107 by arguing for the possibility of a synoikismos between Düzen Tepe  Daems and Poblome, “Adaptive Cycles in Communities and Landscapes”.  Cleymans, Daems, and Broothaerts, “Sustaining People. Reassessing Carrying Capacity through the Socio-Ecological Metabolism of the Ancient Community at Düzen Tepe”. 107  Daems and Poblome, “Adaptive Cycles in Communities and Landscapes”; Waelkens, “Ein Blick von der Ferne”. 105 106

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

195

and Sagalassos, with the population of the former moving to and merging with the latter.108 This process may have induced the necessary economies of scale and associated multiplier effects of increasing returns by concentrating local system potential at Sagalassos rather than having it divided over two different communities. The key factor initiating this process might be associated with the Seleucid dynasty, which gained control over the general area during the third century BCE and enjoyed high popularity at Sagalassos. It is argued that a process of politicisation of the community at Sagalassos was induced through the relationship between the local community and overarching Seleucid administration, whose economic and political policies generated the necessary stimuli and incentives for the dynamics and developments described in this paper. Especially the rapid expansion of the dependent territory of Sagalassos could be explained this way, as the Seleucids would have required a reliable and trusted local partner to control the strategically important north-south corridor along the Lysis river, connecting the Phrygian hinterland with the Lycian coast and therefore would have been either actively intervening or at least passively condoning this development.109 The involvement of the Seleucids can—in the absence of epigraphical sources—only be tentatively posited for now. Still, regardless of whether higher socio-political levels such as that of the Seleucid kings provided the initial impetus or not, it must be stressed that the actual processes for generating the necessary capital and potential needed to sustain the observed system dynamics fit a model of local system dynamics driven by an active community involved in various processes of socio-economic complexity development. Acknowledgements  This research was supported by the Belgian Programme on Interuniversity Poles of Attraction, the Research Fund of the University of Leuven, and the Research Foundation Flanders. The author is a member of the Sagalassos Archaeological Research Project directed by Prof. Jeroen Poblome (University of Leuven, Belgium). The author wishes to thank Prof. Koenraad Verboven and Prof. Jeroen Poblome for the opportunity of presenting this paper in this volume and for their insightful comments and corrections which markedly improved earlier drafts.  Daems and Talloen, “Moving in Together”.  Daems and Poblome, “Adaptive Cycles in Communities and Landscapes”; Poblome, “The Potters of Ancient Sagalassos Revisited”; Poblome et al., “How Did Sagalassos Come to Be”; Waelkens, “Ein Blick von der Ferne”. 108 109

196 

D. DAEMS

Bibliography Acton, Peter. “Industry Structure and Income Opportunities for Households in Classical Athens”. In The Ancient Greek Economy: Markets, Households and City-States, edited by Edward M. Harris, David M. Lewis, and Mark Woolmer, 149–65. New York: Cambridge University Press, 2016. Arthur, Brian. Complexity and the Economy. Oxford: Oxford University Press, 2015. Arthur, W. Brian. The Nature of Technology: What It Is and How It Evolves. Simon and Schuster, 2009. Beinhocker, Eric D. The Origin of Wealth: Evolution, Complexity, and the Radical Remaking of Economics. Boston, MA: Harvard Business School Press, 2006. Bentley, Alex, and Herbert Maschner. Complex Systems and Archaeology. Salt Lake City: Utah University Press, 2003. Blanton, Richard E., Stephen A.  Kowalewski, and Laura M.  Finsten. Ancient Mesoamerica: A Comparison of Change in Three Regions. Cambridge: Cambridge University Press, 1993. Bowman, Alan, and Andrew Wilson. Quantifying the Roman Economy: Methods and Problems. Oxford Studies on the Roman Economy. Oxford: Oxford University Press, 2009. Braekmans, D., P. Degryse, B. Neyt, M. Waelkens, and J. Poblome. “Reconstructing Regional Trajectories: The Provenance and Distribution of Archaic to Hellenistic Ceramics in Central Pisidia (South-West Turkey)”. Archaeometry, 2017, n/a–n/a. https://doi.org/10.1111/arcm.12259. Castellani, Brian, and Frederic Hafferty. Sociology and Complexity Science: A New Field of Inquiry. Berlin, Heidelberg: Springer Berlin Heidelberg, 2009. Chapman, Robert. Archaeologies of Complexity. London: Routledge, 2003. Cioffi-Revilla, Claudio. “A Canonical Theory of Origins and Development of Social Complexity”. The Journal of Mathematical Sociology 29, no. 2 (2005): 133–53. https://doi.org/10.1080/00222500590920860. Cleymans, Sam, Dries Daems, and Nils Broothaerts. “Sustaining People. Reassessing Carrying Capacity through the Socio-Ecological Metabolism of the Ancient Community at Düzen Tepe, SW Turkey”. In Preparation. Costin, Cathy Lynne. “Craft Specialization: Issues in Defining, Documenting, and Explaining the Organization of Production”. Archaeological Method and Theory 3 (1991): 1–56. Daems, Dries. “A Matter of Formalities. A Basic Measure of Social Complexity in the Past”. In Instrumenta 60. Quantifying Ancient Exonomies, Problems and Methodologies, edited by José Remesal, Victor Calvo, and Juan Bermúdez Lorenzo, 89–106. Universitat de Barcelona, 2018. Daems, Dries. “Building Communities. Presenting a Model of Community Formation and Organizational Complexity in Southwestern Anatolia”. Journal of Anthropological Archaeology 56 (2019): 101–110.

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

197

Daems, Dries, Dennis Braekmans, and Jeroen Poblome. “Late Achaemenid and Early Hellenistic Pisidian Material Culture from Düzen Tepe (SW Anatolia)”. Herom 6, no. 1 (2017): 11–47. Daems, Dries, and Jeroen Poblome. “Adaptive Cycles in Communities and Landscapes: The Case of Sagalassos and Düzen Tepe During the Classical/ Hellenistic Period”. Archaeological Review from Cambridge 31, no. 2 (2016): 91–107. Daems, Dries, and Jeroen Poblome. “The Pottery of Late Achaemenid Sagalassos: An Overview”. HEROM. Journal on Hellenistic and Roman Material Culture 6, no. 1 (2017): 49–62. Daems, Dries, and Peter Talloen. “Moving in Together? Synoikismos and Modes of Community Development through Push/Pull Dynamics in SW Anatolia”. BASOR. In Review. Daems, Dries, Mark Van Der Enden, Jeroen Poblome, and Peter Talloen. “The Hellenistic Pottery Repertoire made at Sagalassos, SW Anatolia”. In Daily Life in a Cosmopolitan World: Pottery and Culture during the Hellenistic Period. Proceedings of the 2nd Conference of IARPotHP Lyon, November 2015, 5th - 8th, edited by Annette Peignard-Giros, 81–96. Vienna, 2019. Davies, John. “Ancient Economies: Models and Muddles”. In Trade, Traders and the Ancient City, edited by Helen M. Parkins and Christopher Smith, 225–56. London: Routledge, 1998. De Cupere, Bea, Wim Van Neer, Kim Vyncke, and Hannelore Vanhaverbeke. “Animal Exploitation during the Classical/Hellenistic Period at Tepe Düzen (SW Turkey): Preliminary Results”. 9: 404–10. Oxbow Books, 2017. https:// lirias.kuleuven.be/handle/123456789/581893. Degryse, Patrick, and Jeroen Poblome. “Clays for Mass Production of Table and Common Wares, Amphorae and Architectural Ceramics at Sagalassos”. In Sagalassos VI.  Geo- and Bio-Archaeology at Sagalassos and in Its Territory, 231–54. Leuven University Press, 2008. https://lirias.kuleuven.be/ handle/123456789/219211. Earle, Timothy K. Bronze Age Economics: The Beginnings of Political Economies. Avalon Publishing, 2002. Earle, Timothy, and Kristian Kristiansen, eds. Organizing Bronze Age Societies: The Mediterranean, Central Europe, and Scandinavia Compared. New  York: Cambridge University Press, 2010. Efatmaneshnik, Mahmoud, and Michael J.  Ryan. “A General Framework for Measuring System Complexity”. Complexity 21, no. S1 (2016): 533–46. https://doi.org/10.1002/cplx.21767. Feinman, G.M. “The Emergence of Social Complexity: Why More than Population Size Matters”. In Cooperation and Collective Action: Archaeological Perspective, 35–56. Boulder: University Press of Colorado, 2012.

198 

D. DAEMS

Fletcher, Roland. The Limits of Settlement Growth: A Theoretical Outline. Cambridge: Cambridge University Press, 1995. Freeman, Alan, and Guglielmo Carchedi. Marx and Non-Equilibrium Economics. Cheltenham, UK; Brookfield, WI: Edward Elgar, 1996. Fuller, Benjamin, Bea De Cupere, Elena Marinova, Wim Van Neer, Marc Waelkens, and Michael Richards. “Isotopic Reconstruction of Human Diet and Animal Husbandry Practices During the Classical-Hellenistic, Imperial and Byzantine Periods at Sagalassos, Turkey”. American Journal of Physical Anthropology 149 (2012): 157–71. Garraty, Christopher P. “Investigating Market Exchange in Ancient Societies: A Theoretical Review”. In Archaeological Approaches to Market Exchange in Ancient Societies, edited by Christopher P. Garraty and Barbara L. Stark, 3–32. Boulder: University Press of Colorado, 2010. Gerring, John. Social Science Methodology: A Unified Framework. 2nd ed. Strategies for Social Inquiry. Cambridge: Cambridge University Press, 2012. Harris, Edward M., and David Lewis. “Introduction: Markets in Classical and Hellenistic Greece”. In The Ancient Greek Economy: Markets, Households and City-States, edited by Edward M. Harris, David M. Lewis, and Mark Woolmer, 1–39. New York: Cambridge University Press, 2016. Holland, J. Hidden Order: How Adaptation Builds Complexity. Reading, MA: Basic Books, 1995. Holland, John H. Complexity: A Very Short Introduction. 1st ed. Oxford University Press, 2014. Kaptijn, Eva, Jeroen Poblome, Hannelore Vanhaverbeke, Johan Bakker, and Marc Waelkens. “Societal Changes in the Hellenistic, Roman and Early Byzantine Periods. Results from the Sagalassos Territorial Archaeological Survey 2008 (Southwest Turkey)”. Anatolian Studies 63 (2013): 75–95. https://doi. org/10.1017/S0066154613000057. Kohler, T. “Complex Systems and Archaeology (Santa Fe Working Paper)”. 2011. http://www.santafe.edu/research/working-papers/abstract/4be5898ac43 2f33db80e7e69a01daaa1/. Kohring, Sheila, C. Odriozola, and V. Hurtado. “Materialising “Complex” Social Relationships: Technology, Production and Consumption in a Copper Age Community”. In Socialising Complexity: Structure, Interaction and Power in Archaeological Discourse, edited by Sheila Kohring and Stephanie Wynne-Jones, 100–17. Oxford: Oxbow Books, 2007. Köse, Veli. Nekropolen und Grabdenkmäler von Sagalassos in Pisidien in hellenistischer und römischer Zeit. Brepols, 2005. Leonard, Robert D., and George T.  Jones, eds. Quantifying Diversity in Archaeology. 1st ed. Cambridge u.a.: Cambridge University Press, 1989. Lloyd, S. “Measures of Complexity: A Nonexhaustive List”. IEEE Control Systems 21, no. 4 (August 2001): 7–8. https://doi.org/10.1109/MCS.2001.939938.

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

199

Lucas, Gavin. Understanding the Archaeological Record. Cambridge: Cambridge University Press, 2012. McCormick, Michael. Origins of the European Economy: Communications and Commerce AD 300–900. Cambridge University Press, 2001. Mitchell, Melanie. Complexity: A Guided Tour. Oxford; New  York: Oxford University Press, 2009. Monsieur, Patrick, Dries Daems, and Jeroen Poblome. “Hellenistic and Italic Amphorae from Sagalassos”. HEROM 6, no. 1 (1 May 2017): 97–119. https://doi.org/10.11116/HEROM.6.1.4. Neyt, B., D.  Braekmans, J.  Poblome, J.  Elsen, M.  Waelkens, and P.  Degryse. “Long-Term Clay Raw Material Selection and Use in the Region of Classical/ Hellenistic to Early Byzantine Sagalassos (SW Turkey)”. Journal of Archaeological Science 39, no. 5 (May 2012): 1296–305. https://doi. org/10.1016/j.jas.2012.01.005. North, Douglass C. Institutions, Institutional Change and Economic Performance. Cambridge; New York: Cambridge University Press, 1990. Orton, Clive, and M.  Hughes. Pottery in Archaeology. Cambridge: Cambridge University Press, 2009. Page, Scott E. Diversity and Complexity. Princeton: Princeton University Press, 2010. Parsons, Talcott. Social Systems and the Evolution of Action Theory. Free Press [etc.] cop., 1977. Paulissen, Etienne, Jean Poesen, Gerard Govers, and Jan De Ploey. “The Physical Environment at Sagalassos (Western Taurus, Turkey). A Reconnaissance Survey”. Edited by Marc Waelkens and Jeroen Poblome. Sagalassos II. Report on the Third Excavation Campaign of 1992, 1993. https://core.ac.uk/ reader/34349941. Poblome, Jeroen. “Made in Sagalassos: Modelling Regional Potential and Constraints”. 355–63. Plus, 2006. https://lirias.kuleuven.be/ handle/123456789/160918. Poblome, Jeroen. “The Potters of Ancient Sagalassos Revisited”. In Urban Craftsmen and Traders in the Roman World, edited by Andrew Wilson and Miko Flohr, 377–404. Oxford University Press, 2016. https://doi. org/10.1093/acprof:oso/9780198748489.003.0017. Poblome, Jeroen, Dennis Braekmans, Marc Waelkens, Nalan Firat, Hannelore Vanhaverbeke, Femke Martens, Eva Kaptijn, Kim Vyncke, Rinse Willet, and Patrick Degryse. “How Did Sagalassos Come to Be? A Ceramological Survey”. In Studies in Honour of K. Levent Zoroğlu, edited by Mehmet Tekocak, 527–40. Antalya: Suna & Inan Kiraç Research Institute on Mediterranean Civilizations, 2013. Poblome, Jeroen, Patrick Degryse, Willy Viaene, Raoul Ottenburgs, Marc Waelkens, Roland Degeest, and J. Naud. “The Concept of a Pottery Production

200 

D. DAEMS

Centre: An Archaeometrical Contribution from Ancient Sagalassos”. Journal of Archaeological Science 29, no. 8 (1 August 2002): 873–82. Ragin, Charles C. The Comparative Method: Moving Beyond Qualitative and Quantitative Strategies. University of California Press, 1989. Scheidel, Walter. “In Search of Roman Economic Growth”. Journal of Roman Archaeology 22, no. 1 (2009): 46–70. Shannon, C.  E. “A Mathematical Theory of Communication”. The Bell System Technical Journal 27, no. 4 (October 1948): 623–56. https://doi. org/10.1002/j.1538-7305.1948.tb00917.x. Simmel, Georg. The Philosophy of Money. Reprint ed. Abingdon, Oxon; New York: Routledge, 1978. Sitte, R. “About the Predictability and Complexity of Complex Systems”. In From System Complexity to Emergent Properties, edited by Moulay Aziz-Alaoui and Cyrille Bertelle, 23–48. Springer, 2009. http://www.springer.com/materials/ mechanics/book/978-3-642-02198-5. Smith, Michael. “How Can Archaeologists Make Better Arguments?” The SAA Archaeological Record 15, no. 4 (2015): 18–23. Starr, Chester G. Individual and Community: The Rise of the Polis 800–500 B.C. New York NY: Oxford University Press, 1986. Staubmann, H. “Self-Organization and the Economy: A System-Theoretical Reconsideration of Georg Simmel’s Philosophy of Money”. In Chaos, Complexity, and Sociology: Myths, Models, and Theories, edited by Raymond A. Eve, Sara Horsfall, and Mary E. Lee, 79–90. London: SAGE, 1997. Stroobants, Fran. “The Long-Term Monetization of Sagalassos. Questioning Monetary Production, Circulation and Use from a Regional Perspective”. Unpublished Ph.D. dissertation, University of Leuven, 2017. Tainter, J. “Complexity, Problem Solving and Sustainable Societies”. In Getting Down To Earth: Practical Applications of Ecological Economics, edited by Robert Costanza, Olman Segura, and Juan Martinez-Alier, 61–76. Washington, DC: Island Press, 1996. Tainter, J. “Social Complexity and Sustainability”. Ecological Complexity 3 (2006): 91–103. Talloen, Peter, and Jeroen Poblome. “The 2014 and 2015 Control Excavations on and around the Upper Agora of Sagalassos: The Structural Remains and General Phasing”. Anatolica 42 (1 September 2016): 111–50. Torvinen, A., Michelle Hegmon, Ann Kinzig, Margaret Nelson, M.  Pepples, Colleen A.  Strawhacker, Karen Gust Schollmeyer, and L.  Swantek. “Transformation without Collapse: Two Cases from the U.S. Southwest”. In Beyond Collapse: Archaeological Perspectives on Resilience, Revitalization, and Transformation in Complex Societies, edited by Ronald K.  Faulseit, 262–86. Carbondale, IL: Southern Illinois University Press, 2016.

6  SOCIAL COMPLEXITY AND COMPLEXITY ECONOMICS: STUDYING… 

201

Turner, J. Human Institutions: A Theory of Societal Evolution. Totowa: Rowman & Littlefield, 2003. Vanhaverbeke, Hannelore, Marc Waelkens, Kim Vyncke, Véronique De Laet, S. Aydal, B. Mušič, B. De Cupere, et al. ““Pisidian” Culture? The Classical-­ Hellenistic Site at Düzen Tepe near Sagalassus (Southwest Turkey)”. Anatolian Studies 60 (2010): 105–28. Vermoere, Marleen, Simon Six, Jeroen Poblome, Patrick Degryse, Etienne Paulissen, Marc Waelkens, and Erik Smets. “Pollen Sequences from the City of Sagalassos (Pisidia, Southwest Turkey)”. Anatolian Studies 53 (1 January 2003): 161–73. https://doi.org/10.2307/3643093. Vranić, Ivan. “The Classical and Hellenistic Economy and the “Paleo-Balkan” Hinterland a Case Study of the Iron Age ‘Hellenized Settlements’”. Balcanica, no. 43 (2012): 29–50. Vyncke, Kim. “Düzen Tepe. The Potential of Contextual Analysis and Functional Space Analysis by Means of an Interdisciplinary Archaeological and Archaeometric Research at a Classical-Hellenistic Site”. KU Leuven, 2013. Vyncke, Kim, Branko Mušič, Patrick Degryse, and Marc Waelkens. “The Metal Production at Düzen Tepe (Southwest Turkey): An Archaeological and Archaeometric Study”. Open Journal of Archaeometry 2, no. 1 (16 June 2014). https://doi.org/10.4081/arc.2014.5461. Waelkens, Marc. “Ein Blick von der Ferne: Seleukiden und Attaliden in Pisidien”. Istanbuler Mitteilungen 54 (2004): 435–71. Waelkens, Marc, and Lieven Loots. Sagalassos V: Report on the Survey & Excavation Campaigns of 1996 and 1997. Leuven University Press, 2000. Waelkens, Marc, Kim Vyncke, Jeroen Poblome, Rob Rens, Elisabeth Murphy, Inge Uytterhoeven, Julian Richard, et  al. “The 2010 Excavation and Restorations Campaigns”. In Kazι Sonuçlarι Toplantιsι, vol. 33. Ankara, 2012. Weitzman, Martin. “On Diversity”. Quarterly Journal of Economics 107, no. 2 (1992): 363–405. Wilson, Andrew. “Approaches to Quantifying Roman Trade”. In Quantifying the Roman Economy: Methods and Problems, 213–49. Oxford Roman Economy Series. Oxford: Oxford University Press, 2009. Zuiderhoek, Arjan. “Introduction. Land and Natural Resources in the Roman World in Historiographical and Theoretical Perspective”. In Ownership and Exploitation of Land and Natural Resources in the Roman World, edited by Paul Erdkamp, Koenraad Verboven, and Arjan Zuiderhoek, 1–17. Oxford Studies on the Roman Economy. Oxford: Oxford University Press, 2015.

CHAPTER 7

A Method for Estimating Roman Population Sizes from Urban Survey Contexts: An Application in Central Adriatic Italy Dimitri Van Limbergen and Frank Vermeulen

1   Introduction A considerable amount of scholarly literature exists on how to estimate population figures from archaeological contexts.1 A common method is by multiplying an informed density figure by a particular habitation unit, that is either a (roofed) floor surface from a domestic structure,2 a single 1  Estimating populations from archaeological data has a long history. There is no room to dwell on this history here, but the interested reader is referred to Chamberlain, Demography in Archaeology; Drennan, Berry and Peterson, Regional Settlement Demography in Archaeology. 2  Casselberry, “Further refinement of Formulae for Determining Populations from Floor Area”; Kolb, “Demographic Estimates in Archaeology”; Le Blanc, “An Addition to Naroll’s Suggested Floor Area and Settlement Population Relationship”; Naroll, “Floor Area and Settlement Population”; Wiessner, “A Functional Estimator of Population from Floor Area”.

D. Van Limbergen (*) • F. Vermeulen Ghent University, Ghent, Belgium e-mail: [email protected]; [email protected] © The Author(s) 2021 K. Verboven (ed.), Complexity Economics, Palgrave Studies in Ancient Economies, https://doi.org/10.1007/978-3-030-47898-8_7

203

204 

D. VAN LIMBERGEN AND F. VERMEULEN

housing entity3 or a complete settlement area.4 One issue that commonly complicates such calculations particularly with regard to nucleated agglomerations such as hamlets, villages, towns and cities is the lack of a single fixed density figure that may be used to translate structural finds into demographic values. This is because there is considerable variability in the archaeological record in spatial layout and occupation density of these kinds of settlements in both inter-regional and chronological terms.5 As such, we cannot expect density figures calculated for any type of protohistoric or pre-industrial concentration of housing units to be of any use for estimating town populations in Roman Italy where the general historical information from contemporary authors lacks the detail needed at site level.6 Within Roman urban demographic modelling, multiplying room, house and town surfaces by an average number of people is a method that has been applied frequently in recent years. Such extrapolations are probably the best approach, as more indirect methods such as demographic projections based on cemeteries, theatre and amphitheatre capacities, or aqueduct volumes, have been shown to be more ambiguous.7 Still, the method is certainly not without its problems, as significant debate exists with regard to open spaces within cities or suburban areas and extramural habitation when no adequate excavation or survey information is available. The choice of population density multipliers, and their variability across space and time, can also be problematic. Nevertheless, this method remains the best one available, and urban surveys today are enhancing its potential. The critical point is to establish a reasonable density multiplier, or a range of multipliers for Roman cities. 3  Schacht, “Estimating Past Population Trends,” 119; Kardulias, “Estimating Population at Ancient Military Sites. 4  Curet, “New Formulae for Estimating Prehistoric Populations for Lowland South America and the Caribbean”; Kramer, Village Ethnoarchaeology; Morley, “Cities and Economic Development in the Roman Empire”; Sumner; “Population and Settlement Area”; Zorn, “Estimating the Population Size of Ancient Settlements”. 5  De Roche, “Population Estimates from Settlement Area and Number of Residents”; Kramer, Village Ethnoarchaeology; Postgate, “How Many Sumerians per Hectare”. 6  De Ligt, Peasants, Citizens and Soldiers. 7  Wilson, “City Sizes and Urbanisation in the Roman Empire”; Hanson, Urban Geography of the Roman World discusses the constraints of this method in more detail. For aqueducts in particular, see Duncan-Jones, The Economy of the Roman Empire. For a recent attempt, however, at estimating population size from palaeodemographic and burial data, see McIntyre, “Reconstructing population size in a Romano-British colonia”.

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

205

Until now, these efforts have unsurprisingly been mostly focused on the best-known Italian urban contexts, such as the towns of Pompeii,8 Ostia,9 Rome,10 and Cosa,11 resulting in density figures that range between 105 to 125 (Cosa and Pompeii) and 730 to 840 (Rome and Ostia) persons per urban hectare.12 Given the high degree of standardised town planning in Roman times—that is, the systematic application of an orthogonal grid13 and the application of certain housing modules in (colonial) towns14—we might reasonably expect other Roman towns in Italy to have had similar population densities. On the other hand, a gap of about 700 persons per hectare between the lowest and highest density figure is clearly a significant difference in demographic modelling, especially when exploring wider research questions such as the size of the Italian urban population in Roman times, or aspects such as the carrying capacity of Italian landscapes and regions, and the urban-rural split in Roman society. Towns like Rome, Cosa, Pompeii and Ostia have all been studied intensively over longer periods of time and, especially in the case of the latter two, extensive archaeological excavations have been able to detect their vestiges rather well thanks to an exceptional state of preservation. The situation for the majority of the Roman towns in Italy, however, is radically different and poses some significant problems. It is often the case that so little has remained of these towns—or has been recorded over the years— that we cannot identify much more than their possible location in the landscape. Hence, the detailed urban topography of these sites is virtually unknown to us. In other cases, when we do know the locations of the 8  Beloch, “Le città dell’Italia antica”; Fiorelli, Gli Scavi di Pompeii dal 1862 al 1872; Nissen, Pompeianische studien zur Städtekunde des altertums; Storey, “The Population of Ancient Rome”; Wallace-Hadrill, “Houses and Households”; Wallace-Hadrill, Houses and Society in Pompeii and Herculaneum. 9  Calza, Scavi di Ostia; Meiggs, Roman Ostia; Nibby, Viaggio antiquario ad Ostia; Packer, The Insulae of Imperial Rome; Storey, “The Population of Ancient Rome”. 10  Bairoch, “Urbanisation and the Economy in Pre-Industrial Societies; Brunt, Italian Manpower; Carcopino, Daily Life in Ancient Rome; Hopkins, Conquerors and Slaves; Lo Cascio, “The Population of Roman Italy in Town and Country”; Robinson, Ancient Rome; Stambaugh, The Ancient Roman City. 11  Fentress et al., “Cosa”; Hanson, Urban Geography of the Roman World is the first work to focus on the Roman Empire as a whole. 12  Russell, “The Population and Mortality at Pompeii” argues that most cities in the ancient world had population densities of between 100 and 120 persons per hectare. 13  Castagnoli, Orthogonal Town Planning in Antiquity. 14  Sommella, Italia antica.

206 

D. VAN LIMBERGEN AND F. VERMEULEN

ancient towns, the sites have been subjected to limited modern archaeological research and almost nothing can be said about their extent, internal layout or even their wider structuring. Furthermore, most of the Roman urban centres in Italy developed on hilltops or in other strategic points in the landscape that continued to be occupied post-antiquity. As such, these settlements now lie completely—or at least in part—beneath their medieval and modern counterparts. It goes without saying that in these cases the reconstruction of their original town plans remains problematic in many ways. Usually, the layout of the settlement can only be reconstructed in part by searching for continuity in the medieval and modern town plans. In addition, the amount of available archaeological information is often insufficient to accurately trace the entire course of the circuit wall, or to establish the extent of the built-up area. The location of extramural cemeteries and suburban rural villas may provide some indication in these cases, but generally these only allow an estimation of the limits of a town with a large margin of error. The knowledge about private architecture is also mostly fragmentary, with the position and size of the different residential spaces often remaining unclear and structural remains being limited to wall transects and sections of floor pavements. Thus, establishing an approximate figure for the population of these towns at any moment in their occupational history on the basis of the available archaeological evidence—let alone sketching their demographic evolution over time— remains highly problematic. Our best chances of obtaining more and better data undoubtedly lie in the town areas that were gradually abandoned in post-Roman times and subsequently remained free—in their totality or in part—from more recent constructions. It is obvious that even in these cases not all towns have been studied with the same intensity, or show their traces with the same clarity, but the list of sites in Italy that offer enough archaeological information and are receiving sufficient scientific attention is constantly growing.15 To illustrate what may be achieved with this kind of evidence in terms of demographic reconstructions, we will focus our attention in this paper on the Roman towns of Potentia and Trea, both located in central Adriatic Italy. Over the last fifteen years, these abandoned urban sites have been the subject of intensive aerial photographic and geo-archaeological survey research by a team from Ghent University within the framework of

 Vermeulen et al., Urban Landscape Survey in Italy and the Mediterranean.

15

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

207

the Potenza Valley Survey Project (PVS).16 They conveniently represent the two main categories of towns in Roman Italy: a newly founded colonia and an organically grown municipium respectively (Fig.  7.1). The first part of this paper sets out the criteria for calculating urban population figures from Roman archaeological contexts by discussing the main factors that influence the population density of a Roman town and by touching upon some issues that tend to complicate our calculations. The aggregate evidence for the two towns in our case study is then systematically reviewed. For each town, information is assembled on the extent of the urban surface; the share of public, religious and commercial space; and the dimensions of the insulae and the street grid. When possible, the systematic survey evidence is complemented with other archaeological (excavations), literary and epigraphic evidence. The main purpose of this section is to arrive at realistic figures for the total space taken up by urban infrastructure and public architecture in these two towns. This will establish the

Fig. 7.1  Map of the study area with the indication of all town sites mentioned in the text. (Map by D. Van Limbergen)  Vermeulen, “The Potenza Valley Survey”.

16

208 

D. VAN LIMBERGEN AND F. VERMEULEN

potential areas that were available for private architecture. For obvious reasons our efforts will concentrate on the best documented phase of these settlements, that is the Early/High Imperial period (first to second centuries CE) when both towns experienced their heyday. The following section then concentrates on how to best fill in these residential areas with houses and how to calculate the number of inhabitants. As the data on domestic structures are currently scarce for both towns, their residential layout will be reconstructed using information on the size and nature of private architecture from the wider central Adriatic area during this period. The concluding section of this paper frames the results of our calculations within the wider debate on Roman urban demography and sketches the potential of our methodology for other Roman urban sites in Italy and the Mediterranean.

2   Issues and Guidelines in Calculating Town Populations The first step in any urban demographic reconstruction is always to acquire as much archaeological data as possible on the physical extent of the towns in question. The best way to approximate the urban surface from the archaeological evidence is to calculate the area delimited by the town walls. However, not every town was necessarily provided with a circuit wall; sometimes the natural barriers of a site were considered sufficient for defensive purposes. It goes without saying that in these cases natural barriers offer some guidance, but these cannot be considered as absolute limits. Furthermore, the walled area did not necessarily have to coincide with the actual inhabited area as it was possible that the enclosed zone was not completely built-up, or that suburbs gradually developed because of a lack of space within the original enclosure. As such, assessing the precise extent of a particular town area can be confusing.17 Next, in order to calculate the number of residents that may be fitted within the deduced urban area, a number of variables need to be taken into account. First and foremost, the number of persons per urban hectare could vary considerably because of differences in the private architectural make-up of the town, such as variations in the type of housing and the presence or otherwise of one or multiple upper stories. For instance, the notable differences between the population densities proposed for towns  Hanson, Urban Geography of the Roman World, discusses this issue at length.

17

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

209

such as Cosa (105–125 p/ha) and Pompeii (105–303 p/ha) and cities such as Ostia (290–840) and Rome (580–730) can largely be explained by the dominance of domus-type buildings—with either only a ground floor (Cosa) or sometimes also an upper floor (Pompeii)—in the former towns and the frequent use of multi-storey insulae in the latter. A total of 24 larger (ca. 660 m2) and 224 smaller houses (approximately half the size of the larger ones) were documented for Cosa in the Late Republican period. Assuming a family unit of 10–12 persons for the former houses and a unit of 5–6 persons for the latter, the population of the approximately 14 hectares large town was proposed to have been in the range of 1360–1632 people, that is 105–125 persons per hectare.18 Over the years, many scholars have also made estimates for the population of Pompeii in the first century CE by extrapolating the number of excavated rooms/houses to the total town area (around sixty-six hectares). The most famous of these reconstructions is the one carried out by Wallace-Hadrill who estimated a population of approximately 10,000 people (151 p/ha) for the Campanian town.19 Other estimates range from 12,000 (181 p/ha) to 20,000 (303 p/ha).20 It is clear that the main factor influencing the lower and the higher estimates is the inclusion or otherwise of upper stories in the houses.21 On the other hand, common population estimates for Ostia—with a walled area of about 69 ha, that is about the same size as Pompeii—range from 20,000 to 58,000 people.22 The considerably higher derived population densities for the Roman harbour town, of between 290 and 840 p/ ha, find their explanation in the fact that the people of Ostia were mainly 18  De Graaf, Late Republican-Early Imperial Regional Italian Landscapes and Demography; De Ligt, “The Population of Cisalpine Gaul in the Time of Augustus”; Fentress et al., “Cosa”. 19  Wallace-Hadrill, “Houses and Households”; Wallace-Hadrill, Houses and Society. 20  For the lowest estimate, see Fiorelli, Scavi di Pompeii; for slightly higher estimates, ranging from 15,000 to 18,000 persons (227–272 p/ha), see Beloch, “Città dell’Italia”; Nissen, Pompeianische studien, proposes estimates between 18,000 and 20,000 (272 to 303 p/ha). A figure of between 10,000 and 12,000 is now thought to be the most accurate; see Storey, “The Population of Ancient Rome”. 21  Lazer, Resurrecting Pompeii; Wallace-Hadrill, Houses and Society. 22  The lowest figure is provided by Nibby, Viaggio antiquario: 20,000 inhabitants or 290 p/ha; substantially higher figures are offered in Calza, Scavi di Ostia: 36,000 people or 521 p/ha, and in Meiggs, Roman Ostia: a maximum of 58,000 inhabitants or 840 p/ha. Based on the analysis of the different residential units, Packer, Insulae, estimates the population at a maximum of 27,000 inhabitants (391 p/ha); see also Storey, “The Population of Ancient Rome”.

210 

D. VAN LIMBERGEN AND F. VERMEULEN

housed in residential buildings that had an average height of between two and a half to five stories.23 The same goes for Rome itself, whose population in Imperial times has been estimated—based mainly on the 200,000 recipients of free grain reported in textual sources24—as high as 800,000 to 1,000,000 inhabitants.25 With an urban area of about 1386 hectares enclosed by the Aurelian walls, such a population would have implied densities of between 570 and 720 p/ha.26 Again, these proportionally high densities can be connected with the presence of multi-storey buildings in the city from Republican times onwards.27 As such, it is important to understand whether the towns whose population we are trying to estimate resemble settlements such as Cosa and Pompeii in their private architectural layout rather than those of cities such as Ostia and Rome.28 In other words, it is fundamental to gather as much information as possible on the nature of the residential units in the centres under study. A second parameter that influenced the overall population density of a Roman town was the effective amount of space that was available for private housing. Indeed, the size of the residential area was affected by the extent of the areas reserved for other constructions, such as public, administrative, religious and commercial buildings.29 It is unlikely that Roman towns had a very strict zoning system—that is a system in which buildings with similar functions are deliberately grouped together in certain urban districts—and most sectors would usually have accommodated a mixture 23  De Ligt, “The Population of Cisalpine Gaul in the Time of Augustus”; De Ligt, Peasants, Citizens and Soldiers; Duncan-Jones, The Economy of the Roman Empire. 24  Suetonius, Aug. 40.2; Res Gestae Divi Augusti 15. 25  These are the figures we can find for the population of the city in Brunt, Italian Manpower; Hopkins, Conquerors and Slaves; Stambaugh, The Ancient Roman City; Bairoch, “Urbanisation”; Robinson Ancient Rome; and Lo Cascio, “The Population of Roman Italy in Town and Country”. A considerably lower estimate of less than 500,000 has been offered by Carcopino Daily Life; a figure that also seems to be suggested in Storey, “Population”. 26  Or, according to De Ligt, “The Population of Cisalpine Gaul in the Time of Augustus”, between 440 and 560 p/ha if the suburbs outside the Aurelian walls are included. 27  The presence of such buildings can also be assumed from some of our literary sources, see Yavetz, “The Living Conditions of the Urban Plebs”, referring to Livy 21.62.3, Strabo 5.3.7, and Suetonius Aug. 89. For the archaeological evidence, see Wallace-Hadrill, “Case e abitanti a Roma” and also De Ligt, Peasants, Citizens and Soldiers. 28  Although it seems clear from the available evidence that only the largest cities in the Roman world were provided with apartment buildings (Hanson, Urban Geography of the Roman World). 29  De Ligt, Peasants, Citizens and Soldiers; Wilson, “City Sizes and Urbanisation in the Roman Empire”.

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

211

of living quarters, shops, taverns, workshops and leisure facilities.30 Still, the main area for public architecture was certainly the central forum and its immediate surroundings, occupied by large buildings such as tabernae with porticos, temples, the basilica and the macellum. The public bath complex was also usually situated not far from the central square. Other voluminous constructions were the theatre and the amphitheatre. The position of these spectacle buildings within the urban area could vary according to the morphology of the terrain and the availability of sufficient space. This is why it was not uncommon for amphitheatres to be situated outside the city walls. Although none of these structures would have had standardised dimensions—and thus the proportional share of public buildings would have differed from one town to the other—a recent study by Marta Conventi on the organisation of urban space in some forty newly founded Roman towns in Italy has yielded some interesting results regarding the percentage of space that each of these structures could take up within the town. For example, the forum square could take up a minimum of 0.8 per cent and a maximum of 5.84 per cent, whilst on average, it occupied 2.47 per cent of the total urban area. Moreover, most of the smaller towns—that is, the ones below forty hectares—seemed to have had a proportionally large forum, usually covering about 2 per cent of the urban space, but occasionally 4 per cent or higher. Based on this study it seems that, on average, the basilica occupied about 0.53 per cent, although it never took up less than 0.18 per cent or more than 0.93 per cent. The theatre and the amphitheatre could take between 0.14 and 1.95 per cent and between 1.18 and 2.57 per cent respectively. On average, the former occupied about 0.74 per cent, whilst the medium figure for the latter was 1.85 per cent.31 This means that, on the whole, these structures combined could take up as little as 2.3 per cent and as much as 11.29 per cent of the total town area, with an average of 5.59 per cent. Regarding their variable impact on the overall population density of a town, the cases of Cosa in Italy and the recently studied town of Sabratha in Tripolitania may be considered exemplary. In the case of Cosa the aforementioned density of 105 to 125 persons per hectare refers to the entire 30  Raper, “The Analysis of the Urban Structure of Pompeii”; Wallace-Hadrill, Houses and Society; Laurence, “The Organization of Space in Pompeii”; Paoli, Rome; Storey, “The Population of Ancient Rome”; Hanson, Urban Geography of the Roman World. 31  Conventi, Città romane di Fondazione.

212 

D. VAN LIMBERGEN AND F. VERMEULEN

urban area, including the areas reserved for public and commercial activities, whilst the adjusted figure for the available private space would have been 130 to 155 p/ha, which is approximately 25 per cent higher. Then again, for Sabratha, Andrew Wilson has calculated the density of the residential part of the town—which had a total town size of about 34.6 hectares and a total estimated population of 5730 to 14,326 people—as between 182 and 457 persons per hectare. By extracting the space taken up by the public architecture of the town—that is approximately 3.26 hectares or 9.4 per cent of the urban area—a slightly lower overall density of between 165 and 415 persons per hectare (about 9 per cent lower) was obtained.32 These examples illustrate well that it is important to be as precise as possible in assessing the proportion of an urban area occupied by non-residential buildings. Two final variables that had a significant impact on the number of buildings that could be fitted within a certain urban area were the extent of the space left open for the insertion of minor squares or gardens and town’s street grid.33 For example, the lower estimate of approximately 8000 inhabitants (121 p/ha) for the population of Pompeii proposed by H. Eschebach was inspired by the excavations of W.F. Jashemski, which indicated the notable presence of intramural areas under cultivation.34 The potential impact of street grids depended both on the number of streets in a city and the width of individual streets.35 In addition, towns organised according to a subdivision in larger insulae had a lower number of streets than towns subdivided into smaller insulae. Even if in most cases it is unlikely that we will acquire detailed and complete information on these two parameters through archaeology, any available data might still be valuable to correctly inform our demographic reconstructions. 32  De Ligt, “The Population of Cisalpine Gaul in the Time of Augustus”; De Ligt, Peasants, Citizens and Soldiers (Cosa); Wilson, “City Sizes and Urbanisation in the Roman Empire”; Hanson, Urban Geography of the Roman World (Sabratha). 33  Chevallier, La Romanisation de la celtique du Pô; Conventi, Città romane. 34  Eschebach, Die stadtebauliche entwicklung des Antiken Pompeij; Wallace-Hadrill, Houses and Society. A similarly low figure of 7000 to 7500 people (106 to 113 p/ha) was also proposed by Russell, “Population and Mortality,” 107. 35  Again, the varying dimensions of these streets may be illustrated by the aforementioned study by Marta Conventi, who registered road widths of between 3.1 m and 12 m for the decumanus maximus, and between 3.5  m and 10  m for the cardo maximus. Secondary streets, on the other hand, had smaller widths of between 2.4 m and 6.5 m. The respective average widths were 6.37 m, 7.31 m and 4.44 m (Conventi, Città romane).

