Villa Landscapes in the Roman North: Economy, Culture and Lifestyles 9789048514830

Citation preview

Villa Landscapes in the Roman North

           

17

Editorial Board: Prof. dr. E.M. Moormann Prof. dr. P.A.J. Attema Prof. dr. N. Roymans Prof. dr. F. Theuws Other titles in the series: 1. N. Roymans (ed.): From the Sword to the Plough. Three Studies on the Earliest Romanisation of Northern Gaul Open Access edition: http://dare.uva.nl/record/19675 2. T. Derks: Gods, Temples and Ritual Practices. The Transformation of Religious Ideas and Values in Roman Gaul Open Access edition: http://dare.uva.nl/aup/en/record/172370 3. A.Verhoeven: Middeleeuws gebruiksaardewerk in Nederland (8e – 13e eeuw) Open Access edition: http://dare.uva.nl/aup/en/record/172373 4.  F. Theuws / N. Roymans (eds): Land and Ancestors. Cultural Dynamics in the Urnfield Period and the Middle Ages in the Southern Netherlands Open Access edition: http://dare.uva.nl/aup/en/record/172372 5. J. Bazelmans: By Weapons made Worthy. Lords, Retainers and Their Relationship in Beowulf Open Access edition: http://dare.uva.nl/aup/en/record/172337 6. R. Corbey / W. Roebroeks (eds): Studying Human Origins. Disciplinary History and Epistemology Open Access edition: http://dare.uva.nl/aup/en/record/172272 7. M. Diepeveen-Jansen: People, Ideas and Goods. New Perspectives on ‘Celtic barbarians’ in Western and Central Europe (500-250 BC) Open Access edition: http://dare.uva.nl/aup/en/record/172273 8. G. J. van Wijngaarden: Use and Appreciation of Mycenean Pottery in the Levant, Cyprus and Italy (ca. 1600-1200 BC).The Significance of Context Open Access edition: http://dare.uva.nl/aup/en/record/172274 9. F.A. Gerritsen: Local Identities. Landscape and community in the late prehistoric Meuse-Demer-Scheldt region Open Access edition: http://dare.uva.nl/aup/en/record/172820 10. N. Roymans: Ethnic Identity and Imperial Power. The Batavians in the Early Roman Empire Open Access edition: http://dare.uva.nl/aup/en/record/172930 11. J.A.W. Nicolay: Armed Batavians. Use and significance of weaponry and horse gear from non-military contexts in the Rhine delta (50 bc to ad 450) Open Access edition: http://dare.uva.nl/aup/nl/record/397232 12. M. Groot: Animals in ritual and economy in a Roman frontier community. Excavations in Tiel-Passewaaij Open Access edition: http://dare.uva.nl/aup/en/record/301888 13. T. Derks & N. Roymans: Ethnic Constructs in Antiquity. The role of power and tradition Open Access edition: http://dare.uva.nl/aup/en/record/301890

14. T. D. Stek: Cult places and cultural change in Republican Italy. A contextual approach to religious aspects of rural society after the Roman conquest ISBN 978 90 8964 177 9 15. P. A.J. Attema, G.-J. L.M. Burgers & P. M. van Leusen: Regional Pathways to Complexity. Settlement and land-use dynamics in early italy from the bronze age to therepublican period ISBN 978 90 8964 276 9 16. E.M. Moormann: Divine Interiors. Mural paintings in Greek and Roman sanctuaries ISBN 978 908 9642615

Villa Landscapes in the Roman North E

econ om y, culture an d li festy les

editors

n ico r o y ma n s

&

t o n d er k s

a msterdam u n i v ersit y p ress

This book meets the requirements of ISO 9706: 1994, Information and documentation – Paper for documents – Requirements for permanence. Cover illustration: Roman sarcophagus from Simpelveld (N.) Photo courtesy RMO Leiden. Cover design: Kok Korpershoek, Amsterdam Lay-out: Bert Brouwenstijn, ACVU Amsterdam ISBN 978 90 8964 348 3 e-ISBN 978 90 4851 483 0 NUR 682 © Nico Roymans, Ton Derks, Amsterdam University Press, Amsterdam, 2011 All rights reserved. Without limiting the rights under copyright reserved above, no part of this book may be reproduced, stored in or introduced into a retrieval system, or transmitted, in any form or by any means (electronic, mechanical, photocopying, recording, or otherwise), without the written permission of both the copyright owner and the editors of this book.

to our f rie n d jan slofstra

co n te n ts p re f ace

ix

i n troductor y essay Studying Roman villa landscapes in the 21st century. A multi-dimensional approach Nico Roymans / Ton Derks 1

studies o f late iro n age / roma n tra n sitio n Reflections on the Iron Age background to the emergence of villa landscapes in northern France Colin Haselgrove

45

Exploring villa development in the northern provinces of the Roman empire Diederick Habermehl

61

On the origin and development of axial villas with double courtyards in the Latin West Nico Roymans / Diederick Habermehl

83

essay s o n social reco n structio n Town-country dynamics in Roman Gaul. The epigraphy of the ruling elite Ton Derks

107

Ethnic recruitment, returning veterans and the diffusion of Roman culture among rural populations in the Rhineland frontier zone Nico Roymans

139

Indications for rural slavery in the northern provinces Nico Roymans / Marenne Zandstra 161 The idea of the villa. Reassessing villa development in south-east Britain Jeremy Taylor

179

i n ter p retatio n o f mortuar y e v ide n ce The role of mortuary ritual in the construction of social boundaries by privileged social groups within villa landscapes Laura Crowley

195

Monumental funerary structures of the 1st to 3rd centuries associated with Roman villas in the area of the Treveri Jean Krier / Peter Henrich

211



regio n al studies Roman rural settlements in Flanders. Perspectives on a ‘non-villa’ landscape in extrema Galliarum Wim De Clercq

235

Evaluating settlement patterns and settlement densities in the villa landscapes between Tongres and Cologne Karen Jeneson

259

The villa landscape of the Middle Aare valley and its spatial and chronological development Caty Schucany

275

Roman villa landscapes of the lignite mining areas in the hinterland of Cologne Wolfgang Gaitzsch

285

studies o f i n di v idual sites The Roman villa complex of Reinheim, Germany Florian Sărăţeanu-Müller

301

The Roman villa at Borg. Excavation and reconstruction Bettina Birkenhagen

317

List of contributors

331



p re f ace The archaeology of Roman villas and more generally of Roman rural landscapes has enjoyed widespread interest for some generations now, among both scholars and a wider audience. Some educational presentations of villa sites have even emerged as regular public attractions, as we have seen in the past decade with the villas of Borg in the Trier area, Ahrweiler near Bonn and Jemelle-Malagne in the Belgian Ardennes.1 Despite marked differences in the way they present the Roman past, they all have in common their ability to appeal to the general public, each attracting about 50,000 visitors a year. Why do Roman villas hold such a fascination for people? There are probably different factors at work here. Villas are regarded as the most ‘Roman’ component of the rural landscapes of that time. Through their fascinating archaeological remains (pavements with mosaics, houses with stone foundations, tiled roofs, painted walls, hypocausts, bathing facilities, etc.), they stimulate public imagination about the Roman past. Roman villas evoke a feeling of familiarity through their associations with modern ideals of civilisation and strategies of elite distinction in terms of comfortable living and an appreciation of a rustic life in the country. In our modern post-industrial society, the arcadian ‘Traum vom Lande’ also has broad appeal, as we see in the escape from the city and the building of modern villas. In that sense, villas fit within the modern quest for an idealised past. At the same time, villa attractions help us put modern Western civilisation into perspective through our awareness of the decline of the Roman empire. Underpinning this volume is archaeological research into villas and villa landscapes in the northern provinces of the Roman empire. In 2006 the Archaeological Centre at VU University Amsterdam launched the research programme Villa landscapes in the Roman North. Economy, culture, lifestyles. Funded by the Netherlands Organisation for Scientific Research (NWO), it is scheduled for completion in 2011. Included in the planned output for the programme is this synthesis volume of articles by project members, supplemented by contributions from other archaeologists working in the field of Roman rural archaeology. The aim of this book, which builds on an international symposium organised in Vaals (NL) in November 2008, is fourfold: a. to develop a synthesis on the emergence and development of Roman villa landscapes in the northern provinces. b. to formulate a cohesive body of theoretical concepts and ideas about the social, economic and cultural functioning of villas within the context of the Roman empire. c. to contribute to comparative research into villa landscapes and the phenomenon of regionality of Roman rural landscapes. d. to bring about methodological innovation by devising a classification of villa development trajectories, and to apply new geographical and cartographical techniques (including GIS) in the analysis of villa landscapes. The source material used in this book is primarily of an archaeological nature. Our aim first and foremost is to incorporate in the analyses the explosive growth in archaeological information in recent decades, including that from ‘grey literature’, and to render it productive for broader discussions. But we also seek to use historical and epigraphic data to arrive at a more balanced picture. This volume is simply organised.We start with an extensive introductory paper setting out the parameters for the study of Roman villa landscapes using a series of themes. This is followed by 16 articles with a thematic or regional focus that give added depth to the discussions raised here. They include two contributions which present concrete case studies of recently excavated villa sites.  Cf. the following websites: http://www.villa-borg.de;

1

http://www.bad-neuenahr-ahrweiler.de/bildung_kul-

tur_freizeit/museen/museum_roemervilla/index.html; www.malagne.be.



We hope that this volume will have an impact on research agendas for Roman rural settlements in the 21st century by providing a series of new questions and discussions. All that remains is for us to thank all the people and institutes who have contributed in some way to bringing about this book. First of all, we wish to thank Laura Crowley, Karen Jeneson and Diederick Habermehl, PhD students who have worked on their research with great enthusiasm for over four years and who have helped to create a marvellous working atmosphere in our project group. Our thanks also go to other colleagues at the Archaeological Centre of VU University Amsterdam and at the Research Institute CLUE for their discussions on the many topics touched on in this volume, especially Jeremia Pelgrom, Maaike Groot, Joris Aarts, Philip Verhagen, Hans Renes and Jan Kolen. We are very much obliged to Greg Woolf for his informal role as project advisor and for his stimulating contribution at the symposium in Vaals. Finally, we are especially grateful to Jan Slofstra, who was at the heart of many discussions held here and who gave us continual feedback. As a token of our long-standing friendship we dedicate this volume to him. Bert Brouwenstijn and Jaap Fokkema once again ensured that the illustration material was of a high calibre; Bert also took care of the layout for the book. Thanks also to all the speakers at the Vaals symposium for their stimulating discussions. We would like to thank Amsterdam University Press and in particular Jeroen Sondervan for their always constructive comments and their efforts to make something beautiful of this volume. We extend our gratitude to two anonymous reviewers for valuable comments on the volume’s manuscript. We would also like to express our appreciation to Annette Visser (Wellington, NZ) for translating or correcting the English of the contributions by non-English speaking authors. Nico Roymans / Ton Derks



Studying Roman villa landscapes in the 21st century. A multi-dimensional approach Nico Roymans / Ton Derks 1 Introduction 2 History of villa research 3 Outlining a multi-dimensional approach 4 A landscape-archaeological perspective 5 Rome’s impact on the spatial organisation of rural landscapes 6 Town-countryside dynamics 7 The house as a symbol of elite distinction 8 The economic basis of villa-dominated landscapes 9 The social organisation of production 10 Consumption, self-representation and lifestyle of villa owners 11 Religion and the structuring of rural life 12 The impact of Roman military culture 13 Regional variation in villa landscapes 14 Long-term transformations 15 Villa landscapes and the future of Roman archaeology Abbreviations References

1 i n troductio n The incorporation and integration into the Roman empire had a profound impact on the socio-political, economic and cultural order of the peoples in the Gallic and Germanic provinces. The key material example was without doubt the introduction of Roman-style towns and the associated urban culture. Just as important, however, and inextricably linked to this, was the profound transformation of the countryside, where at least 90% of the population lived. A key factor was the large-scale appearance of Roman villas from the second half of the 1st century AD onwards. Because of their attractive physical manifestation (multi-roomed houses with tiled roofs, stone foundations, plastered walls, hypocausts, bathing facilities, etc.), they count as the most ‘Roman’ component of the then rural landscapes. There is a long tradition of research into the origin, development and social interpretation of villas (see below). In the past decades we have seen a dramatic increase in the number of excavations and a veritable information explosion on Roman rural settlements. However, developing interpretive frameworks for rural settlement research has failed to keep pace with this data explosion. What we need are new, theory-based syntheses to organise the existing evidence and point the way for future research. This study is an attempt to do just that. Our aim is threefold: a. to develop a synthesis of the origin and development of Roman villa landscapes in the northern provinces, b. to formulate a cohesive body of



theoretical concepts and ideas about the social, economic and cultural function of villas within the context of the Roman empire, and c. to contribute to comparative research into Roman villa landscapes. We also wish to create a framework in which the various contributions to this volume can be understood. Defining the concept of ‘villa’ is no easy task. One of the problems facing us is that the term has a wide range of meanings in Roman written sources and does not therefore lend itself to a workable archaeological definition.1 On the other hand, it has proved archaeologically impossible to distinguish clearly between villa settlements and non-villa settlements. Faced with a great variety of settlements, any attempt to categorise them archaeologically will always be a somewhat arbitrary exercise.2 Nor do villas constitute a homogeneous category in either a morphological or a social sense. Provided that we distinguish between a morphological and a relational definition, we can retain ‘villa’ as a scholarly concept. A morphological definition highlights a villa’s material manifestation, which is made up of the following key elements: an agrarian settlement with a Roman-style, multi-roomed main house, partly constructed of stone and with a rectangular ground plan, solid floors, a tiled roof, and sometimes a hypocaust and a set of baths. We can distinguish three main categories of rural settlement:3 a. large and medium-sized villas with a monumental main house of more than 30 or 15 rooms respectively, as well as a number of secondary houses inhabited by dependent families or labourers and other ancillary buildings; b. single farmsteads with a small stone-built main house; and c. ‘native’ or ‘traditional’ settlements with only postbuilt houses. In relational terms, a villa can be defined as ‘a farm, which is integrated into the social and economic organization of the Roman world’.4 This definition alerts us to the specific social and economic organisation of villas and how these were embedded in the Roman empire. If we wish to explore the latter, we need to take account of all other rural settlements, as well as towns. There are two important comments to make about the above classification. Firstly, within the villa landscapes the category of non-villa settlements with only post-built houses seems more widely present than originally assumed. Because of their lesser archaeological visibility they are often underrepresented in our datasets.5 Secondly, opinions vary as to how the large group of single farmsteads with a stonebuilt main house may best be classified. Proceeding from a broad villa definition with the presence of stone-built structures as diagnostic criterion, archaeologists working in the empire’s northern provinces generally refer to them as ‘small villas’. In the Mediterranean provinces, a stricter villa definition is usually adopted, which classifies only the major complexes of our category A as villas. This difference undoubtedly arises from the fact that stone construction was not introduced in the northern provinces until after the Roman conquest, whereas farmhouses with stone plinths were in use in the Mediterranean region well before the conquest. The adoption of a narrow or broad definition of villas has implications for their social interpretation. If we adhere to a strict definition, this implies a direct association with an upper social echelon or elite. If we opt for a broad definition, we should be cautious about automatically associating villas with elites. In most instances the occupants of the stone-built farms of our category B will have been middle-class farmers. However, the morphological and social categorisations of small villas are not always synchronous. The simple villa of Ravensbosch, for example, seems to have belonged to a member of the decurial elite from Xanten, and the famous stone sarcophagus of Simpelveld was found at

1

On the disparity of the meaning of the term ‘villa’ in the

3

literary sources, see Millett 1990a, 91-92; Terrenato 2001, 18 f.; Dyson 2003, 13 ff, esp. 19; DNP vol. 12/2 (2002),

rich 2006, 25; Krause 2006, 273-277. 4

210-221, esp. 213 [Ch. Höcker]; RGA, vol. 32 (2006), 375-387 [F. Reutti]. 2

Cf. Ouzoulias 2006, 200-213; idem 2010; Haselgrove,  this volume; Habermehl, this volume; Jeneson, this vol-

149) emphasize the arbitrary nature of the boundary



Rivet 1969, 177. See also Slofstra 1983, 84 ff.; Mattingly 2006, 370.

5

Several scholars (Slofstra 1991, 163, note 86; Woolf 1998, between ‘native farms’ and ‘villas’.

See Habermehl, this volume; Mattingly 2006, 371; Hen-

ume. 6

Roymans 1996, 61, 73.

G

E in

R

Rh

e

M

A N I

A

EI

FE

L K

S A R D E N N E

Ü

C

v

l a i l

s p e c a s n d l a

N HU

M

SR

eu

O

G

E S E N

se

V

0

A

B

C

D

E

100 km

F

Fig. 1. Simplified map of the distribution of non-villa (A) and villa landscapes (B) in Northern Gaul and the Rhineland. C: excavated rural settlement with byre houses. After Roymans 2007, fig. 18.

a particularly small corridor villa. In this study we will use the terms ‘single farmstead with a stone-built main house’ and ‘small villa’ as synonyms. A ‘villa landscape’ can be defined as a rural landscape that is dominated by villa settlements, if not in a numerical sense, then at least in terms of how they are perceived by rural populations in the light of their monumental nature and social status. A ‘non-villa landscape’ is one in which traditional settlements with post-built houses make up the vast majority of rural sites (more than 95%).6 This volume will focus on landscapes in which villas were a salient feature. However, because the difference between villa landscapes and non-villa landscapes is a relative one, we also include non-villa landscapes such as the Dutch river delta and the Belgian coastal area.7 Geographically speaking, this volume focuses on villa landscapes in the northern provinces of the Roman empire, and more especially in the east-west oriented, fertile loess belt between the Rhine and the Channel coast of Northwestern France (fig. 1). However, individual contributions continually make comparisons with developments in other regions, such as southeastern Britain, Picardy, the Trier region, the Belgian coastal area, the Dutch river delta and the Swiss Mittelland. With regard to time frame, this study focuses on the period from the 1st century to the middle of the 3rd century AD. However, we cannot understand Roman rural landscape formation unless we 7

Cf. Habermehl, this volume.

3

include Late Iron Age developments in our analyses. At that time, the northern half of Belgic Gaul and the Lower Rhineland were populated by less hierarchical, stateless societies lacking urbanised oppida and with almost no access to Mediterranean prestigious imports such as wine and bronze drinking vessels.We can, however, discern there a process of increasing coin usage and a ‘latènisation’ of the material culture.8 The source material used here is primarily archaeological. We have tried to take the recent upsurge in archaeological data, including that from ‘grey literature’, and render it productive for broader discussions, while at the same time continuously bridging the gap between archaeological findings and research conducted by historians in order to arrive at a more balanced picture.

2 histor y o f v illa research There is a long tradition of villa research into the northwestern provinces of the Roman empire (fig. 2).9 In Italy the oldest study of Roman villas dates back to the Renaissance,10 while in the northern parts of the European continent the villa of Buchs in Switzerland is the earliest documented villa ground plan (fig. 3). Published in 1759, it was interpreted as a military mansio.11 The archaeological study of villas did not really get underway, however, until the later 19th century. The later 19th and early 20th century was the heyday of private sociétés savantes, some of which concentrated on archaeological excavations of Roman villas and monumental tumuli. One of the most momentous findings was the almost complete ground plan of the axial villa complex of Anthée (B)12 and the impressive main building of the villa of Basse-Wavre (B).13 The broader interpretive framework for these studies was a historical and philological one, inspired by the texts of classical authors.Villas, and in particular their monumental principal buildings, were seen as the exponents of illustrious Roman civilisation and, proceeding from a normative view of Roman culture, northern villas were regarded as deriving from Italic examples. It was often implicitly assumed that these villas were inhabited by ‘real’ Romans of Mediterranean origin.Villas were contrasted with ‘native settlements’, where people of Gallic or Germanic origin lived.14 The first half of the 20th century saw the first generation of university-trained archaeologists. Employed by national and provincial museums, they continued to carry out villa research. One significant villa excavation was conducted by Franz Oelmann, at Mayen in the 1920s.15 Beneath the monumental main building, Oelmann identified traces of an older pre-Roman structure, which he believed argued for a Celto-Germanic origin of the later villa occupant. A study at Köln-Müngersdorf, published in 1933 by Fritz Fremersdorf, director of the Cologne Museum, also had a great impact.16 This was the first fully excavated villa complex, consisting of a main building, subsidiary buildings, graveyards and fences. Fremersdorf attempted to systematically identify the functions of the secondary buildings in order to arrive at a broader socioeconomic interpretation of the villa. In view of the discovery of habitation traces and finds that preceded the villa’s construction, he too suspected that the builder was of native origin. In Belgium, Robert de Maeyer’s 1937 synthesis marked an important milestone in villa research. It gave an

8

Roymans 1990; idem 2004, chapter 2; idem 2007.

9

Percival 1976 still remains an excellent introduction to

interpretation of villas, Hingley 2008, 157 ff, esp. 164173.

the discussion of villas.

12

Cf. Roymans/Habermehl, this volume, fig. 1.

10

Dyson 2003, 13 ff.

13

De Maeyer 1937, 73-76, fig. 16.

11

Horisberger 2004, fig. 7 and fig. 339. On the British

14

Derks 1998, 3.

Isles, the plans of the Cotterstock and Weldon villas were

15

Oelmann 1928.

published a little earlier (1738). For these and the military

16

Fremersdorf 1933.



Fig. 2. Overview of excavated rural settlements in the northern provinces, with specification of the research period. After Habermehl 2011, fig. 2.2.

overview of all villa ground plans excavated in Belgium up to that time, pointing out the considerable variation in form and size.17 A new feature of post-war villa research was the introduction of systematic regional surveying and mapping programmes of Roman rural landscapes, which added a broader landscape dimension to villa studies. Examples are the work of Hinz and later Gechter and Kunow for the German Rhineland, and Willems for the Dutch river area.18 Agache’s aerial photographic study in the valley of the Somme around Amiens marked a highpoint in French villa research.19 For this study, Agache developed the methods and techniques of aerial photographic surveys, and presented a wealth of documentation on pre-Roman and Roman cultural landscapes. He identified an astonishing villa density on the loess plateaus in the Somme area. The late 1970s saw the beginnings of systematic archaeological research in the lignite mining area of the Cologne hinterland, a vast project that continues to this day.20 Here, the total destruction of the cul17 18

De Maeyer 1937. Cf. De Clercq 2009, 48-51.

new survey data.

Hinz 1969; Gechter/Kunow 1986; Willems 1981, 1984.

19

Agache 1978.

See also the Cüppers/Rüger atlas 1985, which uses the

20

Gaitzsch, this volume.



Fig. 3. Ground plan of the villa at Buchs (CH), excavated in 1759. After Horisberger 2004, fig. 7.

tural landscape offered unique opportunities for archaeological research. Dozens of villa sites have been excavated over a period of several decades. However, because the emphasis is on rescue excavations and data collection, many sites have generated either no publications at all or only interim ones. According to synthesising studies, we are dealing here with a Roman colonisation landscape from the middle of the 1st century AD, showing a clear discontinuity with Late Iron Age habitation. Discussion has focused on whether this colonisation was associated with the founding of the Cologne colonia under Claudius and concomitant land grants to veterans, or with settlers drawn from the Gallic hinterland.21 An important thematic volume published in 1996 and edited by Bayard and Collart examined the question of the origin and earliest development of villas.This publication showed not only that the largely excavated villa sites in Picardy were frequently continuations of pre-Roman fermes indigènes, but also that the spatial organisation of Gallo-Roman villa complexes largely harked back to native conceptions of space rather than to Italic examples. The 1st century AD saw a gradual transformation from enclosed fermes indigènes with post-built wooden houses to enclosed Gallo-Roman villas with highly romanised main buildings constructed partly in stone.22 Another influential work was Colin Haselgrove’s regional study of the Aisne valley. He too presents a picture of the gradual transformation of the Late Iron Age settlement landscape into a Gallo-Roman villa landscape, differentiating between developments on the loess plateaus and the alluvial river valleys.23

21

Gaitzsch 1986; Rothenhöfer 2005, 42-47 (veter

22

Bayard/Collart 1996.

an hypothesis); Lenz 1998; idem 1999a; Heimberg

23

Haselgrove 1996; idem, this volume.

2002/2003 (colonisation by Gallic settlers).



A further landmark in the history of villa research in our study region was the 1980s excavation of a Roman villa complex at Hoogeloon in the southern Netherlands by Jan Slofstra of the VU University Amsterdam. It is in fact the only example of a programmatic scholarly study of a villa site. Slofstra developed innovative theoretical frameworks for rural settlement research that incorporated anthropological theories and concepts (see below).24 Lastly, the past decade has seen the publication of significant synthesising studies on rural settlement research in the Batavian region and a comparison with developments in the villa landscapes of the loess belt. The research programme Rural communities in the civitas Batavorum and their integration into the Roman empire, headed by the present authors, produced two dissertations – by Wouter Vos and Stijn Heeren – as well as a series of articles.25 At the same time, regional studies with an ecological focus were published, such as those of Laura Kooistra and Maaike Groot.26 This review of some of the research highlights regarding Roman villas and villa landscapes in the northern provinces may give an impression of a dynamic field of research in which important advances are being made. The truth of the matter, however, is that although the past two decades have produced a great deal of new fieldwork, the quality of many excavation reports has failed to keep pace with the upsurge in excavation activity. Increasingly, reports are written under contracts that leave little room for viewing the excavation data within a broader disciplinary context. In an interpretative sense, presentday villa archaeology has stagnated: it is under-theorised and has become increasingly isolated from other branches of Roman studies. There is little reflection on the post-colonial critique of traditional villa research and little impetus for new social interpretations of the architecture, decoration and spatial organisation of villas.27

3 outli n i n g a multi - dime n sio n al a p p roach Below we attempt to outline the building blocks of a ‘villa theory’. This is a cohesive body of theoretical ideas and concepts that will help us understand how the villa world evolved and functioned economically, socially and culturally within the context of the Roman empire. We have not opted for a single all-encompassing ‘grand theory’, but rather for a pragmatic, eclectic approach that takes into account social, economic and cultural aspects. There should also be constant attention to structures that change over time through the agency of individual actors and groups. In order to arrive at social models, we use concepts such as social competition, elite distinction and emulation, which were developed by social scientists and recently reformulated by the French sociologist Jean-Pascal Daloz. According to this view, it is critical for elites ‘to demonstrate their dominance vis-à-vis subordinates and to exhibit at least as much supremacy as their peers.’28 One way Roman elites tried to achieve this was through the architecture of their rural residences. Their dominant position in the landscape, their spatial dimensions and layout as well as the hegemonic architecture and interior decoration of larger villas all worked to underpin the social distinction of their elite inhabitants. Villas were thus a sign of superior social rank and this claim to superiority helped to further enhance the elite’s social and political control of the countryside (see also Section 7). 24

Slofstra 1983; idem 1991.

within Roman culture, and its neglect of the role of

25

Vos 2009; Heeren 2009; Roymans 2009a; Derks/Roy-

women and of lower social groups within the villa world.

mans 2006; Roymans/Derks/Heeren 2007.

Villas are automatically associated with civilization, the

26

Kooistra 1996; Groot 2008; Groot et al. 2009.

Hochkultur of the Roman empire and the male landown-

27

Hingley 2005, 88 ff.; Mattingly 2006, esp. chapter 12. Criticisms of traditional villa research are its one-sided

ing class. 28

Daloz 2009, 69.

focus on elite culture, the lack of attention to diversity



Related to this are views on the cultural dimension of the villa world. First and foremost, villas represented a style of consumption, pointing to a set of ideas, values and associated lifestyle of an upper social group which thus defined its symbolic boundaries vis-à-vis lower groups.29 It is interesting to interpret the building of villas as a process of competitive emulation by a community of peers. To some extent we can also apply this emulation model to representatives of middle social groups who adopted elements from elite villas into their own small villas, thereby symbolically claiming a place in the Roman world.30 The economic basis of villas and villa-dominated landscapes is a point of special interest. What generated the wealth needed to build villas on such a vast scale from the mid-1st century onwards? Here we look at the vital role of surplus agrarian production for the urban and military markets, combined with the growing monetisation of the economy. Another theme is the extent to which villas exemplified a specific social organisation of production. While some place great emphasis on the specific nature of villa organisation, speaking – in Marxist-inspired terminology – of a ‘villa mode of production’, others criticise this view and see villas primarily as the product of a specific style of consumption. A closely related theme is the role of different forms of labour within the villa system, especially tenants, wage labourers and slaves. Archaeological research has so far paid scant attention to this rural working class. Villas are not a static phenomenon but must be understood within the forcefield between structure and change. On the one hand, we need to analyse the rise of villa landscapes in relation to Roman social, political and legal institutions, such as the patronage system, municipal organisation and Roman legislation. On the other hand, we should consider agency aspects of the social changes associated with the formation of villa landscapes. This brings us once again to the reasons why people built villas, which was the growing need to display social status and hierarchy in the private sphere in order to secure for themselves a place in the new social context of the Roman empire. A relational definition of villas alerts us to their interaction with both the urban and military worlds. Section 6 presents an analysis of villas as an urban phenomenon which can only be understood in the context of the town-country dialectic. Villas participated in the political, economic and cultural order of the town. Although owners of large and medium-sized villas in particular will also have had a house in town, the relationship between the urban domus and the rural villa is a field that has remained largely unexplored. Furthermore, based on the way rural settlements developed, we can distinguish between: a. villa landscapes as newly colonised landscapes, often initiated by the Roman authorities (e.g. coloniae) and; b. villa landscapes as a transformation of a pre-Roman native settlement landscape.This latter variant is by far the most prevalent, but hybrid forms are possible too. Another key relationship was the one between villas and the military world. In the northern frontier provinces, the countryside had a broad significance for the military community: as a supplier of agrarian products, as a recruitment area for soldiers and as a place for settling veterans.There was a complex articulation between rural, urban and military communities and identity groups, an idea that Mattingly has recently expanded on for Britain.31 The contribution by Roymans in this volume looks at rural developments from a frontier perspective. Finally, the study of villa landscapes involves a global/local dimension. In this volume, we wish to explore the broad regional diversity of villa landscapes. This ties in with a trend within archaeology – a growing interest in locally and regionally specific aspects, partly in response to the present-day globalisation of society. Some ideas from globalisation theory in the modern world can be useful when thinking about the Roman empire.32 In Roman times too, we observe a seemingly paradoxical concurrence of cultural homogenisation and an emphasis on cultural diversity. This study looks at both ‘unity’ and ‘diversity’ in Roman villa landscapes. For the latter, we investigate regional patterns in funerary monuments and the layout and architecture of villa complexes.This regional diversity can be understood as distinctive 29

Millett 1990a, 94, 97; idem 1995, 68; Woolf 1998, 148.

30

Millett 1995, 71. See also the discussion below in Sections 10 and 15.



31

Mattingly 2006.

local responses by groups and individuals to increasing connectivity and homogeneity.33 Hybridity is an attractive explanatory model here.34

4 a la n dsca p e - archaeological p ers p ecti v e ‘… the roofs of villae, perched high upon the overhanging river-banks, the hill-sides green with vines, and the pleasant stream of Moselle gliding below with subdued murmuring.’ Ausonius, Mosella, 20-22. Landscape archaeology offers a key perspective for this study. We take as our starting point the notion that most archaeological data has a spatial dimension. The study of landscape is here not so much an end in itself, but above all a way to better understand how societies functioned economically, socially and culturally. To a significant degree, villa landscapes were the spatial embodiment of the daily rhythm of the lives of local inhabitants. We observe set patterns, such as tilling the soil, grazing cattle, transporting goods over land and water, hunting, and trips into town. At the same time, villa landscapes were part of a complex state society. This means that authorities with domiciles elsewhere had a considerable impact, both directly and indirectly, on how the landscape was organised and used. We will address this topic in greater detail in the next section. Archaeologists often study Roman villa landscapes as a transformation of their Late Iron Age counterparts. We can identify various continuing trends, such as the growing importance of enclosed settlements and a continuation of the highly decentralised habitation pattern of small settlements. Nevertheless, there can be no doubt that these rural landscapes were subject to dramatic and highly visible changes in the course of the first centuries AD. With their monumental main buildings and red-coloured roofs, villas were a prominent feature of the settlement landscape. Added to this were the ubiquitous small cemeteries, monumental graves and sanctuaries. A network of engineered roads with bridges at key river crossings linked the countryside with towns and secondary centres. The landscape was now used more intensively than ever. All of this means that the villa landscapes discussed here were fundamentally different from pre-Roman landscapes in their physical manifestation alone. Here we define the term ‘landscape’ simply as the space that people have organised and assigned meaning to. This implies that landscapes have both a physical and a mental dimension, and that both aspects are deserving of attention. Depending on their field of interest, researchers are able to characterise Roman rural landscapes more specifically. Thus we have ‘landscapes of production’, with an emphasis on economic use, ‘religious landscapes’, with a concern for associations with the supernatural order, and ‘landscapes of power’ or ‘imperial landscapes’, with a focus on manifestations of Roman imperial power. At a more basic level, GIS has proved an important tool for analysing, reconstructing and visualising Roman rural landscapes. In her research, Jeneson explores the methods, techniques and difficulties of trying to integrate complex digital datasets on rural landscapes from three different countries. She combines different spatial datasets in order to produce digital landscape models for the Roman period. Her maps have proved invaluable for evaluating and synthesising these datasets. The final results give us a more reliable picture of the physical dimension of rural landscapes and provide a more solid foundation for reconstructing habitation densities, settlement patterns and road networks.35 We archaeologists are accustomed to studying the agrarian potential and economic exploitation of villa landscapes, and how they differ per region in this respect. An association with the presence of fertile loess 32

We should of course point out key differences between

33

Cf. Giddens 1990, 64.

the two phenomena. Unlike the modern global world,

34

Burke 2009. See Section 13 below.

the Roman empire was also a political entity.

35

Jeneson, this volume; idem, in press.



soils has emerged as a key variable in villa distribution. The cultural dimension of landscapes is less tangible. The term landscape refers to a community’s physical surroundings that have become linked to cultural meaning, world view and a symbolic universe, as locales of specially valued experiences, visions, memories, ideas and ceremonial practices. Such a landscape provides a concrete physical context for the behaviour and identity constructions of individuals and groups, and is the basis for interpretive archaeology. The formation of villa landscapes also implies the formation of new social landscapes. Both for people passing through and for local inhabitants, these were permeated with a new symbolism that pointed to new social relationships, power relationships and links with the town. But how can we say something meaningful about how inhabitants of the time experienced the landscape? Our assumption is that it was the inhabitants themselves who by and large shaped and assigned meanings to the landscape. Individuals and groups availed themselves of opportunities to profile themselves in the landscape by constructing villas, funerary monuments, sanctuaries, field boundaries and the like. In so doing they structured the landscape, using it to express particular ideas and values. This activity will have assumed a different form for members of the elite than for lower social groups, although the evidence is unevenly distributed. For the elite perspective, we sometimes have unique eye-witness descriptions of landscapes, such as Ausonius’ above sketch of the villa landscape in the Moselle valley near Trier. Such landscapes exposed property relations in spectacular fashion by creating highly visible focal perspectives that proclaimed the power of their proprietors.36 Archaeologists are familiar with the notion that landscape is also a temporally layered phenomenon. This ‘historical layeredness’ is the product of a constant process of forgetting and remembering or commemorating, of reusing and effacing.37 This is exemplified in the intense use of villa landscapes to store memory in tangible monuments such as sanctuaries, cemeteries and funerary monuments, and of course the villas themselves. The large-scale building of villas went hand in hand with the introduction of new forms of remembrance and commemoration through monumentalisation and – to a lesser degree – the use of relief images and writing. We see reflected in the monumentalisation of villas a search for greater permanence of private homes, seen as a symbol of the transgenerational vitality and continuity of the families in question.38 The villas and their associated funerary monuments reveal an idealised picture of the family’s inalienable heritage, even though the practice was often quite different. The heads of well-to-do families were often guided by a concern to preserve their good reputation and that of their lineage after their death. Some even had a memorial stone erected near their villa while they were still alive,39 with an inscription designed to preserve their name and that of their family for posterity.40 Others, such as the ignotus from the civitas Lingonum, included provisions in their will about the place, form, size and cost of the funerary monument and about the rituals which they hoped would keep their memory alive after their death.41 To prevent their heirs from disregarding the will, some even demanded penalties for violating its terms. Thus we have provisions aimed at preventing the gift, sale or other kinds of alienation of the family tomb.42 The fact that some conditions were added to epitaphs 36

37 38 39

40

Purcell 1995, 168; Tailor, this volume; Krier/Henrich, 

the ages’ (sparsit per saecula) and of a silence brought about

this volume.

by death (morte silent) and interrupted by the epitaph

Ingold 1993; Bradley 2002; Roymans et al. 2009, 339. Roymans/Habermehl, this volume, Section 6.

CIL XIII 5708.

Cf. CIL XIII 3599 = ILB2 22: epitaph from Gors-

42

Ibid. A comparison with explicit epitaphs elsewhere

Opleeuw of an aedilis from the civitas Tungrorum, who

shows that the penalty in the Lingon’s testament was

erected it at his tumulus when still alive.

primarily a demand aimed at preventing alienation of

For the general argument, Häusle 1980; for an eloquent

the tomb; cf. for example Helttula 2007, no. 253, which

example, Binsfeld 1998 = AE 1998, 953a and b. This

included a penalty for sale, gifting, alienation or the

carmen epigraphicum refers to the ‘high fame’ (sublimem

interment of non-relatives in the tomb.

famam) that the deceased (?) ‘spread of himself through



(aeterno carmine vivant). 41

shows that this concern to preserve the tomb was widespread, especially among groups whose short-term position was less secure, such as freedmen and soldiers.43 For the old aristocratic families of the ruling elite, the continuity of the main residence and a family grave situated on the estate, was guaranteed for the short term. For them too, however, major social changes like those of the late 3rd century brought about disruption.The Bartringen case is a good example of the dismantling of a monumental tower tomb located at a villa. Built in the early 1st century, this monument was demolished in the 3rd century for use as spolia for several new ancillary buildings of the villa.44 Against this background of new remembrance and commemoration practices, it is interesting to explore the extent to which there were fundamental reinterpretations of the past in villa landscapes. We see a distinct need on the part of individuals from upper and middle-class milieus to immortalise their personal reputation and that of their family by creating enduring monumental villas and villa monuments in the landscape.45 This is in strong contrast to the situation in Late Iron Age landscapes, where in the domestic sphere it is precisely the limited differentiation and transience of houses that strikes us and where collective forms of monumentality appeared at most in the form of large fortifications or oppida.

5 rome ’ s im p act o n the s p atial orga n isatio n o f rural la n dsca p es It is a well-known phenomenon that empires can have a direct impact on the spatial organisation of rural landscapes. For the Roman empire, this would be the construction of an engineered road infrastructure, bridges, major waterworks and of course the limes. More contested, however, is the role of the imperial power in introducing a new system of land division. Roman archaeology has presented an almost stereotypical picture of large-scale redivision of land by the central government through a centuriation system. Centuriation implies the confiscation and reorganisation of land by Roman administrators. The model of centuriated landscapes, with its regular parcelling pattern, is associated above all with the founding of colonies and the accompanying land allocations to veterans in the Late Republic and Early Empire. Since the 1950s centuriation was often conjectured in the highly romanised villa landscapes of the northern provinces. The model of centuriated landscapes was based largely on writings of the Roman agrimensores, a supposed link between centuriatio and the settlement of veterans in coloniae, and of course the discovery in the 1950s of the marble cadastre of Orange.46 Reconstructions of ancient centuriation systems, based on analyses of aerial photos and recent historical-geographic data, cropped up everywhere.47 In recent decades, however, archaeologists have become increasingly sceptical as the many reconstructions of centuriation grids have proved hypothetical at best.48 Recently, large-scale excavations in the Netherlands, Flanders and the German Lower Rhine region have revealed growing archaeological evidence for ditch and parcelling systems surrounding rural settlements. The subject of Roman land

43

These were clauses such as Hoc monumentum heredem non

of Roman colonies in Gallia Narbonensis, Christol 1999;

sequetur, often abbreviated to HMHNS (‘this monument

for the cadastre of Orange, Piganiol 1962; Chouquer

is not to be passed down to the heirs’), which meant that the grave monument should stay in the family’s hands.

1983. 47

1983; idem 1987; Peterson 1996; Malvache/Pouchain

This formula occurs inter alia on the tomb of L. Poblicius

1994; Wiedemann/Antrop/Vermeulen 2001.

in Cologne. 44

Kremer 2009; Krier/Henrich, this volume.

45

Cf. Bodel 1997 for Roman Italy.

46

E.g. Mertens 1958; Edelman/Eeuwens 1959; Jacques 

48

Relevant is the recent methodological discussion in  Chouquer 2008, 869 ff.

For the agrimensores, see Campbell 2000; for the settlement of legionary veterans in the centuriated landscapes



division is now back on the agenda.49 The key questions are: what is the role of the central authority vis-à-vis that of local authorities and individual land owners? What contribution can archaeology make to this discussion? For the latter question, it is useful to first distinguish between centuriation and cadastration.50 Centuriation refers to a Roman system of land division, marked in the physical landscape and replacing an older pre-Roman division. We see cadastration as a system of land registration set up for fiscal purposes and which did not necessarily have any implications for the physical organisation of the landscape. A cadastre can also involve landscapes that continued the (pre-)Roman land division. Thus it is not clear whether we should regard the cadastre of Orange solely as a taxation map or whether the grid it features also served as the basis for a physical re-organisation of land. 51 As we have already indicated, the centuriation model is based on the situation in veteran colonies from the Late Republic and Early Empire in Italy and Southern Gaul. The question now is whether evidence for centuriation exists in the colonies founded in Gallia Belgica and Germania Inferior. We know three coloniae in this region: Trier, Cologne and Xanten. We can dismiss Trier because it was a Latin colony where no veterans were settled. We can also disregard Xanten because we have no information at all about Roman land division for the immediate hinterland of Colonia Ulpia Traiana, which was founded in about 100. This leaves the colonia at Cologne founded under Claudius, which was certainly associated with the settlement of retired legionaries.52 There are several microregions in Cologne’s hinterland that have been subject to intensive archaeological investigation in the context of modern lignite mining. Gaitzsch and Jeneson’s contributions show a landscape with an astonishingly dense distribution of predominantly small villas going back to newly founded settlements from the mid-1st century AD. However, the considerable variation in the orientation of plot ditches, settlement enclosures and local roads renders an overall centuriation system unlikely.53 But this may not apply to all of the Cologne territory. Veteran settlement may have been concentrated on the more fertile soils in the Rhine valley in the immediate vicinity of Cologne, where a centuriation system may well have existed. Possible evidence for this is a votive inscription from Cologne, erected by ‘the landowners from the vicus Lucretius in the first scamnum’.54 The term scamnum (‘strip of land’) is a surveying term pointing to the existence of centuriation, or to another parcelling system with strip fields. What do we know of the situation in the peregrine civitates and municipia? Here too, it is generally assumed that landownership was systematically inventoried and registered as part of a census. However, there is no hard archaeological evidence for centuriated landscapes. Nor is there historical evidence for a large-scale expropriation and allocation of land by the Roman authorities.55 We see there a high degree of continuity of native ownership relationships and should not therefore expect to find landscapes centuriated by Rome. There was every scope, however, for small-scale initiatives from below by private landowners seeking to rationalise the way they farmed, and sometimes by local civitas administrators.

49

Cf. the regional studies of Van Londen 2006; Vos 2009,

53

109-116; Heeren 2009, esp. chapter 8.6; De Clercq 2009; Bonnie 2009; Bourgeois et al., 2001. See also Clotuche

höfer 2005, 51. 54

See the methodological discussion in Ferdière 1983, 159 ff.

51

52



CIL XIII 8254 = RSK 158. Vicus Lucretius may have been a district of Cologne. The inscription can be dated

2009, 50-52. 50

Cf. Gaitzsch, this volume, figs 1 and 3. See also Rothen-

to the 2nd century AD. 55

Mattingly 2006, 359-360, arrives at similar conclusions

Chouquer 1983 and 2008, 869 ff., argues for the latter

for Roman Britain. Bonnie (2009, esp. fig. 47) recently

interpretation.The annual taxation for the cultivated land

reconstructed a Roman centuriation in the villa land-

is shown on the cadastral map of Orange. See Piganiol

scape of the civitas Tungrorum, but this too remains hypo-

1962.

thetical.

Tacitus, Hist. 4.65.

Of interest here are the recent studies of archaeologically documented ditch systems in the Batavian and Cananefatian area. We observe the regular presence of ditch systems around rural settlements from the end of the 1st century AD onwards, sometimes in combination with evidence of Roman measurements and measuring techniques.56 However, the ditch systems show a wide diversity in form and orientation and there is no indication at all of an overall regional system. It would appear that land division was carried out by individual settlements that already existed before. We can draw the same conclusions for rural habitation in the northern part of the civitas Menapiorum57 and the villa landscape of the Somme valley around Amiens. The villa complexes examined by Agache vary considerably in their orientation and seem mainly to be based on the local topography and the orientation of pre-villa settlements. Here too there are no signs of uniform land division.58 We know that cadastration was an important tool in the levying of taxes.59 Heeren sees a link between the creation of rectangular ditch systems in the later 1st century AD in the Batavian region and taxation by Rome.60 However, the Roman authorities are unlikely to have carved up the land anew for the sole purpose of levying taxes; we would then expect a more uniform land division with less diversity in form, measurements and orientation of the ditch systems.The construction of rectangular ditch systems around non-villa settlements in the Dutch River area may well have been stimulated by the taxation policy of the Roman authorities, but it is the local landowners who will have taken the final decisions. In the wetlands of the Cananefatian civitas we are confronted with the interesting subject of land reorganisation and water management from the end of the 1st to the 3rd century AD. Here we see the construction of supralocal systems of ditches, dams and culverts, which appear to traverse both clay and peat landscapes.61 The link with water management in this wetland region suggests some kind of central organisation, although not necessarily at civitas level. The Lex Rivi Hiberiensis from the time of Hadrian in Hispania Citerior may offer an interesting parallel.62 This document regulates the creation and maintenance of a regional canal irrigation system in the Ebro valley. Responsibility for organisation, however, lay in the hands of a coalition of pagi from two different civitates, with regulation occurring by consensus among the pagus inhabitants. The writings of Roman agrimensores make reference to a similar local autonomy within land management, in which a key role is set aside for magistri pagi.63 So what conclusion can we now come to about the spatial organisation of Roman rural landscapes? It is important to understand that centuriation in the form of a complete physical reorganisation of the landscape was rare and was confined to coloniae or some of their parts. It seems that Rome’s direct impact on rearranging rural landscapes has been overestimated. Certainly in the peregrine civitates and municipia, it was local authorities and the agency of individual landowners that played a major role. A specific challenge for the future is to deepen this discussion through continued archaeological research into ditch systems around rural settlements in different regions.

56

57

Bloemers 1978;Van Londen 2006; Heeren 2009, chapter

58

Agache 1978.

8; Vos 2009, chapter 3.7.

59

Cf. Piganiol 1962 for the cadaster of Orange.

Oral communication Wim De Clercq, Ghent. See also

60

Heeren 2009, chapter 8.

Clotuche 2009, 50 ff., about archaeological evidence

61

De Ridder 1999.

(remains of ditches) of a ‘Roman-indigenous cadaster’ at

62

Beltrán Lloris 2006.

Onnaing (Northern France).

63

Campbell 2000, 132.7 ff. (Siculus Flaccus).



6 to w n - cou n tr y side d y n amics ‘… As soon as I had escaped from the oppressive atmosphere of the city, and from that awful odour of reeking kitchens which, when in use, pour forth a ruinous mess of steam and soot, I perceived at once that my health was mending. And how much stronger do you think I felt when I reached my vineyards!’ Seneca, Epist. 104.6.64 In Roman times, societies in the northern parts of Gaul were confronted for the first time with urban forms of settlement and with the Roman urban model in particular. Although landed property formed the basis of political power, in the urbanocentric societies of the Roman empire the power centre lay in the towns, which held sway over the countryside through patronage networks, central place functions and fiscal administration.The elite were ‘urban-based’, but valued the countryside as the natural antithesis of the hustle and bustle and pollution of the town, as we see in the works of authors from senatorial elite circles.65 This dream of an idyllic countryside is an important ideological premise for villa construction by the elite.66 The town-countryside dialectic has long been a core theme in the discussion on villas in the Roman empire. Generally speaking, successful urban development relied on the dense distribution and wealth of villas. Conversely, a lack of urban development (e.g. in the civitates of the Menapii, Morini and Frisiavones) was associated with the absence of a villa landscape. Owners of large and medium-sized villas often belonged to both the rural and urban community; they had an urban mind-set and, as members of the decurial elite, frequently owned a house in town as well.67 Their situation contrasted with that of the large group who inhabited small villas. It is difficult to pinpoint the latter’s exact social status, but unlike large villa owners, they did not maintain direct political links with the town. The picture of an idyllic countryside also seems to have appealed to elites in the northern Gallic and Germanic provinces.69 Some texts by late Roman authors such as Ausonius and Sidonius Apollinaris (both of Gallic origin) are nothing short of odes to rural landscapes and to nature.68 Epigraphic data reveals that members of the urban decurial elite had villas in the countryside, in which they invested a substantial share of their wealth. This commitment on the part of upper social echelons to their rural residences was also reflected in a common desire to be buried there. Everyday scenes on funerary reliefs of the rural elite attest to their affection for rural life. Conversely, mosaics or frescos featuring gladiator scenes in the villas of Nennig (D), Orbe (CH) and Maasbracht (NL) bear witness to the close involvement and affinity of villa owners with urban life. However, changes over time are evident in the archaeological data. In the earliest phase the elite invested mainly in urban public works. It was not until the mid-1st century AD that they began investing in private residences in the town and countryside.70 Following the Italic model, the villa elite spent the winter months in town and sought out their rural residences during the summer.71 In her study of Roman houses, Hales points to a tension in Roman elite culture between the urban domus and rural villas.72 The domus was the setting for both the private, domestic life and the public role of a familia. In Roman literary sources, the villa is treated as secondary to the urban domus; the latter was 64 65

Translation by Richard M. Gummere (Loeb edition).

68

Dräger 2001, 125-126.

E.g. Seneca, Epist. 104.6; Pliny the Younger, Epist. 2.17

69

Derks, this volume.

and 5.6.

70

Woolf 1998, 124; Roymans/Habermehl, this volume. Cf.

66

Bentmann/Müller 1970, 9 ff.

67

In particular, the owners of larger villa complexes with

Creighton 2006, 149, for Britannia. 71

a pars urbana are likely to have had an urban residence as

ous country houses of the urban patriciate in Western

well. In archaeological terms, this is perhaps indicated by

Europe during the Late Middle Ages and early modern

the presence in large villa complexes of a second ‘minivilla’, which can be interpreted as the home of a vilicus.



We encounter the selfsame pattern with the sumptu-

period. 72

Hales 2003, 3, 8.

seen as the main seat of a family’s involvement in public life, while the former was more the symbol for the private domain.73 Without a domus a villa owner would be politically invisible in the town. However, we see that domus and villa in fact fulfilled a joint political function. ‘As the domus works in the city, so the villa fulfils a similar role in the country, framing the outsider’s impression of the owner and the owner’s experience of the outside world’.74 In the northern provinces too, rural villas were not solely intended for private repose and leisure. With their architecture and decoration, they were also the setting for a familia’s political role. Although the town was the principle political arena, we must not forget that the economic and political power base of elite families generally lay in the countryside. In all probability, villas played a key role in consolidating and reinforcing that rural power base. Villa owners achieved this by regularly receiving their clientele and political friends. The bronze patronage tablets found in the villa of an urban magistrate at Houthem (NL) provide concrete evidence for this.75 For elite families in fact, villas performed the same sociopolitical function as the domus in town. This gave elite families not only an idyllic and a functional reason for building villas (the dream of an Arcadian countryside and a need for repose and luxury), but also a political one. The town had a direct impact on the emergence of larger villas. Critical here is an analysis of the rich peristyle houses or domus in the civitas capitals of Northern Gaul and the Rhineland. Here we encounter many elements that are also found in the main buildings of large villas: a courtyard with an ornamental garden and a basin, a formal dining room (triclinium) with a mosaic, an elaborate bath suite and an arrangement of smaller rooms in both wings of the main house. In the large axial villa complexes we see a creative reshuffling of elements from both the urban domus and older native settlements, resulting in a new hybrid form.76

7 the house as a s y m b ol o f elite disti n ctio n An important pre-condition for villa construction, and one which also represents a fundamental break with Late Iron Age traditions, was the adoption by emerging elites of Roman ideas about the house as a medium for self-representation and social distinction. Although different regional house-building traditions existed in Late Iron Age societies within our study region, it is conspicuous that everywhere the house played only a minor role in marking differences in social status within local communities. In many settlements it has not been possible to identify elites on the basis of the excavated ground plans of houses.77 Almost everywhere these were small, simple post-built structures with the minimum number of rooms. Another striking feature was the lack of effort made to increase the durability of these houses. They lasted no more than one or two generations, which means that many individuals outlived the house in which they were born. In house building, there was a greater emphasis on expressing communality, transience and collective identities than on social hierarchy.78 This all changed in the Roman period thanks to new elites who employed their homes in strategies of social competition and elite distinction. Both town and countryside saw the appearance of large multi-roomed domus and villas, which marked a radical break with pre-Roman traditions in terms of

73

Hales 2003, 8, 21, 32-35.

74

Hales 2003, 54.

region in Northern France. Here too, however, there is little distinction in house architecture.

75

78

Derks, this volume.

76

Roymans/Habermehl, this volume.

holds in the southern Netherlands during the Iron Age,

77

See, however, Malrain et al. 2002, 137 ff., who try to

see Gerritsen 2003, chapter 3.

For an anthropological perspective on houses and house-

identify Late Iron Age ‘aristocratic’ houses in the Oise



their monumental architecture and interior decoration and which tell us about how elites conceived of their role and place in the Roman world.79 The house became a key element in the competitive display of status and prestige among upper social groups and a powerful tool in building up new positions of power and authority within the new circumstances of the Roman empire.80 Allied to this was a desire to make houses more durable in order to preserve them for future generations. Houses thus became the symbol of a family’s power and vitality.81 Having a large house also meant that guests could be received in worthy fashion, which was critical in the context of intra-elite competition. Daloz points out that having numerous, spacious rooms ‘also raises the issue of care and cleanliness and calls for the presence of servants – who may themselves play an important part in distinction strategies.’82 A further dimension of elite distinction through house building was domination over lower social groups. We see this most clearly in the architecture and layout of the large axial villas. With their division into a pars urbana and a pars rustica, separated from one another by a partition wall, and the hierarchical arrangement of residence and working compound, they were a concrete reflection of the social segregation that existed between the master of the villa and his dependents. The plain quarters in the wings of the working compound stood in stark contrast to the luxury and wealth of the master’s residential compound. Like the urban domus, the villa’s main house was a medium through which the family communicated with the wider community; it functioned as a locus for maintaining patron-client networks through such social rituals as the salutatio. The villa was thus an important means by which the master of the villa continued his political, social and economic control of the countryside.83

8 the eco n omic b asis o f v illa - domi n ated la n d sca p es The widespread emergence of villas during the 1st and 2nd centuries was not just the product of a specific style of consumption by upper and middle social groups. The construction of monumental Roman-style houses with all their amenities also presupposes the availability of considerable sums to pay for building materials and to engage a professional builder and architect. The question is where all that wealth in the form of cash money came from. There were two critical factors here: 1. The emergence of a large military market in the Rhine zone following the stabilisation and extension of the Rhine limes,84 and 2. rapid urbanisation from Augustus onwards, bringing with it an urban market. All of this created powerful consumer demand for agrarian products, resulting in a thriving rural economy. This was further

79

Hales 2003, 245.

be trebled if we consider the size of the military com-

80

Bentmann/Müller 1970.

munity in a broad sense, in other words including slaves,

81

Roymans/Habermehl, this volume.

the wives and children of soldiers, merchants, shopkeep-

82

Daloz 2009, 69.

ers, etc. This comes to an average annual consumption of

83

Slofstra (1983; 1995) points to the importance of villas

between 11,400 and 28,500 tonnes of grain, assuming

in the reproduction of clientele networks.Their architec-

a demand of between 200 and 500 kg per person per

ture and decoration (bath suites, mosaics, wall paintings,

annum. For the villa landscape in the Cologne hinter-

triclinia) indicate a socio-political use in accordance with

land, Gaitzsch assumes an annual yield of 1m3 or 800 kg

Roman elite culture. See also Hingley 2005, 88.

grain per ha, based on approx. 30 ha. of arable land per

For attempts to quantity the demand for grain for the

villa (Gaitzsch, this volume; Kooistra 1996, 98). Estimates

military markets, see Millett 1990a, 56-57; Kooistra

of this kind show that villa landscapes could easily have

1996, 66 ff., 96 ff.; Rothenhöfer 2005, 54-57. In the

satisfied the military community’s demand for grain. See

Lower Rhine limes the number of soldiers is estimated at

also Rothenhöfer 2005, 57.

84

19,000 in the Tiberian period. This figure can certainly



A B C

Voorburg

D E

Nijmegen

Rossum

Xanten

M

eu

ine

se

Rh

Sch

el

dt

Cologne

Maastricht

Tongres

Tournai

E

Bavai

E IF

L

S A R D E N N E

0

50 km

Fig. 4. The supply routes of grain from the villa landscapes in the loess belt between Bavay and Cologne to the military camps along the Lower Rhine.

stimulated by the development of a sophisticated transport infrastructure in the form of a network of land and waterways, and by the Roman taxation system. Although the high settlement densities, open landscapes and technology of Late Iron Age economies show that the latter were already producing at a high level,85 the growth in consumer demand and the taxation in the imperial era will have sparked a new cycle of agrarian expansion and intensification. In almost all regions where systematic surveys have been conducted, the Roman period has emerged as a phase of increasing settlement density and hence demographic growth. In particular the fertile soils of the large loess plateaus, such as in Northern France and the Cologne hinterland, became much more thickly populated.86 The economic basis of villas relied predominantly on the continuing conversion of an agrarian surplus into money via market exchange. There were no monocultures in the modern sense; except for the cultivation of crops for people’s own consumption, they specialised in one or more cash crops. For the villa landscapes of the northwestern provinces, it is generally accepted that these crops were grains, as indicated by the direct correlation between villa-dominated landscapes and the presence of fertile loess soils.87 Growing one or two grains, especially spelt and bread wheat, seems to have been the main source

85

See e.g. Haselgrove, this volume, Section 3.

86

Haselgrove, this volume; Gaitzsch, this volume; Jeneson,

87

Cf. Kooistra 1996, chapter 4; Roymans 1996; Bakels 2009, 167; Kreuz 2004.

this volume.

17

of cash money here. Significantly, the granaries found everywhere on villa sites – and often in non-villa settlements too – were much larger than those known from Late Iron Age settlements.88 In a recent study Georges Raepsaet drew attention to Roman-era agrarian technological innovations in the region between the Seine and Rhine.89 He concentrated in particular on what was probably an autonomous regional development – the introduction of the Gallic harvesting machine, which Pliny and Palladius called a vallus. These authors describe the reaper as being pushed by an animal and used in large fields on fairly level ground, with only the heads harvested and not the straw. The reaper is depicted on funerary reliefs from Buzenol, Arlon, Trier, Reims and Koblenz. The vallus appears to have been widely used on the large villa estates in the region between the Seine and Rhine and may have symbolised technological innovation and agrarian success in the northern loess landscapes. Although it continues to be difficult to establish its precise economic impact, the introduction of the vallus suggests an increase in scale and reduced labour input in grain cultivation.90 What also stands out is the high degree of administration involved in villa estates. Funerary monuments frequently feature scenes with men carrying writing tablets or keeping written accounts.91 The villa economy included other market-oriented activities in addition to grain production. Villas that owed their existence to wine production developed on the steep slopes of the Moselle valley in the Trier region. Relying on iconographic and historical data, various authors have emphasised the importance of the textile industry in the villa landscapes of both Northwestern France and the Trier region.92 This indicates that sheep farming played a significant role. Agache sees wool production as the logical complement to grain production at these villas, although wool may also have been the product of specialised farmsteads in environments that were less suitable for large-scale grain production, especially alluvial river valleys. The present consensus is that the fertile loess belt between Cologne and Northwestern France played a vital role in supplying the Lower Rhine army with grain.93 The army was provisioned via the Rhine, Meuse and probably also the Scheldt routes (fig. 4). The camps located between Bonn and Xanten received their grain via the Rhine.94 Cologne was without doubt a major transport hub, where grain arriving by road from the villa landscape west of Cologne will have been transferred onto boats. The limes camps from Nijmegen to the North Sea coast seem to have been supplied via the Meuse route. Maastricht will have played a key role here as a transfer point for grain from the surrounding loess belt on both sides of the Meuse as well as from Northern France. Evidence for this is circumstantial and consists inter alia of the discovery of large military granaries from the Late Roman period. Such grain stores (controlled by the military?) may also have existed here earlier.95 From the vicus near the Roman army camp at Vechten (NL), we know a votive inscription erected by nautae of Tungrian origin.96 Although we do not know

88

Native settlements in non-villa landscapes could also 

93

have produced a surplus. Cf. the traditional renown of Menapian hams (De Clercq 2006, 469) and archaeo-

Cf. Agache 1978, 356; Wightman 1985, 148-157; Roymans 1996, 59; Eck 2004, 428 ff.

94

The strategic importance of the Rhine for provisioning

logical evidence for specialisation in horse breeding

the Roman military camps with grain is evident from

and barley growing in Batavian settlements (Groot et al.

Tacitus’ report of the Batavian revolt. Cf. Tacitus, Hist. 26-27 and 35.

2009). However, given the material culture of the associated rural settlements, there appears to have been a fairly

95

Bakels/Dijkman 2000. The Maastricht find of sev

limited cash yield.

eral military roof-tile stamps from the Classis Germanica

89

Raepsaet 1995.

among others (Dijkman 2003) suggests a possible naval

90

Cf. Agache 1978, 357.

base and points to the strategic importance of the Meuse

91

Freigang 1997, 310. Cf. also Derks, this volume.

92

Wild 1970, 9; Agache 1978, 358 ff.; Drinkwater 1977/78; idem 1982; Roymans 1996, 64; Henrotay 2006.



for the army. 96

CIL XIII 8815.

what they transported, the most plausible cargo seems to be grain cultivated in the Tungrian villa belt and transhipped in Maastricht. In addition to the Rhine and Meuse routes, there may also have been a Scheldt route for transporting grain to the Rhine delta. This ran via the northern branch of the Scheldt (the later Striene) to the Meuse/Waal estuary, before cutting across to the Rhine via the Corbulo canal. One of the ships found in the Rhine near the Roman army camp of Woerden seems to have traversed this route.The vessel was carrying a cargo of emmer wheat, which given the presence of Orlaya grandiflora, a calcareous weed, was probably cultivated in the lime-rich southern loess landscapes.97 Included in the ship’s inventory were a number of cooking pots made in the Belgian coastal area, which gives an indication of the ship’s route. Supporting epigraphic evidence for grain transports from these parts to the Dutch river area comes from a votive altar erected at Nijmegen by a corn merchant (negotiator frumentarius) of Nervian origin.98 Lastly, it is worth mentioning that contingents of the Classis Germanica were also active on the Scheldt, as the presence of roof-tile stamps shows.99 In the villa landscapes themselves, grain produced by the villas appears to have first been centrally collected in large storehouses in the towns and vici.100 The precise status of these urban storehouses remains unclear. Were they public horrea belonging to the Roman authorities or to the civitas, earmarked for grain purchased on the local market and/or acquired through taxation? Were they private warehouses whose storage capacity was rented by the state or civitas? Or did they belong to large entrepreneurs, like the above-mentioned Nervian negotiator frumentarius M. Liberius Victor, who – hoping for an increase in demand or a better price – temporarily stored their trade goods there? Whatever their status, the horrea attest to the flow of corn surplus from villas to urban centres.101 Thanks to their high agrarian potential, the loess regions were ‘landscapes of opportunity’102 for their inhabitants, who could accumulate capital by producing grain for the market. The desire to avail themselves of such opportunities may also have harboured an element of competition and conspicuous display. The introduction of spectacular technological innovations (such as the reaper), large horrea, symmetrically laid-out working compounds and meticulous bookkeeping bestowed prestige on the master of the villa and was a source of pride. He would gladly have displayed these symbols of successful farming.103 The high villa density in the northern landscapes demonstrates that it was not only elites but middle-class farmers too who were able to accumulate sufficient capital over and above their taxes and who could therefore afford to build villas and funerary monuments.104 All of this suggests a high level of consumption in the villa landscapes between Cologne and Amiens, and that the tax burden in the first centuries cannot have been extreme. 97

Haalebos 1996, 487; Pals/Hakbijl 1992.

98

CIL XIII 8725.

99

Cf. Hanel 1998, 413 with fig. 8, and 414 with notes 130-131; Dijkman 2003. The find of two sandals with a

102

agree, however, with his suggestion of calling non-villa landscapes ‘landscapes of resistance’. 103

100

101

See Purcell 1995; Taylor, this volume; Habermehl 2011, ch. 5.

hobnail pattern in the form of Neptune’s trident leads Haalebos (1996, 493) to believe that the Woerden ship

For this term, see Mattingly 2006, 369, 522 ff. We do not

104

Nor should we underestimate the consumer demand of

was sailed by soldiers of the Classis Germanica.

lower social groups living in traditional non-villa settle-

For examples of large storehouses from the late 1st/2nd

ments (although there was considerable regional varia-

century, see Mertens/Vanvinckenroye 1975 (Tongres);

tion). This was surprisingly high in recently excavated

Matterne et al. 1998 (Amiens) and Vanderhoeven  et al.

Batavian settlements at Tiel and Geldermalsen, where

1997/1998 (Tienen). Cf. also the grain cargo with insects

people had switched completely to imported wheel-

in the Woerden vessel, which suggests that the grain had

thrown pottery in the Flavian period. Cf. Heeren 2009;

been stored for at least a year before being shipped. Pals/

Van Renswoude/Van Kerckhove 2009, 189. On possible

Hakbijl 1992, 295; Haalebos 1996.

tax payment patterns among groups in Belgic Gaul and

For a discussion on ownership and the use of storehouses,

Germania Inferior, see Roymans 1996, 84 ff.

see Dubouloz 2008.



9 the social orga n isatio n o f p roductio n (…) a well-excavated Gallo-Roman villa (…) can only with difficulty, and subtle argument, be made to provide evidence for social structure at any level beyond superficial clichés, and many archaeologists tend to rest content with a description of the physical remains. Wightman 1978, 98. Archaeologists generally adopt a cautious approach to the social organisation of production in villa settlements. This is because of the difficulty of finding tangible evidence but also because discussion of this topic has traditionally been dominated by ancient historians, who apply a villa model based largely on the literary output of the Roman senatorial elite.105 They sketch a picture of villas as centres of large estates (fundi), owned by rich townsmen who invested their wealth in land. The estates were worked by slaves and tenant farmers (coloni), who lived in the villa or in dependent farmsteads. In recent decades criticism has frequently been levelled at the application of this ‘Italic model’ to the northern provinces and at its underlying assumptions.106 The criticism centres on the highly normative representation of Roman elite culture and the associated romanisation discourse, which takes insufficient account of the regional diversity in villas and villa landscapes, including within Italy. It is also felt that this depiction pays little heed to either the archaeological evidence or to lower social groups and non-villa settlements. A further criticism is that little attempt has been made to link up with insights and concepts from the social sciences. In the interpretational framework presented below, we do our best to take these criticisms into account. Our starting point for analysing the social organisation of the villa system is the profound transformation of Northern Gallic societies following their integration into the Roman state system. The precise form that the social organisation of pre-Roman Gaul took is a matter of debate, however. The traditional picture, which we owe chiefly to Caesar, of a highly hierarchical society with a powerful warrior aristocracy at the top controlling large numbers of clients and slaves seems largely to be based on the situation of the Aedui. The archaeological data argues for a high degree of regional differentiation that also allows room for less developed social hierarchies in which collective power networks played a significant role.107 The communities central to this study belonged to this latter group. New power relationships arose here following the introduction of the Augustan administrative reforms. Land ownership constituted the basis of political power and this rested with an urban-based decurial elite, which probably comprised on average one hundred or so families per civitas.We should bear in mind broad changes in the nature of landownership and landholding.108 While the Late Iron Age saw some emphasis on collective, clan-based forms of ownership, in the Roman fiscal administration this was mainly linked to individual families.The granting of Roman citizenship to a limited number of families in the 1st century AD will also have had an impact on ownership relations. This is because citizens enjoyed the protection of Roman law, which placed them in a much better position than non-citizens when it came to safeguarding and extending land ownership. What then is our initial general impression of the social organisation of villa landscapes? It is important to understand that the dense villa distribution in the loess belt does not indicate a simple social divide between a small group of supermagnates and a large throng of landless poor. We agree with Greg Woolf that the large numbers of medium-sized and especially small villas points instead to a broad middle class

105

Cf. the discussion in Slofstra 1983, 85 ff.; Terrenato 2001,

107

18. 106

Mattingly 2006, 367-378;Terrenato 2001, 28-29. See also Wightman 1978, 98; Slofstra 1983, 85 ff.



Thurston 2009; see also Haselgrove, this volume, Section 4.

108

Cf. Wightman 1978, 102; Woolf 1998, 155; Mattingly 2006, 353 ff., for Roman Britain.

of landowning farmers.109 The key argument here is that members of this group will not have wanted to invest large sums in villa construction if this was not their own property but that of a landlord. A key question in the villa debate is the extent to which the emergence of villa landscapes was linked to a new social organisation of production. With the help of historical data from Italy, both archaeologists and historians have attempted to associate the appearance of villas with the introduction of a new ‘mode of production’.110 Slofstra, for example, refers here to a ‘villa system’ or a ‘villa mode of production’.111 A villa is seen as a large-scale agrarian concern, based on a system of dependent labour (slaves, tenants) and producing goods for the market.112 The lords of the villa controlled the labour potential of a rural working class – who are assigned the anthropological label ‘peasant’– via patron-client relations, tenancy relations and other forms of dependency.The villa mode of production is viewed as a product of the new structures of the Roman state system, yet it is also partly understood as a transformation of local pre-Roman social and economic structures. Thus Slofstra speaks of a gradual transformation from a class of ‘tribal farmers’ to ‘peasants’, the latter being defined as small farmers in state societies whose labour is controlled by a social elite. The historian Wickham also describes the socioeconomic organisation in the Roman period and Early Middle Ages in terms of a specific ‘mode of production’, although he does stress that there were many regional variants and no uniform system.113 Slofstra and Wickham’s models offer interesting interpretive frameworks for the social study of villas and villa landscapes, but their approach has also come under fire. Martin Millett and Greg Woolf have reservations about whether the villa mode of production constituted an independent category, suggesting that villas should be analysed primarily in terms of a specific ‘style of consumption’. ‘It is thus more precise to characterize the spread of the villa as an increasing taste for and use of Roman building materials, techniques and styles, rather than as the diffusion of a new social and economic form.’114 Just where the investments needed for this style of consumption came from is then of secondary importance. The money may have been generated by the villa estates themselves, but capital acquired from outside is another possibility. All the same, we must not underestimate the changes in the production relationships associated with the appearance of villas. Although there was no uniform villa mode of production, we can identify a significant combination of new characteristics: an agrarian enterprise aimed at surplus production for the market, and operating within a complex state system that safeguarded the social order. The specific way in which human labour was exploited varied, encompassing tenancy, wage-working115 and slavery. The production relationships differed substantially from those of the Late Iron Age. In villa landscapes we see more distinctive and complex asymmetrical relationships which, much more so than in the Late Iron Age, were aimed at economic exploitation. The degree to which certain families gained control over the labour and production of others seems to be significantly greater than in the pre-Roman societies in our study area. As outlined in Section 7, archaeology has a unique potential to explore the social backgrounds and implications of the adoption of new materials, techniques, forms and concepts in house building and settlement organisation. It can study the way the spatial order of villas and villa landscapes reflects and actively contributes to the shaping of social relations. Thus changes in house construction and the

109 110

Woolf 1998, 163-164.

113

Wickham 2005, 259 ff. and 465 ff.

Slofstra 1991; Wickham 2005 (who talks explicitly about

114

Woolf 1998, 148; Millett 1990a, 94, 97; idem 1995, 68.

the Marxist origin of the term ‘mode of production’). See also the discussion in Terrenato 2001, 18.

See also Martins 2005. 115

On the weak position of day labourers, who were de

111

Slofstra 1991, esp. 170 ff.

facto often worse off than slaves, see Weeber 2000, 243-

112

This implies a strict villa definition that encompasses 

247.

only large and medium-sized villas.



Borg

Levet

Anthée

Oberentfelden

0

50

100 m

Fig. 5. Some examples of axial villas from the northern provinces. See Roymans/Habermehl, this volume, fig. 1. Green: residential compound. Yellow: working compound.



restructuring of settlement space reflect an increasing hierarchisation of social relationships.116 There was an enormous difference in domestic culture between villa sites from the Roman period and settlements from the Late Iron Age, with a tradition of expressing status differences in house construction either undeveloped or absent in the latter. In particular, larger villa complexes with a spatially separate working compound reflect a high degree of landlord control over labour. Archaeological and epigraphic evidence also shows that the use of slaves was not uncommon in the northern provinces, although it occurred on only a limited scale (there was certainly no ‘slave mode of production’).117 Rather, we should assume that villa estates used a combination of tenants and slaves, supplemented by day labourers during peak periods such as harvest time.118 Below we identify four different categories of rural site, based on the situation in the landscape, spatial organisation and architecture, and we propose for each one a modelbased social interpretation. Firstly, there are large villas with a double courtyard (fig. 5). These frequently had a high degree of spatial organisation in accordance with the principles of axiality and symmetry. They featured a spatially demarcated pars urbana, a sign that the owners also had a house in town and were members of the decurial elite. The pars rustica was characterised by a fixed layout and strong uniformity of the secondary houses, as well as by the absence of storage structures, sheds or wells near these dwellings. This layout reflects a distinct social hierarchy with centralised control and organisation of labour. All products went directly to the master of the villa. The secondary houses were built by the landlord and inhabited by his servants and labourers who cultivated land that was directly exploited by the master.119 At harvest time on the estate, tenants (including external tenants) and day labourers may have been called in to provide additional labour. The work was coordinated by a vilicus or bailiff, whose house stood out from the rest through its greater luxury and its location right next to the pars urbana. The spatial layout of the villa complexes reflects the highly dependent position of the people living in the working compound and the formal, impersonal and distant relationship between them and their lord. Although we cannot establish their social status with any certainty, if there is anywhere where rural slavery can be expected, it is in this category of villas. It is no coincidence that the most compelling evidence for the use of slaves (a gravestone for liberti and a set of iron shackles) comes from an axial villa, namely the villa of Liestal-Munzach (CH).120 Secondly, we have small, post-built farmsteads situated in villa landscapes, whose significance seems to have been greatly underestimated (fig. 6). These farmsteads were presumably the lowest category of settlements within villa estates. The inhabitants had a greater degree of autonomy when it came to agrarian management practices; they ran their own cattle and had their own storage facilities in the compound.121 Unlike the families living in the working compounds of larger villas, they were not under any direct physical control. They probably worked tracts of land on the larger villa estates that were indirectly exploited by the landlord as tenant holdings, with the tenants paying rent and perhaps providing seasonal labour. Thirdly, many villa landscapes featured a dominant category of small single-compound villas (fig. 7) with a romanised main building, storage buildings and sometimes one or two secondary houses. The social interpretation of this category of sites is not absolutely clear-cut. We suggest that they were mainly middle-class farmers with their own property who controlled their own labour power and who main116

See also Habermehl, this volume.

117

Roymans/Zandstra, this volume.

120

Roymans/Zandstra, this volume.

118

Cf. Rosafio 1994.

121

Cf. the examples recently excavated at Onnaing, Bohain-

119

The axial villas could fit within a bipartite estate struc-

en-Vermandois and Bray-aérodrome (Northern France;

ture with both a core part that was directly exploited

Clotuche 2009; Ouzoulias 2010, 193-195; idem 196-

by the master of the villa and tracts of land that were

197) and Veldwezelt (Belgium, Vanderhoeven, oral com-

indirectly controlled as tenant holdings. Cf. the discus-

munication).

‘demesne’ in Roman villa estates.

sion in Wickham 2005, 263, 272 ff., on the presence of a



Fig. 6. Examples of post-built Gallo-Roman farmsteads (1st/2nd century) at Onnaing in the Nervian civitas. After Clotuche 2009, fig. 5.

tained no direct links with the town. The inhabitants of the secondary houses possibly had an unfree status; the absence of their own storage facilities might lead us to conclude that their labour power was directly exploited by the occupant of the main house. However, the social distance between them and the inhabitants of the main house was much smaller than in the double compound villas. Relationships were asymmetrical, but less formal and more personal than in the large axial villas. There may also have been cases, however, in which the houses within a small villa belonged to related families. Fourthly, the non-villa landscapes of the Rhine/Meuse delta included many native settlements consisting of a small cluster of traditional byre houses (fig. 8). The farmsteads functioned as relatively inde-



Ha 69

0

50 m

Ha 403

Fig. 7. Ground plan of two small villas in the hinterland of Cologne. After Heimberg 2002/2003, fig. 15.

pendent production units with their own cattle and often a granary. The differences in wealth between the farmsteads remain limited. Presumably, these were communities of predominantly free farmers who emphasised communality in the settlement layout and the associated collective cemeteries.122 The different households may even have been linked through kinship ties. These were frequently ‘military’ families who supplied soldiers to the Roman army from one generation to the next.123 Rural archaeology has its limitations, however. Perhaps the chief of these is our lack of understanding of property relations between settlements. Above we made a distinction between large or medium-sized villas, small villas in the form of single farmsteads, and small non-villa settlements (often described as ‘native’). But to what extent did dependency relationships exist between these categories? Did the larger villas function as centres for estates that controlled smaller settlements or were these smaller settlements owned for the most part by small independent farmers? It is important that we do not assume a uniform social organisation, but remain alert to substantial regional differences (see below). It is only through unique documents that we can catch an occasional glimpse of the ownership relationships of the time. For example, the texts of several bronze tabulae from the small villa of Houthem (NL) show that the villa belonged to a magistrate in the Colonia Ulpia Trajana/Xanten. It is unlikely, however, that this modest villa was his main residence.124 If the ownership relationships of a given moment remain fairly elusive for archaeologists, we have an even harder job coming to terms with the subject of inheritance and transmission of landed property, and hence the dynamics of property relationships.125 We have a general picture of the legal frameworks of land ownership and land transactions in the Roman legal system. Roman law gave landowners considerable flexibility for strategic action when regulating inheritance, which ran through both the male and female

122

123

However, finds of leg-irons in native settlements at 

124

See the discussion in Derks, this volume.

Houten and Someren may indicate the presence of slaves.

125

The question is whether alterations to smaller villas in

Cf. Roymans/Zandstra, this volume.

particular may be seen as reflecting a change of owner-

Roymans 2009b.

ship.



1

0

100 m

2

3

Fig. 8. General ground plan of two native-Roman settlements and a collective cemetery at Tiel in the Batavian river area. After Roymans/Derks/Heeren 2007, fig. 9. 1 settlement Oude Tielseweg; 2 cemetery Passewaaijse Hogeweg; 3 settlement Passewaaijse Hogeweg.

lines, and there was no principle of primogeniture.126 This means that property relations through inheritance and marriage were subject to a constant dynamic. For this reason, we should not view villa estates as static entities, but as ones that could increase or decrease in size. We should also bear in mind elites that owned several estates, after the example of the Roman senatorial aristocracy (the villas of Cicero and Pliny) and the Streubesitz of the Merovingian villa elite. What is remarkable here is the growing role played by wills as legal instruments for regulating the transgenerational transmission of landed property.127 126

For inheritance via the female line, see Gardner 1986, chapter 9.



127

See Section 4 above. As yet, there has been no systematic study of this topic.

The wealthier the family, the more likely it was that the pater familias would draw up a will. A unique example is the text of the ‘testament of the Lingon’, handed down to us via a medieval manuscript.128 Although the situation remains far from ideal and opinions are at times divided, we are convinced that regional and microregional studies into villa landscapes will enable archaeologists to come to interesting conclusions on the points raised above. Here are a few examples. In the Southern Netherlands there is evidence to suggest that it was returning auxiliary veterans who initiated the use of Roman building materials and architectural elements (roof tiles, wooden portico, stone cellars) in native rural settlements. And few true villas here also seem to be linked to veterans of native origin, this time at officer level.129 In the Swiss Aare valley Schucany sketches a model of a hierarchically organised villa landscape, with large axial villa complexes seen as centres of larger domains to which medium-sized and smaller farmsteads located higher up the slopes of the Jura Mountains belonged.130 The loess landscape of the Cologne hinterland, discussed in the contributions by Gaitzsch, Jeneson and Habermehl, had a high density of small villas that evolved out of new settlements from the first half of the 1st century AD. The limited social differentiation between these settlements argues for a picture of small, independent owners with lands assigned in specific allotments to colonists, whether retired soldiers or settlers from Gaul or Germany.131 A point of interest in the villa landscapes of the Trier region is the relationship between villas and vici. The large villa complexes of Borg and Reinheim were located right next to a vicus.132 In the vicus of Bliesbruck, situated several hundred metres from the villa of Reinheim, a large block of strip houses appears to have been built all at once; Sărăţeanu-Müller suspects that the local villa owner was directly involved here in providing accommodation for tenants or day labourers. The general picture emerging from these examples is that of considerable regional variation in the social organisation of villa landscapes, and it is especially at this regional level that questions about the social organisation of production can best be answered. Upper social groups/elites: • old aristocracy • new rich • men with a military career Middle-class groups: free farmers with their own land Lower social groups: • tenants • wage workers • slaves Fig. 9. Simplified scheme of the social composition of Roman rural society in the northern provinces in the first centuries AD.

We can conclude from the above that the social organisation of the villa world was complex and differentiated (fig. 9). It is important to understand that villa owners did not constitute a homogeneous social group. Although the owners of large and medium-sized villas belonged to the upper social echelons, they had a range of origins. They included members of old aristocratic families who succeeded in consolidating their social position, representatives of the new rich who had earned vast sums through commerce and who invested it in land, and ex-Roman army officers who had secured important posi-

128

CIL XIII 5708; Sage 1991.

129

Roymans, this volume; Vos 2009, 237-251.

130

Schucany, this volume.

131

Gaitzsch 1986; idem, this volume; Lenz 1998; idem 

1999a; Heimberg 2002/2003, 69 ff.; Rothenhöfer 2005, 42, 47. 132

Birkenhagen, this volume; Sărăţeanu-Müller, this volume.



tions and wealth as a result of their military career. Then there were the owners of the sizeable category of small villas; exceptionally they belonged to the upper social strata, but the vast majority will have been middle-class farmers who were able to embellish their homes little by little. The lower social groups of rural labourers in the villa landscapes were also fairly heterogeneous in composition and comprised tenants, wage workers and slaves. Some lived in villa complexes, some in other rural settlements, and perhaps in the vici as well.

1 0 co n sum p tio n , sel f - re p rese n tatio n a n d li f est y le o f v illa o w n ers As outlined in Section 7, Roman villas had a dual function: they were used not only to create wealth but also for the purpose of displaying social difference. Villas were an expression of a new style of consumption by upper social groups in which self-representation and the display of a specific lifestyle and its associated values were paramount.133 According to the sociologist Daloz, ‘[w]hat is critical for elites is to demonstrate their dominance vis-à-vis subordinates and to exhibit at least as much supremacy as their peers.’134 There is a wealth of evidence that allows archaeologists to expand on this theme of elite distinction and identify general patterns. The 1st century saw the appearance of a new repertoire of consumer goods, which included villas. For the first generation of villa builders, this may have implied a decision to identify themselves with the culture and lifestyle of the Roman rulers, since it represented a radical break with the traditions of their forebears in the area of house construction. Building villas was a new trend that granted access to new power networks, as well as creating opportunities for marking social boundaries with lower social groups.135 Breaking with ancestral conventions in the domain of house construction probably became the elite’s way of asserting themselves over others. This social differentiation was achieved through the spatial dimensions, topographical situations, distinctive architecture and interior decoration of residences, all well-known instruments in distinction strategies. Roman villas (and urban domus) may be regarded as manifestations of a new strategy of elite distinction, with villas representing a claim to superiority and symbolising superior social rank. For tenants and slaves, villas symbolised their relationship of dependency on their patron and master; only on special occasions were they permitted to enter the pars urbana through the gatehouse. Thus the villa lifestyle served not only to make everyday life more comfortable and enjoyable for villa owners; it was also a symbolic expression designed to authenticate the social standing of upper social groups. The themes of elite distinction and self-representation were also reflected in the burial evidence of upper social groups in the villa landscapes.136 There are different types of data at our disposal: funerary monuments, inscriptions, iconographic data and grave inventories. Each has its own characteristics, but there are four commonly recurring semantic fields which were probably all related to the elite’s social practices: a. Bathing. All large and medium-sized villas boasted a bath suite, an indication of the importance of bathing culture among villa owners. The graves of privileged individuals regularly yield bathing and toilet requisites, such as oil flasks, strigiles and mirrors, pointing to new ideas about bodily appearance and physical beauty as significant components of social distinction. b. Hunting. A popular pastime among villa owners,137 hunting also found expression in the burial ritual, as shown for example by the hunting weapons found in the grave of the man of Bocholtz. In the 133

134



For a consumption perspective on villas, see Millett 1995,

135

Cf. Woolf 1998, 156-157.

68; Woolf 1998, 148; Martins 2005, 132 ff.

136

Crowley 2011; idem, this volume.

Daloz 2009, 69.

137

Cf. Trinquier/Vendries 2009; Lepetz 2009.

testament of the Lingon, the testator stipulated that his hunting weapons should be placed with him on the funeral pyre.138 c. Drinking and dining. Guests for a symposium or dinner party were received in the most formal room in the villa’s principal buildings.139 Drinking vessels and dinner services in the grave inventories of villa owners bear witness to the profound cultural significance of the symposium. The convivial dinner party was also a social context bringing together a host of other facets of elite distinction that have left little trace in the archaeological record: arriving guests displaying their magnificent chariots, clothing and jewellery, the consumption of culinary specialities and expensive imported wines, the display of good manners and linguistic competence, and the keeping of well-trained servants.140 Symposia thus combined two symbolisms: they marked vertical social boundaries with lower groups (which were excluded from participation) on the one hand and horizontal relationships among peers on the other. d. Writing, reading and literacy. Linguistic and literary competence, as expressed in Ausonius’ poems, was a valued aspect of Roman elite culture and was commonly referred to in villa architecture and furnishings. Examples include the recently discovered mosaic of the muses in the villa of Vichten (L) and the busts of philosophers surrounding the water basin in the pars urbana of the villa of Welschbillig.141 Here the villa presented itself as the place where the muses and the arts blossomed. We regularly encounter writing materials in grave inventories, and it is remarkable how frequently writing tablets feature on grave reliefs. We can assume that larger villas had their own library.142 A lost grave relief from Neumagen showed a slave taking a roll of papyrus from a cupboard filled with scrolls (fig. 10).143 Each wall in Room J of a large, sumptuously decorated villa at Mané-Véchen in Brittany featured niches that have been tentatively interpreted as library cupboards.144 Adopting a new strategy of social distinction was not a matter of choice for elites. There is a link with the power play within Roman imperialism. Families wishing to develop, maintain or strengthen their positions of local power within the new Roman regime simply had to embrace the cultural codes of the Roman victors. Failure to do so, or doing so only half-heartedly, entailed the risk of relinquishing their power to upcoming new elites. The new styles of consumption including the building of large and medium-sized villas may thus be explained by a process that Millett described as competitive emulation among upper social groups.145 The question is, however, to what extent this emphasis on emulation and the adoption of a Roman identity can account for cultural changes further down the social hierarchy.146 In the 2nd century, a ‘consumer boom’ occurred through the spread of purchasing power to middle social groups who also began investing in their homes by building small villas.Three mechanisms may explain the development of the simple stone-built single farmsteads that began to spread from the late 1st century onwards. First, peer competition between the owners of these small villas may have been of prime importance rather than emulating the example set by the decurial elite. Not only was it unrealistic to measure themselves against the houses of the ruling elite, which were beyond their financial means, they also hardly

138

De Groot 2006, 43 (Bocholtz grave); CIL XIII 5708

textile business or bookshop. Von Massow 1932, 243 f.;

(testament of the Lingon). 139

Slofstra 1995, on triclinia in villas in the Roman West.

140

Daloz 2009, 78.

141

Vichten: Krier/Reinert 1995; Krier 2002. Welschbillig:

142

143

Blanck 1992, 181; Balensiefen 2002, 106-107, 111. 144

Provost 2007, 91 ff. An alternative interpretation is that the niches served as an archive. For the villa and its exceptionally rich decorations, cf. Boislève 2010.

Cüppers 1990, 665-667.

145

Millett 1990b, esp. 38. For the model, cf. also Brunt 1976.

On the difficulty of identifying them in archaeological

146

For critical remarks on emulation as an overall explana-

terms, Blanck 1992, 182 f.; for private libraries in the Italic

tory framework, James 2001, esp. 200-202, 205-206;

villas of a series of classical authors, Blanck 1992, 152-160.

Alcock 2001, esp. 229; Mattingly 2004; idem 2006, 15;

The relief has also been interpreted as a reference to a

idem 2011, 203 ff; Hingley 2005, 40-48 , esp. 42-43.



Fig. 10. Lost fragmentary funerary relief from Neumagen showing a servant taking a volume from a bookshelf. After Von Massow 1932, 243, Abb. 141.

mixed in the same circles. If the middle class farmers of the small villas did incorporate elements from the homes of the elite (such as wall paintings, a porticus, building in stone), their lower rank and status will have been immediately apparent through their use of cheaper building materials, simpler architecture, the lack of a bath suite, etc. Second, if the small villa could not compete with the large and medium-sized villas of the ruling elite, it may have been distinctive enough to function as a social demarcation with lower social groups. Third, the practical advantages of stone building in terms of durability and upkeep may have been at least as important to the owners of these small villas as the intention ‘to aspire things Roman’.147 While in the early 1st century the new building techniques represented a true innovation, by the later 1st century, specialised labour forces could be contracted in every civitas capital and stone building per se did not necessarily have an elite connotation. Pragmatic considerations may have become more important than ideological ones.

1 1 religio n a n d the structuri n g o f rural li f e Villa landscapes can also be characterised as ‘religious landscapes’, as they are often dotted with larger sanctuaries, small shrines and a range of funerary monuments. This points to a deep-seated cosmological anchoring of villa landscapes, which therefore cannot be understood purely in social or economic terms. Here we will briefly discuss the religious culture of the villa world, in particular the role of religion in structuring the social order and the daily life of rural communities. Surprisingly, this is a theme that has been much neglected in the literature. Clues to the religious life in villa compounds themselves can be found in the form of votive offerings, Jupiter columns and small shrines. Many villas have yielded bronze or terracotta statuettes representing divinities. In the absence of proper contextual information, it is generally assumed that these were erected in a lararium in the main building. The choice of deities was governed by the needs and preferences of the house owners, with almost no evident differences between lararia in Italy and the Roman provinces.148 Especially popular 147

Millett 1990b, 38.

148

Kaufmann-Heinimann 1998, esp. 192 f, with fig. 139.



were Mercury, Jupiter, Minerva, Mars, Victoria, Venus and Fortuna, who could stand for the economic prosperity of the estate, the protection of the house and the reproduction of the family. Isis and Harpocrates are also represented in the cast of deities from the villa of Vallon (CH).149 In the Roman world, the Lares – the actual household gods, depicted as a pair of young men clad in short girdled tunics – were the ultimate embodiment of a family’s fortunes and group identity. If the family moved house, the Lares moved with them. For families that owned several houses, the Lares were located in the main estate.150 The Lares show clearly that, as an identity group, the family had not only a social dimension, but also a religious one. The iconography of the Lar, especially the socially unmarked tunica which was also worn by slaves, highlights the fact that the family was a social construct, encompassing not only the master of the villa and his kin, but dependent servants as well.151 Whereas the lararium was the spatial context in which the inhabitants of the main building conducted their private, personal worship, Jupiter columns and small sanctuaries serving a wider audience are often found outside the residence. In their iconography, Jupiter columns are a clear illustration of cultural hybridity in the Germanic provinces152 and should probably be primarily understood as public expressions of gratitude for the pax romana and for the benefits this brought the master of the villa. Given that few examples could be examined in situ,153 we have yet to determine the extent to which they were real cult places where dedications continued to be made after the monument was inaugurated. This is certainly true of the simple shrines often found in villa compounds. These generally displayed a modest material culture – expensive monumental votive offerings are rare – and they seldom occupied a prominent place. We often encounter such shrines in the central area of the pars rustica of large axial complexes,154 but they are also known from medium-sized and even small villas.155 One question that has been little explored until now is how we should envisage the cult community of these shrines. Did the master of the villa play a key role? Did the shrines serve to moderate the vertical hierarchy between the master and his staff? Or were they used primarily by the labourers living in the villa compound who were dependent on the master? Of interest here is the concentration of pottery sherds from the Voerendaal shrine. Most come from simple terra nigra pots, four of which carry graffiti with the names of individuals, probably representatives of the lowest social group within the villa community.156 Funerary monuments belonging to the villa elite were a special feature of the religious landscape. The spatial relationship between dwelling and monument has yet to be properly examined, but a case study in the Trier region shows that orientation towards a nearby waterway or road was more important when choosing a site for these monuments than their visibility from the villa. This underlines once again that the representative function of funerary monuments was paramount.157 The fact that we seldom encounter prominent funerary monuments right next to one another could suggest that they reflect a single generation of villa inhabitants, and it is not at all clear just how and where the remaining members of the family were buried. The grave complex at the villa of Newel is an exception here. Not only were four tumuli and the foundation of a sizeable funerary monument found within a walled yard, but also the foundations of a Gallo-Roman temple. The graves of the dependent labour force which was not part of the lord’s family have been the subject of little serious study. 149

Kaufmann-Heinimann 1998, 285 f. (GF83).

154

Ebnöther 1995.

150

Dig. 25.3.1.2.

155

For a medium-sized villa, see the example of Voerendaal

151

Foss 1997; Bettini, in press. Cf. Plautus, Aul. 1-8; Cicero,

(cf. note 156); for temples from small villas, Gaitzsch

Leg. 2.27. For the quick adoption of the Lares’ imperial iconography, Boschung 2003, 10 f. 152

Woolf 2001.

153

For a recent example, see BJ 207, 2007, 325 f. (villa HA

2005. 156

Willems 1987, fig. 3; Willems/Kooistra 1988, 145-146, fig. 11.

157

Krier/Henrich, this volume.

488). For Jupiter columns from villas generally, KaszabOlschewski 2003 and Noelke 2005.



Lastly, for the sake of completeness, we would like to look at possible relationships with local sanctuaries located outside villa compounds that villa communities may have taken part in. An interesting case is the matronae cult in the villa landscape of the colonia of Cologne. A dense network of matronae sanctuaries is known from inscriptions and a handful of excavated examples. Recent studies have revealed that these sanctuaries – despite the distinctly native names of the ancestral mothers referred to as matronae or matres – were newly founded in the Roman period and were perhaps linked to a reorganisation of the landscape initiated by the new colonia.158 Epigraphic evidence shows that the organisation of the cult was the responsibility of curiae.159 The question of their social composition and who led them has yet to be addressed. Another research topic still to be tackled is the possible role of sanctuaries in strengthening social cohesion in the countryside. Inscriptions, our main source of information, show that the lion’s share were erected by people with Roman citizenship. However, we need to consider the extent to which other people also visited these sanctuaries and whether these cults may have played a role in strengthening the social cohesion of rural communities.

1 2 the im p act o f roma n militar y culture Given the location of the northern provinces in the frontier zone of the empire, it seems obvious to ask about the impact of the Roman troops stationed there and their culture on the development of villas and villa landscapes. Earlier, in Section 8, we argued that the Roman frontier army provided a substantial market for agrarian products, thereby stimulating the economic development of the countryside. Here we wish to focus on another aspect, the role of veterans as cultural brokers. We archaeologists are familiar with the model, chiefly connected with the founding of coloniae, whereby the Roman authorities in the northern provinces expropriated land on a large scale and reallocated it to legionary veterans. However, to date we have no concrete evidence for large-scale colonisation by veterans. Only in the lignite mining area of Cologne’s hinterland is a colonisation model plausible, although researchers are doubtful whether this involved veterans or civilian settlers of Gallic origin.160 Almost all other regions present a predominant picture of the gradual development of native settlements into villas.161 And yet we may be underestimating the impact that veterans had on the Roman countryside. While it is true that epigraphic data often shows veterans of foreign origin settling in towns and close to army camps rather than in the countryside, there is also a pattern of locally recruited auxilia veterans returning to their settlement of birth. However, we must be aware here of substantial regional differences.162 In the Lower Rhine region, and particularly with regard to the Batavians, there is a wealth of archaeological evidence from the 1st and 2nd centuries for returning veterans. This mostly takes the form of militaria from rural settlements.163 Intense interaction with the army led there to the formation of a typical ‘frontier culture’. This was characterised by a marginal impact of civil Roman architecture in the form of villas. More than 95% of sites were native-type settlements made up solely of byre houses. By contrast, Roman military culture had a surprisingly major impact in the form of mobilia: weapons, military horse gear, jewellery, bronze tableware, seal boxes and writing materials.164 Although we do see the emergence of specific hybrid farmhouses with wooden porticoes and roof tiles, by and large these appear to have been inspired by military architecture and for this reason have recently been labelled ‘veteran

158

Derks 1998, 119 ff; Scheid 1999; Eck 2004, 496-503.

162

Cf. Derks/Roymans 2006; Roymans, this volume.

159

Rüger 1972.

163

Nicolay 2007.

160

Cf. Lenz 1999a; Heimberg 2002/2003, 69-77, and the

164

Roymans/Derks/Heeren 2007; Heeren 2009; Roymans,

discussion in Roymans, this volume, section 7. 161



Cf. Bayard/Collart 1996.

this volume.

farmhouses’.165 We can also assume a military connection for the few ‘real’ villas, perhaps men who had held an officer’s rank in the Roman army. In the Trier region too, there are indications from the Early Roman period for the influence of veteran soldiers on the emergence of villas. Significant examples are the remains of 1st century AD stone funerary monuments featuring cavalry scenes and several inscriptions found in the immediate vicinity of villas; both point to men having ended their military careers as officers.166 The villa landscapes of Northern France, on the other hand, have yielded scarcely a trace of military links through returning veterans. Contrary to the commonly held view, we must conclude that almost nowhere can a direct relationship be established between veteran soldiers and villa construction. The building of villas is evidence of a growing appreciation of civilian lifestyle and culture, and the direct impetus for this came primarily from the towns. It is true, however, that an army career smoothed the transition to the civilian culture of upper social groups.

1 3 regio n al v ariatio n i n v illa la n dsca p es We have repeatedly emphasised that Roman villa landscapes did not form a homogeneous whole in terms of either their physical appearance or social organisation. It is important to understand the heterogeneity and regional variation of rural landscapes, and villa landscapes in particular. Below we will briefly outline regional patterns in house building, settlement morphology and burial traditions as this will then help us to explain this variation. We will also tie in with the discussion on globalization. This diversity offers a clue for understanding cultural identity constructions of groups within the context of the Roman empire. The first point of difference emerges if we make a rough distinction between villa-dominated landscapes and non-villa landscapes, as shown in Figure 1. The holocene Rhine/Meuse delta and the pleistocene sand and clay landscapes of Flanders and the southern Netherlands, as well as the hilly landscapes of the Ardennes and Eifel, belong to the latter category in which villas are scarce. The considerable regional variation in house construction and settlement layout in these non-villa landscapes was due largely to a strong continuation of pre-Roman traditions and practices. As the case of the Batavians shows, this does not automatically mean that these groups were only marginally integrated into the Roman world.167 Quite the opposite – intense interaction with the army led here to a powerful articulation of rural and military identity groups and the formation of a typical ‘frontier culture’. It is surprising, however, to note that regional variation was just as great in villa landscapes, despite the presence of powerful homogenising trends. Everywhere, the formation of villa landscapes went hand in hand with the cultivation of local cultural identities. Thus we see significant distribution patterns for certain villa categories. The contribution by Roymans and Habermehl points out the popularity of axial villas in Northwestern France, Switzerland and the Trier area, while villas with a loose arrangement of buildings within a single enclosed courtyard (Streuhofanlagen) predominate in other parts.168 Also of interest are regional patterns in the funerary monuments erected by villa owners.169 We have already mentioned the ‘barrow landscape’ of Central Belgium, while stone funerary monuments are conspicuously represented in the Rhineland zone east of the Meuse. By contrast, there appear to be almost no monumental graves in the villa landscape of Picardy. How do we explain this regional variation? There were no doubt several factors at work here. Firstly, landscapes had varying potential for generating wealth, for example the presence or absence of fertile 165

Vos 2009, 237-251.

168

Lenz 1998; Heimberg 2002/3.

166

Roymans, this volume.

169

Crowley, this volume.

167

Roymans 2009a and b; Groot et al. 2009.



loess soils. There was also the influence of different regional histories or different forms of taxation by the Roman authorities. Furthermore, there was the specific agency of legionary and auxiliary veterans, which in the Rhine zone was evident in the powerful influence exerted by Roman military culture on the rural world. Above all, however, the considerable regional variation of villa landscapes was determined by the need of communities to cultivate local identities. Hybridity offers an attractive interpretive framework here. According to Eriksen, ‘Hybridity directs attention towards individuals or cultural forms that are reflexively – self-consciously – mixed, that is, syntheses of cultural forms or fragments of diverse origins’.170 Individuals and groups felt a need to emphasise their local character and identity within the new context of the Roman empire. They frequently used old and new cultural expressions to arrive at new creations. Underpinning all this was the need to embed new imported cultural forms in local realities and meanings. Thus unlike neighbouring regions, the villa area around Tongres in Belgium was a true ‘barrow landscape’. Crowley argues that this was not a simple continuation of a Late Iron Age burial tradition; she views tumulus graves as the atavistic response of a rural elite seeking to construct their own identity within the wider Roman world.171 Regional variation was apparent not only in the rural burial practices and in villas, but also in the urban domus. In her analysis of elite town houses within the Roman empire, Hales identifies regional patterns which she sees as local interpretations of a new, ‘Romanised’ cultural language that is expressed in new hybrid creations.172 Regional variation in both town houses and villas arose out of a continual negotiation between local and Roman. As in modern-day society, the cultural phenomena of localism and ‘cosmopolitanism’ were not mutually exclusive during Roman times.

1 4 lo n g - term tra n s f ormatio n s In Northwestern Europe the Early Roman period was a time of profound social change as a consequence of increasing connectivity and integration in the Roman empire, with its law system, urban centres, markets, monetary system and advanced road infrastructure. In many respects the emergence of villa landscapes marked a break with Late Iron Age trends and traditions. Society became increasingly hierarchical, and there was an accompanying trend among the elite and middle social groups to give voice to their privileged social status by adopting a new style of consumption. The prehistoric tradition of building vernacular houses gave way to the construction of more durable, multi-roomed houses, which were built by professional architects and builders. Villas became a symbol of the status and identity of their owners. A second fundamental transformation occurred in the later 3rd and the 4th century when villa landscapes came to an end, literally becoming landscapes of ruins.173 There is a huge body of evidence for discontinuity in the later 3rd century. In almost all regions we observe a severe decline in habitation, as the regional studies presented in this volume also show.174 The widespread demolition of stone funerary monuments in the Late Roman period and their reuse as spolia suggests that old families’ claims to villas had elapsed and that the sites had been abandoned or populated by new inhabitants. In the few villas that continued into the Late Roman period, we observe a growing emphasis on defence through the construction of burgi. However, archaeological research does reveal considerable regional variation in this dynamic, although this needs to be studied in greater detail.175

170 171

Eriksen 2007, 172. See also Burke 2009. Crowley 2009; idem, this volume.

172

Hales 2003, chapter 6, and 181, 246.

173

Effros 2001; Christie 2004, 15 ff.



174

Cf. the contributions by Gaitzsch, Schucany and De  Clercq in this volume; Rothenhöfer 2005, 42.

175

See Van Ossel/Ouzoulias 2000, and the more general discussion in Wickham 2005, 473-481.

A villa system evolved once again in the study region during the Merovingian and Carolingian periods. Similarities with Roman villas are evident above all in the social organisation of production, in the sense that this too involved a system whereby a landlord elite used both free and unfree labour to extract surplus. However, there were also fundamental differences between the Roman and early medieval villa systems.176 The latter entailed a different legal context, did not have market-based production and lacked the town-country interaction that was characteristic of the Roman era, with all its economic, political and ideological connotations. It is therefore hardly surprising that the physical, material manifestation of early medieval villas also differed fundamentally, and that almost nowhere was there a direct continuity from Roman villas. It is not until the Late Middle Ages that we once again encounter an urban-based elite displaying their newly-won social status in their rural residences. Only in the late 13th/14th century did our study region once again feature local and regional elites with residences in both the town and countryside who presented themselves as mediators between the two worlds.177 From the 14th century onward, we see wealthy families from Den Bosch and Antwerp building rural residences in stone, linked to the agrarian exploitation of the land through tenant relationships. However, this elite lived in moated castles with an emphasis on defence, thereby continuing a feudal tradition. Influences of the Roman villa system are nowhere to be seen, in terms of neither material culture nor ideas. These influences are only evident in Western Europe from the Renaissance onwards when urban patricians became inspired by classicism. In Italy the feudal tradition of fortified castles from the early 16th century made way for a new tradition of villas whose architecture and social space were inspired by classical antiquity. This tradition was supported by urban elites who began investing their wealth in the countryside. They were driven not only by economic motives, but also by cultural and ideological ones. They sought to revitalise the classical dream of an Arcadian countryside and of agrarian production and landownership as a solid basis for elite power.178 This tradition was spread by Italian and Spanish architects to more northerly parts, where it set the tone on the landed estates of the elite in the 17th and 18th centuries. Country villas that served as summer residences for urban elites, with no attached land exploited for agrarian purposes, are a more recent development from the 18th century and later, one which bridges the gap to modern-day villa construction.

1 5 v illa la n dsca p es a n d the f uture o f roma n archaeolog y The picture we have presented of the formation of villa landscapes reveals a profound transformation of the Roman countryside in the northern provinces. Stimulated by population growth and consumer demand from the urban and military community, a new cycle of agrarian expansion and intensification laid the economic foundation. This generated wealth, not only for the landowning elite, but also for a large group whom we have called ‘middle-class farmers’. A further driving force behind the emergence of villa landscapes was a new strategy of elite distinction, with houses as a vital element in the competitive display of status. In terms of its basic tenets, this new strategy was obviously inspired by Roman elite culture, yet in its concrete material expression it led everywhere to the creation of new hybrid forms. General Roman principles thus became interwoven with existing local phenomena, and Roman villa landscapes acquired a degree of local character and identity. We should also bear in mind the appear176

Cf. Heimberg 2002/2003, 134. Wickham 2005, 504-514.

togenbosch and its rural hinterland. The town’s old and

177

On the interplay between town and countryside in the

new elites competed fiercely to build the largest and

Late Middle Ages, and the role of local and regional elites, see Bijsterveld’s case study (in prep.) on ‘s-Her-

most prestigious houses. Cf. Janssen 2002. 178

Bentmann/Müller 1970; Cosgrove 1993.



ance of new regional creations that occurred entirely within a Roman setting. An example is the earliest Roman domus in towns like Amiens and Tongres, which makes no reference at all to its post-built predecessors.179 What then is the broader relevance of the study of villa landscapes for Roman archaeology? Three points emerge in the discussions we have raised. The first is that the study of villa landscapes points to relatively high consumption levels among rural populations, which raises the question whether this contradicts the influential model of the consumer town proposed by both sociologists and ancient historians, in particular Max Weber and Moses Finley.180 The basic argument of this model is that ‘the major income of urban consumers was as rural rentiers, not as commercial entrepreneurs’.181 It describes the countryside basically in terms of production for consumption in the town, not in terms of consumption in the countryside itself. The archaeological evidence, however, shows that the level of rural consumption should not be underestimated. Firstly, there is the group of the decurial elite which made large investments not only in their urban domus, but also in their rural residences and the associated burial monuments. Secondly, the villa landscapes show a dense distribution of small villas belonging to a considerable group of middle-class farmers who were able to spend a significant amount of money in their country houses as well. And, finally, even the lowest social groups of peasants appear to have had a consumptive demand that went beyond subsistence. Although this concerns above all petty commodity products such as tools, pottery, clothes, bronzes and glass, it is important to realise that the aggregate consumptive demand of this social group must have been substantial.182 Another point of refinement relates to regional variations. Particularly in non-villa landscapes (as in the civitates of the Menapii, Frisiavones, Batavi and Cananefates) we observe a laborious urban development and a much lower consumption level among rural populations. In these societies social distinctions between the rich and the poor were obviously less developed. Notwithstanding the above nuances, we still feel that the model of the consumer town remains attractive and we share Whittaker’s view that ‘there are no models of the city which can satisfactorily replace Weber’s consumer city, despite its limitations.’183 A second point, which may be seen as a logical corollary of the above reappraisal of rural consumption, is the conclusion that the tax burden cannot have been too extreme in the first two centuries. This is relevant for our general view of the Roman empire. Neo-Marxist and post-colonial literature in particular often present the Roman economy as a highly exploitative system designed to maximise surplus extraction through heavy taxation.184 The appearance of densely inhabited villa landscapes in the northern loess belt suggests, however, that this view needs to be qualified and that Keith Hopkins’ model of a rather low rate of taxation should be re-evaluated.185 A third point is that Millett’s model of ‘competitive emulation’ has been too quickly dismissed as an explanatory mechanism in the post-colonial discussion on the development of villa landscapes.186 It was certainly a major driving force behind the new strategy of elite distinction among peer groups of the social and political upper class, especially in the first generations after the conquest. In villa construction too, however, emulation did not mean servile imitation but instead entailed a creative appropriation that gave rise to new hybrid forms.187 Then there were the competitive strategies employed by a broad middle group of farmers without political influence. On the one hand, they may have used their newly 179

Roymans/Habermehl, this volume.

180

183

Whittaker 1993, 11.

The model, initially introduced by Weber, Bücher and

184

Sombart, is best known through the writings of Finley

185

(1973). For discussion, see Hopkins 1978, Whittaker

186

1990 and 1993, and Wallace-Hadrill 1991.

187

181

Whittaker 1990, 112.

182

Cf. also Hopkins 1978, 54 and 75.



Woolf 1990; Mattingly 2006, 3-20. Hopkins 1980. For a contrasting view, Mattingly 2006, 15. Cf. the example of the axial villas discussed in Roymans/Habermehl, this volume.

acquired wealth to adopt certain elements of the material culture and lifestyle of elite groups.188 But on the other hand two other factors were probably at least as important: emphasising social boundaries with lower class groups and cultivating their own identity as a middle-class peer group. More generally, we thus believe that the post-colonial discourse for understanding rural developments is more of a supplement than an all-encompassing alternative to Millett’s model of cultural change. We would like to end with the general observation that although Roman villa archaeology may have a reputation as a research field with access to extremely rich datasets, the evidence is surprisingly meagre when it comes to more complex questions. What we urgently need therefore are high-quality excavations of villa complexes and rural sites in general, enabling us to sketch a picture of economic and social developments in the countryside, independent of historical models. We need a more balanced view of the material culture and consumption patterns of all the groups inhabiting villa landscapes, not just the upper social and political echelons. Furthermore, relations between villas and other elements of the surrounding landscape, such as sanctuaries and vici, have barely been investigated. How did the cult in these sanctuaries and artisanal production in vici articulate with the villa world? Here lies a clear challenge for local and regional research agendas for the decades ahead.

a b b re v iatio n s AAS Amsterdam Archaeological Studies BJ Bonner Jahrbücher BROB Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek CIL Corpus Inscriptionum Latinarum ILB2 Deman, A./M.Th. Raepsaet-Charlier, 2002: Nouveau receuil des Inscriptions Latines de Belgique (ILB2), Brussels. JRA Journal of Roman Archaeology KJ Kölner Jahrbuch TZ Trierer Zeitschrift

re f ere n ces Agache, R., 1978: La Somme pré-romaine et romaine, Amiens. Bakels, C./W. Dijkman 2000: Maastricht in the first millennium. The archaeobotanical evidence, Maastricht (Archaeologica Mosana 2). Balensiefen, L., 2002: Die Macht der Literatur. Über die Büchersammlung des Augustus auf dem Palatin, in W. Hoepfner (ed.), Antike Bibliotheken, Mainz (Zaberns Bildbände zur Archäologie), 97-116. Bayard, D./J.-L. Collart (eds), 1996: De la ferme indigène à la villa romaine. La romanisation des campagnes de la Gaule. Actes du deuxième colloque de l’association AGER tenu à Amiens (Somme) du 23 au 25 septembre 1993, Amiens (RAP, numéro special 11). Beltrán Lloris, F., 2006: An irrigation decree from Roman Spain: The Lex Rivi Hiberiensis, JRS 96, 147197.

188

It seems doubtful that they were motivated by a ‘desire to appear Romanized’ (Millett 1990a, 94). For these

their own society rather than outside it. Cf. Woolf 1997, esp. 341.

groups, the point of reference should be sought within



Bentmann, R./M. Müller, 1970: Die Villa als Herrschaftsarchitektur. Versuch einer kunst- und sozialgeschichtlichen Analyse, Frankfurt am Main, partly reprinted in F. Reutti (ed.) 1990, Die römische Villa, Darmstadt, 389-435. Bettini, M., in press: Entre « émique » et « étique ». Un exercice sur le Lar familiaris, in P. Payen/E. ScheidTissinier (eds), L’antiquité et l’anthropologie, Turnhout. Bijsterveld, A.-J., in prep.: The interplay of town, countryside and elites: the case of ‘s-Hertogenbosch (c. 1180-1350), in A.J. Bijsterveld/J.D. Keene/J. Naylor/A. Wilkin (eds), Dynamic interactions. Town and countryside relations in Northwestern Europe in the Middle Ages, Turnhout. Blanck, H., 1992: Das Buch in der Antike, Munich. Bloemers, J.H.F., 1978: Rijswijk (Z.-H.), ‘De Bult’. Eine Siedlung der Cananefaten, Amersfoort (Nederlandse oudheden 8). Bodel, J., 1997: Monumental villas and villa monuments, JRA 10, 1997, 5-35. Boislève, J., 2010: Mané-Véchen. Dorures et stucs d’une villa maritime, Archéologia 475, 36-47. Bonnie, R., 2009: Cadastres, misconceptions & Northern Gaul. A case study from the Belgian Hesbaye region, Leiden. Boschung, D., 2003: Die stadtrömischen Monumente des Augustus und ihre Rezeption im Reich, in P. Noelke (ed.), Romanisation und Resistenz in Plastik, Architektur und Inschriften der Provinzen des Imperium Romanum. Akten des VII. internationalen Colloquiums über Probleme des provinzialrömischen Kunstschaffens, Köln 2. Bis 6.Mai 2001, Mainz, 1-12. Bourgeois, J., et al., 2001: Some excavated protohistoric and Roman field systems in sandy Flanders, in F. Vermeulen/M. Antrop (eds), Ancient lines in the landscape. A geo-archaeological study of protohistoric and Roman roads and field systems in Northwestern Gaul, Leuven (Babesch Suppl. 7),139-150. Bradley, R., 2002: The past in prehistoric societies, London. Brunt, P.A., 1976: The Romanization of the local ruling classes in the Roman empire, in D.M. Pippidi (ed.), Assimilation et résistance à la culture gréco-romaine dans le monde ancien. Travaux du VIe congrès international d’études classiques, Madrid, Septembre 1974, Bucuresti/Paris, 161-173, reprinted in P.A. Brunt, 1990: Roman imperial themes, Oxford, 267-281, 515-517. Burke, P., 2009: Cultural hybridity, Cambridge/Malden MA. Campbell, B., 2000: The writings of the Roman land surveyors, London. Chouquer, G., 1983: Localisation et extension géographique des cadastres affichés à Orange, in M. Clavel-Lévêque (ed.), Cadastres et espace rural. Approches et réalités antiques, Paris (Table ronde de Besançon, may 1980), 274-295. Chouquer, G., 2008: Les transformations récentes de la centuriation. Une autre lecture de l’arpentage romain, Annales HSS 63, 847-874. Christie, N., 2004: Landscapes of change in Late Antiquity and the Early Middle Ages.Themes, directions and problems, in N. Christie (ed.), Landscape of change. Rural evolutions in Late Antiquity and the Early Middle Ages, Aldershot/Burlington, 1-37. Christol, M., 1999: La municipalisation de la Gaule Narbonnaise, in M. Dondin-Payre/M.-Th. RaepsaetCharlier (eds), Cités, municipes, colonies. Les processus de municipalisation en Gaule et en Germanie sous le Haut Empire romain, Paris, 1-27. Cohen, A.P., 1985: The symbolic construction of community, London/New York. Clotuche, R., 2009: The Scheldt Valley Commercial Activity Zone. 350 hectares of the Gallo-roman landscape, Britannia 40, 41-64. Cosgrove, D., 1993: The Palladian landscape. Geographical change and its cultural representations in sixteenthcentury Italy, Leicester. Creighton, J., 2006: Britannia. The creation of a Roman province, London/New York. Crowley, L., 2009: Creating a community. The symbolic role of tumuli in the villa landscape of the civitas Tungrorum, in M. Driessen et al. (eds), TRAC 2008. Proceedings of the Eighteenth Annual Theoretical



Roman Archaeology Conference, Oxbow, 113-126. Crowley, L., 2011: Dying in a material world. Self-representation and its implications in privileged mortuary contexts in the Roman villa landscapes between Bavay and Cologne, Amsterdam (Ph.D. thesis VU University Amsterdam). Cüppers, H./Ch.B. Rüger, 1985: Geschichtlicher Atlas der Rheinlande, Beiheft III/1-III/2. Römische Siedlungen und Kulturlandschaften, Cologne (Publikationen der Gesellschaft für Rheinische Geschichtskunde XII, Abteilung 1b Neue Folge). Daloz, J.-P., 2009: The sociology of elite distinction. From theoretical to comparative perspectives, New York. De Clercq, W., 2009: Lokale gemeenschappen in het Imperium Romanum.Transformaties in rurale bewoningsstructuur en materiële cultuur in de landschappen van het noordelijk deel van de civitas Menapiorum, Ghent (unpublished dissertation Ghent University). De Groot, T., 2006: Resultaten van de opgraving van een Romeins tumulusgraf in Bocholtz (gem. Simpelveld), Amersfoort (Rapportage Archeologische Monumentenzorg 127). De Maeyer, R., 1937: De Romeinse villa’s in België. Een archeologische studie, Antwerp. De Ridder, T., 1999: De oudste deltawerken van West-Europa. Tweeduizend jaar oude dammen en duikers te Vlaardingen, Tijdschrift voor Waterstaatsgeschiedenis 8, 10-22. Derks, T., 1998: Gods, temples and ritual practices.The transformation of religious ideas and values in Roman Gaul, Amsterdam (AAS 2). Derks, T., 2009: Ethnic identity in the Roman frontier. The epigraphy of Batavi and other Lower Rhine tribes, in T. Derks/N. Roymans (eds), Ethnic constructs in antiquity.The role of power and tradition, Amsterdam (AAS 13), 239-282. Derks, T./N. Roymans, 2002: Seal-boxes and the spread of Latin literacy in the Rhine delta, in A.E. Cooley (ed.), Becoming Roman, writing Latin? Literacy and epigraphy in the Roman west, Portsmouth (JRA, suppl. 48), 87-134. Derks, T./N. Roymans, 2006: Returning auxiliary veterans in the Roman empire: some methodological considerations, JRA 16, 121-135. Dijkman, W., 2003: DE RE TEGULA. Romeinse dakpanstempels gevonden te Maastricht. Archeologie in Limburg 95, 15-18. Dräger, P., 2001 (ed.): Ausonius, Mosella, Trier, 2001. Drinkwater, J.F., 1977/78: Die Secundinier von Igel und die Woll- und Textilindustrie in Gallia Belgica. Fragen und Hypothesen, TZ 40/41, 107-125. Drinkwater, J.F., 1982: The wool textile industry of Gallia Belgica and the Secundinii of Igel. Questions and hypotheses, Textile history 13, 111-128. Dubouloz, J., 2008: Propriété et exploitation des entrepôts à Rome et en Italie (Ier – IIIe siècles), Mélanges de l’École Française de Rome, Antiquité 120, 277-294. Dyson, S.L., 2003: The Roman countryside, London. Ebnöther, C., 1995: Der römische Gutshof in Dietikon, Zürich/Egg (Monographien der Kantonsarchäologie Zürich 25). Eck, W., 2004: Geschichte der Stadt Köln. Köln in römischer Zeit, Cologne. Edelman, C.H./B.E.P. Eeuwens, 1959: Sporen van een Romeinse landindeling in Zuid-Limburg, BROB 9, 49-56. Effros, B., 2001: Monuments and memory. Repossessing ancient remains in Early Medieval Gaul, in M. de Jong/F.Theuws (eds), The topography of power in the early Middle Ages, Leiden/Boston/Cologne (The transformation of the Roman world 6), 93-118. Eriksen, Th. H., 2007: Creolization in anthropological theory and in Mauritius, in C. Stewart (ed.), Creolization: history, ethnography, theory, Walnut Creek, 153-177. Fehr, H., 2003: Römervilla. Führer durch die Ausgrabungen und Ausstellung am Silberberg, Bad Neuenahr-Ahrweiler, Koblenz (Archäologie an Mittelrhein und Mosel 7).



Ferdière, A., 1983: Recherche des parcellaires fossiles et cadastres antiques: l’exemple de la Gaule du Nord, in M. Clavel-Levêque (ed.), Cadastres et espace rural. Approches et réalités antiques. Table-ronde de Besançon, mai 1980, Paris, 159-183. Foss, P.W., 1997: Watchful Lares. Roman household organization and the rituals of cooking and eating, in R. Laurence/A, Wallace-Hadrill (eds), Domestic space in the Roman world: Pompeii and beyond, Portsmouth RI (JRA, Suppl. Series 22), 196-218. Freigang, Y., 1997: Die Grabdenkmäler der gallo-römischen Kultur im Moselland. Studien zur Selbstdarstellung einer Gesellschaft, Jahrbuch des Römish-Germanischen Zentralmuseums Mainz 44, 277-440. Fremersdorf, F., 1933: Der römische Gutshof Köln-Müngersdorf, Berlin/Leipzig (Römisch-Germanische Forschungen 6). Gaitzsch, W., 1986: Grundformen römischer Landsiedlungen im Westen der CCAA, BJ 186, 397-427. Gaitzsch, W., 2005: Tempel und Jupitersäulen in villae rusticae, Archäologie im Rheinland 2005, 81-83. Gardner, J.F., 1986: Women in Roman law and society, London/Sydney. Gechter, M./J. Kunow, 1986: Zur ländlichen Besiedlung des Rheinlandes in römischer Zeit, BJ 186, 377-396. Gerritsen, F., 2003: Local identities. Landscape and community in the late prehistoric Meuse-Demer-Scheldt region, Amsterdam (AAS 9). Giddens 1990: The consequences of modernity, Cambridge. Groot, M., 2008: Animals in ritual and economy in a frontier community. Excavations in Tiel-Passewaaij, Amsterdam (AAS 12). Groot, M./S. Heeren/L.I. Kooistra/W.K. Vos, 2009: Surplus production for the market? The agrarian economy in the non-villa landscapes of Germania Inferior, JRA 22, 231-252. Haalebos, J.K., 1996: Ein römisches Getreideschiff aus Woerden, Jahrbuch des Römisch-Germanischen Zentralmuseums 43, 475-509. Habermehl, D., 2011: Settling in a changing world. Villa development in the northern provinces of the Roman empire, Amsterdam (Ph.-D thesis VU University Amsterdam). Hales, S., 2003: The Roman house and social identity, Cambridge. Hanel, N., 1998: Ziegelstempel aus dem Areal des Flottenlagers Köln-Marienburg (Alteburg), KJ 31, 401-415. Haselgrove, C., 1996: Roman impact on rural settlement and society in southern Picardy, in N. Roymans (ed.), From the sword to the plough. Three studies on the earliest romanisation of Northern Gaul, Amsterdam (AAS 1), 127-187. Häusle, H., 1980: Das Denkmal als Garant des Nachruhms. Eine Studie zu einem Motiv in lateinischen Inschriften, München (Zetemata 75). Heeren, S., 2009: Romanisering van rurale gemeenschappen in de Civitas Batavorum. De casus Tiel-Passewaaij, Amersfoort (Nederlandse Archeologische Rapporten 36), available online at http://hdl.handle. net/1871/15248. Heimberg, U., 2002/2003: Römische Villen an Rhein und Maas, BJ 202/203, 57-148. Helttula, A., 2007: Le iscrizioni sepolcrali latine nell’Isola Sacra, Rome (Acta Instituti Romani Finlandiae 30). Henrich, P., 2006: Die römische Besiedlung in der westlichen Vulkaneifel, Trier (TZ, Beiheft 30). Henrotay, D., 2006: Un atelier de foulons et de teinturiers à Arlon, Dossiers d’archéologie 315, 109-111. Hingley, R., 2005: Globalizing Roman culture. Unity, diversity and empire, Abingdon/New York. Hingley, R., 2008: The recovery of Roman Britain, 1586-1906. A colony so fertile, Oxford. Hinz, H., 1969: Kreis Bergheim, Bonn (Archäologische Funde und Denkmäler des Rheinlandes 2). Horisberger, B., 2004: Der römische Gutshof von Buchs und die römische Besiedlung im Furttal, Zürich/Egg (Monographien der Kantonsarchäologie Zürich 37). Hopkins, K., 1978: Economic growth and towns in classical antiquity, in Ph. Abrams/E.A. Wrigley (eds), Towns in societies. Essays in economic history and historical sociology, Cambridge etc., 35-77.



Hopkins, K., 1980: Taxes and trade in the Roman empire (200 BC-AD 400), JRS 70, 101-125. Ingold, T., 1993: The temporality of landscape, World archaeology 25, 152-174. Jacques, F., 1983: Cadastrations dans la région Champagne-Ardennes, in M. Clavel-Levêque (ed.), Cadastres et espace rural, approches et réalités antiques. Table-ronde de Besançon, mai 1980, Paris, 330-332. Jacques, F., 1987: Témoins de cadastres romains dans la region de Cassel, Revue du Nord 69 (no. 272), 101-108. Janssen, H.L., 2002: Patrician and aristocratic town residences in ‘s-Hertogenbosch c. 1250-c.1550. The archaeological evidence, in G. Helmig/B. Scholkmann/M. Untermann (eds), Centre-region-periphery. Medieval Europe Basel 2002, II, Hertingen, 140-148. Jeneson, K., 2011: Exploring the Roman villa world between Tongres and Cologne. A landscape-archaeological approach, Amsterdam (Ph.-D thesis VU University Amsterdam). Kaszab-Olschewski, T., 2003: Fragmente einer Iupitersäule aus der ‘Villa rustica Hambach 512’, in P. Noelke (ed.), Romanisation und Resistenz in Plastik, Architektur und Inschriften der Provinzen des Imperium Romanum. Akten des VII. internationalen Colloquiums über Probleme des provinzialrömischen Kunstschaffens, Köln 2. Bis 6.Mai 2001, Mainz, 127-130. Kaufmann-Heinimann, A., 1998: Götter und Lararien aus Augusta Raurica. Herstellung, Fundzusammenhänge und sakrale Funktion figürlicher Bronzen in einer römischen Stadt, Augst (Forschungen in Augst 26). Kooistra, L.I., 1996: Borderland farming. Possibilities and limitations of farming in the Roman period and Early Middle Ages between the Rhine and Meuse, Assen/Amersfoort. Krause, D., 2006: Eisenzeitlicher Kulturwandel und Romanisierung im Mosel-Eifel-Raum, Mainz (RömischGermanische Forschungen 63). Kremer, G., 2009: Das frühkaiserzeitliche Mausoleum von Bartringen (Luxemburg), Luxembourg (Dossiers d’Archéologie du Muséé National d’Histoire et d’Art, 12). Kreuz, A., 2004: Landwirtschaft im Umbruch? Archäobotanische Untersuchungen zu den Jahrhunderten um Christi Geburt in Hessen und Mainfranken, Bericht der Römisch Germanischen  Kommission 85, 97-292. Krier, J., 2002: Peintures romaines de Vichten, Archéologia 395, 44-55. Krier, J./F. Reinert, 1995: Homère et les neuf muses à Vichten, in Luxembourg de la préhistoire au Moyen âge. Dossiers d’Archéologie, Hors série n° 5, 71-73. Lenz, K.-H., 1998: Villae rusticae. Zur Entstehung dieser Siedlungsform in den Nordwestprovinzen des römischen Reiches, KJ 31, 49-70. Lenz, K.-H., 1999a: Die ländliche Besiedlung der frühen und mittleren Kaiserzeit im Hinterland des römischen Köln, KJ 32, 807-821. Lenz, K.-H., 1999b: Siedlungen der römischen Kaiserzeit auf der Aldenhovener Platte, Cologne (Rheinische Ausgrabungen 45). Lepetz, S., 2009: La chasse à la période romaine dans le nord de la Gaule. Étude archéozoologique, in J. Trinquier/C. Vendries (eds), Chasses antiques. Prâtiques et représentations dans le monde gréco-romain (III s. av. – IV s. apr. J.-C.). Actes du colloque international de Rennes, Université de Rennes II, 20-21 septembre 2007, Rennes, 141-159. Martins, C.B., 2005: Becoming consumers: looking beyond wealth as an explanation of villa variability. Perspectives from the east of England, Oxford (BAR British Series 403). Matterne,V./J.-H.Yvinec/D. Gemehl, 1998: Stockage de plantes alimentaires et infestation par les insectes dans un grenier incendié de la fin du IIe siècle après J.C. à Amiens (Somme), Revue Archéologique de Picardie 3/4, 93-122. Mattingly, D.J., 1997: Dialogues of power and experience in the Roman empire, in D. Mattingly (ed.), Dialogues in Roman imperialism. Power, discourse, and discrepant experience in the Roman Empire, Portsmouth RI (JRA, suppl. series 23), 7-24. Mattingly, D.J., 2004: Being Roman: expressing identity in a provincial setting, JRA 17, 5-25.



Mattingly, D.J., 2006: An imperial possession. Britain in the Roman empire, 54 BC-AD 409, London. Mattingly, D.J., 2011: Imperialism, power and identity. Experiencing the Roman empire, Princeton/Oxford. Mertens, J.R., 1958: Sporen van een Romeins kadaster in Limburg?, Limburg 37, 253-260. Mertens, J.R./W. Vanvinckenroye 1975: Een Romeins gebouwencomplex extra-muros te Tongeren, Brussels (Archaeologia Belgica 180). Millett, M., 1990a: The Romanization of Britain. An essay in archaeological interpretation, Cambridge. Millett, M., 1990b: Romanization. Historical issues and archaeological interpretation, in Th. Blagg/M. Millett (eds), The Early Roman Empire in the West, Oxford, 35-41. Millett, M., 1995: Roman Britain, London. Nicolay, J., 2007: Armed Batavians. Use and significance of weaponry and horse gear from non-military contexts in the Rhine Delta (50 BC – AD 450), Amsterdam (AAS 11). Noelke, P., 2005: Iuppitersäulen und -pfeiler in Niedergermanien - Neufunde aus 25 Jahren, in H.G. Horn/H. Hellenkemper/G. Isenberg/J. Kunow (eds), Von Anfang an. Archäologie in Nordrhein-Westfalen, Mainz (Schriften zur Bodendenkmalpflege in Nordrhein-Westfalen 8), 128-137. Oelmann, F., 1928: Ein gallo-römischer Bauernhof bei Mayen, BJ 133, 51-140. Ouzoulias, P., 2006: L’économie agraire de la Gaule. Apercus historiographiques et perspectives archéologiques (Ph.D.-thesis Université de Franche-Comté, available online at http://tel.archives-ouvertes.fr/ docs/00/05/83/PDF/Texte.pdf. Ouzoulias, P., 2010: Les campagnes gallo-romaines: quelle place pour la villa?, in P. Ouzoulias/L. Tranoy (eds), Comment les Gaules devinrent romaines, Paris, 189-211. Pals, J.P./T. Hakbijl 1992: Weed and insect infestation of a grain cargo in a ship at the Roman fort of Laurium in Woerden (Prov. of Zuid-Holland), Review of Palaeobotany and Palynology 73, 287-300. Peterson, J.W.M., 1996: A computer model of Roman landscape in South Limburg, in H. Kamermans/K. Fennema (eds), Interfacing the past. Computer applications and quantitative methods in archaeology, Leiden (Analecta Praehistorica Leidensia 28, 185-194. Piganiol, A., 1962: Les documents cadastraux de la colonie romaine d’Orange, Paris (Gallia, suppl. 16). Percival, J., 1976: The Roman villa. An historical introduction, London. Provost, A., 2007: La « villa » maritime de Mané-Véchen à Plouhinec (Morbihan), Aremorica 1, 85-100. Purcell, N., 1995:The Roman villa and the landscape of production, in T.J. Cornell/K. Lomas (eds), Urban society in Roman Italy, London. Raepsaet, G., 1995: Les prémices de la mécanisation agricole entre Seine et Rhin de l’Antiquité au 13e siècle, Annales. Histoire, sciences sociales 50, 911-942. Rosafio, P., 1994: Slaves and coloni in the villa system, in J. Carlsen/P. Ørsted/J.E. Skydsgaard (eds), Land use in the Roman empire, Rome (Analecta Romana Instituti Danici, suppl. 22), 145-158. Rothenhöfer, P., 2005: Die Wirtschaftsstrukturen im südlichen Niedergermanien. Untersuchungen zur Entwicklung eines Wirtschaftsraumes an der Peripherie des Imperium Romanum, Rahden (Kölner Studien zur Archäologie der römischen Provinzen 7). Roymans, N., 1996: The sword or the plough. Regional dynamics in the romanisation of Belgic Gaul and the Rhineland area, in N. Roymans (ed.), From the sword to the plough. Three studies on the earliest romanisation of Northern Gaul, Amsterdam (AAS 1), 9-126. Roymans, N., 2004: Ethnic identity and imperial power.The Batavians in the Early Roman Empire, Amsterdam (AAS 10). Roymans, N., 2007: On the latènisation of Late Iron Age material culture in the Lower Rhine/Meuse area, in S. Möllers/W. Schlüter/S. Sievers (eds), Keltische Einflüsse im nördlichen Mitteleuropa während der mittleren und jüngeren vorrömischen Eisenzeit, Bonn, 311-325. Roymans, N., 2009a: Becoming Roman in the Rhineland frontier zone. The impact of ethnic recruitment and returning veterans on the romanisation of rural populations, in Kelten am Rhein, Proceedings of the Thirteenth International Congress of Celtic Studies, I, Ethnizität und Romanisierung, Mainz (Beihefte



BJ 58, I), 25-46. Roymans, N., 2009b: Die Bataver. Zur Entstehung eines Soldatenvolkes, in St. Berke et al., 2000 Jahre Varusschlacht. Mythos, Stuttgart, 85-98. Roymans, N./T. Derks (eds), 1994: De tempel van Empel. Een Herculesheiligdom in het woongebied van de Bataven, `s-Hertogenbosch. Roymans, N./T. Derks/S. Heeren (eds), 2007: Een Bataafse gemeenschap in de wereld van het Romeinse rijk. Opgravingen te Tiel-Passewaaij, Utrecht. Roymans, N./F. Gerritsen/C. van der Heijden/K. Bosma/J. Kolen, 2009: Landscape biography as a research strategy. The case of the South Netherlands Project, Landscape Research 34, 337-359. Rüger, C.B., 1972: Gallisch-germanische Kurien, Epigraphische Studien 9, 251-259. Sage, P., 1991: Le testament du Lingon: remarques sur le texte et sur son interpretation, in Y. Le Bohec (ed.), Le testament du Lingon, Lyon (Collection du Centre d’Etudes Romaines et Gallo-Romaines, 9), 17-40. Scheid, J., 1999: Aspects religieux de la municipalisation. Quelques réflexions générales, in M. DondinPayre/M.-Th. Raepsaet-Charlier (eds), Cités, municipes, colonies. Les processus de municipalisation en Gaule et en Germanie sous le Haut-Empire romain, Paris, 381-423. Slofstra, J., 1983: An anthropological approach to the study of romanization processes, in R. Brandt/J. Slofstra (eds), Roman and native in the Low Countries. Spheres of interaction, Oxford (BAR Inter. Ser. 184), 71-104. Slofstra, J., 1991: Changing settlement systems in the Meuse-Demer-Scheldt area during the Early Roman Period, in N. Roymans/F. Theuws (eds), Images of the past. Studies on ancient societies in Northwestern Europe, Amsterdam 1991 (Studies in Prae- en Protohistorie 7), 131-199. Slofstra, J., 1995: The villa in the Roman West. Space, decoration and ideology, in J. Metzler et al. (eds), Integration in the Early Roman West. The role of culture and ideology, Luxembourg (Dossiers d’archéologie du Musée National d’Histoire et d’Art 4), 77-90. Terrenato, N., 2001: The Auditorium site in Rome and the origins of the villa, JRA 14, 5-32. Thurston, T., 2009: Unity and diversity in the European Iron Age. Out of the mists, some clarity?, Journal of Archaeological Research 17, 7-84. Trinquier, J./C. Vendries (eds), 2009: Chasses antiques. Prâtiques et représentations dans le monde gréco-romain (III s. av. – IV s. apr. J.-C.). Actes du colloque international de Rennes, Université de Rennes II, 20-21 septembre 2007, Rennes. Vanderhoeven, A./G. Vynckier/W. Wouters, 1997/1998: Het oudheidkundig bodemonderzoek aan de Zijdelingsestraat te Tienen (prov. Vlaams-Brabant). Interimverslag 1995-1996, in Archeologie in Vlaanderen 6, 133-160. Van Enckevort, H., 2005: The significance of the building program of the Legio X gemina for the integration of the Batavians into the Roman empire, in Z.Visy (ed.), Proceedings of the XIXth International Congress of Roman Frontier Studies, Pécs, Hungary, September 2003, Pécs, 85-93. Van Londen, H., 2006: Midden-Delfland. The Roman Native landscape, past and present, Amsterdam (unpublished dissertation University of Amsterdam). Van Ossel, P./P. Ouzoulias, 2000: Rural settlement economy in Northern Gaul in the Late Empire, JRA 13, 133-160. Von Massow, W., 1932: Die Grabmäler von Neumagen, Berlin/Leipzig. Vos, W.K., 2009: Bataafs platteland. Het Romeinse nederzettingslandschap in het Nederlandse Kromme-Rijngebied, Amersfoort (Nederlandse Archeologische Rapporten 35), available online at http://hdl.handle. net/1871/13283. Wallace-Hadrill, A., 1991: Elites and trade in the Roman town, in J. Rich/A. Wallace-Hadrill (eds), City and country in the ancient world, London/New York, 241-272. Weeber, K.-W., 2000: Alltag im alten Rom. Das Landleben, Düsseldorf/Zürich.



Whittaker, C.R., 1990: The consumer city revisited: the vicus and the city, JRA 3, 110-117. Reprinted in C.R. Whittaker, 1993, Land, city and trade in the Roman empire, Aldershot, VIII.. Whittaker, C.R., 1993: Do theories of the ancient city matter?, in T.J. Cornell/H.K. Lomas (eds), Urban society in Roman Italy, London, 1-20. Reprinted in C.R. Whittaker, 1993, Land, city and trade in the Roman empire, Aldershot, IX. Wickham, Chr., 2005: Framing the Early Middle Ages. Europe and the Mediterranean 400-800, Oxford. Wiedemann, T./M. Antrop/F. Vermeulen, 2001: GIS-analysis of possible ancient normative field systems in the Civitas Menapiorum, in F.Vermeulen/M. Antrop (eds), Ancient lines in the landscape. A geo-archaeological study of protohistoric and Roman roads and field systems in Northwestern Gaul, Leuven (Babesch Suppl. 7), 117-138. Wightman, E., 1978: Peasants and potentates. An investigation of social structures and land tenure in Roman Gaul, American Journal of Ancient History 3, 97-128. Wild, J.P., 1970: Textile manufacture in the Northern Roman provinces, Cambridge. Willems, W.J.H., 1981: Romans and Batavians. A regional study in the Dutch eastern river area I, BROB 31, 7-217. Willems, W.J.H., 1984: Romans and Batavians. A regional study in the Dutch eastern river area II, BROB 34, 39-331. Willems, W.J.H., 1987: Villa’s langs de weg, in P. Stuart/M.E.Th. de Grooth (eds), Langs de weg. De Romeinse weg van Boulogne-sur-Mer naar Keulen: verkeersader voor industrie en handel / Villa rustica: het Romeinse boerenbedrijf in het Rijn-Maasgebied, Heerlen/Maastricht, 46-50. Willems, W.J.H./L.I. Kooistra, 1988: De Romeinse villa te Voerendaal; opgraving 1987, Archeologie in Limburg 37, 137-147. Woolf, G., 1990: World-systems analysis and the Roman empire, JRA 3, 44-58. Woolf, G., 1997: Beyond Romans and natives, World archaeology 28, 339-350. Woolf, G., 1998: Becoming Roman. The origins of provincial civilization in Gaul, Cambridge. Woolf, G., 2001: Representation as cult. The case of the Jupiter columns, in W. Spickermann (ed.), Religion in den germanischen Provinzen Roms, Tübingen, 117-134.



Reflections on the Iron Age background to the emergence of villa landscapes in northern France Colin Haselgrove 1 2 3 4 5 6

Introduction The impact of archéologie préventive Later Iron Age settlement developments Settlements and social organisation Site continuity Conclusion References

1 i n troductio n Thanks to the pioneering research of Roger Agache,1 the courtyard villas of Picardy have come to symbolise the Roman rural landscape in northern France. One of the consequences of this is that the “Picardy model” is often invoked as a yardstick against which to assess settlement developments in other areas of Roman Gaul.2 There are several problems, however, with this approach. First, it rather assumes that the character of rural settlement would be similar from one region to another, although if anything the evidence is to the contrary.3 Second, it downplays the role and significance of other types of rural site that co-existed with villas, although knowledge of the entire settlement pattern is fundamental to understanding the economic organisation of the landscape and its social underpinnings. Third, as JeanLuc Collart emphasises, the “Picardy model” is itself only a snapshot of the landscape at a given moment in time – the late 2nd to 3rd centuries AD, when the deep chalk foundations photographed to such effect by Agache were commonly used. Even in the eponymous region, this model should not be held up as representative of the entire Gallo-Roman period.4 That these should still be issues might seem surprising given the range of Gallo-Roman rural sites discovered and excavated over the last 25 years, the vast majority of them through Archéologie préventive.5 One obstacle is simply the great difficulty of keeping pace with excavated data even at a fairly local level, let alone analysing them on a wide enough scale for inter-regional patterns and differences to be obvious. A second is that a disproportionate amount of data still comes from river terraces and other landscapes targeted by extractive industry, which are not fully representative of Gallo-Roman settlement developments. Last but not least, there are some significant conceptual problems that we still have to overcome. 1

Agache 1978.

advance of development. For a recent review, highlight-

2

Courbot-Dewerdt 2004, 254, 303.

ing the impact of Archéologie préventive on our knowledge

3

See e.g. Wightman 1975, 101-133.

of Gallo-Roman rural settlement, see Monteil/Tranoy

4

As J.-L.Collart noted at the symposium.

2008, especially 66-87.

5

The term used in France for archaeology carried out in



My feeling is that, with honourable exceptions, Roman settlement archaeology has rather left theoretical approaches to languish on one side as new data accumulated. There have been few attempts to develop radically new models for understanding the emergence of Gallo-Roman settlement attributes. The underlying reason, I think, is that many archaeologists still see the transformation of Iron Age rural settlement patterns into the villa-dominated landscapes of the Gallo-Roman world as familiar and fairly unproblematic – what we might call the Asterix model – due to the hierarchical nature of indigenous societies and their strong agrarian base. According to this viewpoint, the exception that proves the rule is in the far north of Gaul, where we encounter seemingly quite different settlement and social patterns associated with the north European byre house tradition.6 I have argued previously that to understand the diversity of rural settlement across Roman Gaul, we need to consider possible longue durée influences on settlement and landscape development, as well as existing social and cultural differences between regions.7 Historical contingencies are also relevant and not restricted to the impact of the Roman conquest. The first part of this paper examines some key changes in later Iron Age rural settlement and economy, which seem to me critical to understanding the subsequent development of ‘villa landscapes’ in northern Gaul. I shall focus on the period from the 3rd century BC onward,8 but will also look back at the earlier first millennium BC landscape – when the settlement pattern was characterised by a high degree of mobility – and ask what implications this might have for later patterns? The second part of the paper will air some other issues, which I feel need discussion. In particular, how strong is the evidence for traditional readings of Iron Age societies as hierarchical? Should we be considering alternative models of social organisation? And what are the reasons for the relative absence of La Tène D2 rural settlements in many areas of northern and eastern France? What significance has this for Gallo-Roman landscapes? My discussion will draw on material from eastern France, privileging three regions with which I am most familiar.9 The first is southern Picardy, particularly the Aisne and the Oise valleys, which have been subject to intensive mineral extraction since the 1960s. The second lies to the south-east of Paris, where Ile-de-France, southern Champagne and northern Burgundy meet. Within this area, the SeineYonne confluence has seen gravel extraction on a similar scale to the Aisne and Oise, whilst settlement in northern Burgundy was recently the subject of an important study by Pierre Nouvel.10 The last area is southern Burgundy, including the Saône and Arroux valleys, where I have more recently been involved in fieldwork.

2 the im p act o f arch  ologie p r  v e n ti v e Before turning to the settlement record, we should consider the impact of Archéologie préventive. First and foremost, even within a region the knowledge it generates is often quite uneven and concentrated in a few hot spots, linked to a specific menace, be this gravel extraction, the building of new towns or infrastructure projects.This is particularly obvious in the Paris region (Ile-de-France), where each cluster of Iron Age settlement is explained by a specific threat.11 Timing can also be a factor; most autoroutes in eastern Picardy 6

See e.g. Bourgeois et al. 2003; Roymans/Gerritsen 2002;

9

The coverage makes no claim to be systematic. For a

Gerritsen 2003.

recent review of later Iron Age settlement and other

7

Haselgrove 1991.

developments covering the northern part of the area, see

8

Approximate dating of the main archaeological sub 10

Nouvel 2004. See also Nouvel 2005.

C2 200-150 BC; La Tène D1 150-90 BC; La Tène D2

11

Marion 2004, fig 3.

90-30 BC; Gallo-Romain préçoce 30 BC-AD 100.



Haselgrove 2007a.

divisions is as follows: La Tène C1 270-200 BC; La Tène

Fig. 1. Number of ‘diagnostics’ (evaluations) by commune in the regions of Ile-de-France, Picardy, Champagne-Ardenne and Burgundy between 2002-2005 (sites of all periods).The much smaller number of full-scale excavations are not shown, but follow a broadly similar pattern (after Rapport au Parlement 2006).

were built in the early years of Archéologie préventive and the impact of linear projects is thus confined to the west. The only recent linear project in south-east Picardy, for the LGV Est, found fewer Iron Age sites than many linear projects, perhaps because of the nature of the landscape it traversed.12

12

See Haselgrove 2007a, fig. 1. Three Iron Age occupation

Gallo-Roman date.

sites were identified over 37 km, compared to six of



As soon as we aspire to regional synthesis, we come up against the fundamental point – oft made, less often heeded – of how representative our information is? Albeit with imperfections, the autoroutes and high-speed rail links do at least offer transects across different types of landscape – plateau top, slopes and valley bottoms. This cannot be said of mineral extraction, which remains rooted in the river valleys. Inter-regional comparisons are just as susceptible to inherent variations in the nature and the pace of development. In the four years (2002–5) following the formation of INRAP,13 Champagne-Ardenne, for example, saw over twice as many evaluations and excavations (sites of all periods) than Burgundy immediately to the south (fig. 1).14 The operations in Champagne included the final phases of the major infrastructure projects at Vatry airport and for the LGV Est, together with construction hotspots around Châlons, Reims, Troyes and many lesser centres.15 These yielded evidence for a wide range of GalloRoman rural settlement sites, of which only a minority qualify as villas, as well as a basis for comparing long-term settlement dynamics in different types of landscape.16 In Burgundy, on the other hand, there are no such hotspots, apart from a slight cluster in Côte-d’Or north of Dijon, and much of the new information for the Roman period is from an urban context.17 A high proportion of the data from Burgundy analysed by Nouvel comes from survey, reducing the degree of chronological and functional resolution, but in many respects providing a more reliable picture of the settlement pattern than in areas like the Aisne valley where the plentiful excavated sites are confined to one environmental niche. In fact, field surveys in both areas indicate a decline in the importance of occupation on the alluvial terraces in Gallo-Roman times, in favour of side valley and plateau locations.18 This means that the sites on the terraces are unlikely to be a representative sample of GalloRoman settlement, at least in the High Empire (there are some indications of a reversal of this trend in the Late Empire). At the same time, large-scale excavations on the Soissonais plateau are beginning to substantiate the earlier fieldwalking results, which implied extensive colonisation of the plateau from the later Iron Age. On an industrial park at Ploisy and Courmelles,19 one Late Iron Age and no fewer than four Gallo-Roman farms have been excavated in an area of just under 1.6 km2. A last point about Archéologie préventive is that it has had more overt impact on Iron Age studies because these started from a manifestly weaker knowledge base. Not until the 1980s did the large-scale excavations of Iron Age rural sites take place, whereas Roman settlements have been the object of a far longer research tradition. Through the work of Agache and many others we already knew a great deal about their layout and appearance – at least those sites built partly or exclusively of stone and/or visible from the air – whilst their general economic and social context has been inferred, not necessarily correctly, by interpolation from other provinces. So the perceived impact of Roman archaeology has so far been far less, partly because much of the data that contradicts accepted trains of thought is still undigested.

3 later iro n age settleme n t de v elo p me n ts One field where understanding has advanced massively concerns Earlier Iron Age settlement.20 Excluding a minority of imposing but mainly short-lived fortified sites, the settlement record of this period 13

14

Institut National de Recherches Archéologiques Préven-

16

Vanmoerkerke 2009,10-11; 378-380.

tives.

17

Rapport au Parlement 2006, vol. 2, 99-107.

Rapport au Parlement 2006, vol. 2, 149-167. Archaeologi-

18

Haselgrove 1996; Nouvel 2004. A marked decline in 

cal work in Ile-de-France is on a vastly different scale

the significance of the alluvial terraces in Gallo-Roman

even from other regions with a high level of develop-

times is equally evident in other areas of eastern France recently examined by Nouvel et al. 2009.

ment. 15



For Vatry, Lagatie/Vanmoerkerke 2005; for the LGV,Van-

19

Duvette 2003; Gransar 2003; Hénon 2005.

moerkerke 2009.

20

Haselgrove 2007b.

was long (and understandably) dismissed as fragmentary and uninformative throughout eastern France, but can now be seen as complex and multi-faceted. The predominantly open settlements took a variety of forms, from seasonal sites linked to specific resources, through farms that moved every generation, to larger settlements that show a stronger attachment to place. Moreover, the periods during the first millennium BC when settlements were at their most numerous on the terraces are those when the climate was warmest and driest.21 Conversely, the periods when sites are hardest to find in the valley bottoms coincide with climatic downturns,22 implying that wetter conditions, as well as impacting on agriculture, caused population to move away from the alluvial terraces. This is not the place to discuss the socio-economic implications of regularly shifting settlement, but it does raise the question of whether similar practices persisted into later periods and how they impacted on site formation at a landscape level. There are certainly hints of significant settlement mobility in the later Iron Age, a good example being at Croixrault (Somme).23 Here, a first enclosure of La Tène C2 date was rebuilt in La Tène C2/D1; a second enclosed settlement was then built 400 m to the north-east and the two linked by fields and ditches. Finally, in La Tène D2, a trackway was built through the western part of the field system, cutting through the original settlement. In general, we have not thought as critically as we should about why so many shifts in focus are apparent at later Iron Age rural establishments – although the valley-bottom was clearly a factor – or why a fair number of Roman sites were inhabited for periods that seem too short to repay the investment of building them. Instead, the tendency is to grasp at the straw of continuity, however thin the evidence on the ground. That said, a key pattern to emerge from the surveys is the greater attachment to place in the Gallo-Roman period than previously (or afterwards), with sites often now occupied continuously for periods of more than two centuries. In northern Burgundy, Nouvel argues for greater stability in site territories at this period.24 This is something we also see in the Aisne valley, with shifts in settlement focus now occurring less often and within more tightly defined territories.25 In the areas of Burgundy, Champagne-Ardenne and Picardy where Gallo-Roman data have been quantified, rural site numbers typically increase from the later Iron Age into Roman times, peaking in the 2nd or 3rd centuries AD. 26 This upward trend conceals a more complex pattern of old sites being abandoned and new ones founded during the later Iron Age, which continued to some extent in the Gallo-Roman period, despite the generally greater attachment to place. In the Aisne valley area, this increase in site numbers went hand in hand with the expansion of settlement onto the adjacent slopes and plateau edges, a phenomenon which I suggested reflected the opening up of the fertile soils of the plateau tops to intensive cultivation on a large scale, itself linked to growth in population.27 The data available from other parts of Picardy support this, as well providing better evidence of the initial phase of colonisation of the slopes and plateau edges during La Tène C1-C2.28 The availability of a mature iron technology and of crops suited to cultivation on heavier soils were evidently the key factors in this process of settlement expansion, but the warmer climate will also have contributed. A similar process of expansion is apparent in lower Burgundy.29 Here, the shift towards the plateaux in

21

Gransar et al. 1999. The periods of maximum settlement

able examples identified by fieldwalking, see Haselgrove/

visibility are Bronze final IIIb (c. 930-800 BC); Hallstatt D2-3/La Tène A (c. 530-400 BC); and from La Tène C2 22

Scull 1992, figs. 7, 9. 26

onward. See also Brun/Ruby 2009, 55.

beek/Van der Leeuw 2003); the lower Burgundy plateau

In Hallstatt C (c. 800-640 BC) and La Tène B2-C1 (c.

(Nouvel 2004); Perthois (Vanmoerkerke 2009), and Valois

325-270 BC). See also Brun/Ruby 2009, 55. 23

E.g. the Aisne valley (Haselgrove 1996); Argonne (Gazen-

(J.-L. Collart pers. comm.).

See Malrain et al. 2005, fig. 18.

27

Haselgrove 1996, 146; 2007a, 502-504.

24

Nouvel 2005.

28

Malrain et al. 2005; Malrain/Pinard 2006.

25

E.g. Limé; Soupart/Duvette 2005, fig. 2. For some prob-

29

Nouvel 2004.



the late Iron Age was accentuated in the Augustan period and settlement achieved its maximum in the early 3rd century AD. The shift away from the valley bottoms after the Iron Age apparent here and in the Aisne valley also occurs in Oise30 and on the chalkland of the Pays D’Othe, although there the base of the slopes in secondary valleys rather than the plateau now became the preferred focus of settlement.31 Whilst it seems likely that the overall rise in settlement numbers and expansion onto the plateaux reflect population increase, this is more difficult to prove. Some possible supporting evidence comes from northern Burgundy, where Nouvel has shown that the total surface area of settlements increased in parallel to site numbers.32 It may also be relevant that the one area without the usual peak in the High Empire is the Dry Champagne. Both at Vatry and along the LGV route, we do see an increase in sites in the late Iron Age, but in neither case is this sustained beyond the 1st century AD and the total number is markedly lower than in the mid first millennium BC.33 Given the widespread evidence of erosion on the chalk by the later Iron Age, the long held view that over-exploitation of the lighter soils put a break on settlement expansion seems likely to be correct. Around the same period as later Iron Age colonisation of the plateaux begins, linear ditches, enclosures and field systems become increasingly common on the alluvial terraces of the major river valleys and elsewhere. Detailed analysis of the palimpsest evidence at Longueil-Sainte-Marie and Verberie, Les Gâts, in the Oise valley, implies that in some instances at least, enclosure was undertaken on a grand scale and not simply piecemeal.34 The labour and cooperation involved in digging and maintaining these ditches may also have served to integrate groups more closely. The form and layout suggests that agricultural concerns were paramount: to separate arable from pasture and for stock management; to drain land that was otherwise too wet for permanent settlement; and perhaps to emphasise property boundaries in a landscape that was fast filling up; in the Aisne valley, for example, many farms were no more than 1 km apart by La Tène D1.35 Pollen analysis shows that by then much of the bottom of the valley was given over to pasture, with some woodland, whilst cereal agriculture was focused further away on the lower slopes and plateaux, another shift with major implications for rural settlement development in GalloRoman times. Crop husbandry underwent major changes in the later Iron Age. Monoculture became the rule – previously other species were often grown alongside the main crop – and bread wheat and millet increased in importance.36 Grain storage also changed, with a reduction in the use of underground silos, the advent of large pottery vessels (dolia), and fewer storage facilities at producer sites.37 Cereals were no longer stored as grain, but on the ear or without the glumes having been removed. Gransar interprets this as indicating that surplus cereal was taken to centralised sites and/or used in long-distance trade. However, whilst he analysed data from all over northern France, terrace sites predominate in the sample and the changes in storage may just reflect changing land use in this particular environmental niche. Before the 3rd century BC, open settlements predominated, but from La Tène C1, more and more were enclosed. Settlement enclosures vary from palisades to very substantial ditches, although these may be more a display of status than for protection. Over time, there is a tendency for enclosures to become more rectilinear and to be subdivided internally into discrete zones. Internal structures include large timber buildings, generally rectangular or with rounded ends; most are post-built, although sleeper-beam construction and continuous wall trenches do occur. Many of these larger buildings are clearly houses, but finds imply that others were barns, byres or workshops. The houses are generally accompanied by smaller storage structures on four, six, or nine posts. A common pattern is for the residential buildings

30

Malrain/Pinard 2006.

34

Malrain/Pinard 2006, 59-63.

31

Nouvel 2005, fig 7.

35

Thouvenot/Gransar 2000.

32

Nouvel 2004.

36

Matterne 2001; Zech-Matterne et al. 2009.

33

Vanmoerkerke 2009, fig. 30.

37

Gransar 2000.



and ancillary structures to be ranged around the enclosure boundary looking on to a central open space, or alternatively for the enclosure to be subdivided, both foreshadowing arrangements common in the Gallo-Roman period.38 Due to the frequent shifts in focus, valley-bottom sites often appear as palimpsests and even ‘simple’ enclosures often prove on excavation to be a composite of minor changes over time and occasional major reorganisations. Equally, whilst many farms seem to be isolated, clusters of sites are sometimes recognisable, as at Saint-Laurent-Blangy,39 a phenomenon that may reflect the pattern of colonisation or particular social networks. Some sites like Herblay comprise two enclosures, one for the inhabitants, the other for agriculture and other activities.40 Not all enclosures were residential. At Braine, La Grange des Moines (Aisne), large quantities of pig and sheep bone, pottery, wine amphorae and other high status items were found in the ditches of a square enclosure and annexe of La Tène D1b date, suggesting this was used for feasting and sacrifice.41

4 settleme n ts a n d social orga n isatio n Few later Iron Age enclosed settlements are likely to have housed social units larger than the extended family, but there is some reason to believe that their status varied. Starting from the size of the enclosed area, but including attributes such as the scale of the ditches, the types of buildings and finds, and even the quality of the meat eaten by the inhabitants, Malrain has argued for a fourfold hierarchy of sites in the Oise valley, which he suggests reflect differences in social status.42 They range from sites like Le Vivier des Grès, where the dwelling lay in its own compound (as at Herblay), which Malrain ascribes to the top of the assumed hierarchy, to the more numerous small enclosed farms and open sites, which are placed at the bottom. Whilst the fine detail can be debated, Malrain’s model appears to reflect some real differences between sites. Even so, the predilection of Iron Age archaeologists to opt for hierarchical interpretations is regrettable – if understandable given Julius Caesar’s account of Gaulish society and the widespread occurrence, albeit mainly in burials, of objects to which we would ascribe high status today and the spectacular acts of consumption that took place in the period. Nevertheless as Hill has argued,43 it is another matter to assume that all Iron Age societies were necessarily hierarchical or that elites existed everywhere. Many attributes of Gaulish societies might be better explained by forms of social organisation based on the principle of heterarchy.44 The segmentary societies described by Fortes and Evans-Pritchard, which lacked central authority and sharp divisions of rank, status or wealth, provide another possible model.45 While this is not the place to go into the specifics of Caesar’s description, the fact that he probably derived much of the detail from his Aeduan allies alerts us the pitfalls of generalising to the rest of Gaul. A possible case of over-readily accepting hierarchical readings of settlement evidence is provided by the La Tène D2 site at Saint-Denis-lès-Sens, in northern Burgundy.46 In general appearance – a group of buildings surrounded by a trapezoidal palisade – this differs little from late Iron Age farms all over eastern France, the only unusual find being a hoard of 242 gold globules-à-la-croix from one of the postholes of the main building (fig. 2). At first glance, then this might seem the epitome of a late Iron Age elite farmstead, but there are certain features that make one uneasy about this reading.47 They include the

38

Haselgrove 2007a, 504-506.

43

Hill 2006.

39

At the Arras Actiparc site; see Jacques/Prilaux 2003.

44

E.g. Crumley 1995; Thurston 2009.

40

Valais 1994.

45

Hill 2006; Sastre 2008.

41

Auxiette et al. 2000.

46

Joly 1994.

42

E.g. Malrain et al. 2002, 137-158.

47

Haselgrove 2009.



Fig. 2. The La Tène D2 settlement and Augustan enclosure at Saint-Denis-lès-Sens (Yonne).

presence of a rare circular building; the proximity of the site to a ford; and many of the ‘status’ finds are in fact from the ditch of the overlying Augustan enclosure, which has no internal structures. Whilst none of this prevents the site having started as an elite residence – which might later have become a place of memory – it does raise doubts. Similar cautions have been expressed over the interpretation of a number of enclosed sites with unusual features in Picardy.48 In pursuing alternative visions of Gaulish societies, we need look no further than the large settlement at Acy-Romance (Ardennes) on the middle Aisne.49 Excluding a cult area in the south-west corner, the site was divided into three quarters, each ranged around a large yard. Many living units were rebuilt up to three times. Based on their analysis of the finds and meat consumption, Lambot and Méniel suggest that each yard was the focus of a discrete occupational group, within each of which were finer gradations of status and wealth.The northern court has the largest houses and its inhabitants were primarily engaged in animal husbandry and manufacturing secondary products. The buildings in the eastern court are simpler and associated with silos, suggesting more concern with arable cultivation. The southern court, nearest the cult area, yielded a concentration of metalworking debris, a transformative activity that may well have required divine assistance and protection. This segregation along socio-occupational lines seems at odds with more conventional readings of Gaulish society50 and again suggests we need to consider alternative scenarios, in this case perhaps a form of caste system.51

48

Malrain et al. 2007.

50

E.g. Buchsenschutz 2004.

49

Lambot 1999; 2002; Méniel 1999; 2001.

51

Haselgrove 2007a, 507, 509.



Very different from Acy-Romance is a La Tène D2 rural site excavated in 2007 at Varennes, La Justice (Seine-et-Marne), just outside the extensive contemporary settlement at Le Marais du Pont, near the Seine-Yonne confluence. The northern end of La Justice is a fairly typical settlement enclosure, but attached to this is a much larger, elongated area some 2 ha in extent.52 Within this space, a very similar range of productive activities to Acy-Romance was performed, including metalworking, while the residential enclosure yielded a number of potentially high-status finds including weaponry and a dispersed hoard of globules-à-la-croix like those from Saint-Denis-lès-Sens. Some 350 m to the south-west of the complex is a cremation cemetery, which yielded further high status finds as well as large quantities of smashed amphorae, the contents presumably consumed in the course of the funerary rituals. Not unsurprisingly given the evidence, the excavator has interpreted La Justice as an aristocratic residence right at the top of the social hierarchy.53 The contrast with Acy-Romance, where agricultural and craft activities were rigidly separated, is interesting, especially with the inhabitants of the nearby aggregation at Le Marais du Pont, engaged in a similar range of craft activities.54 This recalls the situation at Manching (Bavaria), where specialist producer zones in the heart of the oppidum are juxtaposed with palisaded compounds, located towards the periphery, at which similar craft activities were performed,55 raising the question of what this parallel organisation of craft activities might signify. Are we looking at independent social groups, or were the inhabitants of Le Marais du Pont under the control of those who lived at La Justice and others like them? Or given the number of high quality finds from La Justice, should we wary of interpreting it as a residential site? For a third configuration, we may turn to Saint-Laurent-Blangy, where 3 km2 were examined prior to the building of the Actiparc.56 The investigations revealed the remains of five later Iron Age enclosed farms, as well as three cemeteries and a network of tracks. From its location and the presence of bronzeand salt-producing workshops, one of the farms is thought to be of higher status than the rest and it has been suggested that what we have here is an estate belonging to the principal farm.57 In the mid 1st century BC, a Roman fortlet and canabae were built over this farm – viewed as confirmation of its dominant role – whilst other nearby farms continued to be occupied. Without going into more detail, the evidence of these three very different sites leaves little doubt that we need to consider a wider range of possible models of Iron Age social organisation than hitherto and no one model is likely to fit all the societies inhabiting what became Roman Gaul.

5 site co n ti n uit y The final question to be examined here is to what extent late Iron Age rural sites were occupied continuously into the Roman period? Thanks to Archéologie préventive, we know of numerous sites occupied in both periods, but when you look more closely at sequences, unimpeachable evidence of continuity through the 1st century BC and into Gallo-Roman times is rare.This is an issue that I first began to think about in connection with the site I excavated in the 1980s at Beaurieux, Les Grèves (Aisne). Initially, the sequence appeared to be one of continuity during the 1st century BC, but more detailed analysis revealed a chronological gap between the La Tène D1 open site and the enclosed settlement laid out in the second half of the 1st century BC.58

52

Séguier 2008.

56

Jacques/Prilaux 2003.

53

Ibid.

57

Ibid.; Brun/Pascal 2008, 119.

54

Griffisch et al. 2008, 1092-1095.

58

Haselgrove 1996, 155-161.

55

E.g. Collis 1984, 109-136.



Fig. 3. Later Iron Age site continuity in the Aisne valley.

Such gaps can just reflect the vagaries of site formation or even artefact chronologies, but over the years I became aware that none of the other rural sites excavated on the Aisne terraces were occupied throughout La Tène D2 either (fig. 3). The only settlements attributable to the period were short-lived oppida like Villeneuve-Saint-Germain and – on the plateau above – Pommiers and Saint-Thomas, all new builds laid out to a preconceived plan.59 Given the lack of rural farms on the terraces and applying

59



Haselgrove 1995, 82-84; 1996, 147-152; 2007a, 507-511.

a model developed by John Collis,60 I argued that the oppida were inhabited by people who had moved to the fortifications from farms in the environs. The farm at Villeneuve-Saint-German, Les Étomelles, which was abandoned at the end of La Tène D1,61 just as the oppidum was founded 800 m away, is a particularly clear example. Interestingly, the interiors of both Villeneuve and its successor at Pommiers are subdivided by ditches into unequal quarters, which at the former are characterised by distinct types of buildings, artefacts and meat consumption, 62 implying that the inhabitants of both sites were segregated on socio-occupational lines similar to Acy-Romance further up river. Whatever factors caused people in southern Picardy to abandon their farms and live for a while in fortified oppida, it is tempting to link the post-Conquest reoccupation of the river valleys with a group of large, regularly laid-out rural sites, which have been excavated on the terraces.63 Each site comprises a series of separate dwellings with their own yards, pits and wells, laid out around an open space in a manner that anticipates the courtyard villa. At both Bazoches-sur-Vesle and Verneuil-en-Halatte, what seems to be the main dwelling occupied a compound at one end of the courtyard (recalling the arrangement at Herblay and Varennes), leading to the suggestion that the dominant household lived here, whilst the other units were occupied by their social inferiors.64 All but one of these ‘planned’ settlements went on to become villas and, to return to a point made earlier, we clearly have a possible instance of the specific historical conditions that affected this region in the 1st century BC influencing subsequent settlement developments, by causing the rural population to live together in larger groups than previously.65 I now realise that I was in error previously in overlooking the possibility that in the Aisne valley this discontinuity in rural settlement during La Tène D2 was largely confined to the alluvial terraces, in the immediate vicinity of the short-lived oppida.The recently excavated settlement at Ploisy on the Soissonais plateau was occupied at the same time as Villeneuve-Saint-Germain in the valley below and Pommiers.66 The fieldwalking results on the slopes and plateaux also hint at a greater degree of continuity there in the 1st century BC and onward into the Roman period.This makes sense: many plateau sites were recent foundations, specialising in arable farming in a landscape which itself had only been fully opened up to the plough not long ago. A social tradition of settlement mobility geared to the long-exploited valley bottom soils would have had less relevance here. Equally, it would have been more difficult for the inhabitants of these farms, with the attachment to the soil that cereal farming brings, to have relocated temporarily into oppida, than for their counterparts in the valley, where the land was predominantly used for animal husbandry. Although the detailed evidence has yet to be presented, there is no obvious sign of a hiatus between the Late Iron Age farm at Ploisy and the Roman site that followed it.67 It will be interesting to see what

60

E.g. Collis 1984, 83-85.

rebuilt. In Britain, such buildings are usually dismissed as

61

Hénon 2001.

barns or of low status, but from a study of examples in

62

Debord 1990; Auxiette 1996.

southern England, Cunliffe (2008, 113-129) has argued

63

Apart from Beaurieux, examples include the site at Juvin-

that they were actually the focus of many early villas. At

court-et-Damary, Le Gué de Mauchamps; Bazoches-sur-

Beaurieux, the location of the aisled structure away from

Vesle, La Foulerie and Limé, Les Terres Noires in the

the centre of the site (where the main Roman dwelling is

Vesle valley; and Verneuil-en-Hallatte, La Bufosse, on the

thought to lie) and the presence of other Augustan timber

Oise; see Collart 1996; Haselgrove 1996, 156-160; Ben

buildings of similar size probably militates against such

Redjeb et al. 2005, 192-195; Duvette 2005.

an interpretation, but the role of aisled buildings in Gaul,

64

Collart 1996.

which, whilst not as common as in Britain, are found in

65

In passing, it is worth highlighting the presence at Beau-

some numbers in Burgundy, Picardy and eastern France

rieux of a possible timber aisled hall, located against the

(Ferdière 1988), would certainly repay fresh attention.

southern perimeter (Haselgrove 1996, fig. 13C). This

66

Gransar 2003, 41.

was built in the Augustan period and subsequently twice

67

Ibid.



pattern emerges as more plateau sites are excavated in this area and whether continuity proves to be the rule. At present the picture in other parts of Picardy appears to echo Aisne, with more sites abandoned in La Tène D2 than founded and a general dearth of rural settlements of this period,68 despite the sample of sites from Oise and especially Somme being less dominated by the alluvial terraces. Further afield, a similar degree of discontinuity is apparent on the gravels of the Seine-Yonne confluence and on the Saône-Doubs plain, although in both areas more La Tène D2 rural sites have been identified than in the Aisne valley.69 A final comparison may be drawn with the Arroux valley in southern Burgundy. Only a small area was fieldwalked, but the indications are that the pattern is similar, with few sites occupied in La Tène D2, coinciding with the most intensive occupation of Mont Beuvray and then a rapid regeneration of rural settlement from the Augustan period onward in tandem with the growth of Autun.70 Interestingly, fieldwork has identified later 1st century BC occupation over an area of several hectares around the source of the Yonne, only 3 km to the north of Mont Beuvray, clearly contemporary with the oppidum.71 The nature of the activity at the source of the Yonne and at another complex of this date at Poil to the south of Mont Beuvray is unclear – both might be ceremonial foci and/or places of periodic assembly – but it involved consumption of Dressel 1 amphorae on a similar scale to the oppidum itself, a warning against drawing too rigid a distinction between fortified and other sites.

6 co n clusio n Preliminary indications are, then, that there was a significant break in the settlement pattern in the 1st century BC in many river valleys of northern and eastern France and in the environs of other major Late Iron Age centres. This rupture may well relate to the temporary aggregation of the rural population into fortified oppida and was of a different order of magnitude from the discontinuities apparent in the settlement record in the earlier Iron Age, which probably reflect a combination of a tradition of social mobility and the influence of climate. On the recently colonised soils of the valley slopes and plateaux the degree of settlement stability in the 1st century BC may well prove to be rather greater, but we lack detailed information. In general, the attachment to place seems to be much greater in the Gallo-Roman period, but it is clear that in many areas, the alluvial terraces were no longer the principal focus of settlement, at least in the High Empire, and that settlement expansion was concentrated in other areas of the landscape. If we are to understand the development of villa landscapes on a regional basis, these major discontinuities in settlement in certain areas and/or environments and also the imbalances in our existing evidence are factors we need to take into account, whilst also paying more attention to the similarities that exist between Late Iron Age and Gallo-Roman rural settlements, for instance in layout. In general, we need to be more willing to cross the divide between prehistory and Roman archaeology and to consider how regional variations in the character of Iron Age societies and their participation in longer-term processes of social

68

69



Malrain et al. 2005, 142-144. The sample of sites analysed

these areas along with other case studies from Alsace-

includes cemeteries as well as settlements.The cemeteries

Lorraine, lower Burgundy and Champagne, also points

display a marked topographic bias to slope and plateau

to a strong element of settlement discontinuity in the 1st

edge locations, possibly they were one of the ways in

century BC, which is also more marked on the alluvial

which the new settlers expressed their claim to particular

terraces than on the plateaux and which also appears to

tracts of land (Haselgrove 2007a, 498).

start earlier in the eastern part of this zone than in the

Horard-Herbin et al. 2000, fig. 2, for the Seine-Yonne

west.

confluence; Barral 2005, fig. 2, for the Saône valley. The

70

Creighton et al. 2008.

recent survey by Nouvel et al. 2009, which includes both

71

Haupt et al. 2007.

and landscape development influenced Roman settlement patterns. We also need to pay more attention to conceptual issues and be prepared critically to question some habitual assumptions about the nature of Iron Age societies. For too long, archaeology has treated the transformation of Iron Age rural settlement into Gallo-Roman villa landscapes as unproblematic – a form of progress with which we can identify ourselves – but it is clear that Iron Age peoples not only held different beliefs and values to modern Western societies, but also often to each other. For them, the transition was neither an obvious nor an easy one to make.

re f ere n ces Auxiette, G., 1996: La faune de l’oppidum de Villeneuve-Saint-Germain (Aisne): quartiers résidentiels, quartiers artisanales, Revue Archéologique de Picardie 1996, 1/2, 27-98. Auxiette, G./S. Desenne/F. Gransar/C. Pommepuy, 2000: Structuration générale du site de Braine “La Grange des Moines” (Aisne) à La Tène finale et particularités. Présentation préliminaire, Revue Archéologique de Picardie 2000, 1/2, 97-103. Agache, R., 1978. La Somme pré-romaine et romaine d’après les prospections aériennes, Amiens (Mémoires de la Société des Antiquaires de Picardie, 24). Barral, P., 2005: L’habitat rural de la fin de l’Age du Fer (La Tène C2-D1) dans les plaines de Sâone et du Doubs: données socio-économiques et culturelles à partir du matériel céramique, in S. Fichtl (ed.), Hiérarchie de l’habitat rural dans le Nord-Est de la Gaule à La Tène moyenne et finale, Strasbourg (Archaeologia Mosellana 6), 239-274. Ben Redjeb, T./L. Duvette/P. Quérel, 2005: Les campagnes antiques. Bilans et perspectives, Revue Archéologique de Picardie 2005, 3/4, 177-222. Bertrand, I./A. Duval/J. Gomez de Soto/P. Maguier (eds), 2009: Habitats et paysages ruraux en Gaule et regards sur d’autres régions du monde celtique, Chauvigny (Association des Publications Chauvinoises Mémoire 35). Bourgeois, J./I. Bourgeois/B. Charetté (eds), 2003: Bronze Age and Iron Age communities in north-western Europe, Brussels. Braemer, F./S. Cleuziou/A. Coudart (eds), 1999: Habitat et Société. XIXe Rencontres Internationales d’Archéologie et d’Histoire d’Antibes, Antibes. Brun, P./P. Ruby, 2008: L’Age du Fer en France, Paris. Buchsenschutz, O., 2004: Les Celtes de l’Age du Fer dans la moitié nord de la France, Paris. Collis, J., 1984: Oppida. Earliest towns north of the Alps, Sheffield. Collart, J., 1996: La naissance de la villa en Picardie. La ferme gallo-romaine précoce, in D. Bayard/J.-L. Collart (eds), De la ferme indigène à la villa romaine, Amiens (Revue Archéologique de Picardie Numéro Spécial, 11), 121-156. Courbot-Dewerdt, C., 2004 : L’évolution des établissements ruraux entre la fin de l’âge du fer et la mise en place du systeme des villae dans le quart nord-ouest de la france, Ier siecle avant-IIeme siecle apres J.-C, Paris (Unpublished Ph.D. thesis, Université de Paris I). Creighton, J./C. Haselgrove/P. Lowther/T. Moore, 2008: Becoming Roman in southern Burgundy. A field survey between Autun and Bibracte in the Arroux valley (Saône-et-Loire), 2000-3. Internet Archaeology 25. http://intarch.ac.uk/journal/issue25/creighton_index.html Crumley, C.,1995: Heterarchy and the analysis of complex societies, in R. Ehrenreich/C. Crumley/J. Levy (eds), Heterarchy and the analysis of complex societies, Washington (Archaeological Papers, American Anthropological Association 6), 1-6. Cunliffe, B.W., 2008: The Danebury Environs Roman Programme. A Wessex landscape during the Roman era. Volume 1 Overview, Oxford (Oxford University School of Archaeology Monograph 70). Debord, J., 1990: Les fouilles du site gaulois tardif de Villeneuve-Saint-Germain (Aisne), Mémoires de la Fédération des Sociétés d’histoire et de l’archéologie de l’Aisne 35, 137-170.



Duvette, L., 2003: Ploisy, Le Bras de Fer - Zones 3 et 5, Amiens (DRAC/SRA Picardie, Bilan Scientifique 2005), 42–44. Duvette, L., 2005: Les établissements gallo-romains précoces dans les vallées de l’Aisne et de la Vesle, in O. Buchsenschutz/C. Mordant (eds), Architectures protohistoriques en Europe occidentale du Néolithique à l’Age du Fer, Paris, 217-229. Ferdière, A., 1988: Les campagnes en Gaule romaine, Paris. Gazenbeek, S./S. van der Leeuw, 2003: L’Argonne dans l’Antiquité. Étude d’une region productrice de céramique et de verre, Gallia 60, 269-317. Gerritsen, F.A., 2003: Local Identities. Landscape and community in the late prehistoric Meuse-Demer-Scheldt region, Amsterdam (Amsterdam Archaeological Studies 9). Gransar, F., 2003: Ploisy, Le Bras de Fer - Zones 1 et 7, Amiens (DRAC/SRA Picardie, Bilan Scientifique 2005), 40-42. Gransar, F., 2000: Le stockage alimentaire sur l’établissements ruraux de l’Âge du Fer en France septentrionale. Complémentarité des structures et tendances évolutives, in S. Marion/G. Blancquaert (eds), Les installations agricoles de l’Age du Fer en France septentrionale, Paris (Etudes d’Histoire et d’Archéologie 6), 277-298. Gransar, F./G. Auxiette/S. Desenne/B. Hénon et al. 1999: Essai de modélisation de l’organisation de l’habitat au cours des cinq derniers siècles avant notre ère dans la vallée de l’Aisne, in F. Braemer/S. Cleuziou/A. Coudart (eds), Habitat et Société. XIXe Rencontres Internationales d’Archéologie et d’Histoire d’Antibes, Antibes, 419-438. Griffisch, J.-N./D. Magnan/D. Mordant, 2008: Carte Archéologique de la Gaule. La Seine-et-Marne 77, Paris. Haselgrove, C., 1991:The Romanization of Belgic Gaul. Archaeological perspectives, in T. Blagg/M. Millett (eds), The early Roman Empire in the West, Oxford, 45–71. Haselgrove, C., 1995: Late Iron Age society in Britain and north-west Europe: structural transformation or superficial change?, in B. Arnold/D.B. Gibson (eds), Celtic chiefdom, Celtic state, New York, 81-87. Haselgrove, C., 1996: Roman impact on rural settlement and society in southern Picardy, in N. Roymans (ed.), From the sword to the plough. Three studies on the earliest romanization of Northern Gaul, Amsterdam (Amsterdam Archaeological Studies 1), 127-187. Haselgrove, C., 2007a: The age of enclosure. Later Iron Age settlement and society in northern France, in C. Haselgrove/T. Moore (eds), The later Iron Age in Britain and beyond, Oxford, 492–522. Haselgrove, C., 2007b: Rethinking Earlier Iron Age settlement in the eastern Paris Basin, in C. Haselgrove/R. Pope (eds), The earlier Iron Age in Britain and the near Continent, Oxford, 400–428. Haselgrove, C., 2009: Noughts and crosses; the archaeology of ‘Globules-à-la-croix’, in J. van Heesch/I. Heeren (eds), Coinage in the Iron Age. Essays in honour of Simone Scheers, London, 173-186. Haselgrove, C./C.J. Scull 1992: The Romanization and de-Romanization of Belgic Gaul. The rural settlement evidence, in M. Wood/F. Queiroga (eds), Current research on aspects of the romanization of the western provinces, Oxford (British Archaeological Reports International Series 575), 9-24. Haupt, P./I. Klenner/M. Schönfelder, 2007: Prospections sur le site des sources de L’Yonne, Commune de Glux-en-Glenne, Glux-en-Glenne (Rapport annuel d’activité 2007 du Centre Archéologique Européen), 204-209. Hénon, B., 2001: Villeneuve-Saint-Germain. Les Étomelles, Amiens (DRAC/SRA Picardie Bilan Scientifique 2001), 51-52. Hénon, B., 2005: Courmelles. La Plaine du Mont de Courmelles, Amiens (DRAC/SRA Picardie, Bilan Scientifique 2005), 22. Hill, J.D., 2006: Are we any closer to understanding how later Iron Age societies worked (or did not work)?, in C. Haselgrove (ed.), Les mutations de la fin de l’Âge du Fer, Glux-en-Glenne (Collection Bibracte 12/4), 169-179.



Horard-Herbin, M.-P./P. Méniel/J.-M. Séguier 2000: La faune de dix sites ruraux de la fin de l’Age du Fer de La Bassée, in S. Marion/G. Blancquaert (eds), Les installations agricoles de l’Age du Fer en France septentrionale, Paris (Etudes d’Histoire et d’Archéologie 6), 181-208. Jacques, A./G. Prilaux (eds), 2003: Dans le sillage de César. Traces de romanisation d’un territoire, les fouilles d’Actiparc à Arras, Arras. Joly, M., 1994: La ferme gauloise de Saint-Denis-lès-Sens, in L. Faton (ed.), Découvertes archéologiques sur l’autoroute A5, Sens, 40-43. Lagatie, C./J. Vanmoerkerke, 2005: Europort Vatry (Marne). Les pistes de l’archéologie, Langres. Lambot, B., 1999: Organisation spatiale et sociale du village gaulois d’Acy-Romance (Ardennes), in F. Braemer/S. Cleuziou/A. Coudart (eds), Habitat et Société. XIXe Rencontres Internationales d’Archéologie et d’Histoire d’Antibes, Antibes, 383-405. Lambot, B., 2002: Maisons et société à Acy-Romance (Ardennes), in P. Méniel/B. Lambot (eds), Repas des vivants et nourriture pour les morts en Gaule, Reims (Mémoire de la Société Archéologique Champenoise 16), 115-124. Malrain, F./S. Gaudefroy/F. Gransar/G. Auxiette/P. Méniel 2005: La protohistoire récente, Revue Archéologique de Picardie 2005, 3/4, 127-176. Malrain, F./L. Blondiau/C. Chaidron 2007: Les enclos laténiens sont-ils toujours des fermes?, Revue Archéologique de Picardie 2007, 3/4, 17-55. Malrain, F./V. Matterne/P. Méniel 2002: Les paysans gaulois (IIIe siècle–52 av. J.-C.), Paris. Malrain, F./E. Pinard 2006: Les sites laténiens de la moyenne vallée de l’Oise du Ve au Ier siècle avant notre ère, Amiens (Revue Archéologique de Picardie Numéro Spécial 23). Marion, S., 2004: Recherches sur l’Age du Fer en Ile-de-France, Oxford (British Archaeological Reports International Series 1231). Matterne, V., 2001: Agriculture et alimentation végétale durant l’àge du Fer et l’époque gallo-romaine en France septentrionale, Montagnac (Archéologie des Plantes et des Animaux 1). Zech-Matterne, V./L. Bolibya/A. Bouchette/A. Cabanis et al., 2009: L’agriculture du VIe au Ier siècle av. J.-C. en France. État de recherche carpologiques sur les établissements ruraux, in Bertrand, I./A. Duval/J. Gomez de Soto/P. Maguier (eds), Habitats et paysages ruraux en Gaule et regards sur d’autres régions du monde celtique, Chauvigny (Association des Publications Chauvinoises Mémoire 35), 383-416. Méniel, P.,1999: Histoire de l’alimentation carnée et de l’organisation sociale du village gaulois d’AcyRomance (Ardennes, France), in F. Braemer/S. Cleuziou/A. Coudart (eds), Habitat et Société. XIXe Rencontres Internationales d’Archéologie et d’Histoire d’Antibes, Antibes, 405-418. Meniel, P., 2001: Les gaulois et les animaux. Elevage, repas et sacrifice, Paris. Monteil, M./L. Tranoy 2008: La France gallo-romaine, Paris. Nouvel, P., 2004: Des terroirs et des hommes, dynamiques des organisations spatiales dans le bassin de l’Yonne moyenne et leur évolution de la fin de l’Age du Bronze au Haut Moyen Age, Unpublished Ph.D. thesis, Université de Bourgogne. Nouvel, P., 2005: Les établissements ruraux laténiens de basse Bourgogne d’après des donnés de prospections, in S. Fichtl (ed.), Hiérarchie de l’habitat rural dans le Nord-Est de la Gaule à La Tène moyenne et finale, Strasbourg (Archaeologia Mosellana 6), 327-350. Nouvel, P./P. Barral/S. Deffressignes/V. Riquier/J.-M. Séguier/N. Tikonoff/Zehner 2009: Rhythmes de création, fonctionnement et abandon des établissements ruraux de la fin de l’Age du Fer dans l’Est de la France, in I. Bertrand/A. Duval/J. Gomez de Soto/P. Maguier (eds), Habitats et paysages ruraux en Gaule et regards sur d’autres régions du monde celtique, Chauvigny (Association des Publications Chauvinoises Mémoire 35), 109-151. Rapport au Parlement 2006: Rapport au Parlement Mise en ouvre de la loi modifiée du 17 janvier 2001 relative à l’archéologie preventive.Volumes 1-2. http://www.culture.gouv.fr/culture/actualites/rapports/archeo-preventive2006/rapport-archeo.html



Roymans, N./F.A. Gerritsen 2002: Landscape, ecology and mentalities. A long-term perspective on developments in the Meuse-Demer-Scheldt region, Proceedings of the Prehistoric Society 68, 257-287. Sastre, I., 2009: Community, identity, and conflict, Current Anthropology 49, 1021–1051. Séguier, J.-M., 2008: Établissement aristocratique et nécropole de la Tène finale à Varennes-sur-Seine (Seine-et-Marne), Bulletin de l’Association Française de l’Âge du Fer 26, 55-56. Soupart, N./L. Duvette 2005: Limé “Les Sables” (Aisne). Les sépultures et les depots de La Tène, in G. Auxiette/F. Malrain (eds), Hommages à Claudine Pommepuy, Amiens (Revue Archéologique de Picardie Numéro Spécial 22), 289-326. Thouvenot, S./F. Gransar 2000: La gestion du terroir des établissements ruraux de La Tène finale dans la vallée de l’Aisne, in S. Marion/G. Blancquaert (eds), Les installations agricoles de l’âge du Fer en France septentrionale, Paris (Etudes d’Histoire et d’Archéologie 6), 157-167. Thurston, T., 2009: Unity and diversity in the European Iron Age. Out of the mists, some clarity?, Journal of Archaeological Research 17, 7-84. Valais, A., 1994: La ferme des Fontaines à Herblay (Val-d’Oise), in O. Buchsenschutz/P. Méniel (eds), Les installations agricoles de l’Âge du Fer en Ile-de-France, Paris, 113-124. Vanmoerkerke, J., 2009: Le bassin de la Vesle du Bronze final au Moyen Age à travers les fouilles du TGV Est, Reims (Bulletin de la Société Archéologique Champenoise 102 no. 2). Wightman, E.M.,1985: Gallia Belgica, London.



Exploring villa development in the northern provinces of the Roman empire Diederick Habermehl1

1 2 3 4 5 6 7

Introduction Developments in the organisation of settlement space Reorganising settlement space, reorganising relationships Developments in house building Building new houses, building new relationships Developments in storage Discussion

References

1 i n troductio n Settling in a changing world, creating a new place for oneself in the rapidly evolving environment of the Roman provinces, and the impact this had on each and every dimension of the lives of local people – that is the core theme of this paper. In general terms, the development of the northernmost provinces of the Roman empire involved the creation of a new administrative structure that included civitates and their capitals, many other urban and rural centres connected by a network of well-constructed roads, and a series of military camps concentrated along the Rhine in particular. These developments can be linked to significant changes to the economic, social, cultural, demographic and political spheres. New markets opened up, new institutions of power were created, new lifestyles were introduced and people’s mobility increased significantly. Particularly within the context of their local settlements, the rural population had to deal with this changing world and to create a new place within it, by changing not just the way in which they inhabited and worked the land but how they related to each other and to the outside world. These processes of change can be studied by exploring the development trajectories of rural settlements. After all, the adoption of new materials, forms, objects and (spatial) concepts can be regarded as a way of redefining relationships within local communities as well as between them and the outside world. This paper is about individual people, families and communities actively creating a new place for themselves in the changing world of the Roman provinces and the empire. To understand more about significant changes in rural settlement and, ultimately, in local communities, as well as to critically assess the concept of ‘villa’ as often used in Roman archaeology, a study has 1

This study is based on PhD research conducted at 

in the Roman North: economy, culture, lifestyles led by Prof.

the VU University, Amsterdam, which was funded by

Nico Roymans and Dr. Ton Derks. See now Habermehl

the Netherlands Organisation for Scientific Research

2011.

(NWO) as part of the larger programme Villa landscapes



been designed with the following characteristics. First of all, development is the main focus. The reconstruction and interpretation of transformations in rural settlements can shed light on processes of change in rural communities. Secondly, a broad empirical approach has been chosen that reaches beyond the traditional definition of villa. While the latter mainly focused on the use of stone foundations and Romanstyle architecture, I will take a broader perspective that includes other types of settlement where changes have been documented.Thirdly, where possible, a long-term approach is taken. Looking at longer lines of development places us in a better position to pinpoint and understand the transformations taking place during the Roman period: they should not be studied in a diachronic vacuum. Fourthly, in addition to these broad empirical and chronological approaches, a geographical framework is useful as it includes no less than four modern-day countries (the Netherlands, Belgium, Germany and France). This allows us to study a variety of settlements from different regions, including the loess regions, traditionally regarded as ‘villa landscapes’, and the sand and clay regions that are often characterised as ‘non-villa landscapes’. Lastly, it is important to emphasise that this study is intended as a theoretically informed and interpretative study with a social focus. Instead of studying settlements and houses as autonomous objects, we should attempt to explore them in relation to the people who created them and lived their lives there.Viewed from this perspective, the research question then becomes: how and why did people change their immediate living environments, sometimes in very radical ways? In this paper, after a brief assessment of the concept of ‘villa’, I will introduce the research region and dataset. I will then reconstruct and describe a variety of development trajectories in the organisation of settlement space, house building and the construction of granaries throughout the research region. For the sake of clarity, I have schematised development trajectories to make them clearer; good-quality plans and phasings are at the heart of studying and understanding villa evolution. Next I will focus on how we should link these changes to the active and creative people who instigated them. In the last section, I will discuss our findings and views, and attempt to synthesise them.

the ‘ v illa ’ The fact that I have only referred to ‘villa’ a few times up to this point may already reveal my reservations about the word. These stem from the difficulties surrounding the definition and use of the term, its limitations and complex background. In general, the phenomenon referred to as ‘villa’ is only part of a broader and more complex rural development. In my view, the strict and essentialistic application of a villa definition for both data selection and analysis could limit our research scope and our understanding of the real complexity of the processes at hand. We also need to be aware of the potentially misleading and over-simplified associations of ‘villa’: its Roman-ness, its interpretation as the residence of an elite and its functioning as a rationally operating enterprise.2 As we will see, differences between rural settlements were so pronounced that such simplifications would undermine our understanding of complex historical realities. This is not to say that we should, or even could, ignore the term ‘villa’ altogether. The study of monumental villas is and remains an undeniable and vital part of provincial Roman archaeological research. However, we do need to be aware of the problematic nature of the term and the more complex realities that it conceals.

2

See the discussion in Roymans/Derks, this volume. For reflection on the definition of the term ‘villa’, see among others Hingley 1989, 2005; Woolf 1998; Martins 2005;



Slofstra 1991.

data a n d research As outlined above, the main objective of this study is to reconstruct and analyse villa development. This requires the availability of quality archaeological data, including detailed information on both morphology (structure, form and material of the built environment) and chronology (fairly detailed phasing of the development of houses and settlements; absolute and relative datings). As a consequence, only wellexcavated and well-published sites can be used for meeting these objectives. Overall, we could state that the minimum requirement is the availability of a well-documented house plan, preferably well-dated and properly phased. In an ideal situation, data on the broader settlement context and more long-term developments would also be available. For this reason, survey data are not included in the dataset. A related requirement is publication. Generally, only sites that have been published were included in the dataset.3 In addition, my approach to the definition of ‘villa’ means that the general category of rural settlements was inventoried without selecting data on the basis of, say, monumentality or Roman-style architectural elements. As a result, the dataset contains sites that archaeologists will generally regard as ‘villas’, as well as sites that are referred to as ‘native’, traditional settlements. With these objectives in mind, a dataset containing 270 sites was constructed (fig. 1).The research area has been divided into five subregions in order to make the large dataset more manageable and to create an enhanced picture of regionality.The northernmost sand and clay areas comprise the region directly south of the Roman limes (situated along the Rhine). This region has a rich tradition in settlement archaeology, especially from the 1970s onwards.4 A little more to the south, the Flanders region mainly comprises the sand and clay soils of the northernmost parts of Belgium, extending over large parts of the Flemish provinces of East and West Flanders, Antwerp, Brabant and Limburg. Here, settlement archaeology has taken off in recent years and research has been synthesised by Wim De Clercq among others.5 Further south, we enter the loess regions. The Dutch and German loess region comprises the relatively narrow loess belt wedged between the northern sandy region and the mountainous Eifel region in the south. In the Netherlands, only the southernmost tip of the province of Limburg is covered by loess soil. Unlike the previous regions, settlement archaeology is not well-developed here. Much villa research, focusing on monumental traces of habitation (the ‘main buildings’ of villa complexes), was carried out during the second half of the 19th and first half of the 20th century.6 Only a few good examples can be found of large-scale settlement studies (Kerkrade-Holzkuil, Heerlen-Trilandis, Beek-Maastricht-Aachen Airport are among the best researched sites in recent years).7 The Belgian loess region, therefore, encompasses the loess belt between the northern Dutch and Flemish sandy region and the southern mountainous Ardennes region. As in the Dutch and German loess region, much research on monumental buildings was conducted in the early years of archaeology.8 However, more recent work is also available, providing data suitable for this study. The last and southernmost region includes the northern French departments of Nord, Pas-de-Calais, Somme, Oise and Aisne (the latter three together forming the Picardy region). Development-led archaeology has flourished in this region, especially from the 1990s onwards (in connection with gravel extraction, the large-scale construction of roads and railroads and the development of commercial areas).9 One advantage of this has been a focus not only on monumental sites, but on all kinds of settlement – Roman and pre-Roman, simple and complex – which has probably produced a more balanced picture of rural settlement and its development. 3

4

In some cases, ‘grey’ literature (unpublished reports, etc.)

5

De Clercq 2003, 2009, this volume.

could be used. In general, there was no opportunity to

6



return to primary data.

7



See Bloemers 1978; Willems 1981; Slofstra 1991; Roy-

8



mans/Theuws 1991; Roymans 1996; Theuws/Roymans

9

On the development of archaeological research in this

1999; Gerritsen 2003; Vos 2009; Heeren 2009.

See Braat 1934 ; idem 1941 ; Van Es 1972. Tichelman 2005; idem 2010; idem, in prep. De Maeyer 1937; idem 1940. region, see Bayard/Collart 1996; Haselgrove 2007.



Fig. 1. General distribution of sites, included in the dataset of this study.

Table 1 presents the number of inventoried sites per subregion. Fig. 1 shows the distribution of these sites within the landscapes of the research region. Some general patterns can be identified. An especially tight cluster of sites is evident in the Dutch and German loess region. For the Dutch region, however, most of the sites were excavated before 1950, the heyday of villa archaeology in the region.10 The many sites in the German region are due to large-scale lignite mining activity and associated rescue archaeology.11 In addition, a cluster of sites in the Dutch river area relates to the well-developed settlement archaeology of the region. region

number of sites

percentage

Northern sand and clay areas

45

17 %

Flanders (sand and clay region)

19

7%

Dutch and German loess region

84

31 %

Belgian loess region

74

27 %

Northern France

48

18 %

Table 1. Number of excavated rural settlements included in the dataset, presented per sub-region. 10

See Roymans/Derks, this volume.

11

Gaitzsch 1986; idem, this volume; Kunow 1994; Lenz



1998; Lenz 1999; Heimberg 2002/2003.

At the heart of this study is the reconstruction of development trajectories. In the following I will shed light on a variety of development trajectories in rural settlements throughout the research region by exploring a range of well-documented examples. I will then discuss changes in the organisation of settlement space, in house building and in the construction of storage buildings. Next I will focus on the morphologicalarchitectural dimensions of these developments, looking at them from a social perspective, which involves people as active and creative agents in the process, and exploring their backgrounds and implications.

2 de v elo p me n ts i n the orga n isatio n o f settleme n t s p ace The first important element of what we regard as ‘villa development’ involves the reorganisation of settlement space, meaning how the constituent elements of the settlement, which generally include several houses, buildings with economic functions, ditches or walls, graves, ponds and wells, were arranged in relation to each other and how the settlement was defined within the broader landscape. I will discuss several examples from the different subregions of the research area. For the northernmost regions in particular, it is possible to reconstruct a long-term development with regard to the reorganisation of settlement space. For the Iron Age we can reconstruct a process whereby farmsteads became increasingly clustered and stable. Instead of separate farmsteads shifting across a large settlement territory (being rebuilt at a new location each generation), farmsteads were built closer together and were increasingly rebuilt at the same location.12 This process intensified during the Late Iron Age and into the Roman period. Many rural settlements in the northernmost regions consisted of a number of loosely clustered farmsteads during their early phases (1st century BC – (first half of the) 1st century AD) (fig. 2).13 Although these farmsteads formed a coherent settlement, space was not organised beyond the level of the farmstead during this period. At many sites, however, this situation changed during the 1st century AD. At Oss-Westerveld14 and Hoogeloon-Kerkakkers, settlement space was enclosed by ditches, probably during the first half of the 1st century AD, and houses were constructed parallel to these enclosures. In many other cases, settlement space was enclosed during the second half of the 1st century or even the 2nd century AD.15 These settlements now had a common organisational structure, clearly defined by the enclosure, both inwardly and outwardly. In some cases, settlement space was further differentiated. At Geldermalsen, Oss-Westerveld and Hoogeloon, separate compounds were created within the enclosed settlement (fig. 2).These separated farmsteads can be associated with material culture and architecture of a special nature. As we will discuss later, a portico house and a multi-roomed house on stone foundations were constructed at Oss and Hoogeloon respectively. At Geldermalsen, the entrance of the separated compound was accentuated and the associated find assemblage included fragments of military equipment, both indicating its special character. Over time, settlements in these northern regions thus became stable entities, well defined within the landscape. Settlement space was increasingly structured and indications for spatial segregation and dif-

12

13

For the reconstruction of this process, see Gerritsen 

selingh 2000; Vos 2009).

2003.

14

Wesselingh 2000.

See for example the early settlement phases of Rijswijk-

15

For example at Geldermalsen-Hondsgemet, Tiel-Passe-

De Bult (Bloemers 1978 and 1980), Tiel-Passewaaij

waaij, Wijk-bij-Duurstede-De Horden (Vos 2002, 2009)

(Heeren 2006 and 2009), Geldermalsen-Hondsgemet

and Druten-Klepperheide (Hulst 1978; Maas 2007). It

(Van Renswoude/Van Kerckhove 2009), Hoogeloon-

seems that only one farmstead was initially enclosed at

Kerkakkers (Slofstra 1985, 1991; Jeneson 2004) as well as

De Horden and Druten.

several settlements in the Houten and Oss region (Wes-



Wijk-bij-Duurstede-De Horden

0

200 m

Oss-Westerveld

0

100 m

Hoogeloon-Kerkakkers

0

100 m

Fig. 2. The reconstructed development of the settlements at Wijk bij Duurstede-De Horden, Oss-Westerveld and HoogeloonKerkakkers.

ferentiation can sometimes be found. However, farmsteads still seem to have remained separate entities, but integrated into a larger, common organisational structure created by the enclosure. Further south, in the loess regions, we observe similar changes, often resulting in a somewhat different type of settlement which I call ‘compound settlements’. By ‘compound’ I mean an enclosed terrain where buildings were quite strictly ordered, together forming a coherent entity and generally surrounding a central open space. Unlike the settlements described earlier, which consisted of a number of separate farmsteads, compounds were more or less centrally organised. Again, several development trajectories can be reconstructed. Here, however, the late pre-Roman and earliest Roman period remain especially difficult to grasp in archaeological terms. Two well-documented settlements from the German loess region are Jüchen-Neuholz16 and Pulheim-Brauweiler17 (fig. 3). An enclosed compound with rigidly ordered buildings evolved at Jüchen in a number of phases during the 1st century AD, with a fairly small square compound being replaced by a larger one in the 2nd century. Parallel to the emergence of enclosed compounds, there were interesting changes in house building, which I will describe below. At Pulheim, during the first half of the 1st century AD, settlement consisted of small, loosely clustered post-built structures. Similar buildings dated to the pre-Roman period have also been found at other sites in the German 16

Andrikopoulou-Strack 1999; Frank/Keller 2007.

17

Andrikopoulou-Strack et al. 2000.



Jüchen-Neuholz

0

100 m

Frimmersdorf 129

Pulheim-Brauweiler

Kerkrade-Holzkuil

Fig. 3. Development trajectories of the compound settlements at Jüchen-Neuholz, Frimmersdorf 129, Pulheim-Brauweiler and Kerkrade-Holzkuil.

loess region. Then, at about the middle of the 1st century, an enclosure (or partial enclosure) was created, with traditional buildings situated parallel to it. At about the end of that century, a short distance to the west, a new, square compound was subsequently created, with larger buildings constructed in an organised fashion. We find similar shifts from open, fairly unstructured settlement clusters to well-structured,



Verneuil-en-Halatte

Monchy-le Preux

Roye-Le Puits A Marne 1

0

100 m

Fig. 4. Development of the axially organised settlements at Verneuil-en-Halatte, Monchy-le Preux and Roye-Le Puits à Marne I. At the latter, the grey lines represent the early phases (late 1st century BC-1st century AD) and the black lines represent the phases from the 2nd century onwards.

enclosed compound settlements, often with monumentalised or semi-monumentalised buildings, at some other sites, such as Hambach 412,18 Hambach 512,19 Frimmersdorf 129,20 and possibly Voerendaal-Ten Hove21 and Seclin.22 Although such early phases were not documented for many compound settlements, their existence cannot be ruled out. Potential locational shifts in habitation, combined with the non18

Archäologie im Rheinland 2007, 69-71.

19

Kaszab-Olschewski 2006.

20

Lochner 2010.



21

Willems/Kooistra 1986, 1987, 1988; Willems 1995; Kooistra 1996.