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

213

All these factors could change over time, thus affecting also the existing population density of a town. For example, in the case of walled towns it is unlikely that the town area was completely built-up from the beginning. More likely, these towns gradually became crowded; in which case suburbs began to develop only after the space between the walls was completely built-up. Such a scenario may imply lower densities in the initial years of the town’s existence and higher densities in later years.36 Population densities could also change because new and mostly smaller houses were gradually fitted in between older and larger ones, or because older and smaller houses were incorporated into new and larger ones. Another possible evolution was that it became more common to build houses with one or more stories, instead of houses that were only provided with a ground floor. The number of people that could live in a certain town area could also be affected by proportional changes in the organisation of the urban space through time. Thus, residential areas could be turned into administrative, religious or commercial areas—or vice versa—whilst processes of monumentalisation and/or restructuration could increase or decrease the amount of space used for public architecture within a town over time. The same goes for the layout of insulae and streets, which could be modified to fit new circumstances. We will not delve deeply into these issues in the course of this paper, but data on these possible processes become particularly important when tracing urban demographic evolutions over time. The elements on which we will particularly focus in the next section are the ratios of urban infrastructure (insulae, street grid) and public architecture at Potentia and Trea.

3   Urban Survey Data from Central Adriatic Italy 3.1  Potentia The coastal site of Potentia, founded as a colony of Roman citizens, lies in the surroundings of today’s Porto Recanati in the lower Potenza valley (Fig. 7.1). Ten years of intensive aerial photography analysis, large-scale geophysical survey and systematic artefact fieldwalking by the PVS— together with excavations in the area of the western town gate—have resulted in significant advancements in our understanding of the 36  De Ligt, “The Population of Cisalpine Gaul in the Time of Augustus”; De Ligt, Peasants, Citizens and Soldiers.

214 

D. VAN LIMBERGEN AND F. VERMEULEN

topography of this urban settlement.37 In fact, three main phases have been identified in the town’s occupation history. The first phase probably relates to the early years of the colony between 184 and 174  BCE and represents the initial organisation of the town after the arrival of the colonists. In that period, the town area—most likely enclosed by an earthen bank (agger) with a palisade and a ditch—covered about 525 × 300 meters (15.7 hectares).38 According to a passage in Livy (39.44.10), the colony was provided with a three-gated circuit wall in 174  BCE.  This wall has been detected in both aerial imagery and through geophysical prospection and it enclosed a town area of some 16 hectares, regularly subdivided into 48 insulae (Fig. 7.2). Of the three existing town gates only the western gate and a section of the wall itself have been excavated by a team from Ghent University. The latter was built according to the opus quadratum technique with rectangular sandstone blocks of different sizes held together with the use of clay.39 Livy also informs us about the installation in the same period of an aqueduct, a temple dedicated to Jupiter (T) and the first phase of the forum (F) (Fig. 7.2).40 A third phase—probably after the middle of the first century BCE—probably entailed an enlargement of the walled space with some fifty metres to the east. As such, the town was enlarged with twelve insulae and now measured 525  ×  350  metres or about 18.4 hectares intra muros (Fig. 7.3). The reasons for this reorganisation of the town area are not entirely clear but were possibly associated with the necessity to partially rebuild the town in the aftermath of a major earthquake that occurred in the region around the mid-first century BCE, as mentioned by Cicero.41 37  Vermeulen, “Città romane nella valle del Potenza”; Vermeulen, “Reviewing 10 Years of Aerial Photography in the Valley of the River Potenza”; Vermeulen and Boullart, “The Potenza Valley Survey”; Vermeulen and Verhoeven, “The Contribution of Aerial Photography and Field Survey to the Study of Urbanisation in the Potenza Valley”; Vermeulen and Verhoeven, “An Integrated Survey of Roman Urbanisation at Potentia”; Vermeulen, Monsieur and Boullart, “The Potenza Valley Survey”; Vermeulen, Hay and Verhoeven, “Potentia”; Vermeulen et al., “Investigating the impact of Roman urbanisation on the landscape of the Potenza valley”. 38  Vermeulen et  al., “Scavi presso la porta occidentale di Potentia”; Vermeulen and Monsieur, “Le système défensif”. 39  Vermeulen et al., “Scavi presso la porta occidentale di Potentia”. 40  Percossi Serenelli, Potentia; Percossi, “Le fasi repubblicane di Potentia”; Vermeulen and Monsieur, “Système défensif”. 41  Vermeulen and Carboni, “Measuring Urbanisation”; Vermeulen et al., “Scavi presso la porta occidentale di Potentia”; Vermeulen, “Potentia”. Excavations in the area of the temple

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

215

Fig. 7.2  First phase of the town of Potentia, with indication of the initial forum square (F) and the temple for Jupiter (T). (Map by PVS team)

216 

D. VAN LIMBERGEN AND F. VERMEULEN

Fig. 7.3  Map of Potentia, based on the integration of survey and excavation data. (Map by PVS team)

Based on the available epigraphic evidence the town clearly flourished during the Augustan era and the High Imperial period (later first to second centuries CE).42 Evidence of this evolution has been found in the survey results of the PVS within the town area, which suggest that the have indeed identified an important destruction layer, possibly dated to the mid-first century BCE (Paci and Percossi-Serenelli, “Il paradigma della romanizzazione”); traces of an important reconstruction under Augustus have also been found during the excavations of the western town gate (Vermeulen and Monsieur, “Système défensif”). 42  Paci, “Le iscrizioni romane di Potentia”.

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

217

intensity of intramural habitation increased significantly between the Late Republican period and the first to second centuries CE.43 It is possible that in the course of the first century CE residential quarters gradually developed in the immediate suburbium of Potentia, but archaeological evidence for extramural buildings is currently thin. Still, concentrations of Early Imperial domestic pottery and brick building materials found just outside the western and southern town gates suggest that some form of extra-­ urban residential occupation existed in the first two centuries of the Empire, whilst recent geomagnetic surveys in the area immediately outside the western town gate have revealed the traces of a building structure that might be identified as a suburban farm, or another building that was probably unrelated to the predominantly funerary structures discovered in this sector (Fig. 7.3, A).44 The location of the colonial town directly on the coast and within a wetland area just north of the ancient river mouth limits the availability of extramural land suitable for large-scale housing expansion. The absence in the archaeological record of extramural materials from later Roman times suggests that from the third century CE onwards the habitation definitely contracted again to within the walled urban area. The same surveys also attest to an apparent contraction of the habitation within Potentia from the mid-fifth century CE onwards when mainly the forum area and the northern section of the town remained in use, whilst the entire eastern section seems to have been abandoned.45 In fact, these surveys support some of the most significant excavation results that have been obtained over the years, such as the demolition of the temple next to the forum at the end of the fourth century CE, and the insertion of fifth to sixth centuries tombs in the north-eastern corner of the town, in an area that had been occupied up until the early fifth century CE.46 Cemeteries bordered all of the three main exit roads of the town (Fig. 7.3, shaded zones). With over 400 tombs excavated in the 1960s by the Soprintendenza Archeologica delle Marche, the northern one is the largest and best-known cemetery and seems to have been in use from the 43  Vermeulen, “Functional Zoning and Changes in the Use of Space in the Roman Town of Potentia. 44   Vermeulen, “Potentia”; Vermeulen et  al., Archaeological Investigations in the Potenza Valley. 45  Vermeulen, “Functional Zoning”; Vermeulen, “Potentia”. 46  Mercando, “Marche-Rinvenimenti di insediamenti rurali”.

218 

D. VAN LIMBERGEN AND F. VERMEULEN

second century BCE until the fourth century CE.47 Aerial surveys and subsequent excavations have shown the presence here of a series of funerary monuments located along the outgoing road.48 The western cemetery was touched upon during the excavations of the western town gate by the PVS team and can be broadly placed in the first to second centuries CE.49 Furthermore, recent geomagnetic surveys have confirmed the presence of several funerary monuments further along the western exit road of the town50 (Fig. 7.3, B). The presence of the southern necropolis has equally been attested through remote sensing and geomagnetic surveys.51 Finally, a large artisanal-industrial zone dating to the Augustan period occupied the southern part of the northern cemetery and a section of the intramural area of the town (Fig. 7.3, C).52 The attestation of a press floor in association with a kiln producing Dressel 6A wine amphorae and Dressel 6B oil amphorae points towards an important production-related function of the area between the mid-first century BCE and the mid-first century CE.53 The internal organisation of the individual insulae is difficult to reconstruct from the geophysical and aerial survey imagery, but some broad elements of the town’s organisation are sufficiently clear. In its most expansive form the town was subdivided into sixty insulae. Of these, some twenty-four insulae each measured 1 × 2 actus (35 × 70 m), one row of twelve insulae each 1 × 2.5 actus (35 × 87.5 m), another row of twelve each 1 × 1.5 actus (35 × 52.5 m) and a final row of twelve measured 1 × 1 actus (35 × 35 m). On the whole, these insulae took up about 13.2 hectares (132,300 m2) or 71.9 per cent of the total urban area (18.4 hectares), leaving about 5.2 hectares for the actual street grid (28.1 per cent), with streets being on average between five and six metres wide. Within the town area, the forum square itself measured ca.120  ×  30 metres (ca. 3600 m2) (Figs. 7.3, D and 7.4, D). On both of its long sides it was bordered by a series of shops with porticos taking up approximately 1920 m2 (16 × 120 m) along the western side and about 2880 m2 (24 × 120 m) 47  Mercando, Sorda and Capitanio, “La necropoli romana di Portorecanati”; Percossi Serenelli, Potentia. Quando poi scese il silenzio; Edvige Percossi, “La necropoli di Potentia”; Vermeulen, “Potentia”. 48  Percossi, “Necropoli”. 49  Vermeulen et al., “Scavi presso la porta occidentale di Potentia”. 50  Vermeulen et al., Archaeological Investigations. 51  Vermeulen et al., Archaeological Investigations. 52  Mercando, “Marche”. 53  Vermeulen and Monsieur, “Système défensif”.

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

219

Fig. 7.4  Detailed plan of the monumental central area of Potentia, based mainly on the survey results by the PVS. (Map by PVS team)

220 

D. VAN LIMBERGEN AND F. VERMEULEN

along the eastern side (4800 m2 in total) (Figs. 7.3, E and 7.4, E). The northern side was probably dominated by the temple dedicated to Jupiter and a second building, to be identified as another temple or curia (ca. 1 × 1 actus or 35 × 35 m, ca. 1225 m2) (Figs. 7.3, F and 7.4, F).54 A possible basilica flanks the entire southern side (ca. 15 × 30 m, ca. 450 m2) (Figs. 7.3, G and 7.4, G). An area of about 135 × 65 meters (ca. 8775 m2) immediately to the east of the forum also housed a Late Republican/ Augustan temple—measuring approximately 192 m2 (24 × 8 m) and positioned inside a porticoed enclosure of about 1152  m2 (24  ×  48  m) (Figs. 7.3, I and 7.4, I)—and a small macellum taking up about 512 m2 (16 × 32 m) (Figs. 7.3, H and 7.4, H).55 Another temple (ca. 384 m2) was also identified immediately to the east of the Augustan temple (Figs. 7.3, J and 7.4, J).56 On the whole then, the registered central public architecture occupied approximately 1.2 hectare (12,123 m2), about 6.6 per cent of the walled urban area. Based on the provisional plan, however, it is not unreasonable to assume a total central area devoted to public architecture of about 1.9 hectare (120 × 160 m), with possibly an additional main thermal complex and some more temples (Fig. 7.3, grey zone). In the eastern part of the town a semi-circular public building—probably an odeon or a small theatre—occupied approximately 1  ×  1.5 actus (35  ×  52.5  m, 1838  m2) (Fig. 7.3, K). It is very unlikely, however, that an amphitheatre was positioned inside the walls. A possible thermal complex was identified in the southern part of the town but its extensions and detailed plan are unknown (Fig. 7.3, L).57 Only a few bath houses have been excavated in the central Adriatic area.58 A recurrent scheme, however, seems to be the presence of a large bath complex close to the forum of the town—with noted dimensions between 2000 and 5200 m2, or an average surface of 3666 m2—and  Vermeulen and Carboni, “Approaching Roman Fora with Non-Invasive Urban Survey”.  Percossi Serenelli, “Potentia. Fonti letterarie e fonti archeologiche”; Vermeulen and Monsieur, “Système défensif”. 56  Vermeulen and Verhoeven, “Integrated Survey”. 57  Vermeulen and Carboni, “Measuring Urbanisation”. 58  These include the complexes excavated at the towns of Tifernum Mataurense (Catani, “Tifernum Mataurense), Forum Sempronii (Gori and Luni, “Edificio termale a Forum Sempronii”), Ostra (Dall’Aglio, Silani and Tassinari, “Nascita e sviluppo monumentale della città romana di Ostra”), Sentinum (Cavallo, “Sentinum, le Terme Urbane), Septempeda (Sisani, Umbria/Marche), Matilica (Biocco, Città romane), Falerio (Maraldi, Falerio) and Interamnia Praetuttiorum (Staffa, “Teramo”). 54 55

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

221

the location of a second, smaller thermal building in another zone of the settlement. The dimensions of the latter building commonly varied between 300 and 1208 m2, resulting in an average surface of 633 m2.59 As such, with the addition of the average figure for a hypothetical second small thermal building, the public space may have covered a total area of about 2.2 hectare (21,705 m2), or 11.8 per cent of the town. That leaves about 11.1 hectare (110,595  m2) or 60.1 per cent for residential space (see Graph 7.1). Although the survey results do not allow for an assessment of the size and number of houses that occupied a single insula,

F

Ta

Te

M

Th

Forum

Tabernae

Temple

Macellum

Theatre

B

UPB

UPS

SGr

RS

Bath

Unidentified Public Building

Undetermined Public Space

Street Grid

Residential Space

Graph 7.1  Diagram of shares public and private architecture at Potentia

 Van Limbergen, Pots, Presses, People and Land.

59

222 

D. VAN LIMBERGEN AND F. VERMEULEN

rescue excavations in the central part of the town in the 1980s have indicated that the private architecture was of the domus-type, with rooms centred around a courtyard.60 3.2  Trea The remains of the Roman municipium of Trea have been located on an elevated plateau close to the modern town of Treia in the middle Potenza valley (Fig. 7.1). The precise origins of the settlement are not clear but recent surveys in and around the urban site have revealed the presence of Picenian material, suggesting that the Roman town was at least partially built on a pre-existing protohistoric centre.61 As the Itinerarium Antonini identifies Trea as a town lying on the Via Flaminia per Picenum Ancona— that is a diverticulum of the main Via Flaminia that led to Ariminum (Rimini)—the settlement probably possessed the status of praefectura in the third to second centuries BCE. The Liber Coloniarum mentions land allotments in the territory of Trea under the second triumvirate.62 Almost nothing is visible of the site today other than two sections of the circuit wall in quasi reticulatum technique: a wall that was probably erected in the wake of the settlement’s acquisition of municipal status around 49 BCE.63 Excavations in the late eighteenth century, conducted by Fortunato Benigni, uncovered similar wall sections, the remains of a basilica and a sanctuary under today’s monastery of SS.  Crocifisso.64 Later research activities include the surveys carried out by Umberto Moscatelli on and near the town site in the 1970s, and the excavations by Fabrini under the

60  Percossi Serenelli, “Potentia. Fonti letterarie e fonti archeologiche”; Percossi Serenelli, Potentia. Quando poi scese il silenzio. 61  Vermeulen, Slapzak and Mlekuz, “Surveying the townscape of Roman Trea”. Such an origin had already been suggested given the vicinity of the Monte Pitino—an important Picenian centre in pre-Roman times—and the site’s obvious strategic position (Lollini, “La civiltà picena”; Vermeulen et al., “Potenza Valley Survey”; Vermeulen et al., “Investigating the impact of Roman urbanisation on the landscape of the Potenza valley”. 62  Moscatelli, “Municipi romani della V ‘regio’ augustea”. 63  Paci, “Indagini recenti e nuove conoscenze sulle città romane del territorio marchigiano”; Vermeulen et al., “Surveying the townscape of Roman Trea”. Based on the building technique, the construction of the town walls is usually situated either in the first half (Moscatelli, “Municipi romani”) or the second half of the first century BCE (Percossi Serenelli, “Potentia. Fonti letterarie e fonti archeologiche”). 64  Vermeulen et al., “Surveying the townscape of Roman Trea”.

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

223

aforementioned monastery of SS.  Crocifisso between 1985 and 1988.65 On the basis of this knowledge, the total town area of Trea was estimated at approximately thirteen hectares.66 Since 2003, fieldwork by the PVS, using a combination of aerial photography, geophysical prospection, corings and artefact survey, has notably increased our knowledge of the topography and evolution of this Roman settlement. One important contribution from these research activities is evidence of a smaller intramural area than previously thought: eleven hectares. It is also clear that these walls were provided with several rectangular towers, whilst a town gate was found along the southern course of the circuit wall (Fig.  7.5). More importantly, however, thanks to this work

Fig. 7.5  Map of the town site of Trea, based on the survey results by the PVS. (Map by PVS team)  Moscatelli, “Municipi romani”; Fabrini, “Dal culto pagano al culto cristiano”.  Moscatelli, Trea.

65 66

224 

D. VAN LIMBERGEN AND F. VERMEULEN

Fig. 7.6  Detailed plan of the monumental central area of Trea, based on the survey results by the PVS. (Map by PVS team)

some 70 per cent of the town’s infrastructure is known today.67 The central section of the settlement was occupied by the rectangular forum square (ca. 100 × 35 m, ca. 3500 m2) (Figs. 7.5, A and 7.6, n° 1)—including a small temple of approximately 20 × 10 metres (ca. 200 m2), probably the Capitolium of the town or dedicated to the Imperial cult (Fig.  7.6, n° 5)—and its adjacent buildings: a basilica to the west (ca. 35  ×  20  m, 700 m2) (Figs. 7.5, B and 7.6, n° 2), flanked by a curia to the north (ca. 530 m2) (Figs. 7.5, C and 7.6, n° 3), a possible temple to the south (ca. 710 m2) (Fig. 7.5, D) and a series of tabernae flanking parts of the long sides of the square, occupying a combined area of about 4500  m2 67  Vermeulen et al., “Surveying the townscape of Roman Trea”; Vermeulen and Carboni, “Approaching Roman fora”.

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

225

(Figs.  7.5, E and 7.6, n° 4), together with a semi-public domus in the south-east corner measuring at least 30 × 30 metres (ca. 900 m2) (Figs. 7.5, F and 7.6, n° 6). Immediately to the north-east of the forum, along the street leading to the Ancona gate, a recently identified probable macellum occupied an additional space of about 25 × 25 metres (625 m2) (Fig. 7.5, G). Together, the documented buildings grouped in the central section of the town occupied an area of approximately 1.2 hectare (11,465 to 12,000 m2), or 10.6 to 12 per cent of the total urban area. Based on the known plan, however, it is easily possible that the total central area reserved for public space was in the range of 2.3 hectares (23,150 m2) or about 21 per cent of the town (Fig. 7.5, grey zone). This space may have included the main public bath house, and the temples for Minerva and Victoria, whose existence is attested epigraphically.68 Based on the combination of archaeological and epigraphic sources, it seems that the Roman towns of central Adriatic Italy typically housed one large main temple and between two and five smaller temples within their urban area. The larger temples varied in size between 384 and 613  m2—with an average of about 540 m2—and the smaller temples mostly ranged from 117 to 300 m2 (ca. 202 m2 on average), with only two temples of less than 100 m2 (74 m2 on average).69 Earlier excavations in the eighteenth century attested to the presence of at least one other temple—a sanctuary for the gods Isis and Serapis—and a thermal building in the southern part of the town.70 By incorporating the derived average figures for public space, including a small thermal building (633 m2) and a small- to medium-sized temple (74 to 202  m2), the total area occupied by public space may thus have amounted to almost 2.4 hectares (23,985 m2), or about 22 per cent of the town. Recent analysis by the PVS team of older aerial imagery picked up an oval structure near the southern limit of the town, which might possibly be identified as an amphitheatre (Fig. 7.5, H). The inclusion of an amphitheatre into the main urban area was not uncommon in the central Adriatic area. Based on the available information for ten towns, the minimum area occupied by the building would have been 4144  m2 (Interamnia Praetuttiorum) and the maximum 9350 m2 (Fanum Fortunae), with an average of 6390 m2. Percentage wise, the proportions ranged between 0.6  Marengo, Regio V Picenum - Trea, 155–188).  Van Limbergen, Pots, Presses, People and Land. 70  Fabrini, “Culto pagano”; Capriotti Vittozi, Oggetti, idee, culti egizi nelle Marche. 68 69

226 

D. VAN LIMBERGEN AND F. VERMEULEN

per cent (Hatria) and 5.9 per cent (Ancona); on average it took up 2.8 per cent.71 By applying the average surface area of 6390  m2 to the town of Trea, the public space may thus have totalled some three hectares (30,375 m2) or about 27 per cent of the urban area (Graph 7.2). With regard to the internal organisation of the town, there seems to be a series of insulae of 3 × 1 actus (105 × 35 m), bordered by streets that have an average width of four metres (Fig. 7.5).72 The current plan is not detailed enough to determine the exact space taken up by the street grid,

F

Ta

Te

M

Th

Forum

Tabernae

Temple

Macellum

Theatre

B

UPB

UPS

SGr

RS

Bath

Unidentified Public Building

Undetermined Public Space

Street Grid

Residential Space

Graph 7.2  Diagram of shares public and private architecture at Trea  Van Limbergen, Pots, Presses, People and Land.  Vermeulen, “Città romane nella valle del Potenza; Vermeulen et  al., “Surveying the townscape of Roman Trea”. 71 72

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

227

but some suggestions can be made. For example, it is unlikely that the street grid took up as much space in Trea as it did in Potentia (about 30 per cent) since the latter colony had both wider streets and smaller insulae than the former. This implies that the overall occupation ratio of the street grid must have been higher in Potentia than in Trea. In fact, one other well-studied town in the area, Pisaurum, shows a quite different scenario in comparison with Potentia. Streets in Pisaurum were roughly three to four metres wide, whilst the town area itself was subdivided into a series of thirty-five fairly large insulae measuring mostly 2 × 2 actus, but sometimes also 2 × 2.5 actus, 2.5 × 1 actus and 2 × 1 actus. As such, the area reserved for the street grid at Pisaurum was calculated at roughly 10 per cent (1.8 hectare of a total 18 hectares).73 As the situation at Trea appears similar to Pisaurum rather than to Potentia, a share of 15 per cent (1.65 hectares) for the street grid would seem to be more realistic for Trea, thus reducing the total space available for domestic buildings to 6.35 hectares (57.7 per cent). A final interesting feature of Trea is the existence of two orientations in the street grid. While the western part of the town, occupying the higher end of the plateau, has a South-West to North-East orientation, the eastern part, occupying the lower section, has a North-West to South-East orientation. One of the most visible elements of this grid is the six metres wide decumanus maximus, which enters the settlement through the western gate and crosses the entire western section of the site but is abruptly halted by the monumental forum complex that occupied most of the eastern side (Fig.  7.5). Remote sensing activities by the PVS have clearly attested that this road originally continued its course towards a gate in the northern section of the circuit wall but was later obliterated by the construction of the forum and its adjacent buildings, after which the road was diverted along the northern side of the public centre (Fig. 7.5).74 These observations clearly suggest a reorganisation of the urban area in the Early Imperial period as a direct result of the monumentalisation of the town centre, most likely including an enlargement of the original Late Republican core.75  Van Limbergen, Pots, Presses, People and Land.  Vermeulen et al., “Surveying the townscape of Roman Trea”. 75  Most of the Republican material collected during the surveys by the PVS was found in the higher western part of the town (Vermeulen et  al., “Surveying the townscape of Roman Trea”). 73 74

228 

D. VAN LIMBERGEN AND F. VERMEULEN

It is obvious from most of the archaeological evidence that Trea experienced its heyday in the course of the first and second centuries CE, but it is not yet clear if it ever developed suburban quarters.76 Still, considering the relatively small space for housing inside the town walls, it is certainly possible that in the course of the first century CE some extramural quarters developed, in particular to the east and the west of the town where the topography of the location best allowed for it. Artefact surveys on the steep southern slope of the site have attested to surface scatters that could belong to extramural buildings but might also reflect Roman garbage dumps outside the walled area (Fig. 7.5, grey zone, I).77 Again, house plans are not very detailed, but—even more than in Potentia—the registered traces and artefacts point towards the presence of large and richly decorated buildings of the domus type from the first century BCE onwards,78 especially along the main streets of the town, with some of the houses measuring roughly between 1300 and 1600 m2 (Fig. 7.5, squares). 3.3  Private Architecture The residential layout of a Roman town typically consisted of a mixture of houses of various sizes. The well-known and thoroughly studied towns of Pompeii and Herculaneum are good illustrations of this practice. Obviously, we cannot hope to have such detailed information on the houses occupying the towns of central Adriatic Italy. In fact, for most of the towns in the area we only have evidence of isolated wall sections and/or various types of floor pavement segments, associated at best with rooms or partial house plans.79 Clearly it is not possible to deduce either the full plan or the dimensions of these residential units from such fragmentary finds. Here we need to build our picture from the information provided through geophysical and aerial photography imagery, together with the few houses that have been fully excavated over the years. One of the best documented cases is the residential area recently registered by means of aerial photography at Tifernum 76  The site shows traces of occupation between the second century BCE and the early sixth century CE (Vermeulen, “Città romane nella valle del Potenza,” 630). 77  Vermeulen et al., “Surveying the townscape of Roman Trea”. 78  Vermeulen et al., “Surveying the townscape of Roman Trea”. 79  Van Limbergen, Pots, Presses, People and Land.

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

229

Mataurense. Here, the prospections revealed the traces of nine probable houses positioned around the crossing of two paved streets. Two large houses occupied a surface area between 800 and 1000  m2—that is, house A (ca. 950 m2) and house G (ca. 925 m2)—but the other houses were smaller in size, covering surfaces of between either 400 and 600 m2 (560 m2 for house B, 480 m2 for house C, and 470 m2 for house F), or even below 400  m2 (390  m2 for house D and 360  m2 for house E) (Fig. 7.7).80 The same mix is visible on the geophysical imagery from Sentinum where we can clearly see a grouping of houses with differing sizes and shapes in the south-western sector of the town. Admittedly, individual house plans are difficult to discern on the published plan, but in one case the registered traces do indicate the presence of a domus measuring ca. 460  m2, positioned next to a smaller house of about 300 m2 (Fig. 7.8, A, squares).81 Another piece of information comes from Ostra, where a portion of a residential neighbourhood was registered by means of aerial photography. This area—positioned along one of the main streets of the town—consisted of what appears to be four houses flanked by a series of tabernae. Although the published plan only allows for rough estimates of the size of these houses—one house may have measured about 540 m2, two others about 400 m2 and the fourth about 800 m2—the general layout of the quarter seems to confirm both the spatial pattern and the size range individuated for the private architecture of the other towns (Fig. 7.9, B).82 Finally, some useful indications for house sizes may also be derived from a few partially excavated houses in the towns of Pisaurum, Forum Sempronii and Matilica. Two Early Imperial houses were nearly fully uncovered at Pisaurum: the first measured about 625 m2 with at least 19 rooms distributed around a courtyard and the second about 875 m2. The very large “Domus di Europa” at Forum Sempronii extended over an area of approximately 1200 m2. At Matilica two domus complexes have been uncovered, which measured at least 600 m2 and 1200 m2.83 Although it is sufficiently clear that all these structures belong to domus-­ type houses, organised around an atrium and/or peristylium, no detailed  Catani and Monacchi, Tifernum Mataurense.  Medri, “Materiali per una nuova Forma Urbis di Sentinum”. 82  Boschi and Silani, “Aerofotografia e geofisica nella Valle del Misa. 83  Van Limbergen, Pots, Presses, People and Land. 80 81

230 

D. VAN LIMBERGEN AND F. VERMEULEN

Fig. 7.7  Reconstruction of the eastern residential quarter of Tifernum Mataurense, based on aerial photography. (Catani and Monacchi, Tifernum Mataurense, 245, fig. 63)

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

231

Fig. 7.8  Map of Sentinum based on the integration of excavation and geophysical survey data. (After Medri, “Materiali”, 213, fig. 3.1.12)

232 

D. VAN LIMBERGEN AND F. VERMEULEN

Fig. 7.9  Map of the town site of Ostra, with the indication of a residential quarter individuated through aerial photography (B). (After Boschi and Silani, “Aerofotografia e geofisica”, 79, fig. 7)

internal plans can be retrieved from these traces, and thus no information on the precise number of rooms in these houses is available. Therefore, we can merely estimate the number of persons that lived in them. As we have seen—taking as a starting point the estimated population of 10,000 people for the city of Pompeii—Wallace-Hadrill calculated an average occupation of seven to eight persons per average-sized house of 271 m2. This left him with a theoretical occupation rate of 34 to 39 m2 per person.84 If we apply these figures to the aforementioned houses this would imply the following occupation densities: seven to eleven persons for the houses less than 400  m2; ten to eighteen persons for the houses between 400 and 600 m2; twenty to twenty-seven persons for the houses between 800 and 1000  m2; and thirty to forty-seven persons for the houses larger than 1000 m2.  Wallace-Hadrill, Houses and Society.

84

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

233

Thanks to the published reports of a few better excavated houses, we are able to further refine our figures for both the sizes of the individual houses and the number of people that lived in them. A first example is the Late Republican domus recently unearthed at Sena Gallica. This typical atrium-house measured 17 × 27 metres (459 m2) and was subdivided into a total of fourteen spaces: an entrance room (vestibulum) (A), two rooms with a commercial or storage function (tabernae) (B), five bedrooms (cubiculum) (C) distributed around a central courtyard (atrium) (D), two adjacent wing rooms (alae) (E), a tablinum (F) and two dining rooms (triclinium) (G) (Fig. 7.10). The open atrium occupied a space of about 120 m2 (13.5 × 9 m), implying a roofed area of about 339 m2.85 If we again apply Wallace-Hadrill’s ratio of 34 to 39 m2 per person, the number of inhabitants would fall in the range of eleven to thirteen persons. Alternatively, however, we might apply Wallace-Hadrill’s other methodology—that is, assuming an average occupation of one person per ground floor room, excluding the spaces not used for living, such as entrances and

Fig. 7.10  Plan of the domus in the area “La Fenice” at Senigallia. (After Salvini, Area archeologica, 22)

 Salvini, Area archeologica e Museo La Fenice.

85

234 

D. VAN LIMBERGEN AND F. VERMEULEN

passageways, or commercial and/or storage rooms—and consider the number of living rooms to equal the number of inhabitants.86 This would result in a slightly lower occupation of ten persons for the domus in question. As such, an average occupation density of ten to eleven persons might not be unreasonable. The same line of reasoning may be applied to the Late Republican house discovered under the Imperial temple at Urbs Salvia (Fig. 7.11).87 Measuring about 392 m2 (14 × 28 m)—of which 288 m2 is roofed space— and organised into a vestibulum (A) flanked by two tabernae (B), an atrium (C) with four adjacent cubicula (D) and two alae (E) and a tablinum (F) flanked by two triclinia (G), the domus could either have housed

Fig. 7.11  Reconstruction of the Republican domus under the Imperial temple complex of Urbs Salvia. (After Montali, “Considerazioni”, 132, fig. 16)  Wallace-Hadrill, Houses and Society.  Montali, “Considerazioni sulle strutture edilizie dell’area del Tempio-criptoportico”.

86 87

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

235

ten to eleven individuals (Wallace-Hadrill’s first method) or seven persons (Wallace-Hadrill’s second method). Again, an average population of between seven and ten persons can realistically be assumed. Another Late Republican house of similar size is the so-called Casa ad atrio found under the Imperial “Domus dei Coiedii” at Suasa (Fig. 7.12). This domus measured approximately 480 m2 (40 × 12 m) and consisted of an entrance (vestibulum) (A) flanked by a taberna (B) and a bedroom (C), an atrium (D) with three adjacent bedrooms (cubiculum) (E, F, G), a tablinum (H) and a triclinium (I), a garden (hortus) (J) and a backroom (K, L, M).88 If we subtract all the spaces of the house not used for living, Fig. 7.12  The Imperial Domus dei Coiedii at Suasa, with indications of the earlier Casa ad atrio and the Casa del primo stile. (After Campagnoli, “Fasi edilizie”, 320, fig. 1)

 De Maria and Giorgi, “Urbanistica e assetti monumentali di Suasa”.

88

236 

D. VAN LIMBERGEN AND F. VERMEULEN

the number of rooms in the house totals seven. The actual roofed space of the house—that is, without the areas occupied by the atrium (78 m2) and the hortus (125 m2)—covered 277 m2. Depending on which calculation method we apply, the domus may thus have housed either seven persons, or between twelve and fourteen persons. Another well-known Late Republican house, the smaller “Casa del primo stile”, was located somewhat to the south of the “Casa ad atrio” (Fig. 7.12). The first phase of the building consisted of a domus measuring only 189 m2 (27 × 7 m), with an entrance (A), an atrium (B), a tablinum (C), two bedrooms (D, E), a corridor and a likely wide utility space at the back (F). The roofed space was about 169 m2.89 With only three real living rooms, the number of inhabitants must have been relatively low, perhaps ranging between four to five or even only three persons. An argument in favour of the higher range may stem from the supposed presence of a second floor, or at least an upper flat.90 In a second phase, some 266 m2 of space was added to the domus, enlarging it to 455 m2. Theoretically, the building may now have housed between eleven and thirteen persons, but a large part of the newly constructed sector seems to have had an artisanal/utility function, suggesting that the number of inhabitants did not rise dramatically. Taking into consideration the four smaller rooms of the new sector, an estimated figure of seven to nine persons may be a more realistic occupation density. Recently, one of the domus buildings discovered at Tifernum Mataurense, house G, was completely excavated (Fig.  7.13). The Early Imperial house comprised a total of twenty-three rooms occupying a surface of about 925  m2. The house included a vestibulum flanked by two tabernae and another utility space, a tablinum flanked by three other rooms, an atrium/peristylium surrounded by six (bed)rooms to the north and the south and a triclinium flanked by two rooms at the back. The five rooms in the north-eastern corner of the house were occupied by a small thermal installation and probably a second entrance. The number of rooms used for living was probably no more than twelve.91 This means that the house may have been inhabited by about a dozen individuals or, at most, by twenty-three to twenty-seven persons.

 Campagnoli, “Le fasi edilizie”.  De Maria, “Suasa, la città e la sua storia. 91  Tornatore, “Una domus con mosaici a Sant’Angelo in Vado”. 89 90

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

237

Fig. 7.13  Plan of the Imperial domus at Tifernum Mataurense. (Tornatore, “Domus con mosaici”, 883, fig. 2)

A last very well-documented Imperial house is the huge “Domus dei Coiedii” at Suasa (Fig. 7.12): a domus that extended over approximately 3300  m2 at the time of its maximum expansion in the second century CE. Approximately 55 per cent of the complex was taken up by a large porticoed garden/peristylium of about 1800 m2, in which a private thermal installation of about 225 m2 was inserted towards the end of the second century CE.  This left about 1500  m2 for the residential or representational parts of the house, subdivided into a total of forty rooms organised around an atrium (A). Not all of these rooms, however, were used for living purposes: one room contained another thermal installation (AS); another large room of 120 m2 was an audience room (S); four rooms formed a probable hospitium or lodging space (AF-AQ); and a further thirteen rooms were most likely tabernae (AB-AQ) and service or storage spaces (AG-AZ). With the further exception of the entrance of the house

238 

D. VAN LIMBERGEN AND F. VERMEULEN

(K) and the numerous passageways and corridors, this leaves some fifteen rooms available for living.92 Based on an average space of thirty-four to thirty-nine m2 per person, the house would have been inhabited by thirty-­ eight to forty-four individuals. If we apply the ratio of one person per room, however, only fifteen people may have resided in the domus. The comparison between the two calculation methods neatly highlights how houses of similar size may have had slightly different occupation densities due to a differing number of individual living rooms. The same is true for smaller houses as they may have had occupation densities equalling those of larger houses because of a similar number of (smaller) rooms. The “Domus dei Coiedii” in particular is a fine example of how a much larger house does not necessarily imply a proportionally larger household but rather may be linked to a family disposing of more wealth and space. From the available evidence it is also clear that the occupation density of a house was not fixed over time. Whilst we cannot hope to express these diachronic alterations in exact numbers, the process is well attested by the various adaptions of space in a domus through time. Perhaps a first useful order of magnitude may be acquired by setting the absolute minimum occupation at the derived room figures, whereas the higher figures derived from the surface area calculations may be considered as the absolute maximum occupation. Still, in some cases this leaves us with quite large variations in household size, such as seven to fourteen persons for a 480 m2 large house, or twelve to twenty-seven persons for a 925 m2 house, or fifteen to forty-four persons for a 3300 m2 house (Table 7.1, derived D1). Another option might be to use the lower figures from the surface area calculations as maximum occupation densities. Obviously, this gives us overall lower density ranges, but at least the pattern becomes more balanced across the different house sizes (Table 7.2, derived D2). In order to deduce possible urban population densities from these houses, two methods may be considered. A first method uses the average surface area derived from the currently known house sizes—that is, approximately 690 m2 based on a total of twenty-six houses, excluding as such the exceptionally large “Domus dei Coiedii”—and applies the 92  Campagnoli, “Fasi edilizie”). For the Domus dei Coiedii, the corresponding letters on the map are the following: garden/peristylium (BT, AD, AE, BA, BB, BC, BD, BE, BQ, BR, BG) with inserted baths (BL, BI, BP, BM, BS, BN), hospitium (AF, AK, AN, AQ), tabernae (AB, AC, AL, AH, AQ), service spaces (AG, AP, AR, AT, AU, AV, A, AZ), corridors (T, D, F, N, AI), living spaces (AA, U, V, R, Q, P, O, C, E, G, H, I, L, M1, M2).