22

Révillion et al. 1994.

monumental nature of both architecture and material culture, may have meant that these phases were missed by archaeologists. In yet another trajectory, settlements evolved into even more rigidly organised complexes. These are the axially organised settlements, also known as the Anthée type (fig. 4).23 Such complexes are well known from the aerial surveys carried out by Roger Agache in Picardy,24 but in more recent years quite a number have been excavated in detail, shedding more light on their development. At Monchy-lePreux,25 for example, a typical pre-Roman enclosed settlement (consisting of a curvilinear enclosure with an inner residential compound and an outer compound with a different, probably economic, function) was replaced by a new, fairly rectangular enclosure, probably during the early Augustan period. Both the orientation of the enclosure and the position of the entrance are similar to the preceding enclosure. A house on stone foundations, overlooking two rows of secondary buildings, was not constructed until the second half of the 1st century AD but an axial layout may already have been present earlier on. The same can perhaps be said of the settlements of Roye-Le Puits à Marne I26 and Plailly-La Butte Grise.27 At Roye, a rectangular enclosure was created during the earliest Roman period, although the only preserved buildings are those with stone foundations, dated to the 2nd century. And at Plailly an existing compound was extended at around the middle of the 1st century AD, resulting in a new bipartite spatial structure. Somewhat later, space was further reorganised and segmented by ditches. It seems likely that the settlement already had an axial organisation in this phase. However, a monumental house was not built until the 2nd century, on the eastern compound, with two rows of buildings projecting westwards on both sides of a central axis. Unlike the examples described above, the pre-monumental phases have also been documented at some other axially organised settlements. At Verneuil-en-Halatte,28 the first phase of the axial complex could be dated to the Augustan period. Houses, including the most prominently positioned house at the top of the complex, were of the traditional post-built type during this phase. The settlement was monumentalised in the Flavian period. The main residential compound was separated from the working compound by a wall and gate. Indications of early axial complexes were also found at Juvincourt-et-Damary-Le Gué de Mauchamps29 and Famechon-Le Marais.30 At the latter site, the first traces of buildings positioned within an axial spatial structure could be dated to the earliest Augustan period, or even somewhat earlier. These were secondary buildings, situated on the working compound of the complex. A 12 m long building on silex and chalkstone foundations was constructed on the residential compound of this axially organised complex in the Augustan-Tiberian period. Four buildings on chalkstone foundations were built on the working compound in this same period. At Juvincourt, a large, probably long-rectangular complex was laid out during the earliest Roman period, measuring at least 210 by 192 m. Buildings were arranged in accordance with an axial layout along the enclosures on both sides of a central axis. These buildings were of a type indigenous to the area and known from the La Tène period. Despite the considerable size of this early settlement, it did not evolve into a monumental complex. It is remarkable that axially organised settlements were typical of the southern parts of the research region as no such complexes are known in the northern sand and clay areas, Flanders and the Dutch loess region. And in the German loess region, an axially organised complex has been documented only in the southernmost part of the defined region, at Blankenheim-Hülchrath.

23

See Roymans/Habermehl, this volume.

28

Collart 1996, 124 ff.

24

Agache 1978.

29

Carte Archéologique de la Gaule 02, 265-267; Collart 1996,

25

Gricourt/Jacques 2007.

26

Collart 1996; Collart personal communication 2008.

27

Carte Archéologique de la Gaule 60, 375; Gallia Informations

144-146. 30

Collart 1996, 146-149.

1989, 233-235.



3 reorga n isi n g settleme n t s p ace , reorga n isi n g relatio n shi p s The above observations suggest some general trends in the reorganisation of settlement space. It is also important to attempt to understand these developments from a social perspective, focusing on people as active and creative agents in a process of redefining relationships. The first trend concerns the enclosure of settlements, a process that was already underway in the southern parts of the research region during the later Iron Age, but in the more northerly parts generally only from the 1st century AD. Enclosures defined a settlement community, simultaneously including and excluding people, by clearly marking community boundaries. Viewed in this way, enclosures had both a physical function and a symbolic significance. The literature describes the creation of enclosures as reflecting the ‘isolation of the local social group from the wider scale community’.31 From an anthropological perspective, similar interpretations have been made by Thomas, who associated enclosure development with social contraction linked to agricultural intensification and increasing pressure on land. Social groups started connecting themselves with the land in more concrete ways, striving to preserve it among the local group.32 A firmer sense of place developed, establishing a lasting association between the family, the house and the land on which the settlement was located.33 Another important process is the structuring of space. This entailed both the increasingly rigid organisation of settlement space according to an apparent design and at the same time the creation of a socio-spatial hierarchy. Compounds often seem to have been laid out as planned entities at a single moment in time, although this does not mean that they remained unchanged over the years. Buildings were constructed parallel to the often rectangular enclosures, and were sometimes even arranged in axes. Furthermore, certain houses were located in prominent positions within the structure of the settlement, sometimes physically dominating the other houses. At some larger compound settlements, such as Kerkrade-Holzkuil and Hamois-Le Hody, the main house inhabited by the dominant family was situated in a central, prominent position, with its back against the enclosure ditch. It is remarkable that this spatial structure already existed before the main house was monumentalised. With the reorganisation of settlement space, new organisational concepts of symmetry and axiality were also introduced. In architectural studies, axiality is generally associated with authoritarian power, the ‘axis’ often leading to the symbol of power. This was certainly the case in the axial complexes from our region, where the axis led straight up to the centre of local power: the residence of the dominant family who exercised power over people and production. But axiality and symmetry were not solely confined to the large and very rigidly organised axial complexes. At the settlements of Hambach 59, Hambach 127 and Bruges-Refuge, for example, the main residence was located at one end of the settlement compound, with the secondary buildings organised in front on both sides of the compound to create a corridor leading up to the residence. Settlements like Kerkrade-Holzkuil and Hamois-Le Hody had a fairly symmetrical layout, with the main house occupying a central position. Sometimes, movement and experience were guided even more directly. At the axial settlements in particular, the main house was often separated from the rest of the settlement by a ditch or even a wall, thereby further increasing and monumentalising the distance between the main residence, the centre of power, and the other dwellings, where dependent families lived. One can imagine how the large axial complexes guided movement: the rows of secondary buildings formed a sort of corridor leading to the far end of the complex where a large monumental house rose up, with a wall and gate protecting it from free entry. But movement was clearly guided at less grand settlements too. With the settlement entrance located opposite the main house, people would immediately be confronted with

31

Hingley 1984, 25; see also Bowden/McOmish 1987.

32

Thomas 1997.



33

On changing relationships between settlement/house,  land and people, see Gerritsen 2003, 2007.

the monumentality and dominance of that house within the settlement. To direct movement still further, paths were constructed, leading from the entrance straight up to the main residence. Such paths have been documented at for example Hambach 127, Hambach 488 and Voerendaal-Ten Hove. Apart from these general developments, however, there is a marked differentiation in the settlements that developed throughout the research region. Argued from a social point of view, we could say that the asymmetrical social relationships within the axially organised settlements were most complex, most formal and most impersonal. The dominant family created for itself a place that was spatially – and often monumentally – separate from the other families. They exercised direct spatial and social control, looking out over the other families from their monumental residence. The explicit use of axiality can be regarded as indicative of authoritarian power, strictly ordering and controlling space and the axis leading up to the centre of power. Dependent families lived in this rigid spatial structure that breathed inequality and direct control. In some cases, buildings interpreted as secondary houses were highly uniform, barrack-like structures. In these instances, we can suspect that the families living there had no role in the construction of their house, rendering them even more dependent. They may have had the social status of slaves.34 The situation was somewhat different for the category of compound settlements. There, all buildings were built on the same compound around a shared open space. The distance between the residences of families was relatively limited, although the house of the dominant family often occupied a prominent position. We can probably suspect closer, more personal social ties between the families living there. They may have represented different generations of the same family. The character of the asymmetrical or dependency relationships within such settlements remains subject to debate, however. The situation was different again in the multi-farmstead settlements of the north. There, farmsteads seem to have remained relatively separate and independent within the settlements, although in several cases separated compounds of special character could be documented. It seems, however, that while these families became more prominent, they did not directly control the other families in the way that the settlements described above might suggest. Control over space and over people seems less strong, less direct here. In conclusion, the adoption of new organisational concepts was clearly not a simple adoption of Roman ideas on organisation but a conscious and active strategy to create and fix new social relationships within the context of the settlement. By controlling access, movement and experience and by creating spatial hierarchies as metaphors of dependency and domination, new social relationships could be constructed and fixed within the settlement. These were asymmetrical relationships, in which the degree of dependency and the character of the relationships, as reflected in the spatial structure of the settlements, seem to have varied considerably across the research region. Control is a central concept in this context. Indeed, a leader’s status was indicated by his control over others, with architecture an effective way of shaping control within the physical context of day-to-day life.35

4 de v elo p me n ts i n house b uildi n g Having focused on the broader settlement, I will now look more closely at developments in house building. Traditionally, it is this theme that is most explicitly associated with the villa. Again, I will first explore development trajectories in house building throughout the research region and then discuss some avenues for understanding these changes from a social perspective. A number of house development trajectories are evident in the research region. The first category comprises houses that remained fundamentally traditional in form and technique, while non-traditional 34

See Roymans/Derks, this volume, Section 9; Roymans/

35

See also Wilson 1988, 126.

Zandstra, this volume.



Oss-Westerveld

0

Den Haag-Wateringseveld

20 m

Fig. 5. Traditional Alphen-Ekeren-type byre house and portico-house at Oss-Westerveld; traditional house with long fronting portico at Den Haag-Wateringseveld.

elements were added to them. These elements were ‘translated’ to fit the traditional ways of building. We could label this type of house the ‘romanised traditional house’. Examples are especially well documented for the northern sand and clay regions. At Oss-Westerveld, a traditional Alphen-Ekeren type house was surrounded by a porchway of posts, forming a kind of portico around it (fig. 5). The presence of ceramic tiles might also suggest that this portico was covered with tiles. Fairly similar portico houses, with a wooden portico added to a traditional house, were also documented at quite a number of other sites in this region.36 While in most cases the ‘portico’ surrounds the house on all sides, at Hoogeloon and Den Haag-Wateringseveld a sort of post-built hallway fronts the house, exceeding its width considerably (see fig. 5). The introduction of the portico in traditional rural house building has in recent years been associated with military rather than civil architecture.37 As cultural mediators, veterans probably brought new architectural ideas inspired by Roman military architecture to their native settlements and used these in their houses. The second category involves a more profound change involving the adoption of new building techniques and often new materials. At the same time, however, these houses remained relatively small and simple wooden hall-like structures. I prefer to call these buildings ‘timber framework houses’. The rectangular plans of these houses are generally characterised by four walls of relatively large, often square, postholes or the packings and footings that supported these wall posts. Central roof-supporting posts were absent in these houses. Packings and footings represented new foundation techniques. The former generally consisted of gravel, fragments of natural stone, tile and even plaster and pottery, together forming a tight packing to support a timber post. Footings were hewn stone blocks that contained holes for both vertical and horizontal timbers. The elevation of these houses can probably be reconstructed as timber framework structures. The walls consisted of horizontal and vertical timbers forming a sturdy framework and the roof was supported by the walls and a crossbeam spanning between them. Both the sturdier foundations and the new construction techniques for wall and roof will have enabled the roofs to be covered with ceramic tiles. It also seems plausible that the walls were plastered with white chalk. In combination with the red tiled roofs, such houses will already have had a Mediterranean-style appearance. Another new element was the creation of a more or less monumental façade. At Druten, for example, a long façade, similar to the ones described above at Hoogeloon and Den Haag-Wateringseveld, fronted the framework building. In the case of some German framework houses, a single risalith was added to the rectangular building. The emergence of timber framework houses has been documented at settlements like Druten, Pulheim and Jüchen (fig. 6). Post-built houses were replaced by generally larger, 36

See Vos 2009, 238-239, for an overview.

37

Vos 2009, 237 ff., esp. 243; Heeren 2009, 153 ff., esp. 157.



Druten-Klepperheide

Jüchen-Neuholz

Pulheim-Brauweiler

0

20 m

Fig. 6. Reconstructed house development trajectories at Druten-Klepperheide, Jüchen-Neuholz and Pulheim-Brauweiler. In all three cases, framework houses are part of the development trajectory.

sturdier and more monumental framework houses in the second half of the 1st century or first half of the 2nd century AD. 38 The third category comprises the development of multi-roomed houses on stone foundations. It is these buildings that have traditionally been most explicitly referred to as villas.They represent the clearest break with tradition since new techniques, new forms, a range of new materials and new architectural and organisational-spatial concepts were adopted for their construction. It is also in this category that we find a very pronounced heterogeneity, with regard to both form and development. In some cases, a post-built construction was first replaced by a building on stone foundations that had more or less the same spatial structure and dimensions as its predecessor (for example at KerkradeHolzkuil, Cologne-Braunsfeld, Rijswijk-De Bult, Broichweiden-Würselen and possibly also Hamois-Le Hody). Generally, these initial buildings were later extended or replaced by a larger building on stone foundations. At Kerkrade-Holzkuil, for example, a gradual evolution can be reconstructed, whereby a traditional Alphen-Ekeren type house was first replaced by a simple building on stone foundations during the early 2nd century AD. Around the middle of that century, a larger, multi-roomed house with a fronting portico was constructed at the same location. This house was then extended with a risalith and a bath section (see fig. 7). 38

For a more detailed discussion and description, see my thesis: Habermehl 2011.



Rijswijk-De Bult

Hoogeloon-Kerkakkers

Kerkrade-Holzkuil

Kerkrade-Spekholzerheide

Maasbracht

0

20 m

Fig. 7. Reconstructed house development trajectories of Rijswijk-De Bult, Hoogeloon-Kerkakkers, Kerkrade-Holzkuil, Kerkrade-Spekholzerheide and Maasbracht.

Other houses were built as multi-roomed houses from the start, often replacing documented traditional houses. Again, over the years, the existing house cores were extended with rooms, hypocausts and baths, and facades were enlarged or monumentalised. Clear examples of such monumentalisation can be found at Broekom, Maasbracht and Kerkrade-Spekholzerheide (fig. 7). Sometimes, like at Haccourt, Eschweiler-Laurenzberg and Voerendaal-Ten Hove, existing multiroomed houses were torn down so that considerably larger ones could be built. At the latter site, a relatively simple, multi-roomed house was demolished to create a highly monumental, coherent complex.



5 b uildi n g n e w houses , b uildi n g n e w relatio n shi p s For a better understanding of villa development, we need to think about the backgrounds and implications of the house building changes described above. How should these be understood from a social perspective? The first important aspect is the break with tradition. Traditional houses can be regarded as vernacular architecture.They were built using local materials, local forms and techniques (passed on from generation to generation through practical teaching), and a local workforce. In other words, traditional house building was embedded in the local community. The pressure of social conformity and homogeneity with regard to house construction, among other things, meant that the social order was preserved.39 Deviations from standard architecture were not accepted, the house being a supreme symbol of community membership and of continuity between past and present.40 With the construction of new types of houses, in particular the timber framework houses and multi-roomed houses on stone foundations, new materials were imported from outside the local community, new, non-local forms and techniques were adopted and a specialised workforce was brought in from outside. This implies a significant break with tradition and, to a certain degree, with existing relationships within the local community. Another aspect is the monumentalisation of the house. On the one hand, the increasing physical robustness of the house made it a durable structure with a historical dimension. These houses were passed down from the ancestors, thereby becoming ‘lieux de mémoire’, monuments that symbolised the continuity of the community living there and perhaps even represented the broader settlement community. They created a tangible link between the past and the present. On the other hand, houses evolved into more prestigious structures, communicating towards the outside world in more elaborate, visible and conscious ways, simply by being imposing and noticeable. This was achieved in part by the white plastered walls and red tiled roofs that most monumental houses had; these will have been visible in the landscape even from a great distance. It is also interesting to return to changes in the façade, discussed previously. Unlike traditional post-built houses, houses on stone foundations commonly boasted façades that literally acted as the ‘face’ of the house. The rise of these façades can be linked to the changing purpose and significance of the house. Houses became a factor in the creation and maintenance of social relationships, in social competition, and thus acquired a character that focused considerably more on communication with the outside world. The construction of multi-roomed houses also involved a significant reorganisation of domestic space, parallel to the reorganisation of settlement space discussed above. Traditional houses included fairly small, multi-purpose living areas, generally lacking physical barriers that subdivided the space into functionally or socially differentiated areas. By building houses with several separate rooms, domestic and productive activities were increasingly spatially segregated.41 Access and exclusion were more explicitly and monumentally controlled, as the physical and social distance and distinctions between people increased. Multi-roomed houses were the locus for the symbolism of social inequality, domestic space becoming the cultural language of domination.42 A more complex socio-spatial structure was facilitated by means of fixed (walls, hypocausts), semi-fixed ((mosaic) floors and painted wall plaster) and non-fixed elements (objects, furniture).43 In this way, new behavioural patterns, and thus new social realities, were created and fixed within the household.44 Not only did the distance between the members of the household increase, so too did the distance between the household residing inside the house and the outside world. A number

39

Oliver 1997, 121.

42

Hingley 2005, 88.

40

Wilk 1990, 38; Rapoport 1989, XVIII.

43

Rapoport 1994, 460-462.

41

Taylor 2001, 50-52.

44

Hingley 2005, 88-89.



0

20 m

Tiel-Passewaaij

Rijswijk-De Bult

Hambach 512

Verneuil-en-Halatte

Fig. 8. Selection of storage building development trajectories (Tiel-Passewaaij, Rijswijk-De Bult, Hambach 512 and Verneuilen-Halatte).

of authors have argued that the creation of porticoes fronting the house reflected the desire to create a distance between the private home and the public outside world.45 The line of reasoning presented here implies that we should not simply regard the size and appearance of these houses as a reflection of the degree of wealth of their inhabitants. Not only did the luxury of rural houses vary greatly (the presence or absence of hypocausts, bathing sections, triclinia and mosaic floors), but also the degree of complexity in which social relationships were constructed within the household. 45

On such barriers between the house and street, see Robben 1989, 582 ff.



Cologne-Müngersdorf

Hamois-Le Hody

Champion-Le Emptinne

Hambach 512 Bruges-Refuge

0

100 m

Fig. 9. The spatial association between the storage building (black) and the most prominent house (red) within the settlement.

In short, developments in house building should not be regarded as a passive adoption of Mediterranean architectural forms and building practices, but rather as an active social strategy to create new symbols of power and continuity in a changing world. By breaking with traditions, a new social position could be defined and by creating durable, highly visible and prestigious houses, new social relationships within local communities as well as between these communities and the wider world could be constructed, fixed and communicated.

6 de v elo p me n ts i n storage When studying changing social relationships, we should also explicitly focus on the organisation of production. The most interesting buildings to shed light on changing relationships in production are storage buildings or granaries, as they were directly related to production and surplus. The development trajectories of storage buildings can be reconstructed for several settlements in the research region. During the pre- and Early Roman period, granaries were small post-built structures, generally consisting of four or six posts (spiekers in Dutch or Speicher in German; see fig. 8). In the northern



sand and clay regions, we find that new types of larger post-built storage buildings were built during the second half of the 1st or 2nd century. In several cases, these granaries consisted of a core of fairly tightly set posts, surrounded by a kind of porch (fig. 8). At Rijswijk-De Bult, a large storage building was built on parallel, gravel-filled ditches (fig. 8). These structures had a considerably larger capacity than their traditional counterparts (65-80 m2 compared with 5-15 m2 for many traditional granaries). In the more southerly regions, granaries often evolved into buildings on stone foundations, sometimes even partitioned into several rooms. At Hambach 512, for example, a fairly simple post-built granary, set in foundation ditches, was replaced by a rectangular building on stone foundations at precisely the same location. Similar types of building have been documented at quite a number of settlements (e.g. Hamois-Le Hody and Hambach 132). At Verneuil-en-Halatte, granaries were small and post-built during the earliest, Augustan-Tiberian settlement phases. Then, in the Claudian period, a considerably larger post-built storage building was constructed in the working compound. Subsequently, during the Flavian period, a long-rectangular building on stone foundations was constructed at the same location (fig. 8). At Voerendaal-Ten Hove, a large granary on stone foundations was integrated into the highly monumental façade, created in the 2nd century AD. Storage buildings were important elements within rural settlements as they contained the immediate means of existence – the surplus produced. This importance was communicated more clearly in the creation of more conspicuous storage buildings. Another interesting aspect, however, is the association between these storage buildings and the main house. A spatial association between the main monumental house and the now large, monumental storage building can be documented at quite a number of settlements (fig. 9). By associating themselves with that building, the family residing in the main house demonstrated their control over surplus and accentuated the dependent position of the other families. The crops these latter families produced were stored in the large granaries, under the control of the dominant family. The construction of larger and more monumental granaries, as well as their association with the monumental main house, reflects changing relationships of production and control. In fact, it could be suggested that storage buildings were actively deployed in the process of creating new, increasingly asymmetrical relationships within rural communities.

7 discussio n In this paper, I have focused on developments in the immediate living environment (the settlement and house) of the people residing in the countryside during the period around the start of our era. We have seen that significant changes took place throughout the research region both in the way that settlement space was organised and that houses were built. I have argued that by actively changing spatial structures and ways of building, relationships were redefined – within households, within local communities and between these local communities and the broader outside world. Control is a central concept – control over the organisation of space, over movement, experience and, ultimately, over surplus and people. Social relationships thus clearly became more asymmetrical. The nature of these developing asymmetrical relationships seems to have varied considerably, however (as reflected in the highly heterogeneous collection of houses). Generally speaking, the developments discussed here can be related to broader changes within the context of the evolving Roman provinces. One of the most important developments was that of a formal and institutionalised political, administrative system. This much more socio-politically complex system started to emerge in towns, but eventually affected the whole of society as a result of integration into this system.The increase in socio-political complexity in rural settlements can be linked to this broader development. Land was organised and administered in new ways, often changing existing patterns.The existing bases of power and proprietorship will have changed considerably under the Roman system. For some,



this will have provided new opportunities, for others change was less positive as they probably became increasingly dependent. With regard to the development of a more complex market system, demand increased considerably (stimulating production and production specialisation), economic networks widened and money was introduced as a means of economic exchange. Many local communities now moved out of their local economic networks and became more or less integrated into the larger ones, considerably changing their production, income, local economic relationships, attitude towards commodities and exchange and their access to non-local goods. The development of the urban world was also highly important, not only as a political and economic centre, but also as a cultural centre, as a place where a new, urban lifestyle emerged. It seems plausible that these urban centres played an important role in the development of the rural settlements. New building styles, baths and ways of dining evolved in the town and were later introduced in rural settlements, also taking their place in a newly emerging rural lifestyle, or villa lifestyle. People who had fairly direct access to the urban world were probably the first to introduce elements of urban lifestyle in the countryside. People who were less involved, or indirectly involved, in the urban world followed later. Connecting to new, urban lifestyles was probably a strategy to construct a new social position within communities, to differentiate oneself from others and to connect to arenas of power and decision-making, primarily based in towns. In conclusion, I would like to state that villa development, as approached in this paper, involved an active way of constructing, fixing and communicating new social, economic and cultural relationships in a changing world, among other things by reorganising and restructuring settlement and house space, by creating new symbolic systems and by breaking with traditions. Existing structures were actively and creatively manipulated on the level of the rural settlement, within the context of the changing world of the developing Roman provinces. Within this changing world with its far greater socio-political and economic complexity, new social identities as well as new asymmetrical relationships were created, communicated and fixed in the built environment and through the adoption of new lifestyles.

re f ere n ces Agache, R., 1978: La Somme pré-romaine et romaine d’après les prospections aériennes à basse altitude, Amiens. Andrikopoulou-Strack, J.-N. et al., 1999: Eine frührömische Siedlung in Jüchen-Neuholz. Überlegungen zur Siedlungskontinuität in der Lößbörde, Bonner Jahrbücher 199, 141-180. Andrikopoulou-Strack, J.-N. et al., 2000: Der frührömische und kaiserzeitliche Siedlungsplatz in Pulheim-Brauweiler, Bonner Jahrbücher 200, 409-488. Bayard, D./J.-L. Collart (eds), 1996: De la ferme indigène à la villa romaine. La romanisation des campagnes de la Gaule, Amiens (Revue Archéologique de Picardie, special 11). Bloemers, J.H.F., 1978: Rijswijk (Z-H) ‘De Bult’. Eine Siedlung der Cananefaten, Amersfoort (Nederlandse Oudheden 8). Bloemers, J.H.F., 1980: Rijswijk (Z.H.) ‘De Bult’. Een nederzetting van de Cananefaten, Hermeneus 52, 95-106. Bowden, M.C.B./D. McOmish, 1987: The required barrier, Scottish Archaeological Review 4, 76-84. Braat,W.C., 1934: Nieuwe opgravingen van Romeinsche villae, Oudheidkundige Mededelingen uit het Rijksmuseum van Oudheden te Leiden 15, 4-38. Braat, W.C., 1941: Nieuwe opgravingen van Romeinsche villa’s in Limburg, Oudheidkundige Mededelingen uit het Rijksmuseum van Oudheden te Leiden 22, 39-51. Collart, J.-L., 1996 : La naissance de la villa en Picardie. La ferme gallo-romaine précoce, in D. Bayard/J.-L. Collart (eds), De la ferme indigène à la villa romaine, Amiens (Revue Archéologique de Picardie, special 11), 121-156.



De Clercq, W., 2003: L’habitat gallo-romain en Flandre orientale (Belgique). Recherches 1990-2001 dans les civitates Menapiorum et Nerviorum, Revue du Nord 85- 353, 161-179. De Clercq, W., 2009: Lokale gemeenschappen in het Imperium Romanum.Transformaties in rurale bewoningsstructuur en materiële cultuur in de landschappen van het noordelijk deel van de civitas Menapiorum (provincie GalliaBelgica, ca. 100 v. Chr. – 400 n. Chr.), Gent (unpublished PhD-thesis University of Gent). Es, W.A. van, 1972: De Romeinen in Nederland, Bussum. Frank, K./C. Keller, 2007: Jüchen-Neuholz.Vom eisenzeitlichen Gehöft zur Villa rustica, in M. Hegewisch (ed.) Krieg und Frieden. Kelten, Römer, Germanen, Bonn/Darmstadt, 316-324. Gaitzsch, W., 1986: Grundformen römischer Landsiedlungen im Westen der CCAA, Bonner Jahrbücher 186, 397-427. Gerritsen, F., 2003: Local Identities. Landscape and community in the late prehistoric Meuse-Demer-Scheldt region, Amsterdam (Amsterdam Archaeological Studies 9). Gerritsen, F.A., 2007: Relocating the house. Social transformations in late prehistoric Northern Europe, in R.A. Beck (ed.), The durable house. House society models in archaeology, Carbondale, 154-174. Gricourt, D./A. Jacques, 2007: Le mobilier de la villa gallo-romaine de Monchy-le-Preux (Pas-de-Calais), Revue du Nord 89 (nr. 373), 173-196. Habermehl, D.S., 2011: Settling in a changing world. Villa development in the northern provinces of the Roman empire, Amsterdam (PhD-thesis VU University Amsterdam). Haselgrove, C., 2007: The age of enclosure. Later Iron Age settlement and society in northern France, in C. Haselgrove/T. Moore (eds), The later Iron Age in Britain and beyond, Oxford, 492-522. Heeren, S., 2006: Opgravingen bij Tiel-Passewaaij 1. De nederzetting aan de Passewaaijse Hogeweg, Amsterdam (Zuidnederlandse Archeologische Rapporten 29). Heeren, S. 2009: Romanisering van rurale gemeenschappen in de civitas Batavorum. De casus Tiel-Passewaaij, Amersfoort (Nederlandse Archeologische Rapporten 36). Heimberg, U., 2002-2003: Römische Villen am Rhein und Maas, Bonner Jahrbücher 202/203, 57-148. Hingley, R., 1984: The archaeology of settlement and the social significance of space, Scottish Archaeological Review 3, 22-26. Hingley, R., 1989: Rural settlement in Roman Britain, London. Hingley, R., 2005: Globalizing Roman culture. Unity, diversity and empire, Abingdon/New York. Hulst, R.S., 1978: Druten-Klepperhei. Vorbericht der Ausgrabungen einer römischen Villa, Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 28, 133-51. Jeneson, C.F., 2004. Terug naar Hoogeloon. Een nieuwe kijk op de Romeinse nederzetting rond de villa op de Kerkakkers, Amsterdam (unpublished MA-thesis VU University Amsterdam). Kaszab-Olschewski, T., 2006: Siedlungsgenese im Bereich des Hambacher Forstes 1.-4. Jh. N. Chr. - Hambach 512 und Hambach 516, Oxford (British Archaeological Reports, International Series 1585). Kooistra, L.I., 1996: Borderland farming. Possibilities and limitations of farming in the Roman period and Early Middle Ages between the Rhine and Meuse, Assen. Kunow, J., 1994: Die ländliche Besiedlung im südlichen Teil von Niedergermanien, in H. Bender/H. Wolff (eds), Ländliche Besiedlung und Landwirtschaft in den Rhein-Donau-Provinzen des römischen Reiches, Espelkamp, 141-197. Lenz, K.H., 1998: Villae rusticae. Zur Entstehung dieser Siedlungsform in den Nordwestprovinzen des römischen Reiches, Kölner Jahrbuch für Vor- und Frühgeschichte 31, 49-70. Lenz, K.H., 1999: Siedlungen der römische Kaiserzeit auf der Aldenhovener Platte, Frankfurt. Lochner, I., 2010: Eine Tallandschaft im Wandel. Folgen römerzeitlicher Wasserwirtschaft im Elsbachtal, in J. Kunow (ed.), Braunkohlenarchäologie im Rheinland. Entwicklung von Kultur, Umwelt und Landschaft, Weilerswist (Materialien zur Bodendenkmalpflege im Rheinland 21), 167-169. Maas, J.C., 2007: Druten-Klepperheide revisited. Een inheems-Romeinse nederzetting in de civitas Batavorum, Amsterdam (unpublished MA-thesis, VU University Amsterdam).



Maeyer, R. de, 1937: De Romeinsche villa’s in België. Een archeologische studie, Antwerpen. Maeyer, R. de, 1940: De overblijfselen der Romeinsche villa’s in België, Antwerpen. Martins, C.B., 2005: Becoming consumers. Looking beyond wealth as an explanation of villa variability. Perspectives from the east of England, Oxford (British Archaeological Reports, British Series 403). Oliver, P. (ed.), 1997: Encyclopedia of vernacular architecture of the world, Cambridge. Renswoude, J., van/J.Van Kerckhove, 2009: Opgravingen in Geldermalsen-Hondsgemet. Een inheemse nederzetting uit de Late IJzertijd en Romeinse tijd, Amsterdam (Zuidnederlandse Archeologische Rapporten 35). Rapoport, A., 1989: Foreword, in S.M. Low/E. Chambers, Housing, culture, and design. A comparative perspective, Philadelphia, xiii. Rapoport, A., 1994: Spatial organization and the built environment, in T. Ingold (ed.), Companion encyclopedia of anthropology. Humanity, culture and Ssocial life, London, 460–502. Révillion, S./K. Bouche/L. Wozny, 1994: La partie agricole d’une grande exploitation rurale d’époque romaine. Le gisement des ‘Hauts de Clauwiers’, Seclin (Nord), Revue du Nord. Archéologie 76, 99-146. Robben, A.C.G.M., 1989: Habits of the home. Spatial hegemony and the structuration of house and society in Brazil, American Anthropologist 91(3), 570-588. Roymans, N./F. Theuws (eds), 1991: Images of the past. Studies on ancient societies in Northwestern Europe, Amsterdam (Studies in pre- en protohistorie 7). Roymans, N., 1996: The sword or the plough. Regional dynamics in the romanisation of Belgic Gaul and the Rhineland area, in N. Roymans (ed.) From the sword to the plough, Amsterdam (Amsterdam Archaeological Studies 1), 9-126. Siemons, H./J.J. Lanzing (eds), 2009: Bewoningssporen uit de Romeinse tijd in het Wateringse Veld, Den Haag. Slofstra, J., 1987: Een nederzetting uit de Romeinse tijd bij Hoogeloon, in W.C.M. van Nuenen (ed.) Drie dorpen een gemeente. Een bijdrage aan de geschiedenis van Hoogeloon, Hapert en Casteren, Hapert, 51-86. Slofstra, J., 1991: Changing settlement systems in the Meuse-Demer-Scheldt area during the Early Roman period, in N. Roymans/F. Theuws (eds), Images of the past. Studies on ancient societies in northwestern Europe, Amsterdam (Studies in pre- en protohistorie 7), 131-199. Taylor, J., 2001: Rural society in Roman Britain, in S. James/M. Millett (ed.) Britons and Romans: advancing an archaeological agenda, Walmgate (The Council for British Archaeology, research report 125), 46-59. Theuws, F./N. Roymans (eds), 1999: Land and Ancestors, Amsterdam (Amsterdam Archaeological Studies 4). Thomas, R., 1997: Land, kinship relations and the rise of enclosed settlement in first millenium B.C. Britain, Oxford Journal of Archaeology 16 (2), 211–218. Tichelman, G., 2005: Het villacomplex Kerkrade-Holzkuil, Amersfoort (Archeologisch Diensten Centrum, Rapport 155). Tichelman, G., 2010: IJzertijd bewoning en begraving op het löss-plateau bij Beek. Opgraving Maastricht-Aachen Airport, gemeente Beek, Amsterdam (RAAP-rapport 2054). Tichelman, G., in prep.: Onderzoek te Heerlen-Trilandis, Amsterdam (RAAP-rapport). Vos, W.K., 2009: Bataafs platteland: Het Romeinse nederzettingslandschap in het Nederlandse Kromme-Rijngebied, Amersfoort (Nederlandse Archeologische Rapporten 34). Wesselingh, D.A., 2000: Native neighbours. Local settlement system and social structure in the Roman period at Oss (The Netherlands), Leiden (Analecta Praehistorica Leidensia 32). Wilk, R.R., 1990: The built environment and consumer decisions, in S. Kent (ed.), Domestic architecture and the use of space, Cambridge, 34-42. Willems, W.J.H., 1981: Romans and Batavians. A regional study in the Dutch eastern river area I, Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 31, 7-217. Willems, W.J.H., 1986: De Romeinse villa te Voerendaal. Opgraving 1985, Archeologie in Limburg 28, 143150. Willems, W.J.H./L.I. Kooistra 1987: De Romeinse villa te Voerendaal. Opgraving 1986, Archeologie in Limburg 32, 29-38.



Willems, W.J.H./L.I. Kooistra 1988: De Romeinse villa te Voerendaal. Opgraving 1987, Archeologie in Limburg 37, 137-147. Wilson, P. J., 1988: The domestication of the human species, New Haven. Woolf, G., 1998: Becoming Roman. The origins of provincial civilization in Gaul, Cambridge.



On the origin and development of axial villas with double courtyards in the Latin West Nico Roymans / Diederick Habermehl 1 Introduction 2 Spatial layout and geographical distribution 3 Social interpretations 4 The question of the pre-Roman roots 5 The urban link: urbs in rure? 6 The Roman villa as a social house 7 Conclusion and discussion Abbreviations References Appendix 1

1 i n troductio n One of the key material manifestations of the integration of the Gallic and Germanic provinces in the Roman empire was without doubt the introduction of Roman-style towns and – inextricably linked to this – the wide-spread appearance of Roman villas1 on the countryside from the second half of the 1st century AD onwards. There is a long tradition of research into the origin, development and social interpretation of villas. Villas by no means represent a homogeneous category, and we are able to distinguish different types on the basis of their architecture and spatial layout.2 When looking for the origins of this kind of settlement, archaeologists for a long time sought parallels among Italic villas, assuming them to be the models for villas in the north. In the past two decades, however, research has convincingly shown that the northern villa types were not simple replicas of Roman-Mediterranean models. Instead, they displayed a considerable individuality by incorporating local elements into their layout and architecture.This finding ties in with the general view that ‘becoming Roman’ was by no means a standard process, but involved significant variation in the ways in which individuals and groups interpreted a Roman identity in a provincial setting.3 Everywhere, the process of appropriating a Roman identity led to the creation of new cultural forms, which in part recalled Mediterranean traditions but which always incorporated local characteristics.4 We find examples of these hybrid forms not only in Gallo-Roman villa architecture, but also in temple construction, funerary monuments and the naming of deities. 1 

We proceed here from a conventional archaeological definition of a villa: an agrarian settlement containing a Roman-style dwelling, constructed partly of stone and

2 

On the categorisation of villa settlements in the Gallic and Germanic provinces, see Agache 1978; Lenz 1998; Heimberg 2002-2003; Roymans/Derks, this volume, Section 1.

with a rectangular ground plan, solid floors, a hypocaust

3 

and a set of baths.

4 

Mattingly 2004. Woolf 1998; Hingley 2005.



Borg

Levet

Anthée

Oberentfelden

0

50

100 m

Fig. 1. Ground plan of the axial villas at Borg (D), Levet (F), Anthée (B), and Oberentfelden (CH). After Birkenhagen, this volume, fig. 2; Holmgren/Leday 1981, fig. 6; Swoboda 1918, 103; Ebnöther/Monnier 2002, fig. 139.6. Green: residential compound. Brown: working compound.

These latter observations prompt new questions for villa research in the north: how should we visualise the formation process of the northern villa types? Which social groups were the key transcultural mediators in this process? And what motivated the owners to build these villas? The aim of this paper is to contribute to this discussion by analysing one specific subgroup of villas. These are the axial villas of the Anthée type (fig. 1), characterised by a double courtyard, a strong association with an elite and a relatively early starting date.5 We will address the following topics in this paper:

5 



These features, together with their wealth, set them apart

and several dispersed subsidiary buildings within a com-

from the much larger group of small and medium-sized

mon enclosure. See Lenz 1998 on the categorisation of

villas, which were single farmsteads with a main building

villa settlements in the northern provinces.

a. the spatial layout and geographical distribution of double-courtyard villas b. the social interpretation of these villas c. the continuity between double-courtyard villas and pre-Roman rural settlements d. the link with urban elite houses e. the analysis of the Roman villa as a social house. To what extent does the appearance of villas imply fundamental changes in the relationship between the house as a physical entity and the ideas and values of its inhabitants?

2 s p atial lay out a n d geogra p hical distri b utio n Anthée-type villas are a well-defined group within the total spectrum of villa settlements (fig. 1). They feature a long-rectangular ground plan with a double courtyard, one of which is an elite compound containing a residential building, and the other a service compound with a symmetrical arrangement of ancillary structures along each of the long sides. The main access path runs across the central axis of the service compound to the main building, thereby creating an axial arrangement with the emphasis on the broad façade of the main building. The total surface area of such complexes, enclosed by a wall, ranges from approx. 5 to 12 hectares. The spatial layout forms a cohesive whole, which suggests that the villas were conceived and planned as a unity. Particularly for this type of villa archaeologists have long used the twin terms pars urbana and pars rustica. Taken from the writings of Columella, they refer to the residential and agrarian parts of villa complexes respectively.6 We will also employ this terminology, although it is important to realise that Columella did not have this type of villa in mind when he used these terms.7 What do we know about the arrangement of the residential compound? The principal building is often a complex, long-rectangular building with many rooms, including a reception area, living areas and bathrooms.8 The larger villas are of the U-shaped type with wings projecting at right angles from the core building. Others even have a courtyard entirely enclosed by a porticus (fig. 1-d) and thus belong to the category of peristyle houses.9 Each principal building displays a measure of unicity, the result of a long process of rebuilding and expansion to make it more sumptuous. There was usually an ornamental garden in front of the main building, sometimes containing a pond or reservoir (fig. 6). A stone wall often marked the boundary with the service compound, with a gatehouse on the central axis (fig. 1). The main feature of the working area are two symmetrical rows of auxiliary buildings along each of the long sides. There has been little research as yet into their function. However, in some instances it has been argued convincingly that the smaller buildings, too, functioned as dwellings, although craft activities were sometimes carried out there as well.10 The working area may contain up to 20 subsidiary buildings. Assuming that most had a domestic function and were inhabited by labourers with their families, these were settlements with a substantial population, akin to small villages. Therefore, unlike smaller villas, Anthée-type villas were a kind of nucleated settlement, almost always with over 50 inhabitants.

6 

Columella, De re rustica 1.6.1. This Latin terminology is favoured by German and French archaeologists. The German archaeologist Franz Oelmann (1928) seems to have been the first to apply these terms to villas from the northern provinces, like the one at Anthée.

contains known examples; see Agache 1978. 9 

These peristyle houses are known from the villas at Levet (F), and the Swiss sites of Orbe and Yvonand and the French site of Saint-Georges-sur-Arnon. See the appendix.

10 

Cf. Schucany 2006 on the villa at Biberist (Switzerland).

7 

The presence of hearths and kilns in auxiliary buildings

8 

Some principal buildings of villas belonging to vast

at Dietikon suggests that they had a domestic function

complexes are conspicuously simple. Picardy in particular

for craftsmen and farmers. See Ebnöther 1995, 68 ff.

Heimberg 2002-2003, 87.



Cologne

Mainz

Reims

Lyon

0

250 km

Fig. 2. Distribution of double-courtyard villas in the Gallic and Germanic provinces. See also Appendix 1.

The above demonstrates that while Anthée-type villas shared a set of key features, there was also some variation. In part, this was regionally determined. Several villas in Switzerland and central France have peristyle houses, whereas in Picardy there are also complexes featuring a fairly simple principal building. Also in Picardy a group of villas have a trapezoidal layout. The distribution map of Anthée-type villas shows that they are largely confined to the Gallic and Germanic provinces (fig. 2), corresponding to parts of present-day France (mainly Picardy, but also Centre, Cher and Burgundy), Germany (the Trier area), Switzerland and Belgium. They are conspicuously absent in Italy and Narbonensis, as well as the Danube provinces, Spain, and western France (Bretagne, Normandy). Just how representative is the current distribution picture? We see a dense cluster in Picardy and looser concentrations in central France, the Trier area and Switzerland. The clusters in France can clearly be linked to regional traditions of aerial photographical research, which revealed many large villa complexes.11 Excavation intensity would appear to explain the densities in the Trier area and Switzerland.

11

Cf. Agache 1978 for Picardy and Holmgren/Leday 1981 for central France.



Given the dense distribution in several well-researched regions, we may expect a greater density of this villa type in other parts of the present distribution area.12

3 social i n ter p retatio n s Given their size and richness compared with other types of villa, we can generally assume that doublecourtyard villas were the residences of the upper social echelon of Gallo-Roman societies. Evidence to support this are the imposing facades, the large number of rooms, the frequent use of mosaic floors and marble encrustation in the principal building, and above all the many ancillary houses in the working compounds. Also significant is the symbolism of the use of space, with axiality and symmetry as the ordering principles to express power differences and hierarchy. The villa’s spatial order and architecture emphasised the status of the owner and the social inferiority of dependent labourers, and actively contributed to the creation of this social order. Detailed regional surveys of rural landscapes in Picardy and the Swiss Aare valley show that this villa category makes up only a small proportion (approx. 5%) of the total number of rural settlements.13 Nevertheless, some were located close together, occasionally even within view of one another. Woolf has convincingly argued that these villas should be seen as a product of a process of competitive building by a community of peers profiling itself through its ostentatious lifestyle.14 Elites copied and tried to surpass one another’s residences. Emulative competition through investments in public and private buildings was firmly anchored in Roman elite culture, which certainly helps explain common features of this type of villa across larger areas. Although villa ownership relations are barely tangible in archaeological terms, a few general observations can be made. According to Woolf, the dense distribution of villas in Gaul argues for the existence of a broad class of landowners rather than a small group of super magnates.15 The existence of a large tenant group appears unlikely as they would not have invested their wealth in the monumentalisation of villas. Instead, we should think in terms of a hierarchy among villa owners, with double-courtyard villas belonging to the upper echelon within the hierarchy. Frequently, their owners would have belonged to the ruling elite in a civitas, and would accordingly have had a seat on the urban-based council of decurions. This decurial elite, often comprising over one hundred families in each civitas, would generally have owned an urban residence as well.16 It is therefore important to bear in mind that many owners of Anthée-type villas had a house in town and travelled between town and country.17 Family tradition and the owner’s personal preference determined which house was viewed as the principal dwelling. A possible archaeological clue to the periodic absence of owners is the presence of a second ‘mini-villa’, probably the home of a steward, in some of the large villa complexes. An intriguing question is the social status of the working population inhabiting the working compounds of axial villas.The spatial layout of the villa complexes and their internal differences in wealth reveals a wide

12 

There are many villas for which only the main building

urban and rural residences also seems to apply to the nor-

has been traced, while almost nothing is known about

thern provinces. Key arguments are a text from Ausonius

the general layout of the settlement.

(De Herediolo, 29-32; see also section 6 below), epigraphic

13 

evidence from rural sites of individuals holding urban

14 

administrative functions, as well as gladiator scenes on

15 

mosaics or frescos from the villas at Orbe (CH), Nennig

16 

(G) and Maasbracht (NL). Cf. the discussion in Derks,

17 

this volume.

Woolf 1998, 164; Schucany 1999, esp. figs. 4 and 5. Woolf 1998, 156-157. Woolf 1998, 163. See the discussion in Derks, this volume. The Italian model of an elite ‘commuting’ between the



social distance between the inhabitants of the residential and working compounds. The secondary houses were built to a set preconceived plan and show a high degree of uniformity. It is important to note that there were no wells, granaries or sheds around the secondary houses, which means that the occupants had no opportunities for autonomous enterprise and that their labour power was fully and directly exploited by the villa owner. Although it is virtually impossible to obtain archaeological certainty on this point, we should seriously consider the presence of slaves. If we were dealing with tenants, we would expect signs of more autonomous activity. Only in the case of the villa of Liestal-Munzach (CH) do we have additional clues to the presence of slaves in the form of a funerary inscription and a set of leg irons.18 It is almost impossible to gauge the size of the villa estates, especially as they were not static entities. This problem is highlighted by Swiss micro-regional research in the eastern Aare valley around the vici of Solothurn and Olme. There, large villas are fairly evenly spaced at 5-10 km intervals on both sides of the river valley, with smaller villas and unidentified (non-villa?) sites located between the larger villas or somewhat higher up the valley slopes.19 It is difficult to decide, however, whether the smaller villas were controlled by the owners of the large ones, or should be regarded as autonomous split-offs of estates.

4 the q uestio n o f the p re - roma n roots There is broad consensus in the current archaeological literature that the spatial organisation of doublecourtyard villas is of pre-Roman origin.Although these settlements were subject to powerful romanisation during the 1st and 2nd centuries, this would simply have involved the adoption of building techniques, building materials and architectural elements. Lenz refers to the further development of a ‘spätkeltische Bautradition’, and Heimberg views Anthée-type villas as an ‘eigenständige gallische Entwicklung, in der sich noch von Stammesordnungen abhängige Familien- und Gemeinschaftsformen spiegeln mögen.’ In Carroll’s view, these villas reflect ‘Celtic social structures’.20 Two arguments are critical to this point of view. The first is the conclusion that Gallic and Rhineland villa complexes were clearly not copies of villas from Italy or Narbonensis but represented instead a regional phenomenon.21 The second concerns the findings of French studies of fermes indigènes, or enclosed farmsteads. We can be brief about the first argument. The fact that we are dealing with a regional villa design does not necessarily mean that the spatial concept was a pre-Roman one. To investigate this, we need to focus on the second argument, the relationship between double-courtyard villas and pre-Roman enclosed farmsteads. In recent decades, valuable research has been carried out into enclosed farmsteads in northern France. These have been shown to be characteristic of the Late Iron Age and the transition to the Roman period. The ditched enclosures vary considerably in shape: curvilinear, rectangular, or trapezoidal.22 There was a marked trend in the Late Iron Age towards more rectilinear enclosures; there is also increasing evidence of a hierarchical differentiation among rural sites. A later development, in the Augustan period, was the appearance of large enclosed farmsteads with a new spatial organisation, as documented at Bazoches-surVesle, Beaurieux and Verneuil-en-Halatte (fig. 3). The rectangular enclosures contained two rows of timber buildings on opposite sides of a large open space and backing onto the enclosure boundary, a layout

18 

courtyard. Lenz (1998, 55 ff.) refers to Kompaktanlagen.

19 

For recent surveys of Italic villas, see De Franceschini

See the discussion in Roymans/Zandstra, this volume. Schucany 1999; idem, this volume.

20

Lenz 1998, 64; Heimberg 2002/3, 87; Carroll 2001, 76. 

21 

Lenz 1998; Heimberg 2002/3. Italy has a wide range of



2005 and Marzano 2007. 22 

On the typology and chronology of Late Iron Age

villas, which nevertheless share a compact layout, with

enclosures in northern France, see the recent surveys by

spaces for staff and working areas lining an enclosed

Haselgrove 1996, 153; idem 2007, 504 ff.

Fig. 3. Plan of ‘protovillas’ of the Augustan/Tiberian period from northern France. After Ben Redjeb/Duvette/Quérel 2005, fig. 12.

that seems to anticipate the Roman double-courtyard villas in this region.23 Dwelling units at one of the short ends of the enclosures at Verneuil-en-Halatte and Beaurieux may have been part of higher status compounds, although there is no concrete evidence to support this. Traces of older wooden structures have also been found at the villas at Dietikon and Neftenbach in Switzerland. These go back to the Tiberian period, and were probably part of the same kind of rectangular enclosed farmsteads as those known to us from northern France.24 In the villa at Borg near Trier (fig. 1) only the residential compound has been excavated and we have no information as yet about the development of the settlement as a whole. The earliest habitation in the pars urbana began in the second half of the 1st century BC and continued into the Tiberian period. Characteristic features are post-built wooden structures in native style whose orientations differ from that of the later villa.25 23 

Ben Redjeb/Duvette/Quérel 2005, 192 ff; Haselgrove 1996, 156; idem 2007, 506.

25 

Frey 2000; idem 2001 and 2007; Birkenhagen, this volume.

24 

Ebnöther/Monnier 2002, fig, 137, 138.



The long-rectangular basic structure of these enclosed settlements from the Augustan/Tiberian period generally appears to have been retained in the later villas. There are also fundamental differences, however. Augustan-period sites lack a strictly symmetrical layout, with no clear separation between an elite and a working compound. Nor is there evidence of social differentiation between the houses. It seems that house construction had not yet entered into the arena of social competition during this phase. We should also realise that the earliest examples of large enclosed farmsteads or ‘protovillas’ of the Beaurieux type did not appear until the Augustan period and at present are unknown from the Late Iron Age. We observe an acceleration in the process of ‘romanisation’ of Anthée-type villas from the Tiberian/ Claudian period onward. The symmetrical, axial layout characteristic of this type, with a pars urbana and a pars rustica, probably existed from this time onwards. Romanised principal buildings made their first appearance in the Tiberian era. During the Tiberian/Claudian period, a house was built in the elite compound of the villa at Verneuil-en-Halatte, of which only the large cellar remains. The timber-framed house at Borg, dating to about 20-30 AD and erected on stone foundations, boasts a porticus and painted plasterwork, thereby reflecting the adoption of new architectural forms and styles.26 The same applies to the buildings in framework technique at Dietikon and Neftenbach, dating to the Tiberian period and featuring a porticus and painted plasterwork.27 All these wooden buildings stood on the site of the later principal buildings of the villa complex, which were erected in stone in the second half of the 1st century,28 a dating that we can also assume for most other villa sites of this type. The picture that emerges from the evidence presented here is one of a gradual, step-by-step evolution of certain native farmsteads into Anthée-type villas. It was characterised by an increasing taste for Roman building materials, techniques and styles, as well as for architectural principles and ideas about the use of space. The most attractive new element was undoubtedly the incorporation of a pars urbana. The northern double-courtyard villa represents a regional villa design that had its antecedents partly in indigenous social organisation and conceptions of space. However, we should not overestimate the continuity of pre-Roman ideas.29 The spatial organisation of the villas was not a simple continuation of a pre-Roman Gallic form, but represents an essentially new hybrid, a product of the highly competitive, ostentatious elite culture of the 1st century AD, which sought to interpret a Roman identity and lifestyle in its own way.

5 the ur b a n li n k : ur b s i n rure ? With regard to the social interpretation of double-courtyard villas (section 3), we have argued that their owners often belonged to both the rural and urban communities since they owned properties in both places. It is therefore logical to investigate possible influences on these villas from urban elite dwellings. Woolf asserts that villas in the north bore little resemblance to urban homes in terms of spatial and architectural design and that no relationship therefore existed between them.30 We wish to examine this matter further. Little is known at present about urban houses in the northern provinces because most Roman towns were built over in later periods. Added to that, it is almost impossible to study traces of 1st-century AD wooden houses under the remains of stone buildings. Nevertheless, the data at hand does allow us to draw several conclusions. 26 

lo-romaine précoce vers la villa classique, même si cette

27 

dernière conserve les grandes lignes de l’organisation

Frey 2000, 2001 and 2007. Ebnöther/Monnier 2002, 141; Ebnöther 1995; Rychener 1999.

28 

Frey 2001, 44-46; Ebnöther 1995; Rychener 1999.

29 

Collart (1996, 154) also emphasises this: ‘Il semble qu’une véritable rupture caractérise le passage de la ferme gal-



spatiale des établissements antérieurs.’ 30 

Woolf 1998, 157.

Fig. 4. Ground plan of urban courtyard houses at Amiens and Avenches (insula 13), and the villa at Dietikon. Green: garden. Blue: water basin. Brown: portico. After Binet 2007, fig. 3; Carroll 2002, fig. 20; Ebnöther 1995, fig. 21.

Broadly speaking, we can state that axiality and symmetry were well-known ordering principles of Roman architecture and that they were used to express hierarchies and differences in status.We are familiar with these principles from urban fora and from several large rural sanctuaries, like the one at Ribemont-sur-Ancre.31 In the earliest, Augustan/Tiberian phase, we still encounter abundant traces of wooden houses in native vernacular style in the civitas capitals of Belgic Gaul and the Germanic provinces (e.g. at Tongres, Amiens, Arras and Metz).32 From the Claudian period onward, domusstyle houses began to appear, initially in framework technique with horizontal foundation beams, and also in stone in the later 1st century AD.33 Examples of 1stcentury domus are known from Amiens-Palais des Sports (Neronian), Tongres-Hondsstraat (end of 1st century) and Kielenstraat (mid-1st century), Cologne (mid-1st and end of 1st century onwards), Augst and Avenches.34 Although these 1st-century urban homes vary considerably, they are all of the courtyard type. We find long-rectangular houses with projecting wings and probably a garden at the front (Tongres-Hondsstraat and Kielenstraat, AmiensPalais des Sports, Avenches-insula 13; see figs. 4 and 5), as well as peristyle houses with a central courtyard enclosed by a porticus (Avenches, Augst, Cologne). Both variants are also encountered in the elite compounds of Anthée-type villas (fig. 1). If we now compare the elite compounds of double-courtyard villas with 1st-century AD urban courtyard houses, it becomes clear that it was not only individual elements of urban houses (ornamental gardens with a reservoir, porticus, hypocaust, a set of baths, and tiled roofs) that were incorporated into the villas. Elite compounds seem to have translated the full layout of urban courtyard houses to the new rural setting. One example is the parallels between the villas at Borg and Dietikon and the domus of insula 13 at Avenches (figs. 4 and 6). An innovation, however, was to link the pars urbana to a working compound within a long-axial, symmetrical layout.The elite compounds of the villas, with their characteristic walled courtyard, therefore show the clear influence of urban residences.

31 

hoeven 1996. Cologne: Hellenkemper 1975; Eck 2004,

32 

See the contributions of several authors in Hanoune

378ff. Augst: Carroll 2001, 55. Avenches: Carroll 2001,

2007.

55. Recent excavations at Waldgirmes show that we

Agache 1978, 404 ff.; Derks 1998, 189.

33 

should also take into consideration Roman-style urban

34 

Amiens-Palais des Sports: Binet 2007. Tongres-Hond-

houses in the late Augustan period. See Becker et al.

straat: Vanderhoeven 2007. Tongres-Kielenstraat: Vander-

2003.

Brulet 1996, 95.



Fig. 5. Reconstruction of a 1st-century AD urban domus from the town of Tongres. After Brulet/Coquelet 2004, fig. 73.

It is generally accepted that the Gallo-Roman landowning elite invested its wealth first and foremost in the monumentalisation of sanctuaries and public buildings in the town. Only then did they progress, as further disposable income came to hand, to competitive investment in private, Roman-style houses, initially in town (from the later Augustan period onwards) and a little later (from the Tiberian/Claudian period onwards) in the countryside as well.35 As in Italy, cultural reasons – the dream of an idyllic country life – may partly account for elites beginning to invest in rural residences. In Italy, literary works contrasted the tranquillity of rural life with the clamour and stench of the town.36 Social competition was just as important a motivation, however, with villas becoming a focal point in the competitive and emulative lifestyle of elites. A much-neglected topic within villa research is the role of urban architects and builders in the design and spread of the larger villas in particular. They were professional builders and architects, who would have been based in the towns. They contributed to a certain standardisation of villa layout and architecture, as well as to a marginalisation of native vernacular traditions of house building. The enormous surge in villa construction in the countryside must have brought considerable wealth to these builders. Woolf ’s assertion that no relationship existed between villas and urban dwellings in the north seems difficult to uphold for the particular type of villa under discussion.37 The urban domus may have served as a model for the spatial layout of the elite compounds, which were not simple copies, but creative interpre-

35 

Woolf 1998, 126; Haselgrove 1996, 176-177. Millett (1995, 72) describes a similar development for Roman Britain.



36 

Ausonius, De Herediolo 29-32, is the only literary expression of this theme in Gaul. See the next section.

37 

Woolf 1998, 157.

Fig. 6. View of the reconstructed pars urbana of the villa at Borg. Photo Archaeological Park Borg.

tations of urban dwellings. Making the most of greater spatial opportunities in the countryside, the elites experimented with new architectural elements and effects. This resulted in broader villa façades, enlarged gardens, and the creation of views. Their primary purpose, however, was still to create little islands of urban culture in the countryside by building residences marked as urban through their architecture, gardens and modern comforts. The idea of urbs in rure seems to apply not only to villas in Italy, but also to the Gallic double-courtyard villas, thereby justifying the use of the term pars urbana for these residential compounds.38

6 the roma n v illa as a social house The focus of this section is the changing symbolism of the house and social organisation of its occupants following the appearance of larger villas in the countryside. Relevant here is the discussion within cultural anthropology about the concepts of the ‘social house’ and ‘house society’, starting with the work of Lévi-Strauss.39 These terms refer to the phenomenon of the mutual identification of the house and its inhabitants, and focus on the relationship between the physical manifestation of the house on the one hand, and residents, ancestors and house ideology on the other. The house as a physical construct reflects something of the social relationships and ideas of its inhabitants.40 38 

On garden architecture in urban and villa houses, and the notion of urbs in rure, see Purcell 1987.

39 

40 

It is important in this context to note the complex meanings of the term domus in Roman literary sources

Lévi-Strauss (1982) based his concept of the ‘house soci-

(Saller 1994, 80 ff.). The term was used to mean: 1. the

ety’ primarily on the study of noble houses in medieval

physical house, 2. the household community including

France. Cf. Carsten/Hugh-Jones 1995, and also Gerrit-

family and slaves, 3. the broader kinship group and 4. the

sen 2007, 157.

patrimony.



1 3

2

4

8

5

9

0

6

5m

7

10

11

Fig. 7. Late Iron Age houses and two early villa buildings (2, 3) from northern Gaul. 1 Frey 2007, fig.180; 2 Rychener 1999; 3 Frey 2000, fig. 4; 4 Laurelut/Tegel/Vanmoerkerke 2004, fig. 2; 5 Collart 1996, fig. 14; 6 Collart 1996, fig. 14; 7 Haselgrove 1996, fig. 13; 8 Laurelut/Tegel/Vanmoerkerke 2004, fig. 2; 9 Collart 1996, fig. 14; 10 Collart 1996, fig. 14; 11 Haselgrove 1996, fig. 13.

Many anthropological insights regarding the concepts of ‘social house’ and ‘house society’ are difficult to operationalize in archaeology. There are, however, three key potential features of social houses that invite archaeological research.41 These are 1. the identification of people with the house they inhabit and vice versa; 2. the social house and the idea of continuity; and 3. the social house as a unit that retains and transmits material and immaterial property. Moreover, archaeologists are able to add a time dimension that sheds light on changes in the spatio-temporal sequences of houses and house identities. Proceeding from the notion of the social house, we will attempt to make a comparative analysis of Late Iron Age houses and Roman villas in northern France and the Rhineland, highlighting significant differences. Late Iron Age houses, often situated inside enclosed farmsteads, were fairly small, rectangular wooden structures (fig. 7).42 They were usually single-phase houses that existed barely longer than one human generation, which means that most adults survived the house in which they were born.This fact seriously limits the potential role of the physical house as a symbol of transgenerational family continuity. It is almost impossible to identify elite houses on the basis of size or specific architectural features. Haselgrove also points out the highly unstable nature of Late Iron Age rural settlements in northern France.43 Many 41 

Gerritsen 2007, 158.

42 

Laurelut/Tegel/Vanmoerkerke 2004; Lambot/Méniel 1992. Gerritsen (2007) presents a model for Iron Age houses in the southern Netherlands linking the bio-



graphy of the house to the biological life cycle of its inhabitants. 43 

Haselgrove 1996, 164 ff; idem 2007, 508 idem, this volume.

Fig. 8. Reconstruction of the villa at Newel near Trier, with a nearby sanctuary and a group of grave monuments within a separate burial compound. After Wigg 1993, fig. 26.

settlements did not continue without interruption into the Roman period or functioned for only a short period. The 1st century BC in particular was a period of instability. Haselgrove suggests a link between the abandonment of rural settlements and the establishment of large nucleated settlements, which would have involved farmstead relocation. The picture changes after the Roman conquest in response to the new conditions created by the Roman empire. The high degree of stability in the Gallo-Roman settlement pattern contrasts with the marked instability in that of the Late Iron Age.44 It is against this backdrop that we should understand the appearance of villas. As a result of Roman influences, the house became an object for the competitive display of status and wealth.45 This latter element was virtually absent in the Late Iron Age, given the pattern of fairly undifferentiated houses from this period. Villa construction from the mid-1st century AD implied fundamental changes in the traditions and mentalities underlying the villa as a social house. These were: a. an increasing hierarchisation and diversification of household communities. Villas, certainly those of the Anthée type, were occupied by larger, hierarchically structured groups of more than 50 people, which also included people of lower rank (tenants, freedmen, slaves). b. an increasing emphasis on the ideal of permanence and continuity in house construction. The practice of continuously aggrandizing the residential building, whilst maintaining existing orientations and connecting the old and the new, points to a powerful strategy to maintain links with previous generations and the buildings they resided in. The durable house, stretching beyond individual human life cycles, allowed occupants to construct narratives of a permanent social group with a fixed place in time and space. The villa as a whole, but especially the principal building, forged a link between past, present and future generations of inhabitants.Villas were therefore potent symbols of family continuity and vitality.The physical house could accordingly function as a monument of personal and familial commemoration.46 c. the emphasis on durability in villa houses as a strategy to present the members of the social house as an enduring property-holding group. Underpinning this was the ideal image of the villa as the inalienable property of a particular family. The proprietor family arranged for the transmission of material and immaterial property to the next generation within the new frameworks of Roman law.

44 

Haselgrove 1996, 167.

46 

Bodel 1997.

45 

See the discussion in Roymans/Derks, this volume.



This last point presupposes a notion of villa transfer within the context of Roman law. Although villas and the accompanying estates could be sold as commodities to strangers, it is generally assumed that no significant commercial real estate market existed during the Roman empire. Transmissions of landed property to younger generations usually occurred by way of inheritance and marriages, mechanisms which would largely have been driven by economic and political strategies.47 Roman law offered the paterfamilias considerable flexibility in managing the family inheritance and the transmission of landed property to the next generation. ‘Roman inheritance strategies were based on pragmatic decisions aimed at a one-generational transmission of property.’ 48 This focus on the immediate future is best illustrated by the fact that the principles of succession under Roman were not based on primogeniture, but granted equal shares to all children, whether sons or daughters. This pragmatic focus on the immediate future should not suggest, however, that membership of a multi-generational family was a marginal component of Roman aristocratic identity. A testator could even use legal means to ensure that his descendants would own a certain house in perpetuity. This must have been because the domus was regarded as the symbol of the family’s lineage and consequently as a display of the family’s past and current power, and its political ambitions.49 The paterfamilias was the bearer of the ideal of continuity of family property. However, for wealthy aristocratic families that owned several urban and rural houses, not all houses were equally important for transgenerational continuity. One house was regarded as the most important and was most intimately bound up with family identity. It was here that the owner installed his household deities, representing the fertility, wellbeing and hence the continuity of the family.50 Household deities were not inextricably linked to a particular house, however; a subsequent owner might choose to transfer them to another house. These latter observations are based on historical sources from Italy, frequently involving members of the senatorial aristocracy. This of course raises the question of how representative they are of relations in the northern provinces. To what extent does historical and archaeological data support ideas about the villa as a social house and associations with family identity for the north? There is a lack of literary evidence from the early imperial era for Gaul, but a 4th-century poem by Ausonius is instructive. Ausonius, retiring after a long official career in Trier to his father’s villa near Bordeaux,51 is very explicit about the significance that this paternal patrimony had for him: Hail, little patrimony, the realm of my forebears, which my great-grandfather, which my grandfather, which my father tended so carefully, which the last-named left to me when he died all too soon, albeit in a ripe old age. (Ausonius, De Herediolo, 1-4) Ausonius had a house in both the town and the country, but for him, the villa was the symbol of family identity and ancestral tradition. Ausonius’ feelings towards his villa and the close bond with the town closely match the values of the Italic elite culture as expressed in the letters of Pliny the Younger.52 More important for the northern provinces, however, is the archaeological evidence. We can point to the frequent occurrence in rural contexts of remains of elite funerary monuments in the form of mausoleums, tower tombs, or large tumuli.53 Although architectural parts of tombs are often encountered in secondary contexts, there is enough evidence to assume that such monumental graves belonged to villas (fig. 8). Because wealthy villa owners usually owned a house in town, they could choose to be buried in a formal cemetery on the outskirts of town or at their residence in the country. Often they will have 47 

52 

48 

53 

Saller 1994, 151, 161 ff. Hillner 2003, 130. See also Saller 1994, 161 ff.

Pliny the Younger, Epistulae 2.17, 5.6.

Cf. for northern Gaul and the German Rhineland the

49 

regional syntheses by Wigg 1993, Willer 2005 and Mas-

50 

sart 1994. See also Krier/Henrich, this volume.

Saller 1994, 91, 169-170. See also Hillner 2003, 134. Hillner 2003, 135; Saller 1994, 81.

51 

Ausonius, De Herediolo III.1.



Fig. 9. Stone sarcophagus of the lady of Simpelveld with reliefs depicting the woman, her villa and household effects. Photo RMO Leiden.

set out their wishes in a will, as is evident from funerary inscriptions.54 Whenever a grave monument was erected at a villa, this was clearly an important statement symbolising the deceased’s affinity with, and the familial claim to, that villa. Monumental tombs at villas suggest an ideal picture of the villa as an inalienable inheritance linked to family identity. In large villa complexes, this was a matter of consciously creating lieux de mémoires for future generations.55 We find a unique, iconographic expression of the close, intimate bond between villa-owner and villa in the reliefs on the 3rd century AD sarcophagus of the lady of Simpelveld (NL). 56 Reclining on her couch, the lady is looking at depictions of her villa and its household effects in the form of cupboards, furniture and vessels (fig. 9). Located on the inside of the sarcophagus, the reliefs were expressly intended for the deceased. The villa and household goods are shown here to be intimately linked to the lady’s personal identity, a link that continued in the hereafter. We cannot of course view this development of the villa as a social house separately from broader social and cultural changes in the Early Roman period. At the forefront was the emergence of new social arenas for competition, linked to a power base centred mainly on landed property and protected by Roman law. At the same time, we see the disappearance of internal warfare as an arena for acquiring wealth and prestige. With the new political and administrative structure of the Roman empire, Rome had created two key conditions for investing in villa construction: security and stability of landownership. Investing in villas became a new strategy for acquiring status and prestige, and at the same time for shaping a noble family identity. This was the foundation stone for the enormous surge in villa construction from the later 1st century AD.

54 

See the discussion in Roymans/Derks, this volume, Section 4.

1997, 20 ff., on graves near villas. 56 

Galestin 2001.

55 

This ties in with a similar practice in Italy. Cf. Bodel



7 co n clusio n a n d discussio n So what conclusions can we draw about the origin, development and social interpretation of axial villas with a double courtyard? This was a smallish category of villas characteristic of the northwestern provinces, and Gaul in particular. We can attribute the villas to a social elite that generally owned a house in town as well, making them members of both the rural and urban communities. Certain aspects of the spatial layout (a long-rectangular enclosure with houses lining each of the long sides around a central open space) go back to the youngest generation of native farmsteads (fèrmes indigènes). This was reason enough for many researchers to posit a pre-Roman, Celtic origin for these villas. Based on our analyses, we believe that the idea of a pre-Roman, Gallic origin of this villa type should be qualified in the following crucially important ways. Firstly, the above-mentioned native predecessors or ‘proto-villas’ date only from the Augustan/Tiberian period; as yet there are no known examples from the Late Iron Age. Secondly, the farmsteads referred to were subject to fundamental Roman influences from the Tiberian/Claudian era onward, especially through the inclusion of Roman-style elite houses as well as the fixed, symmetrical layout of the villa, with a subdivision into a pars urbana and a pars rustica. This reflects a powerful influence from urban courtyard houses rather than from Italic villas. Thirdly, the replacement of vernacular wooden houses by Roman-style elite dwellings suggests a recasting of the relationship between the physical house and the ideas and values of its occupants. Whereas the transience of houses and their association with the human life cycle were a central feature of pre-Roman society, emphasis shifted in the Roman period towards permanence, transgenerationality and commemoration. This was very much in keeping with Roman ideas. As a result of Roman influence, the house became an object for the competitive display of status and wealth and a symbol of family identity. The idea that double-courtyard villas had a pre-Roman origin both demands qualification and underestimates the impact of Roman culture. This cultural influence comprised more than just the adoption of isolated construction and architectural elements. It also included the design and layout of the walled elite compounds and the associated ideas and values. The elite compounds were inspired by urban courtyard houses. The link with urban houses is a plausible one when we accept that the owners of this type of villa often had close urban connections or were even members of the urban community. This highlights the importance of the town as a locus for the development and spread of new cultural forms and lifestyles. It is interesting to note that it was precisely in villas of this type that we can identify Roman architectural influences at quite an early stage. Although Roman styles and forms did not have a massive impact on the countryside until the late 1st century AD, we encounter clear Roman influences – such as stone foundations, porticoes and wall paintings – in the villa complexes at Borg, Dietikon and Neftenbach in the second quarter of the 1st century AD. These settlements, which were to evolve into large monumental villas, may have been the forerunners in the spread of a new rural lifestyle. Axial villas of the Anthée type can best be seen as a new hybrid creation from the 1st century AD, based on a blend of elements from the Gallic fermes indigènes and Roman-Italic architecture. Elites saw this villa type as an attractive ordering principle to express the new social relations that emerged from the time of Augustus. The demilitarisation of interior Gaul and the introduction of a new administrative order created a relatively stable political system, generating the social and economic conditions for largescale investments in villa construction. Although the new social relations were expressed in the villas in accordance with Roman architectural principles (especially axiality, symmetry, vast size and perpendicularity), the villas were not inspired by Italic villas.57 Anthée-type villas sprang from the tension between adopting a new Roman identity and maintaining local identities. Woolf sees in this field of force the essence of Roman culture, which underpins the many hybrid creations.58

57 

Heimberg 2002/2003, 87.



The above analyses allow us to present a social profile of those who built axial villas of the Anthée type. We should envisage an important landowning elite of native origin that had acquired knowledge of Roman values, tastes and cultural practices at an early stage thanks to their civil networks centred on the town.59 This group selected, ‘translated’ and introduced knowledge into rural communities, thus playing a prominent role as transcultural mediators. Other, simpler villa types (which tended to emerge somewhat later) will have involved other social agents, especially smaller landowners. These final observations bring us to the significance of this study for villa research as a whole. To what extent do our conclusions apply to the many small and medium-sized villas with a single courtyard, which lack a strict separation of the residential part and a strict axial layout? The ideas presented here about the changed significance of the villa as a social house also seem relevant for simpler villas. At the same time, however, we should be aware of fundamental differences. For these smaller villas, it is less likely that the owner/occupant also had an urban home, making the urban link less direct.The smaller villas also appear to have been built somewhat later – in the late 1st or early 2nd century AD – than the doublecourtyard villas discussed here. This time lag suggests that the layout and form of the smaller villas might partly have sprung from the creative emulation of elements from the earliest large elite villas in the countryside. Above all, however, they seem to have continued traditions of older native farmsteads, as in the case of the broad group of single-courtyard villas or Streuhofanlagen.60 This is a matter for further research.

a b b re v iatio n s BSR Bilan scientifique de la région Picardie CAG Carte archéologique de la Gaule

re f ere n ces Agache, R., 1978: La Somme pre-romaine et romaine d’après les prospections aériennes à basse altitude, Amiens. Agache, R., 1983: Typologie et devenir des villae antiques dans les grandes plaines de la Gaule septentrionale, in W. Janssen/D. Lohrmann, Villa-curtis-grangia. Landwirtschaft zwischen Loire und Rhein von der Römerzeit zum Hochmittelalter, München, 17-29. Bayard, D./J.-L. Collart (eds), 1996: De la ferme indigène à la villa romaine. La romanisation des campagnes de la Gaule, Amiens. Bayard, D., 1996: La romanisation des campagnes en Picardie à la lumiere des fouilles récentes. Problemes d’échelles et de critères, in D. Bayard/J.-L Collart (eds), De la ferme indigène à la villa romaine. La romanisation des campagnes de la Gaule, Amiens, 157-184. Becker, A. et al., 2003: Die spätaugusteische Stadtgründung in Lahnau-Waldgirmes. Archäologische, architektonische und naturwissenschatliche Untersuchungen, Germania 81, 147-199. Bénard, J. et al., 1994: Les agglomérations antiques de Côte-d’Or, Paris (Annales Littéraires de l’Université de Besancon 522).

58 

Woolf 1998, 19 ff.. A similar phenomenon are the socalled ‘Gallo-Roman temples’. Derks (1998, 168 ff.)

can be interpreted as a new hybrid form. 59 

The proportion of this group that had acquired this know-

concludes these were new regional creations, that were

ledge through military networks (principally as auxilia

neither copies of models from Italy, nor continuations

commanders in the Roman army) seems to have been

of a pre-Roman, ‘Celtic’ type of temple. The earliest examples did not appear until the Augustan period and

quite small. Cf. the discussion in Roymans, this volume. 60 

Lenz 1998.



Ben Redjeb, T./L. Duvette/P. Quérel, 2005: Les campagnes antiques. Bilan et perspectives, Revue Archéologique de Picardie 3/4, 177-222. Binet, E., 2007: Amiens. L’apport de deux opérations préventives (sites du Palais des Sports-Coliseum et ancient Garage Citroën), in R. Hanoune (ed.), Les villes romaines du Nord de la Gaule: vingt ans de recherches nouvelles, Lille (Revue du Nord, hors série), 43-61. Birkenhagen, B., 2004: Die römische Villa Borg. Ein Begleiter durch die Anlage, Merzig. Bodel, J., 1997: Monumental villas and villa monuments, Journal of Roman Archaeology 10, 1997, 5-35. Brulet, R., 1996: La maison urbaine en Gaule Belgique et en Germanie Inferieure, in P. Gros (ed.), La maison urbaine d’époque romaine en Gaule narbonnaise et dans les provinces voisines, Avignon, 73-97. Brulet, R.. (ed.), 2009: Les Romains en Wallonie, Bruxelles. Brulet, R./C. Coquelet, 2004: Les maisons privées, in D. Bayard/N. Mahéo (eds), La marque de Rome. Samarobriva et les villes du nord de la Gaule, Amiens, 122-127. Buisson, A., 1991: Le tombeau du Lingon. Etude du cadre architectural et archéologique, in Y. Le Bohec (ed.), Le testament du Lingon, Lyon (Collection du Centre d’Études Romaines et Gallo-Romaines, 9), 63-72. Carroll, M., 2002: Romans, Celts and Germans. The German Provinces of Rome, Gloucestershire/Charleston. Carsten, J./St. Hugh-Jones, 1995: Introduction, in J. Carsten/St. Hugh-Jones (eds), About the house. LéviStrauss and beyond, Cambridge, 1-46. Chevrot, J.-F./J. Troadec (eds), 1992: Le Cher, Paris (Carte Archaéologique de la Gaule 18) Coulon, G./J. Holmgren (eds), 1992: L’Indre, Paris (Carte Archaéologique de la Gaule 36) Collart, J.-L., 1996: La naissance de la villa en Picardie. La ferme gallo-romaine précoce, in D. Bayard/J.-L Collart (eds), De la ferme indigène à la villa romaine. La romanisation des campagnes de la Gaule, Amiens, 121-156. Corbiau, M.-H. (ed.) 1997: Le patrimoine archéologique de Wallonie, Namur. Courbot-Dewerdt, C., 2006: An alleged Far West? The Romanisation of the countryside in western Gaul, in B. Croxford et al. (eds), Proceedings of the15th annual theoretical Roman archaeology conference, Birmingham 2005, Oxford, 73-82. Cüppers, H., 1975: Römische Villa Otrang, Mainz. Cüppers, H., 1990: Die Römer in Rheinland-Pfalz, Stuttgart. De Boe, G., 1973: De landelijke bewoning in de Romeinse tijd, Het oude land van loon 28, 85-114. De Boe, G., 1975: Haccourt II. Le corps de logis de la grande villa, Archaeologia Belgica 174, 5-48. De Franceschini, M., 2005: Ville dell’agro romano, Roma. Delmaire, R. (ed.), 1994: Le Pas-de-Calais, Paris (Carte Archaéologique de la Gaule 62/2). Deramaix, I., 2006: Meslin-L’Évêque. Imposante villa hainuyère, La Belgique romaine (Dossiers Archéologie et sciences des origines 315), 64-67. Derks,T., 1998: Gods, temples and religious practices: The transformation of religious ideas and values in Roman Gaul, Amsterdam (Amsterdam Archaeological Studies 2). Ebnöther, Chr., 1995: Der römische Gutshof in Dietikon, Zürich. Ebnöther, Chr./J. Monnier, 2002: Ländliche Besiedlung und Landwirtschaft, in Die Schweiz vom Paläolithikum bis zum frühen Mittelalter 5, Römische Zeit, Basel, 135-178. Eck, W., 2004: Geschichte der Stadt Köln. Köln in römischer Zeit, Cologne. Faure-Brac, O. (ed.), 2002: La Haute-Saône, Paris (Carte Archaéologique de la Gaule 70). Ferdière, A., 1988: Les campagnes en Gaule romaine, Paris. Ferdière, A., 2006: Les transformations des campagnes et de l’économie rurale en Gaule romaine, in D. Paunier (ed.), Celtes et Gaulois, l’archéologie face à l’histoire, 5. La romanisation et la question de l’héritage celtique, Glux-en-Glenne, 109-130. Flutsch, L./U. Niffeler/F. Rossi, 2002: Die Schweiz vom Paläolithikum bis zum frühen Mittelalter. (SPM) V. Römische Zeit, Basel. Frey, M., 2000: Die römische Villa von Borg. Ein reiches Landgut mit vorrömischer Tradition, in A. Haffner/S. von Schnurbein (eds), Kelten, Germanen, Römer im Mittelgebirgsraum zwischen Luxemburg und



Thüringen: Akten des Internationalen Kolloquiums zum DFG- Schwerpunktprogramm ‘Romanisierung’, vom 28. bis 30. September 1998 in Trier, Bonn, 41-50. Frey, M., 2001: Die Terra Sigillata der gallorömischen Villenanlage bei Borg, Saarbrücken (Bericht der Staatlichen Denkmalpflege im Saarland, Beiheft 5). Frey, M., 2007: Die vorrömische und die römische Entwicklung der Villa von Borg, in M. Hegewisch (ed.), Krieg und Frieden. Kelten, Römer, Germanen, Darmstadt, 236-8. Frézouls, E., 1977: Circonscription de Champagne-Ardenne, Gallia 35, 389-418. Galestin, M., 2001: The Simpelveld sarcophagus. A unique monument in a provincial Roman context, in T.A.S.M. Panhuysen (ed.), Die Maastrichter Akten des 5. internationalen Kolloquiums über das provinzialrömische Kunstschaffen, Maastricht, 63-76. Gerritsen, F.A., 2007: Relocating the house. Social transformations in late prehistoric Northern Europe, in R.A. Beck (ed.), The durable house. House society models in archaeology, Carbondale, 154-174. Gricourt, D./A. Jacques, 2007: Le mobilier de la villa gallo-romaine de Monchy-le-Preux (Pas-de-Calais), Revue du Nord 89 (no. 373), 173-196. Hanoune, R. (ed.), 2007: Les villes romaines du Nord de la Gaule.Vingt ans de recherches nouvelles, Lille (Revue du Nord, hors série). Haselgrove, C., 1996: Roman impact on rural settlement and society in southern Picardy, in N. Roymans (ed.), From the sword to the plough. Three studies on the earliest romanisation of Northern Gaul, Amsterdam (Amsterdam Archaeological Studies 1), 127-187. Haselgrove, C., 2007: The age of enclosure. Later Iron Age settlement and society in northern France, in C. Haselgrove/T. Moore (eds), The Later Iron Age in Britain and beyond, Oxford, 492-522. Heimberg, U., 2002/2003: Römische Villen an Rhein und Maas, Bonner Jahrbücher 202/3, 57-148. Hellenkemper, H., 1975: Architektur als Beitrag zur Geschichte der Colonia Claudia Ara Agrippinensium, in Aufstieg und Niedergang der römsichen Welt II.4, Berlin, 783-824. Henrich, P./M.Tabaczek/M. Zelle 2005: Die römische Villa von Gerolstein (Kreis Daun) “Auf dem Hofacker” – “Villa Sarabodis”, Trierer Zeitschrift 67/68, 93-140. Henrich, P./C. Mischka, 2006: Die römische Axialvillenanlage am ‘Römerberg’ bei Gillenfeld/Strohn, Kreis Daun, Funde und Ausgrabungen im Bezirk Trier 38, 18-24. Hillner, J., 2003: Domus, family, and inheritance. The senatorial family house in Late Antique Rome, Journal of Roman Studies 93, 129-145. Holmgren, J./A. Leday, 1981: Esquisse d’une typologie des villas gallo-romaines du Berry d’après les prospections aériennes, Gallia 39, 103-122. Horisberger, B., 2006: Der Gutshof in Buchs und die römische Besiedlung im Furtal, Zürich. Krier, J. 2000: Grandes découvertes à Bertrange – ‘Bourmicht’, Musée Info février 2000, 6-9. Krier, J., 2001: Poursuite de la fouille de sauvetage à Bertrange-‘Bourmicht’, Musée Info 14, 42-43. Lambot, B./P. Méniel, 1992: Le site protohistorique d’Acy-Romance (Ardennes), I: L’habitat gaulois, fouilles 1988-1990, Reims. Laurelut, C./W. Tegel/J. Vanmoerkerke, 2004: Nouvelles réflexions concernant les bâtiments à porche de la fin de l’Âge du Fer, Lunula Archeologia protohistorica 12, 137-144. Lenz, K.H., 1998: Villae rusticae. Zur Entstehung dieser Siedlungsform in den Nordwestprovinzen des römischen Reiches, Kölner Jahrbuch 31, 49-70. Lévy-Strauss, Cl., 1982: The way of the masks, translated by S. Modelski, Vancouver. Malrain, F./V. Matterne/P. Méniel, 2002: Les paysans gaulois (IIIe s. – 52 av. J.-C.), Paris (Collection des Hespérides). Marmol, E. del, 1877: Villa d’Anthée, Annales de la Société Archéologique de Namur 14, 165-194. Marmol, E. del 1881: Villa d’Anthée, Annales de la Société Archéologique de Namur 15, 1-40. Massart, C., 1994: Les tumulus gallo-romains conservés en Hesbaye. Étude topographique, Bruxelles.



Mattingly, D., 2004: Being Roman. Expressing identity in a provincial setting, Journal of Roman Archaeology 17, 5-25. Marzano, A., 2007: Roman villas in central Italy. A social and economic history, Leiden. McKay, A.G., 1975: Houses, villas and palaces in the Roman world, New York. Metzler, J. et al., 1981: Ausgrabungen in Echternach, Luxembourg. Mignot, P., 2006: La villa de Jemelle à Rochefort, La Belgique romaine (Dossiers Archéologie et sciences des origines 315), 72-75. Millett, M., 1995: Roman Britain, London. Nuber, H.U., 1995: Die villa urbana von Heitersheim, in S. Palagyi, Forschungen und Ergebnisse. Internationale Tagung über römische Villen. Veszprém16.-20. Mai 1994, Veszprém, 172-178. Oelmann, F., 1916: Die römische Villa bei Blankenheim in der Eifel, Bonner Jahrbücher 123, 210-226. Oelmann, F., 1928: Ein gallorömischer Bauernhof bei Mayen, Bonner Jahrbücher 133, 51-152. Ossel, P. van/S. Defgnee, 2001: Champion, Hamois. Une villa romaine chez les Condruses: archéologie, environment et économique d’une exploitation agricole antique de la Moyenne Belgique, Namur. Paulke, M., 2010: Die römische Axialhofvilla von Diekirch. Neue Grabungen und Forschungen, Empreintes. Annuaire du Musée national d’histoire et d’art 3, 54-67. Petit J.-P./F. Sarateanu-Müller, 2005: La campagne autour de Bliesbruck, in J.-P. Petit, Bliesbruck-Reinheim: Celtes et Gallo-Romains en Moselle et en Sarre, Paris, 169-190. Quérel, P./M. Feugère, 2001: Le recyclage du bronze, Archéologia 374, 42-48. Rothé, M.-P.(ed.), 2001: Le Jura, Paris (Carte Archaéologique de la Gaule 39). Roymans, N., 2009: Becoming Roman in the Rhineland frontier zone. The impact of ethnic recruitment and returning veterans on the romanisation of rural populations, in Kelten am Rhein. Akten des 13. Internationalen Keltologiekongresses I, Ethnizität und Romanisierung, Bonn (Beihefte der Bonner Jahrbücher 58, I), 25-36.  Rychener, J., 1999: Der römische Gutshof in Neftenbach, Zürich/Egg. Sage, P., 1991: Le testament du Lingon. Remarques sur le texte et sur son interpretation, in Y. Le Bohec (ed.), Le testament du Lingon, Lyon (Collection du Centre d’Etudes Romaines et Gallo-Romaines, 9), 17-40. Saller, R.P., 1994: Patriarchy, property and death in the Roman family, Cambridge (Cambridge Studies in Population, Economy and Society in Past Time, 25). Sărăţeanu-Müller, F., 2007: Die Grossvillen der gallischen Oberschicht. Das Beispiel von Reinheim, in J.-P. Petit/S. Santoro, Leben im römischen Europa. Vom Pompeji nach Bliesbruck-Reinheim, Paris, 201-207. Schucany, C., 1999: Solothurn und Olten. Zwei Kleinstädte und ihr Hinterland in römischer Zeit, Archäologie der Schweiz 22, 88-95. Schucany, C., 2006: Die römische Villa von Biberist-Spitalhof / SO (Grabungen 1982, 1983, 1986-1989), Bern. Steiner, P., 1934: Die römische Villa von Nennig an der oberen Mosel (Kreis Saarburg), Trier. Swoboda, K., 1918: Römische und romanische Paläste, Wien. Vanderhoeven, A., 1996:The earliest urbanisation in Northern Gaul. Some implications of recent research in Tongres, in N. Roymans (ed.), From the sword to the plough. Three studies on the earliest romanisation of Northern Gaul, Amsterdam (Amsterdam Archaeological Studies 1), 189-260. Vanderhoeven, A., 2007: Tongres au Haut-Empire romain, in R. Hanoune (ed.), Les villes romaines du Nord de la Gaule. Vingt ans du recherches nouvelles, Lille (Revue du Nord, hors série, no. 10), 309-336. Waterson, R., 1995: Houses and hierarchies in Island Southern Asia, in: J. Carsten/St. Hugh-Jones (eds), About the house. Lévi-Strauss and beyond, Cambridge, 47-68. Wigg, A., 1993: Die Grabhügel des 2. und 3. Jahrhunderts n. Chr. an Mittelrhein, Mosel und Saar, Trier. Willer, S., 2005: Römische Grabbauten des 2. und 3. Jahrhunderts nach Christus im Rheingebiet, Mainz. Woolf, G., 1998: Becoming Roman. The Origins of Provincial Civilization in Gaul, Cambridge. Zeepvat, R.J. 1991: Roman gardens in Britain, in A.E. Brown (ed.), Garden archaeology, London (CBA Research Report 78), 53-59.



a p p e n di x 1 : list o f a x ial v illas w ith dou b le court y ards i n the n orther n p ro v i n ces f ra n ce Amiens-Montières Agache 1978, fig. 20. Arpheuilles-Mazère Coulon/Holmgren 1992, 120-1. Athies Agache 1978, fig. 16. Beauvais-Rue de Witten-’Les Champs Dolents’ BSR 1996, 50-52, BSR 1999, 46, BSR 2002, 64-65 Behen Bayard 1996, 174. Béhencourt-Lahoussoye Agache 1978, fig. 25. Belleuse-Les-Mureaux Agache 1978, fig. 21; Smith 1997, 164. Bertrange-Bourmicht Krier 2000, 2001. Brussey-Camp Paul/Les Craies Faure-Brac 2002, 147-8. Cachy Agache 1978, fig. 32. Callengeville-Mont Cauvet Bayard 1996, 175 Cambron Agache 1978, fig. 23. Cappy Agache 1978, fig. 26. Cavillon-La Marette Agache 1978, fig. 23. Citernes Agache 1978, fig. 19. Clairy-Saulchoix Agache 1978, fig. 23. Colmier-le-Bas-Les Cloisets Frézouls 1977, 413-4. Crouy Agache 1978, fig. 20. Davenescourt Agache 1978, fig. 18. Diors-Lande Coulon/Holmgren 1992, 60-1. Drucat Agache 1978, fig. 21. Dury Quérel/Feugere 2001. Écoust-Saint-Mein Smith 197, 153. Épaux-Bézu - ZID de l’Omois BSR 2004, 28-30. Estrées-sur-Noye Agache 1978, fig. 14. Étalon Agache 1978, fig. 16. Famechon-’Le Marais’ Collart 1996, 147. Ferin Courbot-Dewerdt 2006, fig. 2. Fignières Agache 1978, fig. 23. Fontaine-le-Sec Agache 1978, fig. 26. Frémontiers Agache 1978, fig. 19. Froissy Agache 1978, fig. 27. Grivesnes Agache 1978, fig. 29. Groises-Bordes Chevrot/Troadec 1992, 304. Haccourt De Boe 1973, 1975. Hédauville-Warloy Agache 1978, ph. 186. Hémévillers (Oise) Agache 1983, 17. Juvincourt-et-Damary-Le Gué de Mauchamps Collart 1996, 144-146. Le Mesge Agache 1978, fig. 18. Levet-Champ des Pois Holmgren/Leday 1981, fig. 6. Levet-La Chaume-aux-Couards Holmgren/Leday 1981, fig. 3. Limé-Les Terres Noires-Les Fraiches Terres BSR 1998, 32-34; CAG 02, 285-291. Limeux-Maisons Pions Chevrot/Troadec 1992, 260. Lux-Champ d’Ogne Bénard et al. 1994, fig. 73.



Malpart Agache 1978, fig. 17. Marchelepot Smith 197, 160; Agache 1978, fig. 28. Martainneville Agache 1978, fig. 33. Mercin-et-Vaux-Le Quinconce CAG 02, 303-309. Mézières-en-Santerre Agache 1978, fig. 25. Molay-Corvée Rothé 2001, 498-9. Monchy-Humières Agache 1978, fig. 20. Monchy-Le Preux Gricourt/Jacques 2007. Namps-au-Mont Agache 1978, fig. 25. Neuville-Saint-Amand – La Vallée de Neuville BSR 2005, 31-33. Ovillers Agache 1978, fig. 26. Parné Ferdière 1988, fig. 2. Plailly - La Butte Grise CAG 60, 375; Gallia Informations 1989, 233-235 Port-le-Grand Agache 1978, fig. 24. Pouligny-Saint-Pierre-La Grave Coulon/Holmgren 1992, 97-8. Quevauvillers Agache 1978, fig. 22. Quincy-Launay Holmgren/Leday 1981, fig. 9. Ribemont-sur-Ancre Agache 1978, fig. 22. Rogy Agache 1978, fig. 16. Roisel - Rue du Nouveau Monde BSR 2004, 119-120. Romegoux Agache 1983, 17. Roye- ‘Le Puits-a-Marne Collart 1996, 132-7. Saint-Georges-sur-Arnon  Holmgren/Leday 1981, fig. 1.3. Saint-Just-en-Chaussée Agache 1978, ph. 148. Saint-Quentin – Parc des Autoroutes C  AG 02, 397-399; BSR 1999, 33-34, BSR 2002, 45-46, BSR 2004, 39-43, BSR 2004, 40-42. Sancerre-Orme aux Loups Chevrot/Troadec 1992, 312. Somain-Derriere L‘Abbaye CAG 59, 409-410. Soyecourt Agache 1978, fig. 23. Thizay-La Bastille Coulon/Holmgren 1992, 155-7. Tilloloy Agache 1978, ph. 185. Translay Bayard 1996, 164-165. Trinquies Bayard 1996, 165. Vaulx-Vraucourt-Chemin de Morchies Delmaire 1994, 328-9. Vaux-sur-Somme Agache 1978, fig. 22. Venette-Bois de Plaisance-zone 1 BSR 2004, 83 ff. Verneuil-en-Halatte-’Le Bufosse’ Collart 1996, 124-132. Villeneuve-sur-Cher-La Grande Pièce Holmgren/Leday 1981, fig. 9.2. Villers-Bretonneux Agache 1978, ph. 180. Ville-sur-Ancre Agache 1978, fig. 21. Vineuil-Grande Pièce Coulon/Holmgren 1992, 183. Vitreux-D’Acey Rothé 2001, 765-6. Warfusée-Nord Agache 1978, fig. 24. Warfusée-Sud Agache 1978, ph. 181. Zouafques CAG 59, 109-110.



b elgium Anthée Del Marmol 1877, 1881. Bruyelle-Haute Éloge Corbiau 1997, 319-322; Brulet 2009, 305-309 Hamois-Champion-Sur Rosdia Van Ossel/Defgnee 2001. Jemelle-Malagne-Rochefort Mignot 2006. Meslin-L’Eveque Deramaix 2006. Haccourt De Boe 1973, 112; idem 1975.

lu x em b ourg Aspelt-Kleppen Carte Archéologique du Gr. D. de Luxembourg 26, 1977, 63-64. Bertrange Krier 2000, 2001. Echternach Metzler et al. 1981. Mersch www.mersch.lu. Diekirch Paulke 2010.

s w it z erla n d Biberist-Spitalhof Schucany 2006. Buchs Horisberger 2006. Dietikon Ebnöther 1995. Genève-Parc de la Grange Ebnöther/Monnier 2002, fig. 130. Liestal-Munzach Drack 1975, 53; Ebnöther/Monnier 2002, fig. 152. Neftenbach Ebnöther/Monnier 2002, fig. 138. Oberentfelden Ebnöther/Monnier 2002, fig. 139.6. Orbe-Boscéaz Ebnöther/Monnier 2002, fig. 139.5. Vicques-Variant Flutsch/Niffeler/Rossi 2002, 400-1. Winkel-Seeb Flutsch/Niffeler/Rossi 2002, 403; Drack 1975, 51. Yvonand-Mordagne Ebnöther/Monnier 2002, fig. 143.

germa n y Blankenheim-Hülchrath Oelmann 1916. Borg Birkenhagen 2004; idem, this volume; Frey 2000, 2001. Fliessem Cüppers 1975. Gerolstein Henrich/Tabaczek/Zelle 2005. Gillenfeld-Strohn-Römerberg Henrich/Mischka 2006. Heitersheim Nuber 1995. Meckel Smith 1997, 159. Nennig Steiner 1934. Reinheim Petit/Sarateanu-Müller 2005; Sarateanu-Müller, this volume. Thür (Mayen-Koblenz) Cüppers 1990, 574-5.



Town-country dynamics in Roman Gaul. The epigraphy of the ruling elite Ton Derks 1 2 3 4 5 6

Introduction Composition and size of the ordo decurionum Urban domus and villas Rural properties and burial places Bonds of friendship and patronage Conclusions

Abbreviations References Appendices 1-5

1 i n troductio n In the western half of the Roman empire, the foundation of towns and the introduction of a system of civic self-administration constitute the most important changes of the Roman era, alongside the rise of villas.1 The precise organisation of the new administration differed from civitas to civitas as in most cases it was partly built upon existing indigenous structures. However, as a result of a process of political integration, referred to in the recent literature as ‘municipalisation’,2 the institutional organisation of the local communities developed along much the same lines and finally became fairly uniform across the Gallic and Germanic provinces. Everywhere the old tribal senate of elders was transformed into a Roman-style city council, variously referred to as senatus, curia or, most often, ordo decurionum.3 The tribal pagus lost its political autonomy and was subordinated to the civitas, its significance in the Roman period being largely restricted to the religious domain.4 And single magisterial positions rooted in local tradition, such as the vergobretus and praetor in Central Gaul or the summus magistratus among the Batavi,5 quickly disappeared and were replaced by the archetypical collegial pairs of duumviri, aediles and quaestores. It is these municipal magistrates, together with the members of the city council with whom they were closely connected, which are the focus of this paper. They are referred to here as the ‘ruling elite’. Who were these men? How many of them were there? 1

Raepsaet-Charlier 1998, 177-179.

Both developments are inextricably intertwined. Cf.  Whittaker 1990, 116, who states that ‘villas were primar-

5

For useful reviews of the evidence, Dondin-Payre 1999,

ily, although not of necessity, an urban phenomenon (…),

150-153; Lamoine 2009, 106-134; Roymans 2004,

a phenomenon of a central civitas town’.

63-64, 200-202. These ‘traditional’ offices probably rep-

2

Dondin-Payre 1999, esp. 127.

resent a first stage in the process of municipalisation and

3

Cf. RE IV, 1901, col. 2319-2352, s.v. decurio, esp. 2319-

may be more Roman than often acknowledged. Cf.

2322 [Kübler]; Dondin-Payre 1999, 159-161.

Dondin-Payre 2003, 147.

4

Wolff 1976, esp. 117; Roymans 1990, 19 ff, esp. 21; 



What evidence do we have to substantiate the general claim that they may be identified with the wealthy owners of the greatest and most luxurious villas in the countryside? How much of their time did they have to spend on exercising their political power? How did they manage to run an agrarian enterprise and at the same time be politically active in town? Should we imagine them constantly commuting between their estate in the countryside and the political arena in the town? Or were they absentee landlords who, like their peers in Italy, lived in town and relied on a bailiff? And where in the end did their loyalties and sympathies most lie – with the town or the countryside? These and other questions spring to mind when trying to conceptualise the social and political relations that tied the Roman villa world to that of the town. Let me be clear right from the start: the available evidence will not allow us to provide definitive answers to all of these questions in the next few pages. Unlike the situation for Italy or the Spanish province of Baetica, which have produced a rich dataset of literary or legal documents shedding light on the life of the ruling elite,6 apart from the Constantinian decree allowing the council of Cologne to call Jews to political engagement in the local ordo,7 we have no legal sources at all for the northwestern provinces. And as far as the literary evidence is concerned, we have to make do with just a few snapshots, which moreover are of a comparatively late (4th- or even 5th-century) date. Thus what has generally been offered so far are inferences based largely on analogical reasoning, using the evidence from the Italian peninsula and the Mediterranean provinces as a model. The question of whether we can validly apply the ‘Italic model’ of town-country dynamics to the villa landscapes of the northern provinces has virtually never been asked, let alone been answered satisfactorily.8 Although inscriptions from the research area may be a more direct source of information than textual evidence from Italy, they are not without their problems either. They are few in number and have an uneven geographical distribution; they are more numerous in some civitates than others9 and, following a general characteristic of epigraphic culture, are more often found in urban contexts than in the countryside.10 They also suffer from a chronological bias with many more inscriptions dated in the 2nd and 3rd centuries than in the 1st. Finally, the content is often formal and stereotypic and thus of reduced informational value.11 The area for which detailed evidence has been collected for this paper (see below) presents no exception to these general trends. Nevertheless, I believe that by paying special attention to the precise find contexts of the inscriptions, we may still gain some interesting new insights into the ways town and country were linked in this part of the Roman world. In order to get the most out of them, I will adopt an interdisciplinary approach which seeks to integrate historical, epigraphic and archaeological evidence. In line with the geographical framework of the project from which this paper springs, the focus will be on the northern villa landscapes between Bavay in France and Cologne in Germany. But before we start discussing the evidence from the area, a few general words must be said on the organisation of the civic administration (Section 2) and on the ruling elite’s places of residence (Section 3). The aim of these sections is, first, to investigate the validity of the ‘Italic’ model for town-country relations in Roman Gaul and, second, to present a frame of reference for the subsequent interpretation of our data (Sections 4 and 5). For the following two sections the geographical scope has therefore been widened to include the three Gallic and two Germanic provinces.

6

One thinks of the correspondence of men like Cicero,

9

Cf. the map in Dondin-Payre 1999, 133, fig. 1.

Horace and Pliny (cf., for instance, White 2010; Hirth

10

Cf., for instance, Wightman 1985, 164, fig. 24; Woolf  1996.

1985; Sherwin-White 1985) and of the municipal charters of Tarente, Urso and Irni (Crawford 1996, nos. 15

11

Cf. Galsterer 2010, 257 f., who states that in the literary

and 25; González/Crawford 1986).

sources ‘often the fullness of life comes to speak’, whereas

7

Eck 2004, 325.

with inscriptions, ‘one always has the feeling of staying at

8

Slofstra 1983, 84 ff, esp. 85, constitutes an exception in

the surface of municipal life’ (my translation).

that the question is posed, but not really answered.



2 com p ositio n a n d si z e o f the ordo decurio n um It should first be pointed out that all magistrates and decuriones exercised power over both the town and country of the civitas, on behalf of which they were nominated. Despite the sometimes vast territories of the civitates, nothing like a multi-scalar administrative hierarchy existed. Magistrates and councillors had their seats in the civic centre of the town, from which they exercised power over the entire territory of the civitas, not just the urban centre. To be eligible for a municipal magistracy or for a seat in the council, candidates had to meet several legal, moral and financial requirements.They had to be a free-born citizen,12 of local origin,13 free from any allegations of infamia,14 and be able to satisfy property qualifications.Women were excluded from the offices and the council. Given these restrictions, the pool from which the municipal elite could be recruited was often limited. Property criteria differed from one community to another (and perhaps also through time) and related to the community’s size and richness. If we are to believe Pliny the Younger, in the early 2nd century AD the lower limit in his home town of Comum was 100,000 sesterces.15 Even if this figure has no generalising value of itself, in the absence of other relevant data and when compared with the property requirements of 400,000 and 1,000,000 sesterces which applied to knights and senators from the Augustan period onwards,16 it may provide a rough indication of property limits for municipal elites elsewhere in the empire.17 With Roman society being pre-dominantly agrarian,18 most magistrates and councillors will have met such minimum requirements for wealth through the possession of landed property.While this was certainly the most accepted way to fulfil the economic criteria, in many cities there were also those who satisfied the capital requirement thanks to ‘new money’ that they had earned through trade or transport.19 Finally, veteran soldiers of the Roman army incidentally turn up as members of the city council or holders of a municipal office. But apart from Roman colonies with veteran settlement (deductio), where veterans of ordinary rank may have dominated the local senate, at least in the early years after the colony’s foundation,20 former soldiers who obtained magisterial or decurional honours generally remained an exception.21 12

13

González/Crawford 1986, ch. 54; CJ 9.21.1; Mouritsen

amassed a fortune by trade or shipping is documented

2011, 73-75.

by two votive altars dedicated to Nehalennia. Stuart/

In exceptional cases, non-residents could be coopted.  E.g. CIL XII 1585, which honours a flamen divi Augusti

14

Bogaers 2001, B37 and B63. 20

Cf. the etymology of ‘decurio’ in Dig. 50.16.239.5: decima

of the Vocontii who had been allected, nomine incolatus,

pars eorum qui ducerentur consilii publici gratia conscribi

into the ordo of Lugdunum. For other examples from the

solita est. Due to the chronological bias of the epigraphic

Three Gauls, often concerning two neighbouring com-

material the presence of veterans in the magistracies and

munities, see Dondin-Payre 2003, 152.

ordines of the Roman colonies at Cologne and Xanten

Infamia could be brought about by dishonourable dis-

is almost invisible; the only exception is Appendix 1, no.

charge from the army, by the pursuit of a ‘dishonourable’

1 (see the comment by H. Galsterer in IKöln2 286). The

profession such as actor, gladiator, pimp or prostitute, by

epigraphic habit was still burgeoning in the early decades

conviction for criminal offence or by condemnation in

of the colonies, whereas in the late 2nd and early 3rd

particular civil cases. Cf. Dig. 3.2; 48.7.1; 50.2.12; RE IX,

centuries, when inscriptions abound, the composition

1916, col. 1537-1540, s.v. infamia [Pfaff].

of the colony’s population and its administration had

15

Plin., Ep. 1.19.

16

Nicolet 1976; Demougin 1988, 76-79.

17

Cf. Jacques/Scheid 1998, 275 f.; Duncan-Jones 1974, 147

contribution of veterans to the civic administration of

f.; Eck 2004, 315 f.; Galsterer 2010, 258.

the communities they settled in has often been overes-

On the limited volume of artisan production and trade

timated (however, see preceding note). Those who did

in the ancient economy, see the classic discussion on the

serve their community as office-holders or councillors

consumer city in Finley 1977; Whittaker 1990.

had most often served in the better paid higher ranks of

For the Batavians, for instance, a councillor who had

the legions rather than the auxilia.

18

19

become much more heterogeneous. 21

Cf. Wesch-Klein 1998, 196-200, who argues that the 



Fig. 1. Plan of the Hauptforum of Augusta Raurica with the curia (C) attached to the basilica (B) (after Balty 1991, 274, fig. 150).

How many seats did the ordo decurionum have? A key document for any discussion of the size and composition of the council is the well-known ‘album decurionum’ of Canusium in Southern Italy.22 The inscribed bronze table, which was found in 1675 in the ruins of what was presumably the towns’ council chamber, lists in four carefully laid-out columns the names of all ordinary and honorary council members active in AD 223, the date of publication. The first column lists the names of the honorary members who had been officially appointed civic patrons of the town – first 31 of senatorial rank (clarissimi viri), then eight of equestrian rank (equites romani).The last column mentions 25 praetextati, young nobles generally under 25 who were waiting their turn, hoping to become full council members in the near future, and who for the time being were permitted to attend the council, albeit without speaking or voting rights.23 If we subtract these 64 names from the total number of 164, the actual ordo came to exactly 100 members. Apart from the much-discussed category of the pedani (to which we will soon return)24 and four members who had entered the council directly by allectio into the magisterial rank of duumvir quinquennalicius,25 all council members

22

CIL IX 338. For an excellent account, Salway 2000.

magistrates and points to these backbenchers’ ambition of

23

On the praetextati, see the discussion in Mouritsen 1998,

holding office soon. Cf. the sons of senatorians who were

239 f. and Laes 2004, 154, 179, who rightly distinguishes

permitted to wear the tunica lato clavo and attend meetings

between the young councillors under the age of 25

of the senate before they had held any senatorial office;

being coopted for exceptional reasons (ex causa) as full

for this, Suet., Aug. 38.2; Nicolet 1976, 34. For the impe-

members of the council (adlecti), despite the legal age

rial age limit of 25 for entry to magistracies or the local

limit of 25, and the larger separate group of praetextati,

council, see Dig. 50.4.8; González/Crawford, ch. 54 (lex

who functioned as a first natural reserve to replenish any vacancies. Whereas the former group comprised quite a

Malacitana) and the discussion in Laes 2004, 158 ff. 24

230, note 4; Taylor/Scott 1969, 548-557.

years old (cf. Laes 2004, 176-177, table 8), the latter generally must have reached the age of legal adulthood and will have worn the toga virilis. The title praetextati is thus to be explained by reference to the toga praetexta worn by



Pedani turns out to be a hapax, derived from pedaneus, itself perhaps a corruption of pedarius. Mouritsen 1998,

few minors who were under 14, some as young as 4 (!) 25

Mouritsen 1998, 241; Salway 2000, 124.

Fig. 2. Photograph of the restored seating space of the curia of Augusta Raurica with on the left the podium for the chairs of the duumviri presiding over the council meeting (author’s photograph).

were former magistrates.26 In keeping with legal prescriptions,27 the magistracies – normally held for one year with the possibility of one or more renewals – are listed in descending rank, according to the dignity and responsibilities connected with the office. The duumviri quinquennales, who were nominated every fifth year of a censorial cycle with the special task of conducting the census (and renewing the ordo),28 head the list, followed by the ‘ordinary’ IIviri (i.e. iure dicundo) who held office in one of the other four years of the census period. The duumviri, regardless of whether they had the special power of censorship, presided over the deliberations in the council and were charged with jurisdiction and holding elections. Next are the aediles, who were responsible for public order, religious buildings and festivals, the inspection of weights and measures, and the markets. The fourth group of magistrates are the quaestores, to whom the community’s public finances were entrusted.29 Concluding the core list of 100 men are the pedani, who held no office and were lowest in rank.The fact that the names of the quinquinnales charged with the publication of the album are mentioned twice – once in the heading of the inscription and once in final position under the quinquennalicii – suggests that the listing within each category observed the rule of seniority.30 Although nominally all councillors had to be re-elected every five years, notwithstanding exceptional cases (e.g. when a decurio had lost his good reputation), it was standard practice for most decuriones to occupy a seat for their life-time.31 26

This is clear from the -icius suffix in the headings of the

the stipulations in the legal charters of the Colonia

office titles that precede their names: quinquennalicii, IIvi-

Genetiva Iulia (Crawford 1996, no. 25) and the Muni-

ralicii, aedilicii, quaestoricii. 27

Cf. Dig. 50.3.1.

28

Salway 2000, 121, with further references in note 20.

29

For the responsibilities connected with the offices, see

cipium Irnitanum (González/Crawford 1986, ch. 18-29). 30

This again is in accordance with the legal prescriptions. Cf. Dig. 50.3.1; Salway 2000, 125.

31

CJ 10.32.8–10.32.294 and CTh 16.1.2–16.1.380.



The crucial question is, of course, to what extent this unique document may be considered representative of the ordines in other towns. Although the album of Canusium has long been taken as evidence for a standard ordo size of 100 members, recent research has increasingly shown that this cannot be the case.32 First, like the property requirement for entry to the council, the size of the ordo seems to have been dependent upon the size and prosperity of the community. Canusium was a thriving Roman colony that was promoted to this status only in AD 160.33 We can expect smaller and less prosperous communities to have had a smaller ordo; for some, this has been established with certainty. Thanks to the discovery of the lex Irnitana, for instance, we know that the council of Irni, a small community in Baetica,34 comprised no more than 63 members, in accordance with an old custom that went back to before the promotion to municipium. Only 30 members are documented for the small Italian town of Castrimoenium, and a similar size has been argued for the council of Petelia in Bruttium.35 In short, even if many communities did have an ordo of 100 councillors,36 the cited examples show that there was room for flexibility. Based on the fact that it was published in bronze instead of the usual whitened wax tablet (tabula dealbata or album),37 a second argument has recently been suggested as to why the album van Canusium cannot be representative. The financial investment and care associated with making lengthy inscriptions in a durable medium like bronze means that this type of publication was reserved for documents that were important for the collective memory of the community. We may conclude from this that the publication of the album in 223 AD was no routine task, but related to a special circumstance or occasion deserving of preservation for the collective memory of the community. Based on the large number of pedani (32),38 Salway proposes the unprovable but attractive hypothesis that the reason was a special beneficium of an imperial grant allowing the Canusine council to increase the maximum permitted number of decurions to 100. Such an imperial beneficium commemorated by the production of the bronze album would account for the unusually high number of councillors without any administrative experience.39 If this suggestion proves to be correct, this means that the council of Canusium also initially had fewer than 100 members. The above gives us every reason to assume, also for the ordines of the communities in the northwestern provinces, that there was no fixed size but rather a considerable variation in time and space. Paradoxically, for the communities for which we have concrete information, the data suggests that their councils were larger than 100 men! According to Caesar, the senate of the Nervii consisted of 600 members, all of whom –with three exceptions – were killed in battle.40 A comparison with the ordo of Canusium, together with the war context of Caesar’s report, could suggest that this was a case of rhetorical exaggeration. I am inclined to attach more credence to Tacitus’ statement that by the end of the 1st century the council of the Treveri had at least 113 members, which was the size of just the pro-Roman faction.41 Lastly, the archaeological remains of the curia of the colonia Augusta Raurica in Augst suggest that the senate of the Rauracenses had at least 100 members in the mid 2nd century. According to older calculations, the five concentric steps on which the seats of the decuriones were arranged during council meetings (figs. 1 and

32

Cf. Nicols 1988; Mouritsen 1998, 231 ff.

38

Mouritsen 1998, 238-241.

33

Salway 2000, 118-120.

39

Salway 2000, 169.

Galsterer 2010, 263, assumes for this town a territory of

40

Caesar, BG 2.28.

about 50 km2 and a few thousand inhabitants.

41

Tacitus, Hist. 5.19: centum tredecim Treverorum senatores.

34

35

Duncan-Jones 1974, 284.

Cf. Chilver/Townend 1985, 100, who remark that this

36

To cite one example for which this size has been proven,

is ‘a curiously precise number, evidently to emphasize

the senate of the Colonia Iulia Genetiva had 100 mem-

the number of senators still remaining after defections.’

bers. Cf. Crawford 1996, no. 25, ch. 75, 97 and 126 (with

Tacitus gives no information on the size of the defected

comment on p. 399).

anti-Roman faction in the senate of the Treveri.

37

Salway 2000, 121. Cf. also Eck 2004, 321.



2) had space for 98 seats. With two seats for the presidents on the rectangular podium (1.74 m x 3.05 m) on the floor in front of the semicircle, this brings us to the round number of 100.42 However, if we use the standard space per seat that is generally adopted as the norm for ancient theatres, the number of available seats and thus the size of the council is nearly doubled.43 Contrasting with such prosperous communities that boasted a large territory and a large council were small and poor civitates with a much smaller council. Ranking among them for certain at the empire’s northern periphery were the Cananefates, the Batavi and the Menapii,44 and possibly also the Traianenses. The territories of these civitates (at least the first three) were comparatively small and the less fertile soils meant there were fewer opportunities for amassing a fortune through landed property and arable farming.45 Villas are hardly known here and those that have been identified are generally small.46 The widelyheld view that the municipal elite consisted of a class of rich landowners who lived in sumptuous houses of the axial type cannot therefore apply here. In less prosperous communities the council must have been smaller or the census for municipal offices significantly lower.

3 ur b a n domus a n d v illas The political work of the ruling elite, especially the ruling magistrates, required them to be in town on virtually a permanent basis. In addition to council deliberations,47 they were expected to attend a range of ceremonial occasions, such as religious festivals, distributions of food or money, sacrifices, offerings, banquets etc. The texts of the municipal laws clearly show that office-holders were allowed to leave the city (i.e. the civitas) but had to nominate a replacement (praefectus) in the event of a lengthy absence. From the provisions in these laws regarding the quorum needed for council decisions, we can gather that it was not uncommon for not all the members to attend council meetings. We do not know how often and what proportion of council members failed to appear, but it seems that decuriones could shirk their obligations in this regard more easily than magistrates. If we have to assume that most magistrates and councillors were landlords who owned one or more rural estates, all of them possessed a town house, an urban domus. Provisions in the late republican municipal laws of Tarente in southern Italy and Urso in the province of Baetica show that maintaining a home in these towns was not a matter of free choice for council members and office-holders, but a legal duty. These laws obliged those who did not already own a house in town to do so within a set period once elected to the position of magistrate or councillor.48 Although such charters have been preserved for only a few towns, it is generally assumed that similar laws with comparable regulations also existed elsewhere.49

42

Laur-Belart 1937, 34, assuming a space per seat of 90 cm.

47

For all measures concerning the seating space as well as

needed the approval of the council. For a general impres-

a balanced discussion of the dating of the curia’s second

sion of other topics dealt with by the ordo, cf. Sherk 1970,

construction phase to c. AD 145 (which fits in nicely with that of the Hauptforum: Bedon/Chevallier/Pinon

esp. 73 ff; Parma 2003. 48

45

46

c. 91 (Lex Ursonensis).

Cf. Balty 1991, 273, note 80, which gives a space per seat of 40-50 cm.

44

CIL I 590 = ILS 6086 = Crawford, no. 15, ll. 26-28 (Lex Tarentina); CIL II.5, 439 = Crawford 1996, 393 ff, no. 25,

1988, 225 f.), one should now consult Balty 1991, 271 ff. 43

All changes to the layout or ownership of public space

49

For Cologne, cf. Eck 2004, 315. Internal evidence to sup-

Buijtendorp 2010, 852, recently assumed for the Canane-

port this is the fact that the laws that have come down to

fates an ordo of 30 members.

us make only sporadic mention by name of the town to

Roymans 1996, 72-84; Galsterer 1999, 266-267; Schalles

which they applied; many articles only refer to the com-

1995, 426; idem 2001. See also above, note 19.

munity in general terms (municipium, colonia) that could

Habermehl 2011, 110, fig. 4.7.

have been used for any other town.



These town houses of the decurial elite may have been pieds-à-terre that were barely distinguishable in archaeological terms from other houses; this will have been particularly true of the homes of backbenchers. The most powerful men, however, had a sumptuous dwelling designed with domestic spaces that were suitable for receptions and audiences, political meetings in smaller circles, and personal archives and libraries. The physical remains of such large, luxurious town houses, often with peristyles and floor areas ranging from 1500 to 3500 m2, have been unearthed by urban archaeologists in almost every civitas capital in the villa landscapes where large-scale excavations have been carried out.50 Thus although all the men of importance in the civitas were urban residents by obligation, living in town quickly became very attractive. Outside the council room, this was where the ruling elite met their social peers. And even in the Roman period the town was the place par excellence where a wide range of new cultural phenomena made their first appearance. For those wishing to belong not just to the political but the cultural elite, the town with its amenities and constant supply of new and exotic products was the ideal location. Living in an urban centre also meant that the politically engaged landowning elite had to cope with the conflicting interests of their negotia publica in the town and the private affairs of their estate in the countryside. Several options were open to them: leasing large parts of their property to tenants, entrusting the management of their estate to an actor or a vilicus, or a combination of the two.51 This model is supported by rich epigraphic evidence and epistolary exchanges from Italy,52 but several inscriptions from Gaul and Germany53 and a few mentions in letters54 by Gallic nobles attest to the fact that business managers acting on behalf of absentee landlords were not unknown in the villa landscapes of the Gallic and Germanic provinces. A recurring theme in the correspondence between Cicero and Pliny the Younger is the escape from the bustle of the town and the attraction of life in the country. Thanks to the requirement to own at least one house in town and one in the country, the ruling elite could at set times exchange the hectic and noisy town life for a peaceful sojourn in the country. For this part of the ‘Italic model’, the most detailed and explicit description we have for Gaul comes from the 4th-century rhetorician and poet Ausonius. After ten years at the imperial court in Trier, where he worked as tutor to the young Gratian, he had returned to his patria Burdigala, where he inherited the small suburban ancestral property (villula) on the death of his father. In the short poem De herediolo, Ausonius sings the praises of the estate that had been in his family for generations. The concluding verses are worth citing in full: ‘This my estate lies not far from the town, nor yet hard by 50

51

Cf., generally, the excellent paper by Vipard 2007. For the

Apart from CIL XIII 4352 (Metz): vil(icus), for which we

state of affairs regarding excavations in the urban centres

cannot be certain whether it refers to a vilicus of a farm

of Northern Gaul, Brulet 1996 and Hanoune 2007, and,

or of an enterprise in town, there are no inscriptions of

specifically for the research area,Vanderhoeven et al. 1992

agricultural vilici from Gaul. An inscription in the floor

(with English summary); Eck 2004, 378-385 (with older

mosaic in the entrance hall of the villa of Laufenburg

literature in ch. 11, note 2); Kienzle 2008. Even with

mentions a cliens who seems to have acted as business

only 10% of the town’s surface having been excavated,

manager (see below, note 72).The inscription from Bavay

the absence so far of sumptuous urban domus in Xanten

(CIL XIII 3572) has not been included; with Dondin-

is noteworthy.

Payre 2003, 150 and Lamoine 2009, 326, note 139, I read

On the likely combination of tenancy and agency, 

vilicus there as the cognomen of the inscription’s dedica-

Aubert 1994, chapter 3; on actores and vilici of agricultural

tor, named Q. Titius Vilicus. I will discuss the inscription

estates generally, Aubert 1994, 132 ff.

in the next section.

52

Cf. Février 1981; Mattusch 2005, 3-12.

53

Inscriptions from Gaul mentioning agricultural business

nius complains about the incompetence of his vilicus. For

managers: CIL XIII 2243 (Lugdunum): actor praedio-

references and discussion, see Aubert 1994, 141-143.

rum horum; CIL XIII 2533 (Ameyzieu, territory of the Ambarri): actor fundi Ammatiaci b(onorum) Flavi Stratonis.



54

In a letter addressed to his friend Paulinus of Nola, Auso-

the town, to rid me of its crowds while reaping its advantages. And so, whenever satiety moves me to change my seat, I pass from one to the other, and enjoy country and town by turns’.55 This and similar passages in other late antique Gallic writing, especially that of Sidonius Apollinaris, show that the classical model of frequent travel and changes of residence between town and country still applied in 4th- and 5th-century Gaul, suggesting that a tradition introduced in the early imperial period still persisted.56

4 rural p ro p erties a n d b urial p laces Let us now turn to the data concerning the ruling elite in the research area. Inscriptions mentioning magistrates and decurions from the Nervii, Tungri, Traianenses and Agrippinenses have been gathered for this paper. There are no such inscriptions for the Cugerni and Ubii, on whose territories the Colonia Ulpia Traiana and Colonia Claudia Ara Agrippinensium were founded in the 1st century. A striking feature is the uneven geographical distribution as well as the widely varying numbers that have been preserved for the four communities, with many times more inscriptions for the Agrippinenses and Traianenses than for the Tungri and Nervii (fig. 3 and Appendices 1-4).This is most likely due to the different epigraphic habits in the Roman frontier zone and hinterland. Despite these inherent limitations and distortions of the dataset, how can we exploit it to say something about the ties that bound the municipal elite to the countryside in this particular part of the empire?57 First, we will use the geographical information of the findspots to identify possible rural properties of the ruling elite. Inscriptions found in secondary contexts do, of course, have to be treated with caution since their findspots only provide approximate indications. This is especially true for inscriptions reused in medieval contexts, which may have been transported over larger distances than spolia from the Late Roman period, as well as for texts found in or near the town since in those cases a rural origin from a suburban villa or sanctuary may be obscured by the secondary urban context. Second, we will use the nature of the inscriptions (funerary, votive, honorary or building inscription) in combination with archaeological information about find contexts to discuss the type of bonds linking the municipal aristocracy with the countryside. In this section, we will use the evidence of the funerary inscriptions for a discussion on burial places of the municipal elite. If all decurions and magistrates were obliged to own a town house, and most of them had one or more residences in the countryside as well, this generally gave them at least two options for their choice of a final resting place – along one of the arterial roads leading out of town or on one of their rural properties. What preferences do we observe and what considerations may have motivated magistrates and decurions to choose a particular option? A first observation to be made is that since the final resting place will have been partly determined by the individual life course and career path, not all members of the decurial elite will have had the same freedom of choice at the end of their life. There was little choice for the surviving relatives of magistrates who had seen a brilliant political career and were decreed, in reward for their services to the community (ob merita), the honour of a public funeral, a burial plot at public expense or even a funerary monument paid by the civitas. As far as we can tell, reception of such honours always went hand in hand with burial outside the walls of the town. There are no examples from the research area of men who received such municipal beneficia, the nearest instance being a scion of the famous Camilli family from the Helvetian capital at Avenches whose funeral was jointly paid for by the Haedui and his home community.58 As the 55

56

Ausonius, De herediolo 29-32. Translation H.G. Evelyn

source-critical remarks.

White (Loeb).

57

For the method, compare Février 1981, 360.

Goodman 2007, 226 ff. esp. 227 f. (with notes 178-179),

58

CIL XIII 5110. The evidence for the practice has been

who provides further references and some pertinent

usefully collected by Wesch-Klein 1993, 196-200.



A Rh

B

in

e

Meu

10 se

se 5 6

0

se

l

le

Mo

13 mm

2

11

12

So

1

4

ppi

i

n

3

Agri

N

i erv

Tu

8

gri

nen

9

s

Tr

ai

an

ens

es

7

e

100 km

Fig. 3. Distribution of inscriptions mentioning magistrates or councillors of the Agrippinenses, Traianenses, Tungri and Nervii. A urban finds; B rural finds (large symbol: more than 5 inscriptions). 1 Cologne; 2 Bonn; 3 Gereonsweiler; 4 Zülpich-Hoven; 5 Nettersheim; 6 Blankenheim; 7 Xanten; 8 Heerlen; 9 ValkenburgRavensbos; 10 Gors-Opleeuw; 11 Vaux-les-Cherain; 12 Bavay; 13 Saint-Hilaire-sur-Helpe-‘Fuchau’

bestowal of such honours by the local ordo decurionum will have been exceptional, such circumstances only applied to a highly select group.Those who received these privileges probably happily enjoyed them and after an entire life centred around the town felt no remorse at being deprived of the alternative. Similarly, veteran soldiers and negotiatores who had settled in town and acquired a seat in the town council only after, or in the final phase of, their professional career might not have had much choice either. Though some of these councillors may have invested in a rural property, most will have had no bonds with the surrounding countryside at all and will have opted for interment in one of the urban cemeteries. Again, such councillors will have formed only a tiny minority of most senates.59 An example from Cologne is the anonymous duumvir who was inscribed in the Voturian tribe and must therefore have come originally from Ostia, Placentia or Bergomum in Italy. He was most probably a veteran soldier or officer.60 If the cases discussed so far are rather exceptional, how might the majority of councillors have looked at the issue? Since burying the dead within the perimeter of the city walls was legally prohibited, opting for a burial place in an urban cemetery meant that the funeral monument would be spatially separated from the decurion’s urban house. In contrast to interment on a rural property, the choice of a tomb just outside the town gates required considerable additional investment in the form of purchasing a burial 59

See above, Section 2 with note 21.

11

60

IKöln2 286 = Appendix 1, no. 1.

plot. After the first generation of councillors had passed away, the most prominent places will have been taken, resulting in plots at conspicuous locations soon becoming extraordinarily expensive. Against this background, the option of being buried on the main estate in the countryside may have had several advantages. While the location of the burial might have been less desirable (though often still along a road), the plot was free and the tomb itself probably less vulnerable to alienation and destruction. For those who owned a property that had been in the family for generations,61 the prospect of being buried in ancestral ground might have had an added emotional value. Finally, building an imposing monumental marker also enhanced the symbolic value of the villa as the embodiment of the family’s aristocratic status and identity.62 So what choices were made by the decurial elite in our research area? Eight funerary inscriptions mentioning members of the local municipal elite are known from the territory of the Agrippinenses. Most of them (six out of eight) have come to light in the urban centre or its immediate vicinity. Although all of these seem to have been found in a secondary context, there are no reasons to assume that their original primary context was rural rather than urban. Three were already reused in antiquity (Appendix 1, nos 1, 4 and 6) and may have been found not too far away from their original spot. Of the remaining three, one was found in Spellen on the right bank of the Rhine some 100 km downstream from Cologne but is said to have been taken there from the Rhenish capital, and little is known about the find circumstances of the other two. It is hard to interpret the two funerary inscriptions found in the countryside, at Nettersheim and Zülpich, as other than approximate indications for the location of decurial estates and the funerary monuments which had been erected there. We have no funerary inscriptions for the Traianenses, but the two we have for the Tungri are both in the countryside, as is the only inscription of the Nervii that may have been funerary in nature. Two of these three texts are only known through manuscripts, with a consequent loss of information. In the late 18th century, an inscription for a duumvir of the Nervii was seen at Saint-Hilaire-sur-Helpe next to the Roman road leading from Reims to the civitas capital of Bavay, which lies about 20 km further to the north. The inscription was set within a tabula ansata and erected by a libertus. It is not entirely clear from the text whether this is a funerary or an honorary inscription, and since there is no additional information on the type of material that supported the inscription (stone, bronze?) or on its dimensions, the interpretation as a funerary monument has to remain speculative.63 Noteworthy is the location on the road.The other inscription that has come down to us via a manuscript concerns a tombstone by an aedilis of the civitas Tungrorum erected for himself and a former slave; the inscription was seen in Gors-Opleeuw, less than 10 km northwest of Tongres on the northern limit of the area with a dense distribution of villas and tumuli. The third and only surviving funerary inscription was erected by Vitorius Florentinus in commemoration of his father, the Tungrian decurio Vitorius Caupius. Its findspot at Cherain in the southeast of the civitas Tungrorum (cf. fig. 3) has sometimes been taken as proof that he was a councillor of a pagus in this part of the civitas,64 but a far simpler explanation is that it indicates the location of his main rural property where his funerary monument was built. As already mentioned, the evidence is sparse and the picture is partly determined by chance, but the significant regional differences that we still observe are remarkable. The relatively high number of funerary inscriptions from the Roman town of Cologne is surprising as they point in a different direction to what comparable investigations into the epigraphy of magistrates and councillors of communities in Gallia Narbonensis and Gallia Cisalpina have shown65 (and what the few inscriptions from the other 61

Cf. Ausonius and his villula.

62

Bodel 1997.

63

Similar methodological remarks are made by Février 

64

Cf. ILB 60 with the corrigenda in ILB2 60.

1981, 360. In the immediate vicinity of the findspot,

65

Février 1981.

of a marble funerary stone have been found. Cf. Carte Archéologique de la Gaule 59, pp. 383 f.

however, Roman cremation pits and a small fragment



civitates suggest). With one exception, onomastic data of the known members of the colony’s municipal elite point to them having local roots. This means that the different picture cannot be explained away by ascribing most of these funerary monuments to ‘foreigners’ who had only one real choice – buying a burial plot in one of the urban cemeteries. The only explanation I can offer is that this apparent preference among the colony’s municipal elite is a sign of the town’s success. Only future finds may help us to see whether Cologne really is an exception in this respect. If the small sample allows us to say anything further, it seems that the decurial elite of the Tungri and Nervii were more inclined to opt for burial in the countryside. For the Tungri such a preference may be confirmed by the archaeological evidence.Vanderhoeven observed that ‘there are no elite graves in the cemeteries at Tongres, although we did find their houses in the town.’66 Instead, from the late 1st century onwards, there are many monumental tumuli in the countryside around Tongres, which were clearly associated with a rich elite; the inscription of the Tungrian decurio at Cherain was found at precisely such a tumulus. A similar argument may perhaps be made for the Nervii, although the empirical basis for cemeteries is much smaller there and caution is therefore needed. However, a recent study of a cemetery along the arterial road leading out of Bagacum to the east concluded that rich burials of the municipal elite are still unknown.67

5

b o n ds o f f rie n dshi p a n d p atro n age

Having evaluated the funerary inscriptions, we now focus on the votive inscriptions and a small set of dedications or honorary inscriptions.We will use this evidence, in particular that of the honorary inscriptions, to discuss another type of relationship that tied the municipal elite to the countryside – namely, bonds of friendship and patronage. Regarding the religious domain, our survey has resulted in seven votive inscriptions and one building inscription made by members of the municipal elite of the Agrippinenses and two votive inscriptions by those of the Traianenses; there are no such inscriptions for the Tungri and Nervii. The findspots and the gods involved allow us to draw conclusions about the precise ways in which magistrates or councillors were involved in religious matters within their community. A building inscription by an aedilis from Cologne (Appendix 1, no. 7) testifies to his duty as officeholder to monitor the building of publicly financed projects, in this case a temple for Apollo in the colony’s urban centre. Direct involvement by a member of the ruling elite in the upkeep of the religious infrastructure is evident in a votive inscription on a small column from Xanten. We are informed that a decurio paid for the restoration of a monument for Jupiter the Best and Greatest. The column, three feet in height, may have served as a support for a statue; whether the restitutio just concerned the statue and/ or its base or also involved work on a temple has to remain unresolved. In addition to these interventions in the urban centres of both colonies, magistrates and councillors were often also involved in financing public buildings in the countryside. An example is provided by the votive inscription from Heerlen by a decurio from CVT who financed the repairs of the public baths in the vicus of Coriovallum. Further, the four dedications by a former duumvir, a quaestor and two decurions of CCAA in the sanctuary of the Aufaniae at Bonn underline the fact that the cult of the Aufaniae, and in particular the one near the legionary base at Bonn, featured among the public cults of the Agrippinenses.68 In all these cases, the sanctuaries and cults involved took advantage of support from local authorities. Finally, three inscriptions may tell us more about the ruling elite’s personal attachments to particular cults. In the case of the

66

Vanderhoeven 1996, 222 f.

67

Loridant/Deru 2009, 119 f.



68

Derks 1998, 119 ff, esp. 130; Scheid 1999, 402 ff, esp. 411 and 414.

0

10 m

Fig. 4. Simplified plan of the villa of Valkenburg-Ravensbos. Shaded area: probable layout of the original plan in the villa’s first phase; a bath suite and a room with hypocaust were added at a later stage. The symbols mark the approximate findspots of the inscriptions (after Remouchamps 1925, fig. 41).

votive altar to Liber Pater and Hercules by the Agrippinensian decurio M.Vannius Adiutor, it is uncertain – given the urban context of the dedication – whether the altar was set up in his capacity as decurio or as a private dedication. Two votive inscriptions from Blankenheim and Gereonsweiler dedicated to the local goddesses Titaca and the matronae Berhviahenae tell a clearer story. The fact that members of the municipal aristocracy turned to these otherwise barely known goddesses69 may be an indication that they came from the immediate vicinity and were familiar with these cults of old. They probably owned an estate close to the sanctuary. The sample of votive inscriptions discussed above illustrates the connections in the religious domain between the ruling elite and the countryside over which they ruled. Whereas in some cases the inscriptions reveal the obligations that went with holding an office, others – especially those to highly local deities – may reveal a personal attachment of the dedicating magistrate or decurio to the cult in question and may point to properties of the municipal elite in the vicinity. To conclude our survey, we will turn our attention to the four remaining texts inscribed on three small bronze tablets which were found during excavations in the villa of Ravensbos near Valkenburg on the presumed southwestern boundary between CVT and the civitas Tungrorum (cf. fig. 3).70 As may be deduced from the plan published in the excavation report (fig. 4), the villa was originally laid out as a simple building of the corridor type with risaliths to which, in secondary instance, a wing with baths was added on the short south west side and a room with a hypocaust on the long south side. Even after these enlargements the villa was still of modest dimensions. As far as the tablets are concerned, two (tablets 1 and 2) were found in the villa’s main hall (room 20) and the third (tablet 3) close to the steps leading to 69

70

Ti[ta]ca may also be mentioned in an inscription from

45 (map), 56 (using the inscriptions); Brulet 2008, 53 and

Koblenz: CIL XIII 7624.

fig. 62 (map) (without argument). Cf. however, Bridger

The river Geul is now generally accepted as the natural

2008, 607 and Abb. 426 (map), who assumes that the

boundary between the Colonia Ulpia Traiana and the

territory of the CVT extended as far south as Maastricht

Municipium Tungrorum. Cf. Raepsaet-Charlier 1994,

(without argument).

11

0

5 cm

Fig. 5. Bronze tablet from the Roman villa of Valkenburg-Ravensbos (tablet 1). Front with inscription mentioning M.Vitalinius, decurio and duumvir quinquennalis of the Colonia Ulpia Traiana.The token of honour was offered by a certain Iulius (Photo author).

the portico at the front of the building. On the basis of their findspot and content, I assume that each of these inscriptions was once on display in the villa’s portico or central reception hall.71 The four texts belong to the rather rare category of inscriptions that document a patron-client relationship.72 In this respect, they are closely related to the tabulae patronatus, the bronze tablets with municipal decrees proclaiming the initiation of a patron-client relationship between the civic community and an individual.73 But unlike those tablets, the patron-client relationships embodied by our texts lack 71

72

After the editio princeps (Remouchamps 1925, 59-64),

tion). Another singular piece of evidence pointing to

the inscriptions have never been re-examined. Given

informal patronage at the lowest levels of society may be

their importance, it has been decided to present a basic

found in the inscription of a floor mosaic in the entrance

description and a renewed reading, based on autopsy, in

hall of the villa of Laufenburg: Germania 24, 1940, 35;

the appendices to this paper.

Rothkegel 1994, 42-44, Abb. 31; Nesselhauf/Lieb 1959, no. 94 = AE 1998, 987.

Cf. Remouchamps 1925, 62, with note 2; Bogaers 1981, 55; Slofstra 1983, 93 f. (with many errors in the transcrip-

1

73

Nicols 1980.

0

5 cm

Fig. 6. Bronze tablet from the Roman villa of Valkenburg-Ravensbos (tablet 1). Back with inscription mentioning T. Tertinius [Cornutus?], decurio, former aedilis, and present duumvir quinquennalis (?) of the Colonia Ulpia Traiana. Dedicated by the Pagus Catual[ensis?] to its best patron (Photo author).

a high political profile and so the size and shape of the tablets, as well as the style of the texts, have been adapted accordingly. Whereas one of the client parties is a community, this is a pagus instead of the entire city; the other three clients are private individuals. If the initiation of the relationship with the pagus presupposes some sort of official collective decision, the latter three concern personal and informal bonds of patronage between individuals. The first of the three tablets bears an inscription on both sides. The best preserved and oldest text is on what may therefore be called the front (fig. 5) and refers to a gift from a certain Iulius74 to his ‘best friend’, by the name of Marcus Vitalinius (the man’s cognomen is missing). Vitalinius was a decurio and 74

Iulius is used here as single name. See the commentary in Appendix 5.



0

5 cm

Fig. 7. Bronze tablet from the Roman villa of Valkenburg-Ravensbos (tablet 2). Dedicated by two peregrini in honour of T. Tertinius Cornutus?, decurio and duumvir of the Colonia Ulpia Traiana (Photo Municipality of Maastricht).

former quaestor of the Colonia Ulpia Traiana and received the honour of the gift while acting as duumvir with censorial power.We learn nothing about Iulius, but his single name suggests that he was a peregrine. Although the bond between the two men was couched in the language of equality as expressed by the stereotypic address to a ‘best friend’ (amicus optimus), the difference in their personal legal status suggests that in reality their relationship must have been hierarchical, with Iulius the subservient client of his socially superior friend and patron.75

75

Amici is just as likely to refer to vertical and horizontal

status need not necessarily be devoid of the human

relationships of friendship, and thus may denote either

emotions of true friendship. Cf. the useful review by

social equality or inequality. Cf. Caldelli 2001; Panciera

D. Konstan (http://bmcr.brynmawr.edu/2002/2002-04-

2001, 15, with examples of unequal relations between

29.html) and now Verboven 2011.

amici. Of course, relations between people of unequal



0

5 cm

Fig. 8. Bronze tablet from the Roman villa of Valkenburg-Ravensbos (tablet 3). The inscription probably honoured the same T. Tertinius Cornutus?, who is mentioned in tablets 1 and 2. The tablet was offered by a private individual as a token of personal friendship (Photo Municipality of Maastricht).

The other three inscriptions all refer to one and the same man, a certain T. Tertinius (again the cognomen is largely missing, but can perhaps be restored as Cornutus).76 Like M. Vitalinius, T. Tertinius was a decurio of the Colonia Ulpia Traiana; his municipal career had in any event brought him to the offices of aedilis and duumvir. The first of the three texts dedicated to him consists of a punched inscription on the back of the tablet we just have been discussing (fig. 6). Since the letters of this inscription have left their mark on the other side, thereby robbing it of some of its original splendour, this text must have been the younger one, made after the inscription for M.Vitalinius on the front had lost its significance.77 Tertinius figures here as patron of the pagus Catual[ensis?].

76

See Appendix 5.

ger ist, geht daraus hervor, daß die eingepunzten Punkte

77

Finke 1927, (306a): ‘Daß die Inschrift der Rückseite jün-

auf der Vorderseite erhaben hervortreten.’



The second tablet from the Ravensbos bearing the second inscription for Tertinius concerns an explicit token of honour (fig. 7). The name or names of the donor(s) are not completely preserved and reconstruction of the text has to remain conjectural. Most commentators have read two single names – Mansuetus and Macrinus. Unless we suppose that their gentilicia have gone unmentioned and the single names functioned as cognomina, both men must have been peregrini without citizenship. This suggests that here too we might be dealing with a relationship of social inequality. That proposition is confirmed by the use of the word HONORI in the opening line of the inscription, which denotes a position of inferiority on the part of the dedicator. Even if the text is not explicit on this point, a patron-client relationship seems very plausible.78 Finally, the third text on the third tablet is too fragmentary to add anything new (fig. 8). We may be happy to establish that the recipient of the tablet was in all probability the same T. Tertinius, as seems to be suggested by the beginning of the first line which mentions the name of the dedicatee. We cannot determine which aspects of his municipal career were mentioned. The donor’s name, Flor[---], has a greater chance of being restored as the nomen Florentius followed by a cognomen. What can we infer from these texts beyond the simple fact that they point to the presence of two councillors in a villa in the extreme southwest of the colony’s territory? Let me state first that M. Vitalinius and T. Tertinius are likely to have been successive proprietors of the villa. We cannot tell what the relationship between the two patrons was and how the property was transmitted from Vitalinius to Tertinius – by sale, exchange or marriage, or by splitting, bequest or inheritance.79 An issue that has elicited frequent comment is the distance between the findspot of the tablets in a villa some 100 km away from the political centre at Xanten, where Vitalinius and Tertinius held office (cf. fig. 3).80 The underlying assumptions are apparently that the addressees of the inscriptions had received the tokens of honour in the town, that their residence in the Ravensbos was simply too far away to be visited while they held office and that the property could therefore only have been acquired after completion of their term of office – in other words, the tablets had only arrived at the Ravensbos in secondary instance as a result of the addressees moving house. While it seems logical that the tablets were offered when the magistrates were still at their post,81 we needn’t assume that the sole place where this could happen was their urban domus. It is much more plausible that receptions and audiences took place in both the rural residences of the ruling elite and their town houses.82 The tablets then testify to the reception of clients in the villa of the Ravensbos itself. Chance has it that the reverse is also visible from another text in our collection, the bronze tabula ansata from Bavay, which the Nervian duumvir Tib. Iulius Tiberinus received from his lictor Q. Titius Vilicus. Although the text is less explicit than that of our tablets (it only has the name of the addressee in the dative and the name of the donor in the nominative with the added appositive lictor), I can hardly see a better explanation than to view this as a token of friendship between two men who had come to know each other very well through their daily, routine contacts in the administration of the Nervian capital in their respective positions as supreme office-holder and lictor.83 We have no information

78

Cf. also a bronze tablet from Verona (CIL V 3401), 

structed after the cognomen of the father. If M.Vitalinius’

dedicated in honour of (honori) a local magistrate by the

cognomen is restored as Tertius, theoretically T. Tertinius

apparitores et limocincti tribunalis eius, i.e. by the civic servants that had to assist the magistrate with the execution

might thus have been his son! 80

precht 1975, 207.

of his office. On these public servants, Weiß 2004; David 79

Cf. Remouchamps 1925, 65; Finke 1927, no. 306; Rup-

2008.

81

Contra Rupprecht 1975, 207.

Even if their gentilicia differ, a family relationship 

82

Cf. also Roymans/Derks, this volume, Section 6.

between the two men and thus inheritance as the form

83

of transmission cannot be completely ruled out given the frequent use in the area of patronymic nomina con-



Cf., for a similar case, see the collective dedication from Verona cited in note 78.

Fig. 9. Distribution of Roman villas (symbol A) in the area just north of present-day Valkenburg in the Netherlands. Striking is the location of the villa of Valkenburg-Ravensbos (symbol B) close to the valley bottom of a tributary of the river Geul, the east-west oriented valley of which is just visible at the bottom of the map. Data and map courtesy of K. Jeneson.

as to the precise findspot of the tablet, but a very attractive suggestion is that it stems from the urban domus of the duumvir, which would make it a nice counterpiece to the tablets from the Ravensbos. We may conclude from this that three of our texts were offered by private individuals from the area around the villa and were presented at receptions during one of the temporary stays of the office-holder in the Ravensbos. While in those instances both parties may have been satisfied to display the token of gratitude that declared their mutual friendship in the private context of the villa’s portico or reception hall, it is difficult to see that this worked in the same way for the pagus too. If the pagus Catual[ensis?] wanted to honour T. Tertinius as its benefactor, we would normally expect the honour to have been bestowed not during a closed meeting in the patron’s private villa, but rather in one of the imposing spaces the pagus had at its disposal. How then do we explain this find in the hall of the villa? It is my guess that the decree of the pagus posited the production of two texts, firstly a comprehensive one destined for public display at a much-frequented spot (locus celeberrimus), either in the urban centre of the colony or in a vicus in the territory of the pagus, and secondly, an abbreviated copy which often just mentioned the cooptation into the ordo, handed out to the patron for display in his private villa. In cases of official appointments of civic patroni this at least was standard practice.84 A delegation appointed by and often 84

Nicols 1980; Panciera 2001. For Gaul, cf. CIL XIII 921,

rensium and the Aureliani, to the provincial governor of

concerning three tabulae patronatus from the late 4th

Maxima Senonia. The bronze tablets were found in his

century offered by the civitas Senonum, civitas Autissiodu-

villa near Agen in Aquitania (Juliot 1898, 30-35).



elected from the ordo decurionum went to visit the intended patron with the request to receive the city in his client network (accipere in clientelam suam) and to accept a copy of the municipal decree as testimony of his patronship to be put on display in his home.85 If we accept this model, the pagus Catual[ensis?] must have sent envoys to the villa of its intended patron in the Ravensbos. Although we do not know where exactly the pagus was situated, an attractive assumption is that the villa was within, or not too far beyond, its boundaries.86 The pagus would then have had good reason to visit Tertinius in his rural property rather than his urban domus. And since men of standing and fortune such as Tertinius will constantly have been sought for support and help, emotional ties with the petitioning community might have been decisive in a patron’s decision to accept or decline the request; ancestral roots may then have played an important role. Finally, with the knowledge of town-country dynamics gathered so far, the function of the villa of the Ravensbos may be re-assessed. If we assume that the villa was just one of several houses belonging to its owner, it is debatable that this was also his main residence. First, given the high density of villas in the immediate vicinity (fig. 9), the villa’s estate cannot have been very sizeable. This observation brought Tessa de Groot to the hypothesis that this and other neighbouring villas ‘applied themselves, alongside farming, to the exploitation of gravel pits and limestone quarries, as a result of which a large acreage of farmland was of less importance’.87 I am not convinced that these secondary activities were carried out on such a scale that they made a significant contribution to the income of this and other villas in the vicinity. Gravel extraction and quarrying, in my opinion, never went beyond the needs of the own farm and never became a real trade. But the whole hypothesis becomes superfluous if we abandon the idea – widespread among archaeologists – that villa estates were composed of continuous blocks of farmland with the villa neatly nestled at its centre. Through sales and purchases, inheritances, bequests and divisions, villa estates may have varied markedly in size. Villa landscapes may have been much more fragmented and dynamic than archaeologists tend to suppose and in the 2nd and 3rd centuries, scattered properties may have been the rule rather than the exception.88 Second, the villa of the Ravensbos is situated almost on the valley bottom of a small stream, a tributary of the river Geul (fig. 9). This is an unusual location for a Roman villa, not one that we would expect if agrarian exploitation was its main goal.89 If the estate had no other farmland at some distance from the villa (something we cannot of course know), we could entertain the possibility that the primary importance of the property for its proprietor resided not so much in its economic revenues (which were certainly welcome), but rather in its function as a pleasant and comfortable residence in the countryside. To conclude this section, I want to briefly present another villa in the territory of the Traianenses. In contrast to the one from the Ravensbos, the villa of Maasbracht has an excellent pedigree for having been the main residence of its office-holding owner. 85

86

Cf. CIL VI 1492: … tabula / hospitali incisa hoc decreto in

87

De Groot 2005, 18: ‘…zich naast de landbouw ook toe-

domo / sua posita permittat …

legden op de exploitatie van grindkuilen en kalksteen-

On the basis of the place name Catualium documented

groeven, waardoor een groot akkerbouwareaal minder

on the Tabula Peutingeriana as one of the stations on the road between Aduatuca Tungrorum and Novioma-

van belang was.’ 88

However, the different properties need not to be as far

gus, the pagus must at least partly be situated on the left

apart as the examples from the literary sources suggest.

bank of the Meuse. The question is whether it stretched

Cf., for example, Ausonius’ properties: urban house in

beyond the river, and if so, how far. On the basis of the

Bordeaux, a villa near Bordeaux, a villa near Cognac

inscriptions from the Ravensbos, the river Geul is gener-

(= fundus Lucanus or Lucaniacus) an estate at Bazas, land

ally accepted as the southern boundary of the colony’s

at Rom near Poitiers and at Buch near Sauveterre-deGuyenne. Aubert 1994, 142, note 80.

territory (Bogaers 1972, 310; Raepsaet-Charlier 1994, 45 (map). If this is right, the pagus’ extreme limit in southeastern direction can not have been beyond that river.

89

An archaeological indication is the presence of just one secondary building, measuring 16x8 m. Cf. Remouchamps 1925, fig. 40.



As is well-known, public service was not without its obligations. All magistrates and decurions were expected to take on certain munera, ranging from financing the construction or repair of a public building to the organisation of games at one of the many religious festivals. The proper fulfilment of such duties could subsequently be exploited for private prestige and status by being depicted in carefully selected floor mosaics or wall paintings in the magistrate’s private house.The excavations of the villa of Maasbracht in the early 1980s produced a very nice example of this.90 In the villa’s cellar, a late 2nd-century addition to the building, a large quantity of painted plaster fragments were found which must have originated from the villa’s main room situated just above. Like most wall decorations from villas in the area, the fragments could be shown to have belonged to a system of panel decoration which divided the wall into three horizontal zones. One of the black rectangular panels of the central zone had a nearly life-size figure of a bestiarius identifiable by the typical puttees wrapped around his lower leg and a heavily clothed upper arm. It has been suggested that what is represented here is the proprietor’s munus of gladiatorial games which he most likely organised in the amphitheatre of the colony at Xanten.91 Depicted in the upper register of the wall are two standing men in Gallic costume fashioned like Roman dress, one with a purse in his right hand, a reference to the family’s wealth, and a seated figure recording notes or accounts on what looks like a wooden tablet. If the latter scene can be taken as a direct reference to the villa-owner’s bookkeeping, following the legal definition of what constitutes a domus (‘wherever one had his seat and bookkeeping and the organisation of his property’),92 this may provide an additional argument for the proposition that this villa was the proprietor’s main residence. To conclude, if we compare the two villas of Ravensbos and Maasbracht, they are not dissimilar: both may be identified as properties belonging to members of the ruling elite, albeit on the basis of different source material. Both show how the political elite of the town had close ties with segments of the population in the countryside, perhaps most intensely with those in their own district. Both examples also demonstrate that the villas of the municipal elite of the Traianenses, which hardly qualify for the category of medium-sized villas, cannot stand comparison with those known in other regions of Gaul. We do not know the urban domus these magistrates had in Xanten, but if the investments for it had to be earned through arable farming, perhaps we needn’t be surprised that the present knowledge of the colony’s town houses has not yet revealed the same luxury and wealth seen in other civitas capitals in Gaul.

6 co n clusio n s To sum up, I hope to have demonstrated that epigraphic research can make a valuable contribution to the social interpretation of villa landscapes. While these sources certainly have their own problems, they also allow us to nuance some of the straightforward archaeological interpretations that have sometimes been presented in the past. I want to conclude by briefly summarising the conclusions that we reached, however banal they may at times seem. 1) For the ruling elite, the foundation of towns and the introduction of a new form of self-governance meant that they settled in the new towns. 2) There was flexibility regarding the property criteria for entry to the council and as a consequence regarding council size. There was also room for accommodating local traditions concerning the size of the council.

90

Unfortunately, the excavation has so far not been pub-

91

Swinkels 1987.

lished. For preliminary reports, Willems 1982; Van Dier-

92

Dig. 50.16.203: ubi quisque sedes et tabulas haberet sua-

endonck/Swinkels/Willems 1987.

rumque rerum constitutionem fecisset.



3) The administrative elite’s possessions in the countryside could include several estates and one or more houses (Ravensbosch); the largest investments will have been made in the main residence, either in the town or the country (Maasbracht). 4) It is conceivable that in poorer regions, such as the civitates of the Cananefates, the Batavians and to a certain degree also that of the Traianenses, some town councillors did not own a home in the country but made do with a house in town. 5) The use of the umbrella term ‘ruling elite’ suggests that the magistrates and councillors who made up this body of politically engaged aristocracy were a closed homogeneous group. We may speak of a community of peers who shared common values, ideals and social practices.93 We have also hinted at differences in terms of hierarchy, seniority, political influence, wealth and status. 6) Most decuriones probably arranged to be buried on an estate in the country and not in town. For the Agrippinenses, however, the available material suggests the reverse. 7) Councillors and magistrates were a unifying link between town and country. Fortunes earned in the countryside were invested in town via munera such as funding the construction or maintenance of public buildings or games and by building private town houses.

a b b re v iatio n s AE L’Année épigraphique BRGK Berichte der Römisch-Germanischen Kommission CIL Corpus Inscriptionum Latinarum CJ Codex Justiniani Dig. Digesta Galsterer, B. and H. 2010: Die römischen Steininschriften aus Köln, Mainz (Kölner Forschungen IKöln2  10). JRA Journal of Roman Archaeology JRS Journal of Roman Studies RE Pauly-Wissowa’s Realencyclopädie

re f ere n ces Aubert, J.J., 1994: Business managers in ancient Rome. A social and economic study of institores, 200 B.C. – A.D. 250, Leiden (Columbia studies in the classical tradition 21). Balty, J.Ch. 1991: Curia ordinis. Recherches d’architecture et d’urbanisme antiques sur les curies provinciales du monde romain, Bruxelles (Académie Royale de Belgique, Mémoires de la Classe des Beaux-Arts, 2nd Ser., vol. 15.2). Bedon, R./R. Chevallier/P. Pinon, 1988: Architecture et urbanisme en Gaule romaine (52 av. J.-C. - 486 ap. J.-C.), Paris (2 vols). Bodel, J., 1997: Monumental villas and villa monuments, JRA 10, 5-35. Bogaers, J.E., 1972: Civitates und Civitas Hauptorte in der nördlichen Germania inferior, Bonner Jahrbücher 172, 310-333. Bogaers, J.E., 1981: Dertien letters op een bronzen plaat, Brabants heem 33, 50-57.

93

Cf. Crowley, this volume.



Bridger, C., 2008: Die civitas Traianensis – das römische Umland von Xanten, in M. Müller/H.-J. Schalles/N. Zieling (eds), Colonia Ulpia Traiana. Xanten und sein Umland in römischer Zeit, Mainz (Xantener Berichte Sonderband, Geschichte der Stadt Xanten 1), 607-626. Brulet, R., 1996: La maison urbaine en Gaule Belgique et en Germanie Inférieure, in La maison urbaine d’époque romaine en Gaule Narbonnaise et dans les provinces voisines. Actes du colloque d’Avignon 11-13 novembre 1994 = Documents d’archéologie Vauclusienne 6, 73-97. Brulet, R. (ed.), 2008: Les Romains en Wallonie, Bruxelles. Buijtendorp, T.M., 2010: Vergeten stad van Hadrianus. Ontwikkeling, uiterlijk en betekenis van het ‘Nederlands Pompeji’, Amsterdam (unpublished PhD thesis, VU University Amsterdam). Burnand,Y., 2006: Primores Galliarum. Sénateurs et chevaliers romains originaires de Gaule de la fin de la République au IIIe siècle, vol. 2. Prosopographie, Bruxelles (Collection Latomus 302). Caldelli, M.L., 2001: Amicus/-a nelle iscrizioni di Roma. L’apporto dell’epigrafia al chiarimento di un sentimento sociale, in M. Peachin (ed.), Aspects of friendship in the Graeco-Roman world. Proceedings of a conference held at the Seminar für Alte Geschichte, Heidelberg, on 10-11 June, 2000, Portsmouth, RI (JRA Supplement 43), 21-29. Chilver, G.E.F./G.B. Townend, 1985: A historical commentary on Tacitus’ histories IV and V, Oxford. David, J.-M., 2008: Les apparitores municipaux, in C. Berrendonner/M. Cébeillac-Gervasoni/L. Lamoine (eds), Le quotidien municipal dans l’occident romain, Clermont-Ferrand, 391-403. De Groot, T., 2005: De Romeinse villa Meersen-Onderste Herkenberg. De resultaten van het waardestellend archeologische onderzoek in 2003 in relatie tot de onderzoeksgeschiedenis en landschappelijke context van het villacomplex, Amersfoort (Rapportage Archologische Monumentenzorg 125). Demougin, S., 1988: L’ordre équestre sous les Julio-Claudiens, Rome (Collection de l’EFR 108). Derks, T., 1998: Gods, temples and ritual practices.The transformation of religious ideas and values in Roman Gaul, Amsterdam (Amsterdam Archaeological Studies 2). Duncan-Jones, R., 1974: The economy of the Roman empire. Quantative studies, Cambridge. Dondin-Payre, M., 1999: Magistratures et administration municipale dans les Trois Gaules, in M. DondinPayre/M.-Th. Raepsaet-Charlier (eds), Cités, municipes, colonies. Les processus de municipalisation en Gaule et en Germanie sous le Haut Empire romain, Paris, 127-230. Dondin-Payre, M., 2003: Le quotidien institutionel des cités dans les Trois Gaules, Cahiers du Centre G.Glotz 14, 143-153. Eck, W., 2004: Köln in römischer Zeit. Geschichte einer Stadt im Rahmen des Imperium Romanum, Köln (Geschichte der Stadt Köln 1), 315-338. Eck,W., 2008: Die Gründung der Colonia Ulpia Traiana in ihrem politischen Kontext, in M. Müller/H.-J. Schalles/N. Zieling (eds), Colonia Ulpia Traiana. Xanten und sein Umland in römischer Zeit, Mainz (Xantener Berichte Sonderband, Geschichte der Stadt Xanten 1), 243-255. Février, P.A., 1981: Villes et campagnes des Gaules sous l’Empire, Ktema 6, 359-372. Finke, H., 1927: Neue Inschriften, BRGK 17, 1*-2* and 1-231. Finley, M., 1977: The ancient city from Fustel de Coulanges to Max Weber and beyond, Comparative studies in society and history 19, 305-327. Galsterer, H., 1999: Kolonisation im Rheinland, in M. Dondin-Payre/M.-Th. Raepsaet-Charlier (eds), Cités, municipes, colonies. Les processus de municipalisation en Gaule et en Germanie sous le Haut Empire romain, Paris, 251-269. Galsterer, H., 2010: Wie funktioniert eine römische Stadt? Die administrative Infrastruktur römischer Gemeinden, Kölner Jahrbuch 43, 257-265. González, J./M.H. Crawford, 1986: The Lex Irnitana. A new copy of the Flavian municipal law, JRS 76, 147-243. Goodman, P.J. 2007: The Roman city and its periphery. From Rome to Gaul, London.



Habermehl, D., 2011: Settling in a changing world. Villa development in the northern provinces of the Roman empire, Amsterdam (unpublished PhD dissertation, VU University Amsterdam). Hanoune, R. (ed.), 2007: Les villes romaines du Nord de la Gaule, Lille (Revue du Nord, Hors série, Collection art et archéologie 10). Hirth, H.J., 1985: Horaz, der Dichter der Briefe: rus und urbs. Die Valenz der Briefform am Beispiel der ersten Epistel an Maecenas, Hildesheim (Altertumswissenschaftliche Texte und Studien 13). Juliot, G., 1898: Inscriptions et monuments du Musée gallo-romain de Sens. Descriptions et interprétations, Sens. Kienzle, P., 2008: Die zivile Wohnbebauung in der CUT, in M. Müller/H.-J. Schalles/N. Zieling (eds), Colonia Ulpia Traiana. Xanten und sein Umland in römischer Zeit, Mainz (Xantener Berichte Sonderband, Geschichte der Stadt Xanten 1), 413-432. Koster, A., 2010: Het grafveld van Noviomagus en de rijke graven van de stedelijke elite, Nijmegen (unpublished PhD dissertation Radboud University Nijmegen). Laes, C., 2004: Children and office holding in the Roman empire, Epigraphica 66, 145-184. Lamoine, L., 2009: Le pouvoir local en Gaule romaine, Clermont-Ferrand. Laur-Belart, R., 1937: Führer durch Augusta Raurica, Basel. Loridant, F./X. Deru (eds), 2009: Bavay. La nécropole gallo-romaine de la « Fache des Près Aulnoys », Villeneuve-d’Asq (Revue du Nord, Hors série, Collection Art et archéologie 13). Mattusch, C.C., 2005: The “Villa dei Papiri” at Herculaneum. Life and afterlife of a sculpture collection, Los Angeles. Mouritsen, H., 1998: The album of Canusium and the councils of Roman Italy, Chiron 28, 229-254. Mouritsen, H., 2011: The freedman in the Roman world, Cambridge. Nesselhauf, H./H. Lieb, 1959: Dritter Nachtrag zu CIL XIII. Inschriften aus den germanischen Provinzen und dem Treverer­gebiet, BRGK 40, 120-228. Nicolet, C., 1976: Le cens sénatorial sous la République et sous Auguste, JRS 66, 20-38. Nicols, J., 1980: Tabulae patronatus. A study of the agreement between patron and client-community, ANRW II.13, 535-561. Nicols, J., 1988: Is there a standard size of the ordo decurionum, Zeitschrift der Savigny-Stiftung für Rechtsgeschichte, Romanistische Abteilung 105, 712-719. Panciera, S., 2001: Onorare l’amico nella sua casa. Amicitia e topografia a Roma e nel suo suburbio, in M. Peachin (ed.), Aspects of friendship in the Graeco-Roman world. Proceedings of a conference held at the Seminar für Alte Geschichte, Heidelberg, on 10-11 June, 2000, Portsmouth, RI (JRA Supplement 43), 11-19. Parma, A., 2003: Per un nuovo corpus dei decreta decurionum delle città romane d’Italia e delle province occidentali, Cahiers du Centre G.Glotz 14, 167-171. Raepsaet-Charlier, M.-Th., 1994: La cité des Tongres sous le Haut-Empire. Problèmes de géographie historique, Bonner Jahrbücher 194, 43-59. Raepsaet-Charlier, M.Th., 1998: Les Gaules et les Germanies, in C. Lepelley (ed.), Rome et l’intégration de l’empire, 44 av. J.-C. – 260 apr. J.-C., Paris, 143-195. Raepsaet-Charlier, M.-Th., 1999: Les institutions municipales dans les Germanies sous le Haut Empire, bilan et questions, in M. Dondin-Payre/M.-Th. Raepsaet-Charlier (eds), Cités, municipes, colonies. Les processus de municipalisation en Gaule et en Germanie sous le Haut Empire romain, Paris, 271-352. Remouchamps, A.E., 1925: Opgraving van een Romeinsche villa in het Ravensbosch (L.), Oudheidkundige Mededelingen van het Rijksmuseum van Oudheden te Leiden 6, 40-77. Rothkegel, R., 1994: Der römische Gutshof von Laufenburg/Baden, Stuttgart (Forschungen und Berichte zur Vor- und Frühgeschichte in Baden-Württemberg 43). Roymans, N., 1990: Tribal societies in Northern Gaul. An anthropologi­cal perspective, Amsterdam (Cingula 12). Roymans, N., 1996: The sword or the plough. Regional dynamics in the romanisation of Belgic Gaul and the Rhineland area, in N. Roymans (ed.), From the sword to the plough. Three studies on the earliest romanisation of northern Gaul, Amsterdam (Amsterdam Archaeological Studies 1), 9-126.



Rupprecht, G., 1975: Untersuchungen zum Dekurionenstand in den nordwestlichen Provinzen des römischen Reiches, Kallmünz. Salway, B., 2000: Prefects, patroni and decurions. A new perspective on the album of Canusium, in A.E. Cooley (ed.), The epigraphic landscape of Roman Italy, London, 115-171. Schalles, H.-J., 1995: Überlegungen zur Planung der Colonia Ulpia Traiana und ihrer öffentlichen Bauten im Spiegel städtischer Architektur des 2. Jahrhunderts n.Chr., Xantener Berichte 6, Köln/Bonn, 379-428. Schalles, H.-J., 2001: Die Wirtschaftskraft städtischer Siedlungen am Niederrhein. Zur Frage der wirtschaftlichen Beziehungen des römischen Xanten mit seinem Umland, in Th. Grünewald/H.-J. Schalles (eds), Germania inferior. Besiedlung, Gesellschaft und Wirtschaft an der Grenze der römisch-germanischen Welt, Berlin (Ergänzungsbände zum Reallexikon des Germanischen Altertums 28), 431-463. Scheid, J., 1999: Aspects religieux de la municipalisation. Quelques réflexions générales, in M. DondinPayre/M.-Th. Raepsaet-Charlier (eds), Cités, municipes, colonies. Les processus de municipalisation en Gaule et en Germanie sous le Haut-Empire romain, Paris, 381-423. Sherk, R.K., 1970: The municipal decrees of the Roman west, Buffalo, NY. Sherwin-White, A.N., 1985: The letters of Pliny. A historical and social commentary, Oxford. Slofstra, J., 1983: An anthropological approach to the study of Romaniza­tion processes, in R.W. Brandt/J. Slofstra (eds), Roman and native in the Low Countries. Spheres of interacti­on, Oxford (BAR International Series 184), 71-104. Stuart, P./J.E. Bogaers, 2001: Nehalennia. Römische Steindenkmäler aus der Oosterschelde bei Colijnsplaat, Leiden (Collections of the National Museum of Antiquities 11 = CSIR Nederland 2). Swinkels, L.J.F., 1987: A gladiatorum munus depicted in a Roman villa at Maasbracht, in Pictores per provincias, Avenches (Cahiers d’Archéologie Romande 43), 191-195. Taylor, L.R./R.T. Scott, 1969: Seating space in the Roman senate and the senators pedarii, Transactions of the American Philological Association 100, 529-582. Vanderhoeven, A., 1996: The earliest urbanisation in northern Gaul. Some implications of recent research in Tongres, in N. Roymans (ed.), From the sword to the plough. Three studies on the earliest romanisation of northern Gaul, Amsterdam (Amsterdam archaeological studies 1), 189-260. Vanderhoeven,A/G.Vynckier/P.Vynckier, 1992: Het oudheidkundig bodemonderzoek aan de Hondsstraat te Tongeren (prov. Limburg). Interimverslag 1989, Archeologie in Vlaanderen 2, 65-88. Van Dierendonck, R.M./L.J.F. Swinkels/W.J.H. Willems, 1987: Rijke hereboeren uit Maasbricht, in P. Stuart/M.E.Th. de Grooth (eds), Langs de weg, Heerlen/Maastricht, 62-67. Verboven, K., 2011: Friendship among the Romans, in M. Peachin (ed.), The Oxford handbook of social relations in the Roman world, Oxford, 404-421. Vipard, P., 2007: Maison à péristyle et élites urbaines en Gaule sous l’empire, Gallia 64, 227-277. Weiß, A., 2004: Sklave der Stadt. Untersuchungen zur öffentlichen Sklaverei in den Städten des römischen Reiches, Stuttgart (Historia Einzelschriften 173). Wesch-Klein, G., 1993: Funus publicum. Eine Studie zur öffentlichen Beisetzung und Gewährung von Ehrengräbern in Rom und den Westprovinzen, Stuttgart (HABES 14). Wesch-Klein, G., 1998: Soziale Aspekte des römischen Heerwesens in der Kaiserzeit, Stuttgart (HABES 28). Whittaker, C.R., 1990: The consumer city revisited. The vicus and the city, JRA 3, 110-117. White, P., 2010: Cicero in letters. Epistolary relations of the Late Republic, Oxford. Wightman, E.M., 1985: Gallia Belgica, London. Willems, W.J.H., 1982: Belangrijke recente vondsten in Limburg, Archeologie in Limburg 15, 14-17. Wolff, H., 1976: Kriterien für latinische und römische Städte in Gallien und Germanien und die ‘Verfassung’ der gallischen Stammesge­mein­den, Bonner Jahrbücher 176, 45-120. Woolf, G., 1996: Monumental writing and the expansion of Roman society in the early empire, JRS 86, 22-39.



a p p e n di x 1 . i n scri p tio n s me n tio n i n g mem b ers o f the ruli n g elite o f the agri p p i n e n ses ( b old : f u n erar y i n scri p tio n s ; italicised : v oti v e i n scri p tio n s ) . findspot

name

office

date

reference

1

Cologne

[--- V]ot(uria tribu) [---]

IIvir

IKöln2 286

2

Cologne

-

aedilicius

IKöln2 287

3

Cologne

Constantianus

decurio CC[AA]

CIL XIII 8333 = IKöln2 288

4

Cologne

-

decurio CCAA

IKöln2 289

5

Cologne? (Spellen)

Veranius Victorinus

decurio c(oloniae) Ag(rippinensis)

CIL XIII 8602 = IKöln2 290

6

Cologne

Deccius Fruendus

decurio CCAA

IKöln2 291

7

Cologne

C. Iul(ius) [---]

[IIvir i]ter(um) aedil(icia) p(otestate)

CIL XIII 8165 = IKöln2 6

8

Cologne

M. Vannius Adiutor

decurio

IKöln2 50

9

Bonn

Tib. Claudius [I]ustus

IIviralis

N 162

10

Bonn

C. Candidinius Verus

decurio CCAA

11

Bonn

T. Macrinius Titianus

decurio CCAA

235

N 164

12

Bonn

Q. Vettius Severus

quaestor CCAA

164

N 165

13

Nettersheim

-

decurio CCAA

14

Zülpich-Hoven

Masclinius Maternus

decurio CA aedilicius duumviralis

15

Blankenheim

Q. Vinius Ursule[---]

decurio CoCAA

CIL XIII 8853

16

Gereonsweiler

Q. Acilius Verus

decurio CCAA

CIL XIII 12013

N 163

CIL XIII 7827 = CSIR Deutschl. III.2, no. 50 352

CIL XIII 7918

a p p e n di x 2 . i n scri p tio n s me n tio n i n g mem b ers o f the ruli n g elite o f the traia n e n ses ( italicised : v oti v e i n scri p tio n s ) . findspot

name

office

reference

1

Xanten

M. R(---) R(---)

decurio CVT

CIL XIII 8617

2

Heerlen

M. Sattonius Iu[cun]dus

decurio CVT

N-L 247

3

Ravensbos

M. Vitalinius [---]

decurio CVT quaestoricius IIvir quinq.

F 306

4

Ravensbos

T. Tertinius [Cor]nu[tus?]

aedilicius [IIvir quinq.?] CVT

F 306

5

Ravensbos

T. Tertinius [Cor]nu[tus?]

decurio IIvir [CVT]

F 307

6

Ravensbos

T. Tertinius [Cor]nu[tus?]

decurio [aedilicius CVT?] IIvir [quinq.?]

F 308



a p p e n di x 3   : i n scri p tio n s me n tio n i n g mem b ers o f the ruli n g elite o f the tu n gri ( b old : f u n erar y i n scri p tio n s ) . findspot

name

office

reference

1

Vaux-les-Cherain

Victorius Caupius

decurio

AE 1921, 66 = ILB 2 60

2

Gors-Opleeuw

C. Gracileius Similis

aedilis

CIL XIII 3599 = ILB 2 21

a p p e n di x 4 . i n scri p tio n s me n tio n i n g mem b ers o f the ruli n g elite o f the n er v ii ( b old : f u n erar y i n scri p tio n ) . findspot

name

office

reference

1

Bavay

Tib. Iul(ius) Tiberinus

IIvir

CIL XIII 3572

2

Saint-Hilaire-sur-Helpe‘Fuchau’

C. Serenius Quartus

IIvir

AE 1997, 1141



a p p e n di x 5 . descri p tio n a n d comme n tar y o n the three ta b lets f rom the v illa o f the ra v e n b os . 94 TA B L E T 1 Depository: Larger part in permanent exhibition at the Römermuseum Xanten (long term loan from RMO, Leiden), two smaller fragments in RMO Leiden Inv.no. RMO, Leiden: l 1932/12.1a (larger part); l 1932/12.1b (two small fragments)

Findspot: Valkenburg-Ravensbos, villa, near east wall of room 20

Bronze tablet which seems to have had the form of a tabula ansata with inscription on both sides. For the text on the other side to be readable, the tablet has to be turned around its horizontal axis. One of the smaller fragments, which bears traces of the letter M, has to be placed at the top left (seen from the front). The other small fragment is plain on both sides and, since it is does not fit to the other fragments, its place within the tablet cannot be determined; it has therefore not been included in the photographs and drawings of figs. 5 and 6. At the end of line 4 on the back, one letter has crumbled off since the tablet’s discovery (compare photos in Van Es 1981, fig. 180 with those in this contribution). Right half is missing; on the long sides as well as on the bottom left and on the top left (i.e. on one of the two small fragments) the original rim has been preserved, whereas it is missing in between, where the ansa may be expected. The tablet’s largest width is at present 12.9 cm, the full height 11.7 cm. The text on the front has been engraved, the other on the back punched; punch marks are visible on the front. Letter height on the front: 10-11 mm; on the back, first line 10-11 mm, third line 10 mm, fourth line 9-11 mm (but first letter P 16 mm), fifth line 7 mm. A plaster cast of the tablet is kept in the depository of the Centre Céramique at Maastricht under inv. no. BC679. Selective literature: Remouchamps 1925, 59-62 with fig. 54-55; AE 1926, 128-129; Finke 1927, 306; Byvanck, A.W., 1935: Excerpta Romana. De bronnen der Romeinsche geschiedenis van Nederland vol. II, ’s-Gravenhage, 83, no. 33; Rupprecht 1975, 207-208; Van Es, W.A., 1981: De Romeinen in Nederland, Haarlem, 236 with fig. 180; Bloemers, J.H.F./L.P. Louwe Kooijmans/H. Sarfatij (eds), 1981: Verleden land. Archeologische opgravingen in Nederland, Amsterdam, 92-93 (with fig.); Bogaers 1981, 55; Bechert,T., 1982: Römisches Germanien zwischen Rhein und Maas. Die Provinz Germania inferior, München, 48, with fig. 42-43 and 119; Bechert, T., 1983: De Romeinen tussen Maas en Rijn, Dieren, 47-48, with fig. 42-43 and 119; Slofstra 1983, 93, no. 2 and 1; Galsterer 1999, 268-269; Brulet 2008, 53; Eck 2008, 251; Bridger 2008, 608, with note 2442.

Front: M · VITALINỊ DEC · C · V · T · Q CIO · IIVIR · Q IVLIVS · AMIC Back : · T · TERTINIO AEDILICIO · C · Ṿ PAGVS · CATVAL NO

M(arco) Vitalinio [---] / dec(urioni) C(oloniae) V(lpiae) T(raianae) q[uaestori]/cio / (duum)vir(o) q[uinq(ennali)] / Iulius amic[o optimo]

T(ito) Tertinio [Cornuto?] / aedilicio [(duum)vir(o)? quinq(uennali)?] / C(oloniae) V(lpiae) [T(raianae) / Pagus Catual[ensis? patro] / no [optimo]

94

I like to thank Ruurd Halbertsma (National Museum

tance and help during the preparation of the manuscript.

of Antiquities, Leiden), Marcus Reuter and Christine

I am grateful to Monique Dondin-Payre (Paris) for

Lincke (Römermuseum Xanten) and Wim Dijkman

discussing with me a first draft of my readings of these

(municipality of Maastricht) for giving me the opportu-

texts. The drawings for publication were made by B.

nity to study the objects in their keeping and for assis-

Brouwenstijn (VU University Amsterdam).



tra n slatio n Front: ‘To Marcus Vitalinius [---], councillor of Colonia Ulpia Traiana, former quaestor, (and present) duumvir quinquennalis. Iulius for his best friend.’ Back: ‘To Titus Tertinius [Cornutus?], former aedile, [active duumvir quinquennalis?] of Colonia Ulpia Traiana.The pagus Catual[ensis?] to its best patron.’

comme n tar y Front 1 Vitalinius: patronymic nomen, derived from the cognomen Vitalis. 2 The only magistracy that can have been mentioned here is the quaestorship; together with the final line and line 4 of the back (see below), this determines the tablet’s original width. 3 Given the available space, quinquennalis can not have been written in full. Most often it is abbreviated to quinq, less frequently occur qq and quinquen. Which of these options is to be preferred, depends on how the dedicator’s name is reconstructed and on whether or not one is prepared to accept a small vacat at the end of the line. 4 Iulius: since this can not have been the nomen (for then the distinctive cognomen would be missing) it functioned either as a single name or as a cognomen. In comparison to Iulius as cognomen, which is rare (cf. CIL XIII 2091, 2112, 4163, 4372, IKöln2 488, and the parallels cited by Finke 1927, 306), Iulius as a single name is not uncommon (e.g.. CIL XIII 278, 474, 914, 1115, 1236, 1841, 2175, 5646, 5785, 11384-5; AE 1978, 495, IKöln2 55, and Kajanto, I., 1982: The Latin cognomina, Helsinki, 61). Should one wish to read Iulius as a cognomen, the nomen must have been a familiar one that was abbreviated to two or three letters, e.g. Ulp(ius) or Aur(elius). More likely, however, is that the dedicator was a peregrine. In the latter case, we have to assume either the longer abbreviation quinquen(alis) or the shorter version with a small vacat at the end of line 3. Amico optimo seems to have been written in full. It seems most unlikely that the tablet mentions the complete municipal career of M. Vitalinius, as has sometimes been suggested (cf. Bechert 1982, 48): Vitalinius had certainly been ordinary duumvir before he was finally elected duumvir quinquennalis. Back 1 For the cognomen, see tablet 2. 2 After the aedileship another magistracy must have been mentioned. From tablets 2 and 3 we know for certain that Tertinius had been duumvir, most probably with censorial power. It therefore seems best to assume that this magistracy was mentioned in the second half of the line. 3 Given the large vacat at the beginning, this line can only have contained the colony’s abbreviated name CVT. 4 The final letter is without doubt an L, as becomes clear from the photograph in Van Es 1981, 235, fig. 180; at present only the serif on the top has been preserved. For the name of the pagus Catual[---] several reconstructions have been suggested: Catual[ium] (Remouchamps 1925, 61), Catual[inus?] (Finke 306; Byvanck 1935, no. 33b; Bogaers 1981, 55; Brulet 2008, 53; Bridger 2008, 607), Catual[iensis?] (AE 1926, 129) and Catual[ensis?] (Eck 2008, 251). Most plausible seems a reconstruction with the suffix –ensis. The location of the pagus is relevant for the reconstruction of the boundary between the CVT and the civitas Tungrorum. Many have sought to establish a relation between the pagus Catual[ensis?] and Catualium, a road station on the Tongres-Nijmegen road mentioned on the Tabula Peutingeriana, and have therefore assumed that the pagus has to be located on the western bank of the river Meuse (cf. Raepsaet-Charlier 1994, 55 f.). If this seems right, the pagus may well have comprised some territory



to the east of the river as well. While the villa of Tertinius need not necessarily to have been situated within the territory of the pagus for him being able to function as its patron, the inscription does make this very likely. The pagus’ most southern boundary may then have been constituted by the river Geul and have coincided with the southern boundary of the CVT. How the pagus got into the possession of the tablet, remains unclear: perhaps Tertinius himself offered the pagus the opportunity to reuse the tablet which he got in his possession when he bought the villa from Vitalinius or when he inherited it at his death.

TA B L E T 2 Depository: Centre Céramique, Maastricht Inv.no.: BC680

Findspot: Valkenburg-Ravensbos, villa, near east wall of room 20

Bronze tabula ansata; right part missing. Since the first publication, a tiny fragment with the letters TI at the end of line 2 and V at the end of line 3, has been lost. Several letters have been engraved so deeply that they got cut through the tablet: this is the case with the letters O and N in line 1, N in line 3, C and V in line 4 and M and V in line 5. Deep engraving also explains why the tablet is often broken precisely on the stroke of a letter.Tablet’s completely preserved height: 10 cm; greatest width, including ansa: 11.8 cm (width ansa c. 3.1 cm); greatest height ansa: c. 6.7 cm. Weight: 35.05 gr. Letter height is for all lines approximately 10 mm. A plaster cast of the tablet is kept in the depository of the Centre Céramique at Maastricht under Inv. no. l 1932/12.2. Selective literature: Remouchamps 1925, 62-63 with fig. 57; AE 1926, 130; Finke 1927, 307; Byvanck, A.W., 1935: Excerpta Romana. De bronnen der Romeinsche geschiedenis van Nederland vol. II, ’s-Gravenhage, 83, no. 34; Rupprecht 1975, 207; Slofstra 1983, 93, no. 3; De Grooth, M.E.Th., 1987:Villa rustica. Het boerenbedrijf in het Rijn/Maasgebied, in P. Stuart/M.E.Th. de Grooth (eds), Langs de weg, Heerlen/Maastricht, 81-86, here p. 83-84 with fig. 6; Galsterer 1999, 268-269; Bridger 2008, 608, with note 2442.

(palm) HONỌ T · TERT NV DEC · II ·VIR MANSVET MACRIN

(palm) Hono[ri (palm)] / T(iti) Tert[ini Cor]/nu[ti?] / dec(urionis) (duum)vir(i) [C(oloniae) V(lpiae) T(raianae)] / Mansuet[us et?] / Macrin[us]

tra n slatio n ‘In honour of Titus Tert[inius Cor?]nu[tus?], councillor, duumvir of Colonia Ulpia Traiana. Mansuet[us and?] Macrinus.’

comme n tar y 2-3 Cornuti: given the width of the tablet, determined with some precision by the first and fourth line, there is room for three letters at the end of line 2. Against the common assumption, NU[---] in line 3 cannot have been the beginning of the cognomen, but is rather the central part of it. One could think of Ve/nulus, Luci/nulus, or Cor/nutus, the latter of which is the most frequent of the three. 4 If we assume that the name of the colony was mentioned somewhere in the text, this can only have been in this line. As a consequence, there seems no room left for the adjective quinquennalis, since even in its most abbreviated form, q.q., it would not fit.



5-6 Most commentators have assumed two single names, although one could read as easily a nomen and cognomen, e.g. Mansuetinius Macrinus. Given the inferior position of the dedicators vis-à-vis the dedicatee as suggested by the opening formula Honori, it might be slightly more plausible to envisage two peregrine men as authors of the text than a single Roman citizen.

TA B L E T 3 Depository: Centre Céramique, Maastricht Inv.no.: BC681

Findspot: Valkenburg-Ravensbos, villa, in front of the building near the stair leading to the portico

Bronze tabula ansata of which the greater part on the right is missing. Complete height 11.4 cm; greatest width, including ansa, 7.6 cm (width of ansa 3.8 cm); height of ansa at juncture with tablet c. 4.2 cm; the ansa is not exactly in the centre of the short side. The bronze is tinny. Weight of the fragment: 18.23 gr. A plaster cast of the tablet is kept in the depository of the Centre Céramique at Maastricht under Inv. no. l 1932/12.3. Selective literature: Remouchamps 1925, 63-64 with fig. 58; AE 1926, 131; Finke 1927, 306; Byvanck, A.W., 1935: Excerpta Romana. De bronnen der Romeinsche geschiedenis van Nederland vol. II, ’s-Gravenhage, 83, no. 35; Rupprecht 1975, 207; Slofstra 1983, 93, no. 4; Galsterer 1999, 268-269; Bridger 2008, 608, with note 2442.

TEṚ[TINIO CORNVTO ?] DEC [AEDILIC ? C V T ? II] VIR [QVINQVENN?] FLOR ΑṂ[ICO OPTIMO]

Tert[inio Cornuto?] / dec(urioni) [aedilic(io)? C(oloniae) V(lpiae) T(raianae) (duum)] / vir(o) [quinquenn(ali)?] / Flor[entius ---us? or Florentiniu ---us?] / am[ico optimo]

tra n slatio n ‘To Tertinius Cornutus?, councillor, [former aedile of Colonia Vlpia Traiana?, duum]vir [quinquennalis?]. Florentius ---us [or Florentinius ---us?] to his best friend’.

comme n tar y 1 Assuming that the dedicatee is identical with the one in tablet 2, his nomen and cognomen can be restored accordingly. 2 At the end of the line, the duumvirate was mentioned, preceded by another magistracy, either the quaestorship or the aedileship; the latter is documented in tablet 1. 3 If Tertinius’ career is listed in ascending order, the duumvirate was mentioned at the end of line 2 and the beginning of this line. Since the ‘ordinary’ duumvirate would leave us with a long vacat, this must have been the duumvirate with censorial power. 4 Given the available space, it seems most plausible that the dedicator had a nomen and cognomen. For the nomen one could think of Florentius or Florentinius.



Ethnic recruitment, returning veterans and the diffusion of Roman culture among rural populations in the Rhineland frontier zone Nico Roymans1 1 2 3 4 5 6 7 8

Introduction Veteran settlement and the epigraphic evidence Roman military equipment in non-military contexts Bronze seal boxes and the spread of Latin writing Auxiliary veterans and Roman-style grave monuments Italian terra sigillata from non-military, rural contexts Veterans and the development of Roman villas Discussion References

1 i n troductio n An old, yet still topical theme of discussion within Roman archaeology is the army’s role as Kulturträger and mediator in the diffusion of Roman cultural forms among indigenous groups in the frontier provinces. For the areas north of the Alps, most studies emphasise the critical role of the army in this dissemination.2 In his recent book on Roman Britain, however, David Mattingly seeks to qualify the army’s importance in this regard. He stresses the ‘essentially self-serving use of material culture within the army, which to a large extent developed a separate version of Roman identity, distinguishing soldiers from civilians rather than uniting them’.3 He points to the marginal impact of Roman military culture on indigenous rural populations in the heavily militarised zones of west and north Britain.4 He bases his conclusion in part on the studies by Simon James and Ian Haynes, who analyse the Roman army as a community with a strongly developed sense of its own identity.5 A major focus in this discussion is the intermediary role of veterans in the spread of Roman cultural forms. Veterans constitute a key group because of their ability to bridge the gap between the military and civilian worlds. Although numerous studies have elaborated on this theme, they have tended to focus rather narrowly on legionary veterans.6 They have done so for pragmatic reasons – legionary veterans 1 

This contribution is a modified version of a paper pub-

4 

Mattingly 2006, 223. We should take into consideration,

lished in the Proceedings of the 13th International Congress of

however, great regional differences, and therefore the

Celtic Studies (= Roymans 2009). Several illustrations and

question arises as to what extent the situation described

the appendix with a list of sites with Italian terra sigillata

by Mattingly is characteristic for other frontier areas (see

have been left out, while a new section on veterans and the development of villas (section 7) has been included. 2 

E.g. Speidel 1995; Von Hesberg 1999; Galsterer 1999.

3 

Mattingly 2006, 167; ibid., 199.

the final dicussion below). 5 

James 1999; Haynes 1999a.

6 

For the Lower Rhine area, see Démougin 1999; Lenz 2006; Bridger 2006.



are much better represented in the epigraphic record than their auxiliary counterparts – but also because of the belief that it was legionary veterans who were the bearers and transmitters of Roman culture, certainly in the early imperial era, and who are therefore most deserving of our attention. Auxiliary veterans are a much neglected group in the study of romanisation processes in the Rhineland frontier zone.7 They merit greater attention because of their dual potential role as transcultural mediators, not only between the military and civilian spheres, but also between Romans and indigenous rural populations. There was of course a sizeable disparity with regard to knowledge of Roman culture between an auxiliary soldier and the average legionary, certainly in the 1st century AD. Unlike auxiliaries, legionaries were Roman citizens from Italy or the heavily romanised provinces of Southern Gaul or Spain. All the same, auxiliaries, too, were profoundly influenced by Roman (military) culture on their discharge from the army. This was reflected inter alia in their equipment, their clothing, their eating and drinking habits, their knowledge of the Latin language and writing, and of the religious and ritual practices within the Roman army. For that reason, and because of their sheer numbers, I wish to contend that we have underestimated the intermediary role of auxiliary soldiers in the romanisation of native populations. In this study I will focus on the behaviour and social impact of auxiliary veterans in the Lower Rhine frontier zone, looking particularly at the 1st century AD. The Roman army set a clear stamp on this region, not only by stationing vast numbers of troops,8 but also by recruiting auxiliaries on a large scale. The practice of conscripting auxiliary troops into closed ethnic units named after their tribe allows us to make a rough estimation of the scale of recruitment in the pre-Flavian period. We know of 33 units in total, corresponding to about 16,500 soldiers.9 Given that these units needed to be constantly supplemented by new recruits, this means that a total of no less than 60,000 men of native origin served in the Roman army in pre-Flavian times.10 Pre-Flavian ethnic recruitment was not evenly distributed across the different tribes, however, and there were undoubtedly changes over time as well. The Batavians (with 10 units) were by far the most important supplier of manpower, followed by the Nervii and Tungri (with 5 units each), and then smaller groups in the hinterland of Cologne and Xanten (the Sunuci, Cugerni, Ubii, Sugambri and Baetasii). Most ethnic units will have started out in the Augustan period as irregular or semi-regular formations, but it was probably not long before they received payment and carried Roman-style weapons.11 For the Batavians, the tradition of recruiting soldiers in fairly homogeneous ethnic groups, with their own commanders, seems to have continued into the 2nd century, but for other groups like the Nervii this practice may have been abandoned shortly after the Augustan era. How can material culture deepen our understanding of the presence of auxiliary veterans in rural populations in the Rhineland frontier zone? I believe that the potential of material culture is far greater than we have supposed. This study elaborates further on this by analysing five different categories of evi7 

Auxilia veterans are central in the studies of Haynes

10 

We assume that three to four generations of soldiers

1999b; idem, 2001; Roxan 2000; Pferdehirt 2002; Derks/

served in the army from the Augustan period onwards,

Roymans 2006.

taking into account the fact that the principle of 25 years

8 

Alföldy (1968, 141) estimates the size of the troops stationed here under Tiberius at 42,000 soldiers, 20,000 of whom were auxiliaries and 22,000 legionaries.

9 

of formal service did not yet exist at this early stage. 11 

An indirect indication for the payment of auxiliaries in the Rhine/Meuse delta is the presence of a horizon of

See Alföldy 1968, especially 141, for a general survey of

Augustan-Tiberian bronze coins in many rural settle-

the pre-Flavian auxiliary units stationed in Germania

ments. See Aarts 2003, 173; Roymans/Aarts 2004, 340;

Inferior and largely originating from the Lower Rhine

cf. also Wolters 1988, 261-272. Also indicative of the

frontier zone and its immediate hinterland. In the pre-

payment of Lower Rhine auxiliary troops is the rapid

Claudian period most of them will have been deployed

replacement in the Augustan period of traditional La

in the Germanic wars, but under Claudius a substantial

Tène swords by Roman gladii or gladius-like swords.

number were transferred to Britannia.



dence: epigraphic material in the form of funerary and votive inscriptions, Roman military equipment from non-military contexts, bronze seal boxes, funerary architecture and Italian terra sigillata. Finally, I will discuss the impact of auxiliary veterans on the development of Roman villas in the Rhineland.

2

v etera n settleme n t a n d the e p igra p hic e v ide n ce

Until recently, archaeological discussion about the possible settlement of veterans has centred almost entirely on epigraphic material. The distribution of inscriptions erected by veterans tells us that most soldiers discharged from the Roman army preferred to settle near the gates of the forts and fortresses of their last army post, as well as in the main towns of the Lower Germanic civitates.12 This is immediately clear if we plot the find spots on a map (fig. 1). The fact that only eleven of the 73 veteran inscriptions listed by Demougin for Lower Germany came from the countryside seems to suggest that few veterans settled there. We should bear in mind, however, that the distribution of monumental veteran inscriptions should be seen as an indicator of the spread of epigraphic culture in general rather than as a measure of the true extent of veteran settlement. In other words, the distribution of monumental inscriptions may be strongly biased towards military and urban centres. No less important is the observation that auxiliary veterans are heavily underrepresented in the inscriptions; only thirteen of the 73 veteran inscriptions can be ascribed to ex-auxiliaries. This latter group appears most elusive in the epigraphic record. A slightly different picture emerges if we take into account the military diplomas (usually surviving only as fragments) issued to retired peregrine soldiers as proof that they had been granted the privileges of citizenship and legitimate marriage. Even though the number of Roman military diplomas from Lower Germany has been surprisingly small to date (only nine or ten examples), the find spots for these diplomas are well-represented in the rural area (fig. 2).13 Apart from one or two from the town of Noviomagus-Nijmegen, one from the Roman colony of Xanten, and one from the army camp of BurginatiumAlt-Kalkar, the remaining six are from the countryside. The most plausible explanation is that these finds were associated with veterans who had returned to their homeland after discharge. Although find spots for military diplomas reveal that auxiliary veterans settled not just in towns and military settlements but also in the countryside, we cannot, on the basis of these documents, arrive at reliable conclusions about the scale of this phenomenon. We are still left with the fundamental problem that soldiers who already had Roman citizenship before beginning military service are not represented in the documents (this relates to all legionaries and, after the end of the 1st century, increasing numbers of auxiliary soldiers). In addition, an unknown proportion of peregrine auxiliary soldiers will not have acquired such a document for financial or other reasons. Even under ideal circumstances – if all military diplomas at one time circulating in Lower Germany had survived – we would still be groping in the dark on this point.

3 roma n militar y e q ui p me n t i n n o n - militar y co n te x ts Given the serious biases associated with the epigraphic evidence for the settlement of auxiliary veterans, we have to explore the potential of other categories of material evidence in order to answer our questions. Significant new insights can be obtained by studying Roman military equipment in non-military contexts. Figure 3 shows the distribution of 1st-century AD Roman gladii or gladius-like swords in Bel-

12 

Derks/Roymans 2006, 124 ff. Demougin 1999, esp. 372.

13 

Derks/Roymans 2006, 126-127.



Fig. 1. Distribution of monumental inscriptions mentioning veterans. After Derks/Roymans 2006, fig. 2. A legionary veterans; B auxiliary veterans; C veterans of unknown type of unit (large symbols indicate 5 or more inscriptions)

Fig. 2. Distribution of Roman military diplomas according to find context. After Derks/Roymans 2006, fig. 3. A urban centre or military settlement; B rural settlement

1

+

army camp rural settlement grave find

+ ++ + + ++ +++ +

river find cult place

Som

Rh

in

e

me

M

M os ell

se

e

eu

0

100 km

Fig. 3. Distribution of 1st century AD Roman gladii or gladius-like swords according to find context. After Roymans 1996, fig. 6, with additions.

gic Gaul and the Rhineland.14 These swords occur frequently in non-military contexts, i.e. settlements, graves, rivers or cult places. Their concentration in the Rhineland zone largely corresponds to the territories of those peoples involved in ethnic recruitment. The most obvious interpretation is to link these weapons to returning auxiliary veterans who brought part of their equipment home, often depositing it in ritual contexts. If we look at the distribution of gladii, however, it is difficult to gauge the true extent of the phenomenon of returning veterans. Recent research in one subregion, the Batavian river delta, has provided important new insights, with the rich potential of metal detection proving vital. Using finds from excavations and from public and private collections, Johan Nicolay has compiled a comprehensive inventory of weaponry and horse gear from civilian contexts in the civitas Batavorum.15 Fragments of these militaria occur in almost every settlement from the 1st to the 3rd centuries (fig. 4). The objects are usually found in the immediate vicinity of traditional byre houses, where they were presumably once kept. Most of

14

Roymans 1996, fig. 6, and appendix 1. Gladius sheaths

Worden’, Kesteren-‘De Woerd’, Ommerenveld-‘Blauwe

have also been included in the inventory. For the

Kamp’, Zoelen-‘De Beldert’, Opijnen-‘Westerbroek II’,

Batavian river area, the distribution map has been

Opijnen-‘Keizershof ’, Wijk bij Duurstede-‘De Horden’,

updated with several new finds from rural settlements

Oosterhout-‘Van Boetzelaerstraat’, Wijchen- ‘Wijchense

published recently by Nicolay 2007, plates 16-22: Tiel‘Passewaaijse Hogeweg’, Echteld-‘Medel’, Rumpt-‘De

Meer’, Nistelrode-‘Zwarte Molen’. 15

Nicolay 2007.

13

urban centre rural settlement cemetery river cultplace

Fig. 4. Distribution of Roman military equipment and horse gear in the Batavian river area according to find context. After Nicolay 2007, fig. 3.6.

the material, especially weaponry fragments, originates from the 1st century AD. Nicolay’s interpretation is that these military items were brought home by returning veterans. On their discharge, soldiers could sell all or part of their equipment back to the army, offer it up in a cult place to Hercules, or keep it at home as personal memorabilia. In this latter case, a considerable quantity of militaria eventually ended up as settlement waste. If Nicolay is correct, the militaria give us an idea of the scale on which veterans returned to their homeland. This has proven to be far greater than previously thought, confirming the notion of an extraordinarily intensive recruitment affecting almost all Batavian families.16 Just how representative is the dense distribution of militaria in rural Batavian contexts for other regions in the Rhineland? The dense concentration of find spots in the Batavian area seems to be due to a combination of factors: the excellent conservation of bronze objects in clay soils, the implementation of a targeted programme to make an inventory of metal detector finds in private and public collections, and of course the historically documented, intensive military recruitment among the Batavians. We can probably expect a similar pattern of finds, albeit less dense, in other regions that experienced substantial ethnic recruitment. In fact, this pattern is reflected in the distribution of Roman gladii in Northern Gaul and the Rhineland.

16 

Demographic reconstructions suggest that almost every

army. Cf. Willems 1984, 236; Roymans 2004, ch. 8.5;

Batavian family must have had one or two men in the

Derks/Roymans 2006, 123.



G

A

E

R

B

M

Rhi

A

C

N

D

I

E

A

F

ne

Wa a l Meuse

0

10

20 km

Fig. 5. Distribution of wooden wax tablets from rural settlements (A) and bronze seal boxes in the civitas Batavorum, according to find context. After Derks/Roymans 2002, fig. 7.6, and Derks/Roymans 2006, fig. 4. B Roman army camp; C civitas capital; D rural centre (vicus); E rural settlement; F sanctuary (large symbols indicate 5 or more seal boxes)



b   n z      b x    n      p      f   n w     n 

Another artefact category relevant to our research is bronze seal boxes. The generally held view is that these boxes were primarily used for sealing private letters written on wax tablets. As we have argued elsewhere, the distribution of the seal boxes allows us to say something about the spread of literacy and – indirectly – about the spread of knowledge of the Latin language.17 Until recently seal boxes were primarily known from Roman army camps. However, as our inventory of finds in the Batavian region revealed, these metal objects also occur regularly in simple rural settlements containing only byre houses (fig. 5). If the boxes originally belonged to letters kept in these houses (fragments of wax tablets have indeed been found in some of these settlements), we can then interpret these finds as archaeological evidence of private correspondence between Batavian soldiers in distant parts of the empire and their families back home. Given the typo-chronology of the seal boxes, this practice seems to have started in the first half of the 1st century AD. It is tempting to connect this evidence with the findings of the study

17

Derks/Roymans 2002.

1

Rh

in

e

Cologne Bonn

M o sell

e

Koblenz

Som

Mainz

me

Worms M

se

100 km

eu

0

Fig. 6. Distribution of 1st-century AD Rhineland gravestones with cavalry scenes. Based on Gerster 1938 and Gabelmann 1973. Large symbols indicate 5 or more gravestones.

of writing tablets from the fort at Vindolanda near Hadrian’s Wall, where a Batavian unit was stationed in about AD 100. The fact that the many text fragments discovered were written in many different hands is concrete evidence that most soldiers were able to read and write Latin. The content of some texts even suggests communication between soldiers and their homeland.18 In this context it is easy to imagine that returning veterans acted as readers, translators and letter writers in rural settlements.19



 x     y v      n   n      n -   y     v    n n

Stone funerary monuments are another interesting material category for the discussion on veterans. Roman legionary soldiers introduced Italian burial practices into the Rhineland in the early 1st century AD, including different variants of stone funerary stelae with commemorative inscriptions and images.20 Subsequently,

18

Cf. Tab. Vindol II 255; II 310; II 346.

19

Although the Batavian case study suggests that we can also expect to find seal boxes in other rural areas in the

1

Rhineland where auxiliary units were recruited, concrete evidence is still scarce. Cf. Derks/Roymans 2002, fig. 7.5. 20

Gerster 1938.

Nijmegen

Rh

in

e

Wesseling Maastricht

M o sell

e

Koblenz

Som

Chameleux Arlon

me

eu se

100 km

Neumagen Nennig

M 0

Bertrange

Fig. 7. Distribution of 1st-century AD Roman tower tombs with cavalry combat scene on a frieze. Based on Gabelmann 1973, 1977 and Krier 2003. Large symbols indicate 2 or more tombs.

we observe auxiliary soldiers, and in particular cavalrymen, adopting the tradition of funerary stelae from the early 1st century onward, but in combination with new figurative scenes produced by Rhineland sculpture workshops. I am referring here to the Rheinische Reitergrabstelen that show a mounted cavalryman riding down a naked, or semi-naked, barbarian, which can be dated to the 1st century AD (fig. 6).21 These stelae belong to auxiliary horsemen who died during active service; it is therefore no surprise that their presence is confined to the military centres in the Rhineland, especially Mainz and Bonn. We then observe that veterans introduced these military funerary monuments and the associated repertoire of images into the civilian sphere. It is interesting to note the presence in the Rhineland of a group of large tower tombs decorated with a relief of cavalry combat, which Gabelmann has been able to largely reconstruct (fig. 7). A few fragmentary inscriptions naming family members of the deceased show that these monuments were erected by veterans,22 as does the distribution pattern. The monuments do not occur in the large military centres of the Rhineland but in the rural hinterland, which is a further clue that we are dealing here with indigenous auxiliary veterans who returned home after 21

Gerster 1938, 19, type B; Gabelmann 1973.

woman, probably the wife or daughter of the deceased

22

A fragment of an inscription from Maastricht mentions

(Gabelmann 1973, 139-140).

a son of the deceased (Panhuysen 1996, 270-273), and a fragmentary inscription from Wesseling mentions a

17

discharge.23 Given the expensiveness of the monuments,24 these were no ordinary soldiers, but probably commanders of cavalry regiments. In three cases where the original find spot is known, this involved villa settlements, which confirms that the deceased came from the upper echelons of society.25 Veterans used these impressive funerary monuments with cavalry combat scenes (comparable in size to the Poblicius tomb at Cologne) to underscore their membership of and affinity with the Roman military community in a rural, civilian setting.

6 italia n terra sigillata f rom n o n - militar y, rural co n te x ts Although the militaria give us some idea of the scale on which veterans returned in the 1st century, it is still not clear precisely when this practice began. It is helpful here to examine the distribution and social use of Italian or Italian-style terra sigillata in non-military contexts in the Lower Rhine frontier zone during the Augustan-Tiberian period. This red slip tableware, produced in Italy and in some workshops in southern Gaul, represents one of the most-studied pottery groups in the Roman world.26 It has been found in large quantities in the earliest Roman army camps in the Rhineland. Up until now, attention has focused heavily on aspects of the pottery’s typo-chronology, production and distribution, with much less work being done on consumption. My research focuses on the occurrence of Italian terra sigillata in indigenous settlements and cemeteries and its interpretation. To this end, an inventory has been made of this ware in the Netherlands, Belgium, the German Rhineland and Northern France. The result is a database of some 44 sites from native contexts.27 By far the majority of these rural sites (40 in number) are clustered in the Lower Rhine frontier zone (fig. 8). Here, the Italian terra sigillata appears to be less rare than is often assumed; almost every recently excavated native settlement in the central and southeastern Netherlands, in which serious attention has been given to ceramic material, has yielded one or more fragments.We can therefore expect a considerable intensification in the distribution pattern in the near future. But how representative is the rare occurrence of Italian terra sigillata in rural contexts in the Trier area and Northern France? Its presence in civitas capitals like Trier, Bavay and Amiens suggests that it also circulated in settlements in the surrounding countryside, following the normal pattern of town-countryside interaction. There are, however, several indications that the distribution of Italian terra sigillata was less dense in these southern areas than in the Lower Rhine frontier region. The almost complete absence of this fine ware in the Gallo-Roman cemeteries in the Aisne/Marne region and the Trier area is signifi-

23 

Alföldy (in Gabelmann 1973, 140) states that in one case – the remains of a monument at Wesseling near Cologne

14,000 sestertii. Cf. Panhuysen 1996, 270, 273, n.1. 25 

Associations with villa settlements are reported for

– we are dealing with a legionary veteran of Italian ori-

the remains of the grave monuments from Wesseling

gin. A fragmentary inscription mentions a woman (prob-

(Gabelmann 1973, 135, 193), Nennig (Gabelmann 1977,

ably the daughter or wife of the deceased) whose name

101-105), and Bertrange in Luxembourg (Kremer 2009,

suggests an origin in Northern Italy. An alternative inter-

128-129). In the other cases the remains come from sec-

pretation is that it concerns a cavalry veteran of Ubian

ondary, Late Roman contexts where they have been re-

origin who had married the daughter of an ex-legionary

used as spolia (Maastricht, Koblenz, Neumagen, Arlon).

who had settled in Cologne. Cf. in this context Tacitus’

26 

remark (Hist. 4.65) on the intermarriage between Ubian

27 

families and legionary veterans of Mediterranean origin. 24 

The text fragment of the tower tomb from Maastricht seems to mention the costs of the grave monument:



Ettlinger et al. 1990. The contexts can be specified as follows: rural settlements 30 sites; grave contexts 9 sites; cult place 1 site. For the database, see the appendix in Roymans 2009a.

Bentumersiel

a

Winsum

b c d e

Nijmegen

Haltern

Riethoven

Rh ine

Cologne Tongres

Arras

Bavai

Amiens

Trier 0

100 km

Fig. 8. Distribution of Italian terra sigillata in the Rhineland frontier zone and adjacent areas. After Roymans 2009a, fig. 9. a Roman military camp; b civitas capital; c vicus; d rural settlement; e idem, with broad spectrum of early-Roman imports

cant. For example, the well-known, fully excavated cemetery of Wederath/Belginum near Trier failed to produce a single piece of Italian terra sigillata. Instead, it is the beakers, cups and platters of the so-called Belgic ware (partly imitating terra sigillata forms) that dominate the rich ceramic spectrum of the cemeteries during the Augustan-Tiberian period. Also significant is the discovery in the town of Tongres that the first phase of civilian habitation in the late Augustan-Tiberian period (phase 2) corresponds to a dramatic drop in the use of terra sigillata compared to the preceding military phase (phase 1).28 All this 28

Vanderhoeven 1996, fig. 9. Future research will show

Italian terra sigillata in rural settlements and cemeteries

us to what extent this trend observed at Tongres is also

may suggest that, there too, the consumption of this fine

relevant for other civitas capitals in Northern Gaul, such

ware was more closely related to the Roman military

as Bavay, Amiens, Arras and Trier. The rare occurrence of

than is generally assumed.