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

239

Table 7.1  Estimated number of residents per house size Size (m2)

N° of rooms

189 392 455 459 480 925 3300

3 7 7 10 7 12 15

D1 (34–39 m2/p) 4–5 10–11 11–13 11–13 12–14 23–27 38–44

D2 (1 room/p) 3 7 7 10 7 12 15

Derived D1 3–5 7–11 7–13 10–13 7–14 12–27 15–44

Derived D2 3–4 7–10 7–11 10–11 7–12 12–23 15–38

density figure obtained from the surface area method, which in this case would result in an occupation of seventeen to twenty persons. Next, the number of houses that may fit in the urban area reserved for private space is then calculated and multiplied by the occupation density figure. For example, at Potentia this would result in a total of 160 houses, or an estimated population of between 2720 and 3200 persons. The population density for the inhabited area only would be 245 to 288 p/ha, whilst the overall population density of the town would be 148 to 174 p/ha. For Trea, this would be a total of 92 houses or 1564–1840 persons, with a density of 246 to 289 persons per hectare for the residential area, and one of 142 to 167 persons per hectare for the whole urban area. However, several objections may be raised against the representative value of these figures. First, the use of an overall average surface figure does not allow for incorporating the occupation densities based on room numbers. As these densities have consistently hinted at lower occupation numbers, we might thus be overestimating the number of people that lived in these towns. Second, this methodology does not take into account the diversity in house sizes that we know to have characterised the private layout of a Roman town. As such, we decided not to use this calculation method. Using an alternative method, we first ranked the houses according to surface bands of 200  m2 from 0 to 1000  m2. The larger houses were grouped together in bands of 1000 to 2000 m2 and more than 2000 m2 respectively. As such, the majority of the houses (ten) could be situated in the 400 to 600 m2 range; five houses ranged between 1000 and 2000 m2; four houses fell into the 200 to 400 m2 and 800 to 1000 m2 bands respectively; and two houses occupied surfaces between 600 and 800 m2. The

240 

D. VAN LIMBERGEN AND F. VERMEULEN

Table 7.2  House sizes in central Adriatic Italy ranked according to consecutive 200 m2 bands, with the exception of the >1000 m2 category 0–200 189

1

200–400 300 360 390 392

4

400–600 400 (×2) 455 459 460 470 480 (×2) 540 560 10

600–800

800–1000

1000–2000

>2000

600 625

800 875 925 955

1200 (×2) 1300 1600 (×2)

3300

2

4

5

1

less than 200 m2 and more than 2000 m2 bands each contained only one house (Table 7.2). However, as this classification did not seem to reflect well the original distribution of house sizes—no house was smaller than 300 m2 in the 200 to 400 m2 band, only two houses were larger than 500 m2 in the 400 to 600 m2 band, and no house was larger than 700 m2 in the 600 to 800 m2 group—we decided to regroup them according to better adapted categories. First, we rearranged categories 2 to 4 into 2 bands of 200 m2 each, that is, 300 to 500 and 500 to 700  m2. We then excluded the 189  m2 house from the calculations as it had been quickly transformed into a 455 m2 house, already classified under the 300 to 500 m2 band. Next, as the more than 2000  m2 band only consisted of the exceptionally large “Domus dei Coiedii”, we also omitted this category and instead added its 1500 m2 living space to the 1000 to 2000 m2 band. The 800 to 1000 m2 group was kept the same. This resulted in a total of twelve houses in the first band (300 to 500), four houses in the second band, three houses in the third and six houses in the fourth (Table 7.3). The use of this classification as the basis for our demographic calculations has several advantages. First, it enables us to propose a model in which the layout of the residential area would be made up of houses of different sizes. For example, ranked according to this classification, it follows that for each very large house of more than 1000 m2, there would be two houses of 300 to 500  m2 and approximately 0.7 house of 500 to 700 m2 and 800 to 1000 m2 each. Although, admittedly, based on a small body of evidence and largely theoretical, it would at least provide us with

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

241

Table 7.3  House sizes in central Adriatic Italy ranked according to adjusted 200 m2 bands 300–500

500–700

800–1000

1000–2000

300 360 390 392 400 (×2) 455 459 460 470 480 (×2) 12 420

540 560 600 625

800 875 925 955

1200 (×2) 1300 1500 1600 (×2)

4 580

4 890

6 1400

a reconstruction akin to the situation encountered in other Roman towns in the region. Second, it also allows us to derive an average figure for each of the subgroups rather than one overall average figure for all the house sizes combined. In this case, the average figures would be 420 m2, 580 m2, 890 m2 and 1400 m2. Third, this ranking also makes it possible to employ the combined occupation figures derived from both room numbers and surface area. For example, we might set the minimum occupation of a 420 m2 house at seven persons, whereas a number of twelve might be a reasonable maximum occupation. Likewise, the minimum figure for a 580 m2 house might be ten, whereas the maximum figure might be fifteen. The situation becomes somewhat more complicated for the two upper categories as we have seen that substantially larger houses do not always imply substantially larger households. The derived densities of twelve to twenty-three persons for a 890  m2 house—and of fifteen to thirty-eight persons for a 1400 m2 house—represent too wide a range for useful calculations, but densities of twelve to seventeen and fifteen to twenty respectively might be considered reasonable alternatives. As such, this methodology was chosen to calculate the number of inhabitants of Potentia and Trea in the Early/High Imperial period. For Potentia, the application of this calculation method would result in a mix of approximately sixty-eight houses in the first category (420 m2), twenty-four houses in both the second (500 to 700  m2) and the third (800 to 1000 m2) categories and thirty-four houses in the upper category

242 

D. VAN LIMBERGEN AND F. VERMEULEN

(1400  m2), fitted in over a residential area of about 11.1 hectares. In demographic terms, such a reconstruction would point towards an estimated population of between 1514 and 2264 persons, that is, a density of 137 to 204 persons per hectare for the residential area only, or 82 to 123 persons per hectare for the total town area (18.4 hectares) when we include the street grid and the public space. For Trea, the calculations resulted in about thirty-eight houses of 420 m2, thirteen houses of both 580 and 890 m2 and nineteen houses of 1400 m2 for the 6.35 hectares residential area. The demographic translation of this reconstruction would be a population of approximately 844 to 1262 people, or a density of 133 to 199 p/ha for the residential area only (6.35 hectares) and a density of 77 to 115 p/ha for the whole town (11 hectares).

4   Conclusion In a recent work on the demographic history of Roman Italy between 225 BCE and CE 100, Luuk de Ligt argued that around the time of the Augustan census (28  BCE) most towns in central Italy had population densities between 120 and 150 persons per hectare. His argument was largely based on a discussion of the structural evidence from Pompeii and Herculaneum, in combination with a comparative study of urban population densities in peninsular Italy in medieval and early modern times.93 In this paper, for the first time, we have been able to examine in detail the Early/High Imperial structural evidence for two central Adriatic Italian towns—known mainly through aerial photography and geo-­archaeological survey—and to discuss the demographic implications of their physical reconstructions. Based on the information that is currently available, these reconstructions seem to point towards the existence of significantly lower urban population densities in the range of 75 to 125 persons per hectare in this area of Roman Italy from the Augustan period onwards. The implications of these results are of potential major importance for the study of urban demography in Italy in Imperial times as we believe that we have shown that a combination of systematic fieldwork in abandoned town areas with the incorporation of all relevant archaeological, literary and epigraphic data from the wider study area can be a great starting point for establishing urban demographic models based on lesser-known archaeological contexts. In the case of Trea in particular, it has been shown that  De Ligt, Peasants, Citizens and Soldiers.

93

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

243

some of the smaller urban agglomerations in Italy—that is, under 15 hectares—might be better defined as service centres rather than population centres, with almost 45 per cent of the settlement area in this case being occupied by public architecture (30 per cent) and streets (15 per cent), resulting in a relatively small habitation component (about 55 per cent), and thus a rather low overall population density (77 to 115 p/ha). As such, we are convinced that the methodology discussed in this paper has the potential to greatly enhance the debate on Roman demography in the future, in particular with regard to discussions on wider urban and rural population sizes or urbanisation rates in Italy in the Imperial period.

Bibliography Bairoch, Paul. “Urbanisation and the Economy in Pre-Industrial Societies: The Findings of Two Decades of Research.” The Journal of European Economic History 18 (1989): 239–90. Beloch, Karl J. “Le città dell’Italia antica,” Atene e Roma 1 (1898): 257–78. Biocco, Emanuela. Città romane. Matelica. Rome: L’Erma di Bretschneider, 2000. Boschi, Federica and Michele Silani. “Aerofotografia e geofisica nella Valle del Misa (Ancona): nuovi dati per la topografia urbana di Ostra.” Rivista di Topografia Antica 22 (2012): 69–82. Brunt, Peter A. Italian Manpower: 225 B.C.–A.D. 14. Oxford: Clarendon Press, 1971. Calza, Guido. Scavi di Ostia: topografia generale. Rome: Libreria dello Stato, 1926. Campagnoli, Paolo. “Le fasi edilizie: revisioni e recenti acquisizioni.” In Archeologia nella valle del Cesano. Gli scavi di Suasa e Santa Maria in Portuno (1996–2009), Atti delle giornate di studio per i venti anni di ricerche archeologiche dell’Università di Bologna, Castelleone di Suasa, Corinaldo, San Lorenzo in Campo 18–19 dicembre 2008, edited by Enrico Giorgi and Giuseppe Lepore, 319–34. Bologna: Ante Quem, 2010. Capriotti Vittozi, Giuseppina. Oggetti, idee, culti egizi nelle Marche. Dalle tombe picene al tempio di Treia (Picus Supplementi VI). Tivoli: Tipigraf, 1999. Carcopino, Jérôme. Daily Life in Ancient Rome: the People and City at the Height of the Empire. New Haven: Yale University Press, 1940. Casselberry, Samuel E. “Further refinement of Formulae for Determining Populations from Floor Area.” World Archaeology 6 (1974): 117–22. Castagnoli, Ferdinando. Orthogonal Town Planning in Antiquity. Cambridge: The MIT Press, 1971. Catani, Enzo. “Tifernum Mataurense: note di topografia urbana, viabilità e bonifica agraria del territorio.” In L’appennino in età romana e nel primo medioevo. Viabilità e popolamento nelle Marche e nell’Italia centro-settentrionale, Atti del

244 

D. VAN LIMBERGEN AND F. VERMEULEN

Convegno di Corinaldo, 28–30 giugno 2001, edited by Marco Destro and Enrico Giorgi, 97–117. Bologna: Ante Quem, 2004. Catani, Enzo and Walter Monacchi. Tifernum Mataurense 2: il territorio. Urbino: Edizioni Tored, 2010. Cavallo, Andrea. “Sentinum, le Terme Urbane. Una rilettura critica dell’edificio.” In Sentinum 295 a.C. Sassoferrato 2006. 2300 anni dopo la battaglia. Una città romana tra storia ed archeologia, edited by Maura Medri, 367–69. Rome: L’Erma di Bretschneider, 2008. Chamberlain, Andrew. Demography in Archaeology. Cambridge: Cambridge University Press, 2006. Chevallier, Ramond. La Romanisation de la celtique du Pô: essai d’histoire provinciale. Rome: École française de Rome, 1983. Conventi, Marta. Città romane di Fondazione. Rome: L’Erma di Bretschneider, 2004. Curet, Antonio L. “New Formulae for Estimating Prehistoric Populations for Lowland South America and the Caribbean.” Antiquity 72 (1998): 359–75. Dall’Aglio, Pier-Luigi, Michele Silani and Cristian Tassinari. “Nascita e sviluppo monumentale della città romana di Ostra (AN).” In Evolução da Paisagem Urbana. Sociedade e Economia, edited by Maria Do Carmo Ribeiro and Arnaldo Sousa Melo, 11–27. Braga: CITCEM, 2012. De Graaf, Peter. Late Republican-Early Imperial Regional Italian Landscapes and Demography. Oxford: BAR Publishing, 2012. De Maria, Sandro. “Suasa, la città e la sua storia. Venti anni di scavi e ricerche.” In Archeologia nella valle del Cesano da Suasa a Santa Maria in Portuno, Atti del Convegno per i venti anni di ricerche dell’Università di Bologna, Castelleone di Suasa, San Lorenze in Campo 18–19 dicembre 2009, eds. Enrico Giorgi and Giuseppe Lepore, 19–32. Bologna: Ante Quem, 2010. De Maria, Sandro and Enrico Giorgi. “Urbanistica e assetti monumentali di Suasa. Novità dalle ricerche recenti.” In Epigrafia e Archeologia romana nel territorio marchigiano, in memoria di Lidio Gasperini, Atti del Convegno, Macerata 22–23 aprile 2013, edited by Gianfranco Paci, 79–142. Rome: Edizioni Tored, 2013. De Roche, C.D. “Population estimates from settlement area and number of residents.” Journal of Field Archaeology 10 (1983): 187–92. Drennan, Robert D., C.  Adam Berry and Christian E.  Peterson. Regional Settlement Demography in Archaeology. New  York: Eliot Werner Publications, 2015. Duncan-Jones, Richard. The Economy of the Roman Empire. Quantitative Studies. Cambridge: Cambridge University Press, 1982. Eschebach, Hans. Die stadtebauliche entwicklung des Antiken Pompeij. Heidelberg: Kerle Verlag, 1970.

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

245

Fabrini, G.M. “Dal culto pagano al culto cristiano: testimonianze archeologiche e documentarie per l’area del SS. Crocifisso a Treja.” Picus 10 (1990): 107–75. Fentress, Elizabeth et al. “Cosa in the Republic and the Early Empire.” In Cosa V. An Intermittent Town, Excavations 1991–1997, edited by Elizabeth Fentress, 13–62. Ann Arbor: University of Michigan Press, 2003. Giuseppe Fiorelli, Giuseppe. Gli Scavi di Pompeii dal 1862 al 1872. Relazione al Ministro della Istruzione Pubblica. Naples: Tipografia Italiana nel Liceo V Emanuele, 1873. Gori, Giancarlo and Mario Luni “Edificio termale a Forum Sempronii.” Picus 2 (1982): 119–30. Hanson, John William. An Urban Geography of the Roman World. Oxford: Archaeopress, 2016. Hopkins, Keith. Conquerors and Slaves. Cambridge: Cambridge University Press, 1978.. Kardulias, P. Nick. “Estimating Population at Ancient Military Sites: The Use of Historical and Contemporary Analogy.” American Antiquity 57 (1992): 276–87. Kolb, Charles C. “Demographic Estimates in Archaeology: Contributions from Ethnoarchaeology on Mesoamerican Peasants.” Current Anthropology 26 (1985): 581–99. Kramer, Carol. Village Ethnoarchaeology. Rural Iran in Archaeological Perspective. New York: Academic Press, 1982. Laurence, Ray. “The Organization of Space in Pompeii,” in Urban Society in Roman Italy, edited by Tim J. Cornell and Kathryn Lomas, 65–82. London and New York: Routledge, 1995. Lazer, Estelle. Resurrecting Pompeii. London and New York: Routledge, 2009. Le Blanc, Steven. “An Addition to Naroll’s Suggested Floor Area and Settlement Population Relationship.” American Antiquity 36 (1971): 210–11. De Ligt, Luuk. “The Population of Cisalpine Gaul in the Time of Augustus.” In People, Land and Politics. Demographic Developments and the Transformation of Roman Italy 300 BC–AD 14, edited by Luuk De Ligt and Simon Northwood, 139–83. Leiden and Boston: Brill, 2008. De Ligt, Luuk. Peasants, citizens and Soldiers. Studies in the Demographic History of Roman Italy 225 BC–AD 100. Cambridge: Cambridge University Press, 2012. Lo Cascio, Elio. “The Population of Roman Italy in Town and Country.” In Reconstructing Past Population Trends in Mediterranean Europe, edited by Kostas Sbonias and John Bintliff, 161–71. Oxford: Oxbow Books, 1999. Lollini, Delia. “La civiltà picena.” Popoli e civiltà dell’Italia antica V (1976):107–95. Maraldi, Lisa. Falerio. Rome: L’Erma di Bretschneider, 2002. Marengo, S.M. Regio V Picenum  - Trea. Supplementa Italica, n.s. 18. Roma: Edizioni di storia e letteratura, 2000.

246 

D. VAN LIMBERGEN AND F. VERMEULEN

McIntyre, Lauren J. “Reconstructing population size in a Romano-British colonia: the case of Eboracum.” Journal of Roman Archaeology 28 (2015): 413–429. Medri, Maura “Materiali per una nuova Forma Urbis di Sentinum.” In Sentinum. Ricerche in corso I, edited by Maura Medri, 199–218. Rome: L’Erma di Bretschneider, 2008. Meiggs, Russell. Roman Ostia. Oxford: Clarendon Press, 1960. Mercando, Liliano. “Marche-Rinvenimenti di insediamenti rurali: Portorecanati (Macerata).” Notizie degli Scavi 33 (1979): 180–280. Mercando, Liliano, S.  Sorda and M.  Capitanio. “La necropoli romana di Portorecanati.” Notizie degli Scavi 28 (1974): 142–430. Montali, G. “Considerazioni sulle strutture edilizie dell’area del Tempio-­ criptoportico.” In Urbs Salvia I. Scavi e ricerche nell’area del Tempio della Salus Augusta, edited by G.M. Fabrini, 119–87. Macerata: Simple, 2013. Morley, Neville. “Cities and Economic Development in the Roman Empire.” In Settlement, Urbanisation and Population, edited by Alan Bowman and Andrew Wilson, 143–60. Oxford: Oxford University Press, 2011. Moscatelli, Umberto. “Municipi romani della V ‘regio’ augustea: problemi storici ed urbanistici del Piceno centro-settentrionale (III-I sec. a.C.).” Picus 5 (1985): 51–97. Moscatelli, Umberto. Trea (Forma Italiae 33). Firenze: Olschki, 1988. Naroll, Raoul. “Floor Area and Settlement Population,” American Antiquity 27 (1962): 587–589. Nibby, Antonio. Viaggio antiquario ad Ostia. Rome: Società tipografica, 1829. Nissen, Heinrich. Pompeianische studien zur Städtekunde des altertums. Leipzig: Breitkopf und Härtel, 1877. Paci, Gianfranco. “Le iscrizioni romane di Potentia.” In Atti del XXIX Convegno di Studi Maceratesi (Porto Recanati, 13–14 novembre 1993), 1–25. Macerata: Centro di Studi Storici Maceratesi, 1995. Paci, Gianfranco. “Indagini recenti e nuove conoscenze sulle città romane del territorio marchigiano.” AnnMacerata 32 (1999): 201–44. Paci, Gianfraco and Edvige Percossi-Serenelli. “Il paradigma della romanizzazione: la colonia di Potentia.” In Archeologia nel Maceratese: nuove acquisizioni, edited by Giulano De Marinis, Gianfranco Paco, Edvige Percossi and Mara Silvestrini, 190–200. Macerata: Fondazione Cassa di Risparmio della Provincia di Macerata, 2005. Packer, James E. The Insulae of Imperial Rome. Rome: American Academy in Rome, 1971. Paoli, Ugo Enrico. Rome: It’s people, life and customs. London: Bristol Classical Press, 1963. Percossi, Edvige. “La necropoli di Potentia. Nuovi rinvenimenti.” Studi Maceratesi 51 (2007): 547–72.

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

247

Percossi, Edvige. “Le fasi repubblicane di Potentia.” In I processi formativi ed evolutivi della città in area adriatica, edited by Giuliano De Marinis, G.M. Fabrini, Gianfranco Paci, Roberto Perna and Maria Silvestrini, 309–30. Oxford: BAR Publishing, 2012. Percossi Serenelli, Edvige. “Potentia. Fonti letterarie e fonti archeologiche.”, In Atti del XXIX Convegno di Studi Maceratesi, Porto Recanati 13–14 novembre 1993, 27–55. Macerata: Centro di Studi Storici Maceratesi, 1995. Percossi Serenelli, Edvige. Potentia. Quando poi scese il silenzio… Rito e società in una colonia romana del Piceno fra Repubblica e tardo impero. Milan: Motta, 2001. Postgate, Nicholas. “How Many Sumerians per Hectare? Probing the Anatomy of an Early City.” Cambridge Archaeological Journal 4 (1994): 47–65. Robinson, O.R. Ancient Rome: City Planning and Administration. New  York: Routledge, 1992. Russell, J.C. “The Population and Mortality at Pompeii.” International Committee on Urgent Anthropological and Ethnological Research Bulletin 19 (1977): 107–14. Salvini, M. Area archeologica e Museo La Fenice. Senigallia: Regione Marche, 2003. Schacht, Robert M. “Estimating Past Population Trends.” Annual Review of Anthropology 10 (1981): 119–40. Sisani, Simone. Umbria/Marche. Rome: Laterza, 2006. Sommella, Paolo. Italia antica, l’urbanistica romana. Rome: Jouvence, 1988. Staffa, Andrea. “Teramo: nuovi dati per la ricostruzione dell’assetto antico della città.” Xenia 19 (1990): 19–30. Stambaugh, John E. The Ancient Roman City. Baltimore: JHU Press, 1988. Storey, Glenn R. “The Population of Ancient Rome.” Antiquity 71 (1997): 966–78. Sumner, William M. “Population and Settlement Area: An Example from Iran.” American Anthropologist 91 (1989): 631–41. Tornatore, M. “Una domus con mosaici a Sant’Angelo in Vado (PU).” In Amore per l’antico. Dall Tirreno all’Adriatico, dalla Preistoria al Medioevo e oltre. Studi di antichità in ricordo di Giuliano De Marinis, edited by Gabriele Baldelli, Fulvia Lo Schiavo, 881–91. Rome: Scienze e Lettere, 2014. Van Limbergen, Dimitri. Pots, Presses, People and Land. the Role of Overseas Export and Local Consumption Demand in the Development of Viticulture and Oleoculture in Central Adriatic Italy. Ghent University, Unpublished PhD thesis, 2015. Vermeulen, Frank. “The Potenza Valley Survey (Marche).” In New developments in Italian landscape archaeology. Theory and methodology of field survey, land evaluation and landscape perception, pottery production and distribution. Proceedings of a three-day conference held at the University of Groningen, April 13–15, 2000, edited by Peter Attema, Gert-Jan Burgers, Ester Van Joolen,

248 

D. VAN LIMBERGEN AND F. VERMEULEN

Martijn Van Leusen and Benoît Mater, 104–06. Oxford: British Archaeological Reports Oxford, 2002. Vermeulen, Frank. “Functional Zoning and Changes in the Use of Space in the Roman Town of Potentia. an Integrated Survey Approach.” In Thinking About Space. the Potential of Surface Survey and Contextual Analysis in the Definition of Space in Roman Times, edited by Hannelore Vanhaverbeke, Jeroen Poblome, Frank Vermeulen, Marc Waelkens and Raymond Brulet, 233–49. Turnhout: Brepols, 2008. Vermeulen, Frank. “Città romane nella valle del Potenza: fotografia aera e prospezioni topografiche.” In Omaggio a Nereo Alfieri. Contributi all’archeologia marchigiana, Atti del Convegno di Studi, Loreto 9–11 Maggio 2005, edited by Giuliano De Marinis and Gianfranco Paci, 613–38. Rome: Edizioni Tored, 2009. Vermeulen, Frank. “Reviewing 10 Years of Aerial Photography in the Valley of the River Potenza (Marche).” In Archeologia aerea, Proceedings of the International Congress, Rome 2009, edited by Giuseppe Ceraudo, 255–66. Foggia, 2011. Vermeulen, Frank. “Potentia: A Lost New Town.” In Urbes Extinctae. Archaeologies of Abandoned Classical Towns, edited by Neil Christie and Andrea Augenti, 77–95. Farnham and Burlington: Ashgate, 2012a. Vermeulen, Frank and Catharina Boullart. “The Potenza Valley Survey: Preliminary Report on Field Campaign 2000.” BABesch 76 (2001): 1–18. Vermeulen, Frank and Francesca Carboni. “Measuring Urbanisation in the Late Republican and Early Imperial Landscape of Adriatic Central Italy.” In Bollettino di Archeologia Online. Edizione Speciale: Multidisciplinary Approaches to Classical Archaeology, Proceedings of the 17th International Congress of Classical Archaeology, Rome 22–25 sept. 2008 (2011). Vermeulen, Frank and Francesca Carboni. “Approaching Roman Fora with Non-­ Invasive Urban Survey: Four Examples in Picenum.” In Proceedings congress “FORUM. Strutture, funzioni e sviluppo degli impianti forense in Italia (IV sec. a.C.–I sec. d.C.), Rome, 2013. Vermeulen, Frank, Sophie Hay and Geert Verhoeven. “Potentia: An Integrated Survey of a Roman Colony on the Adriatic Coast.” PBSR 74 (2006): 203–36. Vermeulen, Frank and Patrick Monsieur. “Le système défensif et la chronologie de la colonie républicaine de Potentia (Marches, Italie).” In Industria Apium. L’archéologie: une démarche singulière, des pratiques multiples. Hommages à Raymond Brulet, edited by Marco Cavalieri, Éric De Waele and Laure Meulemans, 163–83. Louvain-la-Neuve: Presses Universitaires de Louvain, 2012. Vermeulen, Frank and Geert Verhoeven. “The Contribution of Aerial Photography and Field Survey to the Study of Urbanisation in the Potenza Valley (Picenum).” Journal of Roman Archaeology 17 (2004): 57–82.

7  A METHOD FOR ESTIMATING ROMAN POPULATION SIZES FROM URBAN… 

249

Vermeulen, Frank and Geert Verhoeven. “An Integrated Survey of Roman Urbanisation at Potentia, Central Italy.” Journal of Field Archaeology 31 (2006): 395–410. Vermeulen, Frank, Patrick Monsieur and Catharina Boullart. “The Potenza Valley Survey: Preliminary Report on Field Campaign 2001.” BABesch 77 (2002): 49–71. Vermeulen, Frank, Morgan De Dapper, Branko Music, Patrick Monsieur, Hélène Verreyke, Francesca Carboni, Sophie Dralans, Geert Verhoeven, Lieven Verdonck, Sophie Hay, Martin Sterry, Paul de Paepe and Sarah De Seranno. “Investigating the impact of Roman urbanisation on the landscape of the Potenza valley. A report on fieldwork in 2007.” BABesch 84 (2009): 85–110. Vermeulen, Frank, Marco Destro, Patrick Monsieur, Francesca Carboni, Sophie Dralans and Dimitri Van Limbergen. “Scavi presso la porta occidentale di Potentia: notizia preliminare.” Picus 31 (2011): 169–205. Vermeulen, Frank, Gert-Jan Burgers, Simon Keay and Christina Corsi. Urban Landscape Survey in Italy and the Mediterranean, Oxford: Oxbow Books, 2012a. Vermeulen, Frank, Morgan De Dapper, Patrick Monsieur, Paul de Paepe, Lieven Verdonck, Devi Taelman, Jésus Garcia-Sanchez, Sophie Dralans, Dimitri Van Limbergen, Debby Van den Bergh, Possum Pincé, B. Ullrich and Wieke De Neef. Archaeological Investigations in the Potenza Valley, Field Season 2012. Ghent, Ghent University Internal Report, 2012b. Vermeulen, Frank, Božidar Slapzak and Dimitrij Mlekuz.“Surveying the townscape of Roman Trea (Picenum).” In Archaeological survey and the city, edited by Paul Johnson and Martin Millett, 261–92. Oxford: Oxbow Books, 2012b. Wallace-Hadrill, Andrew. “Houses and Households: Sampling Pompeii and Herculaneum.” In Marriage, Divorce and Children in Ancient Rome, edited by B. Brawson, 191–227. New York: Clarendon Press, 1991. Wallace-Hadrill, Andrew. Houses and Society in Pompeii and Herculaneum. Princeton: Princeton University Press, 1994. Wallace-Hadrill, Andrew. “Case e abitanti a Roma.” In Roma impériale. Una metropoli antica, edited by Elio Lo Cascio, 173–200. Rome: Carocci, 2000. Wiessner, P. “A Functional Estimator of Population from Floor Area,” American Antiquity 39 (1974): 343–350. Wilson, Andrew. “City Sizes and Urbanisation in the Roman Empire” In Settlement, Urbanisation and Population, edited by Alan Bowman and Andrew Wilson, 161–95. Oxford: Oxford University Press, 2011. Yavetz, Z. “The Living Conditions of the Urban Plebs in Republican Rome.” Latomus 17 (1958): 500–17. Zorn, Jeffrey R. “Estimating the Population Size of Ancient Settlements: Methods, Problems; Solutions, and a Case Study.” Bulletin of the American School of Oriental Research 295 (1994): 31–48.

CHAPTER 8

Complexity and Urban Hierarchy of Ancient Urbanism: The Cities of Roman Asia Minor Rinse Willet

1   Complexity, Scaling and Cities According to a report of the United Nations published in July 2014 the number of people living in cities has risen dramatically from 746 million in 1950 to 3.9 billion in 2014, 54 per cent of the world’s human population. The report projects that nearly 66 per cent will be living in cities by 2050.1 The city is a global phenomenon and recent scholarship on urbanism has focused on common features of cities in different geographical and chronological settings. This often requires taking a reductionist perspective. Quantifiable aspects, such as population size, crime rates, productivity or features found in some variant form in most cities, such as roads, public buildings and so on, have become pillars in explaining the way cities exist 1  United Nations. Department of Economic and Social Affairs. Population Division World Urbanization Prospects: the 2014 Revision, New  York, 2015 (http://esa.un.org/unpd/ wup/Publications/Files/WUP2014-Report.pdf).

R. Willet (*) University of Leuven, Leuven, Belgium e-mail: [email protected] © The Author(s) 2021 K. Verboven (ed.), Complexity Economics, Palgrave Studies in Ancient Economies, https://doi.org/10.1007/978-3-030-47898-8_8

251

252 

R. WILLET

and develop.2 Critique has focused mostly on the fact that reductionism and the hypothetical-deductive laws derived from sometimes problematic sets of data underestimate the role of human agency and make us lose sight of the diversity, variability and context specificity—in other words of the complex nature—of the actual data upon closer inspection. In archaeology this discussion has been going on at least since the 1980s. In essence it constitutes a dispute between researchers who focus on (the effects of) social structures in society (processualism) and those who focus on the effects of individual agency (post-processualism). John Bintliff, although critical of post-processual approaches, suggested early on that both approaches could be bridged. In effect, the best elements of both approaches could be combined by applying a Braudelian model of processes happening on multiple temporal wavelengths and integrating models of chaos and complexity theory in this.3 Such a combination accounts for the agency of individuals but recognizes their unintended (sometimes major) middle- and longer-term results while also accepting that social structures are more than the sum of their parts. Essentially, the complexity of structures and systems increases through the interaction of their components, driving a system closer to “the edge of chaos”. In such a state minor changes can have major effects, which include rapid collapse, sustained complexity, or the achievement of new levels of complexity. The more elaborate the initial structures are, the higher the risk of chaos or breakdown becomes. Bintliff applied this model to multiple historical scenarios ranging from settlement patterns observed in Classical to late antique Boeotia to the formation of cities and the history of the Roman Empire. Later Alexander Bentley and Herbert Maschner advocated complexity theory as offering a bridge between structural models and agency, arguing its potential to combine our limited knowledge or ignorance on human activity at the micro-scale with “law-like” observations on the macro-scale.4 Individual and unpredictable actions and events on the micro-level affect broader patterns on the macro-level that we can observe both in cities and in regions or environments. It is from these broader patterns that the particularities of context variation and diversity and the possible unpredictability of the cumulative effect of individual behaviours can start to be discerned. Cities and their formation are good examples of how  De Vries, European Urbanisation; Bettencourt, “The Origins of Scaling”.  Bintliff, “Catastrophe, Chaos and Complexity”; Bintliff, “Time, Structure, and Agency”. 4  Bentley and Maschner, Complex Systems and Archaeology. 2 3

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

253

the actions of a number of individual agents produce structures that display great complexity and behave as complex systems. Recent studies on urbanism for various periods around the world suggest that aspects of cities that are related to this complexity (such as population levels, urban amenities, city layout and economic activity) have scalar properties. Luis Bettencourt observed this by creating a predictive model, in which components such as land area, network volume and length (i.e. urban infrastructure), social interactions, socio-economic rates, and so on were tested for hundreds of modern cities in different parts of the world. The basic assumptions in the model include the mixing of populations, incremental network growth, the constraint of human efforts and the proportionality of socio-economic outputs to local social interactions. The result of Bettencourt’s tests showed super-linear relationships in these aspects of cities; for instance for populations of cities in the US the figure of road miles was shown to be incrementally higher for larger cities as was the rate of innovation measured by the number of registered patents.5 In simple terms, this is the result of the “vascular” character of modern cities; a city road grid consists of main roads and highways connecting cities but also of a local level network of roads, streets, paths and alleys connecting neighbourhoods. In this model, self-similarity is expected. For instance, neighbourhoods have access to small local places of worship, but on a higher scale, each city-district or the city as a whole has a large central church, mosque or temple.6 The complexity of a city’s structure shows regularity through scaling. This is conceptually reminiscent of the idea derived from complexity theory that positive feedback to adaptive structures favours the crystallization of persistent inclusive structures or networks, which Bintliff calls the “attractors”. It is, therefore, at least worth assessing to what extent greater complexity of cities is reflected in the scaling behaviours they reflect. Is scaling a result of or a response to fractal-like self-similarity within cities or within a network of cities? While this model has been researched and applied to modern cities, for pre-modern cities these processes are less well attested and understood, which can partly be explained by the limitations of the archaeological record.7 Furthermore, Bettencourt’s model, although it includes constraints on human efforts, implicitly suggests that improvements in urban  Bettencourt, “The Origins of Scaling”, p. 1438; Bettencourt et al., Invention in the City.  Ortman et al., “Settlement Scaling and Increasing Returns”. 7  Stanley et al., “Service Access in Premodern Cities”. 5 6

254 

R. WILLET

structure and services can overcome these constraints. Yet the cities of the past, particularly in the Roman Empire, are clearly limited in size, structure and scale of urban amenities. Only a few megalopoleis became relatively large, while the overwhelming majority remained much smaller than modern cities. To explain the size and economies of cities in the ancient world scholars have often emphasized these cities’ dependence on food supplies from their immediate hinterland or on exchanges based on commerce or on the socio-political status of the city.8 In Malthusian economic thinking, the size and growth or decline of a city depends on its access to resources as (primarily) food.9 Populations increase in size until they reach a balance with available resources, whereby an equilibrium emerges. More recently, scholars have argued that in antiquity, this equilibrium could be “up-scaled” by investments (for instance in infrastructure) or innovations (for instance in resource acquisition), allowing for a “moderate growth” of cities.10 Yet when the evidence for cities in Roman Asia Minor is considered, things seem far more complex than simply rise, stability/equilibrium and fall. Several cities grow in size from the Hellenistic to the Roman Imperial period, such as Ephesos, Pergamon, Herakleia Pontika, Sagalassos and others, and there is a general trend in the increase of the number of cities and settlements in the period from the third century BCE to the third century CE.11 At the same time, however, some cities disappear, move and are refounded, or are merged by synoikismoi, as in the case for instance of Atarneus, Heraclea ad Latmum and others.12 In other words, for the cities of Asia Minor change rather than stasis was the norm. The notion of “equilibrium” does not describe the situation of these cities well. Even “moderate growth” does not hold true for all cities in this area, since not all cities grew, while for example Pergamon expanded explosively in this period. Even the phenomenon of Anatolian urbanism and settlement patterns on the long term is a story of changing developments. The focus of major 8  Finley, The Ancient Economy; Garnsey, Cities, Peasants and Food; Dermody et  al., “A Virtual Water Network”. 9  Beinhocker, The Origin of Wealth, 55. 10  Scheidel, “Demography,” 51. 11  Groh, “Neue Forschungen”; Radt, Pergamon; Hoepfner, “Herakleia Pontike”; Poblome, Exempli Gratia; Broughton, Roman Asia Minor; Marek, Geschichte Kleinasiens. 12   Pirson and Zimmerman, “Das Umland von Pergamon”; Peschlow-Bindokat, “Herakleia”.

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

255

settlements inland (such as Hattusa during the second millennium BCE or the major settlement at Kerkenes Dağ during the first half of the first millennium BCE) during the Bronze Age and Iron Age stands in contrast to the coastal focus of large urban hubs of Classical antiquity. Shifts in large structural factors, such as the climate, influence settlement patterns. For example, John Haldon and his team showed that climatological changes during the first millennium CE had a significant impact on land-use and rural settlement patterns, by combining data from pollen, oxygen isotopes and other environmental, archaeological and textual findings.13 Yet climatic changes do not linearly and predictably direct changes in human habitation and land-use. It is one factor that has a (great) impact, but the way societies respond to such changes can differ. For the Hellenistic and Imperial periods (both falling under the Beyşehir occupation phase,14 major climatic shifts do not seem to take place until the beginning of the Byzantine period in the seventh century CE. The idea to study the societies and their environment as socio-­ecological systems can be traced back to the 1960s in the New Archaeology, which in turn was inspired by geography and has recently received new attention.15 New models have been proposed, in which societies, as “complex adaptive systems”, interact with their environments by adapting to the problems and challenges they face.16 In a complex adaptive system, many variables affect the system, although the key controlling variables may be more limited in number (and sometimes speed, for instance in the case of climate/environmental changes). In this model, adaptive cycles loop between various levels of connectedness, resilience and potential by which the system adapts to new situations as a result of internal or external changes.17 Many of these cycles acting at different spatiotemporal levels interact in a more comprehensive model of “panarchy”.18 Such models potentially allow us to capture the complexity and the changes of a group of cities over time.

 Haldon, The Empire that Would Not Die.  Eastwood et al., “Palaeoecological and Archaeological Evidence for Human Occupance”. 15  Bintliff, “Archaeology at the Interface,” 9–14. 16   Holling, “Understanding the Complexity”; Homer-Dixon, The Upside of Down; Beinhocker, The Origin of Wealth; Poblome, “The Economy of the Roman World as a Complex Adaptive System”. 17  Holling, “Understanding the Complexity,” 396. 18  Holling et al., “In Quest of a Theory of Adaptive Change”. 13 14

256 

R. WILLET

This paper will assess both the complexity and scaling properties of these models by using proxy data on urbanism of Roman Asia Minor collected in the framework of the ERC-funded project “An empire of 2000 cities: urban networks and economic integration in the Roman Empire” at Leiden University (directed by Luuk de Ligt and co-directed by John Bintliff). This project collected various proxy data on the cities of the Roman Empire, including status, size and information on urban amenities, such as the presence and size of public buildings. The objective was to capture the urban network at its supposed peak in terms of its number of cities and their demographic and economic flourishing. For Asia Minor this is assumed to be during the second century and the first half of the third century CE.19 This was a period of relative stability and supposed prosperity under the pax romana before the advent of political instability in the middle and later third century CE. During this period the region was more densely settled by cities than ever before, many of which had elaborate monuments. The objective of this paper is thus first to assess how complexity is observable and how scaling manifests itself. Several physical aspects of cities can be used as a test bed for this, but for the sake of brevity we will discuss only the distribution of cities and the presence and distribution of public buildings focusing on theatres and baths. The scalar properties are assessed through the size of the cities and the hierarchy in these sizes. As a next step, the observed patterns of complexity and urban hierarchy are assessed by whether they allow for a connection between the models, which both allow for scenarios of urban growth and decline. In other words, can increases or decreases in complexity be linked to the scalarity of urban systems? For instance, when cities change and grow or decline does this result in a complexity with scalar properties? First, the stage for these questions must be set by establishing research parameters and critically evaluating the cities of Roman Asia Minor and the research carried out on them.

2   Towards Defining Cities in Asia Minor In modern geography, the term city often describes a settlement with a minimum number of inhabitants. This definition has been applied also to studies of urbanism in past societies. For example, in studying the towns 19  For the full discussion, findings and the data used in this chapter, see Willet, The Geography of Urbanism.