1

seems to suggest that the supply and distribution of terra sigillata in these northern regions was heavily attuned to the demands of the Roman army, and that the – probably cheaper – Belgic ware had a much more dominant position in civilian centres. These observations make the relatively dense distribution of Italian terra sigillata in the Lower Rhine frontier zone an interesting phenomenon, all the more so because the region was completely excluded from imports of Italian wine and other Mediterranean luxuries in the Late Iron Age.29 Classical authors even speak of resistance to the consumption of Roman luxury goods among the Nervii and the Germanic Suebi.30 The Late Iron Age pottery tradition in the Lower Rhine region was completely dominated by handmade domestic ware. Against this background, how can we explain the rapid acceptance of terra sigillata by Lower Rhine groups in the course of the late Augustan-Tiberian period? Everything points to a close military connection,31 but what precise form did this take? The occurrence of Italian terra sigillata at native sites in the Lower Rhine region has generated different social interpretations. 1. Up until the early 1990s, a purely military interpretation prevailed.32 Italian terra sigillata was considered the diagnostic artefact par excellence for tracing Roman military activities. The sites were almost automatically associated with army camps, guard posts, or simply with the Roman army’s lines of march. The possibility that local indigenous people were the consumers of the terra sigillata was not taken seriously, giving rise to a number of questionable military interpretations of sites. 2. In the early 1990s Jan Slofstra proposed an alternative political interpretation, which associated the Italian tableware with native elites. Influenced by core-periphery and prestige-good models, he explained the terra sigillata and related ceramic imports as diplomatic gifts presented by the Roman authorities to local chiefs in the Lower Rhine frontier zone.33 The problem here, however, is that Italian terra sigillata is found so frequently that we need to put its elite character into perspective; most of the excavated sites show no evidence of the presence of an elite. Moreover, Slofstra’s model fails to explain why and how native elites had come to appreciate this Roman pottery within a relatively short time. 3. My alternative is to present a new military interpretation, one that cuts across the traditional divide between Romans and natives and that relates the spread of terra sigillata to the role of ethnic soldiers and veterans as agents in the transmission of Roman material culture and ideas to indigenous groups. What are the main patterns of use for terra sigillata in the earliest Rhineland army camps? Firstly, the general pattern shows that common soldiers owned a cup and a platter for personal use.34 Indicative here are the many graffiti with personal names, a reference to a group of soldiers sharing the same barracks. Secondly, officers owned a large set of fine tableware for banquets and symposia, which played a key role in the shaping of patronage networks in the Roman army. Because officers were in a position to receive guests, they will have had a larger set of fine tableware. Archaeological evidence of such symposia has been found in the army camp at the Kops Plateau at Nijmegen. A refuse pit in a wing of the principia contained the remains of a copious banquet: many drinking cups and platters (mostly of Italian terra sigillata), oil lamps, and the bones of different kinds of meat and fish.35 29 

For distribution maps of Dressel 1 amphorae and bronze drinking equipment in Northern Gaul and the Rhineland, see Roymans 1990, figs. 7.2 and 7.6; Fitzpatrick 1985, fig. 5.

30 

(over 50 kms) from urban centres. 32 

E.g. Boeles 1951; Van Tent 1978, 211; Mertens 1983; Galestin 2001/2002.

33 

Slofstra 1991, 149, 176.

Caesar, BG 2.15.4 (Nervii); 4.2.1, 4.2.5 (Suebi). Cf. Roy-

34 

mans 1990, 163; Fitzpatrick 1985, 311.

35 

31 

Von Schnurbein 1982, 132-134. This find complex dates to the earliest phase of the army

An alternative interpretation connecting the occurrence

camp (the period of Drusus), when mainly legionaries

of Italian terra sigillata in rural settlements to regular

were stationed here. See Van Enckevort/Zee 1996, 53 and

exchange relations between town and countryside is less

38 ff.

plausible, since many sites are situated at greater distances



Fig. 9. Tiel-Passewaaij. Phase I (50 BC-50 AD) of the native settlement with byre houses and the location of a fragment of Italian sigillata (arrow).

This consumption pattern relates primarily to legionary soldiers in the earliest Roman army camps. The question, however, is to what extent auxiliaries adopted the practice at this early stage. Although we know little as yet about the consumption behaviour of the first generations of auxiliaries,36 there are some clues to suggest that they soon began emulating this practice. From Nijmegen we know of several smaller encampments around the fortress on the Kops Plateau that date from the Augustan-Tiberian period and appear to have been used by auxiliary units for varying lengths of time. These have yielded a considerable number of sherds of terra sigillata, including one with an incised Celtic name.37 Phase 1 of the excellently preserved Roman castellum at Valkenburg in the Dutch river delta, built during the reign of Caligula in AD 39, gives us an understanding of the material culture of auxiliary soldiers and the spatial organisation of their barracks. The large numbers of terra sigillata sherds found here suggest that this tableware was the personal property of common soldiers.38 In addition to tableware for everyday personal use, native commanders of auxiliary units, on the other hand, will have been familiar with the Mediterranean symposium tradition practised in officer circles of the Roman army and will therefore have owned a larger set of fine tableware, just like their Mediterranean peers from the legions. Relevant in this context is Tacitus’ story about the Frisian leader Cruptorix, who hosted a large group of Roman soldiers in his ‘villa’ during the Frisian revolt of AD 29.39 Tacitus explicitly states that Cruptorix 36

The problem is that early auxiliary camps, which can be

from these early auxilia camps still awaits publication

expected near the legionary camps, have not yet been studied in any detail. Auxiliary camps from the Augustan-

(personal communication H. van Enckevort, Nijmegen). 38

37

The text of the graffito reads T.M. OLLORIGIS[ ]. See

Glasbergen 1972, 77 ff.; Glasbergen/Groenman-van Waateringe 1974.

Tiberian period are known from Nijmegen and Neuss. 39

Tacitus, Annales 4.73. See also Woolf 2009.

Van Enckevort/Zee 1996, 41-43. The material evidence

11

RIETHOVEN-Heesmortel period I: c. 25 BC-AD 50 period II-III: c. AD 50-c. 175 period IV: c. AD 175-c. 250 Italian sigillata

cemetery: • 1 chalice • 1 platter • 10 cups

settlement: • 7 cups • 6 platters

0

50 m

Fig. 10. Riethoven-‘Heesmortel’. Distribution of Italian sigillata in a native settlement with byre houses and an associated cremation cemetery.



was a former Roman soldier, probably of officer rank. As a veteran he will have spoken Latin and was no doubt well-acquainted with the lifestyle and material culture of the Roman army. Living in his ‘villa’, he will have continued to avail himself of this material culture and lifestyle, exploiting his military connections to gain access to Roman imports like wine, fish sauce and fine tableware. In this sense, Cruptorix may function as a historical role model. If we now return to our Italian terra sigillata and analyse in greater detail the archaeological contexts in which this ware has been found, we can make some interesting observations. Firstly, a large group of native settlements have yielded only one or several fragments of terra sigillata, often in combination with some items of Belgic ware and sometimes fragments of early wine and fish-sauce amphorae. In the settlements excavated at Breda, Wijk bij Duurstede, Oss, Hoogeloon and Tiel (fig. 9), the early imports are associated with byre houses of the native type.We are probably dealing here with personal property taken home by ordinary auxiliary veterans who had learned to use and appreciate this material during their time in the army. The material has nothing to do with elites. Moreover, it is noteworthy that the terra sigillata from rural settlements is dominated by simple platters and cups,40 forms that also predominate in the contemporary camps. Secondly, there are some native sites where the terra sigillata has been found in larger quantities, in combination with a broad spectrum of other ceramics typical of the army camps of the Augustan-Tiberian period, especially Belgic ware, wine and fish-sauce amphorae. The best example is the partially excavated settlement and cemetery at Riethoven, which has produced almost 100 fragments of Italian terra sigillata, representing at least 25 different items, together with the typical broad spectrum of other imports (fig. 10).41 One spot within the cemetery has yielded a concentration of pottery that we can interpret as the remains of a set of symposium tableware. In addition to ordinary cups and a platter, the spectrum includes a decorated chalice and a Dressel 2-4 wine amphora.42 The finds from the adjacent settlement are associated with a small group of native-style houses, situated within a large rectangular enclosure. We might imagine at Riethoven the homestead of a Cruptorix-like figure who had held an officer post in an auxiliary unit, who was familiar with the Roman symposium tradition and who owned a large set of fine tableware and other imports that enabled him to maintain his military network and to receive guests in the Roman style. Other potential sites of the Riethoven type are Winsum and Bentumersiel in the Frisian coastal area.43 To date, these sites have been interpreted as probable Roman military stations. We may conclude that Italian terra sigillata was a characteristic component of the material culture of the Roman army, and it appears to have been used by auxiliary soldiers as well. Its presence therefore in indigenous settlements in the Lower Rhine frontier zone, where town-countryside relations were still poorly developed in Augustan-Tiberian times, seems to suggest close links with the military community. We are probably dealing here with the activities of ethnic soldiers and veterans who took this tableware home, a practice that would appear to have started in the Augustan era. The users of terra sigillata in native settlements were primarily expressing their connection with the Roman army and their membership of the Roman military community and its associated lifestyle.

7

v etera n s a n d the de v elo p me n t o f roma n v illas

A crucial point of discussion in the context of this volume is the impact of auxiliary veterans on the appearance of villas and the formation of villa landscapes. To what extent did their affinity with Roman military culture make them more receptive to the civilian variant of Roman culture, of which villas were 40 

Conspectus 18 platters and Conspectus 22 cups are the

42 

dominant types. See Roymans 2009a, appendix 1.

43 

41 

See Roymans 2009a, table 1, fig, 13, and p. 41.

Roymans 2009a, fig. 13. Winsum: Galestin 2001/2002. Bentumersiel: Ulbert 1977.



Oss 78

Oosterhout H 1

Druten H12

WbD - De Horden H 25

Druten H11

WbD - De Horden H 9

0

5m

Fig. 11. Ground plans of native porticus houses from the Batavian river area. After Roymans 2009b, Abb. 6.

an exponent? The answer is that it didn’t. There is no direct relationship between regions with heavy ethnic recruitment and the appearance of villa landscapes. On the contrary – it is in the zones where recruitment occurred most intensively that we see a scarcity of villas. Once again, the Batavians are a case in point. There we encounter in the first centuries AD a widespread adherence to the tradition of native byre houses.44 Although we can observe some degree of impact of Roman-style architecture on house construction, this often amounted to no more than the addition of a wooden porticus (fig. 11). It is an interesting proposition that such hybrid houses were not inspired by civilian villa architecture, but by the military architecture of the camps and associated settlements, which could then link the porticus farmsteads to returning veterans.45 The question then becomes: how do we interpret the rare ‘true’ villas in this region? Once again, a military connection is a possibility. They may have belonged to returning veterans who had held an officer rank in the army and who were then admitted to the decurial elite of the civitas. Certainly, the villa of Hoogeloon makes such a link appear likely.46

44 

See the discussion in Roymans 2009b; Roymans/Derks/ Heeren 2007.

45 

Vos 2009, 237-251; Heeren 2009, 157-160; Roymans 2009b, 92.



46 

Roymans et al., in prep.

In the villa landscapes of the northern provinces we see a general pattern of villas evolving from native settlements with no involvement on the part of veterans. This does require some comments, however. An issue arising in the lignite mining area of the Cologne hinterland is whether the many small villas excavated there were the result of land grants to retired legionaries at the time of foundation of the colonia under Claudius. The archaeological data does indeed point to large-scale colonisation around the middle of the 1st century AD, but it is debatable whether this involved veterans or settlers from the Gallic hinterland.47 In any event, the excavated settlements did not develop into small villas until several generations later, so there could not have been any direct impact from veterans. Furthermore, reference has been made above to the several examples of remains of monumental tower tombs featuring cavalry combat scenes (fig. 7) found in the immediate vicinity of significant villas, such as those of Bertrange and Nennig in the Trier region. However, we should not associate these 1st-century mausolea with ordinary auxilia soldiers who had served 20 to 25 years in the army, but with members of the decurial elite who in the context of a kind of cursus honorum had progressed through the army ranks before moving on to a civilian career.48 A 2nd-century example of such a person is known to us from a funerary inscription found near the villa of Mersch.49

8 discussio n What conclusions can we draw from the data presented here? Archaeological evidence suggests that Roman military culture had a profound impact on Lower Rhine rural communities, and that the number of ethnic soldiers returning to their homeland after discharge, particularly in the 1st century AD, is much higher than previously thought. We are also able to distinguish different social levels within the group of veterans residing in the countryside. Finds of Roman military equipment, seal boxes, military diplomas, and small numbers of Italian terra sigillata seem to be associated with common auxiliary soldiers, whereas larger numbers of early terra sigillata and above all monumental tower tombs with cavalry scenes suggest men who had held officer positions in auxiliary units. Several factors may have motivated soldiers to return to their homeland in this early phase. 1. Socio-economic motives. Soldiers may have returned to their birthplace because of a claim to their share of the parental property. Moreover, auxiliary soldiers were not given grants of land or money on discharge, and before the end of the 1st century it was rare for them to be awarded Roman citizenship. This seriously restricted the social opportunities for auxiliary veterans embarking on life as civilians, making a return to their homeland an attractive option.50 2. Cultural motives. Despite the lengthy absence from their region of birth, many veterans maintained a close affinity with their homeland and its way of life. Identity, self-image and ‘sense of locality and belonging’51 are the key concepts here. In the Batavian river area, common ex-soldiers who returned to their homeland settled once more in the traditional indigenous byre houses. The close association of this type of house with the stabling of animals points to a deeply-rooted affinity with an ancestral way of life in which cattle and horses played a key role. The evidence for the importance of returning veterans in the Lower Rhine region brings us to the role of auxiliary veterans as mediators of Roman ideas, practices and material culture to indigenous rural 47 

See the discussion in Gaitzsch 1986; Heimberg

citizenship from the later 1st century onwards sig-

2002/2003; Lenz 2006.

nificantly increased an ex-auxiliary soldier’s chances of

48 

improving his social position. See Haynes 2001, 76-77,

49 

about the limited rights and awards of auxiliary veterans

See the discussion in Kremer 2009, 128-129. CIL XIII 4030. Cf. Kremer 2009, 128-129. See also Derks, this volume.

50 

Pferdehirt (2002) points out that the granting of Roman

compared to legionary veterans. 51 

Lovell 1998.



Table 1. Specification of origin in inscriptions of individuals with a Batavian background. Between brackets: the number of people who possessed Roman citizenship, which shows a marked increase in the 2nd century. After Derks 2009, 257. IA = first half 1st century; IB = second half 1st century; II = 2nd century; III = 3rd century

populations. We may assume that army service had profoundly transformed the cultural identity and self-representation of auxiliary soldiers.52 After being discharged, veterans may have acted as interpreters, translators, middlemen, agents, brokers, messengers and commissioners in a wide range of social contexts. Ethnic veterans were probably the first group to introduce Roman pottery, a Roman-style cuisine with new products and new ways of preparing, serving and consuming food, new styles of clothing, knowledge of Latin language and writing, etc. All of this must have had a considerable impact on their social environment. In the 1st century, when Roman civic culture was still poorly developed in the Lower Rhine region, veterans were responsible for a powerful dissemination of the military variant of Roman culture within the rural world. It is also important to understand that returning veterans often brought a family with them.53 Veterans’ wives in particular will have contributed to the romanisation of rural communities. After all, they had become accustomed to Roman army culture in a broader sense during their time in the camp villages. An interesting point of discussion here is the extent to which veterans in the countryside consciously promulgated a Roman identity. Clearly, the use of military material culture by veterans could represent statements about Roman identity and aspirations, but at the same time they could also be used to construct a local, tribal identity, or an identity as a soldier. From the second half of the 1st century, the acquisition of Roman citizenship and the accompanying Roman name reveals the conscious adoption of a Roman identity. Of interest here is the designation of origin of Batavian individuals (usually soldiers or ex-soldiers) known to us from inscriptions (table 1). They refer to their Batavian identity, yet from the beginning of the 2nd century most of these individuals also had Roman citizenship. In a normal situation, being a Roman and being a Batavian do not seem to have been conflicting identities. A further illustration is the 1st-century AD gravestone, found at Mainz, of the cavalryman Annauso (fig. 6). The inscription describes Annauso as a member of the Lower Rhine tribe of the Baetasians, yet the cavalry scene in which, dressed in Roman garb, he tramples down a barbarian foe, leaves no doubt that he assumed a Roman identity at the same time.

52 

Cf. Haynes 1999b.

53 

See Pferdehirt 2002 on wives and children of auxilia



veterans, based on law texts and military diplomas.

The veteran model proposed here for the Lower Rhine frontier region also provides opportunities for comparative research. It is interesting to compare the region with the militarised zones of western and northern England, where we encounter a totally different situation. Here, the culture of the Roman army barely filtered through to the indigenous rural world.54 The key reason may have been the absence of substantial ethnic recruitment in these zones, and thus of veterans who could bridge the cultural gap between the military and rural populations.55 In the 1st century AD, auxiliary veterans are shown to have returned in vast numbers to rural settlements in the Lower Rhine frontier region, where – quite independently of normal town-countryside interactions – they gave a substantial boost to the process of romanising rural populations. We need to remember, however, that we are dealing here with a ‘discrepant’56 military path toward ‘becoming Roman’, one which did not yet guarantee successful integration into the mainstream, civilian variant of Roman culture. An inherent part of military culture is an emphasis on profiling oneself vis-à-vis civilian culture, on mobile, portable wealth, and less on investments in embellishing and enlarging homes.57 This suggests that from the later 1st century AD onwards the dynamic behind the formation of villa landscapes was primarily driven by forces within the civilian, urban culture in combination with economic forces.

re f ere n ces Aarts, J., 2003: Monetization and army recruitment in the Dutch River area in the early 1st century AD, in R. Wiegels/S. Seibel (eds), Kontinuität und Diskontinuität. Germania Inferior am Beginn und am Ender der römischen Herrschaft, Berlin, 162-180. Alföldy, G., 1968 : Die Hilfstruppen in der römischen Provinz Germania inferior, Düsseldorf (Epigraphische Studien 6). Andrikopoulou-Strack, J.N., 1986: Grabbauten des 1. Jahrhunderts n.Chr. im Rheingebiet. Untersuchungen zu Chronologie und Typologie, Köln (Bonner Jahrbücher, Beiheft 43). Boeles, P.C.J.A., 1951: Friesland tot de elfde eeuw, ’s-Gravenhage. Bridger, Cl., 2006: Veteran settlement in the Lower Rhineland. The evidence from the civitas Traianensis, Journal of Roman Archaeology 19, 137-149. Demougin, S., 1999: Les vétérans dans la Gaule Belgique et la Germanie Inférieure, in M. DondinPayre/M.-Th. Raepsaet-Charlier (eds), Cités, municipes, colonies. Les processus de municipalisation en Gaule et en Germanie sous le Haut Empire romain, Paris, 355-380. Derks T., 2009: Ethnic identity in the Roman frontier. The epigraphy of Batavi and other Lower Rhine tribes, in T. Derks/N. Roymans (eds), Ethnic constructs in Antiquity.The role of power and tradition, Amsterdam (Amsterdam Archeological Studies 13), 239-282. Derks, T/N. Roymans, 2002: Seal-boxes and the spread of Latin literacy in the Rhine delta, in A.E. Cooley (ed.), Becoming Roman, writing Latin? Literacy and epigraphy in the Roman west, Portsmouth (Journal of Roman Archaeology, Suppl. Ser. 48), 87-134. Derks, T./N. Roymans, 2006: Returning auxiliary veterans in the Roman empire: some methodological considerations, Journal of Roman Archaeology 16, 121-135. Eck, W., 2004: Aristokraten und Plebs. Die geographische, soziale und kulturelle Herkunft der Angehörigen des römischen Heeres in der Hohen Kaiserzeit, in H. von Hesberg (ed.), Das Militär als Kulturträger in römischer Zeit, Köln, 15-35.

54 

Mattingly 2006, 223.

55 

British cohorts. Cf. Spaul 2000, 189.

Unfortunately, we do not have any information about

56 

the recruitment area of the epigraphically documented

57 

For this concept, see Mattingly 2004. Cf. Mattingly 2006, 184, 211.



Ettlinger, E., et al., 1990: Conspectus formarum terrae sigillatae Italico modo confectae, Bonn (Materialien zur römisch-germanischen Keramik 10). Fitzpatrick, A., 1985: The distribution of Dressel 1 amphorae in Northwest Europe, Oxford Journal of Archaeology 4, 305-340. Gabelmann, H., 1973: Römische Grabmonumente mit Reiterkampfszenen im Rheingebiet, Bonner Jahrbücher 173, 132-200. Gabelmann, H., 1977: Römische Grabbauten in Italien und den Nordprovinzen, in U. Höckmann/A. Krug (eds), Festschrift für F. Brommer, Mainz, 101-117. Gaitzsch, W., 1986: Grundformen römischer Landsiedlungen im Westen der CCAA, Bonner Jahrbücher 186, 397-427. Galestin, M.C., 2001/2002:Winsum-Bruggeburen, second report on the excavation.The Roman pottery, Palaeohistoria 43-44, 435-467. Galsterer, H., 1999: Das Militär als Träger der lateinischen Sprach- und Schriftkultur, in H. von Hesberg (ed.), Das Militär als Kulturträger in römischer Zeit, Köln, 37-50. Gerster, E., 1938: Mittelrheinische Bildhauerwerkstätten im 1. Jahrhundert n. Chr., Bonn. Glasbergen, W., 1972: De Romeinse castella te Valkenburg Z.H., Groningen (Cingvla 1). Glasbergen, W./W. Groenman-van Waateringe, 1974: The Pre-Flavian garrisons of Valkenburg Z.H., Amsterdam/London. Hanut, F., 2004: La terre sigillée italique et les horizons augustéens dans le Nord de la Gaule, in J. Poblome/P. Talloen/R. Brulet/M Waelkens (eds), Early Italian Sigillata. The chronological framework and trade patterns, Leuven (Babesch, Suppl. 10), 157-203. Haynes, I., 1999a: Introduction. The Roman army as a community, in A. Goldsworthy/I. Haynes (eds), The Roman army as a community, Portsmouth (Journal of Roman Archaeology, Suppl. Ser. 34), 7-14. Haynes, I., 1999b: Military service and cultural identity in the auxilia, in A. Goldsworthy/I. Haynes (eds), The Roman army as a community, Portsmouth (Journal of Roman Archaeology, Suppl. Ser. 34), 165-174. Haynes, I.P., 2001:The impact of auxiliary recruitment on provincial societies from Augustus to Caracalla, in L. de Blois (ed.), Administration, prosopography and appointment policies in the Roman Empire, Amsterdam, 63-83. Heeren, S., 2009: Romanisering van rurale gemeenschappen in de Civitas Batavorum. De casus Tiel-Passewaaij, Amersfoort (Nederlandse Archeologische Rapporten 36), online available at http://hdl.handle. net/1871/15248. Heimberg, U., 2002/2003: Römische Villen an Rhein und Maas, Bonner Jahrbücher 202/203, 57-148. James, S., 1999: The community of soldiers: a major identity and centre of power in the Roman empire, in TRAC 8, 14-25. Johnson, A., 1987: Römische Kastelle des 1. und 2. Jahrhunderts nach Chr. in Britannien und in den germanischen Provinzen des Römerreiches, Mainz (Kulturgeschichte der antiken Welt 37). Kremer, G., 2009: Das frühkaiserzeitliche Mausoleum von Bartringen (Luxemburg), Luxembourg (Dossiers d’Archéologie du Muséé National d’Histoire et d’Art, 12). Lenz, K.H., 2006: Veteranen der römischen Armee im Siedlungsbild einer früh- und mittelkaiserzeitlichen Koloniestadt und deren Hinterland. Das Beispiel der Colonia Claudia Ara Agrippinensium (Köln), Germania 84, 61-91. Lovell, N., 1998: Introduction, in N. Lovell (ed.), Locality and belonging, London, 1-24. Mattingly, D., 2004: Being Roman. Expressing identity in a provincial setting, Journal of Roman Archaeology 17, 5-25. Mattingly, D., 2006: An imperial possession. Britain in the Roman empire, 54 BC-AD 409, London. Meadows, K.I., 1994: You are what you eat. Diet, identity and romanisation, in S. Cottam et al. (eds), TRAC 94. Proceedings of the Fourth Annual Theoretical Roman Archaeology Conference, Oxford, 133-140. Mertens, J.R., 1983: The military origins of some Roman settlements in Belgium, in B. Hartly/J.Watcher (eds), Rome and her northern provinces. Papers presented to Sheppard Frere, Oxford, 155-168.



Nicolay, J., 2007: Armed Batavians. Use and significance of weaponry and horse gear from non-military contexts in the Rhine Delta (50 BC – AD 450), Amsterdam (Amsterdam Archaeological Studies 11). Panhuysen, T.A.S.M., 1996: Romeins Maastricht en zijn beelden, Maastricht/Assen. Pferdehirt, B., 2002: Die Rolle des Militärs für den sozialen Aufstieg in der römischen Kaiserzeit, Mainz (Monographien RGZM, 49) Roxan, M., 2000: Veteran settlement of the auxilia in Germania, in G. Alföldy/B. Dobson/W. Eck (eds), Kaiser, Heer und Gesellschaft in der Römischen Kaiserzeit (Heidelberger Althistorische Beiträge und Epigraphische Studien 31), 307-326. Roymans, N., 1990: Tribal societies in Northern Gaul. An anthropological perspective, Amsterdam (Cingula 12). Roymans, N., 1996: The sword or the plough. Regional dynamics in the romanisation of Belgic Gaul and the Rhineland area, in N. Roymans (ed.), From the sword to the plough. Three studies on the earliest romanisation of northern Gaul, Amsterdam (Amsterdam Archaeological Studies 1), 9-126. Roymans, N., 2004: Ethnic identity and imperial power.The Batavians in the Early Roman Empire, Amsterdam (Amsterdam Archaeological Studies 10). Roymans, N., 2009a: Becoming Roman in the Rhineland frontier zone. The impact of ethnic recruitment and returning veterans on the Romanization of rural populations, in O. Dräger et al., Kelten am Rhein. Proceedings of the Thirteenth International Congress of Celtic Studies 1, 25-46, Bonn (Beihefte der Bonner Jahrbücher). Roymans, N., 2009b: Die Bataver. Zur Entstehung eines Soldatenvolkes, in St. Berke et al., 2000 Jahre Varusschlacht. Mythos, Stuttgart, 85-98. Roymans., N./T. Derks/S. Heeren 2007: Romeins worden op het Bataafse platteland. Een synthese, in N. Roymans/T. Derks/S. Heeren (eds), Een Bataafse gemeenschap in de wereld van het Romeinse rijk. Opgravingen te Tiel-Passewaaij, Utrecht, 11-32. Slofstra, J., 1991: Changing settlement systems in the Meuse-Demer-Scheldt area during the Early Roman Period, in N. Roymans/F. Theuws (eds), Images of the past. Studies on ancient societies in Northwestern Europe, Amsterdam (Studies in Prae- en Protohistorie 7), 131-199. Spaul, J., 2000: Cohors 2.The evidence for and a short history of the auxiliary infantry units of the Imperial Roman Army, Oxford (BAR Int.Ser. 841). Speidel, M.A., 1995: Das römische Heer als Kulturträger, in Frei-Stolba/Herzig (eds), La politique édilitaire dans les provinces de l’Empire Romain, Bern, 299-308. Ulbert, G., 1977: Die römischen Funde von Bentumersiel, Probleme der Küstenforschung im südlichen Nordseegebiet 12, 33-65. Vanderhoeven, A., 1996:The earliest urbanisation in Northern Gaul: some implications of recent research in Tongres, in N. Roymans (ed.), From the sword to the plough. Three studies on the earliest romanisation of Northern Gaul, Amsterdam (Amsterdam Archaeological Studies 1), 189-260. Van Enckevort, H./K. Zee, 1996: Het Kops Plateau. Prehistorische grafheuvels en een Romeinse legerplaats in Nijmegen, Abcoude/Amersfoort. Van Tent, W.J., 1978: A native settlement at Jutphaas, municipality of Nieuwegein, Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 28, 1978, 199-239. Von Hesberg, H. (ed.), 1999: Das Militär als Kulturträger in römischer Zeit, Köln. Von Schnurbein, S., 1982: Die unversierte Terra Sigillata aus Haltern, Münster (Bodenaltertümer Westfalens, 19). Vos,W.K., 2009: Bataafs platteland. Het Romeinse nederzettingslandschap in het Nederlandse Kromme-Rijngebied, Amersfoort (Nederlandse Archeologische Rapporten 35). Online available at http://hdl.handle. net/1871/13283. Vossen, I., 2003: The possibilities and limitations of demographic calculations in the Batavian area, in Th. Grünewald/S. Seibel (eds), Kontinuität und Diskontinuität. Germania inferior am Beginn und am Ende der römischen Herrschaft, Berlin/New York, 414-435.



Whittaker, C.R., 1994: Frontiers of the Roman empire. A social and economic study, Baltimore/London. Willems, W.J.H., 1984: Romans and Batavians. A regional study in the Dutch Eastern River Area II, Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 34, 39-331. Wolters, R., 1988: Keltische Münzen in römischen Militärstationen und die Besoldung römischer Hilfstruppen in spätrepublikanischer und frühaugusteischer Zeit, Tyche 3, 261-272. Woolf, G., 1997: Beyond Romans and natives, World Archaeology 28, 339-350. Woolf, G., 1998: Becoming Roman. The origins of provincial civilization in Gaul, Cambridge. Woolf, G., 2009: Cruptorix and his kind. Talking ethnicity on the middle ground, in T. Derks/N. Roymans (eds), Ethnic constructs in Antiquity.The role of power and tradition, Amsterdam (Amsterdam Archeological Studies 13), 207-217.



Indications for rural slavery in the northern provinces Nico Roymans/Marenne Zandstra 1 Introduction 2 Slave labour in the Roman villa system: the historical debate 3 Method and archaeological material 4 Epigraphic evidence 5 Iconography 6 Chains 7 Ground plans of villas and specific buildings 8 Burial evidence 9 Hybridisation 10 Conclusions References Field slaves were the most wretched and impoverished creatures of the Roman empire (Samson 1989, 100).

1 i n troductio n Written sources, in particular the handbooks of the agronomists Varro, Cato and Columella, show that slaves were systematically deployed on the large villa complexes of Italy. Historians long regarded them as the principal form of labour.1 Unfortunately, we have almost no written sources that paint a picture of the situation in the countryside in the northern provinces.2 Our question then is to what extent were slaves deployed there too to work the estates of villa owners? What archaeological evidence is there? Can slavery even be ‘captured’ in archaeological terms? Slavery is a much neglected topic within archaeology. The literature on Roman villas has almost nothing to say on the subject.3 In part, this is because slavery is a vexed ethical issue that does not accord with the picture that many people like to present of a civilised antiquity.4 According to Samson, the prevailing bourgeois and capitalist worldview and its corresponding historical tradition was unable to generate any enthusiasm for archaeological research into slavery.5 All of this serves to make the lack of evidence for slavery a self-fulfilling prophecy: those who do not seek shall not find. The lack of research findings and literature on this topic is certainly also the result of methodological obstacles, however. It has long been assumed that slavery lies outside the scope of archaeology. Slaves are themselves property, and the little that they possess is difficult to tell apart from the possessions of other lower social groups. What could they possibly have bequeathed us in the way of an 1 

4 

2 

5 

Thompson 2003, 79. Cf. Belova 1987, 105.

Webster 2005, 162-163. Samson 1989, 107.

3 

Cf. Samson 1989, 102. See also George 2010, 144.



archaeological footprint, apart from the most obvious indications of slavery such as chains, collars and slave bullae?6 Given that these material categories are only seldom encountered, many archaeologists remain sceptical about the archaeological tangibility of slavery. Yet there is a case to be made, as Webster, Thompson and George above all demonstrate, that archaeologists can make a valuable contribution to discussions on rural slavery in the Roman world.7 In this essay we will investigate the possible archaeological evidence for slavery. We will then identify the extent to which this humblest social group was deployed on villa estates in the northern provinces of the Roman empire. Before examining the material categories that we consider relevant for this study, we brief ly discuss current thinking among historians about the social composition of the workforce on villa complexes in the northern parts of the empire.

2 sla v e la b our i n the roma n v illa s y stem : the historical de b ate Just as the concept of ‘villa’ was not imported complete and ready-made into the northern provinces from Italy, nor was slavery introduced there by the Romans. Written sources make reference to the existence of slaves in the pre-Roman societies of the later northern provinces, where they were no doubt the product of intertribal warfare and raiding.8 Caesar, for example, describes how in Gallic society too slaves were part of the social system, and Strabo mentions slaves as an important export product of pre-Roman Britannia.9 We cannot directly establish the extent to which slaves also populated the countryside in the Roman era. In particular, this is due to the paucity of written sources and epigraphic material, which has prompted some scholars to extrapolate the situation in pre- and post-Roman times – for which we have significantly more written sources – to the Roman period itself.10 The result is a picture of villa staff as chiefly comprising labourers (labelled coloni) whose status matched that of early medieval tenants and serfs (unfree tenants). According to this view, slaves were a minority.11 A characteristic of servi was their unfree status in legal terms, which made them the personal property of their masters and granted them almost no rights.The use of this Latin terminology is not unproblematic, however. Terms like colonus and servus conceal complex dependency relationships that could vary per region across the Roman empire, and with a different regional genesis. We should therefore use such terms with due flexibility; freedom and lack of freedom are relative concepts that can encompass a host of gradations.12 The different views of historians about the composition of the rural workforce have one thing in common: they all assume that slaves were of fairly limited importance.13 Even within the Marxist tradi6 

Shackles, collars and bullae are seen as the most direct

ly identical to that of their free tenant neighbours. Accord-

archaeological evidence for slavery.Webster 2005, 163, n. 14.

ing to Webster (2005, 169-170) dependent social positions

7 

Webster 2005; idem 2008a; Thompson 2003; George 2010.

8 

Belova 1987, 134; Thompson 2003, 121-122; Webster 2005, 169-170.

following integration into the Roman system could even be redefined as slavery in the Roman terminology. 13 

For estimates of the size of the slave population in the Roman empire, see Hopkins 1978, table II.1; Saller/

9 

Shaw 1984, 139; Harris 1999, 64-65. Hopkins estimates

10 

Thompson 2003, 121. Whittaker (1980, 80) regards this

that 35% of the population in Italy were slaves (in 35

as problematic.

BC). The proportion of slaves in the provinces was prob-

Caesar BG 5.45, 6.13, 6.19, 8.30; Strabo IV.5.2.

11 

ably lower, although Harris regards the often cited figure

12 

Whittaker 1980, 83. Wickham (2005, 259-264), focusing

of 10 percent as much too low. He suggests keeping to

on the early medieval period, avoids calling unfree tenants

a figure of 16 to 20% during the high Roman period.

‘slaves’, arguing that although they were unfree in legal

See also the discussion in Ouzoulias 2006, 224 ff. on the

terms, their economic relationship to their lords was effective-

significance of rural slavery in Roman Gaul.

Thompson 2003, 121.



tion, the old view of the Roman empire as resting primarily on a ‘slave mode of production’ has been definitively abandoned as scholars point to the importance of different forms of tenancy that would reach their peak in the 3rd and 4th centuries.14 And yet the supply of slaves appears to have continued unabated, including after the end of the Roman wars of conquest in the 1st century AD. Wightman is convinced that slaves continued to feature among the servants in the villa complexes of Gaul, but she believes it was not economically viable to deploy them in large numbers. Like Whittaker, she says that workers could be found in sufficient numbers among the local rural population. This group undoubtedly included people who were traditionally obliged to provide their services as part of a dependency relationship.15 In social respects, this will have differed little from slavery, but there was a clear distinction in legal terms. Perhaps we should assume the coexistence in the countryside of slavery and tenancy, as well as seasonal labour. This is evident not only from the literary sources we have, but also from legal evidence. For example, the jurist Ulpian wrote in the early 3rd century that landowners had to give a detailed description of their estates for taxation purposes, including the number of slaves and tenant farmers.16 Probably landowners themselves decided on the composition of the labour force needed to work their estates, giving rise to considerable variation. This examination of the different opinions on rural slavery in the Roman world leads us to conclude that the earlier view – that slaves were deployed in large numbers – is incorrect. Nevertheless, it is correct to assume that slaves constituted a normal part of the labour supply at the villa complexes.17 Below we shall see the extent to which the archaeological material supports this hypothesis.

3 method a n d archaeological material In this study we will focus on slaves working in the countryside in the northern provinces. Wherever possible we have tried to filter slaves from urban and military contexts out of the dataset, using the find location of the archaeological material as our main guideline. In the case of inscriptions, additional information could be gleaned from the text and iconography. We will therefore concentrate on the status of slaves (servi), whom we distinguish from groups of ‘tenants’ with different gradations of dependency and who are usually labelled coloni. There is a distinction between the two social groups in legal terms, but this is difficult, if not impossible, to capture archaeologically. Written sources reveal for instance that runaway coloni could also be chained up as a punishment.18 This means that chains found at villa complexes (see below) do not point conclusively to the existence of slaves. Only a few individuals among the group of slaves put to work on villa estates will be archaeologically identifiable.19 This is because only a tiny number will have been in a position to erect inscriptions, and not all will have been shackled; the numbers of chains found is much too small for this. However, compared to the other categories of labourers in villa complexes – ranging from tenants, day labourers to socially inferior members of the villa owner’s family – the presence of slaves seems highly likely, given the number of archaeological phenomena that can be linked to them. The material categories addressed in the following sections are epigraphy, iconography, shackles, settlement layout and house plans, burials and the mobile material culture of slaves.

14 

18 

15 

19 

Belova 1987, 115. Wightman 1978, 106; Whittaker 1980, 79.

Thompson 2003, 238. Webster 2005, 170.

16 

Dig. 50.15.4: Ulp. 3 de cens.; Rosafio 1994, 150.

17 

Rosafio 1994; Thompson 2003, 112-113; Wickham 2005, 259-265; Rothenhöfer 2005, 49.



Fig. 1. Gravestone of the freed slave girls Prima and Araurica, erected by their patron. Found near the villa of Liestal-Munzach (CH). Photo Historisches Museum Basel.

4 e p igra p hic e v ide n ce Epigraphy includes not only the monumental inscriptions on gravestones and votive offerings, but also texts on what are known as instrumentum domesticum, or household goods of all kinds. Of interest to the study of slavery is the category of ownership marks that were added to a wide range of objects. The first and most researched group in this context is that of collars and slave bullae.20 A second group consists of small inscribed objects such as rings, seal rings and stamps (signacula).21 Examples are known from Italic and Southern French villas, but as none have been found so far in the northern provinces, we will not examine this category more closely. A group of small epigraphic sources whose significance still needs to be tested is that of graffiti. Webster shows convincingly that slaves felt a need to add their name or other ownership mark to the few objects they possessed or could regard as their own property.22 Scratching letters or marks onto them was an easy way of doing this. For this reason, it would be useful to take a closer look at the graffiti found at villa sites. This research topic has potential, but for the present there is still little concrete evidence.

20 

Thompson 2003, 238-240.

21 

Cf. Feugère 2008.



22 

Webster 2005, 167-168.

Fig. 2. Gravestone for the slave boy Primulus, probably erected by his master, found at FöhrenHetzerather Weg near a villa site.

To date, monumental inscriptions of slaves (servi) and freedmen (liberti)23 active in the countryside in the northern provinces have constituted the most concrete evidence for the existence of rural slavery. The number remains small, but that does not mean that slaves were barely used in agriculture. We need to realise that slaves were rarely able to afford such inscriptions.24 Further, the epigraphic habit failed to take hold in the Gallic countryside beyond the Rhine frontier.25 It would therefore be wrong to conclude solely on the basis of funerary monuments, votive stones and other monumental inscriptions that there were almost no slaves at villa complexes in this part of the Roman empire. We have collected a total of five inscriptions that can be regarded with certainty as indications of rural slavery.26 Only two can be linked to a specific villa site. The first one is the gravestone erected for the freed slave girls Prima and her ‘sister’ Araurica by their patronus.27 The stone (fig. 1) was incorporated as spolia into the St Lawrence Church in Munzach (CH) and must have come from the nearby villa of Liestal-Munzach.28 The second inscription (fig. 2) is the gravestone erected for the boy Primulus, dated to 23 

Freedmen will have been more likely to be able to afford

gica. Unfortunately, he does not specify these inscriptions

such inscriptions. There was also a trend for freedmen

other than with a general distribution map. We think the

to display their new status through epigraphy. See Joshel

difference in number between our collection and his

1992; Mouritsen 2005. Cf. Belova 1987, 109, 145, who

collection of relevant inscriptions stems from a difference

reports that there are many epigraphic testimonies to

in definition of the term ‘rural slavery’. Ouzoulias seems

freedmen in the countryside.

to wield a broader definition. For instance, in his text he

24 

mentions inscriptions dedicated by or to slaves who were

25 

active in the metal industry. We have narrowed the term

Samson 1989, 100. Lazzaro 1993, 412; Derks 1998, 86 ff. and fig. 3.2.

26

In his study of the agrarian economy of Roman Gaul the ancient historian Pierre Ouzoulias (2006) also handles the topic of rural slavery. He states that there are 19 inscriptions indicating rural slavery from the provinces of Germania Inferior, Germania Superior and Gallia Bel-

'rural slavery' explicitly down to ‘slavery on villa estates’. 27 

CIL XIII, 5312: Prima C(ai) Cotei / lib(erta) ann(orum) XVI et / soror illaeus (sic) / Araurica annic(ula) / et mens(ium) VI h(ic) s(itae) s(unt)./ Patronus po(suit).

28 

Neukom [2009].



Fig. 3. Gravestone for the freedman Iucundus, who had worked as a cattle herder, from Mainz. Erected by his patron. Photo Landesmuseum Mainz.

the 2nd or the 3rd century and found at Föhren close to the Roman road Trier-Ahr valley at some 600 m from a villa site. The stone shows a relief with the representation of a child playing with a dog. The inscription reveals that the boy Primulus and his parents Primula and Sequens were slaves.29 The grave probably belonged to a small necropolis wich also included remains of a monumental funerary pillar. The gravestone erected by a freedwoman of the vilicus Celsus comes from the large cemetery at the civitas capital of the Mediomatrici, present-day Metz.30 Unfortunately, her name has not survived intact. 29 

CIL XIII 4194: D(is) M(anibus) / Primuli infa / ntis Sequen / tis et Primule / filio servo. See also Binsfeld 1974; Schumacher 2001, 245. About the archaeological context of



the gravestone, see Krause 2006, 271. 30 

CIL XIII, 4352: D(is) M(anibus). / Cam[..]a Celsi vil(ici) / lib(erta) Ianuariae filiae / annorum XVII.

Fig. 4. Grave reliefs from Neumagen (below) and Luxembourg (above), with unfree(?) servants tending to the toilette of their mistress. Photo T. Derks and drawing Alexander Wiltheim (17th century).

Given that she was once the slave of a bailiff,31 this inscription points to rural slavery in our study area, even though it comes from an urban context. 31 

A vilicus could also be an imperial freedman who looked

pretation as bailiff of an estate. Men holding such posi-

after certain taxes, or the overseer of a military guard

tions were usually slaves or freedmen themselves. Lazzaro

post. However, the wording of this text suggests an inter-

1993, 101, 178.



a

b

Fig. 5. Funerary monument from Nickenich presumably belonging to a family involved in the slave trade. On the side (b) a depiction of a slave transport. Photo LVR Landesmuseum Bonn.

The remaining two inscriptions were also found in urban contexts, but we have included them in our overview because of their iconography and the occupations they refer to. They are the funerary monument for the freedman Iucundus from Mainz,32 who was murdered by a slave, and that of the young slave Peregrinus from Speyer, the capital of the Nemetes.33 The former (fig. 3) depicts under the inscription itself a bucolic scene, with a flock of sheep and a sheepdog at the shepherd’s feet. The lengthy epitaph informs us that the deceased was a pecuarius, a cattle herder. The iconography of the latter monument also suggests that the boy was a cattle herder, which points to a rural context. The last piece of epigraphic evidence for rural slavery is the ‘testament of the Lingon’.34 In this text, transmitted to us in a medieval manuscript, a member of the Gallic tribe of the Lingones sets out how 32 

CIL XIII, 7070: Iucundus / M(arci) Terenti(i) l(ibertus), /

ser(vus), / ann(orum) X, h(ic) s(itus) e(st). The iconography

pecuarius. / Praeteriens quicumque legis, consiste, / viator, et

of the relief is described in Schumacher 2001, 77, Abb.

vide quam in/digne raptus, inane / querar. Vivere non / potui

26; Espérandieu 5955; Lazarro 1993, 167-168. Lazarro

plures XXX per / annos : nam erupuit (sic) se/rvos (sic) mihi

and Espérandieu incorrectly believe that the boy is hol-

vitam et ipse praecipitem sesse (sic) deie/cit in amnem. Apstu-

ding a bird in his left hand. However, it is not a bird but

lit / huic Moenus quod /domino eripuit. / Patronus the suo posuit. Translation in Boppert 1992b, no. 52. 33 

CIL XIII, 6109: Peregrinus C(ai) Iuli(i) / Nigellionis



a dog leash. 34 

CIL XIII, 5708. For a translation and explanation, see Sage 1991 and Le Glay 1991.

his tomb should be built, appointed and maintained. In this context he specifically assigns a role to his freedmen. We can gather from the wording of the testament that the Lingon loved to reside in the countryside. His tomb was probably erected in the vicinity of his villa (or one of his villas).

5 ico n ogra p h y Roman authors lamented the fact that their contemporaries found it impossible to distinguish between freedmen and slaves on the basis of external attributes such as clothing, appearance and ethnicity.35 Thus it is hardly surprising that we have difficulties deciding whether someone was a slave solely on the basis of how that person is portrayed. Given that our focus in this essay is rural slavery, particularly in the context of villa complexes, we will only cite examples of slaves or ex-slaves of whom it is highly likely that they were deployed on the estates. There was more to estate work than just working the land. Slaves will also have worked in the household, in the main residential building. The servant women shown attending to a lady’s toilette on funerary reliefs from Neumagen and Luxembourg (fig. 4) may well have had an unfree status.36 Given the region where both reliefs were found, it is reasonable to assume that the ladies in question were the main occupants of their villas. The funerary monument from the more northerly situated Nickenich leaves less room for doubt. The narrow left-hand side of the relief shows two ‘barbarians’ being led by chains around their necks (fig. 5b). The man leading them, who is presented above them, is dressed as a soldier and carries a cudgel.37 The enchained men are probably prisoners of war who have ended up in slavery. Because the man leading them is unarmed (except for the cudgel), Schumacher concludes that he was a citizen who had been active in the slave trade.38 Schumacher makes a direct link between this image and the people depicted on the front of the funerary monument. From left to right, three niches show a man, a woman (mother) with a child, and another man (fig. 5). Schumacher feels that the man on the left has the most obvious association with the relief, which is also on the left-hand side of the monument. The family will have owed its prosperity to this economic activity.39 The funerary monument, which is dated to the 1st century AD,40 probably belonged to a nearby, as yet unidentified, villa. This example does at least show that the villa elite in this region were acquainted with the phenomenon of shackled people, most probably slaves, and even had themselves portrayed in association with them.

35 

Schumacher 2001, 71.

36 

39 

Schumacher 2001, 65. A tumulus with surviving inscrip-

Von Massow 1932,Taf. 34, no. 184a;Ternes 1971, 143, fig.

tion was located several dozen metres away from the find

85. Cf. Binsfeld 2006/2007, 172-173; Schumacher 2001,

site for these reliefs, but the relationship to the remains of

203-205.

the stone funerary monument remains unclear. See AE

37 

The short tunica, hitched up at the sides, features on many gravestones of soldiers from the 1st century AD. Neuffer 1932, 26. For this tunica and the paenula (short cloak) as typical clothing for soldiers in funerary reliefs, see

1933, 109; Andrikopoulou-Strack 1986, 37, 42; Carroll 2006, 255-256. 40 

Andrikopoulou-Strack 1986, 43, 62, 65; Wegner 1990, 507-508.

Andrikopoulou-Strack 1986, 65-66 and Boppert 1992a, 72. 38 

Schumacher 2001, 65. For slave traders in the northwestern provinces, see CIL XIII, 8348 (Cologne) and AE 1892, 135 (Mons Poeninus).



6 chai n s The Nickenich relief discussed above (fig. 4) illustrates the use of neckchains in Roman times. Chains around the wrists, ankles or neck were usually used to shackle criminals and prisoners of war, but slaves and even coloni also ran the risk of this kind of punishment. In the original meaning of the word, coloni were (free) peasants, but in the late Roman era they had declined to semi-servile status and were tied to the land. The edict of Constantine from AD 332 shows that coloni were also punished for escaping by being chained up.41 Whenever shackles are found at villa sites, this probably does not point to their use as a punishment device for true criminals. Criminals are more likely to have been held in prisons in the towns.42 The use of chains in Roman times to restrain people is demonstrated beyond a doubt by finds of shackles that were still riveted around wrists or ankles.43 An alternative interpretation is that they were used to chain animals. We know from the post-medieval period that metal hobbles were indeed used to restrain animals, but these had a simple closing mechanism that a person could easily open. The shackles from the Roman era, however, feature ingenious locks with separate keys, making it highly unlikely that they were used on animals. The high degree of craftsmanship and complex design of some shackles also points to them not being used for animals.44 context number army camps 26 canabae/vici 3 total military sites 29 towns 11 small towns 15 villas 15 non-villa settlements 4 total rural settlements 19 7 river contexts hoard finds 6 3 cult sites burials 3 indeterminable 21 total

Table 1. Different find contexts for shackles in the northern provinces. Based on data from Thompson’s catalogue of 1993, with additions.45

114

Table 1 and figure 5 give us an idea of the distribution and different find contexts for shackles from Roman times, based on data from the catalogue in Thompson’s study (with additions).46 Shackles were 41 

Cod. Theod. V.17.1; Thompson 2003, 113-114, 116.

42 

Thompson 1993, 58; idem 2003, 219-221.

43 

These are finds from Pfünz, Remedello Di Sopra, Saint-

Northern France, Switzerland, Belgium, Luxembourg, Germany, the Netherlands and England. 46 

The total number of marked sites is 124. It is not absolute-

Vallerin, Luxé de Pellenz, Bavay and Alta fulla Munts.The

ly certain whether all shackles originate from the Roman

last of these is a Spanish villa site (villa dels Munts). See

period. Added to Thompson’s list are the rural sites of

Thompson 1993, 141; idem 2001, appendix, 151-164.

Houten-Binnenweg (Vos 2009, 145, and fig. 4.21), Budel-

Recently near Martigues in Southern France, the skel-

Duitse School (personal communication M. Bink, BAAC)

eton of a woman was excavated with shackles around her

and Someren-Ter Hofstadlaan (De Boer/Hiddink 2009,

ankles. This find is dated to the Late Iron Age and proves

85-88 and fig. 8.6), the Swiss villa sites of Vallon, Liestal-

that the phenomenon of enchained people already

Munzach and Pomy-Cuarny (Bielman/Brem/Hedinger

existed in pre-Roman society. Duval 2008.

2002, 275-277), a sanctuary in the vicus of Wederath-

44 

Belginum (Cordie 2007, 289, Abb. 235) and the canabae near

45 

the Flavian army camp at Nijmegen (unpublished).

Thompson 1993, 58; Feugère 2008, 14-15. We have confined the area in geographical terms to



a b c d e f

0

200 km

Fig. 6. Distribution of shackles from the Roman period. Based on the catalogue of Thompson 1993, with additions. a military sites; b towns and small towns; c villa settlements; d other rural settlements; e other sites (cult sites, hoard finds, river contexts, burials); f context unknown

clearly a normal phenomenon in the villa landscapes between the Seine and Rhine rivers. Most of them originate from villa sites,47 but in the Southern Netherlands examples are also known from traditional native settlements at Someren, Budel and Houten.48 These are located in a region where intensive military recruitment took place and where there was a practice of returning veterans. The chains found here may well relate to the personal slaves of retired soldiers. Shackles are also known from river contexts and in three instances from sanctuaries. For these categories we could perhaps think of ritual depositions by freedmen in which they symbolically discarded their slave status in the context of a ritualised vow to a deity. The shackle cluster along the limes is more likely to be associated with prisoners of war than with personal slaves belonging to soldiers.49 In total, more leg irons have been found than wrist irons, with neckchains very much in the minority. Thompson sees leg irons (fig. 6) as the characteristic means by which slaves were chained in antiquity. He believes that their frequent appearance in Gaul and Britannia points to the use of slave labour in agriculture.50 Further, it comes as no surprise that chains are also known from urban settlements and small towns, where slaves will have been used for a wide range of heavy work on construction sites, in workshops and the like. 47

Thompson 2003, 118-120.

48

These were leg irons in all three cases.

49

They could also point to punishment practices at military

camps, but Thompson (2003, 125) regards this explanation as less likely. 50

Thompson 2003, 238.

171

Fig. 7. Leg irons from Swiss villa settlements (Vallon, Liestal-Munzach, Cuarny). After Bielman/Brem/Hedinger 2002, 276, Abb. 332.

We must not imagine, however, that slaves wore heavy leg irons as they went about their day-to-day agrarian activities: this would have severely hampered, or made altogether impossible, their ability to carry out productive work.51 Moreover, the number of leg irons found is much too small for this. It is better to view their presence in rural settlements as an indication of temporary punishment for labourers who were entirely under the physical control of their master. This group comprised mainly slaves, but also other dependent individuals. Neckchains appear to have been used primarily when slaves were being transported (fig. 5).

7 grou n d p la n s o f v illas a n d s p eci f ic b uildi n gs The spatial organisation of villa complexes and individual buildings can also yield information relevant to the discussion on rural slavery. The spatial layout of villa sites argues for the existence of strongly hierarchical social relationships within villa communities. This applies in particular to the large axial complexes with a double courtyard, respectively an elite compound and a working compound (fig. 8). The latter compound included along its long sides a tightly symmetrical arrangement of small, highly standardised buildings, which are generally interpreted as dwellings for dependent workers. This type of villa was widely distributed across Gaul.52 Some archaeologists have suggested – sometimes by analogy with ground plans of plantations in Virginia – that it was slaves who did the work at these sites.53 These dwellings for low-status workers at the large axial villa complexes do indeed constitute an intriguing category of evidence. The simple, highly uniform structures were built by the villa owners in accordance with a fixed spatial concept. An important clue for the social interpretation of these dwellings is the fact that there are no signs of any autonomy in the work practices of those who lived there. As a rule, we do not find wells, storage structures or other ancillary buildings near the houses, nor evidence for the stalling of cattle. This suggests that the occupants did not have possessions of their own and that their labour potential was fully and directly exploited by the villa owner. The spatial layout of these villa complexes ref lects a formal, impersonal and distant relationship between the villa owner and the people living in the working compound. The precise social status of the latter group cannot be established with 51 

The find of a skeleton with shackles around its ankles in the Roman tufa mines near Pellenz (Thompson 1993, 141; idem 2003, 142) suggests that mineworkers could do their work enchained. We should bear in mind that a mineworker had a much smaller radius of activity than a



slave working the fields of a villa. 52 

See Roymans/Habermehl, this volume, with a distribution map in fig. 2.

53 

Smith 1997, 295-300; Webster 2005, 156-166, 177.

Fig. 8 Ground plan of the axial villa of Oberentfelden (CH). Redrawn after Ebnöther/Monnier 2002, fig. 139.6. A main house; B bath house; C secondary houses; D nymphaeum

A

B

any certainty, but if there is anywhere in the archaeologically documented settlements where rural slavery can be expected, it is in this category of villa.54 It is less likely that we are dealing here with free tenants, given that the workers did not control their own means of production, were not able to earn income and therefore C could not pay rent. C D With other categories of villa too, the presence of slaves is sometimes posited, although the evidence is often less convincing. For the villa of Köln-Müngersdorf (pubC C lished in 1933), whose main building was located in the centre of an enclosed compound (fig. 8), an attempt has been made to identify the function of the various ancillary buildings.55 The conclusion, based on the archaeo-ecological material and the form of the different structures, was that animal husbandry had played a significant role at this villa. According to Belova, this called for a fixed, stable supply of labour, which was generally made up of slaves.56 A secondary house with a hearth and latrine was interpreted at the time as servants’ quarters (Gesindehaus),57 but this does not mean that slaves were housed there.58 The written sources inform us about another architectural structure associated with slaves – the ergastulum, or slave prison. Columella explains how such a prison could best be created in a cellar of a villa complex, 0 50 100 m with narrow windows high above the floor to prevent shackled slaves from having any contact with the outside world.59 Cellars are regularly found in villa settlements in our study area, but there is no evidence for an interpretation as ergastulum. Although the above interpretations remain hypothetical at present, they do show archaeology’s potential to make a vital contribution to discussion by analysing the spatial order of villa settlements and in particular the dwellings of lower social groups and their associated find assemblages.60

54 

See also the discussion in Roymans/Derks (section 9)

57 

and Roymans/Habermehl, both in this volume.

58 

Horn 1987, 502. Samson 1989, 105-106.

55 

59 

56 

60 

Horn 1987, 502-505. Belova 1987, 111.

Columella Rust. I.6.3. Samson 1989, 103ff. Webster 2005, 170-177.



Fig. 9. The Roman villa of Köln-Müngersdorf with associated cremation burials. After Carroll 2001, fig. 32. A residential main house; B secondary house for servants; C granaries; D cremation burials; E sarcophagi

E

8 b urial e v ide n ce

D

The graves of slaves at or near villa sites cannot in principle be identified in archaeC ological terms,61 although such assumptions B are sometimes voiced. Excavations of the villas of Köln-Müngersdorf and Hambach 69 in the Cologne hinterland unearthed several grave clusters whose location, type C of burial, and quantity and quality of grave goods suggest that they belonged to the lowest social groups. A cluster of 55 cremation graves was found northeast of the enclosure wall at the medium-sized villa of Köln-Müngersdorf, D with a further six graves a short distance away (fig. 9). The majority of the deceased were buried without an urn and with few or no grave goods; the graves are dated to the period between c. 50 and the end of the 2nd century AD.The conclusion reached was that the people buried there belonged to the lowest social strata.62 The graves of the villa owners have yet to be found but these are expected to lie along the arterial road to Aachen.63 A distinct social differentiation is also evident in the graves examined at the Roman villas in the Hambach Forest.We will mention here a small villa (Hambach 69), where a group of simple, almost findless cremation graves, probably belonging to people of low social status, was found along the southern periphery of the villa enclosure. Along the road to the northeast of the villa site, two sarcophagi have been excavated that belonged to people of higher status, possibly the owners of the estate.64 Thus although it is not possible to find evidence for rural slavery from burials at villa complexes, the above examples do offer evidence for a highly developed social hierarchy within local villa communities, based on the location, form and content of the graves. Although we cannot establish the precise nature of the dependency relationships, we can be sure that slaves will have numbered among those buried. A

9 h y b ridisatio n It is interesting to examine whether we can say anything about specific cultural expressions by slaves in the northern provinces. We point here to Jane Webster’s study of the mobile material culture of slaves 61 

Webster 2005, 178.

62 

Thompson 2003, 124. Fremersdorf 1933, 85, 92.

63 

Several sarcophagi from the 4th century AD have been



found inside the villa enclosure at the site, and have been ascribed to a later generation of villa owners. 64 

Gaitzsch 1986, 411-412, 414-417.

from a post-colonial perspective.65 She derives her inspiration from historical and archaeological research into slavery in the New World, with its focus on hybridisation and creolisation. Webster looks in particular at the phenomenon of slaves being able to distinguish themselves in their mobile material culture through their own cultural expressions and in this way arriving at their own identity constructions. This may involve new artefacts such as jewellery and pottery, inspired by examples from both the culture of their masters and their culture of origin. Like New World slaves, slaves at Roman villa complexes will not as a rule have had a common ethnic origin. This will not have prevented them from acquiring or adapting the material culture given to them and thereby giving expression to their own identity. Webster cites as an example the pottery found in a cellar at the villa complex of Gravesend-Chalk (UK). Three items of tableware bore graffiti, which are interpreted as ownership marks. Moreover, six examples of graffiti have been found on pottery sherds, five of which can be identified as names, one written in Greek. Given that slaves in the Roman world often had Greek names, the assumption is that we are dealing here with a slave’s ownership mark. The find assemblage in combination with the building structure, possibly an ergastulum, strongly suggests that slaves were deployed at the villa of Gravesend-Chalk.66 We can conclude that this is a little explored avenue of research but one that is full of potential. Given the small chance of success in isolating material cultural expressions of slaves from that of other dependent labouring groups, it would make sense to focus research on the broader group of the dependent rural working class at villa complexes.

1 0 co n clusio n s The theme of rural slavery has barely featured until now in archaeological research into Roman villas in the northern provinces. There has been an often implicit assumption that slavery was of little consequence here and that, in any case, it has left almost no traces in the archaeological record. However, historians agree that different groups of dependent labourers, including slaves, were active in the countryside in the Roman provinces. We therefore need to bear in mind the possibility that slaves were deployed at villa complexes, as confirmed by inscriptions using the Roman terminology of servus and libertus. The material presented in this paper shows that the importance of slavery has probably been underestimated, as has our ability to say anything meaningful on this subject on the basis of archaeological evidence. With regard to this last point, we join Webster in calling for research focusing on the milieu and cultural expressions of the lowest social groups at villa sites. This will necessitate a combined study of their graves, dwellings and mobile material culture. Using the archaeological evidence as our starting point, we can then investigate the extent to which it is possible to distinguish slavery from other forms of dependent labour within this rural labour class. The most compelling evidence for the existence of rural slavery is to be found in the juxtaposition of different types of evidence.67 A concrete example is the villa site of Liestal-Munzach in northern Switzerland, discussed above. In the section on epigraphic sources we mentioned the gravestone of the freed slave girls Prima and Araurica, found in the immediate vicinity of this villa. The villa featured the characteristic long-axial layout, with the secondary buildings in the working compound probably housing dependent labourers. The find of a set of leg irons on this site lends support to the idea that slaves also featured among their numbers.The combination of the gravestone, the villa ground plan and the leg irons constitutes compelling evidence for the use of slaves at villa complexes in this part of the Roman empire.

65 

Webster 2005.

67 

Cf. Thompson 2003, 238.

66 

Webster 2005, 166-168. See also Webster 2008a, 118-122 and Webster 2008b, 140-141.



re f ere n ces Andrikopoulou-Strack, J.-N., 1986: Grabbauten des 1. Jahrhunderts n. Chr. im Rheingebiet. Untersuchungen zu Chronologie und Typologie, Cologne/Bonn (Beihefte Bonner Jahrbücher 43). Belova, N.N., 1987: Die Sklaverei im römischen Gallien, in E.M. Štaerman/V.M. Smirin/N.N. Belova/ J.K. Kolosovskaja (eds), Die Sklaverei in den westlichen Provinzen des römischen Reiches im 1.-3. Jahrhundert, Stuttgart (Übersetzungen ausländischer Arbeiten zur antiken Sklaverei 4), 103-146. Bielman, A./H. Brem/B. Hedinger, 2002: Kultur und Gesellschaft, in L. Flutsch/U. Niffler/F. Rossi (eds), Die Schweiz vom Paläolithikum bis zum frühen Mittelalter. V Römische Zeit, Basel, 267-304. Binsfeld, W., 1974: Ein römisches Kindergrab, Kurtrierisches Jahrbuch 14, 126-127. Binsfeld, A., 2006/2007: Arbeit – Status – Repräsentation. Sklaven und Freigelassene in Inschriften und Grabdenkmälern des Treverergebietes, Trierer Zeitschrift 69/70, 167-176. Boppert, W., 1992a: Militärische Grabdenkmäler aus Mainz und Umgebung, Mainz (Corpus Signorum Imperii Romani II.5). Boppert, W., 1992b: Zivile Grabsteine aus Mainz und Umgebung, Mainz (Corpus Signorum Imperii Romani II.6). Carroll, M., 2006: Spirits of the dead. Roman funerary commemoration in Western Europe, Oxford (Oxford Studies in Ancient Documents). Cordie, R., 2007: Die Tempelbezirke von Belginum, in M. Hegewisch (ed.), Krieg und Frieden. KeltenRömer-Germanen, Bonn, 286-289. De Boer, E./H. Hiddink, 2009: Opgravingen aan de Ter Hofstadlaan te Someren. Een nederzetting en grafveld uit de Late IJzertijd en Romeinse tijd en erven uit de Volle Middeleeuwen, Amsterdam (Zuidnederlandse Archeologische Rapporten 37). Duval, S., 2008: La défunte aux entraves. L’inhumation d’une esclave de la fin de l’Âge du Fer, Préhistoires Méditerranéennes 14, 1-14. Ebnöther, Chr./J. Monnier, 2002: Ländliche Besiedlung und Landwirtschaft, in Die Schweiz vom Paläolithikum bis zum frühen Mittelalter 5, Römische Zeit, Basel, 135-178. Feugère, M., 2008: Maîtres et esclaves dans l’économie de la Gaule méridionale, Instrumentum 27, 14-16. Fremersdorf, F., 1933: Der römische Gutshof Köln-Müngersdorf, Berlin (Römisch-Germanische Forschungen 6). Gaitzsch, W., 1986: Grundformen römischer Landsiedlungen im Westen der CCAA, Bonner Jahrbücher 186, 397-427. George, M., 2010: Archaeology and Roman slavery: problems and potential, in H. Heinen (ed.), Antike Sklaverei. Rückblick und Ausblick. Neue Beiträge zur Forschungsgeschichte und zur Erschliessung der archäologischen Zeugnisse, Stuttgart, 141-160. Harris, W.V., 1999: Demography, geography and the sources of Roman slaves, Journal of Roman Studies 89, 62-75. Hopkins, K., 1978: Conquerors and slaves, Cambridge. Horn, H.G. (ed.), 1987: Die Römer in Nordrhein-Westfalen, Stuttgart. Joshel, S., 1992: Work, identity, and legal status at Rome. A study of the occupational inscriptions, Norman (Oklahoma Studies in Classical Culture 11). Krause, D., 2006: Eisenzeitlicher Kulturwandel und Romanisierung im Mosel-Eifel-Raum, Mainz (RömischGermanische Forschungen 63). Lazzaro, L., 1993: Esclaves et affranchis en Belgique et Germanies romaines d’après les sources épigraphiques, Paris (Centre de Recherches d’Histoire Ancienne 102). Le Glay, M., 1991: Notes sur le “Testament du Lingon”, in Y. Le Bohec (ed.), Le testament du Lingon, Lyon (Collection du Centre d’Etudes Romaines et Gallo-romaines, Nouvelle Série 9), 57-61. Mouritsen, H., 2005: Freedmen and decurions. Epitaphs and social history in Imperial Italy, Journal of Roman Studies 95, 38-63.



Neuffer, E., 1932: Ein römisches Familiengrabmal von Nickenich bei Andernach, Germania 16, 22-28. Neukom, C., [2009]: Einer der drei grössten römischen Gutshöfe der Schweiz: die Villa Munzach in Liestal, (http://villamunzach.ch consulted on May 5, 2009). Ouzoulias, P., 2006: L’économie agraire de la Gaule. Aperçus historiographiques et perspectives archéologiques (Ph.D.-thesis Université de Franche-Comté, available online at http://tel.archives-ouvertes.fr/ docs/00/05/83/PDF/Texte.pdf. Rosafio, P., 1994 : Slaves and coloni in the villa system, in J. Carlsen/P. Orsted/J.E. Skydsgaard (eds), Land use in the Roman empire, Rome (Analecta Romana Instituti Danici, suppl. 22), 145-158. Rothenhöfer, P., 2005: Die Wirtschaftsstrukturen im südlichen Niedergermanien. Untersuchungen zur Entwicklung eines Wirtschaftsraumes an der Peripherie des Imperium Romanum, Rahden (Kölner Studien zur Archäologie der römischen Provinzen 7). Sage, P., 1991: Le testament du Lingon: remarques sur le texte et sur son interpretation, in Y. Le Bohec (ed.), Le Testament du Lingon, Lyon (Collection du Centre d’Etudes Romaines et Gallo-romaines, Nouvelle Série 9), 17-140. Saller, R.P./B.D. Shaw, 1984:Tombstones and Roman family relations in the Principate. Civilians, soldiers and slaves, Journal of Roman Studies 74, 124-156. Samson, R., 1989: Rural slavery, inscriptions, archaeology and Marx: a response to Ramsey MacMullen’s “Late Roman slavery”, Historia 38, 99-110. Schumacher, L., 2001: Sklaverei in der Antike. Alltag und Schicksal der Unfreien, München. Smith, J.T., 1997: Roman villas. A study in social structure, London/New York. Ternes, Ch.-M, [1974]: Das römische Luxemburg, Küsnacht-Zürich. Thompson, H., 1993: Iron Age and Roman slave-shackles, Archaeological Journal 150, 57-168. Thompson, F.H., 2003: The archaeology of Greek and Roman slavery, London (Reports of the Research Committee of the Society of Antiquaries of London 66). Von Massow, W., 1932: Die Grabmäler von Neumagen, Berlin. Vos, W.K., 2009: Bataafs platteland. Het Romeinse nederzettingslandschap in het Nederlandse Kromme-Rijngebied, Amersfoort (Nederlandse Archeologische Rapporten 35), available online at http://hdl.handle. net/1871/13283. Webster, J., 2001: Creolizing the Roman provinces, American Journal of Archaeology 105, 209-225. Webster, J., 2005: Archaeologies of slavery and servitude. Bringing ‘New World’ perspectives to Roman Britain, Journal of Roman Archaeology 18, 161-179. Webster, J., 2008a: Less beloved. Roman archaeology, slavery and the failure to compare, Archaeological Dialogues 15.2, 103-123. Webster, J., 2008b: Slavery, archaeology and the politics of analogy, Archaeological Dialogues 15.2, 139-149. Weeber, K.-W., 2000: Alltag im Alten Rom. Das Landleben, Düsseldorf/Zürich. Wegner, H.-H., 1990: Nickenich, in H. Cüppers (ed.), Die Römer in Rheinland-Pfalz, Stuttgart, 507-508. Whittaker, C.R., 1980: Rural labour in three Roman provinces, in P. Garnsey (ed.), Non-slave labour in the Greco-Roman world, Cambridge (Cambridge Philological Society Supplementary Volume 6), 73-99. Wickham, Chr., 2005: Framing the Early Middle Ages. Europe and the Mediterranean 400-800, Oxford. Wightman, E.M., 1978: Peasants and potentates. An investigation of social structure and land tenure in Roman Gaul, American Journal of Ancient History 3, 97-128.



The idea of the villa. Reassessing villa development in south-east Britain Jeremy Taylor 1 Introduction 2 Rhythms of change: the architectural development of villas 3 The villa and the public landscape 4 Villas and the ‘romance of storage’ 5 Villas as the locus for agricultural expansion 6 Concluding comments References

1 i n troductio n A look at the landscape of south-eastern Britain in the Roman period shows a widespread, if uneven, pattern of the stone founded, sometimes elaborately furnished buildings that archaeologists call villas (fig. 1). But what are we to make of these buildings and their significance to the landscapes of the western Roman provinces? Over the years much archaeological debate about villas has tended to centre on two rather different perspectives. In one, the villa is seen as primarily an economic institution; the core of an estate whose presence is an indication of a ‘Romanised’ agricultural landscape producing food and other materials for the urban and military markets of the empire. Implicit in some of this work has been a tendency to see villas as de facto indicators of agricultural improvement in the post-conquest landscape. In the second view villas are seen more as social phenomenon in which they are regarded as a form of cultured status display.1 Thus we have seen much focus on the developing architecture and spatial organisation of villa houses leading sometimes to rather sterile speculation on the status of their owners or the detailed nature of their resident household(s). Whilst both approaches have value, alone each represents an unnecessarily constrained view of the idea of the villa and its importance in social discourses about becoming Roman in the western Roman provinces. In an insightful essay on the subject Nicholas Purcell noted that such a distinction was unnecessary to Roman eyes, rightly concluding that ‘the choice of the villa owner is not between on the one hand quietly getting on with the agricultural job and on the other making a splash through elegant decoration; agriculture and elegance are alternative forms of display’.2 In developing the ideal of the villa there was in practice no incompatibility between the care with which the house and the agricultural facilities were designed.3

1 

E.g. Agache 2008; Gros 2001; Mattingly 2006, 370-373.

3 

Purcell 1995, 160.

2 

Purcell 1995, 151.



s

Thame

4 ay

Cr

3 1 ent

Dar

Sto ur

y

le

Gt.

Medwa

Mo

Wey

2

Aru

n

Rother

0

40 km

e

Ous

5

Fig. 1. Distribution map of probable Roman villas in the south-east of England.

In this paper I want to focus on this dual view of the villa and how the study of rural sites more generally can help us to understand the extent to which such ideas were enacted, adapted or ignored in the western provinces. In the overview that follows, I want to focus on three aspects of this process at work. First is the development of rural settlement architecture in order to understand how not just domestic buildings but also entire settlements were manipulated by their occupants as part of an ongoing social discourse across the rural landscape of the Empire.4 Whilst there is a continuing excellent tradition of such work in urban contexts in the western empire,5 there has been less obvious recognition of the potential of such an approach in rural contexts. Second is the understudied evidence for the wider productive infrastructure of such sites in order to assess not so much their economic capacity as the extent to which the conspicuous display of agricultural production, processing and storage may have been an important facet of rural social discourse. Third is to briefly place these insights into the wider context of the evidence for changing patterns of agricultural exploitation in the region in order to see whether the changes noted above are matched by concomitant transformations in wider agricultural practice. In doing so (and in keeping with the focus of this volume), the primary focus is on those rural settlements that saw the development of architectural manifestations that we typically associate with the villa, namely the presence of stone founded buildings, mosaics, hypocausts and baths.Throughout, however, we should remember that these settlements represent only part of a much more diverse landscape in which a wide variety of other social discourses were also involved.Villas, as loosely categorised above, were only ever the minority of rural settlements in the south-east6 even if they represent a heavily over-studied group. To illustrate this I wish to look afresh at some of the archaeological evidence from south-eastern Britain, primarily the modern counties of Kent, Surrey, East and West Sussex, and Hampshire. This area is chosen partly for its simple geographical proximity to the continental studies discussed in the rest of this volume, but also because it was one of the first parts of the island to be incorporated within the Roman Empire and thus might be expected to witness some of the earliest development of villas. 4 

Cf. Taylor 2001, 49.

5 

E.g. Wallace-Hadrill 1994; Creighton 2006; Revell 2008.



2 rh y thms o f cha n ge : the architectural de v elo p me n t o f v illas Most accounts of the rural landscape of the Roman south-east begin by mentioning a group of prominent and early Roman villas on and near the Sussex coast.7 Whilst not wishing to downplay the importance of these sites, most were largely subject to excavation in the 19th and early 20th centuries and as such our evidence for many of them is rather limited. That said the villas at Southwick and Pulborough and the much more recently and extensively excavated palatial complex at Fishbourne were on both a grand scale - and in their emphasis on a central courtyard, extensive reception facilities and in other aspects of their design – more in keeping with southern Gaulish and Mediterranean villas. Despite their evident grandeur and early date of construction, however, they do not form a template that is followed by subsequent developments. With the exception of the early Sussex sites the construction of readily recognisable archaeological traces of villa architecture in the region are primarily a phenomenon of the later 1st and 2nd centuries AD. At present our understanding of the transformation of domestic architecture from timber-built round house to rectilinear buildings in timber, in timber but with stone foundations and ultimately to masonry remains poor, though this will undoubtedly improve with the publication of a number of more recent developer-funded and research excavations such as those at Thurnham, Barcombe, and Beddingham. At Thurnham, for example, the first new building of the Roman period, built in mid-late 1st century AD was a largely timber-framed winged corridor house with a small central range of three rooms.8 Whilst of a form that was totally new to the site, it was not much larger than its predecessor; its ground plan occupied c. 113 m2 while the earlier round house on the site covered roughly 104 m2. It was replaced at the beginning of the 2nd century by a more substantial stone built villa of similar form. More widely, such stone founded villa architecture is constructed primarily from the Flavian period and only becomes widespread during the 2nd century. The architecture of the majority of these buildings tended initially to involve the construction of a simple, though substantial, rectangular building within which lay a single row of rooms. These usually stood behind a portico (sometimes two, front and back) flanked by exaggerated corner pavilions often set forward from the portico to establish a winged façade. These so-called winged corridor villas become a common feature of the rural landscape of south-eastern Britain from the middle of the 2nd century. Whilst our evidence is still fragmentary, it seems the development of villa architecture in the south -east follows a similar trajectory to that seen in parts of northern Gaul9 where after a period in which ephemeral and often difficult to analyse timber buildings remain the norm they are replaced by timber, stone and timber and then masonry buildings of row and winged corridor form. Perhaps more surprisingly, this process of change in south-eastern Britain appears broadly contemporaneous to that in Picardy and some other parts of northern France and Belgium,10 even though Britain was only formally incorporated within the Empire after the Claudian invasion. Furthermore, despite the south-east’s longer tradition of ties with the continent evident in other aspects of its material culture (such as the use of coinage and importation of Gallo-Belgic and Italian drinking and eating vessels) and ceremonial practice (such as richly furnished cremation and inhumation burials), there is little sense that its architectural development was more precocious than other parts of southern Britain.Villas in the East Midlands, for example, also develop in similar ways and at a similar pace from the Flavian to Antonine periods. Together, this 6 

Taylor 2007.

7 

E.g. Black 1987; Drewett et al. 1988; Rudling 2003.

10 

Collart pers comm.; Courbot-Dewerdt 2004; Van Ossel 1992; Woolf 1998.

8 

OAU 1999; Booth forthcoming.

9

 E.g. Collart 1996; Courbot-Dewerdt 2004; 2009; Van Ossel 1992.



suggests that the adoption of stone founded row type and winged corridor architecture was a broadly contemporaneous and widespread process across parts of both northern Gaul and southern Britain. The row type and corridor houses of the late 1st and 2nd centuries in south-east Britain were overwhelmingly of modest proportions at a wider imperial scale and the few truly impressive villas of the region are largely the consequence of long traditions of periodic expansion and embellishment over 100250 years. Furthermore, though we need to be cautious of the rather myopic approach to excavation in the past (where the wider spatial context of the villa was rarely assessed) there is still little evidence that south-eastern villas incorporated large numbers of additional houses to form nucleated settlements or large extended households. Very few of the British villas ever reach the scale of the larger Picardy sites where rows of ancillary buildings and dwellings flank the principal house.The few such examples that do, like Darenth and Bignor, are predominantly a phenomenon of the late Roman period (3rd-4th centuries AD). In most cases the evidence suggests a fairly modest-scale resident population in south-eastern villas. Whilst in the 1st and 2nd centuries many villa houses are modest (though nonetheless striking in the architectural context in which they developed) the construction of baths seems to have been a clear priority from an early date and represents a major focus of investment on many sites. At some sites such as Lullingstone, Darenth, Beddingham, and Keston they are one of the first architectural additions to the first phase of stone-founded house.11 At others, such as Thurnham, Eccles, Minster and Beddington, they appear to be part of the plan from the outset.12 Whilst our understanding of the fine chronology of many sites is comparatively poor, some baths would seem to represent the first or only true masonry building constructed as is suggested by Philp at Hayes or see a substantial masonry bath house attached to a timber building in its initial phase of use as occurs at Rapsley.13 Together this may suggest that a heightened sense of social prestige was attached to bathing and personal grooming in the early Roman period in southeastern Britain, an observation supported by other aspects of the archaeological evidence. In a recent survey of toilet instruments, Eckardt and Crummy noted that, though only adopted at the end of the first century BC they continued to find widespread and habitual use on settlements across southern Britain in the Roman period. Significantly, the use of nail cleaners (along with cosmetic grinders) appears to represent a particularly British social practice long after they were largely abandoned on the continent.14

3 the v illa a n d the p u b lic la n dsca p e In focusing so much on provincial villa plans and the details of their architecture we have potentially missed an equally significant facet of their study, namely their landscape context. In the Mediterranean world at least, the Roman landscape of property depended on the spectacular visibility of a series of focal points of view which announced the power of their proprietors.15 The idea of the villa therefore, at least as represented in the literature of leading Roman writers of the early empire, was of a cultured and productive place that extended to its wider landscape. Given this it would seem valuable to explore the extent to which rural settlements in the provincial landscape of south-east Britain might project such visual statements of social prestige in the wider landscape. The overall distribution of recognised villas is very uneven across the region. In some respects this reflects the intensity of past archaeological investigation. The suburban expansion of south-east London and its neighbouring towns, for example, has contributed to our knowledge of villas in the Darent 11 

Lullingstone: Meates 1979. Darenth: Philp 1984. Bed-

13 

dingham: Rudling 1997; idem 1998. Keston: Philp et al.

14 

1991; idem 1999.

15 

12 

Thurnham, Eccles: Detsicas 1963-1977. Minster: Parfitt et al. 2008; idem 2009. Beddington: Howell 2005.



Hayes: Philp 1973, 87. Rapsley: Hanworth 1968. Eckardt/Crummy 2008, 118. Purcell 1995, 168.

and Medway valleys, and more recently the impact of developing road and rail infrastructure between London and the Channel has transformed our understanding of sites in north and east Kent. It partly, however, appears to reflect genuine differences in the character of the rural landscapes of the region. In Sussex, villas concentrate in three main areas: the fertile coastal plain, the chalk downs and on or near the greensand to the north of the South Downs.16 They are largely absent from the Weald and the fringes of Romney Marsh. In Kent, villas mainly cluster in the north west of the modern county especially in the Medway and Darent valleys and the streams of Swale in north Kent alongside a potentially significant but still little studied group on the Isle of Thanet.17 Again there are few villas around the Weald and surprisingly few around Canterbury and in the Great and Little Stour Valleys. In Surrey, villas are again common on the Greensand belt but there is little evidence of them on the river gravel terraces or along the tributary valleys of the Thames. A closer look at this pattern suggests that simple geographically or economically deterministic explanations do not easily fit the available evidence. Whilst the heavy Wealden and London clays have few villas they are found on heavy soils as at Broadstreet Common and Ashtead in Surrey.18 Equally some but by no means all of the lighter, fertile soils of the region were occupied by villas. There is also little sense that villas cluster around the major towns; indeed there appear to be very few villas close to Canterbury and known examples to the south of London all lie at least 8 km away (though it is important to note that few villas in the south-east of Britain ever lay more than 20 km from the nearest town). Rather, locations along the major river valleys, near the coast or on sites with easy access to the major roads appear more important. Sheldon et al. noted that villas south of London averaged a distance of 7 km from the nearest major road, suggested as being equivalent to an hour on horseback.19 Whilst proximity to a transport route is probable in most cases, this does not seem to extend to intervisibility between them. The villa at Deerton Street, Kent, for example, lies only a little over 1 km from Watling Street and yet was not visible from it.20 The villas of the Medway and Darent valleys lie in locations that were generally secluded from the main roads or coastal routes in their vicinity. Therefore, whilst communications by road or water and accessibility to wider urban centres were considerations, probably more so than the quality of the soil on which they were built,21 this did not extend to a prominent desire for villas to be visible from such places. The evidence available instead suggests three related facets of importance in the location of settlements that were to develop the elaborate architecture of villas. First, the majority of villas (and indeed other rural settlements) subject to modern excavation have shown that they are located on sites already occupied in the Late Iron Age. Along the line of the Channel Tunnel Rail Link (CTRL) in Kent for example, of 30 possible sites of Roman occupation 27 (only 2 of which were villas) appear to have preconquest origins.22 Whilst we should be wary of necessarily equating the presence of Late Iron Age features or material culture with continuity of occupation in the Roman period, it is clear in several of the more recently (and extensively) excavated sites that there is a strong physical relationship between them. Frequently, there is clear spatial ‘respect’ for the layout of the former Iron Age or mid 1st century AD settlements that often had an effect in structuring the subsequent layout of the Roman period villa (fig. 2). At Thurnham in Kent, for example, the small timber-framed row type house of the second half of the 1st century AD was constructed outside the western boundary of the former Late Iron Age settlement, and new boundary ditches were dug that redefined and extended the former settlement enclosure.23 A possible shrine or temple was constructed in the south-east corner of the otherwise empty enclosed space of the former settlement. At Keston the main residential block was constructed within the former 16 

20 

17 

21 

18 

22 

19 

23 

Rudling 2003. Cf. Perkins 2001. Bird 2004. Sheldon et al. 1993, 43.

Wilkinson 2005, 26. Rudling 1998, 103. Booth forthcoming. Booth forthcoming.



Late Iron Age settlement enclosure and a similar arrangement is seen at Beddington.24 At Beddingham and Barcombe in Sussex pre-existing round houses and a Bronze Age round barrow are incorporated within the layout in the Roman period.25 In each case the specific histories are different but in all of them former aspects of the settlement were adapted and re-emphasised within the layout of the later villa. Settlement architecture varies as a consequence, but the importance of former enclosures often leads to the adoption of an enclosed or courtyard plan by the villa in the Roman period. Such plans can be seen at a number of Sussex villas (e.g. Batten Hanger, Up Marden, Chilgrove 1 and 2, Barcombe and Beddingham) but also at sites such as Minster in Thanet. In this respect many of the south-eastern British villas provide a local variation on a theme seen in the development of many Gallo-Roman villas in Picardy and elsewhere in northern France.26 In each there is an element in which the memory of place is respected and thus reflected in the later layout of villas, in so doing giving them a plan form that is distinctive and different from villas seen in southern Gaul and Italy. Whilst the importance of enclosures may explain the particular form some villas took in the southeast, it does not itself explain why these sites, rather than others, developed into villas during the Roman period. Many enclosed settlements existed in the Late Iron Age landscape of the region but did not develop into villas. Whilst a detailed comparative assessment of the portable material culture from these sites might demonstrate differences in status between them in the Late Iron Age or 1st century AD, this has yet to be done and lies beyond the scope of this paper. At present, the nearest we have to such a study comes from the CTRL sites in which it was noted that imports were very rare on all settlements of Late Iron Age and very early Roman date but that the few examples of Terra Rubra, Terra Nigra and Gaulish white wares present were largely confined to Thurnham, the one site to subsequently develop as a villa.27 There are, however, two further facets of the landscape setting of most villa settlements in need of consideration. Whilst good agricultural soils per se cannot be considered to have been a major factor, Bird has noted that villas are frequently set at or near geological boundaries that create varied soil conditions in their vicinity.28 This may well suggest a desire to locate villas at points in the landscape in which mixed patterns of land use could be exploited and or witnessed. In this respect the settings of many villas do reflect our knowledge of their agricultural activities in that in the majority of cases in southern Britain mixed arable and pastoral farming was the norm and had been during the Late Iron Age. Equally, it may reflect local understandings of the desirability expressed by Roman authors such as Pliny the Younger29 for a diverse range of agricultural activities and land use to be present within the purview of the house and its owner. Where the direction of the main architectural façade or façades of villas can be determined, they overwhelmingly show that a wider vista was of importance to their setting. The Medway and Darent valley villas look out across the valley or down along its length. Again specific settings differ from place to place across the region; for example, whilst riverside locations seem important in the Darent Valley, this is not so in the Mole or Cray Valleys.30 But where wider evidence is available, we get a sense of the way that fields, tracks and paddocks in the immediate vicinity were structured to enhance a sense of approach reflecting the importance of these villas as places of reception in the countryside.

24 

27 

25 

Beddingham: Rudling 1998. Barcombe: Rudling/Butler

28 

2002; Rudling 2003.

29 

Keston: Philp et al. 1999. Beddington: Howell 2005.

26 

Haselgrove 1995; idem 2007; Collart 1996; CourbotDewerdt 2004.



Booth forthcoming. Bird 2000; idem 2004, 83. Plin., Letters V.6.

30 

Bird 2004.

a)

b)

c)

0

30m

d)

Fig. 2. Plans of the villas at Thurnham (a), Beddington (b), Chilgrove 2 (c) and Batten Hanger (d). The ditches shown in outline are of Late Iron Age or Early Roman date.



v illas a n d the ‘ roma n ce o f storage ’ 31

4

Whilst many 2nd to 3rd century AD villas in the region are relatively modest establishments, an interesting and ever more evident aspect of their wider settlement architecture is the significance afforded to the construction of one or two further buildings usually in the form of three-aisled halls or other distinctive forms of post-built or stone-founded halls and barns. Often overlooked in our haste to admire the decorative and architectural sophistication of the main ranges of many villas, these buildings were themselves a major expenditure of effort (and cost). Increasingly prominent in the wider landscape of villas from the mid-2nd century AD (though examples such as at Hog Brook in Kent dating to the later 1st century are known), these buildings are a distinctive facet of many British villas in south-eastern Britain from Hampshire to the Humber estuary. Compared to the sometimes modest dimensions and architectural elaboration of the row or corridor houses, many such structures are of a very substantial scale. As Chris Martins has pointed out through his analysis of Dalton Parlours and Winterton elsewhere in Britain they sometimes represented a far more ‘costly’ project than the corridor houses of which we are so enamoured.32 Ranging from 12 to 49 m long, they can, as Cunliffe has noted,33 hardly be regarded as sheds but rather were grand constructions that, at upwards of 6 m high, were built to impress. Indeed, King’s reconstruction of the fallen gable end of an aisled building at Meonstoke in Hampshire suggests a roof apex height of nearly 12 m.34 When we think of these as three-dimensional standing buildings rather than in plan form alone, they sometimes dwarf the ‘main’ buildings with which they are associated, as at Chilgrove 2 in Sussex, and must have constituted imposing, highly visible monuments in the wider countryside. Not all such buildings were of the three-aisled form but were still significant architectural edifices in their own right as can be seen in the granaries at Lullingstone and Horton Kirby (fig. 3).35 The significance of these buildings in the landscape of the villas in Kent may best be seen in their prominent location within the wider setting of each settlement. Most are placed to the fore, to one side but in front of the main villa façade, usually flanking one side of the route of entrance to the villa or placed in a visibly prominent position in its foreground; passage to and from the villa inevitably brought one past these imposing displays of the agricultural potential of the site. At Thurnham, for example, a cobbled track approached the villa from the south-east, heading directly towards a large post-built building before seemingly turning to the west past an aisled building and on to approach the main residential block. At Wingham an aisled building sat on the northern side of a wide courtyard flanking the main house to the west. This courtyard was entered from the east uphill along a trackway from the nearby Roman road.36 A similar situation is seen at Keston where a wide funnel-shaped entrance to the villa is flanked by large ancillary post-built buildings on both sides.37 At Lullingstone, the villa looked down over the nearby river with its granary lying close to a timber revetted river bank or quay and may well even have been approached by water, whilst at Darenth and Bignor the large late Roman villas are flanked by very large aisled buildings arranged in and around a series of courtyards.38 At Deerton Street the main residential block looks across the Hog Brook to a particularly early and well built aisled hall and at Broad Street Common in Surrey aisled buildings were possibly located on the opposite bank of a stream in front or immediately to one side of a contemporaneous and rather modest row type house.39 In different ways all of these places created a sense of processional theatre in which the halls and granaries were to the fore. 31 

36 

32 

37 

33 

38 

Purcell 1995, 169. Martins 2005, 24. Cunliffe 2008a, 117.

34 

King 1996.

35 

Chilgrove: Down 1979. Lullingstone: Meates 1979. Horton Kirby: Philp/Mills 1991.



Jenkins 1985, 97. Philp et al. 1999. Lullingstone: Philp/Chenery 2006. Darenth: Philp 1973. Bignor: Aldsworth/Rudling 1996; Rudling 1998.

39 

Deerton Street: Wilkinson 2005. Broad Street Common: Poulton 2005.

b)

a) 0

20 m

Fig. 3. Plans of the granaries at Lullingstone and Horton Kirby.

Closer inspection of the function of these buildings shows that many (especially in northern Kent) represented substantial displays of agricultural wealth. At Horton Kirby, Philp and Mills estimated that the granary had c. 220 m2 of storage space even taking into account the need for access gangways, and similar calculations for Lullingstone give a figure of about 165 m2.40 With storage in bins or sacks up to a height of only c.1 m, these granaries had the potential to store truly impressive quantities of grain (somewhere between 150 and 220 tonnes). We can rarely be as specific in identifying the functions of aisled buildings, but the general absence of evidence for stalling and the contrasting abundance of evidence for ovens, corn driers and large open spaces within the main hall suggest they were primarily involved in the processing and storage of arable produce or fodder. In south-east Britain the idea of the villa as a visible manifestation of wealth and status can be seen in the emphasis given to architectural display of the main house and construction of baths from the 2nd century but also critically in the scale of their agricultural architecture, technology, and works of agricultural mastery. With a few exceptions though, it is only from the later 2nd century that we start to see the more widespread development of large processing and storage facilities. One of the clearest indications of this is in the construction and use of processing facilities in the form of not only barns, but large corn driers or malting ovens and even mills. Small ovens, corn driers or malting kilns are a common feature on many rural settlements across the region from the early 2nd century, but by the 3rd century some are constructed that are substantial structures in their own right. These were sometimes free standing but were often inserted within large aisled halls or other similar buildings (e.g. at Darenth; Horton Kirby and Belloc Road, Littlehampton)41 where the threshing, parching and cleaning of grain could be carried out under cover, thus facilitating large-scale processing. This factor and their sheer size (the two Darenth examples were each approximately 6.5 m long and 3 m wide) suggest that they can only have been intended to work on an industrial scale both to produce fully threshed grain (especially spelt wheat) and malt for ale production.42 40 

Philp/Mills 1991, 16.

42 

Campbell in Cunliffe 2008a, 53-73.

41 

Gilkes 1993, 11.



Evidence for mills is rather rarer, possibly partly because many forms of draft milling such as the donkey mill do not necessarily leave readily identifiable earth fast archaeological remains and partly because archaeologists have not tended to look for them. Water mills are known from Ickham in Kent and Fullerton in Hampshire,43 but we may have indirect evidence for the scale of crop processing on or near more sites in the presence of large millstones. Defining a point at which a rotary quern is likely to have been so large as to qualify as a mechanically driven or draft driven millstone is difficult, but David King, reviewing a wide range of evidence from eastern England, suggested that anything with a diameter of greater than approximately 56 cm would have been too large to operate by hand.44 More recently, Shaffrey has provided some broad support for this figure.45 Looking at Old Red Sandstone querns she suggested that the presence of a rynd slot indicated that millstones were mechanically rotated and of the 8 specimens she recorded all had a diameter of greater than 60 cm. If we adopt a figure of around 65 cm as a conservative lower limit for evidence of mechanical or draft milling, then we may posit the presence of large mills in or near the villas at Darenth, Keston, Lullingstone, The Mount Maidstone and Thurnham in Kent, as well as at West Blatchington and Littlehampton in Sussex and at Grateley and Houghton Down in Hampshire.46 The architectural scale and increasing visual prominence of these facilities on villas from the late 2nd century suggest that their significance went beyond purely functional agricultural purposes. Many of the early aisled halls were timber-framed buildings that created a simple and flexible space, but Cunliffe’s work in Hampshire shows the considerable skill and degree of preplanning that went into their construction.47 Whilst we tend to downplay the relative sophistication of timber architecture in the Roman period, the hall at Grateley, for example, showed that the 18 aisle posts used were massive baulks of timber that when trimmed, were still 0.6 m square.48 Many of these timber-framed buildings were subsequently converted or rebuilt in stone and sometimes provided with plastered walls and tessellated floors.49 Some were built in stone from the outset. A particularly good example is the 35 m long aisled hall at Hog Brook in Kent, which was constructed in the later 1st century AD (seemingly before the first stone built phase of its neighbouring corridor house); from the start it was a sophisticated and expensive undertaking.50 The wall foundations were 1.2-1.5 m deep and between 0.6-0.9 m wide. The walls were constructed of well-laid courses of mortared Kentish ragstone blocks and the internal piers were built from mortared dressed ragstone and tufa with double layering courses of Roman tile to a height of more than 5 m. The interior was faced with a white lime plaster finish and the building was roofed with tile. For its date it would have been one of the largest and most impressively built private buildings in the Kent landscape, and with alterations it remained in use until the end of the 4th century and possibly beyond. In this respect the architecture and placement of the halls was both a practical and symbolic statement. They were the visible manifestations of abundance where agricultural wealth was produced, stored and displayed in an ordered way.51 To borrow from Purcell, these ‘villas are the barns which define the power of control over surplus in ancient agriculture’.52 The aisled hall was an architectural form that could be adapted to a wide range of purposes and many are likely to have been flexible multi-purpose buildings at any one time. Straight forward equations of

43 

Ickham: Spain 1984; Bennett et. al. 2010. Fullerton: Cun-

47 

liffe 2008b.

48 

Cunliffe 2008a, 115-117. Cunliffe 2008a, 117.

44 

49 

45 

50 

46 

Darenth, Keston, Lullingstone: Meates 1979, 110. The

51 

Mount Maidstone: Kelly 1992, 228. Thurnham, West

52 

King 1986, 109. Shaffrey 2006, 30.

Blatchington: Norris/Burstow 1950. Littlehampton: Gilkes 1993. Grately, Houghton Down: Cunliffe 2008b.



E.g. at Dunkirt Barn, Cunliffe 2008d. Hog Brook: Wilkinson 1997; idem 2005. Perring 2002, 54. Purcell 1995, 169.

architectural type with function or status rarely work, as comparable studies carried out in the East Midlands and Hampshire have shown.53 Critical to better understanding the role of these buildings in rural society in Roman Britain is to look at patterns of usage through time and across space. In doing this with the south-eastern halls, there are some suggestions of clear distinctions in the way they were used between the east and west of the region and beyond. In Kent, as we have seen, they are usually part of a suite of buildings within the settlement and in their initial form were often largely working agricultural buildings or barns as at Darenth, Keston, and Thurnham for example. Whilst there is frequently evidence of the likely provision of accommodation in the buildings, initially the utilitarian appearance of surfaces, drains, hearths and ovens in them suggests little evident prosperity among their inhabitants. During the 3rd and 4th centuries, however, the areas for accommodation within them sometimes become reasonably well appointed spaces, but they are still juxtaposed with more utilitarian working spaces in the attached halls. Usually such buildings are found in small numbers with rarely more than one or two on any site, again suggestive of a resident labour force of moderate size. The relative size of the buildings, their prominent position within the settlement and the physical juxtaposition of accommodation and the productive architecture of the estate in them may suggest that their occupants were themselves on display as a ‘resource’ exemplifying the wealth and culture of the owner. In the west of the region, in West Sussex and extending into Hampshire and beyond, we see a somewhat different pattern. At sites such as West Blatchington, Fishbourne Creek, Thruxton, Houghton Down, Stroud, Clanville and North Warnborough, aisled halls were the principal or only residential building. Most began as large open halls but were then modified to create a range of partitioned domestic rooms usually at one end, some of which were provided with tessellated pavements or even mosaics as at Thruxton along with hearths or braziers. Some, such as Clanville and Houghton Down, had private bath suites added or were provided with separate detached baths as at North Warnborough.54 In each case though, part of the central hall was retained and it seems likely that they functioned as places of communal assembly or reception where both the social and economic life of the settlement could be acted out.55 This tradition has parallels further north in parts of the East Midlands where the retention of a hall for communal gathering and work is again seen alongside domestic rooms and small bath suites under the same roof.56 The differences noted above between the development and use of aisled buildings across the region is a good example of how a similar architectural vocabulary was put to a different purpose according to local social context. In Kent and parts of East Sussex they functioned primarily as ancillary buildings and additional accommodation, whilst in West Sussex and Hampshire they usually acted as the main house in which private and public, domestic and agricultural activities were carried out under one roof.

5

v illas as the locus f or agricultural e x p a n sio n

Given the evidence noted above for the increasing importance during the late 2nd to 4th centuries of the display of agricultural surpluses and their ordered processing and storage, we might expect the presence of villas in the landscape to represent unambiguous evidence for the expansion of agricultural production in their vicinity. Here, however, we need to remind ourselves of Millett’s caution that whilst evidence for expenditure on the architecture or facilities at an individual villa is a reflection of wealth, to necessarily equate this wealth with farming in the immediate locale is a mistake.57 As such we cannot necessarily rely on the evidence from an individual excavated site to provide us with a sense of the degree to which the 53 

55 

54 

Thruxton: Cunliffe 2008c. Clanville, Houghton Down:

56 

Cunliffe 2008a. North Warnborough: Liddell 1931.

57 

Resp. Taylor 2001, 50-52; Cunliffe 2008a, 113-128.

Cunliffe 2008a, 126. Taylor 2001, 52. Millett 2007, 152.



development of villas was associated with wider change or improvements in the productivity of agriculture. Instead we need to look at evidence from the wider agricultural landscape as a whole. A nice case in point is that corn driers and malting kilns appear on many rural sites in this period and little in the record as yet suggests they appear any earlier at villas than at other rural sites in the same region. Where we do see a distinction, however, is in their national distribution in that the spread of corn driers and villa architecture are broadly coterminous.58 This would suggest that at a regional scale those agricultural landscapes that saw the development of larger scale grain processing facilities were also those that saw investment in the creation of villa architecture. In Britain at least, therefore, we rarely see investment in the architectural finery of a rural villa outside landscapes that also saw large scale facilities for the processing, storage and display of arable surpluses. A more detailed assessment of the changing nature of agriculture and the agricultural landscape is, however, still difficult to achieve. Within the region under discussion here, archaeobotanical and zoological data are generally in short supply, partly reflecting the antiquity of many of the excavations and partly because some important recent projects await publication. Consequently, it is still difficult to gain a fuller appreciation of the development of agriculture in the Late Iron Age and Roman periods. On the chalklands of Hampshire the results of the Danebury Environs Programmes indicate comparatively little evidence for innovation, technological or otherwise, in arable agriculture in the transition to the Roman period rather than earlier in the Late Iron Age.59 Campbell, for example, shows that in Hampshire crops were grown as maslins until the latest Iron Age when there was a shift to cultivation of pure crops accompanied by a switch from autumn to autumn and spring sowing that then continued throughout the 1st and 2nd centuries AD.60 This change had the capacity to create greater surpluses and both Van der Veen and Jones, and Campbell cite the increasing numbers of grain rich assemblages and of evidence for their storage and processing through the Iron Age and into the Roman period as support for this idea.61 In the absence of evidence for innovation increasing agricultural production may have been achieved through arable expansion (simply cultivating more of the landscape) a continuation of a long term process that had begun in southern Britain in the Iron Age. In Kent on the south west side of the North Downs, distinctive changes in colluviation may have been caused by ploughing of the upper parts of the valley slopes from the Late Iron Age.62 Weeds in botanical samples from three sites along the route of the CTRL further suggest that a greater variety of soil types were being cultivated in the Roman period,63 but it is important to remember that this picture of expansion is far from uniform. Further botanical work from the CTRL project, for example, suggests that the evidence for expansion in cultivation does not seem to have had a dramatic impact on the availability of woodland resources such as oak for fuel and building materials during the Roman period. Evidence for the pastoral economy is if anything poorer, though the common pattern of a shift to greater emphasis on cattle husbandry in the Roman period is seen at a number of sites. Typically, as at Beddingham, Bignor and The Mount Maidstone, cattle become the dominant species from the 2nd century AD and even in the chalkland areas of Hampshire, where sheep husbandry was traditionally very important, we see a slight shift toward cattle. The possible reasons for this have been much discussed, King for example suggesting it reflects changing patterns in diet but it might equally reflect the increasing importance of arable cultivation as cattle make good draft animals as well as being a resource in their own right.64 Hammon suggests the Hampshire data do indicate breed improvement in sheep and cattle 58 

62 

59 

63 

60 

64 

Cf. Taylor 2007, figs 4.9 and 7.3. Jones 1995; Campbell in Cunliffe 2008a. Campbell in Cunliffe 2008a, 74. This development is also paralled in Gaul: Haselgrove pers. comm.

61 

Van der Veen/Jones 2006; Campbell in Cunliffe 2008a, 72-74.



Booth forthcoming. Booth forthcoming. King 1999; idem 2001; Van der Veen/O’Connor 1998.

during the first two centuries AD, though whether this is through better husbandry or the introduction of new breeds, he cannot say.65 Drawing this material together Campbell and Hammon (2008, 100), however, suggest that we should think of 1st-2nd centuries AD as largely a period of the maximisation of innovations in production that had already taken place in years before the Claudian conquest.66 From this viewpoint the early years after the Roman conquest thus represent a period of expansion and exploitation in arable agriculture rather than innovation. Some improved husbandry was evident but the main transformation, clear from the middle of the 2nd century, was in the use of new methods of processing and storage such as malting and corn drying, above ground storage in granaries and barns, draft or water powered milling, the use of cleaver butchery and new meat curing methods. All of these changes meant that the surpluses that the development of agriculture in the Late Iron Age had made possible could now be mobilised more readily to supply non-agricultural communities. Critical to this of course was the new infrastructure of land and water transport available and the affiliation of villas in the region with this network is an indication of its importance as the grease in the wheels.

6 co n cludi n g comme n ts With the exception of the early Sussex villas, an unusual and still poorly understood phenomenon, most of the earlier villas in the British South-East developed within an architectural repertoire that was common to parts of northern Gaul and southern Britain throughout later 1st to 3rd centuries AD. Most developed on the site of former Late Iron Age occupation and remodelled elements of the existing settlement architecture into their subsequent Roman form. The result of this was the creation of a distinctive, though varied British (and Northern Gallic) pattern of ‘enclosed’ villas that probably reflects the distinctive traditions and identity of many of their occupants (though not necessarily their proprietors). Distinctive regional characteristics in their development are relatively few but possibly can be seen in the early emphasis on the construction of baths and the varied but important role of aisled halls especially in the west of the region. Most of the British villas are on a modest scale both in terms of the domestic architecture of the main house and in the number of ancillary buildings and households within the settlement as a whole. Whilst we should be wary of making too straight forward an equation of this evidence with the nature and scale of land holding and agricultural wealth present, it is perhaps justified to suggest that in south-east Britain there is less sense that either were centralised within the hands of a few powerful households at least until the later 3rd and 4th centuries. Even then south-east Britain had little on the scale of the larger Picardy villas. Whilst the architectural grandeur of many south-eastern villas was unspectacular, it is important to recognise that much effort was invested in the construction and display of the productive infrastructure of these settlements, sometimes to the extent that they even outshone their domestic architecture. In this respect many powerful individuals in south-eastern Britain did adapt an important facet of the Roman idea of the villa noted at the beginning of this paper in the conspicuous display of the agricultural productivity of their estates. It is not really until the middle of the 2nd century, however, that this becomes a prominent and widespread feature of the rural landscape. Initially at least, this transformation appears to have been partly achieved not by wealth generated from new and innovative strategies of agricultural production, but by the expansion, reorganisation or intensification of existing strategies. We still have little understanding of how this was done, but critical to the success of these changes was the development of better means of large scale processing and the 65 

Hammon in Cunliffe 2008a, 89-90.

66

Campbell and Hammon in Cunliffe 2008a, 100.



storage of arable surpluses that could now be mobilised through the newly developed transport network of the Roman province and beyond to further the wealth of those that owned them.

re f ere n ces Agache, S., 2008: La ville comme image de soi. Rome antique, des origines à la fin de la République, in P. Galand-Hallyn (ed.), La villa et l’univers familial, de l’Antiquité à la Renaissance, Paris, 15-44. Aldsworth, F.G./D. Rudling, 1995: Excavations at Bignor Roman villa 1985-1990, Sussex Archaeological Collections 133, 103-188. Bennett, P./C. Sparey Green/I.D. Riddler, 2010: The Roman watermills at Ickham, Kent, Canterbury (Canterbury Archaeological Trust Occasional Paper). Bird, D., 2000:The environs of Londinium: roads, roadside settlements and the countryside, in I. Haynes/H. Sheldon/L. Hannigan (eds): London under ground. The Archaeology of a City, Oxford, 151-174. Bird, D., 2004: Roman Surre, Stroud. Black, E. W., 1987: The Roman villas of South-East England, Oxford (BAR, Brit. Ser., 171). Booth, P., forthcoming: CTRL: The Late Iron Age and Roman Periods. Oxford (Oxford Archaeology). Collart, J.-L., 1996: La naissance de la villa en Picardie: la ferme gallo-romaine précoce, in D. Bayard/J.L. Collart (eds), De la ferme indigène à la villa romaine. La romanisation des campagnes de la Gaule. Actes du deuxième colloque de l’association AGER tenu à Amiens (Somme) du 23 au 25 septembre 1993, Amiens (Revue Archéologique de Picardie, numéro special 11), 121-156. Courbot-Dewerdt, C., 2004: L’évolution des établissements ruraux entre la fin de l’Âge du Fer et la mise en place du système des villae dans le quart Nord-Ouest de la France, Ier siècle avant-IIème siècle après J.-C., Paris (PhD Thesis, Université de Paris I Panthéon-Sorbonne, UFR d’histoire). Courbot-Dewerdt, C., 2009: Feeling like home. Romanised rural landscape from a Gallo-Roman point of view, in M. Driessen/S. Heeren/J. Hendriks/F. Kemmers/R. Visser (eds), TRAC 2008. Proceedings of the 18th Annual Theoretical Roman archaeology Conference, Oxford, 13-24. Creighton, J., 2006: Britannia. The Creation of a Roman Province, London. Cunliffe, B., 2008a: The Danebury environs Roman programme. A Wessex landscape during the Roman era, 1, Overview, Oxford. Cunliffe, B., 2008b: The Danebury environs Roman programme. A Wessex landscape during the Roman era, 2, part 3, Fullerton, Hants, 2001 and 2001, Oxford. Cunliffe, B., 2008c: The Danebury environs Roman programme. A Wessex landscape during the Roman era, 2, part 4, Thruxton, Hants, 2002, Oxford. Cunliffe, B., 2008d: The Danebury environs Roman programme. A Wessex landscape during the Roman era, 2, part 7, Dunkirt Barn, Abbotts Ann, Hants, 2005 and 2006, Oxford. Detsicas, A.P., 1963: Excavations at Eccles, 1962. First interim report, Archaeologia Cantiana 78, 125-141. Detsicas, A.P., 1964: Excavations at Eccles, 1963. Second interim report, Archaeologia Cantiana 79, 121-135. Detsicas, A.P., 1965: Excavations at Eccles, 1964. Third interim report, Archaeologia Cantiana 80, 69-91. Detsicas, A.P., 1966: Excavations at Eccles, 1965. Fourth interim report, Archaeologia Cantiana 81, 44-52. Detsicas, A.P., 1967: Excavations at Eccles, 1966, Archaeologia Cantiana 82, 162-178. Detsicas, A.P., 1968: Excavations at Eccles, 1967, Archaeologia Cantiana 83, 39-48. Detsicas, A.P., 1969: Excavations at Eccles, 1968, Archaeologia Cantiana 84, 93-106. Detsicas, A.P., 1970: Excavations at Eccles, 1969, Archaeologia Cantiana 85, 55-60. Detsicas, A.P., 1971: Excavations at Eccles, 1970, Archaeologia Cantiana 86, 25-34. Detsicas, A.P., 1972: Excavations at Eccles, 1971, Archaeologia Cantiana 87, 101-110. Detsicas, A.P., 1973: Excavations at Eccles, 1972, Archaeologia Cantiana 88, 73-80. Detsicas, A.P., 1974: Excavations at Eccles, 1973, Archaeologia Cantiana 89, 119-134.



Detsicas, A.P., 1975: Excavations at Eccles, 1974, Archaeologia Cantiana 91, 41-46. Detsicas, A.P., 1976: Excavations at Eccles, 1975, Archaeologia Cantiana 92, 157-164. Detsicas, A.P., 1977: Excavations at Eccles, 1976, Archaeologia Cantiana 93, 55-60. Detsicas, A.P., 1989: Excavations at Eccles. A progress report, Archaeologia Cantiana 107, 83-88. Down, A., 1979: Chichester excavations, 4, Chichester. Drewett, P./D. Rudling/M. Gardiner, 1988: The South East to AD 1000, London. Eckardt, H./N. Crummy, 2008: Styling the body in Late Iron Age and Roman Britain. A contextual approach to toilet instruments, Montagnac (Monographies Instrumentum 36). Gilkes, O., 1993: Iron Age and Roman Littlehampton, Sussex Archaeological Collections 131, 1-20. Gros, P., 2001: L’architecture romaine, 2. Maisons, palais, villas et tombeaux, Paris. Hanworth, R., 1968: The Roman villa at Rapsley, Ewhurst, Surrey Archaeological Collections 65, 1-70. Haselgrove, C., 1995: Social and symbolic order in the origins and layout of Roman villas in northern Gaul, in J. Metzler/M. Millett/N. Roymans/J. Slofstra (eds), Integration in the Early Roman West, Luxembourg (Dossiers d’Archéologie du Musée National d’Histoire et d’Art 4), 65-75. Haselgrove, C., 2007: The age of enclosure: later Iron Age settlement and society in northern France, in C. Haselgrove/T. Moore (eds), The later Iron Age in Britain and beyond, Oxford, 492-522. Houliston, M., 1999: Excavations at The Mount villa, Maidstone 1994, Archaeologia Cantiana 119, 71-172. Howell, I., (ed.) 2005: Prehistoric landscape to Roman villa. Excavations at Beddington, Surrey 1981-7, London (MoLAS Monograph 26). Jenkins, F., 1985: The re-excavation of the Roman ‘Villa’ at Wingham,1, Archaeologia Cantiana 100, 87-99. Jones, M., 1995: Patterns in agricultural practice: the archaeobotany of Danebury in its wider context, in B. Cunliffe (ed.), Danebury. An Iron Age Hillfort in Hampshire, 6. A hillfort community in perspective, London (CBA Research report 102), 43-53. Kelly, D.B., 1992: The Mount Roman villa, Maidstone, Archaeologia Cantiana 110, 177-237. King, A., 1996: The south-east façade of Meonstoke aisled building, in P. Johnson/I. Haynes (eds), Architecture in Roman Britain, York (CBA Research report 94), 56-69. King, A., 1999: Diet in the Roman world. A regional inter-site comparison of the mammal bones, Journal of Roman Archaeology 12, 168-202. King, A., 2001: The Romanization of diet in the Western Empire. Comparative archaeological studies, in S. Keay/N. Terrenato (eds), Italy and the West. Comparative issues in Romanization, Oxford, 210-223. King, D., 1986: Petrology, dating and distribution of querns and millstones. The results of research in Bedfordshire, Buckinghamshire, Hertfordshire and Middlesex, Bulletin of the Institute of Archaeology London 23, 65-126. Liddell, D.M., 1931: Notes on two excavations in Hampshire, Proceedings of the Hampshire Field Club 10, 224-236. Martins, C., 2005: Becoming consumers. Looking beyond wealth as an explanation of villa variability, Oxford (BAR, Brit. Ser. 403). Mattingly, D., 2006: An imperial possession. Britain in the Roman empire, 54 BC-AD 409, London. Meates, G. W., 1979: The Roman villa at Lullingstone, Kent, I: The site, Maidstone (Kent Archaeological Society Monograph Ser., 1). Millett, M., 2007: Roman Kent, in J.H. Williams (ed.), The archaeology of Kent to AD 800, Woodbridge, 135-186. Norris, N.E.S./G.P. Burstow, 1950: A prehistoric and Romano-British site at West Blatchington, Hove, Sussex Archaeological Collections 89, 1-53. Oxford Archaeological Unit (OAU) 1999: Thurnham Roman villa, Thurnham, Kent, Oxford (ARC THM98. Detailed archaeological works interim report final). Parfitt, K., 2006: The Roman villa at Minster-in-Thanet, 3. The corridor house, building 4, Archaeologia Cantiana 126, 115-133.



Parfitt, K./D. Perkins/E. Boast/G. Moody, 2008: The Roman villa at Minster-in-Thanet, 5. The main house, building 1, Archeologia Cantiana 128, 309-334. Parfitt, K./E. Boast/G. Moody, 2009:The Roman villa at Minster-in-Thanet, 6.The villa enclosure, buildings 2 and 5, Archeologia Cantiana 129, 333-357. Perkins, D. R. J., 2001: The Roman archaeology of the Isle of Thanet, Archaeologia Cantiana 121, 43-60. Perkins, D.R.J., 2004: The Roman villa at Minster-in-Thanet, 1. Introduction and report in the bathhouse, Archaeologia Cantiana 124, 25-49. Perring, D., 2002: The Roman house in Britain, London. Philp, B., 1973: Excavations in West Kent 1960-1970, Dover, 119-154. Philp, B., 1984: Excavations in the Darent Valley, Kent, Dover (Fourth Research Report in the Kent Monograph Ser.). Philp, B./M. Chenery, 2006: Lullingstone and Shoreham. Discoveries at the Roman villa sites 1982-86, Dover. Philp, B./R. Mills, 1991: The Roman villa at Horton Kirby, Dover. Philp, B./K. Parfitt/J. Willson/M. Dutto/W. Williams 1991: The Roman villa site at Keston, Kent. First report (excavations 1968-1978), Dover (Kent Monograph 6). Philp, B./K. Parfitt/J. Willson/W. Williams, 1999: The Roman villa site at Keston, Kent, Second report (excavations 1967 and 1978-1990), Dover (Kent Monograph 8). Poulton, R., 2005: Excavations near Broad Street Common, Worplesdon, Guildford, in 1994, 1997 and 1998, Surrey Archaeological Collections 92, 29-89. Purcell, N., 1995: The Roman villa and the landscape of production, in T.J. Cornell/K. Lomas (eds,) Urban society in Roman Italy, London, 151-179. Revell, L., 2008: Roman imperialism and local identities, Cambridge. Rudling, D., 1997: ‘Round house’ to villa: The Bedingham and Watergate villas, in R.M. Friendship/D.E. Taylor (eds), From round house to villa, Northampton, 1-8. Rudling, D., 1998: The Roman villas of Sussex, with particular reference to the villas at Beddingham and Bignor, in X. Delestre/A. Woodcock (eds), Proximus 2. Actes de la Table-ronde Archéologique, Dieppe 17 et 18 septembre 1996, Rouen, 97-112. Rudling, D., 2003: The Archaeology of Sussex to AD 2000, Kings Lynn. Rudling, D./C. Butler 2002: Barcombe Roman villa, Current Archaeology 179, 486-489. Shaffrey, R., 2006: Grinding and milling. A study of Romano-British rotary querns and millstones made from Old Red Sandstone, Oxford (BAR, Brit. Ser. 409). Sheldon, H./G. Corti/D.Green/P. Tyres, 1993: The distribution of villas in Kent, Surrey and Sussex. Some preliminary findings from a survey, London Archaeologist 7.2, 40-46. Spain, R J., 1984: Romano-British watermills, Archaeologia Cantiana 100, 101-128. Taylor, J., 2001: Rural society in Roman Britain, in S. James/M. Millett (eds), Britons and Romans. Advancing an archaeological agenda, York (CBA Research report 125), 46-59. Taylor, J., 2007: An atlas of Roman rural settlement in England, York (CBA Research report 151). Van der Veen, M./G. Jones, 2006: A re-analysis of agricultural production and consumption. Implications for understanding the British Iron Age, Vegetation History and Archaeobotany 15, 217-228. Van der Veen, M./T. O’Connor, 1998: The expansion of agricultural production in Late Iron Age and Roman Britain, in J. Bayley (ed.), Science in archaeology. An agenda for the future, London, 127-144. Van Ossel, P., 1992: Etablissements ruraux de l’Antiquité tardive dans le nord de la Gaule, Paris (Gallia, suppl. 51). Wallace-Hadrill, A., 1994: Houses and society in Pompeii and Herculaneum, Princeton. Wilkinson, P., 1997: Swale archaeological survey. Interim report on the Roman villa at Deerton Street, Teynham, Kent, Faversham (unpublished manuscript report Kent archaeological fieldschool). Wilkinson, P., 2005: An archaeological investigation of the Roman aisled building at Hog Brook, Deerton Street, Faversham, Kent, Faversham (unpublished manuscript report, Kent Archaeological Fieldschool).



The role of mortuary ritual in the construction of social boundaries by privileged social groups within villa landscapes Laura Crowley1 1 Introduction 2 Different types of privileged burials 3 Regional patterning 4 Lifestyle themes and self-representation of privileged social groups 5 Community and the definition of social boundaries 6 Concluding observations References

1 i n troductio n This article concerns itself with the privileged burials of the villa landscapes between Cologne and Bavay, in a period spanning the 1st to the 3rd centuries AD. The term ‘privileged burials’ derives from francophone scholarship, where it is employed to describe a burial which fulfils one or more of three criteria: 1) it is indicated by a particularly prominent marker; 2) it is accompanied by ‘valuable’ grave goods;2 or 3) it enjoys a special location in the landscape.3 Because of these characteristics, these burials have previously been referred to as elite burials, a label which is deliberately avoided here, for a number of reasons. The first of these is revealed elsewhere in this volume by Roymans and Derks, namely that the majority of villa settlements with which these burials are associated do not belong to the very top level of luxury dwelling and massive enterprise; they belong, rather, to a broad middle class of villa-inhabitants. The second reason for steering away from the notion of elite burials is more of a general caveat within mortuary archaeology: the degree of wealth evident in a burial is by no means a direct correlate of the means of the deceased in his or her lifetime. Some peoples invest more in burying their dead than in their housing, others inter even their wealthiest members in humble pits; as a consequence, we are not looking at one-to-one reflections of lived economic status. A third reason is the problems inherent in defining a villa elite, or indeed, any elite. The term tends to be employed instinctually, without establishing clear criteria, as such criteria tend to be somewhat arbitrary or forced. This has largely to do with an implicit assumption that the lines were blurry in the past too; a plausible hypothesis, but not a particularly useful one for performing an analysis. And the fourth major reason for labelling these burials privileged rather than elite is that some of the evidence is quite conflicting. While some of these sites are undoubtedly special, the actual financial investment in their construction is questionable, which distances us from the concept of the deceased as a member of the elite. 1

gramme Villa landscapes in the Roman North: economy, culture,

This study is based on PhD research conducted at the VU

lifestyles led by Nico Roymans and Ton Derks.

University, Amsterdam (due for publication in 2011/2012), which was funded by the Netherlands Organisation for

2

The subjectivity of ‘value’ notwithstanding.

Scientific Research (NWO) as part of the larger pro-

3

Ferdière 2004, 35.



Fig. 1. Distribution map showing tumuli, so-called ‘sarcophagi’ and evidence of stone grave monuments in the form of foundations and/or fragments in the study area of Bavay to Cologne. Key individual burials and clusters immediately outside of this region have been included with particular reference to Wigg 1993 and Willer 2005.

The privileged burials are, nonetheless, designed to be a different category of burial. They tend to be spatially differentiated from their more humble counterparts, and distant in form, too. But what do they tell us of villa society, of the occupations and preoccupations of the erectors? How did people represent themselves, how did they want to be perceived? What can we learn of inter-generational relationships? Can we see trends emerging regionally, or is the language and message universal? We approach these questions with the help of certain concepts, namely A. Cohen’s symbolic construction of community,4 the related notion of social boundaries, and the idea of personhood.5 Social boundaries are lines drawn by people to distinguish themselves from other (groups of) people. They are abstract, generally, but can have very material manifestations. Such boundaries can demarcate either vertical, for example class distinctions, or horizontal differences, for example regional ones. They do not have to indicate disharmony; to the contrary, where everybody knows and adheres to their boundaries, social balance can be attained. Cohen holds that such social groups, or communities, are symbolically constructed. It is generally acknowledged that communities physically mark their boundaries, not necessarily by building fences or digging ditches, but by presenting visual cues that unite them internally, while advertising their collective distinct identity from the rest, the so-called ‘Other’. However, Cohen’s theory goes much further than this. He asserts that communities are actually created, as well as being understood and reinforced, by means of symbolism. Furthermore, he ascribes ritual an important

4

Cohen 1985.



5

See Fowler 2004 for an extensive bibliography.

position in the ‘repertoire of symbolic devices’6 available to communities for the affirmation and reinforcement of social boundaries. As mortuary practice is ritual by nature, Cohen’s model is particularly relevant. A further useful approach is that of personhood, useful in that it rattles us out of our Western preconceptions regarding individualism, urging us to regard people of the past from a social perspective. Only then can we begin to approach a sense of people’s roles within society, something which tends to be highlighted in mortuary practice. In the course of this short piece, we will proceed to address two key and closely related topics that emerge from the mortuary data, namely identity construction and selfrepresentation. From this perspective, regionalism performs an important task, as does the depiction of lifestyle themes; we will deal with each of these separately. The concepts described above provide useful tools with which to address these questions, and having done so, we will draw some basic conclusions. This contribution will begin by defining and briefly describing the dataset, privileged burials. Having acquired a feel for the material, we will take a look at the regional patterns that emerge. Subsequently, the prominent themes associated with lifestyle become our focus, together with the self-representation apparent in privileged burials/burial markers. After that, we look more closely at the concepts of community and social boundaries as they relate to our dataset, before proceeding to some surprising features among privileged burials. The article concludes with a concise summary.

2 di f f ere n t t y p es o f p ri v ileged b urials Privileged burials abound in the northern villa landscapes; they are ubiquitous, but far from homogeneous. Four distinct categories of privileged burials have been identified in the dataset, as defined above.7 Commencing with the markers, these are stone monuments and tumuli, the subterranean elements comprise so-called sarcophagi and costly grave gift assemblages. Grave gifts tend to accompany tumulus burials and sarcophagi (cf. below and fig. 4); we rarely find them surviving in the vicinity of a stone monument, nor stumble upon them in isolation. Figure 1 shows the occurrence of these forms of privileged burial, and reveals distinct clustering within the study area. The region shown covers a cross-section of the fertile loess soils which enabled the large-scale farming that was the economic backbone of most villa-enterprises. Where villas occurred, privileged burials occurred also, and in an increasing number of cases direct relationships are traceable.8 Indirect links, too, are evident, and we will discuss these presently. All of our categories of evidence occur either individually or in very limited groups. Chronologically speaking, the tumuli range from the mid 1st to the mid 3rd centuries AD, the tower tombs to the 1st, and the subsequent grave pillars to the late 1st to 3rd centuries. The sarcophagi in question date to the late 2nd and early 3rd centuries AD. As far as the representativeness of the map is concerned, certain points are important to note. The first of these has to do with the history of research. Preceding studies have tended to focus on a modern particular region, resulting in extremely well-researched areas and more neglected ones. As our study region is derived largely from its geological properties, the inclusion of all of these areas will, then, contain a certain degree of bias in terms of intensity of research. Poor dating methods, often employing a distinctly circular reasoning, can be traced in all categories of evidence. This has consequences, not only for establishing developments during our focus period, but also in the decision of whether or not to

6

Cohen 1985, 50.

bordering a main thoroughfare, for example, it may also

7

Criteria 1 and 2 tend to be primary selectors, with crite-

be determined by its situation in the mythical landscape,

rion 3 as an accompanying factor. This has largely to do with the subjectivity of the perception of a location as

a particularly elusive aspect of the archaeological record. 8

See also Krier/Heinrich (this volume).

being special. While it may have to do with immediately



Fig. 2. Tower tomb of Lucius Poblicius from Cologne (left) and grave pillar of the Secundinii from Igel (right) (after Panhuysen 1996, kaart V).

include a particular burial in our study at all. This can result in the perhaps over-cautious exclusion of relevant sites (in particular sarcophagi).9 As mentioned, the stone monuments fall largely into one of two categories: that of what we shall refer to as tower tombs, and that which we call grave pillars (fig. 2).10 The tower tombs, with their military allusions, were introduced in the course of the 1st century AD by the officer classes, probably veterans.11 These were mausolea proper,12 that is, they were architectural structures housing the cremated remains of the deceased. Distinctly Roman-Hellenic in appearance, they were raised only for a relatively limited period in our region. Hot on their heels followed the grave pillars, which can be viewed as the civilian adoption and adaptation of the monumental stone burial trend. Taller and narrower, with iconography on all four sides, rather than only on their frontal aspect, they took the form of pillars, marking rather than containing the burial(s). Issues of scale require us to focus on these two monumental forms, deliberately omitting the multitude of smaller stele that marked graves to a varying extent throughout the region. Furthermore, these two forms can be undisputedly labelled ‘privileged’. The issue of stone monuments and their nigh on universal demolition and reuse as spolia is significant. Taken apart in the late Roman period and incorporated into Late Roman defensive structures of civitas capitals and small towns, we find their fragmentary remains mostly ex-situ. How far they are ex-situ is 3m 2

2m

1

9

10

1

1m

In rare instances, mortuary features of a later date may be

of the problems inherent in researching in this area.

included.

11

Gabelmann 1973; Roymans (this volume).

See Crowley (in prep.) for an overview, as well as Pan-

12

Not in the literal sense of resembling the original mau-

huysen 1996; Andrikopoulou-Strack 1986; Gabelmann

soleum (the burial monument of King Mausolus of

1979. Willer 2005, 3, note 8, lists a confusing repertoire

Halicarnassus), but in that they actually contained human

of different terms used in various languages to describe

remains, rather than functioning as markers of the loca-

monuments which fall into her category of ‘aedicula-

tion where these were interred.

type’ grave markers. In so doing, she illustrates just one



Fig. 3. One of the twin tumuli at Koninksem, near Tongeren (B). Photo author.

virtually impossible to ascertain, which introduces the probability of a dataset contaminated with monuments from originally extra-urban contexts, rather than the rural markers we seek. The evidence for insitu monuments is drawn from their foundations, a form of trace in the archaeological record which has only recently been fully recognised for what it is, and as a consequence has only begun to be identified in modern excavations. The interpretation of foundations and spolia owes a huge debt to the survival of the Igel pillar, located in the civitas Treverorum, not far to the southeast of our region (fig. 2). This is the only intact grave pillar to have survived, its 23 metre high presence providing us not only with an impression of the grandeur of such monuments, but also providing us with epigraphic information and a wealth of iconographic evidence for life and commerce in the countryside. This example helps us to extrapolate from our somewhat meagre dataset of tower tombs and grave pillars. Tumuli are a less problematic category in terms of their original location, but still require a degree of prudence in their analysis.13 Striking in their scale, our burial mounds are typically about 11 m in height, with some examples stretching as high as 15 m, and measuring anything between 10 m and 30 m in diameter (fig. 3). Having been recognised as special entities more or less since their construction, they have not fallen victim to oversight. That fact has worked to their disadvantage too: many show evidence of looter’s tunnels, and we can safely say that many have been robbed of their content.This notwithstanding, some tumuli simply contained less grave gifts than others, some appear to have been constructed above ustrina alone, without any substantial human remains, and some seem to be ‘empty’ of any features whatsoever. There are many theories as to why this may have been the case, but as we are looking at the wider phenomenon of privileging burials, beyond their immediate manifestation, we must also pay attention to associated – from the outside, identical – phenomena in the landscape. Again, we remain focused

13

M. Amand has produced a vast and invaluable corpus on

see the bibliography below. See Crowley (in prep.) for a

the tumuli of our region. For a small sample of references,

more comprehensive list.



Fig. 4. Reconstruction of the wooden funerary chamber of Bocholtz with sarcophagus and assemblage of grave gifts among which a horse bit, an axe, a pair of strigils, a hunting lance, an inkwell, glass vessels and a folding chair (Image courtesy of State Service for Cultural Heritage, Amersfoort).

on the most striking category of tumulus; small low barrows of the type rife in the Iron Age (albeit rare in our region) do not feature strongly in our study.14 The third category of data, so-called sarcophagi,15 is a challenge to locate, as these vessels often bear no trace of an external marker. The term sarcophagus, strictly speaking a limestone coffin for a corpse which is often designed to be free-standing in a hypogeum, is employed in our region somewhat inaccurately to describe coffins of varying stone within which cremated remains are interred, often together with grave gifts (fig. 4).16 Whether the absence of a surface marker was a strategy to safeguard them from plunderers or whether their markers did not survive the intervening millennia is a difficult question. That said, the combination of the strict laws in place protecting burials and the need to re-visit family graves would indicate that they were most likely to have been marked in some way.17 An even more problematic issue, in terms of the distribution map, is their dating.While sarcophagi are a common feature within mortuary ritual from the late 3rd century onwards – when inhumation became the dominant burial trend – those on which we focus predate the masses, and stand out in containing not a corpse, but cremated remains.

14

avoid confusion and provide consistency with extant

Low mounds are less visibly privileged, and reveal less

studies.

specific distribution patterns. 15

Gottschalk 2003.

16

Inaccuracy notwithstanding, we retain the term here to



17

See for example De Groot 2006. 

3 regio n al p atter n i n g One glance at figure 1 reveals distinct regional patterning in our dataset. The tower tombs and grave pillars cluster to the east, along the Lower Rhine, with a handful of outliers along the south west border. Similarly, the tumuli form a concentration towards the west of our region, within the hypothesised boundaries of the civitas Tungrorum in particular. Nestled more or less between these two groupings we find a strip of sarcophagi running from north to south, marking a sort of transitional zone. It is important to bear in mind that we are dealing explicitly with privileged burials here, by definition excluding simple steles, or tombstones, clusters of very low-lying burial mounds, which are essentially hillocks, and small stone cinerary urns. These entities know a somewhat less distinctive distribution pattern, and while they are, to a greater or lesser extent, related to our respective privileged categories of evidence, each are significantly less grand in scale, and can, arguably, be interpreted as representing distinct phenomena. As such, the striking clusters hold. How can we explain this spatial patterning? First of all, let us exclude some possibilities, commencing with means. One could imagine a scenario in which the top category, namely that comprising the wealthiest members of society, chose to erect beautifully crafted massive stone monuments, with the slightly less well-off opting for tumulus-raising, and the group propping up the bottom of the privileged group commissioning sarcophagi.This is not the case. We can assert this with some confidence because we see an entire range of means even within each of the three privileged forms: monuments towering ten metres higher than others; tumuli with volumes of everyday ceramics versus those containing precious metals; roughly-hewn sarcophagi compared with finely engraved ones, to name but a few examples along the apparent spectrum of value, on a purely economic level. Furthermore, we must exercise extreme caution when referring to ‘privileged groups’ building ‘privileged burials’. These two sets certainly overlap, but cannot be assumed to be one and the same. Just as we can extract from the sheer numbers that not every member of a privileged group was accorded a privileged burial, so are there also indications that not every privileged burial is that of a member of the privileged groups. More on this will follow below. Another possible explanatory factor is the availability of raw materials. While we do see a distinct proliferation of stone monuments where stone is readily available, this does not provide an explanation for the cases where resources are imported, nor can investment in such import have been prohibitively costly for those who could afford otherwise extremely lavish burials. The only conclusion is that the manner of privileging a burial was a result of conscious selection, but that conclusion only provides a starting point. Where it really becomes interesting is the reason for such regional choices. I believe that the answer lies on two levels. The first is one of simple administrative geographical borders.18 The second reason is also related to borders, but is significantly different in that they are borders relating to people; it addresses social boundaries. These draw on a sense of community. They mark not only territorial divisions, horizontally between different peer groups, but also vertical divisions, highlighting their status in the social order. Grave form was not prescriptive, and yet the patterning of grave form seems to follow administrational boundaries. What this tells us is that a real social sense of community arose among the ruling elite of the civitates. A relatively small group of qualified landowners, meeting regularly in the capitals for political purposes, these trendsetters developed their own symbols for community, manifesting them in both towns and countryside. All symbolic burials display strong allusions to Roman cultural influence, whether by means of the stone monuments, the sarcophagi, or the tumuli. However, each category has its own particular local twist: the pillars depict a local repertoire of imagery, in a local style; the so-called sarcophagi with their 18

This is not to suggest that method of burial was deter-

are far from simple administrative ones. See Crowley (in

mined by geographical location, but there is a correlation

prep.).

that cannot be ignored.The reasons behind this, however,



rough exteriors bearing cremated remains embody a local peculiarity; and finally the tumuli, with more than a nod to prehistoric barrows, reveal an atavistic use of symbolism to create a sense of longevity and legitimacy otherwise absent in a newly-born community of the early Roman period.19

4 li f est y le themes a n d sel f - re p rese n tatio n o f p ri v ileged social grou p s Moving from the overall form of the burial to its detail, we encounter recurring themes. Be it in pillar iconography or grave gift assemblage, references to aspects of the rural lifestyle abound, regardless of the medium. These include the practice of hunting – for game both hirsute and feathered – the bathing ritual, personal toilet, self-adornment and gaming, as well as literacy and most ubiquitously dining (fig. 4). Not only leisure activities are reflected, but also economic ones.The artefacts embodying this symbolism include spears, arrowheads, strigils, mirrors, jewellery, game pieces, styluses and considerable numbers of vessels for large scale eating and drinking.20 In iconography the themes are manifest in depictions of hunting parties and their prizes, bathing equipment, scenes of women engaged in grooming rituals, detailed observation of clothing and hairstyles, occasionally scenes of tutoring and dining scenes of several varieties, plus business activities such as the payment and receipt of coin (rent or duty-paying),21 the keeping of accounts, the harvest, transportation of goods and the processing of cloth.22 These objects and images all paint an idyllic image of villa life as it is described in the classical texts, but knowing how symbolic mortuary compositions can be and how far these accounts can lie from provincial realities, to what extent can we take this ideal literally? This is where the villa settlements themselves come in, revealing that there is real substance to the mortuary assemblages in terms of the insights they provide into lifestyles lived.23 Widespread evidence for bathhouses as part of villa complexes provide a concrete context for the use of the many strigils and balsam bottles found interred. Likewise, triclinium-style rooms have been identified in the larger provincial villas, their architecture facilitating a particular style of dining which would have required the vessels which adorn our graves so abundantly,24 and faunal evidence from villa settlements reveals that game was consumed in the residential main houses, again lending weight to the evidence for hunting found in graves.25 Such strong links between the settlement and privileged burial evidence lend credibility to the symbolism suggested in the mortuary ritual as well as highlighting how closely dwelling and grave were connected. The homes of the living and those of the dead were integrated in a shared villa estate; as such, it is crucial that they are not viewed in isolation from each other.26 The so-called ‘sarcophagus of the Lady of Simpelveld’ is a key artefact, further accentuating this point (fig. 5).27 The sandstone cinerary urn, of full-body proportions and dated to around the year 170 AD, is elaborately engraved on the interior with images of a villa estate and the furniture and vessels within it. The women depicted reclining on a sort of chaise longue is presumably that of the deceased, a cremated woman aged between 24 and 40 years. Her final resting place is a reproduction of her home environment and valued possessions. The villa tradition-

19

See Crowley 2009 for a detailed discussion of atavism as

22

a strategy in a newly created community. 20

21

See esp. Freigang 1997; also Andrikopoulou-Strack 1986, for discussions of themes. See Crowley (in prep.) for a

Sources are too numerous to mention, with many of

wider bibliography.

the well-published tumulus sites, in particular, providing

23

Bodel 1997.

detailed descriptions and thorough typologies. Beyond

24

Slofstra 1995.

these, Fasold et al. 1998; Struck 1993; Koster 2010; De

25

Rothenhöfer 2005, 253.

Grooth/Mater 1997b, De Groot 2006.

26

Cf. Bodel 1997.

Drinkwater 1981.

27

Cf. Galestin 2001; De Grooth/Mater 1997a.



Fig. 5. View of one side of the interior of the so-called sarcophagus of Simpelveld (photograph courtesy RMO, Leiden).

ally associated with her burial, however, is remarkably modest in its proportions (fig. 6). Hereby we must return to our earlier point that there does not have to be a strict correlation between the two: home and burial. As they comprise a complex, it is reasonable to accept that one could compensate for the other; that a prominent and elaborate burial could be the vehicle for family pride rather than the house, or vice-versa. In such a setting of social mobility as was provided for in the Roman system, it is indeed to be expected that such families would be ambitious, and as success drives success, striving to project that ambition as publicly as possible could form a dominant motivator in the decision to monumentalise either villa, burial or both – but we will return to this point later. If we can distinguish such explicit themes amidst the mortuary evidence, we can only assume that for contemporaries the message was even more succinct. But why was it considered so important to portray oneself, or one’s deceased family member, in such a light? There is clearly more at stake than the writing of a publicly-displayed autobiography. The connotations borne by the grave gifts and iconography alike relate to societal ideals. In order to understand these societal ideals, it is helpful to adopt an approach involving the concept of personhood.28 In its most basic sense, this refers to the notion of being a social person: it is all about relationships and (changing) roles within society. All of the themes mentioned above relate in some way to particular roles. These are strongly linked to facets of personhood such as age and gender. An example of this is the hunt, an activity restricted to the realm of men, adult men, but not too advanced in years as to render them incapable of such a physical pursuit. Personal grooming, despite being part of the daily routine of both men and women, falls firmly into the female sphere when it comes to iconography. The widespread appearance of these model roles in burials across the villa landscapes implies a degree of unity, of shared notions of ideals, that required active participation and signalling, probably resulting in a degree of competitiveness among peers. Even more underlined were the vertical, class, differences. The sheer expense involved in owning a horse for the hunt, in building one’s own private bathing facilities or obtaining such luxurious implements was the exclusive remit

28

See Fowler 2004 for an introduction to the subject.



Fig. 6. Map of the archaeological discoveries at Simpelveld. After Galestin 2001, fig. 1. 1-3 sarcophagi (no. 3: sarcophagus of lady of Simpelveld); 4 villa; 5 secondary building.

Simpelveld

4

1 2

5

3 Stamp

straat E

0

e ijs

rb

ee

k

50 m

of the better-off. Furthermore, only the well-healed of society had the luxury of time during which to engage in activities of leisure rather than labour. Both the enduring iconography of a stone pillar along the roadside, or the conspicuous consumption and deposits in the grave during a well-visited funeral will have sent a clear hierarchical message to the peasant classes, or even slaves, present. Such representation of one’s person explicitly established and reinforced one’s place among a (inter-)regional peer-group, as well as advertising one’s position on a high rung on the socio-economic ladder. But what was the purpose of presenting oneself, or one’s predecessor, as an ideal member of the villa community on a large poster board, as it were, in public at all? Let us first address who it was that was responsible for erecting the monuments in the first place. The truism that the dead do not bury themselves holds for our region and time period too, of course, but the dead – certainly males with means – had arguably perhaps more agency in this cultural setting than ever before or since. A prevailing undertone in classical literature is a concern – to an almost obsessive degree – with posterity. This preoccupation spilled over into the provinces too, as is evident from the words of the Lingonian in his invaluable testament.29 In this document, which has survived in the form of a medieval transcript, a wealthy Gallic landowner describes in minute detail the manner in which he is to be cremated and interred after his death, as well as the specific maintenance demanded by his burial complex subsequently and the active participation required of his descendants in ritual meals and libations at his graveside, for perpetuity. The extensive provision for wills and their consequences in Roman law testifies to their power: they were publicly recited and legally binding.30 Beyond its legal authority, the content of a will also had considerable social power, too. It bound the heirs to the testator in life, before he ever died, creating a mutual obligation, which was cashed in on at death. The obligation then shifted, with the heirs acquiring duties of respect and commemoration. A will is thus both retrospective and forward-looking at the same time, relating to the past of the deceased and the future of the heirs simultaneously.

29

Le Bohec 1991; Le Glay 1991; Buisson 1991.

30

Schoen 2000, 141.



And what of those instances where the deceased took Fate into his own hands, not by taking his own life, but by constructing his own memorial before his death could arrive? This would have been impossible for some categories of burial, such as tumuli, but a real option for erectors of grave pillars such as that of the Secundini brothers (fig. 2).31 The latter bears an inscription telling of the monument’s role in marking the graves of and commemorating members of the previously deceased generation, their peers in age, and also those who have yet to pass away – including themselves. Interpreted by some as an indication of an inherent lack of trust in one’s survivors, it could, and perhaps, should, equally be interpreted as the maximal exploitation of a powerful tool, that of the public poster board advertising one’s membership of a social stratum with all of the positive connotations that brought with it, both explicitly and implicitly inferred. Whether one wielded one’s sceptre from beyond the grave or put it in action while one was still alive, the potential significance and impact of a privileged burial was enormous. These two acts are levelled by viewing them both ultimately as acts of the living, with a view to making a statement that will assert or enhance the status of the living, either by glorifying their dead or glorifying their future dead selves. The burial becomes a vehicle for self-representation, for self-promotion. The obsession with posterity entailed future generations being viewed as an extension of oneself, as those capable of realising one’s aspirations, of honouring one’s memory and ensuring the continuing status of the family and all who once comprised it.32 Such monuments, like testaments, were also firmly forward-facing. Far from representing dusty forgotten grey figures from the past, they actively supported the claim of the living to position and status, to membership of a social class, to the land itself, and urged the living to repay that privilege by rising to the challenge, honouring their proud lineage and ensuring its endurance, both living and dead.

5 commu n it y a n d the de f i n itio n o f social b ou n d aries While such actions may appear to be extremely egotistical (which they may well have been to a point), as a membership card, they were also indicative of the role of the deceased within social communities at various scales. Examples of such scales may include the familia, the regional politically active classes, the inter-regional cultural koiné. Within each scale, the person had a different social role, and therefore depiction of the deceased in a certain role reflects not only facets of his or her social persona within a certain community, but also the nature of that very community. The information provided can inform of regional bonds, blood ties, or shared status and orientation. The correlate of expressing a sense of community is the assertion of social borders. Membership of a club implies that there are those who are excluded. Only when there is an ‘other’ can there be an ‘us’. The exclusivity of privileged burials sends a clear message to those for whom such structures and offerings are beyond their means. They serve as a constant reminder of social and economic hierarchy. Symbols play an important role in reinforcing such hierarchies; these monuments would have comprised part of an environmental context in which the lower social classes were constantly and implicitly put in their place by their ‘superiors’. Despite the negative effect of such a steeply vertical social boundary, shocking to us in an age of political correctness and strived-for equality in rights, there may well have been a degree of pride by association. Servants would have been employed in the raising of burial mounds, were depicted in their roles working for their masters and mistresses, and would have been actively involved in the massive funerary events that took place in association with these markers. While the burials represented domineering symbols of hierarchy, however, they also included the servile classes as members of that hierarchy. 31

Dragendorff/Krüger 1924.

32

Dasen/Späth 2010; Vuolanto 2010.



The servants were, after all, members of that particular villa community, a community offset from those of its neighbouring regions. We see explicit evidence for this in the epigraphy of certain tombstones.33 This is an example of a (hierarchically organised) local community, and it is evidenced in the marked difference in the distribution of various forms of privileged burial. An example of a horizontal community is that of the wealthier classes in general. Despite their clear regional affiliation, as expressed through pillar, mound or sarcophagus, or perhaps because of it, they also formed part of a cultural koiné. Their connections – lifestyle, interests, occupations, etc. – are visible in the recurrence of common themes which transcend burial forms. I say that these distinctive regional burial forms may have even assisted the sense of community on the inter-regional scale because it is a well-known phenomenon that in order to engage in relations with the ‘other’ one must have a clearly defined sense of self; in other words, clearly established (social) boundaries eliminate ambiguity on levels seen to be significant, thus facilitating interaction in mutually beneficial areas.

6 co n cludi n g o b ser v atio n s But the picture is not that clear cut. We have spoken of agency, of the active desires of the deceased, but this individual agency cannot have been universal. Conclusive proof of this is the discovery of infants and children in privileged graves.34 These were deliberately interred by their parents, their families. Considering that the privileged burials in our villa landscapes occur only either individually or in very small groupings, such burials were clearly selected to represent an entire social group. It is significant that children could also perform this important symbolic function. Women, too, are known to have fulfilled this role, challenging any notion that such privileged interment was automatically reserved for the founding father of the villa estate, the first pater familias.35 The motivation behind this selection may have been as simple as that of timing of death, but in any case, we must acknowledge the unavoidable conclusion that it was a truly symbolic action rather than the logical end for the wealthy male. If part of the symbol – the person deceased – on occasion does not verify our expectations, then we must further acknowledge that the social groups, the communities being represented, are a lot more complex than we realise. Complexity is not restricted to the aspects of personhood concerning gender and age, but also to aspects of status and wealth. Boundaries will always be seen by some as being there for incursion and in the climate of real social mobility that characterised the Roman period, movements across both horizontal, territorial boundaries and vertical class ones were possible, and were regularly pursued. Peers ensured that that they exuded the correct image to maintain membership of their socio-economic and political groups, outdoing each other in symbolic profiling. This not only sealed their status, but provided ideals for others, resulting in people from more humble backgrounds striving to attain the heights of their successful neighbours. More than an extrapolation drawn from sociology, this emulation is evident in some of our privileged burials. I mentioned above that not all of those privileged in death would necessarily have been privileged in life, and this is where the concept of aspirational status comes into play. Where a tumulus, today identical in its exterior to many others, has been shown to have been raised over a period of several seasons, we may infer that the raisers did not have the means to dedicate a large taskforce to completing the tomb in one go. It was raised gradually, probably when labourers were not occupied with

33

Cf., for example, the gravestones for slaves - probably

Thisnes. For more detail and further examples see Crow-

erected by their master - from Föhren (G) and LiestalMunzach (CH), discussed in Roymans/Zandstra, this

ley (in prep.). 35

Habets 1895, 281. Examples of female privileged burials

volume, section 4.

include Gors-Opleeuw, Esch/Hoog Keiteren, Niel-bij-

Graves of children: Gors-Opleeuw, Aiseau-Presles,

Sint-Truiden and Helshoven.

34 



sowing or the harvest. The resulting mound was probably just as impressive as that of the neighbours, but the process leading to its ultimate manifestation was a very different one, in terms both of time, relative cost and potentially personal and wider community involvement. Similarly, accepting that the funerary rituals that preceded the interment of cremated remains in such a monumental fashion were well attended, the nature and value of the grave gifts, apart from the functional and religious roles of these, would have been an integral part of the symbolic whole that comprised the burial. We have mentioned above the preponderance of certain themes, but despite the similarities that these themes present across the dataset, there is also considerable heterogeneity. Expressed in virtually all privileged graves, the ubiquitous practice of feasting can be contrasted for example with the theme of the hunt, or that of bathing, which each enjoy a much more limited inclusion. The burials, then, are part of the same cultural koiné, but are executing it at different levels in terms of grave gifts.36 A burial containing the sherds of hundreds of individual vessels is certainly impressive, but is a somewhat different phenomenon than one bearing bronze, or even silver and gold grave goods.37 Size, or in this case, numbers, are not everything. We must distinguish between burials that come about because of the wealth of few, and those that are perhaps the result of the efforts of many. Both represent a community event, but each with very different connotations. The latter suggests privileging for a reason other than attained material status, and we must look to possibilities such as a special symbolic role in the community.As the exteriors of two tumuli with radically different content can appear more or less the same, we are reminded again of the possibility of aspirational acts, of the burial as a product of a collective effort at climbing the social ladder, of creating a monument to where the family is headed rather than to what it has already achieved. Such monuments, whether raised much more slowly and painfully or furnished with more generic gifts, are testaments to ambition and aspiration that grew up from and ensured the successful administration of the Roman regime. In summary, then, the privileged burials of the villa landscapes in our region bear witness to distinct social groupings, symbolically created, strengthened, and expressed through burials of different types: stone monuments, tumuli, or so-called sarcophagi. Within these categories we can ascertain considerable differences in execution, from which we can conclude that some were statements of achievement and establishment, while some were statements of ambition, aspiring to the success of those they emulated and realising the instrumentality of symbols in getting there. One or only a handful of burials acted for the entire family, and as not only men were interred in these special ways, but women and children too, a host of possible reasons for privileging a person’s burial is indicated. One thing, however, is clear: we are looking at close-knit sub-communities spread throughout the countryside within a broader koiné, with intensive social, political and economical ties.

36

37

When we speak of burials here we are referring to the

made here is that before we even arrive at examining the

entire context: the pit/chamber and its contents, infill on

meaning of the different natures of these subsets within

top of and around it, subsequent deposits in the immedi-

the grave assemblage, a distinction needs to be made

ate vicinity, markers, and so forth.

when analysing these contexts which acknowledges the

The difference between vessels employed in the funerary

multiplicity of ways of privileging burials and recognises

banquet and gifts given to the deceased to accompany

the significance of quality and quantity respectively.

them into the Afterlife notwithstanding. The point being



re f ere n ces Amand, M., 1985: De Gallo-Romeinse tumuli, Tongeren (Publicaties van het Provinciaal Gallo-Romeins Museum 31). Amand, M./R. Nouwen, 1989: Gallo-Romeinse tumuli in de civitas Tungrorum, Hasselt (Publicaties van het Provinciaal Gallo-Romeins Museum Tongeren 40). Andrikopoulou-Strack, J.-N., 1986: Grabbauten des 1. Jahrhunderts n. Chr. im Rheingebiet. Untersuchungen zu Chronologie und Typologie, Bonn (Beiheft Bonner Jahrbücher 43). Bodel, J., 1997: Monumental villas and villa monuments, Journal of Roman Archaeology 10, 5-35. Buisson, A., 1991: Le tombeau du Lingon. Étude du cadre architectural et archéologique, in Y. Le Bohec, (ed.), Le testament du Lingon. Actes de la journée d’étude du 16 mai 1990, Paris, 63-72. Cohen, A. P., 1985: The symbolic construction of community, Chichester/London/New York. Crowley, L., 2009: Creating a community. The symbolic role of tumuli in the villa landscape of the civitas Tungrorum, in M. Driessen/S. Heeren/J. Hendriks/F. Kemmers/R.Visser (eds), TRAC 2008. Proceedings of the Eighteenth Annual Theoretical Roman Archaeology Conference, Oxford, 113-126. Crowley, L.M.P., in prep.: Dying in a material world. Symbolism, self-representation and their implications in privileged mortuary contexts in the villa landscapes between Bavay and Cologne, Amsterdam (PhD thesis VU University Amsterdam). Dasen, V./T. Späth (eds), 2010: Children, memory, and family identity in Roman culture, Oxford. De Groot, T., 2006: Resultaten van de opgraving van een Romeins tumulusgraf in Bocholtz (gem. Simpelveld), Amersfoort (Rapportage Archeologische Monumentenzorg 127). De Grooth, M./B. Mater (eds), 1997a: Een huis voor altijd, Maastricht. De Grooth, M./B. Mater, 1997b: Verscholen schatten, begravingen uit Gronsveld, Heerlen en Stein, in M. De Grooth/B. Mater (eds), Een huis voor altijd, Maastricht, 50-59. Dragendorff, H./E. Krüger, 1924: Das Grabmal von Igel, Trier. Drinkwater, J.F., 1981: Money-rents and food-renders in Gallic funerary reliefs, in A. King/M. Henig (eds), The Roman West in the third century. Contributions from archaeology and history, Oxford (BAR International Series 109). Fasold, P./T. Fischer/H. von Hesberg/M. Witteyer (eds), 1998: Bestattungssitte und kulturelle Identität. Grabanlagen und Grabbeigaben der frühen römischen Kaiserzeit in Italien und den Nordwest-Provinzen, Köln (Xantener Berichte 7). Ferdière, A., 2004: Indigènes et “romanisés” à travers la tombe privilégiée en Gaule, Latomus 63, 35-57. Fowler, C., 2004: The archaeology of personhood. An anthropological approach, London. Freigang, Y., 1997: Die Grabmäler der gallo-römischen Kultur im Moselland. Studien zur Selbstdarstellung einer Gesellschaft, Jahrbuch des Römisch-Germanischen Zentralmuseums Mainz 44, 277-440. Gabelmann, H., 1973: Römische Grabmonumente mit Reiterkampfszenen im Rheingebiet, Bonner Jahrbücher 173, 132-200. Gabelmann, H., 1979: Römische Grabbauten der frühen Kaiserzeit, Stuttgart. Galestin, M.C., 2001: The Simpelveld sarcophagus. A unique monument in a provincial Roman context, in T.A.S.M. Panhuysen (ed.), Die Maastrichter Akten des 5. Internationalen Kolloquiums über das provinzialrömische Kunstschaffen im Rahmen des CSIR. Typologie, Ikonographie und soziale Hintergründe der provinzialen Grabdenkmäler und Wege der ikonographischen Einwirkung, Maastricht, 63-76. Gottschalk, R., 2003: Studien zu spätrömischen Grabfunden in der südlichen Niederrheinischen Bucht, Bonn. Habets, J., 1895: Romeinsche villa opgegraven 1879-1880, in J. Habets (ed.), Jaarboek. Uit de nagelaten geschriften van wijlen den Zeer Eerw. Heer Jos. Habets over Wegen en Gebouwen, uit het Romeinsch Tijdperk, in het Hertogdom Limburg, Maastricht (Publications de la Société Historique et Archéologique dans le Limbourg XXXII), 257-296.



Le Bohec, Y., 1991: Le “Testament” du Lingon et le Lingon du “Testament”, in Le Bohec, Y. (ed.) Le Testament du Lingon. Actes de la Journée d‘étude du 16 mai 1990, Paris, 41-55. Le Glay, M., 1991: Notes sur le “Testament du Lingon”, in Y. Le Bohec, (ed.), Le testament du Lingon: Actes de la journée d‘étude du 16 mai 1990, Paris, 57-61. Martens, M., in prep.: Life and culture of the Roman small town of Tienen, Amsterdam (PhD thesis VU University Amsterdam). Morris, I., 1992: Death-ritual and social structure in classical antiquity, Cambridge. Panhuysen, T.A.S.M., 1996: Romeins Maastricht en zijn beelden, Maastricht. Rothenhöfer, P., 2005: Die Wirtschaftsstrukturen im südlichen Niedergermanien. Untersuchungen zur Entwicklung eines Wirtschaftsraumes an der Peripherie des Imperium Romanum, Rahden (Kölner Studien zur Archäologie der römischen Provinzen 7). Schoen, R.P., 2000: Pro facultatibus vel dignitate defuncti. Uitvaartritueel, sociale status en regelgeving in Rome (100 v.Chr. – 300 n.Chr.), Zutphen. Slofstra, J., 1995: The villa in the Roman West. Space, decoration and ideology, in J. Metzler,/M. Millett/N. Roymans/J. Slofstra (eds), Integration in the Early Roman West. The role of culture and ideology, Luxembourg, 77-90. Struck, M. (ed.), 1993: Römerzeitliche Gräber als Quellen zu Religion, Bevölkerungsstruktur und Sozialgeschichte, Mainz. (Archäologische Schriften des Instituts für Vor- und Frühgeschichte der Johannes Gutenberg-Universität Mainz 3). Vuolanto,V., 2010: Children and the memory of parents in the Late Roman World, in V. Dasen/T. Späth (eds), Children, memory, and family identity in Roman culture, Oxford, 173-192. Wigg, A., 1993: Die Grabhügel des 2. und 3. Jahrhunderts n.Chr. an Mittelrhein, Mosel und Saar, Trier. Willer, S., 2005: Römische Grabbauten des 2. und 3. Jahrhunderts nach Christus im Rheingebiet, Mainz am Rhein (Beihefte der Bonner Jahrbücher 56).



Monumental funerary structures of the 1st to the 3rd centuries associated with Roman villas in the area of the Treveri Jean Krier / Peter Henrich 1 Introduction, research questions and aims 2 Mausolea and circular funerary structures of the 1st century 3 Funerary monuments and tumuli of the 2nd and 3rd centuries 4 Conclusions and summary Catalogue References

1 i n troductio n , research q uestio n s a n d aims 1 The focus of research into the funerary monuments in the area of the Treveri has thus far consisted of the stylistic classification of reliefs, the analysis of scenes shown on the reliefs and the architectural construction.2 This is mainly because the majority of funerary monuments have only survived as sculptured blocks reused as spoils.3 Until now it has been virtually impossible to make direct statements about the sepulchral context and location of funerary monuments in terms of topography and their relationship to villas, as relatively few monument foundations have come to light during archaeological excavations. They are also very difficult to find using normal surveying methods, such as surface field-walking. A preliminary analysis of foundations of funerary monuments discovered by chance or during research programmes (see catalogue)4 revealed regularities in the positioning of funerary monuments in relation to the villas.5 In investigations of Roman tumuli, the research emphasis has focused on the typical evaluation of barrow construction, burial rituals and a regional and supraregional analysis of this type of burial. Despite this category of funerary monuments being well-known, there is still an absence of interpretations of the relationship between the burial mounds and villas.6 1

The paper was translated from the German by Dr Clive

4

Bridger, Xanten. 2

3

out in the 1970s in Luxembourg: summarised in Thill

Freigang 1997a, 1997b; Andrikopoulou-Strack 1986;  Baltzer 1983; Kempchen 1995; Kaszab-Olschewski 2007;

This is exemplified by the research programme carried 1975, 302-303 with extensive bibliographical references.

5

Martin-Kilcher 1993. On funerary monuments along 

Langner 2001; Numrich 1997; Von Massow 1932; Zinn

rivers, cf. Leveau 2000. On the position of funerary mon-

2001; Kremer 2009a, 2009b.

uments in general and also in an urban context, cf. Lafon

As the most imposing examples, we can mention the

2002. Graen 2008, 191-196, analysed the phenomenon

numerous funerary monuments reused in the late

across several provinces, but incorrectly assigned most of

Roman fortification walls of Neumagen, Jünkerath and

his examples from the Treveran area to the Rhenish and

Arlon; cf. Von Massow 1932; Numrich 1997; Koethe 1936; Mariën 1945; Mertens 1968, 1973; Lefèbvre 1975, 1990.

Danubian provinces. 6

Wigg 1993, 110-114, 116 does analyse the distances from the burial mounds to the villas and roads, but does not



Rh

in

e

3

10

5

Som

me

16

M se

100 km

eu

0

13

8 9 15 11

se Mo

12

6 17 1 4

7

lle

14 2

Trier

Fig. 1. Distribution of monumental funerary structures associated with villas in the area of the Treveri. 1 Bertrange, 2 Bollendorf, 3 Duppach, 4 Flaxweiler, 5 Gillenfeld, 6 Grevenmacher, 7 Hetzerath, 8 Igel, 9 Kirf, 10 Mehren, 11 Nennig, 12 Newel, 13 Remerschen, 14 Strotzbüsch, 15 Wasserbillig, 16 Weiler-la Tour, 17 Wintersdorf.

The extensive use of geophysical surveying and many excavational and research activities has boosted the number of known foundations of imposing funerary structures in recent decades, enabling more nuanced comments on the siting of monuments in relationship to villas. Included in the analysis are large funerary monuments independent of the type of superstructure, as well as tumuli which differ from other barrows in terms of their visible above-ground features,7 such as their size8 or sumptuous architectural forms.9 Against this background, our central concern here is the location and position of the monuments in relation to the topography and other elements of the cultural landscape, such as villas, roads or waterways. We will document and define the regularities or parameters for choosing the location of this type of monument, as well as their meaning for the self-awareness and self-representation of those buried there and their kin.We will also attempt to explore intentions to monumentalise the cultural landscape in order to increase the imposing architecture of the villas. In addition, we will investigate the nature and extent of changes in the parameters regarding choice of location during the first three centuries.10

mos, are also documented (see Wigg 1993, 48-55).

go into detail about topographical siting and the vis-

7

8

ibility between the burial mounds and other landscape

9

Wigg 1993, 21-45.

elements.

10

We made a conscious decision not to include richly fur-

Cf. Andrikopoulou-Strack 1986, 9-53; Willer 2005, 3-22

nished graves without costly, mortuary structures above

with older literature in n. 5.

ground in this part of the analysis, as these played no role

In the case of particularly imposing burial mounds,

in giving prestigious shape to the cultural landscape.

accessible stone grave chambers, sometimes with a dro-

1

A further concern is the question of continuity from rich Late Iron Age burials to early Roman funerary monuments with regard to location, burial rituals and grave goods, as well as the extent to which regional emphases or differences are evident in the distribution of grave monuments of the 1st and of the 2nd/3rd centuries. The study area covers the central and northeastern part of the civitas Treverorum west of the Moselle and Saar in the territory of the Grand-Duchy of Luxembourg, as well as the northwestern part of Rhineland-Palatinate (fig. 1).11

2 mausolea a n d circular f u n erar y structures o f the 1 st ce n tur y Circular funerary monuments of the 1st century AD that differ from other tumuli in terms of their imposing above-ground appearance are unknown in the eastern part of the research area.12 This could be because although the construction of stone enclosure walls began in the last quarter of the 1st century, this practice did not become widespread until the start of the 2nd century.13 Only then was it possible to build large mounds, resulting in a monumental form of grave that was visible to observers. In the 1st century there are also no indications for monumental funerary towers in the eastern part of the study area. The situation is different in the western parts, however. In recent decades the sites of several monumental, circular funerary buildings of the 1st century have been located in southern Luxembourg.14 The first round building of interest here is the monument of Weiler-la-Tour, which has hitherto been incorrectly dated.15 Its foundations, with an exterior diameter of 7.20 m, were excavated in 1971 in Seitert wood, southeast of the village.16 Only the massive fragment of the console’s cornice from the monumental superstructure of the grave rotunda could be recovered during the excavation; it is definitely a 1st-century structure, and has been dated to the early Flavian period by Numrich.17 The monument was erected on the small plateau of a hilly ridge dominating the landscape and towered some 15-20 m above the site of the large axial courtyard villa of Aspelt-Kleppen/op de Maueren some 300 m away, to which it clearly belonged.18 Thanks to its exposed position, this circular structure – originally measuring 5 m in height – would have been visible from some distance away; it also lay in direct view of the vicus of Dalheim only 3 km away. The important Roman trunk road linking Lyon, Metz, Trier and the Rhine (Via Agrippa) ran just 2.5 km east of the monument. Because of its type (circular) and its topographical location (on a ridge dominating the landscape in all directions), the funerary monument of Weiler-laTour is reminiscent of the circular grave of L. Munatius Plancus on the Monte Orlando near Gaëta south of Rome.19 A comparable grave monument, albeit with a diameter of only 5.20 m, was excavated in 2003/4 on the site of the Gallo-Roman villa of Goeblingen-Miecher.20 This funerary rotunda, whose facing consisted of huge limestone blocks of 1.60 x 0.73 x 0.40 m, was shown to have been built in the second

11

The eastern border of the Treveran area is defined by the

15

Wigg 1993, 188, cat. no. 41 with pl. 43.

Lieser, following the mapping by Heinen 1985, Beilage

16

Thill 1971, 499; 1973b, 291, 297, fig. 2-3; Wilhelm 1974,

1. The tumuli at Nennig were included in the analysis,

16-17, no. 4;Thill 1975, 303; Carte archéologique du Grand-

despite their position on the right bank of the Moselle, as

Duché de Luxembourg, Feuille 26 (1977) 63 (26 D-3); Kremer 2009a, 128 with n. 378.

they provide connections with the funerary monument

12 13 14

at Remerschen and the late Roman grave chamber of

17

Numrich 1997, 44, 162, cat. no. 35.

Bech-Kleinmacher on the left bank of the Moselle.

18

Carte archéologique du Grand-Duché de Luxembourg, Feuille

Ebel 1989, 103, 106-107.

26 (1977) 63-64 (26 D-14 u. D-15).

Wigg 1993, 21, 43.

19

Fellmann 1957; Graen 2008, 27-31, 245-246, cat.no. R 7.

Cf. summary in Krier 2007, 167-168.

20

Krier 2007; Kremer 2009a, 122, 125, 128.



quarter of the 1st century and corresponded to a type of circular sepulchral monument rarely found in Gaul – one with a covered, accessible inner grave chamber. In the case of Goeblingen, it is striking that the funerary rotunda stood just 8.5 m from the main façade of building 1 of the villa, directly in front of the monumental flight of steps to the portico. Since this second domestic building was certainly erected some time after the circular funerary building was built, we must assume that the owners of the Goeblingen estate consciously aligned their new residence with the circular mausoleum, which had already stood for several generations.21 Another question arising from the early Roman funerary building at Goeblingen is the possible connection with the nearby Treveran elite cemetery of Goeblingen-Scheierheck a mere 500 m away. Its five richly furnished mortuary chambers bear impressive witness to the self-consciousness of the pro-Roman, native upper class in the earliest phase of Romanisation, i.e. in the decades immediately following the Gallic Wars.22 It is therefore quite conceivable that the funerary monument of Goeblingen-Miecher, built only a generation later according to Roman custom and Mediterranean prototype, documents the progressive process of Romanisation within the same Treveran elite family.23 Similar funerary rotundas of the second and third quarters of the 1st century are documented by finds of masonry blocks from Luxembourg-Fetschenhaff24 and the palatial villa of Walferdingen-HelmsingenSonnebierg.25 This suggests that this type of monument, which was possibly inspired by the Mainz cenotaph of Drusus the Elder,26 was particularly preferred by the leading families of the western Treveran area in the early Imperial period. A further type of monument, which belongs without doubt to the same chronological and social context and whose architectural form and artistic workmanship clearly goes back to Mediterranean prototypes, has recently been found in the form of an imposing mausoleum of Bartringen-Burmicht, which stood at least 16.35 m high.27 The three-storied mortuary structure, which in typological terms occupies a position between the late Republican funerary monument of the Julii in Saint-Rémy-de-Provence28 and the early Claudian mausoleum of L. Poblicius in Cologne,29 was erected around AD 30 and is distinguished by the animated battle scene on its base30 and the fine artistic quality of the scrolled friezes on the architraves. One of the funerary statues set up in the aedicule of the middle storey displays the head of the deceased ‘capite velato’, which shows that we are certainly dealing here with a Roman citizen who perhaps held the position of priest in Roman Trier. Since all of the monument’s remains were unearthed on the estate of a huge axial courtyard villa whose beginnings reach back to the middle Augustan period,31 the mausoleum must also have originally stood within the country estate, even though its foundations could not be detected during the 1997-2003 excavations.32 Its proximity to the Roman trunk road from Rheims to Trier, which ran just one km further to the north, is certainly not coincidental. At this point we are consciously excluding the putative Early Roman funerary monument of Echternach,33 which Lafon in particular has proffered as a classic example of the architectural alignment of a manorial villa with the mausoleum of the estate’s founder.34 The situation at Echternach is encumbered by too many uncertainties, however, not only with regard to the original location, but also in respect of 21

Krier 2007, 170.

22

Metzler/Gaeng et al. 2009.

28

Rolland 1969.

23

Krier 2007, 170-171; S. Martin-Kilcher, in: Metzler/ 

29

Precht 19792.

Gaeng et al. 2009, 370.

30

Krier 2003.

Wilhelm 1974, 31, cat. nos 166-168; Numrich 1997, 

31

Krier, in: Kremer 2009a, 24.

41-42; Krier 2007, 167-168; Kremer 2009a, 103 (with

32

Krier, in: Kremer 2009a, 30.

fig. 94), 128.

33

Metzler/Zimmer/Bakker 1981, 29, fig. 13, 304-312 with

24

25

ences.

Krier 2007, 168 with n. 26; Kremer 2009a, 128.

26

Krier 2007, 167 with n. 23-24.

27

Kremer 2007 with comprehensive bibliographical refer-



figs. 224-230. 34

Lafon/Adam 1993, 11; Lafon 2002, 117, 119, fig. 1.

the monument’s reconstruction.35 This is based on only seven masonry blocks which were reused in the Late Roman period, making it impossible to draw firm conclusions.

3 f u n erar y mo n ume n ts a n d tumuli o f the 2 n d a n d 3 rd ce n turies Numerous funerary monuments of the 2nd and 3rd centuries have been discovered in the region north of Trier. There are also many features which, owing to stone thefts or purely geophysical investigations, have yet to be dated. The mainly square or rectangular layout of the foundations suggests that most of the monuments are pillar monuments corresponding to the Igel column. A more exact differentiation is not possible – nor is it important for our purposes. There is a conspicuous variety of types of enclosure wall. These could have a simple rectangular design either with or without integrated foundations for a funerary altar or monument, or – as the example of Gillenfeld shows – with a circular enclosure wall and an additional, U-shaped wall in the foreground. This variety can also be seen in the widely varying sizes of the enclosures.36 In contrast to the numerous old excavations and geophysical surveys that often left many questions unanswered, it was possible to investigate salient aspects of the structure and development of the cemetery of Duppach-Weiermühle during the 2002-2004 excavations.The centrally defining element of both the necropolis and the auxiliary buildings of the villa investigated thus far is the 5 m wide private road which connected the villa with the Roman road from Cologne to Trier.37 Thus all individual elements of the cemetery, framed by the funerary enclosure to the south and the grave chamber with forecourt to the north, created a monumental ‘façade’ that corresponded conceptually to the scheme of a sepulchral street.38 Independent of when individual buildings or structures were erected within the cemetery, the basic idea of the relationship to the path of the road remained. The road with its line-up of individual structures bore a direct relationship to the Roman road from Cologne to Trier some 800 m away, guaranteeing the visibility of the grand layout. The fact that this burial ground was intended purely for the villa owner’s family can be deduced from the small number of graves (only four) and the existence of another cemetery in the direct vicinity with no monumental funerary architecture.39 The necropolis was bounded to the north and south by two auxiliary buildings and thus lay in the courtyard and not, as was customary, outside the courtyard walls. Parallels for this can be found in Lullingstone,40 Bierbach,41 Mackwiller42 and Biberist-Spitalhof.43 The examples cited show that the otherwise strict rule of separating funerary and domestic areas was not carried out rigorously. One explanation could be that the owners of the large villa estates originated from the Gallic upper class, which combined a native burial tradition with Italian elements of the 1st century.44 Since the exceptions mainly involve large funerary monuments, we can assume that when deciding where to position the funerary monuments villa owners were motivated by a concern for the aesthetic and monumental appearance of the villa and by their own ideas about design. Such design notions were also paramount in the choice of site for both tumuli at the villa of Nennig. A special architectural feature of this villa is a 256 m long, free-standing portico, which runs parallel and as an extension to the main building with a columned hall and a frontage of around 120 m. It is not attached to the building, however, but ends in rotundas, thus giving observers the impression of a façade 35

Cf. Kremer 2009a, 122 with n. 321-323.

40

Waurick 1973, 127-129; Meates 1979, 25, fig. 2.

36

Cf. the compilation in Henrich/Mischka 2005, 25 fig. 4.

41

Kolling 1986, fig. 17.

37

Cf. Henrich/Tabaczek 2003; Henrich 2003, 48.

42

Lafon 1993, 116, fig. 5.

38

On funerary streets, cf. the papers in Hesberg/Zanker

43

Schucany 2000, 118-124; ead. 1995, 2006.

1987.

44

Cf. above and Lafon 1993, 119.

39

Henrich 2006,126, no. 6.



several hundred metres in length. This centrally defining and dominant element of the villa is further accentuated by both tumuli lying some 300 m away. They are situated on a tangential axis running at right angles from the southern rotunda of the portico towards the Moselle, thereby extending the façade for a further 100 m. To anyone journeying along the Moselle, the villa at Nennig would have appeared as an impressive façade made up of several independent modules of varying depths.45 A comparable situation is also evident at the villa of Gillenfeld/Strohn. Here the two tumuli at the entrance to the Macherbach valley, together with the funerary monument situated higher up, created a linear unity and thus an imposing entrance to the extensive axial villa which stretched out behind it, taking up the whole of the valley. An examination of the locations of the mortuary monuments of Newel, Mehren/Schalkenmehren and Kirf reveals a distinct conceptual distinction. Unlike the examples cited, the funerary monuments here do not form a structural unity with the villas, but are sited at least 100 m away. Nor do they necessarily display an identical alignment. Here the monuments’ visibility from the road or waterway was paramount. Although the relationship to the villa is evident in terms of proximity, there is no direct conceptual connection between the two architectural elements, as is the case in Duppach or Nennig for example.46 Common to both types of location is the fact that the main alignment of the monuments’ façades was always with the roads and that the visibility from the villa’s main building was only of secondary importance. This is also the case in Bollendorf, where the grave monument is located directly on the bank of the River Sauer and thus aligned to that waterway, whereas the affiliated villa lay behind it. The picture for the northeastern part of the Treveran area can be confirmed and supplemented by several examples from Luxembourg. Particularly informative in this regard are the imposing funerary structures of Grevenmacher-Potaschbierg/an der Heck47 and Flaxweiler-Tonn.48 Both stand directly on the important Roman trunk road from Metz to Trier (Via Agrippa) and only 3.5 km apart. The funerary monument of Grevenmacher was erected shortly before the mid-2nd century in a walled enclosure of 30 by 28 m. The type is unusual in that its front faces not only the road, just 10 m away, but also the villa some 500 m further to the north. The tumulus of Flaxweiler, 30 m in diameter and 7 m high, has a stone grave chamber with a 12.75 m long dromos and also lay 10 m north of the Roman trunk road. The villa to which it belonged was visible some 600 m to the southeast. These two examples also bear impressive witness to the fact that owners of large, neighbouring farms competed in their choice of monument type, thereby presenting themselves in an enduring way not only to the local population, but especially to the numerous users of the Roman trunk road linking the Mediterranean with the Rhine. We are dealing here with clear reminiscences of Roman practices from the Mediterranean world. What was true of the busy Roman highways was also true of the more important navigable waterways (Moselle, Sauer). A splendid example is the funerary monument of a major vintner and wine-trader which stood immediately on the left bank of the Moselle in Remerschen-Mecheren.49 Near this monument, which was built in the mid-2nd century, was an associated walled funerary enclosure of 17.3 by 5.1 m, which was still in use during the 4th century. The villa with its domestic and farm buildings stretched directly behind the funerary structures. Since the villa of Nennig, boasting the impressive architecture 45

It is important to bear in mind when analysing the funer-

the tumulus of Strotzbüsch indicates a similar situation:

ary structures of the Nennig villa that the group will have included a monument with a cavalry battle scene.

cf. Henrich 2006, map 23. 47

322-323 with fig. 11.

Unfortunately, only one sculptured block has survived

46

Metzler/Thill/Zimmer 1974; Metzler/Zimmer 1975, 

and the site remains unknown: Gabelmann 1977, 101-

48

Wigg 1993, 148-149, cat. no. 16 with pl. 12.

104.

49

Thill 1970b, 1972a, 1973a; Krier 1992.

The lack of a scattering of rubble in close proximity to



and imposing tumuli described above, was situated only 3.7 km further north on the right bank of the Moselle, we can assume that the estate owners vied with one another. A somewhat different and rather unusual example from Luxembourg is provided by funerary enclosures and monuments investigated between 1970 and 1972 in the Weiler woods on a high plateau above the Moselle valley between Wasserbillig and Lellig.50 This is a proper sepulchral street in the direct vicinity of a 4.5 m wide Roman road, which ran from Trier into the Ardennes and linked the vici of Wasserbillig and Altrier with one another. Over a distance of some 700 m, excavations uncovered two walled funerary enclosures with cremation burials, a similarly walled enclosure with a monumental funerary pillar in its centre and the foundations of further grave stelae on the periphery, as well as the remains of a smaller pillared monument. Since the various burial sites show a clear chronological sequence from the late 1st to well into the 3rd century, they probably all belonged to the same Roman villa, whose location may be assumed in nearby Tempelsklouster. Once again, there is a direct connection with a major road, with the siting of the corresponding villa retreating somewhat into the background.

4 co n clusio n s a n d summar y Since only very few rural funerary monuments of the 1st century are known from the territory of the Treveri, it is still too early to propose a typology of preferred locations. However, it is significant that in all known cases, which are presently concentrated in southern Luxembourg, there is a connection with a large rural estate whose beginnings usually reach back to the earliest Roman period, if not the Late La Tène period. If the funerary monument of the large Roman villa of Borg were to be found, we could therefore expect it to be of one of the Mediterranean types described above. Since we know little as yet, in the Treveran area at least, about the appearance of the contemporary construction phase of these villas, it is difficult to decide in each individual case whether there is a direct architectural relationship between the domestic and farm buildings and the funerary monument, as we assume is the case in, for instance, Weiler-la-Tour/Aspelt. It was essential in this period for the deceased of the native ruling class not only to be buried according to Roman custom, but also to have funerary monuments erected according to Mediterranean prototypes which demonstrated the family’s special social standing to all. An analysis of the locations of funerary monuments of the 2nd and 3rd centuries shows three different types. The first is represented by the examples of Duppach, Nennig and Gillenfeld. Here the funerary edifices were integrated into the overall architectural concept of the villa. They made a key contribution to the appearance and monumentality of the complex as a whole through the axiality and symmetry of the subsidiary and main buildings, as well as of the funerary monuments. In the known examples all or most of the elements described formed a monumental frontal façade which was clearly orientated towards the traffic routes. In the case of the second type of location, the proximity to the main building played only a minor role. Of paramount importance here was a suitable display in the form of monumental funerary pillars along the traffic routes. Observers would have noted a connection with the villas because of the proximity, but in this type the alignment and positioning of the funerary monument did not have to match or be coordinated with that of the villa. The third type, as clearly exemplified by Bollendorf, is distinguished by the main building being situated higher than the funerary monument. Here the waterway can be regarded as the main focus. This is a rare form, with the choice of location dictated by the topography of narrow river valleys. All the types described above have one thing in common. As far as we can deduce from the variable state of research, the main side of the monuments always faced the road or waterway. We therefore 50

Thill 1970a, 1971, 1972b, 1975, 303; Metzler/Zimmer 1975, 320-322 with fig. 10.



have to accept that the main side of these monuments, and in some cases the entire monument, was not visible from the villa.51 Consequently, the imposing character of the sepulchral monuments was clearly paramount in the overall scheme. In her investigation into the relationships of the location of mortuary monuments and villas in Upper Germany and the neighbouring regions, Martin-Kilcher52 recognised that the monuments were erected either close to the main buildings at a slight elevation, with no direct connection to the traffic routes but visible from far away, or on roads that passed by the villas and, in particular, the pars urbana. These results are not directly transferable to our research region as there is no instance in which we can reconstruct a direct connection to a main building at a lower elevation or even to a pars urbana. Certainly, elevated monuments can regularly be seen in the Treveran region, but it is villas as a whole or key locations along the traffic routes that prompted the building of monuments there. Leveau suggests that in addition to the display of luxury, religious considerations also played a role, in particular the giving of offerings to the dead by passers-by.53 This is impossible to ascertain, but as he himself acknowledges,54 such offerings can only have played a minor role in the case of monuments situated along waterways. In his investigations covering several provinces, Graen also deduced that the funerary monuments were erected right on the roadside.55 He also demonstrated that they were mainly located higher than the main building. In the area under investigation, the examples he cites of monuments constructed in valleys, with buildings at a higher elevation because of the topography, relate exclusively to narrow river valleys. His example of Serrig is not in a valley, but on the bank of the Saar and corresponds to our third type defined here. Similarly, in the Treveran area there are no examples of funerary monuments and villas located on opposite sides of valleys. As already described, Graen includes villas, funerary monuments, traffic routes and topography in his investigations, but does not analyse the dependencies and interactions between these individual components in any great detail. The examples and ideas presented here have clearly demonstrated that the erection of monumental funerary structures in the centre of the villa, axial to the villa or even in front of the main building is not the rule, as Lafon maintains,56 but rather the exception. We can also rule out considerations that the funerary monuments marked the centre of estates.57 The results of this study clearly show that funerary monuments of the 1st century are thus far known only from the western part of the study area. Only from the 2nd century onwards are such monuments regularly and extensively found in the eastern Treveran area, as strikingly attested to by reused spoils, in addition to numerous foundations. We see a similar development in the case of the large tumuli, revealing a growing need by the Treveran upper class to show off its private riches through architectural display. Together with large and expansive villas, mortuary edifices were also purposely used to ‘monumentalise’ the landscape. In addition, the analysis shows that no single factor determined the choice of location for a funerary monument. Instead, there was a combination of many, often very individual factors, such as topography, the structure and arrangement of the villa and its origins, alignment with the traffic routes, the type and size of the desired monument as well as a feeling of rivalry among neighbouring estate owners.

51

Cf. also Lafon 2002, 117.

55

Graen 2008, 193-195.

52

Martin-Kilcher 1993, 159, 163.

56

Lafon 2002, 117.

53

Leveau 2000, 202.

57

Lafon 2002, 117.

54

Leveau 2000, 210.



catalogue Place: Bertrange-Bartringen (L) Site: Burmicht Summary of the features and research history: Between 1997 and 2003 the complete, over 6 ha large farmyard of one of the largest axial courtyard villas of the Treveran area was archaeologically investigated in Bartringen-Burmicht. More than 80 blocks in secondary and tertiary use and numerous further fragments of an Early Roman mausoleum hitherto unique in north-eastern Gaul and dating to around AD 30 were uncovered. Location to main building/to next scattering of rubble and to the traffic-routes: During the archaeological work of 1997 to 2003 the certainly massive foundation base of the mausoleum could not be located anywhere within the excavated zone, so that its exact location remains unknown. One cannot doubt, though, that the funerary monument was erected either on the estate (in direct proximity to the main building?) or perhaps near the road, to which the property was aligned. Literature: Kremer 2009a (with extensive bibliographical notes).

Trie r

Fig. 2. Topographical situation of the Roman villa Da

lhe im

of Bertrange.

Place: Bollendorf (D) Site: Swimming baths Summary of the features and research history: In 1974 the 6.85 x 5.78 m foundation of a Roman mortuary monument was documented in the area of the swimming baths of Bollendorf. There are no clues to the dating. Location to main building/to next scattering of rubble and to the traffic-routes: About 330 m to the northwest is a scatter of Roman rubble, which may derive from the main building of the respective villa. The building lies higher than the funerary monument, which is situated on the river Sauer and was entirely visible from the waterway. Literature: Trierer Zeitschrift 40/41, 1977/78, 405.



Fig. 3. Topographical situation of the Roman villa and grave monument at Bollendorf.

Place: Duppach-Weiermühle (D) Site: Bremscheid Summary of the features and research history: From 2002 to 2004 excavations were carried out near Duppach-Weiermühle in a monumentally designed necropolis of a Roman villa with two pillar funerary monuments, a grave-chamber, a funerary enclosure, as well as two smaller monuments. The cemetery was known through chance finds since the beginning of the 20th century. All features of the necropolis were arranged along a 5 m wide road connecting the villa with the Roman road of Cologne to Trier. The foundations of the largest and oldest funerary monument within the cemetery (monument A) were 7.6 by 6.6 m large and were at least 4.6 m deep. The bottom could not be excavated because of rising groundwater and reasons of safety. The monument was surrounded by a 0.6 m wide enclosure wall, which was covered with semicylindrical coping-stones and which enclosed an area of 19.9 by 18.3 m. A sestertius of Hadrian from the foundation trench of the western enclosure wall provides a terminus post quem for the dating of the monument. Both the features, as well as the surviving sculptural fragments show that monument A must have been a pillar monument. Immediately adjacent to and partly including the southern enclosure wall of monument A was the enclosure wall of the second pillar monument (monument B). The 5.2 by 4.2 m foundations were surrounded by an enclosure wall creating an area of 11.8 by 10.8 m. In this case, too, the features show beyond doubt that it must have been a funerary pillar. However, by comparison with the other known types in the region of Trier and beyond, its sculptural decoration was highly individually shaped with a combination of lion and boar, and griffins. The analysis of the sculptures allows a dating of the mortuary monument to the Severan period. To the south of monument B lay two 1.1 by 1.6 m and 1.4 by 1.6 m large foundations (monuments C and D). Their size and modest depth point to the location of two small funerary monuments or cremation cists. To the west of each foundation lay a cremation grave covered with stone slabs. South of monuments C and D and parallel to the road described below was situated a wall with a central thickening (monument E). A square foundation with sides of 1.3 m (monument F) could be located 15 m to the west. A further 8 m to the west lay building III, which had a single room with internal measurements of 8.3 by 5.3 m. A direct connection between monuments E and F with building III can be surmised from the position of both monuments in a straight line in front of the building. Owing to recent operations and considerable erosion all three features had survived very badly. No graves were discovered.



The feature can be interpreted as a funerary garden with monument and covered grave-chamber in the interior and according to finds be dated to after the middle of the 2nd century. In 2004 building II was uncovered some 30 m north of monument A. It had been built into the slope, had internal measurements of 3.2 by 2.3 m and can be interpreted as a grave-chamber. The area in front of the building was subdivided by hedges and a wooden fence. Post-excavation analysis of the features and the sculptured ornamentation has shown that monument A, the oldest element in the cemetery, was erected in the Antonine period. This was followed by monument B in the Severan period and the grave-chamber at the end of the 2nd century. In a third phase the two small monuments C and D and subsequently the funerary garden with the monuments E and F together with building III were constructed. Location to main building/to next scattering of rubble and to the traffic-routes: Along the road directly north of the grave-chamber and south of the funerary monument lay out-buildings of the villa which could be examined archaeologically and also geophysically. The location of the main building remains unknown. Compared to its immediate surroundings the cemetery lies in a topographically exposed position and its whole length was visible from the Roman road of Cologne to Trier. Literature: Henrich 2010.

building VII

building VI

building V

building IV building II

monument A

monuments C/D monument B street monuments E/F building III

building I building X building VIII

street

building IX building XII building XI

geophysically surveyed area excavated area excavated feature geophysical feature

0

100 m

Fig. 4. Groundplan of the Roman grave monuments and building remains at Duppach.



Place: Flaxweiler (L) Site: Tonn Summary of the features and research history: With a diameter of around 30 m and a surviving height of 7 m this is the largest Roman tumulus in Luxembourg and was already excavated at the start of the 19th century.These investigations revealed in the interior an octagonal grave-chamber of 4.5 m diameter built of massive sandstone blocks and accessed by a 12.75 m long, covered entrance (dromos) on the north. Since the finds made then have not survived and the gravechamber itself is no longer accessible, the structure can only be roughly dated to the 2nd or 3rd century. Location to main building/to next scattering of rubble and to the traffic-routes: The name of the enormous earthwork (Tonn) stems from the Latin tumba and the situation 10 m away from the Roman trunk road from Metz to Trier underlines its importance during ancient times. It must have been the burial place of a rich family, whose villa was sited some 600 m southeast of the tumulus. Literature: Carte archéologique du Grand-Duché de Luxembourg, Feuille 23 - Grevenmacher (Luxembourg 1974), 17 (site 23 A-125). - Wigg 1993, 148-149, cat. no. 16 with pl. 12.

Fig. 5. Topographical situation of the Roman villa and nearby grave monument (tumulus) at Flaxweiler.

Place: Gillenfeld (D) Site: Grubenberg Summary of the features and research history: During a geomagnetic survey in 2005 a Roman funerary monument with a circular enclosure wall and an additional, u-shaped wall in the foreground was registered. The monument’s foundations of 5 by 5 m had already been discovered in 1910 and the whole stone material completely robbed out. A dating is impossible. Location to main building/to next scattering of rubble and to the traffic-routes: The funerary monument lies northwest beyond the pars rustica of an extensive axial villa. It stands considerably higher than the main building. Because of the u-shaped wall in the foreground the monument’s main alignment must have been oriented towards the Alf valley rather than towards the villa. Opposite the monument on the other side of the valley ran the Roman road linking Bad Bertrich with the eastwest road from Jünkerath to Andernach. There is also a sanctuary there (Henrich/Mischka 2006). Two tumuli with stone enclosure walls some 180 m and 350 m from the funerary monument are also aligned to the road and were either not or hardly visible from the main building. The monument could be seen afar from the west, the north and the south, i.e. from the direction of the road. Literature: Henrich/Mischka 2005.













Fig. 6. Topographical situation of the Roman villa and associated grave monuments at 



Gillenfeld and the Roman road.

Place: Grevenmacher-Potaschbierg (L) Site: An der Heck Summary of the features and research history: In 1972/73 archaeological investigations were carried out at a large walled funerary enclosure (30 by 28 m) on a plateau west of the Moselle valley. In the centre of the enclosure, foundations of a funerary monument of huge proportions (13.75 by 5.5 m) and unusual groundplan still surviving some 2 m high came to light. During the excavation over 40 m³ of limestone blocks from the monument’s superstructure were recovered, of which around 1,000 display sculptural remains. The surviving architectural pieces allow the reconstruction of a multi-storey monument at least 12 m high, which was decorated on all sides with rich sculptural designs. These include reliefs with mythological scenes, as well as motives from the everyday life of the constructor’s family. Of special importance is that the monument, which was erected around AD 130/140, also possessed relief scenes clearly depicting the growing, making and drinking of wine. Following the monument’s destruction in the second half of the 3rd century, some 20 inhumations were laid within the funerary area during the second half of the 4th century. Location to main building/to next scattering of rubble and to the traffic-routes: The funerary enclosure lies some 10 m from the important Roman road from Metz to Trier. The monument’s main façade faces the road and also a villa lying 500 m further north and surely belonging to it. Literature: Carte archéologique du Grand-Duché de Luxembourg, Feuille 23 – Grevenmacher (Luxembourg 1974), 30 (site 23 C-23). - Metzler/Zimmer/Thill 1974, 119-162. - Metzler/Zimmer 1975, 322-323.



Fig. 7. Topographical situation of the grave monument and the associated villa at Grevenmacher.

Place: Hetzerath (D) Site: Hambuch Summary of the features and research history: During roadworks in 1972 the foundations (6.5 by 7 m) of a funerary monument were uncovered and destroyed. A dating is impossible. Location to main building/to next scattering of rubble and to the traffic-routes: 600 m east of the funerary monument is an extensive spread of rubble, which may be interpreted as the site of a villa and is situated lower than that of the foundations. Some 200 m to the north, beyond the stream of the Rassel, runs the Roman road from the Moselle towards the River Ahr. Literature: Trierer Zeitschrift 37, 1974, 279-280; Steinhausen 1932, 127.

Fig. 8. Topographical situation of the grave monument and the associated villa at Hetzerath.

Place: Igel (D) Site: Village Summary of the features and research history: Today the Igel Column is an in situ free-standing, completely surviving funerary pillar, which has been investigated and documented since the early 16th century. Because of the numerous, partly very well surviving relief scenes, the Igel Column is referred to for comparisons and reconstructions of Roman funerary pillars. The represented scenes and the inscription allow the grave monument to be assigned to the Secundinii family of cloth merchants. Old pictures of the monument, as well as investigations accom-



panying trenching in the immediate and further surroundings provide indications of the sites of further monuments. The funerary monument dates to the beginning of the 3rd century. Location to main building/to next scattering of rubble and to the traffic-routes: Owing to recent building development in the whole area, an exact location of a villa is not possible. In 1911 remains of structures below and some 400 m distant from the Igel Column were interpreted as the remains of a villa, without, however, more precise details of a groundplan being made. The funerary monument lies 200 m from the bank of the Moselle and higher than the features interpreted as the main building. Literature: Dragendorff/Krüger 1924. - Carte archéologique du Grand-Duché de Luxembourg, Feuille 19 – Mertert-Wasserbillig (Luxembourg 1983), 40 (sites 19 B-70 and B-71).

Fig. 9. Topographical situation of the grave monument and the possible remains of a villa at Igel.

Place: Kirf (D) Site: Mischelsbüsch Summary of the features and research history: In 1920 the foundations (3.8 by 3.5 m) and enclosure wall with semi-cylindrical coping-stones were archaeologically investigated. The surviving sculptural remains allow a dating to the 2nd or 3rd century. Location to main building/to next scattering of rubble and to the traffic-routes: Some 250 m from the funerary monument is a scattering of rubble, which is to be interpreted as the villa and which lies below the monument. East of the villa at a distance of 200 m from the monument runs the Roman road towards Trier. Literature: Krüger 1924. - Carte archéologique du Grand-Duché de Luxembourg, Feuille 27 – Remich (Luxembourg 1977), 43 (site 27 C-50).



Fig. 10. Topographical situation of the grave monument and the villa at Kirf.

Place: Mehren/Schalkenmehren (D) Site: Weinfeld Summary of the features and research history: An extensive spread of rubble known since the 19th century on the eastern edge of the Weinfeld Maar was investigated during a geomagnetic survey in 2006/2007. This revealed the main building of a villa with several, non-axially arranged auxiliary buildings over an area of 1.2 ha in all, as well as the foundations and enclosure wall of a sepulchral area with the foundation of a funerary monument. Hitherto, no datable finds or features are known. Location to main building/to next scattering of rubble and to the traffic-routes: The funerary monument is situated on the highest point of the villa’s immediate vicinity, much higher than the main building and the other structures of the villa. It lies 165 m from the main building and 65 m from the nearest building. Near the villa J. Hagen reconstructed a road from the Moselle in the direction of the road joining Jünkerath and Andernach. This hypothesis was confirmed by the evaluation of the settlement data, even though the exact course remains unknown. Owing to the monument’s good visibility in the topographically dominant location, the monument must have been visible from far away and, thus, also from the road. Literature: Henrich/Mischka 2009; Henrich 2006, 32; Hagen 1931, map 2.

Fig. 11. Topographical situation of the grave monument and the associated villa at Mehren.



Place: Nennig (D) Site: Mahlknopf Summary of the features and research history: Since the early 19th century excavations have taken place in the still 10 m high surviving tumulus (diameter 44.5 m), which includes a stone enclosure wall strengthened internally by 32 semicircular, solid buttresses, without the central burial having been discovered. A further, smaller tumulus some 85 m away has been located by aerial photography. A dating to the 2nd century is only possible through the stone enclosure wall. Location to main building/to next scattering of rubble and to the traffic-routes: The tumulus is situated 550 m away from the well-known main building with the Gladiator Mosaic and 300 m from the porticus, which runs parallel to the main building and to the separate bath-house without bonding into the building. The villa and tumuli were readily visible from the Moselle, which today flows some 2 km away. Both tumuli lie lower than the main building on the edge of the Moselle’s lower terrace. Literature: Wigg 1993, 165-166 cat. no. 27 (with older literature). - Carte archéologique du Grand-Duché de Luxembourg, Feuille 27 – Remich (Luxembourg 1977), 40 (site 27 B-254).

Fig. 12. Groundplan of the villa and two associated tumuli at Nennig.

Fig. 13. Topographical situation of the Roman villa and the associated tumulus at Nennig and the Late Roman grave-chamber at BechKleinmacher on the left bank of the Moselle.



Place: Newel (D) Site: Köncherwies Summary of the features and research history: The main building including the whole courtyard and auxiliary buildings of a Roman villa was excavated in 1962/63. Outside the courtyard lay a Gallo-Roman ambulatory temple with a directly adjacent funerary enclosure of 20 by 13 m surrounded by a wall. Inside this four tumuli and the foundations of a funerary monument of 8 by 3.5 m were documented. The whole funerary area dates to the 2nd and 3rd centuries. Location to main building/to next scattering of rubble and to the traffic-routes: The funerary monument and the tumuli lay on the Roman transverse route running from the Moselle valley near Pfalzel and Biewer via Besslich to the Roman road Trier to Bitburg. The funerary area lay higher than the courtyard some 90 m from the villa. Literature: Cüppers/Neyses 1971. - Carte archéologique du Grand-Duché de Luxembourg, Feuille 14 – Rosport (Luxembourg 1985), 40 (sites 14 D-13 bzw. D-12).

burial compound

Fig. 14. Groundplan of the villa and the associated grave monuments at Newel.

Place: Remerschen (L) Site: Mecheren Summary of the features and research history: Following the finds of various decorated relief blocks from this monument situated on the river-terrace of the Moselle in the mid-19th century, excavations in 1970 and 1972 revealed not only the solid foundations of the monument (8.6 by 5.35 m), but also numerous further stone blocks from its superstructure. The remains of reliefs brought to light were enough to allow the reconstruction of a monument over 6 m high from the middle of the 2nd century. All four faces were adorned with rich decoration and various scenes of the everyday life of a major wine-maker and -merchant. Near the monument a walled area of substantial size (17.3 by 5.1 m) was also uncovered, in which two Late Roman graves were found. Location to main building/to next scattering of rubble and to the traffic-routes: The imposing villa with residence and auxiliary buildings belonging to it was unfortunately destroyed by the intensive gravel-extraction in the immediate proximity to the Moselle. Only traces of a largish, partly cellared structure could be investigated. The funerary monument rose between the villa and the bank of the Moselle only 45 m away and presented its main façade towards the river.



Literature: Thill 1970b, 1972b, 1973a. - Metzler/Zimmer 1975, 324-327. - Carte archéologique du GrandDuché de Luxembourg, Feuille 30 – Remerschen (Luxembourg 1979), 17-18 (site 30 A-105). - Krier 1992.

Fig. 15. Topographical situation of the grave monument and the associated villa at Remerschen.

Place: Strotzbüsch (D) Site: Hasselbüsch Summary of the features and research history: The tumulus was investigated in 1821, when a 2.41 by 1.88 by 1.88 m large, vaulted chamber with lateral dromos was revealed. The chamber and dromos were covered by the tumulus and constructed from wellfitting, dressed and non-cemented sandstones. Apart from a chair of white sandstone, no finds are known. Indications of dating are lacking. Location to main building/to next scattering of rubble and to the traffic-routes: There are no Roman sites known in the immediate vicinity of the tumulus. This could be the result of extensive woodland in the neighbouring area to the east, which impedes the finding of ruined sites. Literature: Henrich 2006, 207, no. 350 (with older literature). Place: Wasserbillig - Lellig (L) Site: Weiler Summary of the features and research history: Between 1970 and 1972 chance played a role in the woodlands of Weiler on a high plateau above the Moselle valley between Wasserbillig and Lellig, when several Roman funerary enclosures and monuments were discovered and archaeologically examined. Over a distance of 700 m two walled funerary gardens with cremation burials, another walled enclosure with a monumental funerary pillar in the centre and the foundations of small funerary stelae on the external sides, as well as the remains of a smaller pillared monument were uncovered. Since the different mortuary features are aligned in a clear chronological order from the late 1st to well into the 3rd century, as well as an inhumation grave of the Constantinian period, it is not impossible that they all belonged to the same Roman villa. A Roman building situated in the proximity of a road-junction, opposite the large funerary pillar, seems to belong to the Late Roman period, i.e. at a time when the funerary monument, which lay 18 m away, had already been destroyed. Location to main building/to next scattering of rubble and to the traffic-routes: The features at Wasserbillig-Lellig seem to be a proper row of graves in the direct vicinity of the Roman road, which here measured 4·5 m wide and ran from Trier into the Ardennes and on this stretch linked



the vici of Wasserbillig and Altrier with one another. The corresponding villa is believed to be in an extensive scattering of rubble (Tempelsklouster) situated 465 m away and some 20 m lower down. Literature: Thill 1970a, 1971, 1972b. - Carte archéologique du Grand-Duché de Luxembourg, Feuille 18 – Betzdorf (Luxembourg 1973), 28 (site 18 C-122). - Metzler/Zimmer 1975, 320-322.

Fig. 16. Topographical situation of the grave monuments and the villa remains at Wasserbillig.

Place: Weiler-la-Tour (L) Site: Seitert Summary of the features and research history: In 1971 the foundations of a round funerary structure of 7.2 m in diameter were uncovered in the woods of Seitert, southeast of the village of Weiler-la-Tour. Only one fragment of a huge console’s cornice from the monumental superstructure of the grave’s rotunda could be recovered. The monument stood upon a small, high plateau of a long ridge which dominated the landscape. Location to main building/to next scattering of rubble and to the traffic-routes: The monument stood out some 15 to 20 m above the site of the villa with axial courtyard at AspeltKleppen/op de Maueren some 300 m away, to which without doubt it belonged. The important Roman trunk road of Lyon – Metz – Trier – Rhine ran at a distance of only 2.5 km east of the monument. The Roman vicus of Dalheim also lay within viewing distance (3 km). Literature: Thill 1971, 499; 1973, 291, 297, figs. 2-3; 1975, 303. - Wilhelm 1974, 16-17 no. 4. - Carte archéologique du Grand-Duché de Luxembourg, Feuille 26 – Mondorf-les-Bains (Luxembourg 1977), 63-64 (sites 26 D-3 and D-14, D-15). - Wigg 1993, 188-189 cat. no. 41 with pl. 43.



Fig. 17. Topographical situation of the grave monument and the associated villa at Weiler-la-Tour.