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

257

and cities of Ottoman Anatolia, Suraiya Faroqhi considered a settlement as a town or a city primarily according to its number of registered taxpayers: 1000 inhabitants or more, but including smaller settlements with functional/administrative status. She then translated this into an estimate for the number of inhabitants (3000–5000).20 Jan de Vries in studying the cities of early modern Europe focused mostly on cities of 10,000 inhabitants or more.21 Peter Clark criticized de Vries for this, since it overlooks the functional role of small towns and the development of urban hierarchies or networks.22 Since population figures for antiquity are few in number and often problematic to interpret, the size of the urban area is used as a demographic proxy to define settlements. For republican Italy, Luuk de Ligt employed multiple categories based on size: forty or more hectares for large towns, twenty to forty hectares for medium towns and less than twenty hectares for non-urban settlements.23 In a study of the cities of Roman Asia Minor Hanson used categories from the Barrington atlas to distinguish “major urban centres”. This is not further specified and he in fact lists cities that are less than ten hectares in size.24 Bintliff pointed out that in the evolution from village to polis, the Wobstian threshold (the number of people required in a marriage/mating network to sustain a healthy human population) of about 500 to 600 people is important as this is the level at which a community can theoretically become endogamous and can start laying claims on territory without depending on other communities for its procreation.25 In this model at 500 people the village evolves into what Kirsten would call a “Dorfstadt”. In terms of built-up area 500 people would imply four to five hectares. The problem with settlement area sizes for Asia Minor is firstly that these are not well known for many cities and secondly that epigraphic, numismatic or other historical sources indicate that many very small places  Faroqhi, Towns and Townsmen, 9–11.  De Vries, European Urbanisation, 69. 22  Clark, European Cities and Towns, 2–3. 23  Ligt, Peasants, Citizens and Soldiers, 201. 24  Hanson, “The Urban System of Roman Asia Minor,” 237; Hanson, An Urban Geography addresses size of urban settlements in the Roman Empire in greater detail. However, despite reaching an average size of 54 ha, the fact remains that 596 out of 885 (67 per cent, p. 120–21) of the cities were smaller than 50 ha. This must be a minimum, as many cities are listed by Hanson’s (far from complete) catalogue for which no measurements are provided. It seems likely that few cities larger that were larger than 50 ha are listed among them. See also Donev and Willet, “Book Review” for further discussion of Hanson’s problematic approach. 25  Bintliff, “The Origins and Nature of the Greek City-State,” 47. 20 21

258 

R. WILLET

were nevertheless officially recognized as self-governing communities or cities. Furthermore, towns recognized as cities in the legal sense performed certain functions, for instance levying taxes from their territories.26 For example, the colonia of Parlais may have been inhabited by as few as 500 people, but its legal rank of colonia made it an independent city, free, in principle, to handle its own financial affairs without interference from central authorities.27 Therefore, this legal definition is important for understanding an urban network and how towns functioned in it. The Latin legal statuses of coloniae, municipia and civitates, with or without certain privileges, are attested for the cities of Anatolia but the most important legal status was that of the polis, which continued in relative autonomy from the Classical and Hellenistic periods into the Imperial period.28 Yet to some extent such a legal definition is also problematic, as some villages (komai) clearly fulfilled roles as central places and were adorned with temples and agorai, while there is very little evidence for any “urban” amenities in some legally defined cities. An example of this is Korba, a kome or village of some 3.5 hectares in the territory of the polis of Kyaneai in Lycia.29 Surveys revealed that Korba had a monumental agora and that the cult for Apollo was attested here during Hellenistic and Imperial times. Although Korba was well studied the vast majority of villages, or legally defined towns and cities for that matter, have not been studied with modern archaeological methods, or studied at all for that matter. Furthermore, research has focused primarily on settlements with monuments or historical importance, creating an obvious bias in the knowledge of cities in Asia Minor towards legally defined or monumentalized towns. Researchers are particularly partial to the legally defined cities since they are not only known through historical or epigraphic testimonies but also through local issues of coins.30 The village or small town, although suspected to have played a major role in the logic of settlement, is for the most part not well known in Asia Minor.31 In order to obtain a complete picture of the urban  Epstein, “Introduction,” 2–3; Ligt, Peasants, Citizens and Soldiers, 199.  Levick, Roman Colonies in Southern Asia Minor, 95; Broughton, Roman Asia Minor, 734, 740; CIL III, 326; Plin., Epist. 10.47–48. 28  Dmitriev, City Government. 29  Geppert, “Die Siedlung Korba,” 72–73; Kolb, Burg—Polis—Bischofssitz, 298. 30  Jones, A Numismatic Riddle. 31  Marek Geschichte Kleinasiens, 555; the same could be said for villa-estates of the Roman period, Broughton, Roman Asia Minor, 648–676; Mitchell, Anatolia, 149. 26 27

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

259

network during Roman times, therefore, we focus primarily on self-­ governing towns with an autonomous legal status. As already mentioned, archaeological research on many towns with or without a legal status is problematic. Some poleis, such as Alia in Phrygia, have not even been located.32 Others, such as Midaion in Phrygia or Soatra in Lycaonia, have only been visited and described superficially. The vast majority, even in the case of large cities as Kyzikos, have been studied only for architectural and epigraphic surface remains.33 Other cities, such as Trapezous, Smyrna, Attaleia and many others, are built over and consequently are known only through isolated standing remains, spolia or spot-­ excavations. Despite these problems, however, it must also be noted that many good archaeological research projects are being carried out in the region, which greatly help us to understand urbanisation in the region. Yet for the moment the data are still disparate. The majority of the better studies examples of cities are located in the west, south-west and south-east. There are a few projects in the central region of Anatolia and relatively new projects are currently being carried out in the Black Sea region. The relatively simple question of how many towns and cities dotted the map of Roman Asia Minor reflects the problem posed by the different definitions for cities and the disparate data we have for them. Although Josephus and Philostratus both mention 500 cities for Asia (this could refer to the province, not necessarily Asia Minor), modern literature provides varying answers.34 Broughton, using both historical, epigraphic and numismatic evidence, lists in total 370 official cities in Roman Asia Minor and 8 probable ones.35 Jones lists 282 official cities for Asia as mentioned by Pliny, although he notes that not all communities listed by Pliny may have had urban centres.36 Mitchell on the other hand asserts that the total figure for central Anatolia was 130 cities (Bithynia 13, Pontus 11, Paphlagonia 6, Galatia and Lycaonia 20, Phrygia about 45, Mysia 11 and

 Drew-Bear, “Problèmes de la géographie historique en Phrygie,” 951.  Humann described the site of Midaion in brief terms as a hill “200 Schritt lang, 80 Schritt breit und an 15  m hoch” (Humann and Puchstein, Reisen in Kleinasien und Nordsyrien, 23); the city is known only from its coinage (von Aulock, Münzen und Städte Phrygiens, 33); still the most recent study of the surface remains of the big city of Kyzikos is Hasluck, Cyzicus. 34  Josephus, Bellum Judaicum II, 366; Philostratus, Vitae Sophistarum 548. 35  Broughton, Roman Asia Minor, 697–698, 734–735. 36  Jones, Cities of the Eastern Roman Provinces, 76. 32 33

260 

R. WILLET

Lydia 20), although his exact definition for “city” is unclear.37 Marek follows Pliny as well, in that in early Imperial times he notes 282 cities, which grows to over 600 in late antiquity.38 Hanson listed 180 major urban settlements as the cities for Roman Asia Minor, which he derived from the Pleiades database and the Barrington Atlas, but includes three cities less than 10 hectares.39 The Roman Provincial Coinage database includes many more cities that were issuing coins than the 130 or 176 cities proposed by Mitchell and Hanson (a preliminary count stands at 328 cities). Using this body of evidence and earlier research by Arnold Jones, Louis Robert, Thomas Broughton, Christian Habicht and many individual publications on individual cities, some 446 settlements seem to have been officially recognized communities and many of them probably cities. This number relies on secondary sources, which are, however, widely accepted and although a case could be made to review all the epigraphic material this lies beyond the scope of this research. These will serve as the basis for further analysis on complexity and scaling properties.

3   Complexity and the Urban Pattern The spatial distribution of official cities is uneven (see Appendix, Fig. 8.1 and Table 8.1). Broadly speaking, a dense cluster of cities can be observed starting on the Aegean coast and snaking inward along the rivers Maeander and Kaikos. This cluster tapers east to connect with a second cluster of cities, which runs north-south from Pisidia to western Pamphylia and Lycia (Fig. 8.2). Interestingly, the distribution of cities in this second cluster is denser than in the west. To some extent, this clustering can be explained by the landscape and climate, with broad and fertile river valleys branching in the west of Asia Minor and providing important connections with the Aegean coast, while less densely settled Caria was relatively hillier than the Phrygian highland. The second cluster of cities is demarcated in the north and east by the drier areas of Galatia and Lycaonia, while the outliers of the Taurus Mountains limit this cluster in the east of Pamphylia.  Mitchell, Anatolia, 243.  Marek Geschichte Kleinasiens, 515; Plin., HN 5.150; it must be stressed that Pliny notes 282 cities for the province of Asia, the surface of which is much smaller than the total area of Asia Minor. 39  Hanson, “The Urban System of Roman Asia Minor,” 245, 254–255; 237; Hanson, An Urban Geography, 299–335 (Asia), 354–58 (Bithynia et Pontus), 371–92 (Cappadocia et Galatia and Cilicia, excluding cities on Cyprus), 662–84 (Lycia et Pamphylia). 37 38

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

261

Table 8.1  Number of located official cities and communities (n = 446 and 13 possible) using the provincial division of 117 CE and the geographic regional division Province (200 CE)

Area (in 1000 km2)

No. of official cities and communities (% of total)

Density in official cities and communities per 1000 km2

Asia Pisidia, Lycia et Pamphylia Cilicia Galatia et Cappadocia Bithynia et Pontus Outside provinces within research area

130,388 38,619

198 (46.8%) 86 (20.3%)

1.51 2.23

37,738 255,036

48 (11.4%) 52 (12.3%)

1.27 0.20

64,921

19 (4.5%) 25 (5.8%)

0.29

Note that 31 of the 459 in total are left out since these could not be located. The percentages therefore are based on all the located official cities or communities

Yet Pisidia, Lycia and Pamphylia are quite distinct geographic regions. Lycia and Pisidia have a rather fragmented landscape with many small valleys, rivers and hills/mountains. Pamphylia, however, is dominated by a large and fertile coastal plain, which was settled with several sizable cities, such as Attaleia, Perge, Aspendos and Side. In central Asia Minor there are few cities, a pattern which is observed for later periods as well. This area is broadly speaking characterized by a dry continental climate, which made the region prone to harvest failures until the nineteenth century.40 The region adjacent to the Black Sea is dominated by the Pontic Mountains, which run parallel to the coast. This region is much wetter and colder than the interior and is forested, which is reflected in the role of the coastal cities in the trade of wood in antiquity.41 From a geographical perspective it is noteworthy that in the south-­ west, along the river valleys and along the coast, there is a virtually continuous pattern of cities lying within fifteen km of each other, or roughly a three hour walking distance (Fig. 8.3). Yet for the Black Sea coastal area and on the central plateau the distances between cities are much greater.  Mitchell, Anatolia, 145.  Meiggs, Trees and Timber.

40 41

262 

R. WILLET

Bekker-Nielsen argued that these distances are indicative for the role of cities as central places for their hinterland. Many studies consider a travel time of two hours to be the maximum that farmers would travel each day from their place of residence to their fields. A city or settlement with a hinterland of eight to ten km radius, therefore, could theoretically be the central place for all agricultural enterprises in its hinterland. Such centres are typically considered “agro-towns”.42 Officially self-governing cities, therefore, fulfilled the role of primary central places in areas with the densest distribution of cities, while further inland and to the north and east secondary agglomerations (subject villages, small towns and possibly estate centres) must have served as central places as well. The topography and environment are obvious factors of influence on the spatial urban pattern, which already shows some complexity in the clustering. Yet geographic determinism does not entirely explain the existence of these cities. This becomes obvious when we take a diachronic view. Broughton listed the cities in four periods (late republican, Augustan-­ Julio-­ Claudian, Flavian-Severan and third century CE), in which an increase is recorded in the number of cities from 265 to 368 (see Figs. 8.4 and 8.5).43 The database of the Pleiades website (a mapping project developed from the Barrington Atlas) shows a similar increase in the number of located settlements from the Hellenistic period to the Imperial period, from approximately 320 located settlements in the Archaic-Classical period until 300 BCE to 597  in the Hellenistic period to 1116  in the Imperial period.44 These settlements are not only cities or towns (although these are included) but include also other types of settlements (villages, sanctuaries, small towns etc.) known from ancient itineraries, geographical descriptions and other texts. Oftentimes these are barely archaeologically attested. But still the same pattern is observed as for the cities, namely an increase in the number of settlements from the Hellenistic to Imperial period. Broughton already observed this trend: “The development of the cities from [the Hellenistic age after a period of Persian domination] affords examples of three processes working concurrently: a natural

42  Bekker-Nielsen, The Geography of Power, 9–10, 20–29; Bekker-Nielsen actually uses a 4 km per hour travelling speed; Blok, “South Italian Agro-Towns,” 121–135. 43  Broughton, Roman Asia Minor, 697–706, 734–739. 44  Pleiades. A Community-Built Gazetteer and Graph of Ancient Places, http://pleiades. stoa.org.

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

263

Hellenization, amalgamation, either forced or natural, of small communities, and colonization”.45 Although Hellenization and Romanization are outdated concepts in academia, the idea of different factors acting and interacting and thereby causing the observed pattern of urbanisation is compatible with complexity theory and a panarchy model. The factors quoted here from Broughton, but also found in Jones’ and Magie’s works, are basically the influence of acculturation (for towns this means adopting the institutions of the polis, resulting in legally defined recognizable towns or cities), and the conglomeration of settlements and the foundation of new towns and cities, presumably with migrant settlers. The settling of veterans is attested for several colonies in the Hellenistic and early Imperial period.46 These factors in themselves are not permanent and depend on societal and political contexts. The policies of colonization for example cease in the Imperial period, even though the honorary title of colonia was still sometimes granted for the privileges it supposedly still brought, as for example to Nicomedia during Diocletian’s reign.47 The synoikismos of cities is observed particularly during Classical/Hellenistic times, such as for Alexandreia Troas which was founded in 306 BCE (under the name Antigoneia) from the forceful synoikismos of at least six cities.48 Another example is Keraia in Pisidia, which merged with nearby Kremna in Augustan times when the latter became a colonia.49 The relocation of cities as well, such as in the case of Heraclea ad Latmum during Hellenistic times, led to changes in the urban pattern. The adoption of urban institutions can be explained as a top-down process, whereby Hellenistic rulers and Roman emperors (re) founded cities to increase control and tax-flow or to promote urban culture in regions in their domain.50 Examples of this are the cities of Pontus and Bithynia, organized by Pompey after the Third Mithridatic War in 63 BCE to control the newly conquered lands of Mithridates, or the four cities founded by Hadrian in Mysia (Stratonicea-Hadrianopolis,

 Broughton, Roman Asia Minor, 697.  See Levick, Roman Colonies in Southern Asia Minor, 2–5. 47  CIL III, 326. 48  Schwertheim, “Alexandreia”. 49  von Aulock Münzen und Städte Pisidiens, 34; Robert, Villes d’Asie Mineure, 63–64. 50  Jones, The Greek City, 59–60. 45 46

264 

R. WILLET

Hadrianoutherae, Hadrianoi and Hadrianeia).51 But the benefits of becoming a legally recognized town or city for the urban elites, such as controlling taxes raised from the city’s territory, must also be noted and may help to explain why the total number of official cities increased. The increase of the number of cities reflects an increase in complexity, as the city is inhabited by people not primarily involved in the production and supply of food or other subsistence resources. Cities, therefore, signify an increased division of labour and, more importantly, increased social hierarchy and inequality, whereby the urban elite is sustained by the labour of the rural peasantry. In this sense, society is pushed to greater reliance upon agricultural surplus production and therefore greater complexity. The presence of estates in Phrygia, Bithynia, Pisidia and Galatia in Imperial times is indicative of agricultural exploitation for the economic benefits of a small elite.52 To some extent the increase in the number of cities depended on structural factors determined by landscape and climate but a few critical notes must be made. First, the numerical increase is a simplification of a more differentiated process; some cities are added, but others cease to be, are moved or merged. For example Strabo informs us that the league of cities in Lycia numbered twenty three cities in his time, while Pliny the Elder notes that this league had actually dwindled from seventy to thirty-six cities.53 Although the overall picture displays growth, the local level shows more specificity. Lastly, it must be remembered that the archaeological evidence for many cities is sparse and that the historical record often does not provide founding dates but only allows us to conclude when a settlement officially became a city. Many official cities where probably preceded by an already existing settlement. In addition, a myriad of other settlements existed, which have not (yet) been discovered archaeologically. The increase in complexity, therefore, must be tested through other archaeological proxies, and for this, the presence of public buildings within cities provides a good beginning.

51  Bekker-Nielsen, “Roman Urbanism” also notes the importance of a new road network in conjunction with the Pompeian foundations in Pontus; Boatwright, Hadrian and the Cities, 172. 52  Mitchell, Anatolia, 152. 53  Strabo 14.3; Plin., HN 5.101.

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

265

4   Complexity Within the Cities: Spread of Public Buildings Apart from being aesthetically pleasing, public buildings in ancient cities reflect the activities that took place there. Buildings like theatres, gymnasia or bath-houses and the activities associated with them fulfilled multiple functions (entertainment, sports, education and leisure).54 It remains a point of discussion whether elaborate stone public buildings signify cities with greater complexity as temporary buildings or buildings made from wood that could equally well provide the scenery for public spectacles.55 Yet the construction of public buildings in stone constituted a significant effort in terms of finances, time and labour on the part of the city itself, of private individuals or of the central government.56 Although many types of buildings, such as stadia, amphitheatres, aqueducts, temples and monumental agorai or market buildings formed part of the urban layout for many cities in Asia Minor, we will limit ourselves here to theatres and bath-houses to study the diachronic distribution of public buildings, since these two are best recorded for Imperial Asia Minor. Theatres had their origins in the sixth century BCE in the Greek poleis but theatre buildings were constructed in Asia Minor well into the Imperial period.57 They were often restored, elaborated or expanded, as recorded by inscriptions. Although originally intended for plays, recitals and musical events, in Roman times the theatres became the stage for gladiatorial games, animal hunting and fighting as well, as evidenced by the removal of the front rows of seats and the addition of marble screens to create a protected arena. Amphitheatres specifically designed for these Roman events are rarely found in Asia Minor—only Kyzikos, Pergamon, Anazarbos, possibly Ikonion, Aphrodisias and Ephesos had such buildings, while the latter two were technically speaking adjusted stadia.58 The theatre, however, is a widespread building type during the Imperial period:  Zanker, Die römische Stadt, 101.  Robert, Les gladiateurs, 35; Mathé suggests that for the same reason not all cities had permanent hippodromes or stadiums, since other buildings or spaces could temporarily be used. Even under the empire, when their number increases, these buildings were probably used for only a few days per year (Mathé, “Coût et financement,” 189). 56  Zuiderhoek, The Politics of Munificence; Moretti, “Le coût et le financement”; Mathé, “Coût et financement”. 57  Isler, “Theatre”. 58  Stadia were used for gladiatorial combat as well; Robert, Les gladiateurs, 35. 54 55

266 

R. WILLET

135 towns have a single theatre and 8, such as Pergamon or Laodikeia on the Lykos, host several ones (Fig. 8.6). Many of these theatres are undated or poorly studied. The volumes by Daria de Bernardi Ferrero still provide an important standard work with more recent work by Frank Sear providing additions.59 Studies on individual theatres are taken into consideration as well. The dataset we thus compiled allows us to plot the diachronic distribution of theatres (Figs.  8.7 and 8.8). This shows clearly that the highly urbanized regions were already dotted with theatres in the Hellenistic period (n = 60). By the third century CE their number had not only risen with more cities hosting theatres (n = 95), but several cities had a second theatre (n = 6) or had expanded the cavea (n = 9) to increase the size of the building. Many theatres underwent restorations or conversions by adding a large Roman-style stage building.60 This process could be interpreted purely culturally as a reflection of the adoption of the theatre and associated activities, but the fact remains that most cities did not build theatres and that the increase in theatre size or the addition of more theatres must have constituted a significant effort. Although Roman public bath-houses partly descended from a Greek forerunner (the balaneion), this never became widespread during the Hellenistic period. Bath-houses only became widespread in Imperial times.61 Gymnasia, however, as centres for physical education and paideia were already an established phenomenon in the Hellenistic period and were widespread also in Asia Minor. During the Imperial period gymnasia continued to function and a hybrid building type developed, known as the bath-gymnasium, in which a large palaestra was added to the bath-­building or thermae. Inge Nielsen in Thermae et Balnea regards this building type as particular to Asia Minor and separate from thermae types found elsewhere. She connects the introduction and growth of bath-houses particularly in the Roman province of Asia with the relative prosperity of this province and with its “Romanization”.62 This latter reason is at odds with the fact that public bathing culture in Asia Minor, although perhaps popularized by the Romans, was (architecturally at least) absorbed and reconfigured into the already existing cultural framework. However, the introduction of public bathing in cities demanded efforts to construct these sometimes enormous buildings, such as the six baths and  Bernardi Ferrero, Teatri Classici; Sear, Roman Theatres.  Dodge, “Amusing the Masses,” 247. 61  Nielsen, Thermae et Balnea, 6. 62  Nielsen, Thermae et Balnea, 98–101. 59 60

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

267

bath-­gymnasia at Ephesos, totalling some 41,500 m2 dedicated to bathing in the second century CE.63 Furthermore, these baths would have needed water and fuel for heating, requiring a substantial infrastructure for the water supply, consisting of aqueducts, cisterns or in some rarer cases natural springs, and an organization to bring fuel from the countryside to the city.64 In other words, Roman bathing required greater societal and presumably economic complexity. The distribution of bath-complexes shows a propensity for cities closer to the coast. Overall the densest spread is in the western to south-western parts of Asia Minor, with many cities holding one or more bath-buildings (n = 84; Fig. 8.9). The prevalence of Lycia in this map betrays the incorporation of Andrew Farrington’s study on the baths of Lycia, but since data have been gathered assiduously for the other cities as well, we can still conclude that a relatively high proportion of the cities in Lycia did have bath-buildings. Still, the distribution of baths over Asia Minor largely, although not completely, overlaps with the densest areas of the urban network. The diachronic distribution from the first (n  =  30) to the third centuries (n = 68) shows a dramatic increase in the number of cities with baths, during the timespan of two centuries, mostly in the second century CE (Figs. 8.10 and 8.11). Furthermore, twenty-four cities had multiple bathing complexes by the third century; Ephesos leads with six, followed by Pergamon, Patara and Anemourion, which each have four. Inter-city rivalry must have boosted the monumentalization of cities, particularly where urban hubs were situated in close proximity.65 The data both from theatres and from baths show an increase in number and a distribution pattern that is densest in areas with the densest urban network. This indicates that the pattern of increased urbanisation reflects increased complexity also in terms of urban amenities. Increased urbanisation, however, may have had a limited impact in certain areas, such as Phrygia, where the majority of cities were small, more or less resembling villages or agro-towns.66 Here, however, we need to add a caveat. Archaeological research of ancient cities is still ongoing. Most attention in the past has been on cities in the west, south-west and on the southern coast. The density of public buildings in these areas, therefore, may simply reflect the state of archaeological research with potentially  Groh, “Neue Forschungen,” 104.  Brödner, Römische Thermen, 145–162. 65  Tak, “Longing for Local Identity”. 66  Thonemann, “Phrygia,” 35. 63 64

268 

R. WILLET

more buildings likely to be discovered in central Anatolian and northern cities in the future. However, since we included epigraphic studies, which have been conducted for a long time for most cities in Asia Minor, as much as possible in compiling our dataset, it is more likely that the distribution patterns we detected (see Figs. 8.6, 8.7, 8.8, 8.9, 8.10 and 8.11) reflect historical reality. The date of some buildings, furthermore, is not always obvious, especially when they have not been excavated or fully studied. Some are attested only through sparse remains. This problem, however, is offset with a history of in-depth architectural studies of the well-preserved buildings. For the baths, less well-preserved examples do not always allow clearly determining whether they were private or public, even though the large size of some of them suggests a public function. Many theatres were already constructed in the Hellenistic and Classical periods. This means that if they are a sign of greater societal complexity, the process was already underway with urbanisation before the arrival of the Romans. This also adds to the fact that the lack of construction in some cities during the Roman period does not necessarily reflect economic malaise but could simply reflect the fact that a set of public buildings already existed. A good example of this is the city of Phellos in Lycia, which was situated on top of a mountain. In the small space this site offered, already by Classical and Hellenistic times, a theatre, agora, substantial defensive works and temples had been built (not to mention large monumental tombs), leaving little room for additional Roman building activity.67 Nevertheless, since numerous public building projects clearly date to the Imperial period, as evidenced by the theatres and baths, the heyday of the polis lies in the Roman period in terms of monumentality.

5   The Urban Hierarchy Next, we must examine how this increased complexity manifested itself in the towns and cities, and whether this is accompanied by scaling phenomena. We begin by discussing the urban hierarchy, by which we mean the hierarchy in the size of the cities. The relationship between city size and their place in an urban network has received much attention. This is based on the assumption that the proportionality found between the largest and smaller cities is a result of the underlying socio-economic and geographical context, whereby particularly the “systemness” of an urban network, that  Zimmerman” Eine Stadt und ihr kulturelles Erbe”.

67

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

269

is, the level in which a set of cities in a region forms a system of interdependent cities or a loose collection of autarkic settlements.68 This “systemness” is often related to the economic interdependence or integration of a set of cities and economic development. Urban hierarchies are studied by plotting the sizes of cities against their relative rank in the size hierarchy. The proportional relationship of this hierarchy is then used as a meaningful indicator for the functioning of the urban system. A particular form of urban hierarchy, referred to as the “rank-size rule” or “Zipf’s law”, shows a regular proportion in a rank-size analysis, following a power law. This means that when rank and population size are plotted on logarithmic x- and y-axis they form a straight line. In its simplest form, Zipf’s law is written as x(k) = xm/k, with k as rank, x as the size and xm as maximum size in a set of sizes.69 This law has its origin in the study of linguistics, where George Kingsley Zipf empirically found the law to fit word frequencies for words in a set of natural language utterances. This propelled the study of quantitative linguistics in search of underlying universal principles of language but it also attracted criticism early on. Miller and Chomsky suggested that Zipf’s rule was simply useless for capturing the complexity of language and Mandelbrot even demonstrated that randomized words could generate frequency curves obeying Zipf’s law.70 It is still a debated topic in linguistics but the testing of rank-size relationships (with size being defined by population levels) and urban hierarchies on the cities of Germany from 1870 to 1939 by Zipf himself has made it a household tool for geographers from the 1950s onwards,71 although recently, geographer Peter J. Taylor warned against putting too much emphasis on this “hierarchy fetish”, instead of on the network that cities form.72 Already in 1969, Norman Pounds used the urban hierarchy of towns and cities in Roman Britain and Gaul to identify regional capitals and the chief commercial, administrative and military centres in urban networks, using urban areas as a proxy for population.73 De Vries went one step further by plotting the city population size for early modern cities of Europe  Taylor, World City Network, 16–17.  Cristelli et al. There is More than a Power Law in Zipf”. 70  Miller and Chomsky, “Finitary Models of Language Users,” 463–464; Mandelbrot, “Information Theory and Psycholinguistics”. 71  Zipf, National Unity and Disunity; Miller and Chomsky, “Finitary Models of Language Users,” 463–464; Mandelbrot, “Information Theory and Psycholinguistics”. 72  Ferrer i Cancho, “On the Universality of Zipf’s Law”; Taylor, World City Network, 25. 73  Pounds (1969), p. 154–154. 68 69

270 

R. WILLET

and testing these for Zipf’s law or possible deviations from this curve.74 A deviation commonly encountered is convexity, in which the largest cities are not large enough to fit the Zipf’s law. It has been suggested that this is the result of low system integration, whereby the lack of mobility or economic interconnectedness does not allow the largest cities to grow to their full potential. In other words, infrastructure and economic integration are limiting factors which potentially influence the growth and hierarchy of cities.75 The convexity, however, may also be caused by either an incomplete sample (partitioning), whereby the set of city sizes does not contain the largest city or cities of a dendritic urban network, or the set containing (parts of) multiple autonomous urban networks (pooling).76 As for linguistics, the discussion on Zipf’s law and rank-size analysis in urban studies is not at an end and multiple interpretations of the same data are possible. Taylor stated that Zipf’s law in an urban system is the reflection of the condition of entropy in the system, whereby the straight line gradient on a rank-size plot “is reached when there are a myriad of forces acting on the cities in a country, so many in fact that they can be assumed to have random effects on city growth, which will thereby become proportionate to city size”.77 Paul Krugman discusses Zipf’s rule for the cities of the modern US and suggests random growth of the cities can explain the power law found in the sizes. Yet he goes further by arguing that the unevenness of the natural landscape is also capable of producing this randomness.78 Here one must think of differences in possibilities to move, the presence of natural sources, variation in rural population, religious  De Vries (1984), p. 85–120.  See Marzano (2011), who argues (p.  220) that the distribution of towns in Roman Lusitania followed Zipf’s rule more closely than in other regions in the Iberian Peninsula, which she interprets as greater economic integration caused by the greater mobility and economic integration in that region due to access to navigable rivers. Hanson concludes that convexity found in the rank-size relationship of cities in Roman Asia Minor is the result of partitioning, rather than primacy of a single city (Hanson 2011, pp. 264–265). Note that many of Hanson’s size measurements are at odds with published sizes, with the largest city of Sardis being off by 118 hectares in the best case scenario. Furthermore, the usage of population reconstructions based on sliding population densities of 100 to 1000 people per hectare is at odds with population densities for planned (110  p/ha) and unplanned towns (40–60 p/ha) reconstructed from field survey data and excavations by Simon Price in the same volume (Price 2011, p. 23). 76  Johnson (1980), p. 244–245. 77  Taylor, World City Network, 16. 78  Krugman, “Confronting the Mystery of Urban Hierarchy,” 412, 416. 74 75

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

271

elements in the landscape and so on. In essence, this means that the initial conditions are important for the trajectory of a system.79 This is indeed an attractive interpretation when we reflect on the increasing complexity of cities. Chaos and complexity theory as well assumes that initial conditions influence the development and cycling of a system. For now, however, we will first use Zipf’s rule as a null hypothesis. A total of 168 settlement sizes have been collected and measured for Asia Minor in the Roman Imperial period (Fig.  8.12). Here we assume attested sizes to represent conditions at the height of the Imperial period (approximately the second to third centuries CE). In most cases such accuracy could not be achieved but by and large the collected sizes do represent the Imperial period. The sizes are derived from archaeological publications on individual or multiple towns and the published figures are doublechecked by making measurements in GIS by using plans and satellite imagery. The (often Hellenistic) fortification walls tend to be clearly visible from the air, either directly or indirectly through vegetation. Other large monuments, such as theatres, often clearly form markers as well. Ideally, a well-studied archaeological site yields a published size for the built-up area during the Imperial period, but this is often not the case. Fortunately, an approximation can usually be made for the built-up area by using the (dated) necropoleis as city limits. Classical and Hellenistic fortifications tend to encompass a much larger area than the actual built-­up area, with a higher percentage of the walled area built up for smaller poleis.80 Even here, however, exceptions can be found. Pergamon, for instance expanded beyond its 90 hectare Hellenistic enclosure to a possible maximum size in imperial times of 190 hectares.81 Obviously, for some sites measurements give better results than others (Fig. 8.13). Perge and Kolossai, for example, are clearly visible, while for Midaion in Phrygia only a small hill is visible (see Note 33). Although limited by the state of research, using published size estimates together with own measurements in GIS the set of data provides the best approach to the urban hierarchy. Although not all the measured sites (twenty-two settlements) were cities in the legal sense they are included since they often fulfilled a role as a central place, such as Korba near Kyaneai in Lycia, Notion in Asia or Korasion in Rough Cilicia.  Personal comment from J. Bintliff.  Hansen, The Shotgun Method, 42. 81  Wulf, “Der Stadtplan von Pergamon”; Pirson, “Pergamon”, 120; Pirson expresses doubts on the reconstruction by Ulrike Wulf (230 ha); the figure is derived from Pirson’s discussion of Pergamon during the RAC session of 2016 in Rome. 79 80

272 

R. WILLET

The rank-size plot for Asia Minor (Figs. 8.14 and 8.15) shows a clear convex pattern deviating from a Zipf curve (grey dashed line), which is extrapolated from the largest city in the hierarchy. Here too the analysis is using settlement area as a proxy for population and strictly looking at the Zipf rule. This convex plot is the result of a high number of large- to medium-sized cities: fifty cities measure over forty hectares (30 per cent), forty measure twenty to forty hectares (24 per cent).82 There are seventy-­ eight (46 per cent) small cities under twenty hectares, nearly as many as the medium-sized and large cities. The distribution would presumably tilt more towards the smaller cities if all the other official cities were taken into consideration, since cities for which little archaeological evidence is available are more likely to be smaller. Archaeological projects of the past have focused on sites with an interesting history (Ilion or Troy is a case in point) or with impressive still standing remains. Less spectacular sites, such as Blaundos in Phrygia, Pompeiopolis and Komana Pontika in Pontus and Seleukia Sidera to name just a few, are only recently receiving the archaeological attention they deserve and are providing valuable insights into the lives of smaller cities.83 We can extrapolate small sizes for the remaining approximately 300 legally defined cities in order to replot the curve. If we consider these as belonging to the smallest category, we can model a size for these 300 cities proportionally to the frequency of measured sizes under 20 hectares. For instance, since there are 2 measured cities on 78 of 19 hectares (2.6 per cent) this implies that 8 of the 300 unknown cities had a modelled size of 19 hectares. The resulting graph (Fig. 8.16) essentially creates a secondary convex curve at the lower end. It can be argued that the smallest sizes (especially when generated by modelling) are less reliable: a measuring error of one hectare for a city of forty hectares constitutes a 2.5 per cent error while for a city of four hectares the error margin is 25 per cent. But even if all the sizes under ten hectares are left out both the curves of all measured settlements and the curve including the modelled sizes show clear convexity. In both these graphs, the larger cities show a gradual differentiation; the largest cities are too small, while the middle- to small-­sized cities are too large to fit a simple Zipf’s curve. This can be the result of a lower systemic integration, although it is also possible that the pooling of the different geographic areas of Asia Minor mixed separate urban systems.  Division in size taken from Ligt, Peasants, Citizens and Soldiers.  Filges, “Stadtentwicklung im Gebiet des oberen Mäander”; Summerer, Pompeiopolis; Erciyas, “Komana Pontike”; personal comment from B. Hürmüzlu. 82 83

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

273

A plot for the Roman province of Asia (Fig. 8.17) shows a slightly less convex curve, but even here Zipf’s law does not apply. Although the best fitting power law with an R2 = 0.8474 fits better than the power law curves applied to other data-sets. It can be argued that the province of Asia had an urban network that was more developed and integrated, as reflected by its dense urbanisation. The province had a long history of urbanisation, it was the first in 133 BCE to become a permanent part of the Roman Empire and its large river valleys facilitated connectivity from the coast inland (see above). On the legal level, the integration of this area is well attested via the assize districts or conventus, for which both Pliny the Elder and epigraphic evidence provide extensive evidence.84 These districts grouped cities with a single city serving as a court centre. The provincial governor would tour through these centres over a year to administer justice. Unfortunately, not enough sizes are available to compare the hierarchy of a single district to the overall province. But Dio Chrysostomus (c. 40–c. 115 CE) noted in his Orations for the assize centre of Apamea that the court sessions attracted a lot of people, boosting the prosperity of the city.85 The other cluster of cities in the province of Lycia, Pamphylia and Pisidia (Fig. 8.18) results in a more convex plot. This could be explained by the fractured nature of the landscape, providing we accept the assumption that these distributions are indeed reflecting systemic integration. The problematic economic integration, combined with the founding of Roman colonies in Pisidia, has recently been suggested to explain why Sagalassos began producing its substantial regional tableware.86 At the same time, we do know that on a political level, some integration existed in this region as reflected by the Lycian League (see above), which existed as a Roman protectorate since 168 BCE, but continued as an independently operating group of cities until 43 CE.87 The distribution for the province of Cilicia (Fig. 8.19) has few measured cities and again forms a convex curve. Even when zooming in to a regional level by taking the cities connected by the Maiandros River valley, starting in the west with Miletos to the east as far as Kolossai and including Aphrodisias and 84  Habicht, “New Evidence on the Province of Asia”; Dalla Rosa, “Praktische Lösungen für praktische Probleme”. 85  Dio Chrysostomus, Orationes 34. 15–17. 86  Willet and Poblome, “The Scale of Sagalassos Red Slip Ware Production”. 87  Behrwald, “Lykischer Bund”.

274 

R. WILLET

Alabanda close to the valley, again a convex curve appears (Fig. 8.20). This puts some pressure on the economic or systemic integration hypothesis, as these are all situated on the flatlands of the river valley and one may suppose that these were well connected via land roads, with the greatest distance from Miletos to Kolossai being about 180  km as the crow flies. Basically these examples show that the convexity in the distribution of all measured settlements in Asia Minor is reflected in the individual provincial hierarchies, with Asia having a slightly less convex curve. This means that the pooling of different urban systems is not a likely cause of the observed shape of the curve. The shape of the urban hierarchy in the rank-size analyses could be a reflection of the level of economic integration or systemic integration, but at this point there is no way to check this. Proxies for economic integration could be based on an analysis of proxies for trade, for instance small finds such as ceramics or coins, or on an exhaustive study of property in Asia Minor, but the research and state of publications do not allow for such an analysis yet.88 More importantly, however, and directly related to the question of increased complexity as observed through increased urbanisation, this static image of the urban hierarchy should ideally be elaborated diachronically. Both in the work of Zipf and de Vries the developments over time show interesting patterns whereby changes in the urban hierarchy have been related to economic developments. De Vries’ work also shows that the results can be highly variable, which he partially attributes to weaknesses in defining the elements of a system and its boundaries, such as a province, a nation or geographical unit. He concludes that outside factors impinge on an urban system as well and cannot be disregarded to fully explain the urban hierarchy.89 Unfortunately, as the research stands, evidence for changes in urban size can be provided for only twelve cities. Pergamon has already been mentioned, but Herakleia Pontika, Ephesos, Laodikeia on the Lykos, Perge, Sardis, Ilion, Selinous and Aspendos also show an increase in urban size during the Hellenistic and Imperial period, with Tarsos and Smyrna ostensibly growing as well. For others, such as Phaselis, a decline in size sets in during the late Roman period. Yet for the vast majority, a diachronic sequence of settlement sizes cannot be established. 88  Willet, Early Imperial Tableware, provides a perspective for the inter-city trade of tableware in Roman Asia Minor, which shows a strong regional focus. 89  De Vries, European Urbanisation, 120.

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

275

6   Towards Greater Complexity: A Discussion An increase in complexity is observed through the growth in the number of cities and the accompanying spread and expansion of public buildings. The shape of the urban hierarchies reflects this development as well. Rank-­ size analysis begins to shine a light on what is happening “under the bonnet” of increasing complexity of urban patterns, both for Asia Minor as a whole and at the provincial level. Economic or systemic integration, or the lack thereof, can be suggested to explain the shape of the curves, but cannot be advanced as anything more than a hypothesis at this time. It may be worthwhile to approach this curve from another angle, already hinted at by Krugman. If the random factors of the geography and history are indeed responsible for creating an urban hierarchy that conforms to Zipf’s law, then the convex curves that are found in Asia Minor may still be the result of randomness (it is all a coincidence). Yet the maps of both of the spread of cities and public buildings show the direction of distribution starting for the west and south-western areas inland. With most of the larger cities situated in the west and many smaller ones in close proximity, clearly this is a different urban system than the Central Plateau or Black Sea Coastlands. Here, it is of interest that in all the rank-size graphs, the largest cities were situated on a plateau, which suggests they fulfilled similar regional roles. The convexity found in all curves might be indicative of the path-dependent development of these urban patterns, as suggested by the spread of cities and public buildings. This makes it likely that the convex shape is not the product of pure randomness, but it may be related to the stages of the process of urbanisation, particularly since the province of Asia has a slightly less convex curve. A complementary model from ecological colonization theory, as discussed by Serge Frontier, may shed some light on this. In a study on the ecology of plankton, Frontier noted that the ecological niches occupied by a community (plants to herbivores to carnivores, all dependent on each other’s existence) are seemingly random and, when described in a rank-­ frequency graph, follow a straight log-log curve, as a Mandelbrot curve and similar to Zipf’s law.90 Yet in some instances the number and frequency of species in an ecosystem do not follow the Mandelbrot curve. This suggests that there are cases where a self-ordering process takes place in the ecosystem, namely the self-ordering of information (species diversity) and cost. Frontier recognized three stages of ecosystem maturation:  Frontier, “Utilisation des diagrammes rang-frequence,” 39.

90

276 

R. WILLET

1. System with fewer species in high frequency, where few species are dominant in the system in terms of frequency. The curve then bends slightly right with a number of low frequency species having less difference in frequency. 2. Diversity increases to a maximum with more species in less divergent frequencies. In this stage, information management is optimized as a result of the system spending part of its energy on maintaining biomass and part of this self-ordering effect or information management. 3. Ageing of the system, diversity decreases and the curve is straightened again. The system is on the edge of chaos or to speak in terms of the panarchy model: the system has become more brittle in terms of its resilience, energy is conserved to a maximum in ordering and exogenous shocks might be able to release this energy again. When applied to the discussion on cities and complexity this model suggests that the curves of rank-sizes may be related to the maturation of the system. In the first stage, we could envision the dominance of a few very large cities and many tiny settlements, a situation perhaps best reflected in Iron Age Anatolia, although this lies beyond the scope of this paper. The convex curves encountered in our study would reflect the second stage of maturation: a period when the system has most diversity, that is many cities fill in the landscape most efficiently as large urban niches are broken up and a more effective cover of settlements is established. This is taking place in south-western Anatolia and along the coasts while cities are moving more inland. The third stage of maturation is an emergent trophic hierarchy which is sustained by a complex resource feedback, which is less efficient as stage two but creates maximum vertical structures and is marked by the Zipf rule type of curve. Particularly the fact that the province of Asia best fits a power law suggests that western Imperial Anatolia started to transform towards this stage, while outside the region, new primate-cities were (re-)emerging, such as Rome, Antioch on the Orontes and so on. There was still great diversity in urbanism in Anatolia and the local/regional roles of cities must not be underestimated but the effect of Roman administration and economic globalization and commercialisation was starting to pull the system towards a greater interdependence of towns. Hierarchy in an urban system, therefore, is an emergent property (with self-organizing elements) of the system’s maturing but depends at the

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

277

same time on the specific initial conditions and history of the cities in the system. The official grants of self-government to cities and the variation in their status probably created trophic effects. It is noteworthy that particularly cities that were or are suspected to be centres for the assizes rank among the largest cities of each province. The observed increase in complexity and regional variation, I suggest, are a reflection of the maturation of the settlement pattern. The importance of other proxies, however, must be emphasized as they can shed more light on the administrative and economic nature of the interdependence between cities. Urban patterns need to be studied diachronically over a longer period, in order to help us disentangle the structural from the eventual. Lastly, it is permissible to speculate a bit on the processes that caused the growth in the number of autonomous cities, public buildings, size of cities and, to an extent, variation in size. Civic autonomy is connected with the power over taxation it provided to urban elites. Indeed, the public buildings researched here are strongly tied to the style of living of urban elites as well. It is difficult to escape the obvious conclusion that all these developments in urbanism are related to the concentration of wealth at these cities. Van Bavel recently argued that the emergence of factor markets (that distribute land, labour and capital) eventually leads to elites that benefit from these markets and tend to bend institutions to take more advantage of these, thereby increasing inequality in the distribution of wealth.91 The research on Anatolian rural settlements during Hellenistic and Imperial times is limited; it is difficult to say whether these developments happened in tandem with growth in the countryside, or that they happened at the cost of rural life, or both. Yet at the same time, we must be cautious with too optimistically interpreting the emergence of these monumentalized urban hubs, despite the romantic beauty of their ruins.