Place: Wintersdorf an der Sauer (D) Site: Zinzigbüsch Summary of the features and research history: In 1879 and 1881 the remains of four statues were found some metres above the River Sauer on a 6.3 m wide and 5 m deep terrace paved with large sandstone slabs. The terrace was surrounded by an enclosure wall open to the river and covered by semi-cylindrical coping-stones. Hettner reconstructed individual monuments with free-standing statues on the terrace. However, the so-called terrace could have been the foundation of a pillared funerary monument. Location to main building/to next scattering of rubble and to the traffic-routes: In the immediate and wider surroundings of the funerary monument walls and probably building remains, allegedly including an 11 m long ‘corridor’ leading to a building, were certainly documented, but a definite allocation of function is not possible. Literature: Hettner 1903, 57. - Steinhausen 1932, 377-378. - Carte archéologique du Grand-Duché de Luxembourg, Feuille 19 – Mertert-Wasserbillig (Luxembourg 1983), 28 (site 19 A-120).



re f ere n ces Andrikopoulou-Strack, J.-N., 1986: Grabbauten des 1. Jahrhunderts n.Chr. im Rheingebiet. Untersuchungen zu Chronologie und Typologie, Köln/Bonn (Bonner Jahrbücher, Beiheft 43). Baltzer, M., 1983: Alltagsdarstellungen der treverischen Grabdenkmäler, Trierer Zeitschrift 46, 7-151. Béal, J.-C./G. Charpentier, 2006: Nouvelles remarques sur le mausolée d’Andance (Archèche), in J.-C. Moretti/D. Tardy (eds), L’architecture funéraire monumentale. La Gaule dans l’empire romain. Colloque Lattes 2001, Paris (Archéologie et histoire de l’art 24), 337-354. Cüppers, H., 1990, Die Römer in Rheinland-Pfalz, Stuttgart. Cüppers, H./A. Neyses, 1971: Der römerzeitliche Gutshof mit Grabbezirk und Tempel bei Newel (Kreis Trier-Land), Trierer Zeitschrift 34, 143-225. Dragendorff, H./E. Krüger, 1924: Das Grabmal von Igel. Römische Grabdenkmäler des Mosellandes und angrenzender Gebiete I, Trier. Ebel, W., 1989: Die römischen Grabhügel des ersten Jahrhunderts im Treverergebiet, Marburg (Marburger Studien zur Vor- und Frühgeschichte 12). Fellmann, R., 1957: Das Grab des Lucius Munatius Plancus bei Gaeta, Basel. Freigang, Y., 1997a: Die Grabmäler der gallo-römischen Kultur im Moselland. Studien zur Selbstdarstellung einer Gesellschaft, Jahrbuch des Römisch-Germanischen Zentralmuseums 44, 277-440. Freigang,Y., 1997b: Die Bedeutung der Bekleidung und Attribute auf Grabmälern im nordöstlichen Teil der Provinz Gallia Belgica, in B. Djuric/I. Lazar (eds), Akten des IV. internationalen Kolloquiums über Probleme des provinzialrömischen Kunstschaffens Mai 1995, Ljubljana, 107-117. Gabelmann, H., 1977: H. Gabelmann, Römische Grabbauten in Italien und den Nordprovinzen, in U. Höckmann/A. Krug (eds.), Festschrift für Frank Brommer, Mainz, 101-117. Graen, D., 2008: „Sepultus in villa“. Die Grabbauten römischer Villenbesitzer, Hamburg (Antiquitates 46). Hagen, J., 1931: Die Römerstrassen der Rheinprovinz. 2. Auflage, Bonn. Heinen, H., 1985: Trier und das Trevererland in römischer Zeit, Trier (2000 Jahre Trier I). Henrich, P., 2003: Die römischen Grabdenkmäler von Duppach-Weiermühle. Ein Vorbericht der Grabungen 2003, Funde und Ausgrabungen im Bezirk Trier 35, 46-52. Henrich, P., 2006: Die römische Besiedlung der westlichen Vulkaneifel, Trier (Trierer Zeitschrift, Beiheft 30). Henrich, P., 2010: Die römische Nekropole und die Villenanlage von Duppach-Weiermühle, Vulkaneifel, Trier (Trierer Zeitschrift, Beiheft 33). Henrich, P./C. Mischka, 2005: Die monumentale römische Grabanlage von Gillenfeld, „Grubenberg“, Kreis Daun, Funde und Ausgrabungen im Bezirk Trier 37, 21-28. Henrich, P./C. Mischka, 2006: Der römische Tempelbezirk von Gillenfeld, „Etzerath“, Kreis Daun, Funde und Ausgrabungen im Bezirk Trier 38, 25-33. Henrich, P./C. Mischka, 2009: Die römische Villa mit Grabdenkmal und die frühneuzeitliche Wüstung „Weinfeld“ bei Mehren/Schalkenmehren, Landkreis Vulkaneifel, Funde und Ausgrabungen im Bezirk Trier 41, 14-25. Henrich, P./M. Tabaczek, 2003: Die Gräberstraße und ein Nebengebäude der römischen Villa von Duppach-Weiermühle, Archäologie in Rheinland-Pfalz 2003, 56-59. Hettner, F., 1903: Illustrierter Führer durch das Provinzialmuseum in Trier, Trier. Kaszab-Olschewski,T., 2007: Ländliche Alltagsszenen auf Grabreliefs der Nordwestprovinzen, in E. Walde/B. Kainrath (eds), Die Selbstdarstellung der römischen Gesellschaft in den Provinzen im Spiegel der Steindenkmäler. Akten des IX. Internationale Kolloquiums über Probleme des provinzialrömischen Kunstschaffens Innsbruck 2005, Innsbruck (IKARUS 2), 173-182. Kempchen, M., 1995: Mythologische Themen in der Grabskulptur. Germania Inferior, Germania Superior, Gallia Belgica und Raetia, Münster (Charybdis 10). Koethe, H., 1936: Straßendorf und Kastell bei Jünkerath, Trierer Zeitschrift 11, Beiheft 1, 50-106.



Kolling, A., 1986: Die Villa von Bierbach, Homburg (Forschungen im römischen Schwarzenacker 2). Kremer, G., 2009a: Das frühkaiserzeitliche Mausoleum von Bartringen (Luxemburg), mit einem Beitrag von J. Krier, Luxemburg (Dossiers d’Archéologie du Musée national d’Histoire et d’Art 12). Kremer, G., 2009b: Der Grabbau eines flamen aus Mersch und die kaiserzeitlichen Grabbauten des Moselund Rheingebietes, in D. Boschung (ed.), Grabbauten des 2. und 3. Jahrhunderts in den gallischen und germanischen Provinzen. Akten des Internationalen Kolloquiums Köln 22. bis 23. Februar 2007, Wiesbaden (ZAKMIRA 7), 109-135. Krier, J., 1992: Das Grabdenkmal eines römischen Winzers an der Mosel bei Remerschen, in J. Lichardus/A. Miron (ed.), Der Kreis Merzig-Wadern und die Mosel zwischen Nennig und Metz, Stuttgart (Führer zu archäologischen Denkmälern in Deutschland 24), 256-259. Krier, J., 2003: Ein neuer Reliefblock aus Bartringen und die Grabmonumente mit Reiterkampfdarstellungen an Mosel und Rhein, in P. Noelke et al. (eds), Romanisation und Resistenz in Plastik, Architektur und Inschriften der Provinzen des Imperium Romanum. Neue Funde und Forschungen. Akten des VII. Internationalen Kolloquiums über Probleme des provinzialrömischen Kunstschaffens Köln 2. bis 6. Mai 2001, Mainz, 255-263. Krier, J., 2007: Ein neuer Grabrundbau des 1. Jahrhunderts n. Chr. in Goeblingen (Luxemburg), in E. Walde/B. Kainrath (eds), Die Selbstdarstellung der römischen Gesellschaft in den Provinzen im Spiegel der Steindenkmäler. Akten des IX. Internationalen Kolloquiums über Probleme des provinzialrömischen Kunstschaffens Innsbruck 2005, Innsbruck (IKARUS), 159-171. Krüger, E., 1924: Der römische Grabturm mit Balustrade im Mischelsbüsch, Germania 8, 32-39. Lafon, X., 2002: La localisation des monuments funéraires, in C. Landes (ed.), La mort des notables en Gaule romaine, Lattes, 113-120. Lafon, X./A.M. Adam, 1993, Des morts chez les vivants? Tombes et habitat dans la France du nord-est, in A. Ferdière (ed.), Monde des morts, monde des vivants en Gaule rurale. Actes du colloque Archéa/Ager 1992, Tours (Revue archéologique du Centre de la France, supplément 6), 113-120 Langner, M., 2001: Szenen aus Handwerk und Handel auf gallo-römischen Grabmälern, Jahrbuch des Deutschen Archäologischen Instituts 116, 299-356. Lefèbvre, L., 1975: Les sculptures gallo-romaines du Musée d’Arlon, Bulletin de l’Institut Archéologique du Luxembourg 51, 1-88. Lefèbvre, L., 1990: Le Musée luxembourgeois Arlon, Bruxelles (Musea Nostra 20). Leveau, Ph., 2000: Mausolées au bord de fleuves. Aristocratie commerçante et travaux de correction du Rhône, Caesarodunum 33/34, 1999/2000, 201-215. Mariën, M.E., 1945: Les monuments funéraires de l’Arlon romain, Arlon. Martin-Kilcher, S., 1993: Situation des cimitières et tombes rurales en Germania superior et dans les régions voisines, in A. Ferdière (ed.), Monde des morts, monde des vivants en Gaule rurale. Actes du colloque Archéa/Ager 1992, Tours (Revue archéologique du Centre de la France, supplément 6), 153-164. Meates, G.W., 1979: The Roman Villa at Lullingstone, Kent, Vol. 1: The site, London (Monograph Series of the Kent Archaeological Society 1). Mertens, J., 1968: Nouvelles sculptures romaines d’Arlon, Bulletin de l’Institut Archéologique du Luxembourg 44, 22-35, with pl. I-XIX. Mertens, J., 1973: Le rempart romain d’Arlon, Bruxelles (Archaeologicum Belgii Speculum VII). Metzler, J./C. Gaeng et al., 2009: Goeblange-Nospelt. Une nécropole aristocratique trévire, Luxembourg (Dossiers d’Archéologie du Musée national d’Histoire et d’Art 13). Metzler, J./J. Zimmer/G. Thill, 1975: Grosses gallo-römisches Grabdenkmal mit Bezirk und Bestattungen bei Grevenmacher („Heck“), Hémecht 26, 119-162. Metzler, J./J. Zimmer, 1975: Archäologischer Rundgang durch Luxemburg, Hémecht 27, 307-330. Metzler, J./J. Zimmer/L. Bakker, 1981: Ausgrabungen in Echternach, Luxemburg. Numrich, B., 1997: Die Architektur der römischen Grabdenkmäler aus Neumagen. Beiträge zur Chronologie und Typologie, Trier (Trierer Zeitschrift, Beiheft 22).



Precht, G., 19792: Das Grabmal des L. Poblicius, Köln. Rolland, H., 1969: Le mausolée de Glanum (Saint-Rémy-de-Provence), Paris (Gallia suppl. 21). Schucany, C., 1995: Eine Grabanlage im römischen Gutshof von Biberist-Spitalhof, Archäologie in der Schweiz 18, 142-154. Schucany, C., 2000: An elite funerary enclosure in the centre of the villa of Biberist-Spitalhof (Switzerland) – a case study, in J. Pearce/M. Millett/M. Struck (eds), Burial, society and context in the Roman world, Oxford, 118-124. Schucany, C., 2006: Die römische Villa von Biberist-Spitalhof/SO (Grabungen 1982, 1983, 1986-1989). Untersuchungen im Wirtschaftsteil und Überlegungen zum Umland, Remshalden. Steinhausen, J., 1932: Ortskunde Trier-Mettendorf, Bonn (Archäologische Karte der Rheinprovinz 1.1). Thill, G., 1970a: Ummauerter römischer Friedhof bei Lellig (1. Jahrhundert n.Chr.), Hémecht 22, 371-382. Thill, G., 1970b: Um eine „versunkene“ Römervilla bei Remerschen, Hémecht 22, 455-467. Thill, G., 1971: Neue Grabungsergebnisse in „Weiler“ zwischen Wasserbillig und Lellig, Hémecht 23, 489506. Thill, G., 1972a: Das Grabdenkmal eines gallo-römischen Großwinzers und Weinhändlers bei Remerschen, Hémecht 24, 209-210. Thill, G., 1972b: Nachtrag zum Grabungsbericht über den Gräberbezirk „Weiler“ (Wasserbillig-Lellig), Hémecht 24, 367- 369. Thill, G., 1973a: Zum römerzeitlichen Grabdenkmal auf « Mecheren » bei Remerschen (ein Nachtrag), Hémecht 25, 195-199. Thill, G., 1973b: Travaux et réalisations au Musée d’Histoire et d’Art, Luxembourg 1971 (Rapport du Conservateur), Publications de la Section Historique de l’Institut Grand-Ducal 88, 289-299. Thill, G., 1975: Zur archäologischen Denkmalpflege in Luxemburg, Hémecht 27, 297-306. Von Hesberg, H./P. Zanker (eds), 1987: Römische Gräberstraßen. Selbstdarstellung – Status – Standard, München (Bayerische Akademie der Wissenschaften. Philosophisch-Historische Klasse, Abhandlungen Neue Folge, Heft 96). Von Massow, W., 1932: Die Grabmäler von Neumagen. Römische Grabmäler des Mosellandes und der angrenzenden Gebiete II, Berlin. Waurick, G., 1973: Untersuchungen zur Lage der römischen Kaisergräber in der Zeit von Augustus bis Constantin, Jahrbuch des Römisch-Germanischen Zentralmuseums Mainz 20, 107-146. Wigg, A., 1993: Grabhügel des 2. und 3. Jahrhunderts an Mittelrhein, Mosel und Saar, Trier (Trierer Zeitschrift, Beiheft 16). Wilhelm, E., 1975: Pierres sculptées et inscriptions de l’époque romaine. Musée d’Histoire et d’Art Luxembourg, Luxembourg. Willer, S., 2005: Römische Grabbauten des 2. und 3. Jahrhunderts im Rheingebiet, Mainz (Bonner Jahrbücher, Beiheft 56). Zinn, F., 2001: Untersuchungen zu den Wagenfahrtdarstellungen auf provinzialrömischen Denkmälern, Kölner Jahrbuch 34, 141-266.



Roman rural settlements in Flanders. Perspectives on a ‘non-villa’ landscape in extrema Galliarum Wim De Clercq1 1 Introduction 2 The Menapian civitas in extrema Galliarum 3 The rural settlement landscape 4 A model for the socio-economic development of a ‘marginal’ region 5 Peasants farming for Rome? 6 Material culture and social identity in a non-villa landscape 7 Conclusion References

i i n troductio n For a long time, Belgian archaeology of the Roman period was dominated by intellectual discourses often rooted in colonial and ideological backgrounds, leaving little or no place for the agency of indigenous groups nor for the cultural diversity which developed under Rome’s global umbrella. The contact of these societies with Rome was primarily seen as a one-way trajectory in which superior Roman culture was adopted or not adopted by native groups. Differences as reflected in material culture were very much viewed as culturally relevant levels of ‘romanisation’.2 This idea was particularly prevalent in research into rural landscapes, which up till the early 1980s mainly focused on the study of vici, villae and related phenomena situated in the Belgian part of the fertile loess belt stretching from Picardy in northern France across central Belgium to the German Lower Rhineland. A picture emerged of a profoundly Romanised rural landscape crowded with villae, tumuli and some vici, and crowned by the presence of the municipium of Atuatuca Tungrorum. The western and northern part of the country, however, remained a vast terra incognita, notwithstanding early excavation work in the Saxon Shore fort of Oudenburg3 and two inventorying projects which mapped significant amounts of Roman-period material culture in the Waasland and coastal areas.4 This unbalanced picture was partially adjusted in the early 1980s when excavations in the Antwerp and Kempen region in the northern part of Flanders revealed ‘native’ settlements consisting of post-built

1 

This paper originates from my PhD research on the

2 

De Maeyer 1937 and even the more recent ideas devel-

transformations in the local communities of the civitas

oped in Vermeulen 2001 (note 7 below) and Nouwen

Menaporium between 100 BC and 400 AD (De Clercq

2006 and 2008. The latter still leaves little or no place for

2009). I would like to thank Nico Roymans for inviting

the agency of local communities.

me to publish in this volume and for his critical remarks

3 

on the draft.

4 

Mertens/Van Impe 1971. Thoen 1967; Thoen 1978.



Fig. 1. Position of the study area (yellow) in the civitas Menapiorum in Belgic Gaul.

stable-houses.5 This research largely benefited from neighbouring work by Dutch colleagues working in the Maas-Demer-Scheldt area and the Dutch Kempen. They developed social anthropological interpretation models which allowed us to understand the regional diversity at work in the Romanisation process. Slofstra for instance emphasised the importance of clientship relations and dependent labour in understanding villa and non-villa modes of production, while Roymans stressed the importance of the force of old cultural traditions and the nature of the landscape, as well as the conservative nature of the pastoral societies inhabiting the sand landscapes.6 In the late 1980s Vermeulen carried out an intensive field survey and some limited excavations in the region between the rivers Lys and Scheldt, south of Ghent, a micro-region situated in the civitas Menapiorum. This research, combined with some small-scale excavations, shed light for the first time on the nature of the Roman-period occupation of the sandy soils of west and north-western Flanders. He suggested that society was only very superficially Romanised.7 Differences in material culture and cultural diversity were interpreted in an economic framework, such as access to markets and opportunities

5 

civitas was economically and culturally on a low rung of

6 

the ladder.’

De Boe/Lauwers 1981; De Boe 1988. Slofstra 1991; 2002; Roymans 1996, 1999.

7 

Vermeulen 1992; 2001, 20-21: ‘the northern part of the



connected with the global economic network or the army. The pattern of a settlement landscape on the sandy soils, consisting of so-called native-Roman settlements with a manifest absence of villae as centres of organised agricultural production, as well as the near-absence of other forms of material culture relating to Mediterranean-style patterns of consumption and display of identity, were considered signs of a ‘less-Romanised’ marginal society living on the edges of Gaul. During the past decade, archaeological research in the north-western inland areas of the civitas Menapiorum has increased exponentially, resulting especially for the Roman period in a broad set of context-based data in the area bordering the North Sea coast (fig.1). Although more excavation work is definitely still needed, the north-western sand region has now become one of the best documented areas in Belgium. This record constitutes an important gain in information but first and foremost it provides an adjusted interpretative framework on the socio-economic and cultural transformation of the area in an attempt to understand its integration in the wider context of the Roman Empire.

2 the me n a p ia n ci v itas i n e x trema galliarum Based on geography and on historical sources including Roman and medieval text fragments, we can reasonably assume that the civitas territory of the Menapii stretched along the southern North Sea coast from the mouth of the river Scheldt in the north – nowadays the Oosterschelde – to the mouth of the river Aa in the south.8 Its eastern borders separating it from the civitas Nerviorum were marked by the rivers Scheldt and Lys. The research presented in this paper focuses on the northern part of the civitas, a region situated in the modern-day territories of Belgian Flanders and the south-western Netherlands. The civitas capital Castellum Menapiorum, the modern-day Cassel in northern France, was eccentrically positioned in the southern part of the civitas, leaving the northern parts far away from the administrative centre. In political and geographical terms, the civitas in point was situated on the northern fringes of the province of Gallia Belgica. When in 69 AD the Batavian rebel Julius Civilis sent his troops south over the Meuse to ravage and loot the territories of the Menapii and their southern neighbours, the Morini, Tacitus associated these regions with extrema Galliarum, the extremes of Gaul.9 Virgil even called the Morini extremi hominum.10 These expressions leave little doubt as to the ephemeral character attributed to these tribes from both a Roman political and geographical point of view. Contemporary descriptions of the landscape, when considered from a productive perspective, also implicitly echoed this marginal character. Caesar reported dense forests and vast marshes11 and later on Strabo wrote about a landscape of vast forests of thorny shrubs. He described the northern parts of Gaul, generally speaking, as densely forested areas in which half-wild boars were raised.12 The famous Menapian ham celebrated by Martial in the 1st century13 and later still mentioned in Diocletian’s Price Edict, also suggests the presence of forests to raise pigs.14 Although forests could possess an economic value, for instance as silva glandaria,15 the scarce contemporary accounts for the region paint an unequivocal picture of a produc-

8 

De Laet 1961, based on medieval text sources. His

10 

demarcation was partially contested by Bogaers (1971).

11 

The use of Thiessen polygons produces another outline

12 

of the civitas. We have chosen to rely on the delimitation

13 

suggested by De Laet which coincides well with natural

14 

boundaries and the distribution of sanctuaries.

15 

Virgil, Aeneis, VIII, 727. Caesar, BG 6.7. Strabo, Geographica, 3. 4-5. Martial, Epigrammata, XIII, 54. Edictum de pretiis rerum venalium 4.8. Meiggs 1982.

9 

Tacitus, Hist., 4. 28.



tive – i.e. agricultural (arable) and ecological –marginality which seems to correspond in general terms with the landscape context of at least the northern part of the civitas.16 Based on geomorphology, the region can roughly be divided into two different types of landscape. The first comprises the coastal and estuarine areas. It is of a Holocene nature and consisted in Roman times of marine clay and peat, influenced by tidal movements of the North Sea. It was a dynamic landscape, penetrated by creeks and probably difficult to inhabit on a permanent basis. Nevertheless, this area was suitable for activities such as pastoralism, salt and peat extraction, fishing and shell harvesting. More stable occupation and crop cultivation was confined to the locally dried-out peat areas. The second landscape type is of a Pleistocene nature and is situated between the coastal plain and the Scheldt basin. This relatively flat sand landscape is intersected by several smaller brooklets and depressions which alternate with somewhat dryer sand ridges. Due to the high acidity and the compaction resulting from migration processes of humic acids and iron, these sandy soils are infertile on both a chemical and physical scale; they could not produce large crop yields.17 In the region between Bruges and Ghent, a tertiary clay formation (cuesta) rises above the sand landscape; it is considered even more infertile than the sandy soils, suitable only for woods or poor pastures. Obviously neither the marine nor the sandy landscapes enabled a large-scale surplus production in crops and cereals in particular, in contrast to the more generous loess soils located further east in Nervian and Tungrian territories. The presence of the Nervian negotiator frumentarius M. Liberius Victor at Nijmegen18 or the Nervian trader and baker Tertinius Secundus in Cologne19 indicate the important economic potential of these neighbouring soils east of the river Scheldt for creating a cereal surplus. The epigraphic evidence also testifies to a market-oriented mode of production and a trade in cereals with the Rhine army. Given the ecological parameters described, we cannot assume a similar situation for the northern part of the Menapian civitas. However, the civitas held a stipendiary status which meant that taxes had to be paid to the Roman state in one way or another. It is likely that from the early 1st century onwards a part of this levy was paid in manpower as auxiliary forces for the army, likewise the neighbouring civitates.20 Pre-Flavian military recruitment in the area can be implicitly assumed from historic sources; later, in 122 and 124 AD, a Cohors I Menapiorum was explicitly mentioned. Stationed in Britannia, it was involved in the construction of Hadrian’s Wall.21 However, we should not overestimate the importance of military recruitment as a method of tax payment in the civitas Menapiorum, especially compared with the situation in the Batavian or Nervian civitates, which delivered substantially more cohortes and alae to Rome. Also, payment in manpower diminished over time.22 It is more logical to assume that the Menapian tribute was largely paid in kind (e.g. as annonna militaris) and that commodities such as animal products (wool, salted meat, dairy products) or salt played a primary role in this transaction. I will suggest later on that this resulted in an economic intensification and presumably also an increasing social stratification in the local communities.

16 

Groenewoudt (2009, 149-151) offers a good discussion on the term marginal landscape, defining it as an area of

19 

CIL XIII 8338. Cf. Galsterer/Galsterer 2010, no. 430, who think of a trader in millstones.

low agricultural productivity and low population density

20 

during a given period. He also points to the relativity

21 

Alföldy 1968; Roymans 1996, 22-23; 86-88. The Cohors I Menapiorum is mentioned on two military

of the concept since it is influenced by economic, geo-

diplomas: one from Stannington (UK) (CIL XVI 70 =

graphical, social and cognitive parameters, varying over

CIL VII 1195 = RIB 2401, 6) and one from Brigetio

time. On this subject, see also Turner/Young 2007. 17 

Langohr 2004, 182-183.

18 

CIL XIII 8725.



(Hungary) (CIL XVI 69). 22 

Aarts 2005, 65.

3 the rural settleme n t la n dsca p e settleme n t chro n olog y Based on well-dated excavated settlement contexts, the occupation density in the northern part of the civitas in the period 200 BC-400 AD, expressed in occupation phases of 25 years, increased at a fast pace during the 1st century AD and reached a peak in the 2nd century (fig. 2). While this picture is no doubt partially biased by wider patterns of the importation of well datable material culture, it is also increasingly confirmed by radiocarbon and dendrochronological dating. When observed as a whole rather than in occupation phases, both the Pleistocene and marine landscapes show a tendency towards the discontinued occupation of sites with Late Iron Age roots after the Flavian period and the intensified creation of new settlements in the Flavian era. Remarkably, Late Iron Age sites on the Pleistocene sand are rare, as are their Early Roman equivalents. This seems to be a historic fact in large parts of this northern area, notwithstanding intensive excavation work in recent years and a later Iron Age pottery style which is well recognisable and very similar to the northern French style group. This image contrasts profoundly with the more loamy, fertile soils in the south of Flanders and with the adjacent parts of northern France which seem to represent a more sustained continuation of settlement activity.23

Fig. 2. Density of rural settlements over time in the northern part of the Menapian civitas, expressed in phases of 25 years and divided over the Pleistocene and coastal landscapes.

23 

For example, the vast settlement landscapes revealed by research in the Lille region: Quérel 2007.



During the Flavian and Antonine periods, the landscape seems to have been most densely populated. However, in the inland Pleistocene area the vast majority of farms were already abandoned before AD 200 and few of them resumed afterwards. Those which survived were often restructured or defended, or new sites were created at other locations nearby. This major event seems to go hand in hand with the installation of military camps along parts of the coast and at strategic inland positions. For the 3rd century and unfortunately based on little contextual data, Hugo Thoen noted thriving habitation of the Flemish coastal plain,24 which we should understand more objectively as a peak in the occurrence of a few well datable find categories such as samian pottery. However, this picture could be related to the strong Roman military presence in the area during this century, thereby not necessarily reflecting a local economic pattern. In later Roman times settlement density fell dramatically.

f arm structure o n the p leistoce n e sa n d A common feature of rural settlements in the northern Pleistocene part of the civitas is that they are enclosed by a ditch system.The enclosed area could be subdivided into yards or compounds. Excavations, aerial photography and field control have revealed large numbers of them, pointing to an intense occupation. The enclosures have different shapes and cover areas of about one ha on average. In most cases the enclosure was formed by one or more ditches; rather exceptionally a palisade or a combination of both elements was used (fig. 3). The shape of the ditch system varied substantially. In some cases we observe rectangular enclosures, in other cases curvilinear examples and combinations of both systems. Some multi-period sites such as the complex at Aalter-Langevoorde (see below) clearly show the use of rectangular enclosures well into the Late Iron Age (c. 150 BC) as is frequently attested in Northern France, where rectangular Late Iron Age ditched farmsteads clearly precede Roman-period villa complexes.25 Apart from the rectangular enclosures, other sites (such as a newly created farm complex from the beginning of the 2nd century AD at Kluizen) show ditch systems combining curvilinearity and angularity, pointing to an often unsystematic method of marking settlement boundaries. Within the study region we observe a general absence of systematic ordering of the landscape or of settlement layout, while in certain specific cases we see a more systematic approach and the use of Roman measures such as the actus. A more systematic pattern emerges however in the internal structure of the farms. In most cases the living areas were clustered close to the enclosure ditch and consisted of a timber house, a well and some outbuildings such as a four- to nine-post granary. However, these areas only took up a small part of the total enclosed area of the complex, which remained mostly bare of important structures.26 Both in structure and general morphology, the data available shows a clear parallel to the northern French ‘fermes indigènes’ but seems to contrast with the settlement structure attested on the sandy soils of the Kempen and Meuse-Demer-Scheldt area in northern Flanders and the southern part of the Netherlands, where farm buildings appear to have been more equally distributed across the settlement and enclosures seem to have been rarer or reserved for elite sites.27

24 

where these ‘low-density’ settlement areas are frequently

25 

considered to be ‘off-site phenomena’ and hence remain

Thoen 1981. Collart 1996; Malrain/Matterne/Méniel 2002, 153-154.

26 

This observation is an important part of the evaluation process in the context of developer-led archaeology,



underevaluated (De Clercq et al. 2011). 27 

Slofstra 1991.

house secondary building well / pool cremation burial

0

100 m

enclosure

Fig. 3. Enclosed rural settlements from the North-Menapian sandy landscape. 1 Aalter-Langevoorde, 2 Bruges-Refuge, 3-6 Evergem-Kluizendok.



One of the best documented farms is that of Aalter. This complex is the only example at present which shows a continuity of occupation from 200 BC to 200 AD. A rectangular enclosure was dug in about 200 BC. In its first stage at least part of it had a ritual meaning since many complete and half pots were deposited at one spot close to a post-lining. At its second stage it became a settlement enclosure. Just before the beginning of our era, the enclosure was abandoned and shortly afterwards a new large enclosure was dug. This consisted of two parallel ditches which were laid out in a rectangular configuration, covering an area of some two hectares, which considerably exceeds the mean surface of the farms, based on the current evidence. Alongside the southern enclosure a series of constructions including a house and a nine-post – later extended to a 12-post – granary were built. On the opposite side in between the two enclosure ditches a heavy four-post construction was found, presumably another granary or even a tower. The granary occupied a prominent, central position in the site layout and, like the Iron Age sites in northern France,28 we may assume a certain mise-en-scène, a socially constructed spatial layout displaying a successful harvest and hence the prosperity of the family. The house itself was rather small, suitable for a family of six members. It was reconstructed on the same spot during the Flavian era. A large dump of pottery in the enclosing ditch near the house and granary contained a set of high-quality wheel-thrown Belgic beakers of Marne origin, a fire-steel and fragments of white and red wall plaster, which is unknown throughout the northern region. The 2nd century is documented by one or possibly two houses with a byre section, which seems to indicate that cattle-raising or the collection of dung for arable use had now gained in importance. Granaries were not found for this period. The houses had a roof of slate imported from Luxembourg, which is the only known example in Belgium so far of such a roof covering. The site was abandoned before the end of the 2nd century. Another interesting farm was excavated at Bruges (fig. 4). It was a newly created site from Early Roman times, which became an enclosed multi-compound farm during the Flavian and Antonine periods.29 By the beginning of the 3rd century only one or two compounds seem to have survived. During the middle of the 2nd century its plan was very similar to that of stone-built villa complexes known from the fertile loess regions of central Belgium and Northern France. However, despite the plan and the fact that all constructions were still built in timber, it is again the social construction of settlement space that is most conspicuous. On the spot where in villa complexes we would expect to find the owner’s house and hence an expression of status, we encounter only an ordinary timber-framed building, which does not differ from the other houses on the site in terms of architectural elements or finds. However, if we consider the spatial arrangement of granaries, these outbuildings appear remarkably clustered around the house, again pointing to a spatial mise-en-scène showing a successful harvest linked with symbolic capital. A significant feature is the appearance of deepened byre sections on the farm from AD 150 onwards. The complexes of Aalter and Bruges show a transformation from a Late Iron Age or Early Roman complex to a Roman-period farm. In the case of Aalter, an existing site evolved into a farm inhabited by one family, a situation which remained so until the late 2nd century. In the case of Bruges, an Early Roman compound farm evolved into a multi-compound complex in which several families seem to have lived together during the 2nd century, albeit in different compounds and houses. This suggests different social systems at work at the local level. However, identity is expressed in both cases not only in the consumption and deposition of specific kinds of material culture such as Belgic beakers or wall-plaster, but also in the spatial layout of the farm buildings and the position of the granaries in particular. Both sites therefore seem to indicate the importance of and the attempts to increase agricultural production within this area. They also point to the link this generated with social status and increasing social hierarchy at the local level. In the case of Bruges, an intra-site social hierarchy is even apparent. The efforts 28 

Courbot-Dewert 2008.

29 

Hollevoet/Hillewaert 1997/1998 (2002).



0

50 m

houseplan

ditch

well

cremation burial

granary

Fig. 4. The farm at Bruges-Refuge. After Hollevoet/Hillewaert 1997/1998.



Fig. 5. The small rural village of Evergem-Kluizendok (2nd/3rd century AD). After Laloo et al. 2009.

to increase agricultural production must have had their impact on the landscape. Several pollen diagrams show modest percentages of cereals and the appearance of heath land, which is a landscape type suitable for extensive grazing, and typical of sandy soils which have been reclaimed but quickly abandoned for agricultural purposes due to their poor quality. Furthermore, the density of occupation observed for the 2nd century (fig. 2) is many times higher than it was in the very Late Iron Age, also indicating the intensive reclamation of the landscape, even in the poorest parts. This was particularly felt in the Kluizendok project in Ghent Harbour, the largest archaeological operation in Belgium so far, consisting of an area of 170 ha in a low-lying sandy and very infertile area (fig. 5).30 The area was systematically trenched and later about 16 ha was excavated. Totally unexpectedly, a native-Roman settlement landscape was revealed, consisting of a cluster of several 2nd century contemporaneous compounds, most of which had already disappeared before the early 3rd century. The small village lay close to a small dirt road and was characterised by simple farmsteads consisting of timber-framed houses with a deepened byre section in some cases, a well, some small outbuildings and an enclosure surrounding the complex (fig. 5). The environmental data found in the 21 wells indicates that the settlement was located in a dense oak forest. Apart from some isolated finds and structures, no older

30 

Laloo et al. 2009.



or younger settlement traces of importance were registered, pointing to the fact that this landscape was not put into cultivation until the early 2nd century, which is late compared to the general pattern (fig. 2). We may therefore assume that the colonisation of this landscape by a small village community only began once better soils elsewhere had been taken. In the early 3rd century only two farms survived and the forest quickly regenerated to once again cover the entire landscape.

house - b uildi n g traditio n s i n the p leistoce n e area The Roman-period farmhouses in the north-western part of the civitas were constructed entirely of timber, and incidentally covered with a ceramic or slate roof. Houses sheltered not only families but in most cases also some of the animals under the same roof. After a long-standing prehistoric tradition of twoaisled timber-framed house building, architecture on the sandy soils developed during the 1st century AD onwards from the old two-aisled construction to a one-aisled design. The later Iron Age and Early Roman house of the Alphen-Ekeren type with its heavy ridge posts was abandoned during the Flavian era for a one-aisled construction with a cruciform layout of the posts: the weight of the roof gradually shifted from the central series of ridge posts towards the heavy posts on the sides which supported the trusses. This type is by far the most common house type found in the region (fig. 6). Apart from a logical search for more space inside houses, this rapid architectural evolution also reflects an economic and socioeconomic evolution compared to older periods. The more equal distribution of the roof weight, for instance, allowed for the construction of an attic and thus greater storage capacity or housing possibilities. From the middle of the 2nd century onwards to the beginning of the 3rd century, house architecture evolved towards a more equal distribution of the roof weight within one-aisled constructions. This involved a transformation from the cruciform plan of ridge- and truss-bearing posts to the use of more and equally spaced posts. More importantly, this further morphological evolution in house plans can be linked to the appearance of deepened byre sections, which have been revealed at several farms spread over the sandy landscape of northern Flanders. They appear in the youngest evolutions of the cruciform plan but dominate in successive types (15 out of a total 20 known examples in the region), which links them clearly to a specific architectural design. On average, houses also became larger over time (from an average of 90 to 130 m² from c. AD 10 to AD 250 in a total of about 100 houses). When considered in detail, however, this trend appears to be determined by the appearance of deepened byre sections from 150 AD onwards and not by other house types. While the younger houses contained stables with an average area of 30 m², the actual living areas continued to be small over time (c. 60-80 m²), indicating that the size of the households remained more or less the same, not reaching beyond six people. For the medieval and post-medieval periods, deepened byre sections (or potstallen) were related to attempts to increase animal and agricultural production and to collect the dung to fertilise the poor sandy soils. Also for the Roman period they point to new agricultural mechanisms at work, such as the overwintering and breeding of larger numbers of animals, to the breeding of other species and/or to the deliberate accumulation of dung to manure the fields in order to increase crop yields. In any event, it seems to have been related to some kind of intensification, probably due to population growth31 and/ or taxation. While the architectural transformation processes at work clearly point to rapid economic development, the social structuring of the house continued to have its prehistoric three-fold functional structure

31 

Willems 1984, 246; Bloemers 1978, 103 ff; Roymans 1996, 71 ff.



0

20 m

Fig. 6. Typical house plans from the late 1st and 2nd century AD in the northern half of the Menapian civitas.



with parts for living, stabling and work. Building or demolishing a house was a process that was accompanied by offerings. Hoards are predominantly found in the postholes of the heavy roof posts supporting the ridge or in the pits where the trusses near the entrance were positioned. Offerings sometimes consisted of complete or deliberately broken pottery vessels and frequently of whetstones, probably pointing to the strength of the posts or to a good harvest. In both cases they make hopeful references to the continuity of the house and family.32

f u n erar y p ractices i n the p leistoce n e area Apart from large cemeteries situated near central places (vici) such as Oudenburg33 or WaasmunsterPontrave,34 most of the cemeteries known at present seem to have been characterised by a small number of graves (about 12-20). In several cases one or two small rectangular enclosures with an entrance to the south occupy a central position in the cemetery. They contain one or two burials with material culture that often differs from the surrounding burials in quantity, quality (e.g. the presence of iron razor blades), and type (Brandgrubengräber). It therefore seems that these cemeteries should be seen as family or local group cemeteries where only certain people, perhaps founders or local chiefs, could be buried in an enclosed space. Their social status at the local level was obvious and expressed not only in terms of the enclosure and the grave goods, but also the grouping of common burials surrounding them. The use of enclosures and the grave type were embedded in local Iron Age traditions. In central places such as Waasmunster-Pontrave, non-local funerary traditions were revealed in the discovery of the remains of two monumental grave pillars. As centres of economic transaction, these villages also acted as places for the exchange of ideas and it appears that people present in the vicus expressed certain aspects of identity through means and materials foreign to the local rural environment. This probably suggests the presence of intermediate groups or resident aliens35 using Roman-style material vocabulary to express their identity.

settleme n t t y p es f rom the coastal area Important new insights have been gained from settlement on the coastal plain in recent years.They reveal a differentiated spectrum of occupation types. A first category consists of Flachsiedlungen, established on the dried-out peat. The second consists of artificially elevated platforms. The first type of habitation was well documented at Borsele-Ellewoutsdijk, where nine spatially separated farms have been found, each consisting of a timber longhouse, some fences and smaller outbuildings, all built on dried-out peat soil.36 The site was occupied between the middle of the 1st and the middle of the 2nd centuries AD. Study of the dung found in the stable part of the Borsele houses revealed that the animals were sheep or goats. The botanical remains found in the excreta prove that the animals grazed on lower salt-marshes with a strong marine influence. Archaeozoological and botanical research also indicated the presence of a mixed agricultural regime with the cultivation of barley, which tolerates salty soils well, although the economic emphasis was clearly on pastoral activities, probably resulting in the production of dairy products and salted meat. Another type of habitation was revealed only recently. A small, artificially raised platform was found at Serooskerke on Walcheren, indicating human presence possibly in a temporary context.37 The small 32

De Clercq 2009, 334-338.

35 

Mattingly 2006, 525.

33 

36

34 

37 

Hollevoet 1993. Van Hove 1996.

Sier 2003. 

De Clercq/Van Dierendonck 2009.



mound was built from peat and clay. On the platform there were remains of a hearth and several layers of waste consisting of large quantities of shells and some pottery, the whole dating to around 200 AD. Interestingly, the small platform was attached to a large dike, 7 m wide at its base, which could be traced over more than 90 m. If we return to the Flemish coast, we encounter a similar elevated structure that has been partially excavated at Steene near Ostend. Again a small artificially raised platform was discovered, linked to a dike which seems to have been circular or curved. The platform itself was built from clay and peat. Soil features indicate the presence of some kind of construction, while layers containing late Flavian and early 2nd century pottery, many shells and sheep bones were found on the slopes. Another large dike was found nearby at Raversijde. 38 It is unclear for the moment to what extent we can speak of an organised, large-scale attempt to reclaim land from the sea during Roman times. There is no doubt, however, that the construction of large dikes required considerable investment in time and manpower and the construction of these earthworks over long distances had to be coordinated by a regional or local authority. The discovery of the platforms opens up a new perspective on research in this area and points to the fact that these dynamic landscapes were suitable for habitation, albeit perhaps on a temporary basis. The find at Zeebrugge of natural sandy platforms (donken), which also attracted occupation, seems to be another indication of temporary human presence in the marine area, in this case in relation to the production of salt (see below).

house - b uildi n g traditio n s i n the coastal area In the coastal area, two house-building traditions were observed. Five houses were two-aisled, as one would expect, following the Iron Age and Roman period traditions further east and south on the Pleistocene land. Four houses, however, were partially three-aisled. The best preserved house of this type was large, with a length up to 25 m. The roof was supported by an A-frame construction. This method of building houses is unknown in the adjacent Pleistocene territories in the south and east, but it is well known from the Early Iron Age onwards in the coastal regions further north, such as near the Meuse estuary and even further north in Germanic territory. Two other three-aisled houses were found in the region, one at Colijnsplaat close to the Oosterschelde mouth, dating from the Claudian period, and another further north-east at Poortvliet dating from the Flavian period.39 The material culture found at Colijnsplaat points to a strong northern, Germanic influence on the site.

4

a model f or the socio - eco n omic de v elo p me n t o f a ‘ margi n al ’ regio n

a p roducti v e la n dsca p e o f f arms Based on the settlement structure, comparisons with other regions and to a lesser extent palaeo-environmental data (which is heavily influenced by the poor preservation conditions), we can develop a model for the socio-economic development of the north Menapian region. The underlying assumption is that the rise in numbers of people unable to produce their own food (the military, city dwellers) and the need to pay taxes to the Roman state (see above) led to the creation of different kinds of surplus. We

38 

Demey/Vanhoutte/Pieters 2010.

39 

Excavations J. van den Berg, AWN; De Clercq/Dieren-



donck 2009.

can assume that in an agricultural society this surplus was primarily of an agrarian nature and that taxes were paid directly in kind to the state or army, or that commodities resulting from diverse agricultural activities were converted into money at markets. It is evident that, unlike the loess soils further east in Nervian territory, the acid, sandy soils did not permit large-scale surplus production of cereals. This is also reflected in settlement structure, in which a mixed agricultural regime appears with crop and cereal production (as echoed in the small four- or nine-post granaries) and animal husbandry (as reflected in the deepened byre sections). Larger granaries such as horrea are absent. Often the single granary occupied an important position in the spatial layout of the complex, pointing to the importance of the symbolic capital cereals possessed for these communities (see above). As far as the storage capacity of the granaries is concerned, almost all households were able to produce sufficient cereals to sustain themselves in both consumption and seed reserves. Based on the study of these structures, a real surplus was not reached or it was extracted from the farm directly after the harvest. However, if we consider palynological or botanical data, ‘native’ species of cereals with a low turnover such as barley continued to be produced. Dutch research based on largescale palaeo-environmental and archaeozoological data originating from farms in the Dutch river area clearly confirms that farms situated on the sand or clay did generate surplus. They did so not primarily in cereals but by animal husbandry, which gained in importance over time; for the Batavian area this was clearly expressed in horse breeding.40 The sandy north Menapian region probably fits into this model. While animal bone is absent due to the acidity of the soil, there are several other indications such as deepened byre sections, enclosures and palaeo-environmental data which point to the importance of animal husbandry, a production which could result in meat, dairy and textile products. Some texts mention the presence of large herds of half-wild swine in the northern parts of Gaul, while Menapian ham seems to confirm the importance of swine herding, as was still the case in the Middle Ages for the region north of Ghent.41 Sheep, goats and cattle will also have played an important part. They could have been kept in the already mentioned deepened byre sections, which can also be linked to economic intensification (see above).

the me n a p ia n salt co n n ectio n w ith rome An important point of interest is the socio-economic development of the coastal area.42 Although settlement structure and material culture suggest at first glance that the region and its local communities were only marginally integrated in economic terms into the Roman Empire and attracted only marginal interest from Rome, I will argue that precisely the opposite was the case: the region attracted the full interest of the Roman state and its agent in the field, the army. The key factor was the production of salt, which is attested at many places such as Zeebrugge, Dudzele, Leffinge, Middelburg and Koudekerke.43 In addition, salt production is borne out by the find of different types of ceramic containers at inland sites such as Evergem-Kluizendok near Ghent (fig. 7). Furthermore, related to the production of salt, there is increasing evidence of fish sauce production in the Oosterschelde estuary and the large-scale harvesting

40 

Groot/Heeren/Kooistra/Vos 2009; Vossen/Groot 2009.

43 

This Flavian extraction site contained an unusual large

41 

number of colour-coated imported pottery from the

42 

A more detailed paper on the socio-economic devel-

Cologne workshops compared to contemporaneous sites

opment of the coastal plain will be published in the

(more than 20% of the total MNI, while less than 2% is

proceedings of the ‘Landscapes or Seascapes’ conference,

normal), indicating a strong connection with the Rhine-

held in Ghent in June 2010 and organised by the Dept.

land and the Cologne area.

Thoen/Soens s.a..

of Medieval History of Ghent University.



Fig. 7. Salt container with briquettage remnant, found at Evergem-Kluizendok. After Laloo et al. 2009.

of shells such as mussels and cardium edule for both human consumption and the production of lime. All these elements point to an intensive exploitation of the coastal area. It appears however that the transaction of this commodity did not lead to the accumulation of financial or symbolic capital. Either the status achieved through salt transactions was expressed in another less visible way, or the capital generated was drawn off by people or institutions outside the region. Theoretically, this could be to middlemen or to specialised traders, to civil servants at the civitas level or to the army. In previous debates on the Belgian coastal plain it was often forgotten that the exploitation of salt as one of the key mineral resources in the empire was an imperial prerogative. This has important consequences for our ideas about the transformation of the coastal wetlands as it turns this landscape or seascape into a specific taskscape. The Roman empire had different opportunities for using its right to this resource. It could collect the salt directly or indirectly by and for the army, which used it as part of salaries or to preserve and supply food. Another possibility is that the empire had less direct control but took a part of the profits by imposing a tax on commerce in salt or by renting salt exploitation to private contractors. A combination of these possibilities also seems plausible. Interesting parallels can be drawn in this respect with British regions such as the Fenlands, the Somerset Levels or Dorset, where similar patterns of habitation, reclamation and extraction were revealed. In these regions, the Roman state seems to have had a serious interest in salt. However, as in the Menapian coastal area, this does not appear to have led to an accumulation of local wealth. Rippon therefore suggests that the salt-marshes of the Som-



erset levels were an imperial estate from which salt was extracted.44 Fincham interprets the systematic reclamation of land in the Fenlands in the context of developing a taskscape aimed at the production of salted meat for the Roman army.45 Gerrard recently suggested that the famous Dorset BB1 industry was an army-controlled supply line in salted products, a line which stretched far in distance and time.46 In all three cases an important intervention by the Roman state is suggested. If we apply the British situation to the Menapian coastal area, we may also presume an important role for the Roman army in the region, not only because of the arguments outlined above, but also because of contemporary epigraphic information. The close relationship between the Menapian civitas and the Roman army became well established during the Flavian period. When other northern tribes resisted Roman authority in the wake of Julius Civilis, the Menapii and the Morini remained silent and chose the side of Rome. Moreover, in the wellknown Rimini inscriptions, the salinatores of the civitates Menapiorum and Morinorum honoured Lucius Lepidius Proculus, centurion of the 6th Legion Victrix stationed at Novaesium on the Rhine and one of the three legions that defeated Civilis in 70 AD. The inscriptions therefore point to a direct relationship between the Menapian coastal area and this military region, passing through salt transactions, which may have been used as a form of taxation in kind to fulfil the duty of civitas stipendiaria. Aarts has already argued how the taxes raised in Gallia Belgica and Germania Inferior were spent in these regions,47 predominantly by the army which could register these transfers in pactiones. In this context we can easily assume that salt was transferred as a form of tax payment to the army or the army was involved in controlling the supply of a basic mineral. As for the term salinator itself, debate has focused on the functions of salt workers, salt contractors or salt merchants. Hocquet has clearly defined the different functions and nomenclature in the salt industry, pointing out that salinatores were neither workers nor contractors exploiting the salt resources (these were the conductores or salarii), nor merchants (who were the negotatiores), but a group of servants operating at the level of the civitas who controlled salt production and trade and who transferred part of the profits to the Roman treasury.48 They were also responsible for the continuum in salt supply to the salarium, the city’s storehouse where the salt was kept. The find at Tongres of an altar erected by Catius Drousus, a Menapian sal(inator?), and dedicated to Jupiter and the genius of the Municipium Tungrorum could confirm this interpretation.49 Despite claims that altar inscriptions at the temples of Colijnsplaat and Domburg50 reflected the presence of merchants passing through the region and crossing the Channel, it can equally be argued that these individuals were also interested in the commodities originating from the region itself. After all, merchants in salt and allec constitute the largest group among all the professions mentioned on the more than 300 altars. These people again originated from the Rhineland and Trier. The installation of the army itself in the region on the border with the Pleistocene sand during the later 2nd and the 3rd century not only met strategic goals and the need for security, but also gave the army an opportunity for direct control of economic activities in the coastal wetlands. Sheep and especially salt no doubt sum up the fundamentals of socio-economic development in the coastal zone under an umbrella of security and control by the presence of the Roman army on the border of the coastal plain.

44 

49 

45 

50 

Rippon 2000, 69. Fincham 1999; idem 2002, 84-89.

46 

Gerrard 2008.

Vanvinkenroye 1994; Nouwen 1997. Stuart/Bogaers 2001; De Clercq/Van Dierendonck 2009, 54.

47 

Aarts 2000, 12.

48 

Hocquet 1994.



5

p easa n ts f armi n g f or rome ?

The aspects outlined above shed light on the socio-economic development of Menapian society and its integration in the wider context of the Roman empire. The case of the coastal wetland area is indeed a very specific one, and it probably attracted the full and direct interest of the Roman state. The situation is probably even more complex for the sandy soils. Both micro-regions, which were marginal from an agrarian and productive point of view, were intensively exploited during Roman times in accordance with the possibilities offered by the landscape. Marginal landscapes (even those considered as such in contemporary Roman sources) became specific taskscapes in local terms and were exploited on an unprecedented scale, one which would not be repeated before the High Middle Ages. Although written sources relating to social structure are absent for the Roman period, in many respects the situation must have been similar to the Middle Ages, for which a peasant society and a ‘commercial survivor’ economy is suggested for the region.51 In both periods a predominantly rural-based society had to function in a socio-political and economic system in which taxes had to be paid to an elite authority, whether it be Rome, noblemen or clergy. In addition, a growing section of the population (soldiers and city dwellers) was unable to produce its own food. Farmers had to produce surplus to meet these demands, thereby adding to the stress on the landscape. Marginalities could thus become opportunities. Salt production or the appearance of deepened byre sections illustrates how these landscapes became specific taskscapes and landscapes of opportunity. However, this process still did not lead to full integration into the Roman economy and its monetary system. Van Heesch has demonstrated the limited use of coins in the civitas compared to neighbouring areas.52 Coins are totally absent at most sites and their use seems to have been confined to the military and to very few central places. We could therefore suggest that in economic terms local communities continued to act on a partially autarkic basis, with locally produced commodities being exchanged by barter at local or regional markets for other commodities, which in turn were converted by middlemen into money, or which entered army supply lines or the wider economic network of the Roman empire. Rural production was intended first and foremost to ensure the basic survival of the nuclear or extended family, although to achieve this goal some part of the production had to be commercialised in one way or another, at least for taxation reasons. External demands such as taxes and commodities for non-local consumers not only stimulated and intensified internal production in the local communities, they probably also affected the social structure of society itself to some degree. Social structure itself is poorly understood at present, but some aspects of identity such as status at the local and/or family group level can be assessed in the material culture and spatial layout of farms and cemeteries (see above). Societies with a low degree of hierarchy mainly operating at the basic household and family level but which are also integrated (albeit partially) into a market system are labelled ‘peasantries’ by anthropologists.53 Labour on the farm is organised at the family level and, apart from the social subsistence of the family group, it primarily serves the physical subsistence of the family by generating food and a small surplus for taxes or exchange for other goods. Erdkamp described this system for the Roman period as ‘subsistence and a little bit more’.54 The development of ‘peasantries’ in the study region has been suggested for the Middle Ages, and was also proposed by Slofstra for the Meuse-Demer-Scheldt region in the Roman period.55 Although such a social structure is difficult to assess archaeologically, we may assume that this process was also at work in the study region.

51 

54 

52 

55 

Thoen 2004; Thoen/Soens s.a.. Van Heesch 1998.

53 

Ellis 1988, 12.



Erdkamp 2005, 96-100. Thoen 2004; Slofstra 1983.

6

material culture a n d social ide n tit y i n a n o n - v illa la n dsca p e

The settlement landscape in the north-western sandy part of Gallia Belgica belongs to the ‘non-villa landscapes’. It is a landscape of less fertile soils where post-built farms clearly prevailed over the stone-built complexes which characterised the ‘villa landscapes’ on loess soils. The two types of settlement landscape represent fundamentally different economic systems as well as highly differentiated expressions of identity in material culture, including architecture, food types and funerary traditions. When contemporary sources attribute a productive marginality to the non-villa communities of the Menapii, they implicitly suggest a social marginality as ‘concealed communities’.56 Particularly in Belgian archaeology, with both types of landscape in close proximity to each other within a small research area, these marked differences in archaeological visibility combined with historical sources have produced up until the early 1990s a biased view of the past whereby this image of social marginality and seeming resistance to Roman cultural values was actively cultivated. Seen from an ecological point of view, marginal landscapes could become landscapes of economic opportunity in a new imperial context if used to meet the demands of levies or more directly for the exploitation of certain strategic resources, as was presumably the case for the Menapian coastal plain and its salt. In this case the marginal landscapes became landscapes not only of opportunity but of domination. Especially where salt is concerned, the sources reveal a strong connectivity between the region and Rome from the Flavian era onwards. This link is not expressed in local consumption patterns, however, and architectural styles merged with Roman-style traditions, which demonstrates in turn a high degree of disconnectivity and the persistence of local ways of expressing identity. It seems that the local Menapian coastal population barely profited from the riches generated by the salt, fish sauce and chalk. Elite material culture markers or strong clustering of imported material culture were confined to the very few and barely known central places on the border with the Pleistocene or to military strongholds. We suggested that the Roman state was directly involved in the coastal region by renting its exploitation or by taxing it in kind. This relationship could operate directly through the army or indirectly via markets and middlemen in the local centres, suggesting an important role in the region for a small number of intermediate groups of local or non-local origin, so-called ‘resident aliens’.57 The latter were more profoundly engaged in the Roman economy and trade and also expressed their identity through a non-local vocabulary. The clustering of imported material culture at sites such as Koudekerke in the coastal area or the funerary pillars of the central place of Waasmunster on the Pleistocene could well be an expression of identity of these few people. Unlike the central places, the material expression of status or cultural identity in native communities operated pre-eminently in accordance with the traditional set of culturally determined values embedded in the local habitus. In this context, Roman-style material culture such as a decorated samian vessel may have been adapted as an elite material culture marker, but in a local cultural context for a local ‘audience’ and without its related Mediterranean connotation of being part of a fixed set of tableware used for dining. It can be argued that status in these farming communities was achieved not, or not solely, by possessing and using Mediterranean-style material culture, but also by the number of animals a family owned, the number of wives a man had, the number and size of the granaries or the dunghill in the yard, or the offerings made during important social events such as funerals, house construction or house abandonment. We cannot determine whether these societies were putting themselves behind, but the way in which social discourses were materially translated was negotiated in another, locally defined way. Therefore we cannot simply assume that material culture from a non-villa landscape or, more broadly, 56 

Turner/Young 2007.

57 

Mattingly 2006, 525.



material culture from so-called marginal societies should automatically be linked with social marginality and geographical and socio-economic disconnectivity. Economy and social group structure in the local communities of the Menapian non-villa landscape probably remained embedded in pre-Roman social relations, being incompatible with a villa-based model of social stratification, dependent labour, market-oriented production and its related cultural framework of ways of expressing certain aspects of identity.58 Roymans has already pointed out that anthropologists label these kinds of peasant society ‘conservative’, meaning that the maintenance of regional social group traditions and identities generates only slow cultural changes since aspects of ‘Roman’ culture were translated from and into the native cultural-ideological background.59 Recently he has argued that the Batavi, like the Menapii, clung to ancestral traditions such as living in byre houses as well as regenerating old funerary traditions to define themselves within the new context of the Roman empire.60 Although many people from Batavian farming communities living in byre houses were deeply integrated into the culture of the Roman army, were literate and possessed Roman citizenship (the latter two are often regarded as expressions of successful integration into the Roman state and culture),61 they also deliberately and explicitly used a native vocabulary to express their social and cultural identity in the Roman world, pointing to the ambivalent way of expressing identity under Rome’s global umbrella.These expressions of ‘discrepant identities’ represent the wide range of responses to Rome’s impact on the native communities it conquered, varying from revolt and resistance to adaptation and integration.62 Woolf rightly points out that Roman imperial culture was indeed not homogeneous but constituted rather a ‘structured system of differences’ leading to ‘cultures of exclusion’ in different levels of society.63

7 co n clusio n The archaeological research of the last decade has profoundly enriched our picture of the landscapes of the northern part of the civitas Menapiorum, whether this be the Holocene coastal plain or the Pleistocene sandy hinterland. Both landscapes witnessed intensified exploitation and settlement activity along the pre-determined lines of the natural landscape. Survival strategies were at least partially – and often out of necessity – integrated into the Roman economy and its tax and consumer system, but the ways in which this happened were based on criteria defined by the landscape, local identities and the power of Rome. A probable consequence was the transformation of societies towards peasant societies – farming communities with a low social hierarchy, working in a predominantly autarkic society but also engaged in the creation and the extraction of surplus from their communities. Native traditions expressed in house building, pottery production and funerary traditions strongly persisted but evolved within their own local cultural backgrounds. It can be argued that the ‘marginal’ character of the Menapian communities – even when expressed explicitly in contemporary sources and suggested by material culture patterns – is a concept which needs to be carefully contextualised. Rather than being economically isolated and socially marginalised, these communities were economically well-articulated with Rome using the specific opportunities offered by the landscape they lived in. Ecological, economic or social marginality are therefore relative and scaledependent concepts and the negotiation of certain aspects of identity has to be understood in specific local contexts. Interpretations of the strong persistence of native traditions in local communities must involve not only economic or geographical parameters, but also social and cultural factors, thereby seeking a balance between native agency and Roman domination. 58 

61 

59 

Roymans 1996, 72-84; idem, 1999; Hingley 1997; Mat-

62

Mattingly 2006, 524-528.

tingly 2006.

63

Woolf 1997; idem 1998, 18-19.

Slofstra 1991.

60 

Roymans 2009.



Roymans/Derks/Heeren 2007.

re f ere n ces Aarts, J., 2000: Coins or money? Exploring the monetization and functions of Roman coinage in Belgic Gaul and Lower Germany 50BC-AD450, Amsterdam (unpublished PhD thesis VU University Amsterdam). Aarts, J., 2005: Coins, money and exchange in the Roman world. A cultural-economic perspective, Archaeological Dialogues 12, 1-28. Alföldy, G., 1968: Die Hilfstruppen in der römischen Provinz Germania Inferior, Düsseldorf (Epigraphische Studien 6). Bloemers, J.H.F., 1978: Rijswijk (Z.H), ‘De Bult’. Eine Siedlung der Cananefaten, Amersfoort (Nederlandse Oudheden 8). Bogaers, J.E., 1971: Germania Inferior, Gallia Belgica en de civitates van de Frisiaviones en de Tungri, Helinium 11, 228-237. Collart, J.L., 1996: La naissance de la villa en Picardie. La ferme gallo-romaine précoce, in D. Bayard/J.L. Collart (eds), De la ferme indigène à la villa romaine. La romanisation des campagnes de la Gaule. Actes du deuxième colloque de l’association AGER tenu à Amiens (Somme) du 23 au 25 septembre 1993, Amiens (Revue Archéologique de Picardie. Numéro Spécial 11), 121-156. Courbot-Deweirdt, C., 2008: Une mise en scène des activités agricoles au sein des villae gallo-romaines? La disposition des bâtiments des établissements ruraux de Gaule septentrionale (Ier siècle-IIIe siècle après J.-C.), in P. Madeline/J.-M. Moriceau (eds), Bâtir dans les campagnes. Les enjeux de la construction de la Protohistoire au XXIe siècle, Caen (Bibliothèque du pôle rural. Enquêtes rurales, Numéro Hors Série 1), 335-350. De Boe, G., 1988: De inheems-Romeinse houtbouw in de Antwerpse Kempen, in F. Brenders/G. Cuyt (eds), Van beschaving tot opgraving. 25 jaar archeologisch onderzoek rond Antwerpen, Brussel, 47-62. De Boe, G./F. Lauwers 1980: Een inheemse nederzetting uit de Romeinse tijd te Oelegem, Brussel (Archaeologia Belgica 228). De Clercq, W., 2009: Lokale gemeenschappen in het Imperium Romanum. Transformaties in de rurale bewoningsstructuur en de materiële cultuur in de landschappen van het noordelijk deel van de civitas Menapiorum (Provincie Gallia-Belgica, ca. 100 v. Chr. - 400 n. Chr.), Gent (unpublished PhD thesis, Ghent University). De Clercq, W./R.M.Van Dierendonck, 2010: Extrema Galliarum. Noordwest-Vlaanderen en Zeeland in het Imperium Romanum, VOBOV-info 64, 34-75. De Laet, S.J., 1961: Les limites des cités des Ménapiens et des Morins, Helinium 1, 20-34. Demey, D./S.Vanhoutte/M.Pieters, 2010: Een Romeinse dijk met woonplatform in Stene bij Oostende (Prov. W.-Vl.), in A. Bosman/W. De Clercq/J. Hoevenberg/F. Vilvorder (eds), Romeinendag. Journée d’Archéologie Romaine, Louvain-La-Neuve, 95-98. De Maeyer, R., 1937: De romeinsche villa’s in België. Een archeologische studie, Antwerpen/’s-Gravenhage. Ellis, F., 1988: Peasant economics. Farm households and agrarian development, Cambridge. Erdkamp, P., 2005: The grain market in the Roman Empire, Cambridge. Fincham, G., 1999: Poverty or power? The native response to Roman rule in the Fenland, in P. Baker/C. Forcey/S. Jundi/R. Witcher (eds), TRAC 98. Proceedings of the Eight Annual Theoretical Roman Archaeology Conference, Oxford, 46-51. Fincham, G., 2002: Landscapes of imperialism. Roman and native interaction in the East-Anglian Fenland, Oxford (BAR British Series 338). Galsterer, B./H. Galsterer, 2010: Die römischen Steininschriften aus Köln - IKöln2, Mainz (Kölner Forschungen 10). Gerrard, J., 2008: Feeding the army from Dorset. Pottery, salt and the Roman state, in S. Stalibrass/R.Thomas (eds), Feeding the Roman Army. The archaeology of production and supply in NW-Europe, Oxford, 116-127. Groenewoudt, B., 2009: An exhausted landscape. Medieval use of moors, mires and commons in the Eastern Netherlands, in J. Klápštĕ/P. Sommer (eds), Medieval rural settlement in marginal landscapes. Ruralia



VII 8th-14th September 2007 Cardiff, Wales, UK, Leuven, 149-180. Groot, M./S. Heeren /L. Kooistra/W.K. Vos 2009: Surplus production for the market? The agrarian economy in the non-villa landscapes of Germania Inferior, Journal of Roman Archaeology 22, 231-252. Hingley, R., 1997: Resistance and domination. Social change in Roman Britain, in D. Mattingly (ed.), Dialogues in Roman Imperialism. Power, discourse and discrepant experience in the Roman Empire, Portsmouth, Rhode Island (Journal of Roman Archaeology, supplementary series 23), 81-102. Hocquet, J.-C., 1994: Production et commerce du sel à l’âge du Fer et à l’époque romaine dans l’Europe du Nord-Ouest, Revue du Nord 76, 309, 9-20. Hollevoet,Y., 1993:Ver(r)assingen in een verkaveling. Romeins grafveld te Oudenburg (prov. West-Vlaanderen). Interimverslag, Archeologie in Vlaanderen 3, 207-216. Hollevoet,Y./B. Hillewaert, 1997/1998 (2002): Het archeologisch onderzoek achter de voormalige vrouwengevangenis Refuge te Sint-Andries/Brugge (prov. W. Vl). Nederzettingssporen uit de Romeinse tijd en de Middeleeuwen, Archeologie in Vlaanderen 6, 191-207. Laloo, P./W. De Clercq/Y. Perdaen/Ph. Crombé (eds), 2009: Het Kluizendokproject. Basisrapportage van het preventief archeologisch onderzoek op de wijk Zandeken (Kluizen, gem. Evergem, prov. Oost-Vlaanderen, december 2005-december 2009), Gent (Universiteit Gent, Ugent Archeologische Rapporten 20). Langohr, R., 2004: Maldegem-Vake. Selection of contributions from soil science, in F. Vermeulen/K. Sas/W. Dhaeze (eds), Archaeology in confrontation. Aspects of Roman military presence in the Northwest. Studies in honour of Prof. Em. Hugo Thoen, Gent (Archaeological Reports Ghent University 2), 181-189. Malrain, F./V. Matterne/P. Méniel, 2002: Les paysans gaulois (IIe siècle -52 av. J.-C.), Paris. Mattingly, D, 2006: An imperial possession. Britain in the Roman empire 54 BC - AD 409, London. Meiggs, R., 1982: Trees and timber in the ancient Mediterranean world, Oxford. Mertens, J./L.Van Impe 1971: Het Laat-Romeins grafveld van Oudenburg, Brussel (Archaeologia Belgica 135). Nevell, M., 1999: Iron Age and Romano-British rural settlement in North West England. Theory, marginality and settlement, in M. Nevell (ed.), Living on the edge of empire. Models, methodology & marginality, Manchester (Archaeology North West 3, Council for British Archaeology North West), 14-26. Nouwen, R, 1997:To I(uppiter) O(ptimus) M(aximus) and the Genius of the Mun(icipium) Tung(rorum). Atuatuca, the first known municipium of Gallia Belgica?, in XI Congresso Internazionale di Epigrafia Greca e Latina, Roma, 417-426. Nouwen, R., 2006: De Romeinen in België (31 v.C. - 476 n. C.), Leuven. Nouwen, R., 2008: Het Romeinse leger en het romaniseringsproces in de Lage Landen, Pax romana. De Romeinse vrede in de Lage Landen, Kunsttijdschrift Vlaanderen 320, 3-11. Purcell, N., 1995:The Roman villa and the landscape of production, in T.J. Cornell/K. Lomas (eds), Urban society in Roman Italy, London, 151-179. Quérel, P., 2007: Apports de l’archéologie préventive à la connaissance du terroir antique du Mélantois (France), in A. Bosman/M.-H. Corbiau/W. De Clercq/J. Hoevenberg (eds), Romeinendag. Journée d’Archéologie Romaine 2007, Namen, 53-58. Roymans, N., 1996: The sword or the plough. Regional dynamics in the romanisation of Belgic Gaul and the Rhineland area, in N. Roymans (ed.), From the sword to the plough. Three studies on the earliest romanisation of Northern Gaul, Amsterdam (Amsterdam Archaeological Studies 1), 9-126 Roymans, N., 1999: Man, cattle and the supernatural in the Northwest European plain, in Ch. Fabech/J. Ringtved (eds), Settlement and Landscape. Proceedings of a conference in Arhus, Denmark, 1998, Copenhagen (Jutland Archaeological Society), 291-300. Roymans, N., 2009: Die Bataver. Zur Entstehung eines Soldatenvolkes, in St. Berke et al., 2000 Jahre Varusschlacht. Mythos, Stuttgart, 85-98. Roymans, N./T. Derks/S. Heeren (eds), 2007: Een Bataafse gemeenschap in de wereld van het Romeinse rijk. Opgravingen Tiel-Passewaaij, Utrecht.



Rippon, S., 2000: The transformation of coastal wetlands. Exploitation and management of marshland landscapes in North West Europe during the Roman and Medieval Periods, Oxford. Sier, M., (ed.) 2003: Ellewoutsdijk in de Romeinse tijd, Bunschoten (ADC-Rapport 200). Slofstra, J., 1983: An anthropological approach to the study of Romanization processes, in R.W. Brandt/J. Slofstra (eds), Roman and native in the Low Countries. Spheres of interaction, Oxford (BAR Int. Ser. 184), 71-104. Slofstra, J., 1991: Changing settlement systems in the Meuse-Demer-Scheldt area during the EarlyRoman period, in N. Roymans/F. Theuws (eds), Images of the Past. Studies on ancient societies in Northwestern Europe, Amsterdam (Studies in Prae- en Protohistorie 7), 131-200. Slofstra, J., 2002: Batavians and Romans on the Lower Rhine. The Romanisation of a frontier area, Archaeological Dialogues 9, 16-38. Stuart, P./J.E. Bogaers, 2001: Nehalennia. Römische Steindenkmäler aus der Oosterschelde bei Colijnsplaat, I. Textband, II. Tafelband, Leiden (Collections of the National Museum of Antiquities at Leiden). Thoen, H., 1967: De Gallo-Romeinse nederzetting van Waasmunster-Pontrave, Brussel (Oudheidkundige Repertoria. Reeks B: De Verzamelingen III). Thoen, H., 1978: De Belgische Kustvlakte in de Romeinse tijd. Bijdrage tot de studie van de landelijke bewoningsgeschiedenis, Brussel (Verhandelingen van de Koninklijke Academie voor Wetenschappen, Letteren en Schone Kunsten van België. Klasse der Letteren Xl, 88). Thoen, H., 1981: The third century Roman occupation in Belgium. The evidence of the coastal plain, in A. King/M. Henig (eds), The Roman West in the third century. Contributions from archaeology and history, Oxford (BAR - International Series 109(ii)), 245-257. Thoen, E., 2004: Social agrosystems as an economic concept to explain regional differences. An essay taking the former county of Flanders as an example (Middle-Ages-19th century), in B. van Bavel/P. Hoppenbrouwers (eds), Landholding and land transfer in the North Sea area (late Middle ages-19th century). Turnhout (Corn Publication Series. Comparative Rural History of the North Sea Area), 47-65. Thoen, E./T. Soens, s.a.: Elevage, prés et paturage dans le comté de Flandre au Moyen Age et au début des Temps Modernes. Les liens avec l’économie rurale régionale, Gent. Turner, S./R. Young, 2007: Concealed communities. The people at the margins, International Journal of Historical Archaeology 11, 297-303. Van Heesch, J., 1998: De muntcirculatie tijdens de Romeinse tijd in het noordwesten van Gallia Belgica. De civitates van de Nerviërs en de Menapiërs (ca. 50 v.C. - 450 n.C.), Brussel (Koninklijke Musea voor Kunst en Geschiedenis. Monografie van Nationale Archeologie 11). Vanvinkenroye, W., 1994: Een Romeins votiefaltaar te Tongeren, Limburg, Het oude Land van Loon 73, 225-238. Vermeulen, F., 1992: Moderate acculturation in the fringe area of the Roman Empire. Some archaeological indications from the Civitas Menapiorum, Bulletin van het Belgisch Historisch Instituut te Rome 72, 5-41. Vermeulen, F., 2001: Cultural biography of the protohistoric and Roman landscape of the Civitas Menapiorum, in F. Vermeulen/ M. Antrop (eds), Ancient lines in the landscape. A geo-archaeological study of protohistoric and Roman roads and field systems in Northwestern Gaul, Leuven (Babesch, supplement 7), 17-23. Vossen, I./M. Groot, 2009: Barley and horses. Surplus and demand in the civitas Batavorum, in M. Driessen/S. Heeren/J. Hendriks/F. Kemmers/R.M. Visser (eds), TRAC 2008. Proceedings of the Eighteenth Annual Theoretical Roman Archaeology Conference, Oxford, 85-100. Willems, J., 1983: Romans and Batavians. Regional developments at the imperial frontier, in R. Brandt/J. Slofstra (eds), Roman and native in the Low Countries. Spheres of interaction, Oxford (BAR International Series, 184), 105-128. Woolf, G., 1997: Beyond Romans and natives, World archaeology 28, 339-350. Woolf, G., 1998: Becoming Roman. The origins of provincial civilization in Gaul, Cambridge.



Evaluating settlement patterns and settlement densities in the villa landscapes between Tongres and Cologne Karen Jeneson1 1 Introduction 2 Reconstructing the settlement landscape 3 Evaluating settlement densities 4 Archaeological practices – past and present 5 Conclusions References

1 i n troductio n This article focuses on the Roman settlement landscapes of the fertile loess soils between current-day Tongres (Belgium) and Cologne (Germany). The commonly presented view regarding these landscapes is that in the course of the later 1st century AD they became almost completely dominated by villa-type settlements, characterized by one or more stone-built structures.2 In the rare instance that a different type of rural settlement is suggested, it is not quantified, which means that there are at present no estimations of the proportion of villas to other types of rural settlement for this region. With regard to settlement density, an average density of one villa per square kilometre seems to be the general consensus for the overall area.3 This article aims to challenge these perceptions of the composition of the rural settlement landscape and of habitation density. Based on an extensive inventory, and combining archaeological information with spatial dimensions, a new dataset for these landscapes has been compiled which demonstrates both higher settlement densities and a much higher proportion of ‘post-built’ farms than is commonly thought. This dataset is the result of a landscape-archaeological study that formed part of the research project ‘Roman villa landscapes in the north. Economy, culture, lifestyles.’4 The aim of this study was to reconstruct 1 

This paper is based on the results of the PhD study

Hambach area; the latter number is taken from Gaitzsch

‘Exploring the Roman villa world between Tongres and

1986. Wendt and Zimmermann 2008, 7, cite Willems

Cologne. A Landscape-archaeological approach’, which

1988 and Gaitzsch 2002 for their calculations of land per

will be defended in 2012 (Jeneson, in prep.). This project

settlement. Nonetheless, the possibility of higher settle-

is funded by the Netherlands Organisation for Scientific

ment densities are alluded to by various researchers; see

Research (NWO).

for example De Groot 2006, 4-5, who acknowledges

2 

For the problem of defining the term villa, see Roymans/Derks, this volume, section 1.

3 

For example, a density of one settlement per 2 km2 is

that the current dataset for Roman South Limburg is not representative of the original distribution. 4 

A research project led by Nico Roymans and Ton Derks

used for the Heerlen Basin area in Kooistra 1996, and

of the Archaeological Centre of VU University Amster-

a density of one settlement per 0,5 km2 is used for the

dam, and funded by the NWO.



G

E R M

A N

Rh e

I

in

A

COLOGNE TONGRES

EI

FE

L K

S A R D E N N E

Ü

C

v

l a i l

e s a p c s n d l a

H

M

S UN

R

eu

O

S

G

E S

se

V

0

100 km

Fig. 1. Location of the study area. D

E

F

and analyse the Roman landscapes on the loess soils between the Meuse and Rhine, incorporating the results of more than 150 years of archaeological activity. Although both archaeological inventories and synthesizing work have been done for the Roman landscapes in each of the three countries in the study area,5 reconstructions of the settlement landscape across national borders have been lacking. An all-encompassing inventory of Roman sites in the study area was therefore carried out, including the process of data categorization, to ensure a homogeneous, reliable result. A GIS was used for the registration, mapping and analysis of data. The results of the inventory, the subsequent reconstruction of the settlement landscape, and the settlement density map will be presented and analysed in this paper. I will put forward the argument that particular archaeological practices, rather than any other factor, are responsible for serious biases in current views on the composition and average settlement density of Roman rural landscapes in the northern provinces, and that it is possible, based on the results of recent investigations, to reconstruct different scenarios. The region under scrutiny lies within an extensive loess belt, stretching from the north of France, via central Belgium and Dutch South Limburg, to the Rhine in Germany. The study area measures approxi-

5

See for example Künow 1994, Gechter/Künow 1986,

the Hambach microregion, see various publications by

Lenz 1999, and Heimberg 2003 for the German region;

Gaitzsch, for example 1986 and 1988, and his article in

De Maeyer 1940, Mertens 1964, De Boe 1973, and

this volume.

Vanderhoeven 2002 for the Belgian region; Byvanck 1947, and De Groot 2006 for the Dutch region. For



Fig. 2. Civitas centres and small towns within the study area around 100 AD.

mately 25 by 125 kms. Located just west of the Rhine, the land here is intersected by the Meuse river and many smaller rivers, including the Ruhr, Erft and Geul, and numerous small brooks and streams. Because of its fertile soils, good water situation, temperate climate and slight elevation (from only a few metres above sea level to just over 300 m), this region has been favoured by farmers ever since the Neolithic. After its conquest by the Romans, it was rapidly developed along the standard Roman model, with the foundation of several major towns to serve as centres for political administration, as well as the construction of main roads, connecting it to the rest of the empire. Figure 2 shows the situation at the end of the 1st century AD. Another important trend here was undoubtedly the development of large surplus-producing farms, or villas, that are archaeologically recognizable by the presence of one or more buildings made of stone and ceramic building material, such as roof tiles. Although we observe a considerable variation in the spatial layout of villa sites,6 they mostly consist of a rectangular enclosed compound containing all the buildings.

2 reco n structi n g the settleme n t la n dsca p e Although inventorizing the Roman sites proved to be a straightforward, albeit rather time-consuming task, reconstructing the settlement landscape was more problematic. As the basic dataset comprises the results of more than 150 years of archaeological activity, performed by a wide range of actors, the interpretation of the find material was not always reliable. For example, because many early investigators were keen to discover villas, they gave this label to virtually any site where roof tile fragments and Roman pottery were found. I therefore decided to perform a data characterization, which entailed assessing the material from each individual site, using nine defined categories of find material, together with guidelines set up specifically for the region. The aim of this process was to produce homogeneous criteria for the classification and interpretation of the archaeological evidence. The result is a dataset of just over 3000 sites, 2333 of which could be attributed to one of five main categories defined in this study (settlement,

6 

See for example the contributions of Roymans/Derks and Habermehl in this volume.



Fig. 3a. The reconstructed Roman settlement landscape with the distribution of stone-built rural settlements.

Fig. 3b.The reconstructed Roman settlement landscape with the distribution of post-built and ‘type-undefined’ rural settlements.

burial, road network, specialist craft and specific features). The remaining sites which did not provide sufficient information to allow for a reliable characterization were recorded as ‘unidentified Roman evidence’. A conscious decision was made in this first phase not to label settlement sites as ‘villa’ or ‘farm’, but to use more neutral terms: a. Settlement – stone-built: settlement with at least one building made of stone, ceramic building material, concrete and plaster b. Settlement – post-built: settlement consisting only of buildings made of perishable materials, particularly wood and thatch.



Doing so enabled me to circumvent the discussion regarding the definition of a villa,7 and ensured that all settlement evidence was characterized along the same guidelines, as a specific set of find material was defined for each of the two types of settlement. Interestingly, it turned out that of the 1301 settlement sites, 1247 sites (or 96%) were of the ‘stone-built’ type, and only 41 (or 3%) were of the post-built type. Of these 41, the majority are dated to the Early Roman period, and were replaced by stone-built settlements within less than a century. It seems therefore that the landscape in this area truly was a ‘villa-dominated landscape’. However, it became clear that the reconstruction of the settlement landscape required further work. Combining archaeological details with spatial information, the new settlement dataset was reviewed, using the average dimensions of a typical rural complex. Furthermore, a methodology was designed to allow for the re-interpretation of non-settlement sites, taking into account burial evidence, specific site features and even unidentified sites. The basis for both re-interpretations was provided by modern large-scale excavations of Roman rural settlements in our study region. As mentioned earlier, all villas have a distinct built area, separated from the fields and meadows by ditches. When excavated completely, these built areas can be measured and used to establish the average dimensions of a Roman rural settlement. Armed with this knowledge, an informed decision can be made as to whether two settlement sites, found in close proximity to each other, should be interpreted as a single settlement or as two separate settlements.8 In a GIS environment the precise distance between the two sites can be measured precisely. As many villa settlements consist of both stone-built and post-built structures, this was a particularly important step in the reconstruction process, as it allowed me to decide whether a site with a post-built structure formed part of a villa settlement, or whether it constituted a separate farm, without any stone-built structures. This yielded a more reliable reconstruction of the proportion of stone-built and post-built farms. The result of this process was that of the 41 post-built settlements in the first dataset, only 26 could be said to represent a separate settlement of the post-built type. With 1145 rural settlements of the stone-built type,9 it seems that the domination of villas was almost complete (98% stone-built versus 2% post-built). Non-settlement sites were also included in the reconstruction of the settlement landscape, using an additional characteristic of Roman rural settlements in the region. For example, burials are nearly always found on the borders of the built area. We can therefore assume that burials in rural contexts are indicators of settlements, and that burials found with no apparent settlement evidence nearby can be regarded as representing a settlement. Here too working in a GIS environment allowed me to apply the spatial information on the average dimensions of a rural settlement to the burial evidence, so that I could make an informed decision as to whether a burial belonged to a settlement site nearby or represented a separate, as yet undiscovered settlement. In total, 344 burial sites were interpreted as indicators of settlements. The majority of these consist of ‘regular’ burials, i.e. without any ostentatious form of display. As it was not possible to determine the type of settlement in these cases, they were recorded as ‘settlement-undefined’. Special burial sites, such as monumental tumuli and stone grave markers, were seen as representing stone-built settlements. Sites that according to my guidelines could not be interpreted other than ‘site with Roman find material’, such as coins or Roman pottery, were also re-interpreted based on the spatial information of the average size of sites, the assumption being that such sites could represent small settlements without any stone or ceramic building material and only little material culture to begin with. It can therefore be argued that these sites could be the remains of small post-built settlements. A total number of 403 sites were thus re-interpreted as settlement sites. 7 

Cf. Roymans/Derks, this volume, section 1.

8 

9 

The re-interpretation of settlement evidence using spa-

If the distance between two sites is more than 250 m,

tial dimensions actually resulted in a reduction of the

they are interpreted here as representing two different

number of stone-built rural settlements compared to the

settlements. See the discussion in Jeneson, in prep., sec-

first dataset.

tion 4.



The reconstruction of the rural settlement landscape resulted in a new dataset with 1238 (64%) stonebuilt settlements, and 290 (15%) post-built settlements; in addition there are 139 (7%) settlements that are likely to have been post-built, and 262 (14%) for which the type cannot reliably be established. (see figs 3a and 3b).10 If all of the ‘undefined’ sites turned out to be of the post-built type, this would lower the proportion of stone-built farms from 98% in the original dataset to 64%. Nonetheless, even in this case the post-built farms seem to have been a minority in the study area. Surely this is the ultimate proof that the region was dominated by villas. I will argue here that archaeological recording methods in the past have had a major impact on the dataset, with the high visibility of the remains of stone-built structures resulting in a bias towards this type of rural settlement. To substantiate this claim, I will discuss the settlement density map, exploring some potential explanations for areas of higher settlement density. In doing so, I will show that it is not possible to identify a particular environmental factor, but that there is an obvious difference in archaeological recording practices per country, which is most likely the cause. I also believe that these recording practices are responsible for the popular picture of a rural landscape consisting almost exclusively of villas.

3 e v aluati n g settleme n t de n sities The settlement density map (fig. 4) based on the reconstructed rural settlement dataset clearly shows great variation in density across the study area. In most parts it was less than one settlement per square kilometre. Much higher densities, of three to four settlements per square kilometre, are found only in the more eastern part of the region, between the vici of Juliacum and Tiberiacum. The main question is whether the map is a reliable reflection of settlement density in the Roman period. If so, we can expect to find specific conditions in the areas of high density that made them more attractive to Roman farmers. For example, it is generally accepted that environmental factors such as climate, soil type, the presence of water, and elevation determined the suitability of a region for farmers. In this line of reasoning, we can expect the highest densities where environmental conditions differed substantially from, and were demonstrably better than, those in areas with lower densities. Identifying such factors can be done by visually inspecting the map of the study area and comparing the settlement density map in terms of several environmental factors. Better environmental conditions in the areas of high settlement density should be identifiable on the map. Figure 5 indicates roughly the area where the fertile loess soils are found. In line with the above, we can expect soil conditions to be substantially better in the areas of high settlement densities than those in the rest of the study area. However, the map shows that this is not the case, as loess soils are found throughout the study area. The only exception is an area located northwest of the city of Tongres, where a different type of loess is found that is harder to work and therefore less suitable for arable farming. The second variable is that of water. Figure 6 shows the most important rivers and streams in the study area. The assumption here is that proximity to running water was important to Roman farmers. Regions where rivers and streams were sparse are thought to be less attractive to settlers. Looking at the map, it is obvious that this factor cannot be responsible for the differences in settlement density, as in some parts high density is found both in close proximity to water and further away; conversely, low settlement density is found in water-rich areas and in areas where there are hardly any rivers or streams nearby. The third environmental factor is elevation. It is generally assumed that farmers avoided areas at higher altitudes, although it must be said that the maximum elevation in the study area made arable 10 

It should be mentioned at this point that most of these

early, middle or late Roman period are therefore not pos-

additional settlements are not dated beyond the generic

sible. It is assumed that the reconstruction reflects the situa-

‘Roman period’. More detailed reconstructions for the

tion in the middle Roman period, i.e. the 2nd century AD.



Fig. 4. Settlement density map based on the reconstructed rural settlement dataset.

Fig. 5. Map showing the location of loess soils in comparison to the settlement density distribution within the study area.

farming very viable; it goes without saying that cattle farming could be practised at virtually any location. However, it turns out that the main concentration of higher settlement density is spread across an elevation range of 50 to 220 m, the same range in which we also find many areas with lower settlement density. Therefore it cannot be claimed that the higher density zones are due to a particular elevation. The last factor taken into account here is the presence of natural stone, a key ingredient in Roman architecture. It could be argued that natural stone deposits at a particular location are responsible for higher settlement numbers. But the zones with the highest densities in the research region are actually located far from any such deposits, as natural stone is completely lacking there. The stone types used there were brought in from other areas located outside the study region. 11 On the other hand, the pres11 

Rothenhöfer 2005, 106-107.



Fig. 6. Map showing rivers and brooks in comparison to the settlement density distribution within the study area.

ence of limestone and silex deposits in parts of Dutch and Belgian Limburg do not seem to have led to substantially higher settlement numbers. To conclude, the regional variation in settlement density cannot be explained by any one of the natural environmental factors discussed above, which means other variables should be examined. For example, the limes located at the Rhine was a major stimulus for the development of the countryside in the study area, with the army camps and new towns in this zone functioning as important consumers of agricultural produce. It can therefore be argued that proximity to the limes zone would have had an impact on settlement numbers, the assumption being that areas of high demand were more attractive to the suppliers of desired products. However, the density map shows that this argument does not apply to the study area, as the region closer to the Rhine had lower densities than the zones situated further away. Proximity to a civitas capital in general is also thought to have influenced settlement patterns, based on the same assumption that proximity to an area of high demand was an important attraction. It is generally accepted that ten Roman leugae, or 22 kilometres, would have meant approximately one day’s travel for an ox-drawn cart laden with produce. This distance was therefore taken as a measure of proximity to the two civitas capitals in the region (current-day Tongres and Cologne). It was expected that the land within this zone had a significantly higher settlement density, given the assumption that proximity to a major town would have stimulated settlement. However, inspection of the map in figure 7 shows that in fact the reverse seems to be the case. Proximity to one of the two main arteries of the region, the Rhine and Meuse rivers, could also explain higher settlement density, based on the assumption that location close to a main artery would be preferable to location further away. Map 7 shows that this was not the case in the study area. In fact, the only thing that the map seems to prove is that in relation to the last two factors, the area of the highest densities is found in the most unfavourable location, at least 30 to 50 kms from the nearest main river, limes zone and main town. For a farmer these distances would have meant being at least a day’s travel, and more likely two, away from the nearest main market or port. In light of the above, we must conclude that the high settlement densities found in the eastern-central part of the study area cannot be explained by the presence of, absence of or proximity to any environmental factor, whether natural or cultural. Nonetheless, we could still argue at this point that the observed differences are a reflection of the situation in the Roman period and that there were other variables at



Fig. 7. Map showing the zones of 22 kms, or approximately 10 Roman leugae (in green) around the two civitas capitals on the density map of the study area.

work that are yet to be discovered. However, it is the author’s belief that differences in the archaeological practices in each of the three countries in the study area, in the past and today, are responsible for the observed settlement density patterns, rather than any environmental variable.

4 archaeological p ractices – p ast a n d p rese n t In Dutch and Belgian Limburg, interest in the locally found remains from the Roman period was at its peak in the early 20th century. Classically educated individuals, often clergymen and doctors, searched their own backyards for what they considered to be the obvious remains of Roman culture, such as villas, rich graves and elements like statues, frescoes and mosaics. Much of the work done in those days can be attributed to just a handful of people.12 It goes without saying that the way in which the research was carried out is hardly comparable to the archaeological practices of today.Typically only the stone foundations of buildings were laid bare; techniques such as systematic field walking, coring or remote sensing were unknown. In the German region all focus was on the remains of the Roman military presence along the Rhine, to the neglect of the Roman countryside. All this changed after 1950. In post-war Dutch and Belgian Limburg, interest in the remains of the Roman villa world faded away, as all efforts were focused on rebuilding the economy and the country itself. In Dutch Limburg known villa sites became protected monuments where no further research could be carried out. In Germany, however, interest in non-military themes, such as rural settlements, increased after 1945, and in many areas large-scale systematic surveying programmes were set up to capture the settlement landscapes of the past, including the Rhineland. In the area around Jülich in particular, one such programme was conducted in this period.13 In addition, volunteers from the region were involved in field-walking activities for many years. The Valetta Treaty has been implemented in the Netherlands since the late 1990s, resulting in new research carried out in Dutch Limburg regarding Roman rural settlements. In fact, it was not until the 12 

For example, Habets, Goossens and Peeters for Dutch

13 

Hinz 1969.

Limburg. See De Groot 2006; Jeneson, in prep., Section 1.



Fig. 8. Map showing the location of the lignite mines in the German Rhineland on the settlement density map of the study area.

year 2000 that a villa settlement was excavated completely (i.e. the entire settlement rather than just the main building) and published in full.14 Even more importantly, 2009 saw the excavation of a Roman settlement at Heerlen-Trilandis. Located within the so-called ‘villa landscape’, it featured several contemporary post-built houses and not a single stone-built structure, making this the first Roman ‘non-villa’ rural settlement to be completely excavated in the region.15 Two such settlements have also been excavated at Kesselt and Veldweselt in Belgian Limburg in recent years.16 These recent excavations hint at the existence of many more of these ‘non-villa’ rural settlements, which evidently had not yet been discovered by the earlier archaeological work in these two countries. In the meantime, open-cast mining activities in the vicinity of the towns of Jülich and Bergheim in the Rhineland must be mentioned as being of the utmost importance for the archaeology of the region. This type of mining entails the large-scale destruction of complete landscapes, together with the archaeological remains they contain. However, the entire range of archaeological activity, from survey to excavation, is carried out on a regular basis on the land surrounding each of the three open-pit mines. The use of systematic surveys stands out in particular. The land to be destroyed is surveyed several times, under different circumstances. The activities are carried out by one archaeological institute, and all sites are recorded uniformly within a single database.17 I argue here that this mode of operation, and in particular the practice of systematic field surveys in combination with remote sensing techniques, is the only explanation for the area of high settlement density in the eastern-central part of the study area. Figure 8, showing the location of the three open-pit mines on the settlement density map, substantiates this claim. The fact that not all of the area located within the mining zone has the same high settlement density can be explained, again by focusing on the history of archaeological activities there. Figure 8 shows a low settlement density at the two large mines in the east (the Garzweiler-Frimmersdorf mines). This is because these mines were exploited much earlier in the 20th century, prior to the era of intensive archaeological

14 

Tichelman 2005.

15 

G. Tichelman (RAAP-Zuid, Weert), oral communication.



16 

Vanderhoeven et al., 2006; A. Vanderhoeven (Tongeren), oral communication.

17 

See Gaitzsch, this volume.

Fig. 9. Proportions per country per research type, for the overall dataset of the study area.

prospection. In contrast, settlement densities are highest at the location of the Hambach mine, situated between Jülich and Bergheim. Work here did not start until the late 1970s, and field surveying has been part of the systematic research approach since the beginning. In addition, until the moment mining activities begun here, the region was covered by a large forest, protecting the sites beneath from farming activities for centuries. This could also explain why settlement density here is even higher than in the Weisweiler region (located west of Jülich), where the land has been farmed and ploughed for hundreds of years. The different archaeological traditions can also be compared by examining the variable ‘type of research’ per country, as shown in figure 9. The percentages indicate that non-invasive practices, such as field survey and aerial photography, are the most dominant types of research in the German part of the study area. In the Dutch area, on the other hand, invasive methods such as excavations and trial trenches are the most frequent. Another argument to support my claim that the observed differences in settlement density are the result of specific archaeological practices, rather than a reflection of the situation in the Roman period, is that each of the areas with a higher settlement density can be attributed to a specific person or institution that carried out archaeological research there, in the past or today. For example, the areas of higher settlement density in Dutch Limburg can all be attributed to three key figures from the period prior to 1940,18 or to the work of current-day council archaeologists.19 An important implication of the suggestions above is that the observed high densities in the German lignite mining areas should be seen as the rule, rather than the exception. If archaeological practices are responsible for this variation, it then follows that in areas where investigations have been more intense, the resulting density is a better indication of the situation in the past than in other areas where less activity took place. This would mean that a habitation density of two to three settlements per square kilometre should be accepted for the entire study area.

18

Respectively, Habets from Berg & Terblijt, Goossens from

Rolduc near Kerkrade, and Peeters from Heerlen. See also

19 

Council archaeologists from the towns of Maastricht and Sittard.

note 12.



Fig. 10. Map of the reconstructed landscape west of the current-day town of Bergheim, shown with a grid of 1 by 1 km, where the ratio of stone-built to post-built settlements seems to be 1:1.

The next question to answer is whether archaeological practices are responsible for the current views on the composition of the settlement landscape. This is an issue which has already been addressed by scholars, in particular those working in the Mediterranean area, where field survey results are often the only datasets available, including ‘legacy data’.20 It has been pointed out that the high visibility of remains of stone-built structures, either buried or scattered on the surface, inevitably results in a bias towards this type of settlement, and an underestimation of smaller, post-built sites, where much less durable material was present to start with. So even when systematic field surveys are carried out, the difference in the find assemblages of the two main types of settlement means that the results are often biased in favour of stone-built sites. In areas where archaeological activity primarily takes the form of excavations, such as in Dutch Limburg, the past focus on stone-built structures influenced where subsequent investigations were carried out, and this in turn has been translated into predictive models, which are used by archaeologists today to determine where new research is to be carried out, and more importantly, where not. Thus the chance of a post-built settlement being discovered and excavated is very small. In fact, the only reason why the post-built settlement at Heerlen-Trilandis was excavated, was because roof tile fragments had been found on the surface of the site, and it was therefore registered as a villa site in Archis, the national archaeological database. It is not too far-fetched to conclude that because of all this, it is nearly impossible to evaluate this perception of a villa-dominated landscape. I will now give one last example to demonstrate the likelihood that post-built rural sites are underrepresented in most parts of the study area. As described earlier, in the methodology for the reconstruction of the settlement landscape, non-settlement sites were re-interpreted in terms of settlements, arguably of the post-built type. In the reconstructed landscape shown in figure 10, this results in a proportion of stonebuilt to post-built sites of almost 1:1. It should come as no surprise that this ratio can only be reconstructed within the area of the Hambach lignite mine. Here many sites consisting of little more than a few pottery fragments have been detected by means of repeated surveying. It can be argued that these sites represent small, single farm settlements, scattered amongst the larger stone-built farms. Unfortunately 20 

See for example Goodchild 2007, chapter 3, and Witcher 2008.



they are rarely excavated, making it difficult to substantiate this claim. However, for other similar regions too, scholars are suggesting that the current views regarding the proportions of stone-built to post-built rural settlements are in need of adjustment, and it is hoped that future fieldwork will address this issue.21 The example above touches upon an issue that has not yet been addressed here, and that is generally absent in the discussion of Roman rural landscapes in this part of the empire,22 namely the labour power required for the large-scale production of cereals. Large-scale arable farming is dependent on seasonal labour, especially for ploughing and harvesting. In a landscape like the one shown in figure 10, the stonebuilt type settlements are all assumed to have produced surplus cereals. Another common assumption is that more than one family lived at a villa. Even though it is difficult to determine with certainty the function of the secondary buildings within the enclosed settlement, it is generally thought that at least one or two families lived there, with the men working permanently on the farm. But in light of recent work on the demand for labour during the ploughing and harvesting season, we can assume that with an average size of at least 50 hectares, many more men would have been needed at certain times of the year.23 The question is where these people came from, and where they lived the rest of the year. Although it lies beyond the scope of this article to discuss this issue more thoroughly, I suggest that the landscape in figure 10 shows a potential scenario in which small, possibly dependent farmers lived on plots of land in between the stone-built settlements. For most of the year they were self-sufficient, but they probably worked on the fields of the villa settlements during the ploughing and harvest season. This shows that in order to really understand the social structure of these landscapes and the workings of the agricultural economy, it is vital to fully reconstruct the settlement landscape. If an entire layer of this landscape is missing, it is impossible not only to reconstruct the demographic situation, but also to build social and economic models that reflect the original situation. It is difficult to examine the social hierarchy of the rural population in this region if we only have information about the settlements of part of that population.

5 co n clusio n s This article set out to challenge the current perception of the Roman landscapes on the loess soils between Tongres and Cologne, in particular ideas about the settlement pattern and the average density of rural settlement. The result is a new reconstruction of the settlement landscape, based on a wide range of archaeological evidence in combination with spatial analyses of this evidence. I have demonstrated that in this new reconstruction, settlement density was much higher in places than the generally accepted one settlement per square kilometre. I have also shown that the number of post-built sites equalled the number of stone-built sites in some areas. More importantly, I have shown that past and present archaeological practices rather than environmental factors are behind the observed variation in settlement density, with the use of systematic field surveys seemingly a decisive factor. In addition, the traditional focus on stone-built structures has meant that hardly any attention has been given to post-built settlements, as though they were absent from these landscapes, a view that is challenged by new discoveries and the reconstruction presented here. The implication of these conclusions is a need to reconsider certain assumptions about the landscapes in this part of the empire. Admitting higher settlement densities than the one per square kilometre generally accepted so far has important consequences not just for the amount of land per rural settlement, but also for population numbers and for a whole range of other issues, such as the production potential of 21 

Cf. the discussion in Ouzoulias 2006, 147-150 and Ouzoulias 2010, for the north of France.

22 

23 

Kooistra 1996 and Goodchild 2007, for example, calculated the need for additional labour during the seasonal

Some noteworthy examples notwithstanding, for exam-

activities of ploughing and harvesting, based on different

ple Kooistra 1996.

sized estates.



the farms, individually and as a whole. Reconsidering the proportion of non-villa, post-built settlements throughout the area also has consequences for various related issues, not the least of which is the social organization of rural society. Nonetheless, I do think that the study area can still be regarded as a ‘villadominated landscape’. Even with a much higher proportion of post-built settlements, the stone-built type still represents the majority (65 %). There are of course many more issues that need to be addressed regarding these landscapes. A formidable problem is the lack of well-dated sites, currently only one-third of the entire dataset. Substantial new data is needed before we can make a detailed reconstruction of the development of these landscapes. The theme of settlement hierarchy also deserves further attention, as diversity has been established even amongst the stone-built settlements.24 Fortunately, archaeological researchers have embraced the possibilities offered by GIS for spatial data analysis on a regional and interregional scale, and new ways of characterizing settlement landscapes using a variety of analysis and modelling techniques are currently being developed. This has been evidenced recently by French scholars working in this area,25 and by the fact that an entire session at the 35th Conference on Computer Applications and Quantitative Methods in Archaeology, held in Berlin in 2007, was devoted to the subject of identifying settlement patterns and territories. It is hoped that this article has demonstrated the potential of (low-tech) modelling and spatial analyses in a GIS environment. It goes without saying that more archaeological evidence is needed to substantiate the claims made in this article. I hope that archaeologists will acknowledge the possibilities suggested here, and will translate them into new research strategies for the region. Only then will it be possible to change the commonlyheld opinions about rural landscapes in this part of the empire.

re f ere n ces Bertoncello, F., 2002: Villa/vicus. De la forme de l’ habitat aux réseaux de peuplement, in P. Garmy/Ph. Leveau (eds), Dossier Villa et vicus en Gaule Narbonnaise, Revue Archéologique de Narbonnaise 35, 39-58. Bertoncello, F./L. Nuniger, 2005: Analyse statistique des données et modèle gravitaire. Étude de cas sur la modélisation des réseaux d’habitats, Ecole Thematique Géomatique, Analyse et Modelisation Spatiales en Archeologie, 12-17 Sept. 2005, Dijon. Bertoncello, F./L. Nuninger, in press: From archaeological sherds to qualitative information for settlement pattern studies, in Niccolucci, F. (ed.), Beyond the artefact. Digital interpretation of the past. CAA2004, Computer Applications and Quantitative Methods in Archaeology, 13-17 April 2004, Prato  (Italy), Budapest. Also downloadable via http://univ-fcomte.academia.edu/LaureNUNINGER/Papers/377913/ From_Archaeological_Sherds_to_Qualitative_Information_for_Settlement_Pattern_Studies Byvanck, A.W., 1947: Excerpta Romana III. De bronnen der Romeinsche geschiedenis van Nederland. Part 3, ’s-Gravenhage. Boe, G. de, 1973: De landelijke bewoning in de Romeinse tijd, Het oude land van Loon, 85-114. Favory, F./J.J. Girardot/C. Raynaud, 2003a: Typologie archéologique de l’habitat rural, in S. van der Leeuw/F. Favory/J.L. Fiches (eds), Archéologie et systèmes socio-environnementaux. Études multiscalaires sur la vallée du Rhône dans le programme Archaeomedes, Paris, 201-237. Favory, F./J.J. Girardot/C. Raynaud, 2003b: La dynamique spatio-temporelle de l’habitat rural galloromain, in S. van der Leeuw/F. Favory/J.L. Fiches (eds), Archéologie et systèmes socio-environnementaux. Études multiscalaires sur la vallée du Rhône dans le programme Archaeomedes, Paris, 301-321. 24 

Cf. Habermehl, this volume.

25 

Cf. Bertoncello 2002; Bertoncello/Nuninger 2005 and in press; Favory/Girardot/Raynaud 2003a and 2003b.