 Bavel, The Invisible Hand, 260–265.

91

278 

R. WILLET

Appendix: Figures

Fig. 8.1  Map displaying the towns and cities with official self-government (n = 446 and 13 possible; note that only 428 are located and plotted on the map)

Fig. 8.2  Heat map of the self-governing cities and communities in the late second to early third centuries CE; the radius used for this heat map was 15 km

Fig. 8.3  The official self-governing cities with a 15  km radius buffer drawn around them

Fig. 8.4  Number of cities as listed by Broughton, Roman Asia Minor in Hellenistic/late republican Anatolia

Fig. 8.5  Number of cities as listed by Broughton, Roman Asia Minor in Flavian-­ Severan times and the third century CE in Anatolia

Fig. 8.6  Map of all theatres located in Asia Minor (n  =  143  +  11 possible theatres)

Fig. 8.7  Map of the datable theatres up to c. 30 BCE (n = 59 + 1 place with multiple theatres)

Fig. 8.8  Map of the datable theatres up to c. 300 CE (n = 95 + 6 places with multiple theatres + 9 places with expanded seating areas)

Fig. 8.9  Map of all baths in Asia Minor (n = 84 + 2 possible)

Fig. 8.10  Map of all the datable baths until c. 100 CE (n = 23 plus 7 places with multiple bath-houses)

Fig. 8.11  Map of all the datable baths until c. 300 CE (n = 44 plus 24 places with multiple bath-houses)

284 

R. WILLET

Fig. 8.12  Map displaying the sizes of all measurable cities and settlements in Roman Asia Minor (n = 224)

Fig. 8.13  Kolossai from the air; note the semi-circular recess of the theatre on the east of the central hill. The city probably extended north towards the river, where the necropolis was located

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

285

Fig. 8.14  Rank-size plot for the settlements of Asia Minor (n = 168) using area in hectares

Fig. 8.15  Rank-size plot on logarithmic scales for the settlements of Asia Minor (n = 168) using area in hectare. The dashed grey line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fit

286 

R. WILLET

Fig. 8.16  Rank-size plot on logarithmic scales for the settlements of Asia Minor (n = 168) together with modelled sizes of the unmeasured official cities (n = 300), both using area in hectare. The dashed grey line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fitting trend line following a power law

Fig. 8.17  Rank-size plot on logarithmic scales for the self-governing settlements of the Roman province of Asia (n = 55) using area in hectare. The grey line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fitting trend line following a power law

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

287

Fig. 8.18  Rank-size plot on logarithmic scales for the self-governing settlements of the Roman province of Lycia et Pamphylia, Pisidia (n = 46), using area in hectare. The grey dashed line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fitting trend line following a power  law. This province incorporated large parts of the region of Pisidia in the second century CE

Fig. 8.19  Rank-size plot on logarithmic scales for the self-governing settlements of the Roman province of Cilicia (n = 18) using area in hectare. The grey dashed line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fitting trend line following a power law

288 

R. WILLET

Fig. 8.20  Rank-size plot on logarithmic scales for the self-governing settlements of Maiandros River valley (n  =  11) using area in hectare. The grey dashed line represents the curve the set of sizes would have if it matched Zipf’s law. The black dotted line represents the best fitting trend line following a power law

Bibliography Bavel, Bas J.P. van. The Invisible Hand? How Market Economies have Emerged and Declined since AD 500. Oxford: Oxford University Press, 2016. Behrwald, Ralf (2006), “Lykischer Bund.” In Der Neue Pauly, edited by Hubert Cancik and Helmut Schneider, Consulted online on 12 February 2018. https://doi.org/10.1163/1574-9347_dnp_e713500. Beinhocker, Eric D. The Origin of Wealth: Evolution, Complexity, and the Radical Remaking of Economics. Boston, Mass: Harvard Business School Press, 2006. Bekker-Nielsen, Tønnes. The Geography of Power: Studies in the Urbanization of Roman North-West Europe. Oxford: B.A.R., 1989. Bekker-Nielsen, Tønnes. “Roman Urbanism in the Pontic Frontier Zone”, In The Economics of Urbanism in the Roman East, edited by Rinse Willet. Heidelberg, forthcoming. Bentley, R.  Alexander, and Herbert D.  G. Maschner, eds. Complex Systems and Archaeology. Salt Lake City, Utah: University of Utah Press, 2003. Bettencourt, Luis A. “The Origins of Scaling in Cities” Science 340, (2013): 1438–1441. Bintliff, John L. “Archaeology at the Interface: an Historical Perspective”, In Archaeology at the Interface: Studies in Archaeology’s Relationship with History,

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

289

Geography, Biology and Physical Science, edited by J. L. Bintliff and C.F. Gaffney, 4–31. Oxford: B.A.R., 1986 Bintliff, John L. “Catastrophe, Chaos and Complexity: the Death, Decay and Rebirth of Towns from Antiquity to Today”, Journal of European Archaeology 5 (1997): 67–90 Bintliff, John L. “The Origins and Nature of the Greek City-State and its Significance for World Settlement History”, In Les Princes de la Protohistoire et l’emergence de l’Etat., Actes de la table ronde internationale organisée par le Centre Jean Bérard et l’École Française de Rome, edited by P.  Ruby, 43–56. Naples and Rome: Ecole Française de Rome, 1999. Bintliff, John L. “Time, Structure, and Agency: The Annales, Emergent Complexity, and Archaeology”, In: A Companion to Archaeology, edited by J.L. Bintliff, 174–194. Malden and Oxford: Wiley-Blackwell, 2004. Blok, Anton. “South Italian Agro-Towns”, Comparative Studies in Society and History 11 (1969): 121–135. Boatwright, Mary T. Hadrian and the Cities of the Roman Empire. Princeton: Princeton University Press, 2000. Brödner, Erika. Römische Thermen und antikes Badewesen. Darmstadt: Primus, 2011. Broughton, Thomas Robert Shannon. “Roman Asia Minor’” In An Economic Survey of Ancient Rome. Vol. IV, edited by Tenney Frank, 499–916. Baltimore: Johns Hopkins Press, 1938. Ferrer i Cancho, Ramon. “On the Universality of Zipf’s Law for Word Frequencies”, In Exact Methods in the Study of Language and Text, edited by P. Grzybek and R. Köhler, 131–140. Berlin: De Gruyter, 2007. Clark, Peter. European Cities and Towns 400–2000. Oxford; New  York: Oxford University Press, 1995. Cristelli, Matthieu, Michael Batty, and Luciano Pietronero. “There Is More than a Power Law in Zipf”. Scientific Reports 2, no. 812 (8 November 2012) Consulted online on 12 February 2018. https://doi.org/10.1038/srep00812. Dalla Rosa, A. “Praktische Lösungen für praktische Probleme: Die Gruppierung von Conventus in de Provinz Asia un die Bewegungen des Prokonsuls C. Iulius Severus (Procos. 152/53)”, Zeitschrift für Papyrologie und Epigraphik 183 (2012); 259–276. De Bernardi Ferrero, D. Teatri Classici in Asia Minore. 4 Voll., Roma: L’Erma di Bretschneider, 1966–1974. De Vries, Jan. European Urbanization, 1500–1800, London: Routledge, 1984. Dermody, B. J., R. P. H. van Beek, E. Meeks, K. Klein Goldewijk, W. Scheidel, Y. van der Velde, M.F.P. Bierkens, M. J. Wassen and S.C. Dekker. “A Virtual Water Network of the Roman World”, Hydrology and Earth System Sciences 18 (2014): 5025–5040.

290 

R. WILLET

Dmitriev, Sviatoslav, City Government in Hellenistic and Roman Asia Minor, Oxford: Oxford University Press, 2005. H. Dodge, Hazel. “Amusing the Masses: Buildings for Entertainment and Leisure in the Roman World”, In Life, Death, and Entertainment in the Roman Empire, edited by David S.  Potter and David J.  Mattingly, 205–255. Michigan: Michigan University Press, 1999. Donev, Damjan and Rinse Willet, “Book Review of An Urban Geography of the Roman World, 100 BC to AD 300, by J.W.  Hanson”, American Journal of Archaeology 121.4 (2017). https://doi.org/10.3764/ajaonline1214.Donev. Drew-Bear, Thomas. “Problèmes de la géographie historique en Phrygie: l’exemple d’Alia”, In Aufstieg Und Niedergang Der Römischen Welt, II.  Prinzipat 7.2, edited by H.  Temporini and W.  Haase, 931–952. Berlin and New  York: De Gruyter, 1980. Eastwood, W. J., N. Roberts, and H. F. Lamb. “Palaeoecological and Archaeological Evidence for Human Occupance in Southwest Turkey: The Beyşehir Occupation Phase”. Anatolian Studies 48 (1998): 69–86. Epstein, Stephen R. “Introduction. Town and Country in Europe, 1300–1800”, In Town and Country in Europe, 1300–1800, edited by S.R.  Epstein, 2–3. Cambridge: Cambridge University Press, 2001. Erciyas, Burcu. “Komana Pontike: A City or a Sanctuary?”, In Mithradates VI and the Pontic Kingdom, edited by J.M. Højte, 289–312. Aarhus: Aarhus University Press, 2009. Faroqhi, Suraiya. Towns and Townsmen of Ottoman Anatolia. Trade, Crafts and Food Production in an Urban Setting, 1520–1650. Cambridge: Cambridge University Press, 1984. Filges A. “Stadtentwicklung im Gebiet des oberen Mäander. Die Lydisch-­ Phrygische Grenzregion am Beispiel von Blaundos”, In Stadt und Stadtentwicklung in Kleinasien, edited by E. Schwertheim, E. Winter, 35–50. Bonn: Habelt, 2003. Finley, Moses I. The Ancient Economy. Updated edition with a foreword by Ian Morris, Berkeley: University of California Press, 1999. Frontier, S. “Utilisation des diagrammes rang-frequence dans l’analyse des ecosystems”, Journal de Recherche Océanographique 1 (1976): 35–48. Garnsey, Peter. Cities, Peasants and Food in Classical Antiquity: Essays in Social and Economic History. Edited by Walter Scheidel. Cambridge: Cambridge University Press, 2004. Geppert, Karin. “Die Siedlung Korba im Norden des Yavu-Berglandes”, In Lykische Studien 7: Die Chora von Kyaneai, edited by F.  Kolb, 5–73. Bonn: Habelt, 2006. Groh, Stefan. “Neue Forschungen zur Stadtplanung in Ephesos”, Jahreshefte des Österreichischen Archäologischen Institutes in Wien 75 (2006): 47–116.

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

291

Habicht, Christian, “New Evidence on the Province of Asia”, The Journal of Roman Studies 65 (1975): 64–91. Haldon, John. The Empire that would not die. The Paradox of the Eastern Roman Survival, 640–740. Cambridge (MA) and London: Harvard University Press, 2016. Hansen, Mogens Herman. The Shotgun Method: The Demography of the Ancient Greek City-State Culture. Columbia: University of Missouri Press, 2006. Hanson, John W. ‘The Urban System of Roman Asia Minor and Wider Urban Connectivity’. In Settlement, Urbanization, and Population, edited by Alan K.  Bowman and Andrew I.  Wilson, 229–75. Oxford: Oxford University Press, 2011. Hanson, John W. An Urban Geography of the Roman World, 100 BC to AD 300. Oxford: Archaeopress, 2016. Hasluck, Frederick W. Cyzicus. Being Some Account of the History and Antiquities of that City, and of the District Adjacent to it, with the Towns of Apollonia Ad Rhyndacum, Miletupolis, Hadrianutherae, Priapus, Zeleia, etc., Cambridge: Cambridge University Press. Hoepfner, Wolfram. Herakleia Pontike—Ereğli. Eine Baugeschichtliche Untersuchung. Vienna: Böhlaus, 1966. Holling, C.S. “Understanding the Complexity of Economic, Ecological, and Social Systems”, Ecosystems 4 (2001): 390–405. Holling, C.S, Lance H. Gunderson and Donald Ludwig, “Chapter 1: In Quest of a Theory of Adaptive Change”, In Panarchy: Understanding Transformations in Human and Natural Systems, edited by L.H. Gunderson and C.S. Holling, 3–22. Washington, DC: Island Press, 2002. Homer-Dixon, Thomas, The Upside of Down. Catastrophe, Creativity and the Renewal of Civilisation, Toronto: Random House of Canada, 2006. Humann, Karl and Otto Puchstein, Reisen in Kleinasien und Nordsyrien ausgeführt im Auftrage der Kgl. Pruessischen Akademie der Wissenschaften. Band I Textband, Berlin, 1890. Isler, H.-P. “Theatre. II Architecture”, In Der Neue Pauly, edited by Hubert Cancik, Helmut Schneider, Consulted online on 12 February 2018. https:// doi.org/10.1163/1574-9347_bnp_e1206910. Johnson, Gregory A., “Rank-Size Convexity and System Integration: A View from Archaeology” Economic Geography 56 (1980): 234–247. Jones, Arnold H.  M. The Greek City from Alexander to Justinian. Oxford: Clarendon, 1940. Jones, Arnold H.M. Cities of the Eastern Roman Provinces. 2nd ed., Oxford: Clarendon, 1971. Jones, Tom. “A Numismatic Riddle: The So-Called Greek Imperials”, Proceedings of the American Philosophical Society 107 (1963): 308–347.

292 

R. WILLET

Kolb, Frank. Burg—Polis—Bischofssitz. Geschichte der Siedlungskammer von Kyaneai in der Südwesttürkei, Mainz am Rhein: Philipp von Zabern, 2008 Krugman, Paul. “Confronting the Mystery of Urban Hierarchy”, Journal of the Japanese and International Economies 10 (1996): 399–418. Levick, Barbara. Roman Colonies in Southern Asia Minor, Oxford: Clarendon, 1967. Ligt, Luuk de. Peasants, Citizens and Soldiers. Studies in the Demographic History of Roman Italy 225 BC–AD 100, Cambridge: Cambridge University Press, 2012. Bettencourt, Luis A., José Lobo and Deborah Strumsky. “Invention in the City: Increasing Returns to Patenting as a Scaling Function of Metropolitan Size”, Research Policy 36 (2007): 107–120. Mandelbrot, Benoit. “Information Theory and Psycholinguistics: a Theory of Word Frequencies”, In Readings in Mathematical Social Sciences, edited by Paul F.  Lazarsfield and Neil W.  Henry, 151–168. Cambridge, Mass: MIT Press, 1966. Marek, Christian. Geschichte Kleinasiens in der Antike, München: C.H. Beck, 2010. Marzano, Annalisa. “Rank-Size Analysis and the Roman Cities of the Iberian Peninsula and Britain: Some Considerations”, In Settlement, Urbanization, and Population, edited by Alan K.  Bowman and Andrew Wilson, 196–228. Oxford: Oxford University Press, 2011. Mathé, Virginie. “Coût et financement des stades et des hippodromes”, In (ed.), L’argent dans les concours du monde Grec, edited by Brigitte Le Guen, 189–223. Saint-Denis: Presses Universitaires de Vincennes, 2010. Meiggs, Russel. Trees and Timber in the Ancient Mediterranean World, Oxford: Oxford University Press, 1982. Miller George A. and Noam Chomsky. “Finitary Models of Language Users”, In Handbook of Mathematical Psychology. Volume II, edited by R. Duncan Luce, Robert R.  Bush and Eugene Galanter, 419–491. New  York and London: Wiley, 1963. Mitchell Stephen. Anatolia: Land, Men, and Gods in Asia Minor. Vol. I: The Celts in Anatolia and the Impact of Roman Rule, Oxford: Oxford University Press, 1993. Moretti, Jean-Charles. “Le coût et le financement des théâtres grecs, Coût et financement des stades et des hippodromes”, In L’argent dans les concours du monde Grec, edited by Brigitte Le Guen, 147–187, Saint-Denis, 2010. Nielsen, Inge. Thermae et Balnea. The Architecture and Cultural History of Roman Public Baths. Vol. I Text, Aarhus: Aarhus University Press, 1993. Ortman, Scott G., Andrew H.F.  Cabaniss, Jennie O.  Sturm and Luis M.A. Bettencourt. “Settlement Scaling and Increasing Returns in an Ancient Society”, Science Advances 1 (2015), e1400066–e1400066. Peschlow-Bindokat, Anneliese. “Herakleia”, In Stadtgrabungen und Stadtforschung Im Westlichen Kleinasien—Geplantes und Erreichtes, edited by Wolfgan Radt, 101–112. Istanbul: Ege Yayinlari, 2006.

8  COMPLEXITY AND URBAN HIERARCHY OF ANCIENT URBANISM… 

293

Pirson, Felix. “Pergamon—Bericht über die Arbeiten in der Kampagne 2007”, Archäologische Anzeiger (2008): 83–155. Pirson, Felix and Martin Zimmerman. “Das Umland von Pergamon— Wirtschafliche Ressourcen, ländliche Siedlungen und politische Repräsentation”, In Pergamon. Panorama Der Antiken Metropole. Begleitbuch Zur Ausstellung, edited by R.  Grüssinger, V.  Kästner and A.  Scholl, 58–65. Berlin: Michael Imhof, 2012. Poblome, Jeroen, ed. Exempli Gratia. Sagalassos, Marc Waelkens and Interdisciplinary Archaeology, Leuven: Leuven University Press, 2013. Poblome, Jeroen. “The Economy of the Roman World as a Complex Adaptive System. Testing the Case in Second to Fifth Century CE Sagalassos”, In Structure and Performance in the Roman Economy. Models, Methods and Case Studies, edited by Paul Erdkamp and Koenraad Verboven, 97–140. Brussels: Latomus, 2015. Pounds, Norman J.G. “The Urbanization of the Classical World”, Annals of the Association of American Geographers 59 (1969): 135–157. Price, Simon. “Estimating Ancient Greek Populations. The Evidence of Field Survey”, In Settlement, Urbanization, and Population, edited by Alan K.  Bowman and Andrew Wilson, 17–35. Oxford: Oxford University Press, 2011. Radt, Wolfgang. Pergamon: Geschichte und Bauten einer antiken Metropole, Darmstadt: Philipp von Zabern, 1999. Robert, Louis. Les gladiateurs dans l’Orient Grec, Paris: Champion, 1940. Robert, Louis. Villes d’Asie Mineure. Études de Géographie anciennes, 2ème ed., 63–64. Paris: de Boccard, 1962. Scheidel, Walter. “Demography”, In The Cambridge Economic History of the Greco-­ Roman World, edited by Walter Scheidel, Ian Morris and Richard P.  Saller, 38–86. Cambridge: Cambridge University Press, 2007. Schwertheim, E. “Alexandreia. Stadt in der Troas”, In Der Neue Pauly, edited by Hubert Cancik and Helmut Schneider. 2006. Consulted online on 12 February 2018. https://doi.org/10.1163/1574-9347_dnp_e113940. Sear, Frank, Roman Theatres. An Architectural Study, Oxford: Oxford University Press, 2006. Stanley, B.W., T.J. Dennehy, M.E. Smith, B.L. Stark, A.M. York, G.L. Cowgill, J.  Novic and J.  Ek. “Service Access in Premodern Cities An Exploratory Comparison of Spatial Equity”, Journal of Urban History 41 (2015): 1–24. Summerer, Lâtife. Pompeiopolis I: Eine Zwischenbilanz aus der Metropole Paphlagoniens nach fünf Kampagnen (2006–2010), Langenweißbach: Beier and Beran, 2011. Tak, Herman. “Longing for Local Identity: Intervillage Relations in an Italian Mountain Area”, Anthropological Quarterly 63 (1990): 90–100.

294 

R. WILLET

Taylor, Peter J. World City Network. A Global Urban Analysis, London and New York: Routledge, 2004. Thonemann, Peter. “Phrygia: an Anarchist History, 950 BC–AD 100”, In Roman Phrygia. Culture and Society, Greek Culture in the Roman World, edited by Peter Thonemann, 1–40. Cambridge: Cambridge University Press, 2013. von Aulock, Hans. Münzen und Städte Pisidiens. Teil II, Tübingen: Ernst Wasmuth, 1979. von Aulock, Hans. Münzen und Städte Phrygiens. Teil II, Tübingen: Ernst Wasmuth, 1987. Willet, Rinse and Jeroen Poblome, “The Scale of Sagalassos Red Slip Ware Production—Reconstructions of Local Need And Production Output of Roman Imperial Tableware”, Adalya 18 (2015): 133–157. Willet, Rinse. “Early Imperial Tableware in Roman Asia Minor: a perspective on the diachronic patterns and morphological developments”, Internet Archaeology 50 (2018). https://doi.org/10.11141/ia.50.17. Willet, Rinse. The Geography of Urbanism in Roman Asia Minor. Sheffield: Equinox Publishing, 2020. Wulf, Ulrike. “Der Stadtplan von Pergamon. Zu Entwicklung und Stadtstruktur von der Neugründung unter Philetairos bis in spätantike Zeit”, Istanbuler Mitteilungen 44 (1994): 135–175. Zanker, Paul. Die römische Stadt: eine kurze Geschichte. München: Beck, 2014. Zimmerman, Martin. “Eine Stadt und ihr kulturelles Erbe. Vorbericht über Feldforschungen im zentrallykischen Phellos 2002–2004”, Istanbuler Mitteilungen 55 (2005): 215–270. Zipf, George K. (1941), National Unity and Disunity. the Nation as a Bio-Social Organism. Bloomington, Ind. Principia Press, 1941. Zuiderhoek, Arjan. The Politics of Munificence in the Roman Empire. Cambridge: Cambridge University Press, 2009.

PART III

Epidemics

CHAPTER 9

Disease Proxies and the Diagnosis of the Late Antonine Economy Colin P. Elliott

No reliable estimate for the mortality of the Antonine plague exists, nor will it ever. Instead, historians are left to assemble scattered bits of literary evidence, archaeological proxies, economic truisms and a handful of Egyptian wages and prices together in order to lend a sense of empirical plausibility to what are essentially speculative arguments. This description is not meant to demean the exercise—we do what we can with what we have; nevertheless, the entire operation hinges upon speculation—a fact which does not change even when the analysis is quantitative or even bolstered by the computational power of computer programs. The discipline of ancient economic history has been trending towards quantification for decades now—bringing many positive developments. Scholars must take care, however, to ensure that the quest for better numbers never supercedes the ever more important task of model-building through critical dialogue between argument and evidence.1 The greatest benefit of quantitative modelling—especially involving the sorts of impressively 1

 Howgego, “Some Numismatic Approaches to Quantifying the Roman Economy,” 289.

C. P. Elliott (*) Indiana University, Bloomington, IN, USA e-mail: [email protected] © The Author(s) 2021 K. Verboven (ed.), Complexity Economics, Palgrave Studies in Ancient Economies, https://doi.org/10.1007/978-3-030-47898-8_9

297

298 

C. P. ELLIOTT

complex programs being crafted by many of my co-contributors to this volume—is the inherently qualitative exercise of constructing the models themselves and justifying the relationships, connections and causalities which are built into them. Even for contemporary epidemiologists, modelling disease impacts is fraught with uncertainty and guesswork. Ancient plagues and their effects are even more difficult to understand. It should not surprise us, therefore, that the scholarship has produced a wide range of estimates for mortality during the Antonine plague years (however long the plague lasted—as we do not know for certain!). Only two studies, to my knowledge, have put forward firm numerical estimates for mortality which fall below 10 per cent. On the basis of their diagnosis of the plague as smallpox, Robert and Maxwell Littman suggested that mortality could not have been more than –7 to 10 per cent.2 Their conclusions are not based upon much empirical evidence; their estimate is extrapolated from mortality patterns of smallpox more generally. The Littman’s estimate was meant to be a check upon the minimalism of James Frank Gilliam, who suggested a total mortality of no more than 2 per cent.3 There have since been many detractors and few have rushed to Gilliam’s defence.4 Scholars in the 1960s and 1970s believed plague mortality to be in the range of 1 and 10 per cent, whereas the past two decades have seen a sustained barrage of mortality estimates above 20 per cent, with some recent figures ballooning past 30 per cent.5 Scholars who have gone against these high-mortality estimates have avoided offering any estimates for mortality and have, instead, stuck to mostly qualitative arguments.6 Geoffrey Kron, for example, argues on the basis of a thorough comparative study of nutrition and hygiene in the Roman Empire and eighteenth- and nineteenth-century Western Europe that any smallpox outbreak in the former would have been “far from catastrophic”.7 Across several articles and book chapters, Christer Bruun has used a positivistic approach to show that there is plenty of evidence indicating a late second-century CE crisis (or crises), but there is no reason

 Littman and Littman, “Galen and the Antonine Plague,” 252–255.  Gilliam, “The Plague under Marcus Aurelius,” 250. 4  Exceptions being Salmon, Population et Dépopulation, 1974; Bruun, “The Antonine Plague and the ‘Third-Century Crisis’,” 209. 5  Paine and Storey, “The Alps as a Barrier to Epidemic Disease”. 6  Greenberg, “Plagued by Doubt”; Kron, “Nutrition, Hygiene and Mortality”. 7  Kron, “Nutrition, Hygiene and Mortality,” 239. 2 3

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

299

to believe that the Antonine plague was the sole, or perhaps even the chief, instigator.8 While the qualitative nature of the objections to high-mortality plague may be partly to blame for their lack of traction, it is equally true that a high-mortality Antonine plague makes for a convenient deus ex machina to explain the healthy amount of evidence for various demographic and economic problems in the mid- to late second century CE.9 I suspect a great deal of the preference for such high estimates is due to several sources from Roman Egypt (and also, perhaps, what has been said about these sources). In the past two decades, several scholars have staked much upon this evidence, while, simultaneously, the most damning aspect of Gilliam’s “minimalist” view is the fact that he virtually ignored Egypt altogether.10 More recent scholarship suggests that Egypt actually offers a great deal of indirect evidence for plague, but also, that this evidence may be sufficient for quantitative study.11 Therefore, it should not surprise us that William Harris recently announced the post-mortem for Gilliam’s view and other minimalist interpretations: “no minimalist position such as that defended by Gilliam merits further consideration as far as Egypt is concerned”.12 We also have Dominic Rathbone who, while not taking aim at Gilliam directly, is as unequivocal as Harris: “No one disputes that the Antonine Plague, which was carried into Egypt in 166/7 CE, caused over the next decade a dramatic aggregate population loss, probably of around 20–30 per cent to judge from some attested cases, including over twenty Delta villages”.13 Rathbone points to the unquestionable evidence that Egypt was in crisis 8  Bruun “The Antonine Plague in Rome and Ostia”; Bruun, “The Antonine Plague and the ‘Third-Century Crisis’”; Bruun “La Mancanza Di Prove”. 9  I wish I were the first to apply this calque to the Antonine plague. See Cleary, The Roman West, 466. 10   Duncan-Jones, “The Impact of the Antonine Plague”; Scheidel “A Model of Demographic and Economic Change”; Scheidel, “Roman Wellbeing”; Rathbone, “Villages, Land and Population”. 11  So convincing, in fact, that James Greenberg’s critique was mostly concerned with the interpretation of the data rather than questioning whether the datasets were actually valid in the first place. See Greenberg, “Plagued by Doubt”. 12  Harris, “The Great Pestilence,” 334. 13  Rathbone, “Roman Egypt,” 700, emphasis added. Rathbone offered a more cautious view in earlier work: “The Antonine plague thus does seem to have affected at least Lower and Middle Egypt from about 166 CE into the late 170s, and to have been characterized by sporadic, brief but devastating outbreaks in individual localities”; see Rathbone, “Villages, Land and Population,” 119.

300 

C. P. ELLIOTT

around the time of the Antonine plague. Only very recently did scholars begin asking whether the evidence for crisis might be reinterpreted as related to climate and/or agricultural productivity rather than exclusively concerned with plague.14 My answer to this question (yes, absolutely) was provided in a recent Past & Present article wherein I argue that demographic evidence usually interpreted as plague related is more likely the result of agricultural, political and climatological causes.15 At the time I did not attempt to untangle the wage and price data which is regularly used to support high estimates of plague mortality—a situation I will now  rectify. First, however, we must come to grips with the theoretical context for how scholars normally interpret these data.

1   Price Proxies and Plague Either explicitly or (much more often) implicitly behind every discussion of the plague’s impact upon wages, prices, growth, contraction, mortality and fertility, among other things (an elephant in the room, really), is the concept of equilibrium. With the Antonine plague, economic historians typically work with either (or both) of two approaches to equilibrium: neo-classical and neo-Malthusian.16 The neo-classical concept of “general equilibrium” dictates that prices in a perfectly competitive market tend to shift with the forces of supply and demand. General equilibrium, in short, predicts that markets will clear and offers an explanation in the abstract for why this occurs. In practical terms, economists and economic historians are more likely to use “partial equilibrium”—which assumes, for heuristic purposes, that phenomena or conditions other than those under study remain constant (ceteris paribus). When economist and historian Peter Temin first began to focus on the Roman world a decade and a half ago, it was still controversial to say, as he did, that the Roman Empire featured “an economy where most resources are allocated by prices that are free to move in response to changes in underlying conditions. … markets in the early Roman Empire typically

14   Wilson, “The Mediterranean Environment,” 265. See also the work of Blouin, Triangular Landscapes. 15  Elliott, “Climate Change, Plague and Local Violence”. 16  The clearest, most conscientious application of both frameworks that I have read is in Jongman, “The New Economic History”.

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

301

were equilibrated by means of prices”.17 If millions of working males died during the Antonine plague, then the neo-classical theory of general equilibrium predicts that (ceteris paribus) the surviving workers, due to a sudden trough in the supply of workers, would have charged much higher prices for their labour. Land values, moreover, should  have fallen—and limited papyrological evidence suggests that this is indeed what occurred.18 There are additional assumptions at work which make the application of the theory slightly more complex: the market must have been a competitive market, the surviving labourers must have  known that they could charge more, additionally, they must also have exhibited profit-­maximizing behaviour in spite of any cultural or institutional disincentives to do so.19 Without getting into a diversionary discussion, let me simply state that anthropological and sociological studies of the ancient world offer serious arguments which suggest that many of these secondary assumptions were partially or perhaps even completely absent in the societies of the Ancient Mediterranean. Nevertheless, whether the theory itself is accurate or applicable, this is typically how neo-classical notions of equilibrium help to explain the impact of the Antonine plague. Neo-Malthusian equilibrium also addresses supply and demand, but only insofar as it relates to the ability of human societies to feed themselves. Thomas Malthus was deeply concerned about the demands that exponential population growth might place upon the food supply, which only grows arithmetically.20 In the Malthusian worldview, the size of the population is fixed by the food supply. When population grows beyond what can be sustained by the food supply, it must either be checked “positively” through mortality from plagues, famines, wars and so on or “preventively” by means of restrictions on the birth rate. Malthusian equilibrium describes a stable population rather than stable prices—a society in which food production is maximized and population levels are stagnant. Practically speaking, this equilibrated society must be one in which income does not rise above the level of subsistence.21 The Antonine plague and the sparse demographic evidence which accompanies it are too seductive 17  Temin, “Market Economy” 170 Expanded upon in Temin, “Price Behaviour,” 204 Contra Bang, “Trade and Empire,” 25 n.59; Haley, Baetica Felix, 11–12. 18  Scheidel, “Roman Wellbeing,” 280–285. 19  Hence, related neo-classical simplifications such as perfect competition, perfect knowledge and profit maximization. 20  Malthus, Thomas. An Essay on the Principle of Population. London, 1798, 4.5.1. 21  Clark, A Farewell to Alms, 23.

302 

C. P. ELLIOTT

for neo-Malthusians to ignore. The Roman Empire appears to have experienced radical population increases over a short period of time, especially in cities which so happened to share characteristics like those Malthus wanted to see implemented in the cities of early nineteenth-century Europe: “we should make the streets narrower, crowd more people into the houses, and court the return of the plague … we should build our villages near stagnant pools, and particularly encourage settlements in all marshy and unwholesome situations”.22 Hence, the Antonine plague was a long-overdue “positive check” which brought population down suddenly and, in line with Malthusian equilibrium, raised incomes in the short term as the per-capita share of the food supply was increased.23 The operation of both forms of equilibrium appears to be present in prices for various goods found in several collections of Egyptian papyri. These prices were first assembled as part of a monumental work by Allan Chester Johnson, later reworked by Hans-Joachim Drexhage and, finally, further culled and intensified in an important study by Rathbone of wheat, wine and donkey prices.24 It is Rathbone’s dataset of wheat prices, especially his graphs, which historians tend to use as the basis for their studies of the Antonine plague. These data, insofar as the term applies (more on this below), seem to confirm the existence of both inflationary and non-inflationary periods in Roman Egypt: notably, stable prices between 45 and 160 CE, price inflation between 161 and the early 190s CE, and then another stable period from the 190s to 270 CE. The prices, however, do not exactly show inflation from 161 to the 190s; rather, after a gap in the sources during that crucial period, it seems that prices from the 190s are roughly twice what they were in the previous period. Rathbone, who is chiefly responsible for noticing and interpreting this important finding, was most interested in disproving the long-held view that currency debasement and price inflation occurred concurrently in the third century CE.25 Subsequent scholars, however, have used his findings to confirm the severity of the Antonine

 Malthus, Thomas. An Essay on the Principle of Population. London, 1798, 4.5.1.  Scheidel, “Roman Real Wages in Context”; Scheidel, “Roman Wellbeing”; Scheidel, “A Model of Demographic and Economic Change”. 24  Johnson, Roman Egypt; Drexhage, Preise, Mieten/Pachten, Kosten und Löhne; Rathbone, “Prices and Price Formation in Roman Egypt”; An updated list of wheat prices is found in Rathbone and Von Reden, “Mediterranean Grain Prices”. 25  Rathbone, “Monetisation Not Price Inflation”. 22 23

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

303

plague.26 Indeed, if the Roman economy operated according to neo-­ classical and/or neo-Malthusian rules, then price inflation is exactly what would be expected after millions of plague deaths. Rathbone is helpfully transparent by including all prices in his graph— state and private, certain and uncertain. Not all of these prices, however, are actually “data”—converting them requires that we apply some higher standards. Most crucially, prices which cannot be quantified precisely dated to a single year or which do not clearly describe wheat must be eliminated entirely, or graphed using some form of stock chart which accounts for ranged in data. Ranged prices or dates absolutely should not be graphed as single points, whether by mean, median or mode, as this would mislead readers. I offer a second modification merely for clarity. The two broad categories of data—private prices and state prices—should be graphed separately, as the former is a truer reflection of supply and demand while the latter is less likely to be such. Lastly, there is the issue of chronological range. I am most interested in looking at plague years and the prices immediately surrounding those years and so I have eliminated the information from before 100 CE. With these criteria applied, we start with the following representation of private prices (Fig. 9.1). My criteria for inclusion as “data” managed to eliminate several of the most ambiguous entries. What was already a fairly sparse dataset has now become even sparser. And, in fact, this is partly why I have dissected the data in the way that I have—to show that the information which can actually be transcribed as “data” does not conclusively indicate sudden price inflation in mid- to late second-century Roman Egypt. Like Rathbone, I avoid adding any trend lines to my graph, and the explanation for why I made this choice illuminates some of the problems with the dataset. R-squared values indicate the degree to which a trend line correlates to data (a value of 1 equals perfect correlation while a value of 0 indicates no correlation whatsoever).27 Trend lines with high R-squared values, say .8 or .9, would merit inclusion because they leave little variance to factors outside of the model. As it stands, many ancient historians (presumably the main readership for this volume) could assume that R-squared simply 26  Synthesis in Temin, The Roman Market Economy, 74–75. Earlier sources: Rathbone, “Prices and Price Formation in Roman Egypt”; Bagnall, Currency and Inflation, arguing for some changes in prices, partly due to plague and partly due to monetary factors, is Harper, “People, Plagues, and Prices”. 27  R-squared is a statistical measure which shows how much variance in a dataset is explained by the variable(s) in the model.

304 

C. P. ELLIOTT

Fig. 9.1  Private wheat prices (exclusive of uncertain dates/prices/commodities) from Roman Egypt, 100–270 CE. (Sources: Rathbone, “Prices and Price Formation in Roman Egypt,” 217–233; Rathbone and Von Reden, “Mediterranean Grain Prices,” Table A8.12. I have only used prices in which both the year and the price was confirmed with absolute certainty. An odd state price of 6 denarii per artaba in 246 CE was ultimately left out because it is unclear whether the nominal parity between denarii and tetradrachmai held at this time—hence, the price is uncertain)

indicates some sort of “average”, but this is far from accurate. In fact, trend lines really should be avoided altogether unless R-squared values are high enough to justify their inclusion. Not long after first presenting this paper at the authors’ workshop at Sagalassos, Turkey, in September of 2015, I became aware of Kyle Harper’s work on plagues and prices which includes discussion of Egyptian wheat prices along similar lines to my own here and, hence, must be factored into this discussion (Fig. 9.2).28 Despite the fact that Harper and I were initially unaware of each other’s work, it is interesting that we came to similar conclusions about the shape of the dataset. The two main differences in our results derive from Harper’s 28  I wish to again express my deepest gratitude for Kyle’s generosity in sending his JEH paper in which he dissects this data, as well as links to the data tables, ahead of publication. This section would be impoverished if not obsolete (!) without his generosity.

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

305

Fig. 9.2  Nominal wheat prices (drachmai/artaba) to 275 CE (Harper, “People, Plagues, and Prices” 816)

inclusion of uncertain dates, as long as the range of possible dates is limited to less than a century, and a wider chronological scale.29 He plots a price of 7 dr. per artaba at 115 CE, for example, in order to handle the uncertainty of the price’s date, which is believed to be somewhere between 100 and 130 CE. Observers will also note several additional prices (12, 18 and 20 dr. per artaba) at 187/8 CE, a year which acts as a mean for the suspected range of these prices (175–200 CE). While it is absolutely correct to factor these prices in our reckoning of the overall trends in prices, it is not in my view statistically accurate to graph prices with such broad date ranges as single points, as though the uncertainty (and complexity this uncertainty creates) was absent.30 It is partly a question of preference: should we err on the side of more data (which may be less accurate but more conducive to observing trends and patterns) or on the side of more precise data (which may be more accurate but be less useful for establishing trends and patterns)? There is a sharper division between Rathbone, Harper and myself concerning what these data may be telling us. I think it is fair to say that Rathbone and Harper are more optimistic while I advocate a degree of pessimism. Rathbone, for example, unquestionably sees the Antonine  Harper, “People, Plagues, and Prices,” 815.  My concerns echo those of Wilson, “Quantifying Roman Economic Performance,” 154.

29 30

306 

C. P. ELLIOTT

plague as the best explanation for the change in prices, although he is careful to acknowledge that “we lack the evidence to explain quite how it happened”.31 Harper is similarly optimistic but is also careful to say that plague is “a plausible causal factor in the escalation of prices”.32 In Fig. 9.2 we see a linear trend line stopping at the last price point in 160 CE, which suggests a slightly rising wheat price before the plague. A second price band begins with the first price point in 187/8 CE and runs until the early 270s.33 The low R-squared value (0.06412) of the first trend line tells us that the line does not strongly correlate to the data. To give a sense of the problem, if I create a linear trend line for the entire dataset used by Rathbone, which would translate to a small annual increase in inflation each year (~1.4 per cent), the R-squared value is 0.3063—still mostly unrepresentative of the data, but less so compared to a line which stops at 160 CE and which implies that there are two distinct price bands. What this suggests is that a (purely hypothetical) small annual inflation rate from 100–270 CE is actually more representative of the data (which, I stress, is limited) than a punctuated surge in inflation between the mid-160s and the early 190s CE. What we can only know for certain is what is shown by the points in all three figures: the final “low” price is dated to 160 CE, followed by a thirty-two year gap until the first “high” prices.34 While the gap in the data by no means eliminates the link with plague, it does not confirm it either. The statistical analysis bears this out. Moving beyond private wheat prices, the second overarching category of prices is state prices—administrative rates set by officials, usually for tax and tribute payments. If the same stringent criteria are used with private prices, a graphic representation looks something like this (Fig. 9.3). Readers will immediately notice, especially after the previous discussion of trend lines, that I have included one here. This is mostly to show what a highly correlated R-squared value (0.9681 in this case) would look like. The line supports two key insights, both of which can be seen just as easily without the polynomial trend line. All data on the graph show that state prices of wheat are roughly unchanged through 216 CE—an astounding degree of stability if there indeed was a devastating demographic shock 31  Rathbone, “Roman Egypt,” 713; Rathbone and Von Reden, “Mediterranean Grain Prices,” 178. 32  Harper, “People, Plagues, and Prices,” 816. 33  Harper, “People, Plagues, and Prices,” 20. 34  Rathbone also dates the second band “from the AD 190s”. See Rathbone and Von Reden 2015: 177.

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

307

Fig. 9.3  State prices of wheat in Roman Egypt, 100–270 CE. (Sources: Rathbone, “Prices and Price Formation,” 217–233; Rathbone and Von Reden, “Mediterranean Grain Prices,” Table A8.12 I have chosen to use a third-order polynomial as, with such a limited amount of data, I wanted to use as few constants as possible)

due to plague. It seems unlikely that state prices changed during the years for which data is unavailable—those in which, according to literary sources, plague was most severe. Econometrics, however, only show us what may have happened statistically—it is a great danger to allow such models to dictate history. Historians can, however, treat the model as a counterfactual which demands explanation: what sorts of historical conditions would explain an essentially static state price through the second and early third centuries CE? Even without my regression line, the discrepancy between trends in state prices and private prices is easy to observe, and Rathbone noticed and commented upon it in his 1997 study and again in a 2015 piece

308 

C. P. ELLIOTT

co-authored with Sitta von Reden.35 He suggested that the stability of state prices was at least partly due to preventive policies and infrastructure which allowed the state to keep grain in reserve during good years and release these stocks in bad years.36 Rathbone also argued, however, that these policies and practices would have had a stabilizing effect on markets. I am not so sure. The wheat crop’s dependency on the Nile’s inundations meant that supply could vary greatly year to year. If the state price was generally held to around 8 dr. per artaba, then this would have caused major problems for supply.37 In years of plenty, the state price would have been artificially high, likely encouraging mass sell-offs of wheat. Market sales would have been avoided in favour of transactions with the state. More problematically, in years of dearth, suppliers would have avoided selling to the state wherever possible—they would have instead taken their wheat to markets where higher prices could be obtained. Tax payers could have  (and almost certainly did) paid an aederatio in cash (i.e. until the early third century when such cash payments were banned).38 Discouraging or even outlawing such activity—thus transforming normal markets into grey or black markets—was unlikely to overcome high discrepancies between market and state prices, especially in years when the harvest was particularly bad. There is evidence for such black-market transactions in the Roman Empire, although not from the Antonine period. Julian’s fourth-century CE invective against the population of Antioch decries “rich men” who “secretly sold the corn in the country for an exaggerated price”, suggesting the ease with which official prices could be subverted.39 The contemporary evidence is more circumstantial. A tax account drafted around 185 CE lists taxes in kind paid by the meris of Herakleides, an administrative division of the Arsinoite nome (Fayyum).40 Both the amount of grain collected and still owed survives; of 814,862 artabas owed, only 223,581 artabas had been collected.41 The shortfall is astounding, and while it may have been caused (at least partly) by sellers avoiding state sales and ­tributes,  Rathbone and Von Reden, “Mediterranean Grain Prices,” 178.  Rathbone, “Prices and Price Formation,” 198. 37  Discussion on artabas as a measurement in Rathbone 1991: 468–469; Mayerson 2000. 38  P. Col. 6.123. 39  Julian. Mis. 368a–370d. See also Amm. Marc. 22.12–14 and other sources mentioned in Wiemer 1995: 269ff. 40  P. Oxy 66.4527. 41  van Minnen 2001. 35 36

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

309

it is equally valid to speculate about the influence of droughts, civil disturbances or the other sorts of disruptive forces I highlighted in the previous section. In many cases, such forces would augment both price and supply disturbances resulting from state intervention. In short, while I agree with Rathbone that grain storage may explain the stability of prices, especially in years without augmenting disturbances, it seems more likely that price controls merely pushed instability onto markets, whether in terms of prices or (more likely) supplies. There is one more complication that arises from using wheat prices as a proxy for economic growth and the impact of the Antonine plague. The role of wheat as the primary foodstuff, as well as the degree to which its price and supply was a regular concern in the Roman world, makes wheat a poor good for measuring price inflation. Price inflation, which I emphasize at the risk of being banal, is a rise in the general level of prices. There is a good reason why modern central banks measure price inflation with a consumption basket: an aggregation of goods which filter out, as much as possible, price fluctuations associated with specific items or even specific industries. The “core” consumer price index in the United States, for example, excludes some food and energy costs. For the Roman world, it is realistic to assume that household expenditures on staple foods would have been much higher compared to modern times—making wheat prices more reflective of overall consumption. At the same time, prices of a single good—especially one so prone to seasonal variation and climate—are more likely to show volatility. While it is useful, therefore, to observe trends in wheat prices, price changes in a single good can only reveal so much about general price levels.42 Rathbone anticipated this problem (not everyone who has subsequently used his data has done the same!) by including prices for wine and donkeys. To summarize his findings: wine prices show a moderate, gradual inflationary trend while donkey prices are more difficult to pin down due to differences in age and sex as well as the fact that donkeys were capital rather than consumer goods.43 Through most of the second century CE (through around 180 CE) donkeys sold for between 50 and 280 dr. while a higher price band (between 300–800 dr.) appears to have run from about 197 CE through sometime in the middle of the third century.44 There is certainly a case here for price  Similar thoughts offered by Scheidel 2014a: 213.  Rathbone, “Prices and Price Formation,” 198–210, 223–239. 44  Rathbone, “Prices and Price Formation,” 210. 42 43

310 

C. P. ELLIOTT

i­nflation over several centuries; there is not, however, solid evidence for a plague-related disruption in the mid- to late second century. The wheat prices, therefore, should be considered in light of Occam’s razor and historians should prefer the most obvious explanations for high wheat prices: either a dearth of supply or an increase in demand. For the situation in mid-second-century CE Egypt, this means that either there was a population boom (in contrast with anecdotal evidence for population decline) or a period of frequent famines. Famines, however, come and go; so how should historians explain the fact that wheat prices never return to mid-second-century levels? This crucial question demands further study to be sure, but I can propose a few possibilities here, each of which touch upon larger methodological questions. Neo-Malthusian reasoning would suggest that the rise in real prices would be temporary as population recovered and equilibrium was restored.45 The fact that prices remain high implies, within the neo-­ Malthusian framework, that the population loss was effectively permanent. Permanent population losses fit into a neo-Malthusian framework as “checks” which restore equilibrium. In other words, the high prices seen from the late second century CE onwards were indicative of the “normal” equilibrium and the low prices witnessed during the first two centuries of Roman administration were indicative of an unsustainable disequilibrium.46 Helpfully, Walter Scheidel’s work on the Antonine plague in Egypt considers land rents.47 Prices for land are cut in half over a nearly two-­ century period from the early 100s to the 260s CE. The neo-Malthusian explanation for this drop is that falling land prices correlate with population declines.48 Although if we take the neo-Malthusian model at face value, the fact that land prices fall continuously would indicate a steady decline in population rather than a positive check and subsequent recovery. If that is the case, then we should expect commodity prices (including wheat) to drop back down, but they do not. The clean simplicity of the neo-Malthusian framework is one of the reasons it is so attractive, but this simplicity also makes it easy to falsify.

 See chapter 1, p. 5.  This would fit remarkably well with the argument in Scheidel, “Roman Wellbeing”. 47  Scheidel, “In Search of Roman Economic Growth”, critiqued in Bagnall, “The Effects of Plague. More recent thoughts in Scheidel, “Roman Wellbeing”. 48  Hansen and Prescott, “Malthus to Solow”. 45 46

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

311

There is also a potential monetary explanation—the same explanation which Rathbone rejects, at least partly. If the value of Roman currency was primarily a function of precious metal content, it would be easy to connect debasements under Marcus Aurelius and Commodus with price inflation, especially if these debasements continued.49 In fact, the silver content in Roman coinage continued to plummet through the third century, making this explanation seem all the more likely. It also helps that other proxies (wine, donkeys) show higher prices—something more suggestive of a general rise in prices. At the same time, if the monetary explanation were true, we would expect to see what the quantity theory of money tells us should happen: prices moving with debasements from decade to decade, if not year to year. Rathbone’s aforementioned work, however, casts doubt of such a tight relationship.50 So just as with the neo-Malthusian explanation, the superficial simplicity of a monatarist explanation belies methodological complications and problems. Another possibility is that prices and wages stayed high because of low market integration whether systematically or, perhaps, specifically relating to Egypt. The “low” prices of grain, about an average of 8 dr. per artaba, coincide with the period when Egypt was the main source of Roman grain. It is interesting that the destination for Egyptian grain seems to have shifted towards the eastern Mediterranean sometime during the third century.51 From this time, it appears that Rome was fed once again from the western breadbasket of North Africa, as was the situation prior to the annexation of Egypt in the late first century BCE.52 In fact, whenever literary accounts of food crises or famines from between the fourth and sixth centuries CE reveal the source of the Roman grain supply, it is always Africa.53 There are hints here that the “market integration” which was present seems to have been largely a function of imperialism (supply via tribute, distribution via political networks)—this possibility requires further study.54 A new quantitative study of ceramic distribution by Paul Reynolds suggests that trade in a variety of goods was already dividing into eastern and western regions 49  See Elliott, “The Acceptance and Value of Roman Silver Coinage; Butcher and Ponting, “The Beginning of the End”; now see Elliott, “Silver Debasement, Climate and Plague”. 50  Rathbone? “Monetisation Not Price Inflation”. 51  Stathakopoulos, Famine and Pestilence, 52. 52  Euseb. Hist. eccl. 8.14.6. Connection established by Stathakopoulos, Famine and Pestilence, 178–179. 53  Summary in Stathakopoulos, Famine and Pestilence, 51–52. 54  Erdkamp, The Grain Market in the Roman Empire; Bang, The Roman Bazaar.

312 

C. P. ELLIOTT

(and smaller sub-regions) as early as the second century CE.55 The point here is that many factors probably influenced the changes in prices in mid-­ second-­century Egypt. The phenomenon need not have been caused by plague or, at least, exclusively caused by plague.

2   Wages, the Military and Plague Mortality As difficult as it is to use Egyptian wheat prices to hypothesize about plague mortality, it is even more difficult to understand the limited evidence for wages in the second century CE.  There is, however, a good theoretical reason to at least attempt to track wages: a shock which suddenly and severely reduced population numbers would have affected supply and demand unequally—the resulting imbalance would have  almost certainly been reflected in demands for higher wages, especially in those professions which required less skill.56 Scheidel’s long-term (third century BCE to eleventh century CE) survey of daily wages for unskilled rural Egyptian labourers essentially duplicates the bands that we see in the Egyptian wheat prices.57 Insofar as his process and a multiplicity of sources can overcome the inherent uncertainty of some of the data, Scheidel has offered an impression of “real” wages—wages measured on a constant standard and which, hence, are adjusted for inflation. What he finds is that real wages appear to have been exactly the same in the periods immediately before and after the Antonine plague: 4.9 litres of wheat per day in both periods. Furthermore, there is evidence of wage adjustments in the face of disease-related demographic changes during the Justinianic plague in the mid-sixth century CE. Daily wheat wages from 570 CE through the early eighth century fall between 7.7 and 13.4 litres. This evidence resonates with laws which were passed under Justinian against servants who demanded double or triple their normal wages.58 The situation became so bad that, according to John of Ephesus, the high costs of labour meant that members of the court at Constantinople could not afford to launder their expensive clothing—a hyperbolic statement, to be sure, but one which probably belies some underlying truth.59  Reynolds, “The Supply Networks”.  Explained most effectively in Temin, “The Contribution of Economics,” 65. 57  Scheidel, “Real Wages in Early Economies,” 453. 58  Nov. Iust. 122. 59  Little, “Life and Afterlife,” 21 Quantitative data appears in Scheidel, “Real Wages in Early Economies”. 55 56

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

313

Scheidel also summarizes price and wage changes in real terms across a variety of commodities and wage types.60 Interestingly, only prices for donkeys (as a whole; not distinguishing for age/gender) and monthly wages show any kind of marked rise; donkey prices rise by around 50 per cent and wages rise roughly 20 per cent. All other prices and wages (wheat, oil, wine, land prices and land rents, among others) either show little to no increase in real terms or, as especially seen in prices and rents for land, there are notable declines. Again, the neo-Mathusian case receives some support. Scheidel rightfully recognized the two most plausible explanations for these prices: “either … any demographic contraction caused by the Antonine plague was much less severe than that precipitated by the Black Death, or … institutional arrangements … prevented Egyptian workers from fully benefitting even from a substantial shift in the land/ labour ratio”.61 His conclusion, informed by a comparison between the collectivistic, short-term-oriented Mamluk landholders who doggedly attempted to preserve rents even after the Black Death reduced the labour on their estates on the one hand, and the seemingly weaker position of second-century CE landholders who lacked the institutional or market power to exploit workers to the same degree as the Mamluks on the other hand, was that we cannot rule out such “institutional arrangements”—a term unfortunately left unspecified and unexplained.62 The first explanation for the problematic wage evidence—that the demographic effects of the Antonine plague were much less severe than most now believe—is left completely alone by Scheidel and dismissed outright by Harris in his interventi on the strength of the “wigwam” argument into which this evidence is normally built.63 A much more thorough case than the one I make here is required to deconstruct this wigwam; moreover, a new, more plausible wigwam will be needed to account for the evidence—a task which will take many hands and much time. In the meantime, it may be possible to draw upon several wage rates for Roman legionaries which are extant in the literary sources in order to think more broadly  about the impact of plague mortality. Tremendous perils accompany the use of these numbers in economic analyses of any 60  Scheidel, “Model of Demographic and Economic Change”. See graphic form in Scheidel, “Roman Wellbeing,” 284. 61  Scheidel, “Roman Wellbeing,” 285. 62  Scheidel, “Roman Wellbeing,” 286. 63  Harris, “The Great Pestilence” 336.

314 

C. P. ELLIOTT

kind, however. I will discuss these difficulties in turn, but I should start with the most crucial issue: the army almost certainly was affected by plague much differently compared to the general population. After all, the hypothesis offered by several ancient sources (and which is repeated almost verbatim in some modern historical narratives) is that Lucius Verus’ army was the principal vector by which the plague was brought from the East.64 Jerome’s image of a Roman army “slaughtered almost to extinction” may be hyperbolic, but his point is corroborated by the fourth-century historian Eutropius, who says that almost all of the soldiers fell sick.65 Furthermore, a bronze inscription memorializes a group of Mithraists, probably soldiers, who died of an unspecified mortalitatis causa in Virunum in 184 CE.66 The direct evidence for legionary mortality is no better than any of the other evidence for the Antonine plague—a few literary sources and some ambiguous, limited archaeological evidence—but what we do know of the living conditions of mid-second-century CE legionaries suggests high susceptibility to density-dependent pathogens. The armies of Verus and Marcus Aurelius were cultural and ethnic “melting pots”. After Marcus Aurelius staved off German invasions along the Danube with troops which had already been to Parthia and back, he immediately incorporated newly pacified German auxiliaries into the army; he then marched right back to the Near East after the usurpation of Avidius Cassius in Egypt.67 The soldiers in Rome’s diverse armies, furthermore, probably traded with local merchants, socialized in local establishments, and they certainly slept with local women.68 Roman soldiers must have been an important vector for the spread of all manner of disease in the Roman world. It was certainly the case that manpower shortages affected military recruiting during the Justinianic Plague.69 Therefore, there is no reason to assume that the army did not experience roughly similar, if not higher, mortality rates compared

64  Bray, Armies of Pestilence, 12; Grant, The Antonines, 32; Goodman, The Roman World, 82; Freeman, Egypt, Greece, and Rome, 709; Gourevitch, Limos Kai Loimos, 57; Stathakopoulos, Famine and Pestilence, 94. 65  Jer. Chron. 236, 237; Eutr. 8.12. 66  Beck, “Qui Mortalitatis Causa Convenerunt”. 67  Birley, “Hadrian to the Antonines,” 178. 68  McGinn, Prostitution, Sexuality, and the Law, 127–129; Phang, The Marriage of Roman Soldiers, 52–61. 69  Lee, “The Empire at War,” 118.

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

315

to Egyptian labourers. Is there a connection between the wages of soldiers and the impact of the Antonine plague? According to the most thorough study of Roman legionary pay yet produced, “we can be certain” that legionaries were paid 300 denarii per annum from Domitian to Septimius Severus.70 This is where certainty ends as the literary sources quantify wages in a variety of ways: sometimes ancient authors provide us with annual figures for military expenses, sometimes we are given a figure of pay per head, other times we are given an increase in some relative proportion to the previous wage (and we cannot be exactly sure what the “previous wage” was). Each soldier’s wage was primarily composed of a regular stipendium (salary), paid out mostly in cash but perhaps partially in kind as well. The ratio between cash and kind payments is unclear and certainly changed over the period in question. It is also unknown how far donativa (gifts) supplemented official salaries. Donativa were true “bonuses” in the early Principate—additional pay on top of the normal salary; yet by the second century CE, donativa seem to have become an expected, regularized part of the total compensation package. There is still evidence of exceptional donativa: Septimius Severus, for example, gave a 1000 sesterces donativum to the legions from Upper Pannonia in 193 CE.  Septimius was formerly governor of the province and used the legions there to secure his claim to the purple in Rome in 193 CE. The donativum is best tied to Septimius’ accession and the precarious position he was in as two other men had also claimed the purple. His ability to subdue rival claimants directly benefited from a generous donativum to his most loyal troops. It is possible but unlikely that donativa such as these were used as incentives for new recruits, as the viaticum—a three-aurei subsidy to cover travel expenses for new recruits—already existed.71 In short, it is not immediately clear how a recruitment crisis, perhaps due to a high-mortality plague, would be dealt with by army paymasters. Finally, there are eight documents which may attest rates of pay for soldiers in aggregate form, but it is unclear how to break apart the numbers to calculate exact rates of per-capita pay for certain types of soldiers.72 Hence, while most scholars have agreed that

70  Alston, “Roman Military Pay,” 115. The same rates appear in Duncan-Jones, Money and Government, 34; see also Speidel, “Roman Army Pay Scales Revisited”. 71  Phang, Roman Military Service, 168. 72  They are listed in Alston, “Roman Military Pay,” 115–120.

316 

C. P. ELLIOTT

Septimius’ pay rise elevated wages to 450 denarii per annum, the number is unrepresentative of total compensation. Military wages also present methodological problems: (1) legionary wages are administrative wages, not market wages,73 (2) as a consequence of their nominal nature, these wages give little sense of actual purchasing power (“real wages”),74 and (3) it is reasonable to hypothesize that provincials from regions with lower wages may have been supplementing the military at this time, reducing upward pressure on army wages.75 If the Antonine plague indeed killed some 15–30 per cent (or more) of the population, and this mortality rate was at least paralleled in the Roman army (if it was not greater, as we probably should assume) then we would expect that a reflexive, economically sensitive Roman state would have been required to increase army pay sometime between the middle of the 160s and the late 180s. Even in the absence of such economic rationalism, however, the need to obtain recruits during what several sources suggest was a manpower shortage presumably would have incentivized paymasters to offer higher pay.76 No such increase in base salary, however, occurs until the late 190s CE. In nominal terms, legionary pay stayed at 300 denarii per  annum from 83/4 CE until it finally rose to 450 denarii in 197 CE. The nominal pay rate, however, can only support a superficial analysis. Ideally, donativa would also be taken into account, as would the amount of actual silver paid (or available to exchange via bronze coinage) to legionaries. As for donativa, apart from a few specific cases, the evidence is almost entirely silent on general trends; it is unclear how often donativa were awarded, to which soldiers they were given, what the amounts were and how the amounts changed over time.77 Obviously these are all significant problems, but we must work with what is available. What little I believe can be said about legionary base pay is produced below (Fig. 9.4). Two lines are shown; the dotted line is the nominal rate of per-capita pay for legionaries, while the solid line is the amount of physical silver either paid directly to soldiers or which could be obtained by exchanging other coin denominations. Assuming that denominational exchange rates  Again, it is Temin who explains the concept best; see Temin, “Price Behaviour,” 190.  Scheidel, “Roman Real Wages in Context”. 75  Scheidel, “In Search of Roman Economic Growth,” 60. 76  SHA Marc. 21.6–7; Oros. 7.15.6, 27.7. See also ILS 2304, as discussed in DuncanJones, Structure and Scale, 72. 77  A list of all donativa attested in literary sources appears in Duncan-Jones, Money and Government, 257. 73 74

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

317

Fig. 9.4  Two measures of Roman legionary pay (160–211 CE). (Sources: Fineness figures through 192/3 CE come from Butcher and Ponting, “The Beginning of the End”; Elliott, “The Acceptance and Value of Roman Silver Coinage”. Figures from 193/4 through 211 are from Gitler and Ponting, The Silver Coinage of Septimius Severus and His Family, 55–57. Weights are those of Duncan-Jones, Money and Government, 227; Walker, The Metrology of the Roman Silver Coinage, 3–18)

held for soldier pay allows us to circumvent the possibility that some soldiers were not paid directly in silver. The solid line, which tracks wages in terms of silver, confirms that Rathbone was absolutely correct to believe that Septimus Severus’ pay rise simply returned wages to their previous levels in terms of purchasing power.78 In grams of silver, Septimius’ pay increase was not a pay rise in the context of the period as a whole. Rathbone argued that this was because price inflation had made the purchasing power of Roman currency fall between 160 and 190 CE. It may indeed be the case that price inflation occurred during that period, but it is also  Rathbone, “Monetisation Not Price Inflation,” 323.

78

318 

C. P. ELLIOTT

worth noting that in the same period the salary of the average Roman legionary in terms of silver bullion was also nearly halved by the mid-­190s— the opposite of both neo-Malthusian and neo-Classical expectations.79 How much did the silver content of coinage matter to the actual purchasing power of imperial coinage during the late second century? Unfortunately, the question can only be answered by engaging in a discussion about whether Roman currency circulated according to “metallist” or “chartalist” principles. Elsewhere, I make a case for a predominantly “metallist” situation—that the silver content in Roman coinage affected its purchasing power.80 If I am correct about this  assumption, then the graph indicates declining real wages. Even if my metallist predilections are mistaken, however, nominal salaries too were held steady until the late 190s CE. Both possibilities, despite their differences, do not show what we would expect to see among an army “slaughtered” (as Jerome claims) by plague: increases in base salaries. It does remain possible that officials passed out donativa rather than salary increases—unfortunately, we will never know. The evidence which does exist, however, even when interpreted through both neo-Malthusian and neo-Classical equilibrium, provides no support for the idea that the Antonine plague was both catastrophic and widespread. Although military salaries are state payments, the pressures on the state to raise army pay to attract scarce recruits would have been enormous, yet it appears that there was no rise in base salary until 197 CE.81 Even without the plague, the period’s confluence of civil and foreign wars—far more than had occurred in previous decades—substantially increased the demand for soldiers. Moreover, a period of entrenched warfare meant that soldiers faced much greater risk of death, dismemberment or disability, making recruitment all the more difficult. The conclusion we must draw from changes in military pay—if any conclusion should be drawn from such a problematic proxy—is reminiscent of Scheidel’s conclusion from his own study of Egyptian wages and  Indeed, if I had included the previous decade (150s CE), the first entry for 150 CE would have seen legionary pay at just over 787 grams of silver compared to just over 436 grams of silver in 196/7 CE. 80  Elliott, “The Acceptance and Value of Roman Silver Coinage”; Elliott, Economic Theory and the Roman Monetary Economy, 126–140. See also Bransbourg, “Fides et Pecunia Numerata. Part I”; Bransbourg, “Fides et Pecunia Numerata, Part II “. Also helpful is Scheidel, “Coin Quality, Coin Quantity, and Coin Value”; Haklai-Rotenberg, Aurelian’s Monetary Reform”. The issue is addressed with less clarity in Katsari, The Roman Monetary System. See also Crawford, “Money and Exchange”. 81  Kehoe, “Contract Labor,” 127. 79

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

319

prices: either the Antonine plague did not kill nearly as many people as scholars tend to suspect and/or the Roman state, whether for lack of information or desire, was unable or unwilling to account for the dynamics of the labour market.82 I do not think this latter conclusion is a stretch: as a category of wages, Roman military salaries fit the pattern expected of non-market wages; they show inflexibility in the face of changes to supply and demand as well as uniformity across provinces and regions, something seen in government salaries even in modern market-­oriented economies. From 84 CE, a uniform deduction of 100 sestertii per stipendium to cover a legionary’s wheat ration was removed regardless of the price of wheat (in Rome or elsewhere) and regardless of where the legionary was stationed.83 At the same time, both literary sources and archaeological evidence suggest that the costs of living across the Roman Empire varied wildly from place to place, whether or not a plague decimated the military population in the mid- to late second century CE. Did the Roman state, like so many states throughout history (again, inclusive of post-industrial market-oriented states), have the ability or even the inclination to pay optimum wages (i.e. wages which most closely resemble the point of intersection between supply and demand)? Instead of further evidence that “the Romans thought that in broad terms their empire had an integrated economy”, is it not equally likely, if not more plausible, that Roman state officials gave little thought to macro-economics at all?84 Regardless of which scenario is the most accurate, my aim here is to expose some contradictions in prevailing views of the period. The status quo is based upon two contradicting positions and cannot continue: if the Roman economy was integrated and plague was severe, then salaries would have had to have risen; however, the little evidence available suggests wages were mostly stagnant or perhaps even declined during the plague years. Either the plague was not severe or the Roman economy was not as integrated as we have been led to believe; or perhaps these explanations are both reasonable. I also have suspicions that the state’s role in the grain market was significant enough to dramatically swing prices during the Antonine and Severan period, but these too are discussed elsewhere.85  Scheidel, “Roman Wellbeing,’ 285.  Roth, The Logistics of the Roman Army at War, 14–15; Speidel, “Roman Army Pay Scales Revisited,” 360. 84  Rathbone, “Earnings and Costs”, 310. 85  Elliott, “Disease, Grain Supply and State Inflexibility”. See also Erdkamp, The Grain Market; Erdkamp, “The Food Supply of the Capital”; Erdkamp, “Economic Growth”; 82 83

320 

C. P. ELLIOTT

3   Conclusion: On Proxies and Quantification One of the more unsatisfying aspects of working with quantitative data is that such data are often used destructively: to falsify hypotheses rather than in support of new arguments. This paper, in which I offer  adjustments or augmentations of existing perspectives on the Antonine plague without offering much to support my own view (than the plague had a moderate if not minor impact), is no exception. As I mentioned before, some of my corroborating views on the remarkable events of the mid- to late second century CE can be found elsewhere with a more comprehensive framework—one which involves the interplay of climate, geography, state institutions and markets—to emerge as additional work is published.86 My chapter here is contextualized within a volume broadly centred upon “complexity economics”—a school of economics in which economic complexity is explored, usually through computational modelling. While I do not use computational modelling in this chapter, I do investigate the thinking and assumptions which are modelled and computed; after all, computers compute—they do not think. The products of computational models, like econometric equations, are a direct product of the quantitative inputs and the assumptions of the programming. The use of such an approach in the study of the ancient economy will, like most quantitative approaches, produce a healthy pursuit of new evidence; at the same time, as with any approach, there are excesses which must be avoided. In addition to improving the inputs of such programs, scholars must continue to think about the assumptions which underlie the models and programs they create. As I have argued here, for example, the often co-dependent assumptions of Roman market integration and a “severe” Antonine plague do not seem to work with the available price and wage evidence from Egypt. These sorts of conclusions stymie the simple economic  models, creating a demand for models (and programs) which can accommodate additional economic complexity. Some of the methodological hallmarks of New Institutional economics, particularly transaction costs and bounded rationality, are certainly helpful in this regard. Computer models too, as the studies in this volume prove, also offer tremendous potential. At the Silver, “The Plague under Commodus”; Bang, The Roman Bazaar; Garnsey, Famine and Food Supply. 86  Elliott, “Silver Debasement, Climate and Plague”; Elliott, “Disease, Grain Supply and State Inflexibility; Elliott, “Climate Change, Plague and Local Violence”; Elliott, “The Acceptance and Value of Roman Silver Coinage”.

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

321

same time, caution is important. One of the disastrous results of the Finleyan turn in ancient history was a deep-seated mistrust of social science methodology. It would be equally disastrous, however, to eschew humanistic methodologies of comparative history, culture-based study and other non-experimental approaches, even when constructing the computational models required for studying complexity. Humanistic methodologies, being qualitative in nature, may not produce the precise numbers currently in demand, but they do, in my view, produce better arguments and more thoughtful assumptions. Without these, quantitative data are devoid of meaning.

Bibliography Alston, R. “Roman Military Pay from Caesar to Diocletian”, The Journal of Roman Studies 84 (1994): 113–123. Bagnall, Roger S. Currency and Inflation in Fourth Century Egypt. Chico, CA: Scholars Press, 1985. Bagnall, Roger S. “The Effects of Plague: Model and Evidence”, Journal of Roman Archaeology 15 (2002): 114–120. Bang, Peter F. “Trade and Empire. In Search of Organizing Concepts for the Roman Economy”, Past & Present 195 (2007): 3–54. Bang, Peter F. The Roman Bazaar: A Comparative Study of Trade and Markets in a Tributary Empire, Cambridge: Cambridge University Press, 2008. Beck, R. “‘Qui Mortalitatis Causa Convenerunt’: The Meeting of the Virunum Mithraists on June 26, A. D. 184”, Phoenix 52 (1998): 335–344. Birley, A. R. “Hadrian to the Antonines”, In The Cambridge Ancient History. Vol. 11, edited by Alan K.  Bowman, Peter Garnsey and Dominic W.  Rathbone, 1032–194. Cambridge: Cambridge University Press. Blouin, Katherine. Triangular Landscapes: Environment, Society, and the State in the Nile Delta under Roman Rule, Oxford: Oxford University Press, 2014. Bransbourg, Gilles. “Fides et Pecunia Numerata. Chartalism and Metallism in the Roman World Part I: The Republic”, American Journal of Numismatics 23 (2011) 87–152. Bransbourg, Gilles. “Fides et Pecunia Numerata, Part II: The Currencies of the Roman Republic”, in American Journal of Numismatics 25 (2013): 179–242. Bray, R.  S. Armies of Pestilence: The Impact of Disease on History, Cambridge: Cambridge University Press, 2004. Bruun, Christer F.  M. “The Antonine Plague in Rome and Ostia”. Journal of Roman Archaeology 16 (2003): 426–34. Bruun, Christer F. M. “The Antonine Plague and the ‘Third-Century Crisis’”. In Crises and the Roman Empire: Proceedings of the Seventh Workshop of the

322 

C. P. ELLIOTT

International Network Impact of the Empire (Nijmegen, June 20–24, 2006), edited by Daniëlle Slootjes and Olivier Hekster, 201–17. Impact of Empire. Leiden; Boston: Brill, 2007. Bruun, Christer F. M. “La Mancanza Di Prove Di Un Effetto Catastrofico Della « Peste Antonina » (Dal 166 d.C. in Poi)”. In L” Impatto Della « Peste Antonina », edited by Elio Lo Cascio, 123–63. Pragmateiai 22. Bari: Edipuglia, 2012. Butcher, Kevin and Matthew Ponting. “The Beginning of the End? The Denarius in the Second Century”, Numismatic Chronicle 172 (2012): 63–83. Clark, Gregory. A Farewell to Alms: A Brief Economic History of the World. Princeton: Princeton University Press, 2007. Cleary, Simon E. The Roman West, AD 200–500. An Archaeological Study. Cambridge, 2013. Crawford, Michael H. “Money and Exchange in the Roman World”, The Journal of Roman Studies 60 (1970): 40–48. Drexhage, Hans-Joachim. Preise, Mieten/Pachten, Kosten und Löhne im römischen Ägypten bis zum Regierungsantritt Diokletians. St. Katharinen: Scripta Mercaturae, 1991. Duncan-Jones, Richard. Structure and Scale in the Roman Economy. Cambridge; New York: Cambridge University Press, 1990. Duncan-Jones, Richard. Money and Government in the Roman Empire. Cambridge: Cambridge university press, 1994. Duncan-Jones, Richard P. “The Impact of the Antonine Plague”. Journal of Roman Archaeology 9 (1996): 108–36. Elliot, Colin P. “The Acceptance and Value of Roman Silver Coinage in the Second and Third Centuries AD”. Numismatic Chronicle 174 (2014): 130–52. Elliott, Colin P. “Climate Change, Plague and Local Violence in Roman Egypt”, Past & Present 230 (2016): 3–31. Elliott, Colin P. Economic Theory and the Roman Monetary Economy. Cambridge: Cambridge University Press, 2020. Elliott, Colin P. “Silver Debasement, Climate and Plague in Second-Century A.D. Egypt.” In Debasement: Manipulation of Coin Standards in Pre-Modern Monetary Systems, edited by Kevin Butcher. Oxford: Oxbow, forthcoming-a. Elliott, Colin P. “Disease, Grain Supply and State Inflexibility in the Late Second Century A.D.” In Disease in the Ancient World, edited by R.  Arnott and R. Breitwieser. Oxford: Oxbow, forthcoming-b. Erdkamp, Paul. The Grain Market in the Roman Empire: A Social, Political and Economic Study. Cambridge; New York: Cambridge University Press, 2005. Erdkamp, Paul. “The Food Supply of the Capital”, In The Cambridge Companion to Ancient Rome, edited by Paul Erdkamp, 262–277. Cambridge, 2013 Erdkamp, Paul. “Economic Growth in the Roman Mediterranean World: An Early Good-Bye to Malthus?” Explorations in Economic History 60 (2016): 1–20.

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

323

Freeman, C. Egypt, Greece, and Rome: Civilizations of the Ancient Mediterranean. Oxford, 2014. Garnsey, Peter. Famine and Food Supply in the Graeco-Roman World: Responses to Risk and Crisis. Cambridge: Cambridge University Press, 1988. Gilliam, J.  F. “The Plague under Marcus Aurelius”. The American Journal of Philology 82, no. 3 (1961): 225–51. Gitler, Haim, and Matthew Ponting. The Silver Coinage of Septimius Severus and His Family, 193–211 A.D: A Study of the Chemical Composition of the Roman and Eastern Issues. Milano: Ennerre, 2003. M. Goodman, Martin. The Roman World 44 BC–AD 180. London, 2013. Gourevitch, Danielle. Limos Kai Loimos: A Study of the Galenic Plague. Collection Pathographie, 10. Paris: de Boccard, 2013. Grant, Michael. The Antonines: The Roman Empire in Transition. London, 1994. Greenberg, James. “Plagued by Doubt: Reconsidering the Impact of a Mortality Crisis in the 2nd c. A.D.” Journal of Roman Archaeology 16 (2003): 413–25. Haklai-Rotenberg, Merav. “Aurelian’s Monetary Reform: Between Debasement and Public Trust”. Chiron 41 (2011): 1–39. Haley, Evan W. Baetica Felix: People and Prosperity in Southern Spain from Caesar to Septimius Severus. Austin: University of Texas Press, 2003. Hansen, G.D. and E.C. Prescott. “Malthus to Solow”, American Economic Review 92 (2002): 1205–1217. Harper, Kyle. “People, Plagues, and Prices in the Roman World: The Evidence from Egypt”. The Journal of Economic History 76, no. 3 (September 2016): 803–39. Harris, William V. “The Great Pestilence and the Complexities of the Antonine-­ Severan Economy”. In L” Impatto Della « Peste Antonina », edited by Elio Lo Cascio, 331–38. Pragmateiai 22. Bari: Edipuglia, 2012. Howgego, Christopher. “Some Numismatic Approaches to Quantifying the Roman Economy”, In Quantifying the Roman Economy: Methods and Problems, edited by Alan K.  Bowman and Andrew Wilson, 287–295. Oxford: Oxford University Press, 2009. Johnson, Allan Chester. Roman Egypt to the Reign of Diocletian. An Economic Survey of Ancient Rome. Volum II. Baltimore: Johns Hopkins Press, 1936. Jongman, Wim. “The New Economic History of the Roman Empire”, In Quantifying the Greco-Roman Economy and Beyond, edited by François de Callataÿ, 169–88. Bari: Edipuglia, 2014. Katsari, Constantina. The Roman monetary system: the Eastern provinces from the first to the third century AD. Cambridge; New  York: Cambridge University Press, 2011. Kehoe, Dennis. “Contract Labor”. In The Cambridge Companion to the Roman Economy, edited by Walter Scheidel and Walter Scheidel, 114–30. Cambridge: Cambridge University Press, 2012.

324 

C. P. ELLIOTT

Kron, Geoffrey. “Nutrition, Hygiene and Mortality: Setting Parameters for Roman Health and Life Expectancy Consistent with Our Comparative Evidence”. In L” Impatto Della « Peste Antonina », edited by Elio Lo Cascio, 193–250. Pragmateiai 22. Bari: Edipuglia, 2012. Lee, A.D. “The Empire at War”, In The Cambridge Companion to the Age of Justinian, edited by M.  Maas, 113–33. Cambridge: Cambridge University Press, 2005. Little, Lester K. “Life and Afterlife of the First Plague Pandemic”, In Plague and the End of Antiquity: The Pandemic of 541–750, edited by Lester K.  Little, 3–32. Cambridge: Cambridge University Press, 2006. Littman, R.  J., and M.  L. Littman. “Galen and the Antonine Plague”. The American Journal of Philology 94 (1973): 243–55. Mayerson, P. “The Monochoron and Dichoron: Standard Measures for Wine Based on the Oxyrhynchition”, Zeitschrift für Papyrologie und Epigraphik 131 (2000): 169–72. McGinn, T. A. J. Prostitution, Sexuality, and the Law in Ancient Rome. Oxford: Oxford University Press, 2003. Minnen, Peter van. “P. Oxy. LXVI 4527 and the Antonine Plague in the Fayyum”. Zeitschrift Für Papyrologie Und Epigraphik, no. 135 (2001): 175–77. Paine, R. R. and G. R. Storey. “The Alps as a Barrier to Epidemic Disease during the Republican Period: Implications for the Dynamic of Disease”, In L” Impatto Della « Peste Antonina », edited by Elio Lo Cascio, 179–91. Pragmateiai 22. Bari: Edipuglia, 2012. Phang, Sara Elise. The Marriage of Roman Soldiers (13 B.C.–A.D. 235): Law and Family in the Imperial Army. Leiden; Boston: Brill, 2001. Phang, Sara Elise. Roman Military Service: Ideologies of Discipline in the Late Republic and Early Principate. Cambridge: Cambridge University Press, 2008. Rathbone, Dominic. “Villages, Land and Population in Graeco-Roman Egypt”, Proceedings of the Cambridge Philological Society 36 (1990): 103–42. Rathbone, Dominic. Economic Rationalism and Rural Society in Third-Century A.D.  Egypt: The Heroninos Archive and the Appianus Estate. Cambridge; New York: Cambridge University Press, 1991. Rathbone, Dominic W. “Monetisation Not Price Inflation in Third-Century Egypt?” In Coin Finds and Coin Use in the Roman World: The Thirteenth Oxford Symposium on Coinage and Monetary History, 25.-27.3.1993, a NATO Advanced Research Workshop, 321–40. Berlin: G. Mann, 1996. Rathbone, Dominic. “Prices and Price Formation in Roman Egypt”. In Économie Antique Prix et Formation Des Prix Dans Les Économies Antiques, edited by Jean Andreau, Pierre Briant, and Raymond Descat, 183–244. Saint-Bertrand-­ de-Comminges: Musée archéologique départemental, 1997.

9  DISEASE PROXIES AND THE DIAGNOSIS OF THE LATE ANTONINE… 

325

Rathbone, Dominic. “Roman Egypt”, In The Cambridge Economic History of the Greco-Roman World, edited by Walter Scheidel, Ian Morris and Richard P. Saller, 698–719. Cambridge: Cambridge University Press, 2007. Rathbone, Dominic W. “Earnings and Costs: Living Standards and the Roman Economy (First to Third Centuries Ad)”. In Quantifying the Roman Economy: Methods and Problems, edited by Alan K.  Bowman and Andrew Wilson, 299–326. Oxford: Oxford University Press, 2009. Rathbone, Dominic W. and Sitta Von Reden. “Mediterranean Grain Prices in Classical Antiquity”, In A History of Market Performance: From Ancient Babylonia to the Modern World, edited by Robert J. Van der Spek, Jan Luiten van Zanden and B. van Leeuwen, 149–235. London: Routledge, 2015. Reynolds, P. “The Supply Networks of the Roman East and West: Interaction, Fragmentation, and the Origins of the Byzantine Economy”. In Trade, Commerce, and the State in the Roman World, edited by Andrew Wilson and Alan K. Bowman, 353–96. Oxford: Oxford University Press, 2018. Roth, Jonathan P. The Logistics of the Roman Army at War (264 B.C–A.D. 235). Leiden; Boston: Brill, 1999. Salmon, Pierre. Population et Dépopulation Dans l”Empire Romain. Bruxelles: Latomus, 1974. Scheidel, Walter. “A Model of Demographic and Economic Change in Roman Egypt after the Antonine Plague”. Journal of Roman Archaeology: An International Journal 15 (2002): 97–114. Scheidel, Walter. “In Search of Roman Economic Growth”. Journal of Roman Archaeology 22, no. 1 (2009): 46–70. Scheidel, Walter. “Real Wages in Early Economies: Evidence for Living Standards from 1800 BCE to 1300 CE”. Journal of the Economic and Social History of the Orient 53 (2010): 425–62. Scheidel, Walter. “Coin Quality, Coin Quantity, and Coin Value in Early China and the Roman World”. American Journal of Numismatics 22 (2010): 93–118. Scheidel, Walter. “Roman Wellbeing and the Economic Consequences of the “Antonine Plague” (with a Contribution by John Sutherland)”. In L”impatto Della “Peste Antonina” (Incontri Capresi Di Storia Dell”economia Antica (5th: 2008: Anacapri, Italy, and Rome, Italy)), edited by Elio Lo Cascio, 265–95. Bari: Edipuglia, 2012. Scheidel, Walter. “Roman Real Wages in Context”. In Quantifying the Greco-­ Roman Economy and Beyond, edited by François de Callataÿ, 209–18. Bari: Edipuglia, 2014. Silver, Morris. “The Plague under Commodus as an Unintended Consequence of Roman Grain Market Regulation”. The Classical World 105, no. 2 (2012): 199–225.

326 

C. P. ELLIOTT

Speidel, M. “Roman Army Pay Scales Revisited: Responses and Answers, In De l”or pour les braves  ! Soldes, armées et circulation monétaire dans le monde romain, edited by Michel Reddé, 53–62. Bordeaux, 2014. Stathakopoulos, Dionysios Ch. Famine and Pestilence in the Late Roman and Early Byzantine Empire: A Systematic Survey of Subsistence Crises and Epidemics. London: Routledge, 2017. Temin, Peter. “A Market Economy in the Early Roman Empire”. The Journal of Roman Studies 91 (2001): 169–81. Temin, Peter. “The Contribution of Economics”. In The Cambridge Companion to the Roman Economy, edited by Walter Scheidel and Walter Scheidel, 45–70. Cambridge: Cambridge University Press, 2012. Temin, Peter. The Roman Market Economy. The Princeton Economic History of the Western World. Princeton, NJ: Princeton University Press, 2013. Temin, Peter. “Price Behaviour in the Roman Empire”, In Quantifying the Greco-­ Roman Economy and Beyond, edited by François de Callataÿ, 189–207. Bari: Edipuglia, 2014. Walker, David Richard. The Metrology of the Roman Silver Coinage. Part 3. From Pertinax to Uranius Antoninus. Oxford: British Archaeological Reports, 1978. Wiemer, Hans-Ulrich. Libanios und Julian. Studien zum Verhältnis von Rhetorik und Politik im vierten Jahrhundert n. Chr. München: Beck, 1995. Wilson, Andrew. “The Mediterranean Environment in Ancient History: Perspectives and Prospects”. In The Ancient Mediterranean Environment between Science and History, 259–76. Leiden; Boston: Brill, 2013. Wilson, Andrew. “Quantifying Roman Economic Performance by Means of Proxies: Pitfalls and Potential”. In Quantifying the Greco-Roman Economy and Beyond, edited by François de Callataÿ. Pragmateiai 27. Bari: Edipuglia, 2014.

CHAPTER 10

Measuring and Comparing Economic Interaction Based on the Paths and Speed of Infections: The Case Study of the Spread of the Justinianic Plague and Black Death Lars Börner and Battista Severgnini

1   Introduction The motivation for writing this chapter is twofold. On the one hand, this study is inspired by the analysis of the determinants of long-run growth of Europe and the Mediterranean. Economic historians have identified various stages in the development of societies in this area. For instance, the

L. Börner (*) Martin-Luther-University of Halle-Wittenberg, Halle, Germany DAFM, King’s College London, London, UK e-mail: [email protected] B. Severgnini Copenhagen Business School, Frederiksberg, Denmark e-mail: [email protected] © The Author(s) 2021 K. Verboven (ed.), Complexity Economics, Palgrave Studies in Ancient Economies, https://doi.org/10.1007/978-3-030-47898-8_10

327

328 

L. BÖRNER AND B. SEVERGNINI

period of the so-called Commercial Revolution1 during the late Middle Ages has been identified as a phase of urbanisation characterized by an increase in trade activities and the formation of economic and trade institutions that would become the foundation of the rise of the West during the following centuries.2 Other studies reach further back in time to investigate the Roman economy and society from a contemporary perspective and analyse the role of the Roman Empire in the later development of modern civilizations.3 Important elements are state formation, political organization, Roman legislation, a common language and religious values, and the creation of an early market economy based on a strong transportation system and infrastructure, for instance, in the form of Roman roads and safe sea trade routes. However, whereas there has been an attempt to build a link between these or other influential past epochs and more recent economic developments, very little has been done to develop a comparative perspective between past civilizations using quantitative methods. A comparison of potential legacies between the late Roman Empire and the rise of the Commercial Revolution would be interesting, since the historical sequence from the Ancient Period to the Middle Ages is understood as a cultural, institutional, and structural break from the waning Roman Empire rather than a smooth evolutionary process of growth and development.4 In fact, the surge of the Roman Empire in the form of the Justinianic Empire during the sixth century has been interpreted as a final phase of prosperity and growth, and the subsequent centuries have been interpreted as a phase of decline, with the political and economic disintegration of the Mediterranean and Europe until the abovementioned Commercial Revolution. This phase of disintegration and structural break has been used as an argument  Lopez, The Commercial Revolution.  Cipolla, Before the Industrial Revolution; Epstein, Freedom and Growth; Greif, Institutions. 3  Brown, Poverty and Leadership in the Later Roman Empire; Burdick, The Principles of Roman Law and Their Relation to Modern Law; Aldrete and Aldrete, The Long Shadow of Antiquity; Temin, The Roman Market Economy; Michaels and Rauch, “Resetting the Urban Network”. 4  The fall of the Roman Empire has been widely discussed, for instance, by Gibbon, Decline and Fall; Burckhardt, Die Zeit Constantins des Großen; Rostovtzeff, The Social and Economic History of the Roman Empire; Mazzarino, La fine; Ward-Perkins, The Fall of Rome; Goldsworthy, The Fall of the West, and many others. However, some scholars understand this period rather as a smoother process of relative decline than a total disintegration. For instance, see Pirenne, Medieval Cities; Brown, The Making of Late Antiquity; McCormick, Origins of the European Economy. 1 2

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

329

to understand the starting point of the rise of the West during the Commercial Revolution rather than the legacy of the Roman Empire.5 Comparisons between past civilizations and epochs are rare and difficult to carry out, and the insights are often rather controversial. Whereas a comparison between recent developments and past civilizations gives at least a clear benchmark with regards to quantitative statistics of the actual or recent developments, comparisons between past civilizations leave us with guesses about the economic developments of two different periods. For instance, studies comparing standards of living (and related urbanisation and population size numbers) between the Commercial Revolution during late Middle Ages and the economic heydays of the Roman Empire have led to controversial conclusions. Lo Cascio and Malanima, for example, argued that both periods might have been comparable in GDP numbers per capita.6 Scheidel has claimed that standards of living must have been clearly lower during the Roman Empire. Nevertheless, he derives similar urbanisation rates (which is typically a proxy for economic development) but concludes, based on a Malthusian perspective, that Roman society must have been under stronger economic pressure than in the later epoch, with little scope for further population growth. It was thus trapped in a Malthusian low-level equilibrium.7 Such challenges of comparative studies in a very long-run perspective due to missing quantifiable information bring us to the second part of the motivation for this study, that is, the use of an alternative and indirect measurement to understand economic dynamics within societies, namely, the spread of diseases. In a previous work, we showed that the spread of diseases can be used to better understand economic and social interaction in societies.8 In particular, we showed that the speed of the spread of the Black Death along trading routes can be used to identify determinants of trade during the period of transmission. Following such a methodology, it can be shown that “economic exchanges” can be explained not only by the transportation technology but also by institutions in the form of political borders or religious events, which have an impact on the speed of the 5  Cipolla, Before the Industrial Revolution; Epstein, Freedom and Growth; Greif, Institutions. A different perspective is provided by McCormick, Origins of the European Economy. He argues that the rise of Charlemagne in the eighth and ninth centuries (before the Commercial Revolution) can be considered as “The Origin of Europe”. 6  Lo Cascio and Malanima, “Ancient and Pre-Modern Economies” GDP. 7  Scheidel, “Demography”; Scheidel, “Real Wages in Early Economies”. 8  Boerner and Severgnini, “Epidemic Trade”.

330 

L. BÖRNER AND B. SEVERGNINI

spread of the disease and correspond with medieval trade flows. Furthermore, the identification of key variables of the determinants of trade activities allows one to estimate relative trade intensity between cities and regions. In addition, the speed of the spread of disease can be linked to long-run economic growth dynamics. Thus, the speed of the spread can be used as a proxy variable to measure economic and social interaction. These results are especially useful for understanding the abovementioned historical period, since the Black Death spread at the peak of the Commercial Revolution all over Europe and the Mediterranean. Therefore, the results reflect the role of economic activities during a very important epoch. Based on these findings, we can raise the question of whether such an analysis is also possible for the spread of the Justinianic Plague, which was disseminated throughout the Mediterranean during the middle of the sixth century. Furthermore, we can investigate whether we can create a comparative perspective based on the speed of the spread of these two diseases. Such a connection would indeed enable us to compare the two already-mentioned epochs: the end of the Roman Empire, at the time of its recent surge based on Justinian’s Empire, and the time of the Commercial Revolution, the heydays of medieval trade. Looking at the spread of the Justinianic Plague, we find that the diffusion process follows a similar pattern as the spread of the Black Death. In particular, we show that the speed is determined by the trade geography and technology available at the time. It correlates in a positive way with the Stanford ORBIS data set,9 which measures the average speed of transportation between Roman towns during the Roman Empire. However, comparing the speeds of both diseases, we find that the average speed on land and sea trade routes is much faster during the Black Death than during the Justinianic Plague. The contamination of the Mediterranean took nearly twice as long as during the late Middle Ages. We argue that this difference can be linked to differences in the economic strength of both epochs. However, the question of whether this difference can be mainly assigned to the relative economic decline and disintegration of the Roman Empire or is indeed related to the new strengths of trade, trade institutions, and economic competition among the economic powerhouses of the time, namely, Genoa and Venice, cannot be fully answered. What we 9

 ORBIS. The Stanford Geospatial Network Model of the Roman World, orbis.stanford.edu.

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

331

can offer is a new quantitative method (that measures social and economic interactions in relative but not absolute terms) that strongly suggests that we can indeed find a strong difference. To achieve these results, we collect data on the spread of the diseases, match them with the destinations along the historically known trade routes, estimate the speed between destinations based on the time and distance (which is based on GIS data), and try to relate it to various institutional and geographical variables. To estimate the coefficients of the explanatory variables of the speed of the Black Death, econometric regression techniques were used that control for non-observable missing variables and non-linearity. Due to the small number of observations of the spread of the Justinianic Plague, we mainly have to work with simple descriptive statistics, correlation coefficients, and qualitative interpretation of the data and use these key figures to create a quantitative comparison. We then discuss the results in the historical context. Our chapter contributes in several ways to the existing literature: First, it contributes to a better understanding of the spread of diseases. It shows that both plagues were determined by the social and economic interactions in societies. Building on our existing results, we show that the Justinianic Plague followed a similar pattern as the Black Death. These results support claims by scholars who argue that both diseases were spread by a Bubonic Plague. The bacillus of the Yersinia pestis was transmitted by fur fleas living on rats, which could only travel long distances beyond the local city walls on merchant cargo of ships and caravans.10 Second, it contributes to the discussion of the decline of the Roman Empire and the rise of the West during the period of the Commercial Revolution. Our heterodox approach indeed identifies a strong difference between the economic strength of Justinian’s Empire and the period of the Commercial Revolution measured by our proxy variable for social and economic interaction between cities and regions. However, the question of whether the identified difference measures a “smooth” decline of the Roman Economy, as recently described by scholars of late Antiquity,11 or a sharp economic and political disintegration of the Roman Empire,12 or even just shows the 10   Cipolla, “The Plague”; McCormick, Origins of the European Economy; Sallares, “Ecology”, among others. 11  Brown, The Making of Late Antiquity; McCormick, Origins of the European Economy. 12  Rostovtzeff, The Social and Economic History of the Roman Empire; Ward-Perkins, The Fall of Rome; Goldsworthy, The Fall of the West.

332 

L. BÖRNER AND B. SEVERGNINI

new strengths of the Commercial Revolution13 cannot be answered conclusively. The remainder of this chapter is structured as follows. In Sect. 2, we describe the spread of the Black Death and discuss the main causes of its spread as discussed in the related literature. Next, we present our methodology: We show how we measure the speed of the spread of the plague, determine the causes of the speed, and link it to trade and economic growth. In Sect. 3, we describe the spread of the Justinianic Plague and elaborate again on the causes. A main goal is to identify similarities of both diseases to be able to make a quantitative comparison. Section 4 compares the spread of the two diseases based on quantitative output and qualitative observations. In addition, differences in the spread are discussed and interpreted in the historical economic contexts. Finally, Sect. 5 concludes.

2   The Black Death 2.1  The Spread of the Black Death The Black Death entered the Black Sea area in 1346 via the Silk Road and spread throughout the Mediterranean and Europe along the main trade routes until 1351.14 Figure  10.1 visualizes the geographical dispersion over time. Starting with the year 1346 (yellow lines), each year is marked with a different colour. The first documented infected area is in the Middle East, starting with Tabriz, a town along the Silk Road. Next, in 1347 (red lines), the Mediterranean started to be contaminated. Starting in Kaffa on the Black Sea (one of the end points of the Silk Road), the Eastern and Western Mediterranean were reached via Constantinople. The disease reached the hinterland close to the Mediterranean coast by the end of the year. It then continued along water trade routes to North West Europe, including England, Northern France, Belgium, the Netherlands, Norway, and the first cities on the Baltic Sea in 1348 (blue lines). During the same time, the disease spread farther inland via France to Switzerland and  Lopez, The Commercial Revolution.  The list of historians who contributed to the identification of the place and time of the diffusion of the plague is long and reaches back to the late nineteenth century. A recent and comprehensive description of the spread can be found in Benedictow, The Black Death. 13 14

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

333

Fig. 10.1  The spread of the Black Death

Southern Germany and then spread throughout Italy. The Black Death continued to spread to the West from the Adriatic coast via the Balkans and the river Danube, infecting Hungary and Austria. It then propagated farther into the Iberian Peninsula. During 1349 (green lines), the plague contaminated further parts of the Baltics and large parts of Germany and Eastern Europe. The disease approached the region from the South and from the coastlines of the North and Baltic seas. Finally, in 1350 (orange lines) and 1351 (grey lines), we have further evidence of the spread in the very north of the Baltic Sea and Russia. The map covers all towns for which we have information and can show the first infection. We do not cover any additional observations where the disease returned, which would create additional feedback loops.

334 

L. BÖRNER AND B. SEVERGNINI

2.2  How Did the Black Death Spread? The Black Death was transmitted to human beings by flea bites on black rats. The fleas and rats were carriers of the Bubonic Plague.15 These fleas travelled with merchant goods along sea and land trade routes. Neither the fleas nor the rats could travel on their own. Thus, human social interaction among cities and regions was required to spread the disease. We also tested this argument by looking at the pattern of diffusion of the disease and we found an s-shaped curve, typical for diffusion through social interaction.16 The main cause of spatial social interaction during the period of the Commercial Revolution was economic activity in the form of trade between cities and regions. Other social movements were comparably marginal or can be excluded as a transmitter of the plague. One potential transmitter could have been war activities, but except for two local/ regional cases, no war activities could be observed during the spread of the Black Death. Another transmitter could have been pilgrims. However, no special effect along pilgrim routes could be identified.17 Finally, migration flows could be incorporated as an additional effect. However, strong migration flows can be mainly observed after the spread of the Black Death.18 In addition, if we assume an occasional flow of migrants in the years of the spread of the plague, then they travelled with merchants along the trading routes and typically on their boats. Thus, it is reasonable to claim that the Black Death was mainly spread by merchant activities along trade routes.19 In addition, Benedictow has shown based on case studies 15  For a recent DNA recovery from victims of the Black Death and identification of the bacterium Yersinia pestis, see Bos et al., “A Draft Genome of Yersinia Pestis from Victims of the Black Death”; or Callaway, “Plague Genome”. 16  Boerner and Severgnini, “Epidemic Trade”. 17  We tested this argument in the empirical exercise we describe in the next section. 18  Domar, “The Causes of Slavery or Serfdom”. 19  Since the 1970s, the causal link between the dispersion of the Black Death and trade has been confirmed by several studies in economic history (Cipolla, “The Plague and the PreMalthus Malthusians”), medieval history (Herlihy and Cohn, The Black Death and the Transformation of the West; Bridbury, “Before the Black Death”.), demography (Livi-Bacci, “The Nutrition-Mortality Link in Past Times”; Livi Bacci, Concise History of World Population.), and medicine (Pollitzer, Plague), and most recently by Benedictow, The Black Death. There exist a small number of contributions, most recently represented by Cohn, “The Black Death: End of a Paradigm”, that claim that there is no such relationship (they neither confirm the existence of bacillus in form of the Bubonic Plague nor the transmitter

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

335

that the Black Death indeed spread faster along major trade routes than along less important trade links to the hinterland.20 Finally, an important aspect of the spread of the Black Death is that it could not be stopped or influenced by late Medieval towns. For instance, the introduction of quarantines for merchants and their trade goods was only introduced as a health policy later on and indeed helped to prevent the outbreak of diseases once the causes were recognized.21 2.3   A Quantitative Approach to Study the Spread of the Black Death22 Medieval historians have documented very precise data on the spread paths and dates of the infection of medieval cities. In this way, the dispersion of the Black Death can be quantified in the Mediterranean and Europe in the period from 1346 to 1351. In particular, we can measure how fast the disease spread along trading routes, looking at the time and distance from one destination to the next. Such an analysis results in more than 200 city pairs where the average speed in kilometres per day can be measured. Furthermore, looking at the descriptive statistics of this sample, a large variation in speed can be measured. This variation makes it interesting to empirically exploit and figure out what determined the differences in the speed of infection between different routes and destinations. In particular, it can be tested whether trade indeed mattered for the spread of the plague, as claimed by a large number of scholars from varies disciplines, as outlined in the previous section. To do this, first, a theoretical model must be created that replicates the medieval trade environment and is able to determine (and predict) the main trade-promoting and trade-restricting variables between two trading destinations. Looking at the large menu of trade models that have been created in economic theory over the last few decades, an Armington type

in the form of rats, so they ultimately reject the link between social diffusion and the spread along commercial trade routes). However, neither Cohn nor any other scholar offers any clear alternative explanation. 20  Benedictow, The Black Death. 21  Cipolla, “The Plague”; Slack, “The Disappearance of Plague”. 22   As outlined, this section is based on  our past work and  summarizes Boerner and Severgnini, “Epidemic Trade”.

336 

L. BÖRNER AND B. SEVERGNINI

of trade model is meaningful to work with.23 This model assumes that ­cities (or their population) have a preference for consuming a variety of products and that different regions or cities have an interest in trading with each other if the foreign products are sufficiently different from the locally produced goods. Indeed, we can observe a large number of products traded in regional and long-distance markets, including bulk goods and luxury products. At the same time, merchants face trading costs. These costs should not surpass the purchase prices of these products on the local market; otherwise, there will be no demand. Such costs are typically related to the physical time to travel, political borders, and other cultural factors, such as different languages, that make it more costly to trade. Indeed, we can characterize medieval trade routes using such characteristics, and we can identify different trade costs related to the trade geography and technology. For instance, transportation along water routes was typically cheaper and faster than along land trade routes. Furthermore, the area of investigation can be characterized by a larger number of political territories and political borders that had to be crossed. Also, a large number of different languages were spoken, which enriched the cultural variety and may have created cultural barriers to trade. Under these assumptions, the trade intensity increases when the supplied products are sufficiently diverse and the trading costs sufficiently low. In a next step, such a mechanism can be tested in an empirical analysis.24 The dependent variable that is going to be explained is the speed of transmission of the plague between two cities. This is used as a proxy for relative trade intensity between cities.25 The more trade occurs between two cities, the more merchants travel with their cargo between the cities and the more likely and faster a city will become infected, assuming the other trading partner in a city pair has already been contaminated. As explanatory variables, we collected information on the city characteristics and trade routes of the identified city pairs, containing geographical, social, political, religious, and climate variables. Looking at the 23  Armington and other scholars: Armington, “A Theory of Demand for Products Distinguished by Place of Production”. 24  We follow here the empirical trade literature, which mainly follows an empirical gravity model approach; see Anderson and van Wincoop, “Gravity with Gravitas”. 25  Please note that we measure relative trade intensity by comparing the speed between different city pairs. In addition, the speed of transmission is measured in one direction only. Thus, we measure commercial flow from the outbreak in the corner of the Eastern Mediterranean to the rest of the Mediterranean basin and Europe.

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

337

geographical variables, we considered the physical location (altitude, longitude, and latitude) of the cities, the distance between them, and whether the centres are located by a sea or by a river. To check the effective time to travel, we looked at the Stanford ORBIS data set, which provides information about the travelling time during the Roman Empire. This also works for the late Medieval period, since transportation technologies mainly changed after the outbreak of the Black Death.26 Concerning the social and political variables, we checked whether bishops, princes, and universities were in those cities. Finally, we controlled for the time period, since trade and the dispersion of the disease could be influenced by religious seasons such as Advent and Lent, by the climate, or by any other transformation of the bacterium over time. The main results of our econometric model, also confirmed by several robustness checks, consistently show that the different means of transportation, geographical distance, regions divided by political borders, the institutions, and the religious holidays played an important role in explaining trade. More precisely, the speed of the disease positively correlates with the physical time to travel, following the Stanford ORBIS data set. Alternatively, water transportation was much faster than overland routes. Furthermore, long-distance trade implied a more intense commercial exchange, while political borders and religious periods such as Lent and Advent decreased the amount of trade.27 Furthermore, controlling for the estimated trade effects of the identified variables, we find additional geographical effects related to specific cities. In particular, strongly export-focused trade hubs show a faster transmission of the plague. For instance, we can document this for the city of Bordeaux, with a strong export for wine, or Bergen for fish. In contrast, cities with a strong excess of imports during the period of investigation, such as Alexandria, encountered the plague significantly faster. Finally, we show that the population growth28 of cities during the centuries following the Black Death is correlated with the speed of transmission. These findings are in line with the idea that the period of the Commercial Revolution is relevant for long-run growth.  Pryor, Geography, Technology, and War.  Controlling for temperature did not have any significant effect on the speed of transmission of the Black Death. 28  Population size has been established as a proxy variable for economic growth in the literature on pre-modern growth; cf. Acemoglu, Johnson, and Robinson, “The Rise of Europe”. 26 27

338 

L. BÖRNER AND B. SEVERGNINI

3   The Justinianic Plague 3.1  The Spread of the Justinianic Plague Before we get into a detailed description and analysis of the Justinianic Plague, we must note that the source material and information we can derive is much smaller than for the Black Death. Many insights must be indirectly derived, and some precise local dates of the outbreak are still debated.29 Contemporary witnesses inform us about the spread of the plague all over the Mediterranean.30 However, the detailed paths of infection must be carefully reconstructed based on the existing fragmented source evidence. The most comprehensive analysis has been made by Stathakopoulos.31 We rely on his insights. Additional or alternative dates referring to other sources are indicated in the following section. Figure 10.2 sketches the outbreak of Justinianic Plague. It shows the routes of the spread and marks them for each year in different colours. The outbreak of the plague was first documented in the town of Pelusium in the middle of July in 541. Pelusium was likely infected from Southern Egypt, although no exact place of origin can be assigned.32 The disease spread farther along the coast in two directions, to the East and West. To the East, the plague arrived in Gaza along the sea trade route in the middle of August. From Gaza, the disease spread farther inland into various directions: to Nessana at the end of October, Rehovot at the beginning of November, and Eboda in the middle of December. From Rehovot, it travelled to Jerusalem, where the plague broke out at the end of February or beginning of March 542 and Zora around the 21st of March. Along the coast, the disease travelled from Gaza to Antiochia, where it must have arrived in early spring. Again, from here, we can 29  A research programme is urgently needed that could discover more locations and reveal information about the dates of the outbreak of the Justinianic Plague, as proposed by McCormick, “Toward a Molecular History of the Justinian Pandemic”. 30  Stathakopoulos, Famine and Pestilence in the Late Roman and Early Byzantine Empire; Horden, “Mediterranean Plague in The Age of Justinian”. 31  Stathakopoulos, Ibid. 32  The port of Clysma, located at the Red Sea, has been identified as a possible gateway for the plague. (Tsiamis, Poulakou-Rebelakou, and Petridou, “The Red Sea and the Port of Clysma.”) Such a path is likely because Pelusium was infected before Alexandria. This indicates that the disease did not travel along the Nile but came via the Red Sea and likely from Asia. For a broader discussion on the origin with a different perspective (and a likely origin from Africa), see Sarris, “Integration and Disintegration in the Late Roman Economy”.

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

339

Fig. 10.2  The spread of the Justinianic Plague (Red = 541; Green = 542; Blue = 543; Orange = 544–547)

­ ocument a spread farther along the commercial land and river routes to d Apamea, Epiphanea, and Emesa. The outbreak of the disease along this route can be documented for June. Along the coast, the disease spread to Myra in spring 542.33 From Pelusium, the disease spread to the West, to Alexandria, the largest regional harbour and hub in the Eastern Mediterranean, in the middle of September 541. From Alexandria, the disease crossed the Eastern Mediterranean to the north to Constantinopolis (potentially passing Myra), arriving at the end of March or beginning of April in 542. From Constantinopolis, the plague travelled inland, spreading to Central Anatolia, where it can be documented for Sykia in early summer of 542.34 Furthermore, the disease spread along the sea trade routes crossing the Aegaean and Adriatic, reaching Illyricum in 543. From Alexandria, the plague continued to move West, arriving first in Sicily, which was the gateway to the Western Mediterranean. The disease 33   Alternatively, Myra was directly infected via the ship route from Alexandria to Constantinopolis; cf. Stathakopoulos, Famine and Pestilence in the Late Roman and Early Byzantine Empire. 34  To locate the lost city of Sykia, we follow recent findings by Barchard, “Sykeon Rediscovered”.

340 

L. BÖRNER AND B. SEVERGNINI

can first be documented in the Sicilian town of Mazara by the end of December. Ships traveling from Alexandria to Sicily first had to pass Messana. Thus, we can assume (and reconstruct) that the plague likely broke out in Messana already at the beginning of December. From Sicily, the disease spread to North Africa, where the plague can first be documented in the town of Sufetula by the end of January 543. It also travelled to Italy, but probably not before spring 543.35 From Northern Africa (likely from Carthage), which was still the granary of the European Western Mediterranean, the plague spread to Rome, where the outbreak is dated to November 543, then to Southern France in 543, and to Spain, probably also in 543.36 Whereas we are well informed about the time of the outbreak of the plague in Rome, we only know the year 543 for France, and we can estimate the same year for Spain.37 In addition, there is evidence that the plague arrived in Ireland in 54438 and may have made its way from there to Wales in 547.39 This description covers the first outbreak of the Justinianic Plague. Over the next few decades, several additional rounds of spreads in Europe and the Mediterranean can be documented.40 However, for this comparative analysis, we only cover the first outbreak. 35  Stathakopoulos, Famine and Pestilence in the Late Roman and Early Byzantine Empire states that major sources about the Southern Italian city of Naples do not reveal any infection by spring 543. Thus, the disease likely spread to Southern Italy by summer 543. Thus, we can take Naples as a proxy variable for an infection. 36  The plague must have arrived along the most important sea trade routes on the French and Spanish coast. We have no information identifying at which port the disease first broke out. To estimate the speed of the spread, we take two of the most important harbours at the time, Massalia and Tarraco, as proxies. 37  Since it is not entirely clear from the source material whether the disease arrived in Spain in 542 or 543, Kulikowski tends to estimate the outbreak of the plague already for the end of 542 due to strong commercial ties between the Eastern and Western Mediterranean (Kulikowski, “Plague in Spanish Late Antiquity”). However, given that the spread only seems to have arrived in Sicily by the end of December and Tunisia by January 543, it seems reasonable to assume the disease arrived not before spring 543 (similar to Southern Italy) on the Western European Mediterranean coast. 38  According to Dooley, “The Plague and Its Consequences in Ireland”, the plague spread from Southern France via Narbonne along the river Garonne to Bordeaux to Ireland. The ORBIS data set does not report any direct connection from Bordeaux to Ireland. We assume that the traders followed the coastline up to the Bretagne and then crossed the Atlantic to Ireland. 39  Maddicott, “Plague in Seventh-Century England”. 40  Stathakopoulos, Famine and Pestilence in the Late Roman and Early Byzantine Empire.

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

341

3.2  What Determined the Spread of the Justinianic Plague? Again, we can ask what drove the spread of the Justinianic Plague. Whereas past scholars could only assume, based on reports by contemporary witnesses and on the behaviour of the spread of the plague, that the diseases was a Bubonic plague,41 recent scholars have identified the bacillus Yersinia pestis as the cause of the plague based on DNA analysis.42 The carriers of the bacillus must have also been fleas and rats that travelled with merchants on their goods. Otherwise, they would not have been able to spread over long distances. Thus, the Roman transportation technology and trade was crucial for the diffusion all across the Mediterranean.43 The claim that Justinianic Plague was mainly disseminated by merchant ships and caravans along the major trading routes has been made by various scholars working on the topic.44 However, we must raise the question of whether we need to consider other social movements during the spread of the disease. We are unaware of any strong migration flows or any established pilgrim routes, as during the late Middle Ages. However, we must consider the role of military movements during the period of the spread.45 The first wave of reconquering the Western Empire by Justinian occurred before the outbreak of the plague.46 Nevertheless, we can assume that major soldier movements occurred during this period, particularly in the Western Mediterranean, to protect the regained territories and borders of the Empire. In particular, the military expedition by Totila, the king of the Ostrogoths, which started in 542 in Northern Italy and reached Naples by spring 543,47 must have had an impact on social and military movements. Thus, the spread of the plague on the Italian Peninsula in 543 may have been affected by war activities during this year. Finally, we can raise the question of whether the Romans understood the causes of the spread of the plague and whether there was a way to stop or slow it down. Also, during Antiquity, physicians did not know how to  Horden, “Mediterranean Plague in The Age of Justinian”; Sallares, “Ecology”.  Harbeck et al., “Yersinia Pestis DNA from Skeletal Remains from the 6th Century AD”. 43  McCormick, “Rats, Communications, and Plague”; Rosen, Justinian’s Flea, 185–86. 44  Stathakopoulos, Famine and Pestilence in the Late Roman and Early Byzantine Empire, 136; Orent, Plague; Horden, “Mediterranean Plague in The Age of Justinian”; Rosen, Justinian’s Flea, 185–86. 45   Stathakopoulos, Famine and Pestilence in the Late Roman and Early Byzantine Empire, 136. 46  Boss, Justinian’s Wars. 47  Jacobsen, The Gothic War. 41 42

342 

L. BÖRNER AND B. SEVERGNINI

treat or prevent a Bubonic Plague.48 The arrival of the plague was rather seen as a supernatural event.49 Justinian himself declared in 542 the end of the plague as the end of a punishment of God.50

4   A Comparative Perspective 4.1  The Determinants of the Justinianic Plague, a Quantitative Perspective Looking at the spread of the Justinianic Plague, we can investigate whether we can identify a similar pattern of determinants as detected in the spread of the Black Death. Since we do not have many observations, we cannot make a profound quantitative study. Instead, we must approximate and qualitatively interpret the dates of the outbreaks and routes of the spread. To build a rough data set from which we can extract simple descriptive statistics and create a comparative perspective with the empirical results of the spread of the Black Death, we again collect city pairs. We measure the average speed of infection between cities and regions in kilometres per day. To measure the distance between two destinations, we again use the Stanford ORBIS data set to measure the effective length of the fastest trading route. As in the previous study, we split the month into four units and match the date of outbreak with one of these four slots. In case we only have rough estimates about the outbreak (for instance the month), we assume it happens in the middle of the month. Finally, in case we only know the year, we only add this information to a separate data set, where we incorporate these very vague data as well in order not to bias the output of the more precise data set we already have. In this latter case, we assume again that the outbreak happened in the middle of the year, but we also check alternative specifications, that is, an outbreak at the beginning or at the end of the specific year. The data collection gives us approximately eighteen “rather precise” city pairs of observations and another five rather speculative ones. Thus, we obtain a total of twenty-three observation points. However, we will only work with twenty-two observations, since the city pair that describes  Rosen, Justinian’s Flea, 211–12.  Rosen, 216–217; Stathakopoulos, Famine and Pestilence in the Late Roman and Early Byzantine Empire, 146–54; Stathakopoulos, “Crime and Punishment”. 50  Horden, “Mediterranean Plague in The Age of Justinian”. 48 49

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

343

the connection from Ireland to Wales contains two speculative dates. Therefore, we skip this observation. These remaining twenty-two observations are compared to the data set describing the spread of the Black Death, where we can rely on more than 200 city pairs as observation points. In a first step, we investigate whether the main determinants of the spread as identified in the study of the Black Death can be linked to the spread of the Justinianic Plague. A main insight is that the speed of the spread depends on the transportation technology and related trade geography of the time. One simple way to test this is to separate land from sea trade routes and check whether the average speed of transmission is faster on sea than on land.51 Indeed, we find that the average speed on land is much slower (3.6 km per day) than along the sea trade route (7.4 km per day). A more sophisticated approach is to study whether the spread of the disease correlates with the average speed of transportation following the Stanford ORBIS data set. Again, we find support for such a relationship. We find a significant positive coefficient.52 This supports the claim that the spread was also promoted by the trade or at least the trade transportation technology of the time. A second important determinant of the spread of the Black Death was political borders. Measuring political border effects during the spread of the Justinianic plague is challenging for two reasons. First, the area of investigation where we have information about the spread of the plague is mainly covered by Justinian’s Empire. Thus, we have only a few data points with borders we could investigate. We have a “grey” area in the Western Mediterranean, which was reconquered by Justinian’s army and could be characterized by political instability and unclear border lines. Second, we have strong measurement uncertainty for data points outside or at the border of the Justinian Empire. For example, for the observations for the Italian Peninsula, France, Spain, and Ireland, we only have yearly data. Thus, we must take this into account when investigating the effects of political borders. Despite these reservations, we can bundle city pairs, in which both observation points of a city pair fall within the area of the established 51  Due to the few observations we have, we bundle the land and the river trade routes together. The river routes in the Black Death data set are slightly faster than the land trade routes but much slower than the sea trade routes. 52  The coefficient is 0.33 and is significant at a 5 per cent confidence level. This is an acceptable result, given the rough data and the small number of observations.

344 

L. BÖRNER AND B. SEVERGNINI

Eastern Empire or alternatively in this grey area. Next, we can empirically test whether the spread in one area is significantly faster than in the other. Such an analysis becomes, then, mainly a comparison between the established Eastern Mediterranean Empire and Western Europe. To derive statistical results, we run a t-test where we compare the average speed of city pairs belonging to the established Eastern Mediterranean Empire with city pairs where at least one city is outside the Empire or belongs to the newly regained territory. However, we do not find any significant difference in the speed of transmission.53 These results are interesting, since they contribute to a discussion that goes beyond the question of political borders for the area and period of investigation. These findings touch on the question of how economically (and socially) disintegrated the Western Mediterranean was during late Antiquity. Older studies have argued that based on the political instability, trade must have been seriously interrupted.54 However, recent investigations have shown that trade in the Western Mediterranean, particularly among North Africa, Spain, France, and Italy, was still intensive.55 Thus, our findings support this line of reasoning; political borders or insecurity along the borders of the reconquered area did not seem to have hindered the spread of the disease, which was strongly driven by economic activities. However, we must end this discussion with the remark that a potentially higher activity of soldiers in this area could have increased the social interaction, and this could have compensated for a reduction in trade activities. 4.2  A Comparative Perspective Over Time Taking this analysis one step further, we can ask whether we can identify differences in the speed of the spread of the Justinianic Plague in comparison with the speed of the diffusion of the Black Death and analyse whether such differences could be related to different intensities of social and economic interactions over time. 53  These results are based on a conservative estimate that the plague only arrived in the middle of 543 in Spain and France. If we replace these numbers with an arrival already by the end of 542 or the beginning of 543, we might expect an even faster transmission in this politically unstable area. 54  For instance Rostovtzeff, The Social and Economic History of the Roman Empire. 55  Kingsley and Decker, “New Rome”, 3–4.

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

345

To be able to do this, we must first discuss the similarities of both diseases and rule out that the two plagues were spread by different types of carriers. As already discussed in the previous sections, both plagues were transmitted by the same type of bacterium.56 The disease, in turn, could only be carried by social movements between the cities and regions. The transportation technology available was the same during both epochs. The correlation between the speed of the spread of the plague and the time to travel based on the existing trade technology (found for both the Black Death and the Justinianic Plague) indeed supports such a view. This transportation technology was mainly used for trade. Other strong social movements in the form of migration or pilgrims can also be excluded for the period of the Justinianic Plague. However, there may have been major soldier movements during the period, particularly in the Western part of the Justinian Empire, to preserve the borders of the regained territories. However, in the Eastern Mediterranean, we should expect a rather safe area where trade between the different regions could flow. Since we cannot observe any significant differences in the speed of transmission between the Eastern and Western Mediterranean, as shown before, it is reasonable to use the aggregate average speed of transmission of Justinian’s Plague, compare it with the speed of the transmission of the Black Death, and discuss and derive potential differences in economic interaction over time.57 Comparing the spread of both diseases, we find that the diseases broke out in different corners of the Eastern Mediterranean basin, but the contamination took place in the whole area.58 Thus, we can ask how long it 56  However, we cannot exclude that the Yersinia pestis mutated or evolved during this period of time in such a way that it had an (so far unobserved) impact on the speed or form of spread of the infection. For a recent discussion of the evolution of the bacterium, see Demeure et al., “Yersinia Pestis and Plague”. 57  A final comparative remark concerning the climate should be made: We previously noticed that temperature did not have an impact on the speed of transmission in the study of the Black Death. However, we must mention that the spread of the Justinian Plague falls into the coldest decade measured during the last 2000 years (for instance see the data collection by Luterbacher et al. “Review of 2000 Years of Paleoclimatic Evidence). Consequently, we might need to take this into account. Unfortunately, we are unable to quantify and compare this in a meaningful way given the few observations we have for the Justinian Plague. 58  Since we identified different geographical starting points of the spread of the disease the reader might ask if this had any effect on the routes infected and estimated determinants of the speed of the spread, that is, due to the path-dependent order of the infection. Studying the spread of the Black Death we could not identify such effects: Neither could we find any

346 

L. BÖRNER AND B. SEVERGNINI

took to spread from the North to the South (or vice versa) and from the Eastern to the Western Mediterranean corner. Comparing these numbers, we find that it took roughly eight months from the South (Pelusium in July) to the North (Constantinople in April) to cross the Eastern Mediterranean during the Justinianic Plague.59 In contrast, it took only two months for the Black Death to travel from the North (Constantinople in June 1347) to the South (Alexandria in August 1347). Similarly, we find differences in the spread from the Eastern Mediterranean to the West. It took only two months for the Black Death to reach Sicily (August 1347) from Constantinople, and it took nine months to reach Valencia at the Western end of the Mediterranean Basin. On the contrary, it took Justinianic Plague approximately seventeen months to travel from Pelusium to Sicily (December 542) and it did not spread to the Western Mediterranean coast until spring or summer 543, so nearly two years. Finally, if we add Ireland to this comparative analysis so that we have a geographical observation from North West Europe, we find similar differences. Whereas the Black Death took approximately ten months to reach Ireland from Sicily, the Justinianic Plague took approximately 1.5  years (assuming the plague arrived in Ireland in the middle of the year). These differences are also reflected in the average speed of infection, especially along the sea trade routes. The average speed of spread along land routes is comparable, with an average speed of 4.5 km per day during the Black Death and 3.6 km per day during the Justinianic Plague. Along sea trade routes, we find an average speed of 10.6 km per day during the Black Death and 7.4 km per day during the Justinianic Plague. These differences are more pronounced if we only look at the average speed of dispersion in the Mediterranean. In the period of the Black Death, the average speed of transmission on Mediterranean Sea trade routes was about 12.6 km per day.

specific effects in the estimates or residuals of the regression analysis nor based on an empirical network investigation we just started in a new project. Furthermore, there is no evidence that specific trade routes were infected and others not. The starting point only had an impact on the timing of the infection not the selection. 59  If we replace Pelusium with Alexandria as the major hub in the South Eastern Mediterranean, the infection would still last six months.

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

347

4.3  A Comparative Discussion of the Differences in Speed of Transmission Following our methodology, we want to discuss whether we can place these results in a meaningful historical comparative context, especially in a comparative economic perspective. A first point that must be discussed is the question whether both economies were market economies, in the sense that the allocation of products was determined by specialization in production and exchange of goods across different towns and regions in the Mediterranean. Such a system is obviously driven by trade (and related social and economic interaction). Thus, a hypothesis that can be discussed is whether the period of the Commercial Revolution was more characteristic of a market economy than the Roman Empire during the Justinianic era and, consequently, whether this is the reason why the disease spread much slower. As already outlined, the period of the Commercial Revolution is described or even defined as a phase of intensification of trade. The frequent interregional bulk trade and exchange of luxurious products has been widely documented.60 In contrast, the structure of the Roman economy has been discussed more controversially. A traditional line of research has argued that the Roman Empire rather consisted of local economies and that the interregional trade was very limited and mainly organized by the state to feed the local populations of the largest cities, such as Rome.61 More recent research, however, argues that we can find frequent interregional trade and specialization of production in different regions and cities.62 For instance, grain was produced and exported from Egypt and Tunisia, and oil and wine were produced and shipped from Italy and Spain to the rest of the Empire.63 Also, recent studies on market integration in the Roman Empire have conjectured that price formation in different regions was driven by an underlying market mechanism.64 Evidence of specialization and interregional trade can also still be documented for late Antiquity: in the Eastern Mediterranean in the triangle of Egypt, ­Palestine/ 60  Lopez, The Commercial Revolution; Freedman, Out of the East; Abulafia, The Mediterranean in History. 61  Finley, The Ancient Economy. 62  Kehoe, “The Early Roman Empire: Production”, finds such regional specialization of production (for instance, ceramics, linen, and wool) already during the early Roman Empire. 63  Bowman and Wilson, The Roman Agricultural Economy. 64  Temin, The Roman Market Economy.

348 

L. BÖRNER AND B. SEVERGNINI

Syria, and Istanbul; in the Western Mediterranean between North Africa and the European Mediterranean; and, finally, between the Eastern and Western Mediterranean.65 Consequently, an attempt to explain the difference in the speed of the transmission based on the differences in the basic functioning of the economy is unconvincing. An alternative explanation might not lie in the differences of economic systems but rather in the trade intensity between the cities and regions. An indicator of such trade intensities could be the variation and difference in the population size and related urbanisation rates. An increase in the urbanisation rates typically indicates an increase in trade and specialization in production. Cities must import basic foodstuffs for consumption and raw and semi-finished products for local production, and they must export processed and final products. An increase in urbanisation in pre-modern societies is typically not only a sign of increase in trade but also of wealth in general. However, looking at the urbanisation rates of both periods, we are left with a puzzle. Comparative demographic studies have found that the urbanisation rates in the Mediterranean were comparable during the Commercial Revolution and the Ancient Roman Empire.66 Other studies suggest that the urbanisation rates during the Roman Empire were even higher and comparable with the rates in early Modern Europe.67 In line with such comparable urbanisation rates are similar GDP estimates for both epochs by Lo Cascio and Malanima.68 However, Scheidel argues that standards of living must have been clearly lower during the Roman Empire. He argues that Roman society was trapped in a low-level Malthusian equilibrium, where real wages were determined rather by population size than economic development.69 Such a line of argumentation would fit with our findings of a much slower transmission of the Justinianic Plague, which would be in line with a lower trade intensity and explain the lower standards of living during the Roman Empire. In contrast, the Commercial Revolution can be described by higher trade intensities and

65   Kingsley and Decker, Economy and Exchange; Decker, “Food”; Ward-Perkins, “Specialisation”; Mundell Mango, “Beyond the Amphora: Non-Ceramic Evidence for Late Antique Industry and Trade”; McCormick, Origins of the European Economy. 66  Scheidel, “Demography”, 80. 67  Wilson, “City Sizes”. 68  Lo Cascio and Malanima, “Ancient and Pre-Modern Economies” GDP”. 69  Scheidel, “Demography”, 55–56; Scheidel, “Roman Real Wages”.

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

349

higher standards of living with the potential for further economic growth and development. Such an interpretation brings us to a more dynamic perspective and interpretation of these differences in speed and likely differences in trade intensities and economic welfare. Whereas the period of the Commercial Revolution indeed stands for a phase of economic growth and prosperity, the reign of Justinian I represents the end of the glory of the Roman Empire. Even if it stands for a recent surge, it might not have reached the economic strength of the Roman Republic or the Roman Empire. In this way, a decrease in trade activities might simply just show a decline of the strength of the Roman Empire. The slower transmission of the plague can then be matched to a de-population and de-urbanisation during late Antiquity that was motivated by an increasing need for the agricultural self-sufficiency of the population (and a general decline in trade and the market economy).70 Nevertheless, Justinian’s empire represented a sophisticated economy with many very advanced institutional properties. Therefore, in a final discussion, we must investigate how favourable the Justinianic period was compared to the period of the Commercial Revolution with regard to trade-supporting or trade-restricting institutions. Modern empirical trade studies have identified key factors. Political borders, as previously discussed, are one potential determinant of trade, but there are many others. In particular, common economic policy, legislation, and cultural variables have been identified as trade-supporting institutions. Comparing these favourable variables with both institutional regimes, we should expect that Justinian’s Empire performed very well with respect to trade. Not only can we document a large territorial empire (without any borders), but Justinian I also realized the codification of the Roman law and implemented it throughout the Empire to simplify jurisdiction and improve legal security.71 He enforced a currency minting monopoly, with 70  The list of theories for such demographic and sectorial shifts is long. In particular, institutional failures that led to a decline of the state have been prominently discussed. For instance, see Rostovtzeff, The Social and Economic History of the Roman Empire. Exogenous shocks due to ongoing migration flows and war activities (Goldsworthy, The Fall of the West) or permanent arrivals of diseases since the Antonine Plague starting in AD 165 have also been discussed. (McNeill, Plagues and Peoples; Erdkamp, “Urbanism”, sees the de-urbanisation as more related to a social and political shift away from the city to the countryside than to economic factors.) 71  Rosen, Justinian’s Flea, 119–20.

350 

L. BÖRNER AND B. SEVERGNINI

the gold coins being accepted and used by traders beyond the Empire.72 The stable currency guaranteed security, particularly for wholesale trade. Finally, Justinian I showed considerable political and military strength, which supported the security along the trade routes and exchange on the markets in the Empire. Such forms of economic and political unification cannot be documented for the period of the Commercial Revolution in the Mediterranean. For instance, political and legal fragmentation was the rule rather than the exception. However, the Justinianic period also documents the effort to establish monopolies,73 for instance, for silk trade,74 strong restrictions for foreign merchants on local markets, the right for local trade guilds to fix prices, and a ban on certain products being sold and exported to markets outside the Empire.75 Thus, Justinian’s empire can be also interpreted as an overregulated state with many restrictions that hindered free markets and competition in the Mediterranean. Consequently, the competition among many merchant groups from various states and city states during the Commercial Revolution, with Genoa and Venice at the forefront, might have supported a much stronger trade intensity in the Mediterranean several centuries later.76

5   Conclusion This chapter studied the spread of plagues from a comparative perspective. It investigated the determinants of the speed of the spread of the Black Death and the Justinianic Plague. In addition, it compared the speed of infection during both periods of time. We found that the Justinianic Plague followed a similar pattern as the Black Death, which we have already investigated in previous work. In particular, we detected that the speed of transmission between two destinations along a trade route is indeed determined by the trade technology and trade geography, which is the physical time to travel between two destinations following the ORBIS data set. This supports claims made by other scholarly work that argues that both diseases were spread by human interaction, particularly through 72  Evans, The Age of Justinian, 236; Hendy, Studies in the Byzantine Monetary Economy, 398–99. 73  Bury, History of the Later Roman Empire, 356–57. 74  Baker, Justinian, 320–21. 75  Evans, The Age of Justinian, 226–27, 236; Stantchev, Spiritual Rationality, 24. 76  Lane, Venice; Epstein, Genoa & the Genoese.

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

351

trade activities. Furthermore, we found that the Black Death spread approximately twice as fast as the Justinianic Plague in the Mediterranean. This result indicates that given the existence of the same trade technology and spread of the same type of bacterium that the intensity of social and, in particular, economic interaction between cities and regions was much more intensive during the late Medieval period than during the reign of Justinian. This can be related to differences in levels of economic activity between the period of the Commercial Revolution and the Roman Empire. Alternatively, these results can be interpreted as economic snapshots of different time dynamics during the two epochs: on the one hand, the Justinianic period in a phase of decline of the Roman Empire, with fewer trade activities and a stronger focus of local and agricultural activities, and on the other hand, the Commercial Revolution, a period of economic growth and an increase in trade activities that may have led to long-run growth, representing the origin of the rise of Western Europe. However, in conclusion, we must stress that the analysis of the Justinianic Plague relies on very fragmented data. More information on the dates of the spreads in different destinations would allow us to make stronger analytical statements and derive comparative insights.

Bibliography Abulafia, David. The Mediterranean in History. Los Angeles: J.  Paul Getty Museum, 2011. Acemoglu, Daron, Simon Johnson, and James Robinson. “The Rise of Europe: Atlantic Trade, Institutional Change, and Economic Growth”. American Economic Review 95 (2005): 546–79. Aldrete, Gregory S., and Alicia Aldrete. The Long Shadow of Antiquity: What Have the Greeks and Romans Done for Us? London: Continuum, 2012. Anderson, James E., and Eric van Wincoop. “Gravity with Gravitas: A Solution to the Border Puzzle”. American Economic Review 93 (March 2003): 170–92. Armington, Paul S. “A Theory of Demand for Products Distinguished by Place of Production”. Staff Papers (International Monetary Fund) 16 (1969): 159–78. Baker, G. P. Justinian: The Last Roman Emperor. New York, NY: Cooper Square Press, 2002. Barchard, David. “Sykeon Rediscovered? A Site at Kiliseler Near Beypazarı”. Anatolian Studies 53 (2003): 175–79. Benedictow, Ole Jørgen. The Black Death, 1346–1353: The Complete History. Woodbridge, Suffolk, UK; Rochester, N.Y., USA: Boydell Press, 2004.

352 

L. BÖRNER AND B. SEVERGNINI

Boerner, Lars, and Battista Severgnini. “Epidemic Trade”. London School of Economics Political Science, Economic History Working Paper 212, 2014. Bos, Kirsten I., Verena J. Schuenemann, G. Brian Golding, Hernán A. Burbano, Nicholas Waglechner, Brian K. Coombes, Joseph B. McPhee, et al. “A Draft Genome of Yersinia Pestis from Victims of the Black Death”. Nature 478, no. 7370 (12 October 2011): 506–10. Boss, Roy. Justinian’s Wars: Belisarius, Narses and the Reconquest of the West. Stockport: Montvert, 1993. Bowman, Alan K, and Andrew Wilson, eds. The Roman Agricultural Economy: Organization, Investment, and Production. Oxford Studies on the Roman Economy 3. Oxford: Oxford University Press, 2013. Bridbury, A.  R. “Before the Black Death”. The Economic History Review 30 (1977): 393–410. Brown, Peter R. L. Poverty and Leadership in the Later Roman Empire. Hanover: University Press of New England, 2002. Brown, Peter. The Making of Late Antiquity. Cambridge (Mass.): Harvard University Press, 1978. Burckhardt, Jacob. Die Zeit Constantins des Großen. Basel: Schweighauser, 1853. Burdick, William L. The Principles of Roman Law and Their Relation to Modern Law. Clark: The Lawbook Exchange Ltd., 2004. Bury, J.B. History of the Later Roman Empire from the Death of Theodosius Ii to the Death of Justinian. New York: Dover, 1958. Callaway, Ewen. “Plague Genome: The Black Death Decoded”. Nature News 478, no. 7370 (25 October 2011): 444–46. Cipolla, “The Plague and the Pre-Malthus Malthusians.” Journal of European Economic History 3 (1974): 277–84. Cipolla, Carlo M. Before the Industrial Revolution: European Society and Economy, 1000–1700. London: Routledge, 1993. Cohn, Samuel K, Jr. “The Black Death: End of a Paradigm”. The American Historical Review 107 (2002): 703–38. Decker, Michael. “Food for an Empire: Wine and Oil Production in North Syria”. In Economy and Exchange in the East Mediterranean During Late Antiquity. Proceedings of a Conference at Somerville College, Oxford, 29th May, 1999, edited by Sean Kingsley and Michael Decker, 69–86. Oxford: Oxbow Books, 2001. Demeure, Christian E. et  al. “Yersinia Pestis and Plague: an Updated View on Evolution, Virulence Determinants, Immune Subversion, Vaccination, and Diagnostics”, Nature: Genes and Immunity 20 (2019): 357–70. Domar, Evsey D. “The Causes of Slavery or Serfdom: A Hypothesis”. The Journal of Economic History 30 (1970): 18–32.

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

353

Dooley, Ann. “The Plague and Its Consequences in Ireland”. In Plague and the End of Antiquity: The Pandemic of 541–750, edited by Lester K. Little, 215–30. Cambridge: Cambridge University Press, 2007. Epstein, Stephan R. Freedom and Growth: Markets and States in Pre-Modern Europe. Routledge Explorations in Economic History. New  York: Routledge, 2000. Epstein, Steven A. Genoa & the Genoese: 958–1528. Chapel Hill: University of North Carolina Press, 1996. Erdkamp, Paul. “Urbanism”. In The Cambridge Companion to the Roman Economy, edited by Walter Scheidel and Walter Scheidel, 241–65. Cambridge: Cambridge University Press, 2012. Evans, J. A. S. The Age of Justinian: The Circumstances of Imperial Power. London: Routledge, 2002. Finley, M. I. The Ancient Economy. Updated with a New Foreword by Ian Morris. Berkeley: University of California Press, 1999. Freedman, Paul. Out of the East: Spices and the Medieval Imagination. New Haven: Yale University Press, 2008. Gibbon, Edward. The History of the Decline and Fall of the Roman Empire. London: W. Strahan and T. Cadell, 1776. Goldsworthy, Adrian K. The Fall of the West: The Death of a Roman Superpower. London: Hachette, 2009. Greif, Avner. Institutions and the Path to the Modern Economy: Lessons from Medieval Trade. Cambridge; New York: Cambridge University Press, 2006. Harbeck, Michaela, Lisa Seifert, Stephanie Hänsch, David M.  Wagner, Dawn Birdsell, Katy L. Parise, Ingrid Wiechmann, et al. “Yersinia Pestis DNA from Skeletal Remains from the 6th Century AD Reveals Insights into Justinianic Plague”. PLOS Pathogens 9, no. 5 (2 May 2013): e1003349. Hendy, Michael F. Studies in the Byzantine Monetary Economy, C. 300–1450. Cambridge; New York: Cambridge University Press, 2008. Herlihy, David, and Samuel Kline Cohn. The Black Death and the Transformation of the West. Cambridge, Mass.: Harvard University Press, 1985. Horden, Peregrine. “Mediterranean Plague in The Age of Justinian”. In The Cambridge Companion to the Age of Justinian, edited by Michael Maas, 134–60. Cambridge: Cambridge University Press, 2005. Jacobsen, Torsten Cumberland. The Gothic War: Justinian’s Campaign to Reclaim Italy. Yardeley, PA: Westholme, 2009. Kehoe, Dennis P. “The Early Roman Empire: Production”. In The Cambridge Economic History of the Greco-Roman World, edited by Walter Scheidel, Ian Morris, and Richard P.  Saller, 541–69. Cambridge; New  York: Cambridge University Press, 2007.

354 

L. BÖRNER AND B. SEVERGNINI

Kingsley, Sean A., and Michael Decker, eds. Economy and Exchange in the East Mediterranean During Late Antiquity. Proceedings of a Conference at Somerville College, Oxford, 29th May, 1999. Oxford: Oxbow Books, 2001a. Kingsley, Sean A., and Michael Decker. “New Rome, New Theories on Inter-­ Regional Exchange. An Introduction to the Eastern Mediterranean Economy in the Late Antiquity”. In Economy and Exchange in the East Mediterranean During Late Antiquity. Proceedings of a Conference at Somerville College, Oxford, 29th May, 1999, edited by Sean A. Kingsley and Michael Decker, 1–27. Oxford: Oxbow Books, 2001b. Kulikowski, Michael. “Plague in Spanish Late Antiquity”. In Plague and the End of Antiquity: The Pandemic of 541–750, edited by Lester K.  Little, 150–70. Cambridge: Cambridge University Press, 2007. Lane, Frederic C. Venice: A Maritime Republic. Baltimore: John Hopkins University Press, 1973. Livi Bacci, Massimo. A Concise History of World Population. Cambridge, MA: Blackwell, 1997. Livi-Bacci, Massimo. “The Nutrition-Mortality Link in Past Times: A Comment”. The Journal of Interdisciplinary History 14 (1983): 293–98. Lo Cascio, Elio, and Paolo Malanima. “Ancient and Pre-Modern Economies’ GDP in the Roman Empire and Early Modern Europe”. In Quantifying the Greco-Roman Economy and Beyond, edited by François de Callataÿ, 229–51. Pragmateiai 27. Bari: Edipuglia, 2014. Lopez, Robert S. The Commercial Revolution of the Middle Ages, 950–1350. Cambridge; New York: Cambridge University Press, 1976. Luterbacher, Jürg et al. “A Review of 2000 Years of Paleoclimatic Evidence in the Mediterranean”. In The Climate of the Mediterranean Region, edited by Piero Lionello, 87–185. London: Elsevier, 2012. Maddicott, John. “Plague in Seventh-Century England”. In Plague and the End of Antiquity: The Pandemic of 541–750, edited by Lester K.  Little, 171–214. Cambridge: Cambridge University Press, 2007. Mazzarino, Santo. La fine del mondo antico. Milano: Garzanti, 1959. McCormick, Michael. “Rats, Communications, and Plague: Toward an Ecological History”. The Journal of Interdisciplinary History 34 (2003): 1–25. McCormick, Michael. Origins of the European Economy: Communications and Commerce, A.D. 300–900. Cambridge: Cambridge University Press, 2001. McCormick, Michael. “Toward a Molecular History of the Justinian Pandemic”. In Plague and the End of Antiquity: The Pandemic of 541–750, edited by Lester K. Little, 290–312. Cambridge: Cambridge University Press, 2007. McNeill, William Hardy. Plagues and Peoples. New York: Peter Smith, 1976. Michaels, Guy, and Ferdinand Rauch. “Resetting the Urban Network: 117–2012”. The Economic Journal 127 (2017).

10  MEASURING AND COMPARING ECONOMIC INTERACTION BASED… 

355

Mundell Mango, Maria. “Beyond the Amphora: Non-Ceramic Evidence for Late Antique Industry and Trade”. In Economy and Exchange in the East Mediterranean During Late Antiquity. Proceedings of a Conference at Somerville College, Oxford, 29th May, 1999, edited by Sean A.  Kingsley and Michael Decker, 87–106. Oxford: Oxbow Books, 2001. Orent, Wendy. Plague: The Mysterious Past and Terrifying Future of the World’s Most Dangerous Disease. New York: Simon & Schuster, 2004. Pirenne, Henri. Medieval Cities: Their Origins and the Revival of Trade. Translated by Frank Davis Halsey. Princeton: Princeton University Press, 1925. Pollitzer, Robert. Plague: A Manual for Medical and Public Health Workers. World Health Organization Monograph Series 22. Geneva: World Health Organization, 1954. Pryor, John H. Geography, Technology, and War: Studies in the Maritime History of the Mediterranean, 649–1571. Cambridge: Univ. P, 1992. Rosen, William. Justinian’s Flea: Plague, Empire, and the Birth of Europe. New York: Viking, 2006. Rostovtzeff, Michael Ivanovitch. The Social and Economic History of the Roman Empire. Oxford: Clarendon Press, 1926. Sallares, Robert. “Ecology”. In The Cambridge Economic History of the Greco-­ Roman World, edited by Walter Scheidel, Ian Morris, and Richard Saller, 15–36. Cambridge: Cambridge University Press, 2007. Sarris, Peter. “Integration and Disintegration in the Late Roman Economy: The Role of Markets, Emperors, and Aristocrats”. Late Antique Archaeology 10, no. 1 (1 January 2013): 167–88. Scheidel, Walter. “Demography”. In The Cambridge Economic History of the Greco-­ Roman World, edited by Walter Scheidel, Ian Morris, and Richard P.  Saller, 38–86. Cambridge UK; New York: Cambridge University Press, 2007. Scheidel, Walter. “Real Wages in Early Economies: Evidence for Living Standards from 1800 BCE to 1300 CE”. Journal of the Economic and Social History of the Orient 53 (2010): 425–62. Scheidel, Walter. “Roman Real Wages in Context”. In Quantifying the Greco-­ Roman Economy and Beyond, edited by François de Callataÿ, 209–18. Bari: Edipuglia, 2014. Slack, Paul. “The Disappearance of Plague: An Alternative View”. The Economic History Review 34 (1981): 469–76. Stantchev, Krasimir Stefanov. Spiritual Rationality: Papal Embargo as Cultural Practice. Oxford: Oxford University Press, 2014. Stathakopoulos, Dionysios Ch. “Crime and Punishment: The Plague in the Byzantine Empire, 541–749”. In Plague and the End of Antiquity: The Pandemic of 541–750, edited by Lester K.  Little, 99–118. Cambridge: Cambridge University Press, 2007.

356 

L. BÖRNER AND B. SEVERGNINI

Stathakopoulos, Dionysios Ch. Famine and Pestilence in the Late Roman and Early Byzantine Empire: A Systematic Survey of Subsistence Crises and Epidemics. Aldershot: Ashgate, 2004. Temin, Peter. The Roman Market Economy. The Princeton Economic History of the Western World. Princeton, N.J.: Princeton University Press, 2013. Tsiamis, Costas, Effie Poulakou-Rebelakou, and Eleni Petridou. “The Red Sea and the Port of Clysma. A Possible Gate of Justinian’s Plague”. Gesnerus 66 (2009): 209–17. Ward-Perkins, Bryan. “Specialisation, Trade, and Prosperity: An Overview of the Economy of the Late Antique Eastern Mediterranean”. In Economy and Exchange in the East Mediterranean During Late Antiquity. Proceedings of a Conference at Somerville College, Oxford, 29th May, 1999, edited by Sean A. Kingsley and Michael Decker, 167–78. Oxford: Oxbow Books, 2001. Ward-Perkins, Bryan. The Fall of Rome and the End of Civilization. Oxford: Oxford University Press, 2006. Wilson, Andrew. “City Sizes and Urbanization in the Roman Empire”. In Settlement, Urbanization, and Population, edited by Alan K.  Bowman and Andrew Wilson, 161–95. Oxford: Oxford University Press, 2011.

Index1

A Actor-Network-Theory (ANT), 75 Adaptation, 37, 44, 167, 168, 191, 194 Aegean, 65, 67, 260 Aerial photography, 213, 223, 228–230, 232, 242 Africa, 311, 338n32 Agency, 25, 30–34, 75, 76, 252 Amphitheatre, 204, 211, 220, 225, 265 Anatolia, 258, 259, 276, 280 Ancona, 225, 226 Antonine Plague, 8, 15, 297–303, 305–306, 309, 310, 312–316, 318–320, 349n70 Arab peninsula, 71 Arabs, 21, 68 Armington trade model, 335 Asia, 14, 259, 266, 271, 273–276, 338n32

Assize (conventus), 273 Austrian economics, 130, 130n25 B Balkans, 68, 333 Bang, Peter Fibiger, 12, 62, 67, 113, 115 Banks / banking, 40, 126, 214, 309 Barygaza, 71 Bath bath-house, 266 bathing complex, 267 Behavioural Economics, 8 Berenike, 71 Betweenness, 45, 57, 60, 65, 70, 71, 79, 80 Bithynia, 263, 264 Black Death, 16, 74, 313, 327–351 Black markets, 308

 Note: Page numbers followed by ‘n’ refer to notes.

1

© The Author(s) 2021 K. Verboven (ed.), Complexity Economics, Palgrave Studies in Ancient Economies, https://doi.org/10.1007/978-3-030-47898-8

357

358 

INDEX

Britain, 38, 42 Bubonic plague, 331, 334, 334n19, 341, 342 C Caria, 260 Çatalhöyük, 76 Causal factors, 13, 164, 167–170, 172, 186, 187, 191, 193 Causal loop, 135–137, 141, 151–154 Central Europe, 59 Ceramics, 173–177, 274, 347n62 Chaos / chaos theory, 252, 271, 276 See also Complexity / complexity theory China, 21, 33, 75 Cilicia, 273, 287 Climate / climate change, 255, 300, 309, 320 Closeness, 57, 60, 65, 81, 82 Cloth, 70 Clustering coefficient, 57, 58, 61, 70, 71 Coinage, currency, 40 See also Money, quantity theory of Collapse, 63, 64, 64n40, 169, 252 Colonia, 258, 263 Commercial Revolution, 16, 328–332, 334, 337, 347–351 Comparative advantage, 7, 108 Comparative evidence, 13 Complexity / complexity theory, 33, 53–78, 127–138, 164, 165, 167, 169, 171, 188, 194, 251–277, 305 See also Chaos / chaos theory Complexity economics, 2, 3, 8, 11, 12, 16, 46, 53, 54, 72, 125–127, 129, 130, 163–195, 320 Complex system / complex adaptive system, 33, 38, 55, 125–154

Compliance (of social or formal rules), 30, 34, 36 See also Defection (of social or formal rules) Computational modelling, 11, 105–122, 320 See also Data modelling Congruence / incongruence (of formal and informal institutions), 10, 23 Connectivity, 11, 45, 45n48, 46, 60, 61, 63, 65–68, 73, 74n73, 166, 167, 170, 171, 273 Constantinople, 62n34, 312, 332, 346 Cosa, 205, 209–211 Cotton, 70, 70n62 Cultural beliefs, 2, 23, 26–28, 27n11, 30, 35, 46 D Data modelling, 2, 5 See also Computational modelling Defection (of social or formal rules), 35 See also Compliance (of social or formal rules) Degree (in network analysis), 60, 73 Demographic modelling, 204, 205 Demography, urban, 208, 242 See also Population, urban Density figures, 14, 203–205, 239 network, 46 population, 204, 205, 205n12, 207–211, 213, 239, 242, 243, 270n75 Depositum, 145, 148–152 Digital data, 4, 5 Dimensionality, 166, 167, 170, 171 Disease, 4, 37n31, 297–321, 329–335, 337–341, 338n32,

 INDEX 

340n35, 340n36, 340n37, 343–345, 345n58, 347, 349n70, 350 See also Epidemic; Pandemic; Plague Distribution (of goods), 107, 118 Diversity, 72n64, 128, 142, 150, 166, 167, 170, 184, 188, 191, 239, 252, 275, 276 Domestic economies, 190 Düzen Tepe, 13, 163–195 E Eastern Mediterranean, 45, 67, 68, 114, 339 Ecological colonization theory, 275 Economic development, 1, 3, 7, 9–11, 16, 28, 32, 46, 63, 68n52, 191, 269, 274, 328, 329, 348 Economic growth / performance, 4, 5, 24, 27, 55n6, 72n64, 186, 309, 330, 332, 337n28, 349, 351 Economic integration, 12, 61, 105–122, 256, 270, 270n75, 273, 274 Economic rationalism, 316 See also Market integration Economic theory, 7, 335 Econophysics, 132–134, 132n46, 137, 141, 142 Efficiency, 24, 25, 192 Egypt, 7, 15, 42, 299, 310–312, 314, 320, 347 Emergence, 10, 14, 35, 46, 55, 73, 74, 127, 128, 137, 138, 142, 147, 149, 154, 165, 168, 190, 277 emergent properties, 13, 14, 33, 167, 276 Emptio venditio, 126, 145, 148 Energy, 5, 21, 32, 76n82, 169, 178, 194, 276, 309

359

Enforcement, 10, 23, 25–27, 30, 32, 34, 36, 38, 41, 44, 46, 74 Epidemic, 16, 135 See also Disease; Pandemic; Plague Equilibrium / equilibrium theory, 6, 8, 15, 30, 54, 128, 167, 254, 300, 302, 310 See also General equilibrium; Partial equilibrium Evergetism, 43, 45 Exchange, 26, 32, 33, 36, 37, 39, 40, 45, 46, 60, 62, 68–71, 74, 91, 108, 109, 125–154, 164, 181, 184–186, 190–193, 254, 316, 329, 337, 347, 350 Exchange networks, 185, 188, 194 Exogenous shock, 7, 27, 276, 349n70 F Famines, 301, 310, 311 See also Food supply Fanum Fortunae, 225 Feedback (loops), 169, 190, 333 Flow models, 140 Food supply, 26, 254, 301, 302 See also Famine Formal institutions, 22–24, 26–28, 27n12, 38, 39, 43 Fortification, 173, 186, 271 Forum Sempronii, 220n58, 229 Frankincense, 70 Free-riders / free-riding, 23, 28–30 G Galatia, 259, 260, 260n39, 264 Game theory, 32 General equilibrium, 6, 7, 300, 301 See also Equilibrium / equilibrium theory

360 

INDEX

Geomagnetic / geophysical survey, 214, 218, 223, 231 Gift exchange, 36, 39, 42 Grain, 7, 62, 63, 70, 110, 210, 308, 309, 311, 319, 347 Gymnasia, 265, 266 H Harbours, 11, 34, 59, 60, 71, 209, 339, 340n36 Hatria, 226 Herculaneum, 14, 228, 242 Hierapolis, 62 I India, 11, 21, 33, 71, 75 Indian Ocean, 11, 69, 71 Inflation, 15, 154, 302, 303, 306, 309, 310, 312, 317 Informal institutions, 23, 24, 27, 32n23, 47 See also Non-designed institutions Information, 3, 4, 7, 10, 12, 14, 23, 29, 30, 37, 46, 56, 57, 69, 71, 71n64, 74, 77, 107, 108, 112–115, 117–120, 129, 131, 133, 142, 143, 166–168, 170, 191, 192, 204, 206–208, 210, 212, 225, 228, 229, 232, 242, 256, 275, 276, 303, 319, 329, 333, 336–338, 340n36, 342, 343 Infrastructure, urban, 13, 207, 213, 253 Innovation, see Technology Institutional change, 4, 10, 13, 14, 22–32, 37, 38, 46, 154 Institutional development, 185, 194 Institutions / institutionalisation, 2, 12, 13, 22–32, 35, 37–41, 44, 46, 47, 108, 125, 126, 130,

145–147, 149, 151–154, 169, 186, 190, 192, 193, 263, 277, 320, 328–330, 337, 349 Interamnia Praetuttiorum, 220n58, 225 Investment, 6, 13, 14, 169, 180, 181, 186, 189, 254 Italy, 7, 26, 205, 206, 208, 211, 242, 243, 257, 333, 340, 344, 347 J Justinian I, Emperor, 75, 349, 350 Justinianic Plague / Justinian’s Plague, 327–351 K Keynesian models / Keynesian economics, 6 Kiln, 178, 179, 182, 189, 218 See also Pottery / potters L Labour division / specialisation of, 194 Levant, 67 Loans, 41, 126, 145, 149–152 Locatio conductio, 126, 144, 148, 150, 152 Lycaonia, 259, 260 Lycia, 258, 260, 261, 264, 267, 268, 271, 273, 287 M Mainstream economics, 2, 6, 55n8, 127, 129, 130, 133, 167 Malthus, 302 See also Malthusianism

 INDEX 

Malthusianism, 254, 301, 302, 329, 348 See also Malthus Market economics, 6, 12, 16, 25, 26, 39, 61–63, 68–70, 71n64, 72, 107–116, 118, 120, 121, 132–134, 132n46, 185, 190–192, 265, 277, 300, 301, 308, 309, 313, 316, 319, 320, 336, 347, 350 Market economy, 12, 107–111, 130, 328, 347, 349 Market integration, 63, 107, 111, 114, 117, 118, 311, 320, 347 See also Economic integration Matilica, 220n58, 229 Mechanisms of complexity, 164, 166, 167, 193 Mediterranean, 16, 45n48, 65, 68, 68n54, 69, 71, 73, 77, 108, 109, 208, 328, 332, 340 Monetisation, 41, 43, 45, 139 Money quantity theory of, 311 role of, 41, 43, 44n47, 45, 47, 139 Monte Testaccio, 72 Muziris, 71 Myos Hormos, 71 Mysia, 259, 263 N Nash equilibrium, 29 Neoclassical economics, 7, 9, 126, 128, 130 Nesting / nested cluster, 66, 67 Network measures, 81 Network theory / network analysis, 56–59, 57n13, 70n59, 72–77 New Institutional Economics (NIE), 1, 8, 10, 22n2, 24, 108, 125 Non-designed institutions, 10

361

See also Informal institutions Numismatics / numismatic data, 6, 257 O ORBIS data set, 340n38, 342, 350 Ostia, 205, 209, 210 Ostra, 220n58, 229, 232 P Palmyra, 74 Panarchy, 255, 263, 276 Pandemic, 7 See also Epidemic; Disease; Plague Partial equilibrium, 300 See also Equilibrium / equilibrium theory; General equilibrium Path dependence, 31 Path length, 58, 71 Performance, 2 See also Economic growth / performance Periplus Maris Erythraei, 70n62 Phrygia, 259, 264, 267, 271, 272 Pisaurum, 227, 229 Pisidia, 260, 261, 263, 264, 273, 287 Plague, 298, 300–319, 331 Po (river), 64, 79–82 Polis, 257, 258, 263, 268 Political economy, 190, 191 Pompeii, 4, 14, 205, 209, 210, 212, 228, 232, 242 Pontus, 263, 272 Population growth, 301, 329, 337 levels, 253, 269, 301 urban, 242 (see also Demography, urban) Post-processualism, 252

362 

INDEX

Potentia, 13, 206, 213–219, 221, 227, 228, 239, 241 Pottery / potters, 13, 63, 64n39, 73, 93, 94, 164, 174–179, 181–185, 188, 189, 191, 193, 217 Power laws, 55, 59, 72, 269, 286, 288 Prices (of goods), 107, 112, 120 Problem-solving, 168, 169, 194 Processualism, 252 Production, 2, 4, 5, 12, 13, 21, 23, 25, 26, 28, 32, 39, 54n2, 63, 72, 73, 77, 106, 107, 112, 114–120, 130, 164, 168, 174–184, 187–194, 264, 301, 347n62, 348 Property rights, 13, 22, 32, 42, 44, 125, 169, 186 Proxy data, 1–17, 55, 55n6, 256 Public building, 14, 211, 220, 251, 256, 264–268, 275, 277 Q Quantification, 5, 297, 320–321 Quantitative modelling, 297 Quantitative Narrative Analysis, 76 R Rank-size analysis, 14, 269, 270, 275 Rationality / rational actors, 107 agents, 28, 29, 112, 120 Remote sensing, 13, 218, 227 Resilience, 21–47, 67n50, 68, 255, 276 Resource distribution, 25, 32 Resource exploitation, 178, 188, 189 Resource procurement, 13, 164, 175–178, 187, 193 Rheinzabern (Tabernae), 11, 73, 93, 94 Riverine networks, 61 Rivers / river routes, 59, 64, 72

Roads, 13, 42, 45, 62, 64, 217, 218, 227, 251, 253, 264n51, 274, 328 Roman law, 32, 41, 142–150, 152, 154, 349 Rome, 5, 6, 15, 21, 26, 42, 46, 62, 63, 109–112, 205, 209, 210, 276, 311, 314, 315, 319, 340, 347 S Sabratha, 211, 212 Sagalassos, 13, 163–195, 254, 273, 304 Scale / scalar properties, 253 Self-organisation, 37 Self-similarity, 14, 37, 37n32, 38, 46, 253 Sena Gallica, Senigallia, 233 Sentinum, 220n58, 229, 231 Shipwrecks, 6 Silk, 75, 350 Slaves, slave-mode of production, 26 Small-world network, 58, 70 Social complexity, 163–195 Social hierarchy, 14, 37, 264 Social interaction / social and economic interaction, 331 See also Network theory / network analysis Social norms, 10, 23, 29, 30, 32n23, 46, 47, 169 Social roles, 22, 34, 35, 37, 43 Social rule system, 28, 31, 32, 34, 46 Social sanctions, 23, 27, 35, 36 Social system, 8–10, 32–34, 37, 38, 44–46, 54n2, 165, 169, 171 Social system theory, 10, 32–38, 32n24 Socio-ecological system (SES), 255 Socio-economic complexity, 13, 164–170, 174, 175, 186–195 Stadia, 265

 INDEX 

Statistics, 77, 111, 118, 132, 331, 335, 342 Structural models, 64, 252 Structure, 4, 9, 13, 14, 22, 31, 32, 37, 37n32, 38, 43, 45–47, 56, 58, 60, 67, 72–74, 114, 116, 127, 129, 130, 134–137, 139, 140, 164, 168, 169, 173, 179–181, 184–186, 189–192, 203, 208, 211, 217, 225, 229, 252–254, 276, 347 Supply and demand, 5, 6, 9, 108, 114, 118, 120, 169, 186, 191, 300, 301, 303, 312, 319 Surveys, archaeological, 207 Synoikismos, 194, 263 System theory, 10, 32 T Tableware, 12, 106, 107, 112–118, 120, 121, 176, 177, 181–183, 185, 273, 274n88 See also Ceramics Taprobane, 71 Taxes, 6, 22, 38, 39, 258, 264, 306, 308 Technology technological development, 13, 169, 170, 186 technological innovation, 22, 169, 170, 189 Temin, Peter, 12, 61, 67, 107–113, 116, 120, 121, 300 Terra sigillata, 73 Theatre, 14, 204, 211, 220, 256, 265–268, 271, 281, 282, 284 Tifernum Mataurense, 220n58, 228–230, 236, 237 Tin, 70 Trade communities, 74

363

diasporas, 74 Trade geography, 343 Trade routes, 350 See also Trade geography Transaction costs, 22, 192, 320 Trea, 206, 213, 222–228, 239, 241, 242 Tyrrhenian Sea, 67 U Urban hierarchy, 251–277 Urbanisation / surbanisation, 4, 10, 14, 186, 243, 259, 263, 267, 268, 348 Urbanism, 251–277 Urban networks, 14, 256, 258, 259, 267–270, 273 Urban topography, 205 Urbs Salvia, 234 V Village, 13, 33, 174, 188, 204, 257, 258, 262, 267, 299 Visualisation(s), 4, 12, 57, 58, 60, 83–85, 87, 90–92, 126, 127, 131–134, 137–143, 139n65, 145, 149, 151 W Wages, 15, 144, 148, 297, 300, 311–319, 348 Walls (of cities), 211, 331 Warfare, 63, 318 Well-being, 4, 5 Z Zipf’s law, 269, 270, 273, 275, 285–288