Gaitzsch, W., 1986: Grundformen römischer Landsiedlungen im Westen der CCAA, Bonner Jahrbücher 186, 397-427. Gaitzsch, W., 1988: Geländeprospektion und Flächenstruktur römischer Siedlungen im Hambacher Forst, Kreis Düren, Archäologisches Korrespondenzblatt 18, 373-386. Gaitzsch, W., 2002: Römische Landbesiedlung. Hof- und Wirtschaftsräume, in W. Menghin/D. Planck (eds), Menschen, Zeiten, Räume. Archäologie in Deutschland. Begleitband zur gleichnamigen Ausstellung, Stuttgart, 267-273. Gechter, M./J. Kunow, 1986: Zur ländlichen Besiedlung des Rheinlandes in römischer Zeit, Bonner Jahrbücher 186, 377-396. Goodchild, H., 2007: Modelling Roman agricultural production in the Middle Tiber Valley, Central Italy, (unpublished PhD thesis, University of Birmingham). De Groot, T., 2006: Roman villae in the loess area of the Dutch province of Limburg. An analysis of their number, distribution and preservation, Berichten van de Rijksdienst voor het Oudheidkundig Bodemonderzoek 46, 275-302. Heimberg, U., 2002/2003: Römische Villen an Rhein und Maas, Bonner Jahrbücher 202/203, 57-148. Hinz, H., 1969: Kreis Bergheim, Düsseldorf. Kooistra, L.I., 1996: Borderland farming. Possibilities and limitations of farming in the Roman period and Early Middle Ages between the Rhine and Meuse, Amersfoort. Künow, J., 1994: Die ländliche Besiedlung im südlichen Teil von Niedergermanien, in H. Bender/ H. Wolff (eds), Ländliche Besiedlung und Landwirtschaft in den Rhein-Donau Provinzen des römischen Reiches. Vorträge eines internationalen Kolloquiums vom 16.-21. April 1991 in Passau, Espelkamp, 141-197. Lenz, K.H., 1999: Siedlungen der römischen Kaiserzeit auf der Aldenhovener Platte, Rheinische Ausgrabungen 45, 69-85 / 106-113. Maeyer, R. de, 1940: De overblijfselen der Romeinsche villa’s in Belgie. De archeologische inventaris, Antwerpen. Mertens, J., 1964: Enkele beschouwingen over Limburg in de Romeinse tijd, Archaeologia Belgica 75, 5-42. Ouzoulias, P., 2006: L’économie agraire de la Gaule. Aperçus historiographiques et perspectives archéologiques (unpublished PhD thesis, Université de Franche-Comté, downloadable via http://tel.archives-ouvertes.fr/tel-00011567 Rothenhöfer, P., 2005: Die Wirtschaftsstrukturen im südlichen Niedergermanien. Untersuchungen zur Entwicklung eines Wirtschaftsraumes and der Peripherie des Imperium Romanum, Rahden (Kölner Studien zur Archäologie der römischen Provinzen 7). Tichelman, G., 2005: Het villacomplex Kerkrade-Holzkuil, Amersfoort (ADC-rapport 155). Tichelman, G., in prep: De Romeins-inheemse nederzetting Heerlen – Trilandis. Vanderhoeven, A., 2002: Aspekte der frühesten Romanisierung Tongerens und des zentralen Teiles der civitas Tungrorum, in Th. Grünewald/S. Seibel (eds), Kontinuität und Diskontinuität, Berlin (RGA-E Band 35), 119-144. Vanderhoeven, A. et al., 2006: Veldwezelt. Vues socio-économiques nouvelles sur les campagnes, Dossiers Archéologie et sciences des origines 315 (La Belgique Romaine), 60-63. Wendt, K.P./A. Zimmermann, 2008: Bevölkerungsdichte und Landnutzung in den germanischen Provinzen des römischen Reiches im 2. Jahrhundert n.Chr. Ein Beitrag zur Landschaftsarchäologie, Germania 86, 1-36. Willems, W.J.H., 1988: Die große Villa rustica von Voerendaal (Niederlande), in M.E.Th. de Grooth (ed.), Villa Rustica. Römische Gutshöfe im Rhein-Maas-Gebiet, Freiburg, 8–13. Witcher, R.E., 2008: (Re)surveying Mediterranean rural landscapes. GIS and legacy survey data, Internet Archaeology 24.



The villa landscape of the Middle Aare valley and its spatial and chronological development Caty Schucany

1 Introduction 2 The villa landscape and its organization 3 Biberist 4 Bellach 5 The development of the villa landscape and its social interpretation References

1 i n troductio n The landscape to be discussed - the Middle Aare valley on the south foot of the Jura range – comprises a stretch of land, roughly 60 km long, in the centre of the Swiss plateau (fig. 1).1 In antiquity, this area was part of the civitas Helvetiorum. The valley is positioned to the east of Lake Biel and between the Jura and the Napf hills, a prominent rise of the Alpine foothills. It is a landscape strongly moulded by glaciers, a river valley with a width of 20-25 km with the Aare meandering at a height of 420-430 m above sea level. In the northeast it is bordered by the first chain of the Jura range, to the south by the hills of the lateral moraines, which cover the molasse ranges and were left untouched by the glaciers.2 The settlements lie above the Aare plains, at around 440-460 m above sea level, preferably on the gently sloping foothills of the southern Jura range, or less often, on the opposite side of the valley on the lateral moraines, created by landslides after the glaciers’ retreat. The hillsides are all drained by many streams. The ground is composed of moraine material and is thus fertile. Climatically, the valley is exposed to both storm bringing west winds as well as the cold bise from the northeast. In the winter, it is often covered by a dense fog.3 In antiquity, when the Aare had not yet been diverted into Lake Biel, as was the case after the Jura water correction in the 19th century, the river would have periodically flooded the plains, like in the Middle Ages and the early modern period. In a cultural sense, the western part of the middle Aare valley belonged to the hinterland of Aventicum (Avenches), capital of the civitas Helvetiorum, possibly forum Tiberii from the 2nd decade AD onwards and was elevated to the status of colony around 70 AD − colonia Pia Flavia Constans Emerita Helvetiorum Foederata.4 However, the eastern part of the valley, as the cooking vessels suggest, is oriented to the east,

1 

I would like to sincerely thank Andrew Lawrence, Basel for the as ever excellent translation into English.

4 

On the issue of cultural identity: Schucany 1999, 88; Schucany et al. 1999, 120; for Avenches see Archäologie

2 

Schweiz 24, 2001. The elevation to the status of colony

3 

cannot merely be viewed as a punishment of the civitas

See Schucany 2006, 47. Schucany 2006, 33.



10 km

N

Rhine

Augusta Raurica

Vindonissa Olten

re

Holderbank Aa

Jura

Salodurum

Fig. 1. The Middle Aare valley in Switzerland in the Aventicum

Napf

Roman period. The contour line of the grey area corresponds approximately with the 500 m asl line.

towards Vindonissa, where a legionary camp stood in the 1st century AD. The border zone between both regions is not just defined by the diffusion of the different types of cooking vessels, but also by a thinning out of the settlement.5 The centre of each respective part of the valley was a vicus, Salodurum (Solothurn) in the west, epigraphically proven as a vicus,6 and Olten in the east, whose ancient name or status is still unknown.7 Both small towns were built where the Aare can easily be bridged, with an inhabited area of 5 ha at most where 1000-1500 residents could have lived. While the settlement in Olten started in the late Augustan period, Salodurum was not founded until the Tiberian period. In the late 3rd and early 4th century AD, both places were turned into a castrum, no doubt in order to secure the important Aare crossings. A further small town lay in Holderbank (fig. 1) in the side valley leading to the Upper Hauenstein, a pass connecting the Aare valley with the Upper Rhine, or Aventicum with Augusta Raurica. As it covers an area of just 2 ha, only 500 inhabitants would have lived there.8 The settlement started in the second quarter of the 1st century AD and carried on at least until the 2nd century.



   v     n     p   n        n  z   n

Apart from the vici of Salodurum, Olten and Holderbank, which can be classed as small towns, we know of three different types of rural settlement. First, there are at least 30 large villas with enclosures of more than 5 ha and splendid residences of over 60 m length. Next we know of just under 40 middle-sized villas with enclosures up to 3 ha and residences of up to 40 m length, and finally we know a large number of small farmsteads with enclosures of up to 1 ha. A number of settlements cannot be classed into one of these three categories with any certainty. The majority of these will be small farmsteads with small enclosures Helvetiorum but far more as a strengthening of the Hel-

for some time. The elite of the civitas also made up the

vetian elite, which had long possesed Roman citizenship.

elite of the colony; the magistrates still came from the

At the least those belonging to Galba’s clientele and

same families. See Frei-Stolba 1999, 67-91.

who were initially on the losing side to Vitellius, the

5

Schucany 2001; Schucany, in prep.

later opponent of Vespasian in the civil war of 69 AD.

6

Walser 1980, no. 141.

Furthermore, through his father, this emperor had a close

7

Schucany 1999, 89; Ebnöther/Schucany 1998, 92.

relationship to the city’s ruling class; his son Titus even

8

Ebnöther/Schucany 1998, 92.

seems to have lived with his grandfather in Aventicum

7

lying above 500 m above sea level, since larger villas are hardly ever found above this level. In the Jura range there are several find spots, lying clearly higher and which we sweepingly term “forest dwelling”.They were probably used in connection with the nearby forests for timber and coal or for cattle rearing and pastoral economy; a few could be interpreted as sanctuaries. The number of the large and possibly the middle-sized villas will roughly correspond to the erstwhile reality, as they yield clearly visible traces and can be determined with ease.9 However, for the small farmsteads, a number of unknown cases must be reckoned with. The mapping of the settlements shows that the large villas are equally distributed, giving the impression that they were the centres of larger domains to which middle and smaller farmsteads belonged. According to the natural borders, like the Jura and rivers, and according to the equidistant lines to the next large villas, these domains would have comprised of 2000 ha. The basic assumption of this study is that, aside from the large villa, these domains were made up of one to three middle-sized villas and the same amount of smaller farmsteads. Starting with a population of 100-120 persons for the large villas, 30-40 for the middle-sized ones and 10 for the small farmsteads, a domain would have comprised of 150-200 people.10 In the following, two such domains will be presented in their organisation and their development: Biberist and Bellach (fig. 2). They show that these domains were anything but static structures.

3

b i b erist

Taking the rivers Aare and Emme as natural borders and using middle lines between the villa and its next neighbours as the east and west borders, the domain of the large villa of Biberist-Spitalhof can be estimated at 1800 ha.11 According to the field names, which reflect the landscape of the 16th century at the latest, around 530 ha (29%) were permeable soils, suitable for agriculture, while 280 ha (16%) were saturated and used as meadows and pastures. The zones above 500 m above sea level would have been wooded as they are now (740 ha resp. 40%).Woodland would have taken up around more than half of the domain (990 ha). Apart from the large villa in the northeast corner of the domain, we know of a middle-sized villa in the southern centre in Lohn as well a small farmstead in Biberist-Oberwald. The villa’s personnel of 120-130 can be calculated on the basis of the labourers’ houses, probably between 8 and 10, and on the basis of the number of people living in middle-sized villas and small farmsteads.12 Every year, approximately one third of the eligible land would have been ploughed. The chronological development shows that the organisation of the domain changed over the course of time.13 Firstly, on the spot of the later middle-sized villa, in the third quarter of the 1st century, a farmstead enclosing around 1 ha can be made out. This first farmstead would have only cultivated the surrounding fields in the northeastern corner of the later domain. At around 80 AD the farmstead was transformed and converted into a villa with a rectangular, axially laid-out courtyard. The no-doubt increased personnel would have ploughed the new land in the northwest corner of the domain. At the beginning of the 2nd century, the axial villa was once again enlarged and a middle-sized villa was

9 

At the end of the 19th and beginning of the 20th cen-

living contemporary can be calculated (Martin-Kilcher

tury, during a period of archaeological pioneering, a

1976, 102; 133-137). For the population of large villas

myriad of local and regional historians combed acres,

and smaller farmsteads see below.

meadows and woods searching for the archaelogical past

11 

of their village (see Jahrbuch der Schweizerischen Gesellschaft

12 

für Urgeschichte 1, 1908, 69). In the case of the large and middle-sized villas they were often successful. 10 

In the completely analyzed cemetery of the villa of Courroux, a complex of middle size, around 35 people

Schucany 2006, 273-281. Around 80 persons in the large villa, 35 persons in the middle-sized villa and 10 persons in the farmstead (Schucany 2006, 278-280).

13 

On the development of the complexes: Schucany 2006, 253-270.



Jura

OberdorfHerrenkeller

OberdorfBellvuestrasse

Lommiswil

Bellach

BellachMannwil

BellachFranziskanerhof

LangendorfHüslerhof

Salodurum

SelzachSpielhof BiberistSpitalhof

BiberistOberwald

Em me

Aare

Oberwald

Fig. 2. Reconstructed extension of the domains of the vil-

Lohn

Biberist

las of Biberist and Bellach and recon-

mountain/woodlands meadow arable field unidentified 0

1 km

structed landuse pat-

vicus large villa middle-sized villa small farmstead/uncertain site forest dwelling

terns in the western part of the Middle Aare valley.

established in Lohn. The personnel in Lohn would have cultivated the good soils south of Oberwald. The people living in the farmstead Oberwald, which was occupied from this time onwards, may have been familiar with forestry. This constellation persisted until the turn of 3rd century, when the farmstead was evidently abandoned. Large changes and renovations in the residential villa could be attributed to a change in ownership.14 Shortly after the middle of the 3rd century, the large villa was pillaged.15 The villa in Lohn was abandoned at this time at the latest. In the later 3rd century, probably after a hiatus of several years, maybe a decade, a partial reconstruction took place in the area of the large villa. Apart from the ruins of the vilicus house to the north of the erstwhile gate, living quarters and a granary were built, forming a small estate. The inhabitants, maybe 10 or 20, were insufficient to cultivate the whole area of the previous domain. They would have kept to the fields in the north, whose harvest could have easily been carried by the much smaller granary.16

4

b ellach

Taking the river Aare, a prominent stream gully towards the vicus Salodurum as well as the Jura foothills as natural borders and using a middle line between the villa and its next neighbour to the west as its

14 

Schucany 2006, 263.

15 

Contemporary burnt layers in various houses, the remains of deceased and weapons in the debris of the



building next to the gatehouse all point to a violent end (Schucany 2006, 266-269). 16 

Schucany 2006, 269.

vicus large villa middle-sized villa small farmstead uncertain site forest dwelling

late 1st cent. BC early 1st cent. AD middle 1st cent. AD late 1st cent. AD 1st cent. AD

130

Jura

0

125

5 km

92

Olten

165 86

Holderbank

84

135

90

83

101

106

109

122

105

82

re Aa

54 51 30

37

116

Salodurum

24

44

115

77 114

41

70

9

62

Napf

47 15

Fig. 3. The villas from the 1st century BC and the 1st century AD in the Middle Aare valley (nos. after Schucany 1999, Abb. 4-5; Schucany/Wullschleger, in press).

western boundary, the domain of the large villa Bellach-Mannwil can be estimated at c. 2200 ha.17 Even if an analysis of field names is lacking, the structure of the landscape can be described as follows.The steep south slope of the first Jura chain was definitely wooded as it is today (around 740 ha resp. 34%). In the Aare plain to the south, riparian woodland is a distinct possibility (around 90 ha resp. 4%). Subtracting the four stream gullies, arable land of around 1200 ha (55%) would have been available. Apart from the large villa Bellach-Mannwil in the southern centre, we know of three middle-sized villas, Selzach-Spielhof, Bellach-Franziskanerhof and Langendorf-Hüslerhof, all situated on more or less the same contour line as the large villa on a hypothetical road running parallel to the river Aare. Further up the slope, away from the road, two small farmsteads, Lommiswil-Katzenstuden and OberdorfBellvuestrasse, as well as, higher in the Jura, a so-called forest settlement, Oberdorf-Herrenkeller, are documented. The personnel of the large villa and the middle-sized villas as well those in the farmsteads a little higher up would have been charged with cultivating the surrounding fields. The people inhabiting the farmsteads higher up the hill (Oberdorf-Herrenkeller), probably situated in the woods, would have worked in the forestry. In total, personnel of around 230 persons can be estimated.18 Of the villa Bellach-Mannwil itself, only the residence was excavated along with one outhouse.19 The former had a front of over 60 m length, thus enabling us to class this complex into the highest size 17 

Schucany/Wullschleger, in press. I would like to thank

19 

Jahrbuch für Solothurnische Geschichte 48, 1975, 217-229;

Mirjam Wullschleger, Solothurn, for references and dis-

Archäologie des Kantons Solothurn 8, 1993, 115; Jahrbuch der

cussions.

Schweizerischen Gesellschaft für Ur- und Frühgeschichte 75,

18 

Around 100 in the large villa, 3x35 in the middle-sized

1992, 211; 76, 1993, 203; Archäologische Schweiz 15, 1992,

villas and 3x10 in the small farmsteads (see above refer-

185.

ence 10 and 12).



bracket. For the time being, no further statements can be made on the size and the organisation of this villa. According to the finds it was inhabited from the 1st to the mid 3rd century. With the exception of the two villas Bellach-Franziskanerhof and Langendorf in the east, only scarce documentation from old excavations exists on the other complexes of the domain. Thus nothing can be said about their chronological development.20 The two small farmsteads are even lacking finds which might yield a more precise dating than just “Roman”. Of the villa of Bellach-Franziskanerhof we only know of an outhouse and a quite large bath building, permitting us to class this complex to the middle size bracket.21 According to the finds, this part of the estate was constructed in the beginning of the 2nd century and persisted until the 4th or possibly the 5th century. The villa of Langendorf-Hüslerhof is the one best known to us.22 It deals with a settlement of the irregular courtyard type with an area of around 3 ha, consisting of a main building and two outhouses. Two pits with pottery from the Late La Tène period hint at a predecessor dating to the 1st century BC. In the middle of the 1st century AD, after a hiatus of over three generations, a building was erected belonging to a smaller farmstead. Towards the end of the 1st century a main building and two outhouses were constructed, standing for a generation until the beginning of the 2nd century. After a hiatus of a generation, not until the second half of the 2nd century, the establishment was reinstated: the main building was renewed and repaired, and the same happened to the two outhouses. A shed and a Grubenhaus were added. The villa stayed in this form until the middle of the 3rd century. The domain of the villa of Bellach-Mannwil saw a dynamic history. It is still unclear whether the Late Iron Age farmstead in the southeastern corner of the later domain was its starting point. Around the middle of the 1st century AD the complex was enlarged with the construction of the large residence in Bellach-Mannwil in the southern centre of the domain along with the courtyard in the southeastern corner of Langendorf-Hüslerhof. It is likely that the villa in Selzach-Spielhof was also erected around this time, in the southwestern part of the domain. The 2nd century saw a densification of the settlement with the construction of the villa of Bellach-Franziskanerhof in the southeastern corner and of a small complex high up in the Jura hills in the northeastern corner. In the 3rd century evidently only the large villa and both villas in the southeastern corner of the domain existed. The whole domain ceased to exist after the middle of the 3rd century AD; the people inhabiting Bellach-Franziskanerhof cultivated probably just the surrounding fields in the southeastern corner.

5 the de v elo p me n t o f the v illa la n dsca p e a n d its social i n ter p retatio n For the moment, a still too large a number of villas can only be roughly dated, if at all, due to the sparse finds. This leads to an inconclusive picture, especially for the early and late periods. So the chronological development of the villa landscape (fig. 3-4) can only be insufficiently described for the earliest and

20 

Selzach-Spielhof: Archäologie des Kantons Solothurn 5, 1987, 161; Lommiswil-Katzenstuden: Jahrbuch für Solothurnische Geschichte 10, 1937, 233; Oberdorf-Bellvues-

Ur-und Frühgeschichte 4, 1911, 180-181; 65 1982, 190-191; 66, 1983, 274-275. 22 

Harb/Wullschleger 2010. I would like to thank Mirjam

trasse: Jahrbuch für Solothurnische Geschichte 33, 1960,

Wullschleger, Solothurn, for important references from

230; Oberdorf-Herrenkeller: Jahrbuch für Solothurnische

her work.

Geschichte 29, 1956, 178-179. 21 

Spycher 1981; Spycher 1982; Spycher 1983; Archäologie des Kantons Solothurn 1, 1981, 93; 3, 1983, 119-121; 7, 1991, 111-112; Jahrbuch der Schweizerischen Gesellschaft für



vicus large villa middle-sized villa small farmstead uncertain site forest dwelling

1st cent. AD 2nd cent. AD 3rd cent. AD still in the 4rd cent. AD 0

127

Jura

5 km

96 93

92 91

Olten

109

Holderbank 136

103

133

75

63 33

18

19

36 37

99

61

76

re Aa

Salodurum 44

40

Napf

Fig. 4. The villas from the 1st century BC and the 1st century AD, the new villas from de 2nd and 3rd centuries AD and the villas still occupied in the 4th century AD in the Middle Aare valley (nos. after Schucany 1999, Abb. 4-5; Schucany/Wullschleger, in press).

the latest periods, i.e. the for the late 1st century BC/early 1st century AD and for the period after the middle of the 3rd century. On the other hand, the data basis for the 1st and 2nd century AD is better and yields a substantiated picture.23 We know of only two settlements from the 1st century BC (fig. 3), both from the western part of the region in question: Apart from the already mentioned Late Iron Age farmstead in Langendorf-Hüslerhof (no. 37), there is also a farmstead with a continual development in Messen (no. 15), which was transformed into a large villa in the course of the 1st century AD.24 In the early 1st century, middle-sized villas in the eastern part of the region can be added (nos. 90, 114-116, 122). The founding of the legionary camp of Vindonissa may have played a role in this matter.25 The stationing of at least 6000 men and countless riding and draft animals required a enormous increase in corn. A part of this corn will have been sought after in the nearby and faraway environs. Around the middle of the 1st century, the southern foothills of the Jura were populated with larger and smaller establishments (nos. 30, 44, 77, 82, 84, 109, 135, 165). At this time at the latest, the expansive domains, controlled by a large villa, would have taken root. The smaller complexes would have been subordinated to the larger ones. In the second half of the 1st century further larger villas were added (nos. 9, 41, 62, 70, 101), especially in the area south of the Aare. 23 

See also Ebnöther/Monnier 2002, 137-141.

24 

Archäologie und Denkmalpflege in Kanton Solothurn 3, 1998,

25 

For the latest summary on Vindonissa, see Fellmann 2006, 422-426.

58-83.



In the 2nd century (fig. 4), a densification of the settlement landscape becomes apparent (nos. 18, 19, 33, 40, 61, 63, 75, 91, 93, 99, 103, 133, 127, 96). Even if there are exceptions, like the hiatus mentioned in the example of Langendorf-Hüslerhof (no. 37), this period is still a time of prosperity for the villa landscape. In the 3rd century, the only new foundation can be found in Oberbuchsiten-Eichacker (no. 76).26 After the mid 3rd century a massive rupture can be observed and many complexes seem to have been subsequently abandoned. This could have to do with the well-known political troubles of the third quarter of the 3rd century, triggered by the economic crisis. This situation would have in turn finished off the intensive overproduction of the large land ownerships relying on safe trade routes and organised markets. New excavations have however shown that, in the late 3rd and early 4th centuries, a certain rebuilding took place within the ruins, even if this was on a smaller scale in half-timbered construction (nos. 36, 44, 92, 96, 109), traces of which might have escaped earlier excavators.27 Even so, these farmsteads, as they must be termed, returned to subsistence agriculture and cultivated only the immediate surrounding lands.28 New foundations such as Balsthal-St.-Wolfgang (no. 136) are markedly seldom, not only in the middle Aare valley but also elsewhere in the Swiss Plateau. Despite the insufficient data basis, the largest villas still seem the earliest within the respective estates. The native elites were thus responsible for the organisation of the agriculture, cultivating and slowly extending their inherited or even allocated terrains along with the families under their control.

re f ere n ces Ebnöther, Ch./J. Monnier, 2002: Ländliche Besiedlung und Landwirtschaft, In L. Flutsch/U. Niffeler/F. Rossi (eds), Die Schweiz vom Paläolithikum bis zum frühen Mittelalter (SPM) V. Römische Zeit, Basel, 135177. Ebnöther, Ch./C. Schucany, 1998: Vindonissa und sein Umland. Die Vici und die ländliche Besiedlung, Jahresbericht der Gesellschaft Pro Vindonissa 1998, 67-97. Fellmann, R., 2006: Windisch/Vindonissa, in M. Reddé/R. Brulet/ R. Fellmann/J.K. Haalebos/S. Von Schnurbein (eds), Les fortifications militaires. L’architecture de la Gaule Romaine, Bordeaux, 422-426. Frei-Stolba R. 1999: Recherches sur les institutions de Nyon, Augst et Avenches, in M. DondinPayre/M.-Th. Raepsaet-Charlier (Eds) Cités, municipes, colonies. Le processus de municipalisation en Gaule et en Germanie sous le Haut Empire romain, Paris, 29-95. Harb, P./M. Wullschleger, 2010: Der römische Gutshof von Langendorf, Kanton Solothurn. Ausgrabungen an der Hüslerhofstrasse 1993, 1994 und 2002, Basel (Antiqua 46). Martin-Kilcher, S., 1976: Das römische Gräberfeld von Courroux im Berner Jura, Derendingen/Solothurn (Basler Beiträge zur Ur- und Frühgeschichte 2). Schucany, C., 1999: Solothurn und Olten – Zwei Kleinstädte und ihr Hinterland in römischer Zeit, Archäologie Schweiz 23, 88-95. Schucany, C., 2006: Die römische Villa von Biberist-Spitalhof/SO (Grabungen 1982, 1983, 1986-1989). Untersuchungen im Wirtschaftsteil und Überlegungen zum Umland, Remshalden (Ausgrabungen und Forschungen 4). Schucany, C., in prep.: L’occupation du territoire et la répartition de céramique culinaire: deux moyens pour déterminer les limites et l’organisation spatiale de la cité des Helvètes?, in C. Batigne Vallet (ed.),

26

 Jahrbuch für Solothurnische Geschichte 16, 1943, 198-202;

on coins. No pottery is available.

Jahrbuch der Schweizerischen Gesellschaft für Ur- und Frühge-

27 

schichte 33, 1942, 93. However, the dating is based solely

28 



See also Ebnöther/Monnier 2002, 150-151. See also Ebnöther/Monnier 2002, 151-154.

Les céramiques communes comprises dans leur contexte régional. Table ronde, 2 et 3 février 2009 à Lyon, Paris (Publication de la Maison de l’Orient et de la Méditerranée). Schucany, C./S. Martin-Kilcher/L. Berger/D. Paunier, 1999: Römische Keramik in der Schweiz, Basel (Antiqua 31). Schucany, C./M. Wullschleger, in press: Zwischen Aventicum und Augusta Raurica. Beidseits des Jura, der Kanton Solothurn in römischer Zeit, in C. Bélet-Gonda/J.-P. Mazimann/A. Richard/F. Schifferdecker (eds), Le peuplement de l’Arc jurassien de la Préhistoire au Moyen Âge. Journées archéologiques frontalières de l’Arc Jurassien. Actes 2007. Spycher, Hp., 1981: Archäologische Ausgrabungen auf dem Franziskanerhof, Bellacher-Kalender 1981, 107-109. Spycher, Hp., 1982: Ausgrabungen auf dem Franziskanerhof in Bellach, Bellacher-Kalender 1982, 120-122. Spycher, Hp., 1983: Funde aus der Römerzeit in Bellach. Abschluss der Ausgrabungen auf dem Franziskanerhof, Bellacher-Kalender 1983, 91-98. Walser, G., 1980: Römische Inschriften in der Schweiz für den Schulunterricht photographiert und erklärt,II. Teil: Nordwest- und Nordostschweiz, Bern.



Roman villa landscapes of the lignite mining areas in the hinterland of Cologne Wolfgang Gaitzsch 1 2 3 4 5

Introduction Settlement development and population density Economy Other research foci: wells and the Roman road network Conclusion

Abbreviations References Appendix 1

1 i n troductio n The Rheinische Bodendenkmalpflege (Rhineland Archaeological Heritage Service) has conducted a comprehensive investigation of numerous Roman rural settlements in three large opencast mining areas within the Jülich loess belt: the Hambach mine (HA), located 30 km west of Cologne, the Inden mine (WW) situated just to the south of the small Roman town of Juliacum (Jülich), and the Garzweiler mining area (FR) located on the Aachen-Neuss Roman road in the hinterland of the Rhine frontier (fig. 1). In terms of natural features, this is the core area of the Jülich loess belt in the southern part of the Lower Rhine basin. Conditions for archaeological surveying have been far from favourable.The loess layer on the Hambach plateau is to date less than a metre thick, with considerable losses to the archaeological record due to erosion; on the other hand in the Elsbach valley (Garzweiler mine), the colluvial topsoil is seven metres deep, rendering the archaeologically relevant layers inaccessible. Before mining operations began, the Hambach Forest was densely wooded, which further restricted surveying opportunities.The Hambach mine was opened up in 1978. Since that time, managing the archaeological operation, i.e. the day-to-day rescue work, has been the responsibility of an independent agency, the Titz outpost of the Rhineland Archaeological Heritage Service. The Hambach research area alone covers an area of approx. 100 km2­­.­ The varying scale and intensity with which the three opencast mining operations have proceeded in recent years has had a growing impact on the planning of Roman excavations. In 2003, for example, a second mining front was established in Hambach, affecting several settlements at one and the same time. A large villa at Hambach 488 was particularly endangered because of its location at the intersection of both mining fronts (fig. 1 and 3). The use of two large excavators at the first pit face (cf. fig. 2) accelerated the rate at which surface soil was lost and boosted the level of archaeological activity. This made excavation planning a highly complex affair, demanding a great deal in terms of organisation and personnel. The Inden mining operations reached the southern periphery of the small town of Juliacum (fig. 1 and 4). Several settlement sites were identified through systematic surveying, while others were discov-



ered in the village of Alt-Inden. These included WW 122,1 a large, prestigious villa adjacent to a Roman and Merovingian cemetery containing more than 800 burials, and a subterranean, 600 m-long water conduit using qanat construction methods.2 The Elsbach valley, which was opened up in Augustan times, disappeared from the map in the Garzweiler mining operations.3 Roman excavations account for 30 to 50% of the annual volume of excavations performed by the Titz agency. There has been no significant increase in the number of sites (about 150) known in the Hambach mining area in the early 1990s, although this number has increased by 10 to 20 % in the other two mining areas. There has been a marked improvement, however, in our detailed understanding of many sites as a consequence of repeated traditional surveys and modern scientific surveying methods, used primarily in the context of projects supported by the Lignite Foundation.4 The number of villas discovered in the Hambach settlement area through excavations has doubled since 1989.5 Excavations have identified a total of 42 settlement sites, 33 of which have been extensively investigated. The compounds of 18 villas have been completely or almost completely excavated. In addition, nine settlement sites have been largely investigated in the Inden mine and 18 in the Garzweiler mine. Five compounds have been identified in the Garzweiler mine, and six partially identified in Inden. A total of 69 rural settlements have been revealed through extensive excavations in the three settlement areas, as they are defined by the boundaries of the lignite mines.These excavations have given us a clearer picture of Roman settlement patterns as a whole and a more precise understanding of the early stages of development in particular. We now also have chronologically reliable information on Late Antiquity.6

2 settleme n t de v elo p me n t a n d p o p ulatio n de n sit y Settlement of the Hambach region began when the area was opened up by the Via Belgica, the Roman road linking Bavay with Cologne, and then developed southwards from the Augustan period on, as early graves close to the road and in later villas show. Examples are graves HA 260 and HA 503, belonging to small settlements on the Via Belgica, and HA 230 at Winterbach, where a system of field ditches has been identified. The extension of some individual compounds began in the Tiberian-Claudian period, initially with one or two-aisled wooden post-built structures on the sites of principal buildings erected on stone foundations after the middle of the 1st century. Villas HA 132, HA 512 and HA 59 exemplify this development. In the 2nd century we encounter a densely populated landscape dominated by small villa sites. The villa compounds range in size from one to more than six hectares (HA 132). The 18 fully investigated compounds occupy a total of about 30 ha of land. The extent of the associated farmland, until recently assumed to be 50 ha per villa, can now be confirmed on the basis of new excavations and the most recent calculations by Karl Peter Wendt of the University of Cologne.7 To determine population density, Thiessen polygons were created that took account of 28 settlement sites covering an area of 34 km2. This is about half of the Hambach study area, which corresponds in size to the Aldenhovener Platte in the Inden mine, which has been assessed by Karl Heinz Lenz.8 If all sites are taken into account, 1 

References to individual sites in each of the three mining areas have generally been kept to a minimum; for

gie im Rheinischen Braunkohlenrevier (Cologne, published since 1994).

literature on these sites, see the appendix at the end of

5 

this article.

6 

2 

In this construction method, vertical shafts are dug to construct a tunnel through which a water conduit runs; see Päffgen 2005.

3 

Becker 2005; Lochner 2006; Lochner in press; cf. also fig. 1.

4 

Cf. Geschäftsberichte der Stiftung zur Förderung der Archäolo-



For the previous status of research, see Gaitzsch 1991. For evaluative studies already published or in preparation, see fig. 1 (green number) with references in table 1 as well as Geschäftsberichte (note 4).

7 

Wendt 2008; now also Zimmermann et al. 2009.

8 

Lenz 1999a. A detailed analysis of a similar survey area was recently published: Wessel/Wohlfahrt 2008.

A

B 00/30

C 00/30

D

131

E 55/14

42

76/136

503

56

59 412 382 516

132 512

500

39

Fig. 1. Distribution of Roman rural settlements in the concession areas of three opencast mines in the hinterland of Cologne. FR: Frimmersdorf/Garzweiler; HA: Hambach; WW: Weisweiler/Inden. A Via Belgica and other Roman roads; B scientifically evaluated and published micro-regions (Aldenhovener Platte/Merzbachtal, Schlangengraben/Indetal, Elsbachtal); C sites that have been the subject of doctoral dissertations and Master’s theses; D boundary of extraction area in 2007; E concession boundary.

Hambach has less than one villa (0.8 villas) per 100 hectares of land, compared with 1.7 villas per km2 on the Aldenhovener Platte. If we consider only the ‘well-documented’ settlements (i.e. those that have been identified beyond doubt and excavated), this figure drops to an average of one to two villas per km2. The settlements were located on average 1 to 1.5 km apart (fig. 1 and 3). In some cases the distance is smaller, for example about 500 m between HA 127 and HA 523 and between HA 130 and HA 431.



Fig. 2. Mining front of the Hambach concession with archaeologists investigating a Roman well (HA 127) just in front of large excavator in action. Photograph: H. Haarich, LVR-ABR.

Villas WW 122,WW 123 and WW 126, in exposed locations on the steep western bank of the Inde river, are characterised by even smaller distances (fig. 4): these three principal buildings are only 1.2 km apart. This means that the population density was three to four times higher than, say, that of the NetherlandsLimburg area, where 150 to 200 ha of land are estimated per compound.9 Larger properties in the Inde valley contrast with the small villas in the Hambach forest, which are primarily agrarian in nature. One example is WW 122, the large, prestigious villa mentioned above, which is 150 m long and incorporates both a residential and farming section on a 260 metre-wide area of land. Coin finds and three predecessor buildings show that settlement began here in the early 1st century, with the two-phase bathhouse originating from the end of that century. WW 123 and 126, two smaller flanking settlements, seem to have been subordinated to the central villa. Habitation continued into the 4th century, during which time a burgus was located in villa WW 126 (fig. 4). A Merovingian settlement followed in the 5th century, attested to by an associated cemetery with over 700 burials. A destruction layer in a second burgus in villa WW 112 testifies to habitation ending in about 355. Here there was a military link to the Late Roman fort in the small town of Jülich just 3 km downstream. Settlement development on the Inde shows parallels with the Elsbach valley in the Garzweiler mine area. A paved road was built in the riverside area at around the turn of the millennium. A villa analogous to the known La Tène settlement at Hambach 382 was erected in the immediate vicinity of a fortified Late La Tène settlement at Jülich-Bourheim (WW 94/376), without there being any direct continuity of habitation. Characteristic of early structures in the three mining areas are house types belonging to the Iron Age building tradition and attributed to native or migrant settlers (either northern Gallic or Elbe Germanic

9 

Jeneson, this volume.



Fig. 3. Southern part of the opencast

A

mining area of Hambach.

B

A excavated rural settlements; B surveyed

C

sites; C cemeteries and grave monu-

D

ments; D single burials; E Roman roads

E

(1-3).

influences, depending on one’s interpretation of the data).10 These were so-called ‘protovillas’ constructed in the Claudian era.11 At present, four such settlements are known in the Garzweiler mine (FR 49, 80, 129, 131), with Pulheim-Brauweiler and Jüchen-Neuholz located outside the study area.12 As in Pulheim and Kerpen-Sindorf, the period of habitation continued into the 2nd and 3rd centuries and did not involve stone conversion. Covering less than one hectare, these compounds were clearly smaller than their average Hambach counterparts. The buildings were either one-aisled post-built structures, with the posts positioned in front of the long or narrow sides to form trapezoidal entrances, as in HA 59 and FR 129, or houses with wall ditches, like HA 412 and Kerpen-Sindorf,13 also from the first half of the century. There were also post-built structures with earth cellars (HA 382 and Jüchen-Neuholz). The population of the Hambach area can be estimated at 1,000 to 2,000 people, depending on whether we assume ten, 15 or more people per settlement site. Wendt estimates 16 to 24 inhabitants per 1 km2.14 The considerably lower figure of four people per km2 was calculated on the basis of preserved grave finds, but we should assume a much larger number of non-preserved burials – five to six times as many, if we assume 15 people per villa. HA 132 leads the way with 81 confirmed burials (including Late Roman ones), followed by HA 127 with 52 preserved graves, predominantly from the middle Roman 10

Heimberg 2002/2003, 64 ff.; Frank/Keller 2007; Kießling 2008.

11 

On the concept of ‘proto-villa’, Slofstra 1991, esp. 163f,

rural settlements as ‘enclosed non-villa settlements’. 12 

Pulheim-Brauweiler: Andrikopoulou-Strack et al. 2000; Jüchen-Neuholz: Frank/Keller 2007.

177, 184; Heimberg 2002/2003, 64 ff.; also Frank/Keller

13 

2007. Habermehl (this volume) describes this category of

14 

Aeissen/Görür 2002. Wendt 2008; Wendt 2010.



Fig. 4. Rural settlements to the south

+

A

E

of the small town of Juliacum.

B

F

A investigated villas; B surveyed sites;

C

G

C excavated sites; D brick kilns; E

D

H

Roman water conduits; F burgus; G Late Iron Age fortification; H surveyed area 1991-1996.

imperial period. The two-hectare settlement of HA 382 shows a cross section: 23 graves, from the mid1st century to the early 3rd century, as well as two inhumation burials, the latest of which dates to the beginning of the 5th century. If we consider the number of burials in conjunction with the duration of habitation, the burial data is clearly not demographically representative.

3 eco n om y There are no definite findings on whether the size of the settlement compounds is proportionate to that of the surrounding land, or whether the level of economic activity determined the size of the compounds. The average compound size is 1.8 ha per villa. The relationship between agriculture and animal husbandry is reflected in the types of construction. Individual studies have contributed to our understanding of the function of the farm buildings. Larger compounds such as HA 512 more frequently featured stables, large hall-like post-built structures or wooden buildings on stone foundations, more than 20 m long, or – as in the case of HA 127 and HA 131 – with internal stalls for livestock. Given the estimated number of inhabitants, if about two-thirds of the 50 ha was used as arable land, it probably produced more than twice as much grain as was required. Ursula Heimberg has arrived at these figures on the basis of the storage capacities of the Voerendaal villa.15 Assuming a harvest of two m3 per hectare, the granaries of HA 206 and HA 127 had an estimated storage capacity of about 140 m3 and the 15 

Heimberg 2002/2003, 128 f. esp. note 131.



large horreum of WW 112 about twice this quantity. Even with smaller harvests, the Hambach region would have been able to produce enough grain annually to feed 6,000 to 7,000 people. On the immediate periphery of the settlement compounds themselves, the Hambach excavations have revealed enclosed plots of 1 to 1.5 ha. These are either square or rectangular in shape, such as HA 412, or, as in the case of the new villa HA 488, have a plot on the northern side, as well as a series of associated field ditches south of the compound. Here, and to a greater extent at HA 127, subdivisions of the surrounding farmland have been identified. These are documented as being 200 m long in the case of HA 59 and up to 600 m long in the case of HA 132.

4 other research f oci : w ells a n d the roma n road n et w or k The investigation of Roman wells has become increasingly important, not just for settlement chronology and archaeological finds, but also for palaeobotany, microfauna and zoology. In addition to the detailed knowledge of timber construction in antiquity, the dating of construction timbers in particular has produced an unparalleled and remarkably dense and reliable data set (fig. 5), which has enabled us to investigate periods of habitation lasting on average 150 to 200 years. Site

Date +/-5

Site

Date +/-5

FR 110

49 BC

FR 125

130

FR 100

13 BC

FR 131

130

FR 109

2

FR 00/30

130/140

NW 95/1000

2

HA 512

131

HA 59

4/40

HA 59

139

FR 107

10

FR 116

149

NW 98/15

20/30

HA 500

155

FR 49

30-60

Neu-Inden/Altdorf

159

FR 100

32

WW 88/76

159

FR 120

34

NW 90/51

175

HA 125

46

FR 07/5

177

HA 512

50

FR 143

210/220

FR 127

58

FR 109

214

FR 115

60

FR 110

214

HA 382

60

WW 99/150

214

HA 53

65

HA 224

219

NW 94/1005

75

FR 49

220-250

WW 87/65

78

FR 143

223/233

FR 93/178

95

HA 425

236

FR 131 (99/32)

96

HA 512

238

FR 78

99

HA 382

241

FR 44

102

HA 59

320

FR 81

102

HA 500

393

FR 49

130

WW 38

400

Table 1. Dendrochronological datings of Roman wells in the Rhineland lignite mining areas (cf. also fig. 5).

A total of 152 wells have been identified in the mining areas (as of 2008), distributed across 61 settlement sites: 28 in HA, 17 in WW and 16 in FR. This comes to an average of three to four wells per settlement. Heading the list are two villas (HA 132 and HA 382), each with seven wells. The wells were between 20 and more than 30 m (WW 94/376) deep!



Fig. 5. Distribution of excavated and dendrochronologically dated Roman wells (dots), construction timber and wooden water conduits (squares) in the concession areas of the three opencast mines in the hinterland of Cologne. FR: Frimmersdorf/Garzweiler; HA: Hambach; WW: Weisweiler/Inden.

If we include construction timbers from the Elsbach valley, we currently have more than forty datings, thanks to the work of the Cologne Laboratory of Dendroarchaeology under the direction of Burghart Schmidt (table 1). The figures document a continuous settlement development, beginning as early as the Augustan era in terms of road construction and in the Tiberian-Claudian period in terms of settlement, and persisting into the third quarter of the 2nd century. The subsequent 40-year gap in construction can be attributed to a greatly reduced tree population as a result of tree felling and climate changes. Wells and water conduits were once again being built in the first half of the 3rd century. In one exceptional instance there was a hiatus of 150 years, reflecting the decline and abandonment of the majority of settlements.16 The latest wells were built at the end of the 4th century in the Hambach 500 glassworks and in the Matronae sanctuary of Eschweiler-Lohn (WW 38). In the 4th century only a few settlements were once again in use. A change in function is evident, with crafts replacing or supplementing agricultural activity. Glass was manufactured on a large scale, followed by metal recycling. So far eight glassworks have been identified and investigated. With its two workshops, Hambach 132 occupied a key position: it is there that the Late Antique cemetery with its associated grave good products was found. Production began at the Hambach glassworks in the mid-4th century and, as the latest burials in Hambach 132 show, probably continued into the mid-5th century. The glassworks did not manufacture their products in isolation from each other.There are commonalities in both the chemical composition of the glass and in the manufacture of particular types of vessel, which suggests organisational and economic links.17

16 

The vegetation images compiled by the Archaeobotany Laboratory at the University of Cologne are based on well finds from Hambach and a comprehensive analysis by Wolf-Dieter Becker of sediments from waterlogged



sites in the Elsbach valley. Becker 2005; Hallmann-Preuß 2002/2003; Kaszab-Olschewski 2006. 17 

Gaitzsch et al. 2000, and now Brüggler 2009.

Fig. 6. Part of the Via Belgica between Colonia Claudia Ara Agrippinensium (Cologne) and the Vicus Juliacum (Jülich) with the reconstructed sites of Roman league stones in and just outside the opencast mine of Hambach. Arrow: archaeologically documented in situ find.

The Roman road network is the second research focus in the lignite opencast mining areas (fig. 1, 3 and 6).18 The special circumstances of comprehensive, large-scale research in the opencast lignite mines enable us to focus on two questions about landscape organisation which would be very difficult to answer elsewhere. The first concerns the extent to which the Roman road network was a continuation of older, late prehistoric roads and the second concerns the nature of the network of secondary roads. Nine km of the Via Belgica, the main Roman road between Cologne, Heerlen and Tongres, ended up as part of the Hambach mine – this amounts to about one sixth of the German section of the road. Countless sections of road have been documented. No older roadways have been discovered beneath the Roman road, but instead there have in some cases been discoveries of post-built structures that can be attributed to the Late La Tène settlement layer of HA 382 – hence after LT D1. The in situ find of a league stone has contributed to the ancient topography and survey of the road. Together with itineraries, this has enabled a reconstruction of the locations of league stones as far as Cologne (fig. 6). In addition to the important Via Belgica, several secondary roads helped to open up the Hambach settlement area (fig. 1-3). One road measuring more than 7 km in length and running in a northeasterly direction towards Thorr (Tiberiacum) (fig. 3.1) has been systematically surveyed and verified through excavations in the forest and on the edges of the mining area. Grave finds point to an early dating. A second, more or less parallel road (fig. 3.2) was only recently documented in greater detail following the discovery of a crossroads during mining operations. A third road (fig. 3.3) ran from the southeast in a northerly direction. Paved and unpaved farm tracks such as the 100 metre-long main thoroughfare in HA 127 were situated in and near several settlements.



co n clusio n The resulting settlement picture is representative of the settlement landscape in the western foreland of the CCAA to the west of the river Erft.19 The landscape is dominated by small agricultural farms that can be regarded as small villas. Larger villa complexes such as WW 122 are the exception. Continuity of settlement from the Late La Tène onwards cannot be established. The villas were newly established along uniform lines in connection with the development and establishment of the CCAA. 18 

the urban area and villa, see Spiegel 2002, esp. 719 ff., fig.

19 

16; Dodt 2006.

Gaitzsch 2004, 2008. For the current status of research on the apparently less densely populated settlement of the ager coloniae between



Whereas the Hambach settlement area appears to be chronologically and structurally homogeneous, the study areas further to the north and west differ clearly from each other with regard to the older settlement stages and the Late Antique transitional horizon. In addition to current excavations in the Inde valley (WW 128), research in Garzweiler II is yielding promising findings for the Late La Tène and early Roman development of the settlements. These were located close to the Rhine frontier and the military camp in Neuss. In the Inden study area, the Late Roman horizon has been identified in villas WW 112 and WW 117 and a Merovingian connection has been found in the cemetery of Alt-Inden (WW 04/61). More detailed findings are expected from the forthcoming mining operations in the village of Pier, which has a Matronae sanctuary20 adjoining it to the east.

a b b re v iatio n s AK AR BJ FMRD RA

Archäologisches Korrespondenzblatt Archäologie im Rheinland Bonner Jahrbücher Die Fundmünzen der römischen Zeit in Deutschland Rheinische Ausgrabungen

re f ere n ces Becker, W.D., 2005: Das Elsbachtal. Die Landschaftsgeschichte vom Endneolithikum bis ins Hochmittelalter, Mainz (RA 56). Brüggler, M., 2009: Villa rustica, Glashütte und Gräberfeld. Die kaiserzeitliche und spätantike Siedlungsstelle HA 132 im Hambacher Forst, Mainz (RA 63). Brüggler, M., 2010: HA 132 – Villa rustica und spätantike Glashütte, in J. Kunow (ed.), Braunkohlenarchäologie im Rheinland. Entwicklung von Kultur, Umwelt und Landschaft, Weilerswist (Materialien zur Bodendenkmalpflege im Rheinland 21), 91-94. Dodt, M., 2006: Bäder römischer Villen in Niedergermanien im Lichte neuer Ausgrabungen im rheinischen Braunkohlenrevier, BJ 206, 63-85. Frank, K./C. Keller, 2007: Jüchen-Neuholz. Vom eisenzeitlichen Gehöft zur Villa rustica, in G. Uelsberg (ed.), Krieg und Frieden. Kelten, Römer, Germanen, Bonn, 316-324. Franke, T., 1999: Ein Matronenheiligtum in Inden-Pier, Kreis Düren, BJ 199, 117-140. Gaitzsch, W., 1983: Römische Siedlungsplätze im Verlauf der antiken Straße von Köln nach Jülich, in G. Bauchhenss (ed.), Archäologie in den Rheinischen Lößbörden, Cologne/Bonn (RA 23), 347-362. Gaitzsch,W., 1991: Römische Hof- und Wirtschaftsräume im Rheinischen Braunkohleabbau am Beispiel des Hambacher Forstes, in B. Gramsch/G. Wetzel (eds), Archäologische Erkundung und Rettungsarbeit in Tagebaugebieten Mitteleuropas, Berlin (Veröffentlichungen des Museums für Ur- und Frühgeschichte Potsdam 25), 125-134. Gaitzsch, W., 1993: Brand- und Körpergräber in römischen Landsiedlungen der Jülicher Lößbörde, in M. Struck (ed.), Römerzeitliche Gräber als Quellen zu Religion, Bevölkerungsstruktur und Sozialgeschichte, Mainz, 17-39. Gaitzsch, W., 2002: Römische Landbesiedlung. Hof- und Wirtschaftsräume, in W. Menghin/D. Planck (eds), Menschen, Zeiten, Räume. Archäologie in Deutschland, Berlin, 267-273.

20 

Franke 1999.



Gaitzsch, W., 2004: Die Via Belgica zwischen Elsdorf und Jülich, in H. Koschik (ed.), “Alle Wege führen nach Rom...”. Internationales Römerstraßenkolloquium Bonn 2003, Pulheim (Materialien zur Bodendenkmalpflege im Rheinland 16), 175-196. Gaitzsch, W., 2008: Von der Via Belgica zur Sophienhöhe. Straßenforschung im Tagebau Hambach, in J. Kunow (ed.), Erlebnisraum Römerstraße. Via Belgica, Aachen (Materialien zur Bodendenkmalpflege im Rheinland 18/2), 31-43. Gaitzsch, W./A.-B. Follmann-Schulz/K.H. Wedepohl/G Hartmann/U. Tegtmeier, 2000: Spätrömische Glashütten im Hambacher Forst, BJ 200, 83-241. Gottschalk, R., 2003: Studien zu spätrömischen Grabfunden in der südlichen Niederrheinischen Bucht, Bonn (Diss. Bonn). Trade publication in press Hallmann, B., 2001: Zur Besiedlung von Schlangengraben und Indetal in römischer Zeit, in H. Koschik (ed.), Archäologische Talauenforschungen, Mainz (RA 52), 228-259. Hallmann-Preuß, B., 2002/2003: Die Villa rustica Hambach 59. Eine Grabung im Rheinischen Braunkohlenrevier, Saalburg Jahrbuch 52/53, 283-535. Heege, A., 1997: Hambach 500. Villa rustica und früh- bis hochmittelalterliche Siedlung Wüstweiler (Gem. Niederzier), Kreis Düren, Cologne/Bonn (RA 41) Heimberg, U., 2002/2003: Römische Villen an Rhein und Maas, BJ 202/203, 57-148. Hinz, H., 1969: Die Ausgrabung auf dem Kirchberg in Morken (Erft). Von der Steinzeit bis ins Mittelalter, Düsseldorf (RA 7). Kaszab-Olschewski,T., 2006: Siedlungsgenese im Bereich des Hambacher Forstes 1.- 4. Jh. n. Chr. Hambach 512 und Hambach 516, Oxford (BAR Int. Ser. 1585). Kaszab-Olschewski, 2010: Die Wirtschaftsbeziehungen einer Villa rustica (HA 512) im Spiegel ihrer Funde, in J. Kunow (ed.), Braunkohlenarchäologie im Rheinland. Entwicklung von Kultur, Umwelt und Landschaft, Weilerswist (Materialien zur Bodendenkmalpflege im Rheinland 21), 95-98. Kießling, P., 2008: Die Villa rustica HA 412 und das römische Gräberfeld HA 86/158, Bonn (PhD thesis, Rheinische Friedrich-Wilhelms-Universität Bonn), available online at Deutsche Nationalbibliothek (urn: nbn:de:hbz:5-13546) or http://hss.ulb.uni-bonn.de:90/2008/1354/1354-1.pdf Köhler, B., 2006: Villa rustica Frimmersdorf 49 und Villa rustica Frimmersdorf 131. Studien zur römischen Besiedlung im Braunkohlentagebaugebiet Garzweiler I. (PhD thesis, University of Cologne) available online at Deutsche Nationalbibliothek (urn:nbn:de:hbz:38-16888) at http://kups.ub.uni-koeln.de/ volltexte/2006/1688/ Kunow, J., 1994: Die ländliche Besiedlung im südlichen Teil von Niedergermanien, in: H. Bender- H. Wolff (ed.), Ländliche Besiedlung und Landwirtschaft in den Rhein-Donau-Provinzen, Espelkamp (Passauer Universitätsschriften zur Archäologie 2), 53-72, 141-197. Lenz, K.H., 1999a: Siedlungen der römischen Kaiserzeit auf der Aldenhovener Platte, Cologne (RA 45). Lenz, K.H., 1999b: Die ländliche Besiedlung der frühen und mittleren Kaiserzeit im Hinterland des römischen Köln, KJ 32, 807-822. Lochner, I., 2007: Germanische Einwanderer am südlichen Niederrhein, in G. Uelsberg (ed.), Krieg und Frieden. Kelten, Römer, Germanen, Bonn, 325-327. Lochner, I., 2010: Eine Tallandschaft im Wandel. Folgen römerzeitlicher Wasserwirtschaft im Elsbachtal, in J. Kunow (ed.), Braunkohlenarchäologie im Rheinland. Entwicklung von Kultur, Umwelt und Landschaft, Weilerswist (Materialien zur Bodendenkmalpflege im Rheinland 21), 167-169. Lochner, I., in press: Das Elsbachtal während der römischen Kaiserzeit. Landschaftsarchäologische Untersuchungen einer ländlich geprägten Region in den rheinischen Lößbörden (Dissertation, in press) Päffgen, B., 2005: Die Quanat-Wasserleitung der villa rustica in Alt-Inden, AR 2005, 89-90. Rothenhöfer, P., 2005: Die Wirtschaftsstrukturen im südlichen Niedergermanien. Untersuchungen zur Entwicklung eines Wirtschaftsraumes an der Peripherie des Imperium Romanum, Cologne (Kölner Studien zur Archäologie der römischen Provinzen 7).



Slofstra, J., 1991: Changing settlement systems in the Meuse-Demer-Scheldt area during the Early Roman period, in N. Roymans/F. Theuws (eds), Images of the past. Studies on ancient societies in Northwestern Europe, Amsterdam (Studies in prae- en protohistorie 7), 131-199. Spiegel, E.M., 2002: Ausgrabungen in einem römischen Siedlungsplatz mit zwei spätantiken Burgi in Köln-Widdersdorf, Kölner Jahrbuch 35, 699-782. Wagner, G., 2006: Die Gräber der römerzeitlichen Villa Hambach 382, Cologne (MA thesis; PhD thesis in prep.) Wendt, K.P., 2008: Bevölkerungsdichte und Landnutzung in den germanischen Provinzen des Römischen Reiches im 2. Jahrhundert n. Chr., Germania 86, 191-226. Wendt, K.P./A. Zimmermann, 2010: Zur Bevölkerungsdichte in den römischen Rheinlanden, in J. Kunow (ed.), Braunkohlenarchäologie im Rheinland. Entwicklung von Kultur, Umwelt und Landschaft, Weilerswist (Materialien zur Bodendenkmalpflege im Rheinland 21), 87-90. Wessel, I./C. Wohlfahrt, 2008: Archäologische Forschungen auf der Rheinbacher Lößplatte, Mainz (RA 62). Zimmermann, A./K.P. Wendt/T. Frank/J. Hilpert, 2009: Landscape archaeology in Central Europe, Proceedings of the Prehistoric Society 75, 1-53.



a p p e n di x 1 : roma n rural sites i n the three lig n ite e x tractio n areas i n the hi n terla n d o f colog n e FR: Frimmersdorf/Garzweiler; HA: Hambach; WW: Weisweiler/Inden Site

References Hambach mine (HA)

HA 10

Gaitzsch, AK 18, 1988, 373-387

HA 21

Gaitzsch 1983, 358

HA 22

ibid.

HA 23

Heimberg 2002/2003, 135

HA 26

Gaitzsch 1983, 358

HA 31

BJ 188, 1988, 422

HA 34

Gaitzsch 1993, 26; Gaitzsch, in J. Kunow (ed.), Materialien zur Bodendenkmalpflege im Rheinland 18/1 (2008) 31 ff. Abb. 9

HA 47

Gaitzsch 1983, 349; Heimberg 2002/2003, 135

HA 50

W. Linden, Römisches aus der Bürgermeisterei Rödingen. Verwaltungsbericht 1926, 54 no. 24

HA 53

ibid., no. 21

HA 56

A. Schmidt, Hambach 56. Eine römerzeitliche Villa mit spätantiker, lokaler Münzprägestätte (MA-thesis University of Bonn 2002); Heimberg 2002/2003, 135

HA 59

Rothenhöfer 2005, 45 ff.; Hallmann-Preuß 2006

HA 66

Heimberg 2002/2003, 135

HA 68

Dohrn-Ihmig 1983, in G. Bauchhenss (ed.), Archäologie in den rheinischen Lössbörden, Cologne (RA 24), 229-231

HA 69

Kunow 1994, 166 f.; Rothenhöfer 2005, 45 ff.; Heimberg 2002/2003, 135

HA 74

Heimberg 2002/2003, 136

HA 75

Gaitzsch et al. 2000

HA 101

Gaitzsch, AR 1995, 76-79; BJ 198, 1998, 417 f.

HA 104

Gaitzsch 2004

HA 103

BJ 199, 1999, 440-442

HA 111

Gaitzsch et al. 2000; Heimberg 2002/2003, 136

HA 125

BJ 198, 1998, 404 f.

HA 127

Heimberg 2002/2003, 136; BJ 204, 2004, 324; BJ 205, 2005, 309-311

HA 130

Gaitzsch/Haarich, AR 2004, 81 f.; Heimberg 2002/2003, 136

HA 131

Gaitzsch 2004; id. AR 2003, 139-141

HA 132

Gaitzsch et al. 2000; Rothenhöfer 2005, 164-170; Brüggler 2009, 2010

HA 133

H. Hinz, Kreis Bergheim (1969) 240 no. 2; AR 2007, 111 ff.

HA 162

BJ 183, 1983, 644 f.

HA 206

BJ 194, 1994, 425; Heimberg 2002/2003, 136

HA 224

KJ 33, 2000, 131-142; Heimberg 2002/2003, 136

HA 230

Gaitzsch 1993, 31

HA 235

Gaitzsch 1983, 356

HA 240

Gaitzsch 1983, 358 f.

HA 252

Heimberg 2002/2003, 136

HA 260

Dohrn-Ihmig, Ausgrabungen im Rheinland `79/80, 1981, 119-121; Gaitzsch 1983, 360

HA 264

Heimberg 2002/2003, 136

HA 303

BJ 186, 1986, 635 f.; Heimberg 2002/2003, 136

HA 382

Heimberg 2002/2003, 137; Wagner 2006



Site

References

HA 403

Heimberg 2002/2003, 137

HA 412

Heimberg 2002/2003, 137; Kießling 2008; AR 2007, 69 ff.

HA 415

Gaitzsch 1993, 33

HA 417

Gaitzsch 2004

HA 425

Heimberg 2002/2003, 137

HA 427

Gaitzsch 2004

HA 488

Gaitzsch/Haarich, AR 2004, 81 f.; Gaitzsch/Weiner, AR 2006, 151 f.; Gaitzsch/Haarich/Janssens, BJ 207, 2007, 325 f.

HA 493

Schwellnus/Hilbert, Ausgr. im Rheinland ‘78 (1979) 46 f.

HA 500

Heege 1997

HA 503

Heimberg 2002/2003, 137; Lochner 2007

HA 512

Kaszab-Olschewski 2006, 2010

HA 516

ibid.

HA 77/213

Gaitzsch 2004

HA 89/232

ibid.

HA 91/26

ibid.

HA 93/328

Gaitzsch/Hermanns, BJ 195, 1995, 513-515; Gaitzsch, in J. Kunow (ed.), Materialien zur Bodendenkmalpflege im Rheinland 18/1 (2008) 31 ff. Abb. 9

HA 95/130

Gaitzsch 2004

HA 95/146

ibid.

HA 95/267

ibid.

HA 97/193

ibid.

HA 02/203

ibid.

HA 02/210

Gaitzsch/Janssens, BJ 205, 2005, 311-314

HA 08/16

Gaitzsch/Janssens, AR 2007, 111 ff.; AR 2008, 111-114 Frimmersdorf/Garzweiler mine (FR)

FR 42

H. Hatch, 1992: Frimmersdorf 42. Eine römische Villenanlage in der Jülicher Börde (MA-thesis University of Cologne)

FR 49

Köhler 2006

FR 75

Gaitzsch, BJ 187, 1987, 580

FR 78

ibid. 586-588

FR 80

Gaitzsch, BJ 188, 1988, 406-408

FR 100

Lochner

FR 107

Arora/Schmidt, AR 1996, 55 f.

FR 109

Lochner

FR 110

Lochner

FR 115

Arora/Lürken, AR 1993, 51-53; Arora/Schmidt, 1996, 55 f.

FR 116

Lochner

FR 120

Lochner

FR 125

Arora/Schmidt, AR 1996, 55-58; Lochner

FR 127

ibid.

FR 129

Arora, AR 1997, 53-55; Heimberg 2002/2003, 138; Lochner 2010; Lochner, in press

FR 131

Heimberg 2002/2003, 138; Köhler 2006

FR 55/14

Hinz 1969, 21-62; Heimberg 2002/2003, 142

FR 76/136

Rech, Ausgrabungen im Rheinland `76, 1977, 106-108

FR 98/22

Arora/Lochner, AR 1999, 97 f.

FR 00/30

J. Englert, Frimmersdorf 2000/30. Drei römische Brunnen aus dem Braunkohletagebau Garzweiler (MA-thesis University of Cologne 2003) Geilenbrügge, Arch. in Deutschland 3, 2006, 49

FR 02/126



Site

References

FR 02/131

Arora, AR 2002, 111 f.

FR 07/5

Schuler, AR 2007, 96-98 Weisweiler/Inden mine (WW)

WW 2

Lenz 1999a, 205 no. 69

WW 32

Lenz 1999a, 194 no. 67

WW 37

Lenz 1999a, 200 no. 68; Lenz 1999b; Heimberg 2002/2003, 138

WW 39

Gottschalk 2003

WW 46

Bentz et al. BJ 184, 1984, 606

WW 79

Lenz 1999a, 154 no. 38

WW 80

Lenz 1999a, 156 no. 40; Heimberg 2002/2003, 138

WW 109

Päffgen/Schürmann, BJ 197, 1997, 291

WW 112

Päffgen/Willer, Von Anfang an. Arch. in NRW. Ausstellungs-Kat. (Cologne 2005), 465-471; Heimberg 2002/2003, 138

WW 117

Hallmann 2001, 238 ff.; FMRD VI 2 (in press)

WW 118

Päffgen, Von Anfang an. Arch. in NRW. Ausstellungs-Kat. (Cologne 2005), 427-430; Dodt-Päffgen, AR 2004, 98-101

WW 120

FMRD VI 2 (in press)

WW 122

Dodt/Päffgen, AR 2005, 86-88; Päffgen ibid. 89 f.

WW 50/82

BJ 148, 1948, 384

WW 69/82

W. Piepers, Ausgrabungen an der alten Burg Lürken, RA 21 (1981); Lenz 1999a, 176 no. 63; Lenz 1999b

WW 70/47

Lenz 1999a, 126 no. 23

WW 72/50

Lenz 1999a, 143 no. 32

WW 72/54

Lenz 1999a, 147 no. 35; Lenz 1999b

WW 75/67

Lenz 1999a, 153 no. 36

WW 79/23

Lenz 1999a, 153 no. 37

WW 81/45

Lenz 1999a, 205 no. 69

WW 83/94

Lenz 1999a, 173 no. 58

WW 87/65

BJ 189, 1989, 225 ff.; Lenz 1999a, 167 no. 43

WW 87/80

Lenz 1999a, 155 no. 39

WW 88/76

BJ 189, 1989, 225 ff.; Lenz 1999a, 160 no. 42

WW 92/235

Hallmann 2001, 241

WW 93/9

Hallmann 2001, 232 f.

WW 93/130

Heimberg 2002/2003, 138

WW 94/141

Hallmann 2001, 242 f.

WW 94/356

Gaitzsch/Päffgen, BJ 196, 1996, 582

WW 94/376

Päffgen, AR 1995, 47-50; Böwing/Päffgen, ibid. 65-67

WW 98/137

Päffgen, AR 1998, 88

WW 99/146

Päffgen/Schmitz, AR 2002, 103 f.

WW 99/150

Päffgen/Schürmann, BJ 201, 2001, 418 f.



The Roman villa complex of Reinheim, Germany Florian Sărăţeanu-Müller 1 2 3 4 5 6

Introduction The layout of the villa complex The chronological development of the residential area The working part: general layout The working part: detailed description of some annexes Conclusions References

1 i n troductio n The large villa complex of Reinheim is situated on the German side of the German-French border between the villages known today as Reinheim (Saarland), in Germany, and Bliesbruck (Moselle), in France. In ancient geographical terms, the villa is situated about 75 km to the east of the caput civitatis of the Mediomatrici (Metz), and just a few hundred yards to the north of the small town (vicus) of Bliesbruck (fig. 1). The most prominent feature of the countryside around the villa is the small stream named the Blies, which runs through rolling hills rising no higher than 400 m above sea level. The presence of Triassic limestone has produced highly fertile farmland which must have played an important role in the lives of the rural population in the Roman period. Almost 100 archaeological sites from the Roman period have been mapped within a radius of approximately 12 km around the villa of Reinheim. Systematic archaeological excavations have been carried out at only a few of these sites – on the German side the villae rusticae of Bliesdahlheim, Böckweiler, Erfweiler-Ehlingen and Wittersheim and on the French side the villae rusticae of Großwald and Heidenkopf. Scheduled archaeological excavations to examine the vicus on the northern outskirts of the township of Bliesbruck began at the end of the 1970s.1 In the course of several excavations an artisan and merchant district, a market place, public baths and numerous pits and shafts were uncovered and documented. Wells and remains of wooden aqueducts indicate how the vicus was supplied with water. Several graves that were documented to the north and south of the vicus prove the existence of two cemeteries. The excavated rows of houses were all built from limestone and as a rule had cellars and one or more rooms with hypocaust heating systems. A courtyard served as a workshop. Judging by the finds, the vicus seems to have been inhabited from the middle of the 1st century to the first half of the 5th century. The settlement attained its largest size around the middle of the 3rd century. In the second half of the 3rd century the vicus fell victim to Germanic invasions. Although the vicus recovered economically in the first half of the 4th century, albeit with a reduced populated area, it was destroyed by fire around the middle of the

1

Cf. Petit et al. 2005.



B

lie

s

0

200 m

Fig. 1. Map of the villa complex of Reinheim, the vicus of Bliesbruck and the cemetery (in grey) in between.

4th century, probably due to disturbances brought on by the civil war. Judging by the findings, the end of Roman life in the vicus of Bliesbruck dates back to the first third of the 5th century.2

2

Petit 2000, 446.



Fig. 2. Aerial photograph of the villa complex of Reinheim viewed from the northeast. Part of the courtyard with the annexes B12 and B13 was lost through gravel extraction, which resulted in the pond visible in the centre of the photograph. Beyond the pond, the excavated features of the annexes B11, B9 and B8 are visible, whereas in the centre of the courtyard’s short south side the rebuilt gatehouse is recognisable; on the western long side the annex B1 has been rebuilt while features of the annexes B2-B6 are being excavated. The three tumuli in the forefront are modern reconstructions of the Early La Tène mounds which were excavated in the early 1950s some 100 m to the east.

2 the lay out o f the v illa com p le x The first systematic archaeological examinations in the vicinity of the Roman villa began in the summer of 19873 and were carried out, first of all, on the slight hill known to the local inhabitants as Heidenhübel (‘pagan hill’).4 The remains of an impressive palace-like building were brought to light. These remains turned out to be the main building of the villa complex. South of the main building is a walled courtyard almost 300 m in length and 135 m wide (cf. fig. 6).5 The archaeological remains of the villa complex suffered immense damage up to the beginning of the 1980s due not only to the extensive agricultural use of the land, but also to gravel quarrying. The transition zone between the main residence and the courtyard was especially affected. Only the remains of a dividing wall (EW) detected near the banks of the modern

3

In 1987, an archaeological park (Europäischer Kulturpark/

4

The name Heidenhübel probably goes back to the numer-

Parc archéologique européen Bliesbruck-Reinheim) was

ous wall remains and fragments of roof tiles that came to

set up between the two villages as a German-French col-

light every time the land was ploughed. Excavations were

laboration. This project was prompted by the discovery of a burial chamber in Reinheim dating back to the Early La

carried out here as early as 1840-1841 and 1870. 5

Cf. Miron et al. 1993; Sărăţeanu-Müller 2000.The building

Tène period and the discovery of various building struc-

complex on the Heidenhübel is referred to as area A (pars

tures from the Roman period.

urbana), and the terrain south of it as area B (pars rustica).



gravel pond point to a division between the two parts of the villa (pars urbana and pars rustica). On the north side, almost a quarter of the courtyard was lost through gravel quarrying (fig. 2). The main building extends 80 m from east to west and more than 60 m from north to south (cf. fig. 6). It consists of a rectangular central part with two wings on either side that form two protruding risalits. On the northern side, in the alignment of the main building’s central axis, approximately 10 m away from the façade, a huge plinth measuring 3 x 3 m was found – this had most probably been part of a monumental structure. The remains of a wooden well were recorded approximately 8.5 m north of the northeast risalit (cf. fig. 3). The outline of the well is square with sides of approximately 1 m in length. It consisted of stacked elements each made of four planks mortised at the corners. Dendrochronological analysis of the wooden planks dates the building of the well to around 90 AD. 6 The rectangular courtyard could be accessed via a gatehouse on the short, northwest side. On the outside, six single buildings stood on either side of the long sides of the enclosure walls. The walls between the annexes had been severely damaged by ploughing and most of them had been taken down to the foundation level by modern farmers. By means of geophysical measurements and traditional archaeological surveys, the courtyard was recorded as having had an area of 40,000 m². However, no wall structures or stone plastering could be detected. Numerous pits ranged in a row were documented parallel to and about 50 m north of the southern wall. No finds were made in the infill of the pits. We are probably dealing here with planting pits of a kind of ‘green wall’ intended to mark off a certain area.7 The whole complex formed the centre of an estate belonging to a lord. The main building served as the owner’s living quarters, whereas the annexes served as utility buildings, with the odd one also being used as living quarters.8

3 the chro n ological de v elo p me n t o f the reside n tial area The grand villa of Reinheim was built in the second half of the 1st century AD. The additions to the main building (including the porticos, the hypocaust heating system, bathroom and cellar) were added in stages with plan amendments that can be broken down into four main settlement periods.9 After several building projects in the second half of the 1st century, the main building acquired an H-shaped floor plan with a central section and two side wings in the first half of the 2nd century (cf. fig. 3). The west wing was redesigned in the first half of the 2nd century (cf. fig 4). After fire damage, probably towards the end of the 2nd century, the main building, now in its third settlement period, reached its biggest extension and was fitted with the most opulent of furnishings (cf. fig. 5). In its last settlement period, i.e. during the second half of the 3rd century, the role of the villa complex changed. The structure of the building remained the same but the remodelling of some of the components of the main building and the annexes indicate that the villa assumed a new role. In the middle of the 4th century the entire building was destroyed by another fire, which marked the end of the villa complex of Reinheim. Numerous finds, including coins and pottery (e.g. Argonne ware) from the first half of the 4th century, were salvaged from the cellar. Incidental finds from the second half of the 4th century indicate that the vicinity of the villa’s west wing was sparsely inhabited.

6 7

Miron et al. 1993, 113.

structed on a scale of 1:1 as part of the visualisation plans

The pits were discovered when a drain within the court-

carried out by the supporting organisation – the council of the Saar-Pfalz-Kreis.

yard was shifted. The documentation and mapping has been done by my former colleague Peter Buwen. 8

The gatehouse (B7) and an annexe (B1) were recon-



9

A detailed discussion of all building phases will be presented in the final publication.

well

3

2

1

5 8

4 6

7

9

shaft

22

11

26

14

23

86

19/85

83/87 81 76

78 0

79

80

20 m

Fig. 3. Simplified floor plan of the main building of the villa complex of Reinheim. Settlement period I.

settleme n t p eriod i ( mid - 1 st ce n tur y to the b egi n n i n g o f the 2 n d ce n tur y ) The finds prove that the main building was first inhabited during the second half of the 1st century AD (fig. 3). The main building consisted of a central part and two wings. In the northern half of the east wing seven rooms were grouped symmetrically around one central room (5). Rooms one to five had terrazzo flooring (opus signinum). The central section of the east wing consisted of a single room (14). The southern part, however, was divided into several rooms, only one of which had underfloor heating (room 80). Room 14 and all of the rooms in the southern part, with the exception of room 80, had clay floors. A corridor on the west side (11) formed the connection to the other parts of the building within the complex. Excavation work in the southern part of the east wing has not yet been completed. It appears that the central component of the main building was rectangular, almost 45 m long and approximately 19 m wide. It is subdivided into three rooms: 22, 23 and 26. Remodelling work in the west wing, which took place in the Roman era, only gives us a vague idea of its floor plan. Similar to the east wing, it was divided into three building structures.The entire complex that originated on the Heidenhübel seems to be H-shaped, and the main building kept this floor plan right up until the end of the settlement.



well

84

3

63 60

61 54

2

56

52

1 4

5

53

8

7

water basin

55

6

9

21

shaft

22 51

58

11

26

14

23 69

45

91

39

86

42

41

19/85

40

34

38

35

33

83/87 81 76

78 0

79

80

20 m

Fig. 4. Simplified floor plan of the main building of the villa complex of Reinheim. Settlement period II.

settleme n t p eriod ii ( b egi n n i n g o f the 2 n d ce n tur y to the b egi n n i n g o f the 3 rd ce n tur y ) Extensive remodelling work could only be observed in the area of the west wing (fig. 4). The east wing and the central part of the main building remained unchanged when compared with the previous building phase. Even if the east wing of the main building formed a separate component, the division of the west wing into three parts remained unchanged. An approximately 40 m long water tank, made from limestone slabs, is situated between the risalits of the side wings, about 3.5 m in front of the north façade. The west wing was divided into three areas: the bathing complex in the north, the living area in the south and the entrance in the centre. The outlines of the north and south parts of the west wing are equivalent to those of the east wing. The north and south walls of the newly created part of the building are in alignment with the north and south walls of the east wing. It is clear that the owner wanted the main building to be symmetrical. The bathroom was located in a separate section of the building and consisted of several small rooms. As the entire complex was in poor condition, the functionality of the various rooms could only be established in a few cases. Parts of a semicircular water gutter were discovered in rooms 52, 53 and 84. The imprint of a drain is still visible. Surviving fragments of painted wall plaster and terrazzo flooring indicate that the complex was luxurious. Although a heating room (praefurnium) could not be discovered, fragments of box tiles (tubuli) prove the existence of underfloor heating.



The living section was divided into seven rooms, two of which (33 and 34) had underfloor heating, heated from the same room (room 38). In addition to this, there was a connecting flue between the two hypocaust rooms. The other four living rooms (39-42) had clay floors. An almost square cellar room (45), with two lighting shafts on the west wall and two alcoves on the south wall, had been added later on the northern side of the living section.The masonry, preserved to a depth of 1.30 m, had been largely covered with white plaster. A painted red grout line gave the impression of regular blocks. The cellar rooms were not built on top of foundations but rather directly on top of the natural subsoil, which consists here of brown gravel. The floor of the room was made of boards. Charred remains of the 1.5 m wide wooden staircase were found parallel to the east wall. The abundance of charcoal remains from the roof beams and timber floor boards that were found in the cellar beneath the fallen roof tiles are an indication of fire damage. Bronze coins were found on top of the charcoal layer, the youngest of which dates back to the reign of Commodus (180 - 192 AD). One of the most interesting discoveries recorded were two latrines in the west wing, both built in a similar way.The first latrine is situated on the west side of the living area. It is a narrow room approximately 0.60 m wide and almost 3.50 m long. The bottom stone layers of both parallel walls are still preserved. The terrazzo flooring of the latrine slopes down slightly to the north. At the end of this, in a northwest direction, a ditch was uncovered that had a layer of green clay at the bottom.The building style of the sewer can be partly reconstructed. The walls were made of vertical standing limestone slabs and had slabs placed over the top as covers. The latrines were torn down towards the end of the 2nd century AD during extensive remodelling of the main building. In their place a heating room for the underfloor heating system (47) was built. A large proportion of the findings came from the layers filling the latrines and the sewer, in particular pottery that on first inspection seems to date to between the 1st century and the end of the 2nd century AD. The second latrine was built on the northern side of the bathing complex. This was also a narrow room (84), 0.70 m wide and almost 3.50 m long. The terrazzo floor with a 0.40 m incline crosses the bathing area and runs into a sewer with a clay base. The sewage was probably disposed of in the river Blies, situated just 100 m away. No data could be collected on how water was supplied to the first latrine; however, a pipeline was recorded at the second latrine that channelled the sewage from the west part of the bathing section to the latrine. Earth samples from both latrine sewers, analysed by the Institute for Physical Geography at the University of the Saarland, prove that these facilities were used as latrines.

settleme n t p eriod iii ( f irst hal f o f the 3 rd ce n tur y ) Between the end of the 2nd and the beginning of the 3rd century the main building was damaged by fire. As a result, extensive remodelling work followed that led to the biggest extension of the main building (fig. 5). Several smaller rooms (36-37, 46-48, 50) were added around the living quarters of the west wing and all of the living rooms were fitted with opus signinum floors. The only part of the east wing that remained unchanged was the northern part. In the wing’s central section, room 14 was replaced by rooms 12, 13, 15, 16 and 17. In the southern section, all rooms apart from room 80 were re-designed. Due to difficulties in the 19th century, only two rooms, 19 and 20, could be recorded. Both rooms had opus signinum floors. The west side of room 64 was formed into an apse. The water tank situated on the northern side of the building was abandoned and filled with building rubble and charcoal. A great deal of the pottery found came from the layers filling the tank; it dates back to between the middle of the 1st century and the end of the 2nd century AD. New colonnades were built on the north and south of the central part of the main building. The walls leading down to the south appear to have formed a kind of pergola (28). The southern ends of these walls have been destroyed by gravel quarrying.



84 3

63

60

61

64

5

53

52

54

6

10 9

55

65

4 7

8 70

62

1

2

56

21 30 59 58

22

12

51 11

26

13

16

15

23 72 69 48

45

41

39

43

17 24

42

47 46

44

40

20

19 31 29

37

34

38

36

35

33

28

71

18

81 76

78 79

32

80

50

0

20 m

Fig. 5. Simplified floor plan of the main building of the villa complex of Reinheim. Settlement period III.

settleme n t p eriod i v ( seco n d hal f o f the 3 rd ce n tur y to the middle o f the 4th ce n tur y ) In the second half of the 3rd century the functionality of several areas within the main building changed, which is probably an indication that the villa complex took on a new role. Due to modern day damage around the area of the main building’s central section it is only possible to say something about its west part where a clay floor was recorded beneath the layer of damage. Two bread ovens once stood upon this floor. Again, due to modern day damage, it is also difficult to conduct a stratigraphic evaluation of the area immediately to the south of the main building. The remains of a clay floor in room 29, where a garden had probably once stood, again indicate that the function of the main building had completely changed in this building phase. Around the middle of the 4th century the entire main building and the annexes in the courtyard fell victim to a fire. The odd finding in the west wing proves there was a settlement in the second half of the 4th century. The youngest Roman coin found in the villa of Reinheim dates back to 395 and was minted during the reign of Theodosius.



A

B13 B6

B12 B5

B11 B4

B10 B3

B9 B2

B8 B1

0

B7

50 m

Fig. 6. Complete plan of the villa complex of Reinheim with main residence (pars urbana) and working part (pars rustica) with numbered annex buildings (B1-B13).

4 the w or k i n g p art : ge n eral lay out The entire courtyard gives an impression of harmony and symmetry (fig. 6). The gatehouse (B7) on the short south side indicates that the villa owner wanted control over who entered his estate and that he wished to direct the visitors’ attention towards the main building. A pathway, probably an avenue,



Fig. 7. Annexe B3 of the villa at Reinheim. In the background are annexes B2 and B1.

led directly to the residential building on the northern edge of the complex. On the long sides of the courtyard, single buildings that were used partially as living quarters and partially as utility buildings stood together in pairs opposite each other. The wall that surrounded the entire courtyard was not secured to the single buildings. This means that the annexes were built first and the enclosure wall was added later, built in the same style as the annexes. On the east side, approximately 10 m from the southeast corner of the pars rustica, there is a single building (B8) measuring about 22 m in length from north to south and almost 15 m in width, which served as living quarters. The next single building (B9) is a singleroomed building about 10 m (N-S) in length and almost 8 m wide that had to be entered via ramps which led to an almost 2.30 m wide passage. The third and fourth buildings (B10, B11) on the east side were the same size. The first of these (B10) has not been excavated yet, but geomagnetic surveying has revealed its square outline. The east side of B11 was damaged due to gravel quarrying but the west side facing the courtyard shows the same ramp and passage as B9. The next two annexes (B12 and B13) were also damaged by gravel quarrying. Whereas there is no data available on B12, an emergency excavation in 1964 at least revealed evidence that B 13 was a 22 m long (NS) and 12 m wide building. On the west side of the courtyard onlookers would have seen an architectural complex quite similar to that of the east side. Opposite building B8 stood a single-roomed annexe measuring 14 x 9.50 m (B1; fig. 7), which also had a ramped entrance with a 2.30 m wide passage. The next four annexes (B2, B3, B4 and B5; fig. 7) are all 10 m x 8 m in size. Whereas a passage could be documented on the east side of B2, it was not possible to discover a passage in the other two buildings due to modern day damage; however, in both cases remains of ramps leading to the courtyard were discovered. The fifth annexe (B5) recorded in 2009 continues to be examined in 2010. A spacious building (B6), the last of the buildings situated on the west side, was the same size as its counterpart on the east side, 22 m in length (NS) and 12 m wide (EW). Two buildings or rather inner courtyards, equal in size, were recorded west of this building. An examination of the meticulous building techniques of the excavated areas revealed that the buildings were all built in one go. Remains of sandstone columns indicate that colonnades had probably stood in front of each of the single buildings. It has not been demonstrated that the single buildings were connected by a single colonnade. Fragments of plaster from the wall showing a red grout line prove that the exterior walls of the annexes were decorated in this way. All of the annexes had tiled roofs.



Fig. 8. Excavation photo of the gatehouse (B7) of the villa at Reinheim.

5 the w or k i n g p art : detailed descri p tio n o f some a n n e x es The gatehouse (B7) has a length of 5.60 m from east to west and a width of 5.5 m from north to south (fig. 8). The gate openings are 2.30 m wide. Whereas the foundations of the courtyard walls are approximately 0.35 m deep, the gate’s limestone foundation is 0.70 m deep. Sandstone blocks about 0.10 m thick were preserved at each corner of the gatehouse and on the edges of the gate openings. Interlocking them with the above-ground limestone masonry not only had a decorative effect but, more importantly, reinforced the corners and edges. Fragments of plaster from the wall showing a red grout line with the same decoration technique as the annexes in the courtyard, while pieces of roof tile indicate the existence of a tiled roof. A wedge of sandstone, salvaged from the rubble and measuring 7.4 cm across the lower edge, is an indication of arched windows. According to a drawing of the reconstructed arch, the internal measurement of the arch was 12 Roman palms (palmus), with each palm measuring 7.4 cm. Four-metre wide stone paving indicates that the gatehouse was connected to a road which ran 5 m south of and parallel to the courtyard wall. The rectangular annexe B1 has a length of almost 14 m (NS) and a width (EW) of about 9.5 m.10 Gates on both the east and the west sides of the building made it accessible from both sides. The two access ramps were built in layers, where pottery from the second half of the 1st century and the first half of the 2nd century AD was found. The foundations were made of loose limestone, on top of which was

10

Cf. Sărăţeanu-Müller 2010.



Fig. 9. Annexe B6 of the villa at Reinheim. Excavation photo facing south. To the left of the picture you see the paved path and the northern sandstone block from the gatehouse.

a solid foundation consisting of four layers of stone. Then came a recess, followed by the rising wall. The numerous tile fragments that were salvaged from beneath the fallen stones indicate the presence of a tiled roof. Fragments of limestone plastering with red painted grout lines show how the exterior wall was decorated. Sandstone wedges probably came from arched windows. Ramps made the annexe accessible for carriages which would enable goods or other items to be loaded and unloaded easily. The opening of the gate on the east and west side measures approximately 2.3 m and is almost as wide as the ramp. The presence of a fireplace in the opening can be interpreted as a change in function at a later date. The oven in the northern section and the charred grains of corn that were found in the southeast corner of the building are indications that annexe B1 was used as a bakery in the final settlement period from the end of the 3rd to the middle of the 4th century. Annexe B6, almost 22 m long (NS) and 12.15 m wide (EW), is a single-roomed building (fig. 9). Three post holes with an average diameter of 0.60 m were found in the middle of the building. They served as supports for the roofing structure, not as a room divider. Extensive farming of the land in modern times caused severe damage to the wall structures. The walls of two rooms which shared a façade facing south were documented west of the annexe; they probably served as an open courtyard. The walls of their short north side could no longer be documented due to damage, but both interior courtyards show the same depth (22 m) as annexe B6. The width (EW) in both cases is about 19.50 m. Remains of the enclosing wall from the main building (pars urbana) were recorded to the north and east of the annexe’s northeast corner. In both cases the wall was not interlocked with the walls of the annexe. No definite evidence proving the presence of a gate or a passageway was found. Unlike annexes B1, B2, B3, B9 and B11, there are no indications of a ramp. The preserved height of the walls is a further factor that



Fig. 10. Annexe B8 of the villa at Reinheim. Excavation photo facing south.

leads one to doubt the presence of a passageway. A paved pathway, almost 3.70 m wide, coming from the west, passed through a courtyard gate approximately 3.40 m south of the building. The pathway is flanked with larger lime stones and at its time of usage was paved with brown gravel. A huge, almost square sandstone block with sides of approximately 1.10 m was found as part of the gatehouse (fig. 9). The first building on the eastern side of the courtyard (B8; fig. 10), which measures 22 m in length (NS) and almost 15 m wide, is the only building that is subdivided into more rooms. Based on the findings recorded so far11 two main periods of use can be identified: in the first period (1st – 3rd century) the building served as a residence. Painted fragments of wall plaster, parts of sandstone columns and underfloor heating are clear indications of this. In the first phase a large, 22 x 9.50 m, rectangular building was built consisting of a larger central section and two side wings of equal size. At a later stage a hypocaust-heated room was built on to the eastern wall of the annexe. The underfloor heating system only took up part, approximately one third, of the entire room. At that stage, the heating flue was located within the central room. A stone-paved pathway, about 3.5 m wide, also belongs to this phase. It comes from the south, 1.5 m away from the east side of the building and has a gradual incline to the north. The pathway could be seen almost up to the room with the terrazzo flooring. Its profile reveals a slight curve. The stone paving was covered with a layer of gravel, with the curb marked by larger stones. This pathway probably led onto the paved road on the south side of the courtyard that connected the villa with the vicus of Bliesbruck. In its second period of use (end of the 3rd – middle of 4th century), this building

11

The archaeological examination is not yet complete.



was used as a workshop for the production and processing of bronze. Before the fire, which findings place in the middle of the 4th century, all of the rooms apart from the hypocaust room had clay floors. In the middle of the central room was a double oven. Further, smaller ovens or fireplaces were recorded both inside and outside the building. The analysis carried out by the Institute for Physical Geography at Saarland University confirms that the double oven was used for the manufacturing of bronze.

6 co n clusio n s The villa complex of Reinheim fits typologically into the category of large long-axial villas.12 Several dozen of this category have been documented in the German and Gaulish provinces by means of excavations, electromagnetic surveys or aerial photography.13 Common to all of these villas are the axial layout, the spatial division into a main building and a courtyard, and their large size. It is not uncommon for the main building – the actual residence – to appear palatial and in many cases to have luxurious furnishings and fittings such as mosaic flooring and marble-clad bathrooms. Such grand villas were only owned by the local upper class, which was completely devoted to the Roman way of life. As this type of longitudinal villa is not found outside the German or Gaulish provinces, they are probably of local Celtic origin. Judging by the findings, it is assumed that the villa of Reinheim was not built until after the middle of the 1st century AD, which places it among the later axial villas. The main residence could be reached by entering through a gatehouse on the south side of the courtyard via a central axis, most probably designed as an avenue. A partially recorded dividing wall (EW) indicates a division between the pars urbana and the pars rustica. The long sides of the courtyard wall are not connected to the annexes, which indicates that the enclosure wall was built afterwards: the annexes were built in the courtyard’s first period of use and the courtyard was later enclosed by a limestone wall built using the same building technique. The interior surface areas of long-axial courtyard villas vary in size between 8,000 m2 to 160,000 m2. The courtyard belonging to the villa of Reinheim was 40,000 m2. A distinct feature are the rows of single, same sized and equally spaced buildings used as annexes or utility buildings, which faced each other on the long sides. The presence of ramps, carriage passageways and single-roomed buildings suggest that the villa of Reinheim was a farm estate. It is easy to imagine carriage loads of goods being loaded and unloaded in these buildings. It is also conceivable that servants lived on the upper floors of the annexes. It goes without saying that the villa of Reinheim must have belonged to a powerful family, so wealthy they could afford such a grand estate. Both the size of the estate and its proximity to the ancient town of Bliesbruck indicate that the owner had a certain economic influence over the vicus. Although there is no definite proof, we believe that the villa’s owner had the row of seven utility buildings on the northern edge of the artisan district all built at the same time. All seven of the buildings were used for living and working.14

12

Cf. Sărăţeanu-Müller 2007.

13

Cf. Roymans/Habermehl, this volume, fig. 2.



14

Petit et al. 2005, 113-114.

re f ere n ces Miron, A./F. Müller/A. Schäfer, 1993: Die Villa von Reinheim. Ein archäologisch-numismatischer Vorbericht, Blesa 1, 107-128. Petit, J-P. (ed.) 2000: Le complexe des thermes de Bliesbruck (Moselle). Un quartier public au cœur d’une agglomération secondaire de la Gaule Belgique, Paris (Blesa 3). Petit, J.-P. et alii, 2005: Bliesbruck-Reinheim. Celtes et Gallo-Romains en Moselle et en Sarre, Paris. Sărăţeanu-Müller, F., 2000: Die gallo-römische Villenanlage von Reinheim, Saarpfalz, Homburg (Blätter für Geschichte und Volkskunde, Sonderheft). Sărăţeanu-Müller, F., 2007: Die Großvillen der gallischen Oberschicht. Das Beispiel von Reinheim, in J.-P. Petit/S. Santoro (eds), Leben im römischen Europa. Von Pompeji nach Bliesbruck-Reinheim, Paris, 201207. Sărăţeanu-Müller, F., 2010: Die Großvilla von Reinheim, Saarpfalz-Kreis: das Nebengebäude B1. Ein Vorbericht, in Denkmalpflege im Saarland 2, Saarbrücken, 181-201.



The Roman villa at Borg. Excavation and reconstruction Bettina Birkenhagen 1 2 3 4 5 6

Introduction General layout of the villa complex Chronological development of the pars urbana Principles of reconstruction of the residential area Reconstruction of single parts of the residence Future perspectives

References

i i n troductio n The Archaeological Park of the Roman villa at Borg is situated in a wooded area between the modern villages of Borg and Oberleuken in the German Bundesland Saarland, close to the point where the present-day borders of Germany, France and Luxembourg meet (fig. 1).1 The villa stands on the fertile limestone soils (Muschelkalk) of the Saar-Mosel-Gau between the rivers Moselle and Saar. This site was directly adjacent to the important Roman road between the civitas capitals of Augusta Treverorum/Trier and Divodurum Metromatricorum/Metz. Constructed in the Augustan period, this road was part of the long-distance link between Cologne and Lyon.2 A road settlement is known at ‘Auf Schiffels’, about 300 m to the north of the gate that gave access to the villa complex. As recent excavations have shown, it consisted of a road station (mansio) and several auxiliary buildings.3 Although most of what we presently know about the villa is the result of archaeological fieldwork carried out during the past twenty-five years, research at the villa actually began around 1900, when Johann Schneider, a local teacher from nearby Oberleuken, first discovered traces of a Roman settlement and carried out the first small-scale excavations.4 After these initial explorations, the site sank into oblivion until it was rediscovered in the early 1980s.5 The Kulturstiftung Merzig-Wadern, in close cooperation with the Landesdenkmalamt des Saarlandes, the Arbeitsamt and the municipality of Perl, started large-scale excavations in 1987 when the site became seriously threatened by the clandestine excavations of treasure hunters.6 The excavations showed that the villa is an example of the axial type, with a clear distinction 1 

Acknowledgements: Figure 1 is published with permission by the Landesamt für Kataster-, Vermessungs- und Kartenwesen license no. 23233/09 from December 17, 2009. The drawing of figure 4 was made by K.-D. Brill.

responsible at the Landesdenkmalamt des Saarlandes, it was still identified as a vicus. Cf. Miron 1997, 12. 6 

From 1993 to 2000 the excavations were part of the Deutsche Forschungsgemeinschaft research programme

2 

entitled ‘Kelten, Germanen, Römer im Mittelgebirgs-

3 

raum zwischen Luxemburg und Thüringen’. Cf. Haf-

4 

fner/Von Schnurbein 2000.

Cf. Cüppers 1990, Abb. 25. Birkenhagen 2010. Hettner 1901, 364; Hettner 1902, 441.

5 

In 1983, when the site was described by the archaeologist



Fig. 1. The area of the Archaeological Park of the Roman villa at Borg with the Roman road Trier-Metz and the road station at ‘Auf Schiffels’.

between a residential area (pars urbana) and a working area (pars rustica) (fig. 2). Over the past two and half decades, attention has concentrated on the residential area, which is now almost fully excavated. With a total surface area of 7.5 hectares, the villa is certainly one of the biggest in the Saar-Mosel region. The villa of Borg stands out for its rather exceptional preservation conditions. In some places, the walls of the building were still standing up to a height of 1.80 m!7 This was largely because the site was abandoned after the end of Roman occupation in the 5th century; it gradually reverted to woodland and has not been used for agricultural purposes ever since. This means that most of the site has never been levelled or ploughed, which has left many structures and finds more or less in situ. This of course lends the site a high scientific potential.

2 ge n eral lay out o f the v illa com p le x As mentioned above, the villa belonged to the longitudinal axial type and was divided into a residential part in the east and a working part in the west. The two parts were separated by a partition wall. Access to the main building was through a gate, positioned more or less in the centre of the dividing wall. The working part (c. 400 x 150 m) was surrounded by its own wall, outside which 17 auxiliary buildings were built, nine on the long northern side and eight on the slightly shorter southern side. The working part – and the villa as a whole – was accessed through another gate right on the main road that ran along the front of the working area. Here the course of the wall was determined by the road, whose construction must have pre-dated that of the wall of the compound. Remarkably, the main gate on the road and the gate between the working part and the main building do not align (fig. 2).



9

8

7

6

5

4

3

2

pars urbana

pars rustica

10 0

11

12

13

1

14

15

16

17

100 m

Fig. 2. Simplified plan of the villa at Borg. The pars urbana (green) is separated from the pars rustica (brown) by a wall. The buildings 1-2 and 16-17 belong to the pars rustica even if they are situated very close to the residential building. While the residential area has been almost completely dug, the working part is largely known from geophysics (buildings 16-17) and field surveys (after Miron 1997, appendix 1, with additions).

In its most luxurious phase, the residential building consisted of three wings which were symmetrically laid out around a central rectangular courtyard and were connected to each other by a surrounding gallery. The central part, on the axis of the complex, comprised the representational building with a large reception hall. The two side wings were used for a suite of baths (southern wing) and living quarters (northern wing). Two kitchens were found at the extreme ends of each side wing. If these perhaps functioned in different periods, the one in the north seems to have belonged to an early phase, the other, next to the baths, to a later phase.8 In front of this residence was a big water basin (c. 30 x 10 m) that does not align well with the main building. Between the water basin and the gate to the working part, a small foundation (c. 6 x 5.50 m) was discovered that may have belonged to a small temple or a funerary monument. The foundation consisted of large sandstone blocks, which formed an entrance to the east. A paved floor was discovered inside. As the first investigations were only superficial, further excavations are needed before we can make definitive statements about the function of this structure.9

3 chro n ological de v elo p me n t o f the p ars ur b a n a The excavations of the pars urbana have allowed us to distinguish between three main phases of continuous habitation (fig. 3): a post-built settlement dating from the Late La Tène and Augustan period, followed by the first, timber-framed building with a partial stone footing dated to the Tiberian period, and finally, the stone-built phases, the earliest of which go back to the mid-1st century AD.10 Traces of the Late La Tène settlement were found under the two side wings of the later stone-built residence. 7 

finds, the foundation cannot be dated with any precision.

8 

Perhaps new information may be gained through future

Brück 1992, 114. The southern wing was partly turned into a kitchen as late as the 3rd century.

9 

The walls of the excavation trench were just outside the

research in the immediate vicinity. 10 

Frey 2001, 31-173.

foundation. Since the excavation has not yielded any



Fig. 3. Simplified multi-period plan of the pars urbana of the villa at Borg. A settlement traces from the Late La Tène period; B Augustan-Tiberian timber-framed residence with wooden beams and portico; C Roman stone buildings belonging to three or more different phases.

This consisted of four buildings whose orientation is slightly different from the later stone buildings. A fireplace and a floor have been documented for building II, which suggests that the building functioned as living quarters. The other three buildings probably represent granaries (fig. 3 and 5).11 The timberframed building that succeeded this settlement (fig. 4) lies under, and bears a striking resemblance to, the 11 

Frey 2007, 290-295.



Fig. 4. Detailed excavation plan of the Tiberian timber-framed predecessor of the later stone building of the villa at Borg.

representational building of the stone residence. It is almost the same size (c. 17 x 10 m) and has a similar room layout and exactly the same orientation. It may thus be seen as its true predecessor.12 Although we should assume that the stone residence was not built all at once, it is not yet possible to provide a detailed sequence or the exact dates for the different rebuildings and expansions (fig. 5). At least parts of the two side wings seem to have been built contemporaneously with the central building of the first stone-built phase. Initially, these single parts were not connected. In succeeding construction phases, the two side wings were expanded eastwards and were linked to the central part by a surrounding gallery to form a single complex. The first kitchen was probably built in one of the early construction phases. For reasons that are still not clear, a second kitchen was built at a later stage in a building directly attached to the bath. The gatehouse that connected the residential area with the working part also saw several phases of reconstruction and expansion. Although the details of how the pars urbana developed have not yet been clarified, the villa clearly underwent its largest expansion and possessed its most luxurious amenities at the end of the 2nd and the beginning of the 3rd century AD. For the 4th century, some traces of a change of use could be established. For example, traces of metalworking were found in one part of the gatehouse, while a bone carver may have had his workshop in another. Ovens were installed in other parts of the buildings, sometimes even in doorways.13 However, 12 

Brück 1992, 116 f.

as well as in auxiliary building 1.

13 

For instance, in the northern porticus, in the gate house



Fig. 5. Development of the construction phases of the settlement at Borg. A Late La Tène and Augustan post-built settlement; B Tiberian timber-framed building with partial stone footing; C first stonebuilt phase dating from the mid-1st-century; D second stone-built phase dated to the 2nd century; E final stone-built phase dated to the late 2nd or early 3rd century.

not all parts of the residence were affected by such re-occupation, and some zones were simply no longer inhabited and started to decline (e.g. the earlier of the two kitchens). At the beginning of the 5th century AD, the complex was abandoned altogether and the last residents left the area.There are no traces of violent destruction that could be related to Germanic invasions. As far as the post-Roman history of the area is concerned, there have been no indications for a postRoman settlement in or directly around the villa. Habitation continued later in the modern villages of Borg and Oberleuken. Residents of these villages returned to the site in order to exploit the villa ruins as a quarry.



4 p ri n ci p les o f reco n structio n o f the reside n tial area The excellently preserved traces of the stone-built phases in particular made this an attractive site for educational programmes for a wider audience. Several options were discussed, and after initial experiments to consolidate the preserved stone foundations by adding just a few stone rows of modern masonry, the decision was finally taken to completely rebuild one of the buildings of the main residence of the villa on site.14 It was clear from the outset that two conditions had to be met to warrant this expensive solution: 1) the scientific results of the excavations had to be taken fully into account in order to ensure a high quality product; 2) the re-built structures had to be economically feasible, i.e. they had to have the potential to become cost-effective in the long term. With these guidelines, the entire pars urbana was re-built between 1997 and 2008 (fig. 6).15 However, any complete rebuilding, even if based on detailed knowledge of excavation results, leaves room for conjecture and requires decisions on issues that most archaeologists do not usually have to bother about. This case was no different. The following principles have been used to justify the solutions chosen. Rebuilding of the complex meant first of all having to opt for a single construction phase. It would have been impossible, for instance, to show all six major and all eight minor reconstruction phases of the bath building.16 As the residence reached its largest expansion with its most luxurious amenities in the 2nd and early 3rd century AD, it seemed logical to choose this phase for the rebuilding.17 Next, for reasons of authenticity, it was decided to rebuild the residence on top of the original Roman foundations where possible, although separated from them by a kind of frame. Incidentally, the original foundations had to be removed, however. There were two reasons for this. Firstly, some foundations were taken away to ensure that the preceding structures of the Late La Tène settlement or of the first timberframed building were examined satisfactorily.18 Secondly, parts of the original foundations were in such poor condition that they could not support the weight of the new building; they were therefore removed for reasons of stability. This applied to the entire north wing and the gate. Since it would have been too time-consuming and expensive to use only materials and methods that are archaeologically or historically documented, the buildings have generally been reconstructed using currently available materials and modern construction techniques. Within these limitations, an attempt has been made to match as closely as possible the visual appearance of the original Roman building.Thus modern roof tiles from Italy were used as they bore a closer resemblance to Roman ones than the tiles customarily used in Germany.19 For the typical terrazzo flooring (opus signinum), several materials and methods were tested before the final decision was taken.The stone gutters around the u-shaped courtyard that are designed to take away rainwater run-off from the roof are replicas of original examples found in

14 

For a summary of the options, Kühnen 1997, 84. The

17 

In order to overcome the problem of being able to show

decision to rebuild the complex was taken after a long

just one phase of the residence’s development, computer

consultation process. It initially concerned only the bath

models are being developed that will show interested

and central part of the residence. The necessary funds for rebuilding were provided by the Ministerium für Wirtschaft des Saarlandes and Landkreis Merzig-Wadern resp. Kulturstiftung Landkreis Merzig-Wadern. 15 

The positive feedback from visitors after the first build-

visitors other construction phases of the building. 18 

The foundations of the Tiberian timber building were covered with sand and have been preserved under the pavement of the modern building.

19 

Although roof tiles have been found in great quantities

ings (bath and modern tavern) were completed has been

during the excavations, so far there are no examples of

important for the project’s completion.

brick stamps.

16 

For a first description of the baths’ phasing, see Brück 1997, 60 ff.



Fig. 6. Exterior views of the rebuilt complex looking over the water basin to the east towards the residence (top), and to the west towards the gate which connected the pars urbana and the pars rustica (bottom).



Fig. 7. Internal views of the reconstructed baths of the villa at Borg with frigidarium (top left), tepidarium (top right), caldarium with hot water basin and labrum (bottom left), and rest room (bottom right).

situ. The same applies to the ‘half tons’ that crown the wall of the water basin, as well as the two modern copies of the nymph that decorate both short sides (fig. 6).20 Several sources were used to estimate the height of the residence’s single parts. First of all, it was assumed that the width and depth of the excavated foundations corresponded to the weight and height of the building. Part of the bath building, for example, must have been considerably higher than the rest of the residence. Second, architectural rules on proportion known from ancient authors like Vitruvius have been applied wherever possible.21 One example is that a room’s height should be in proportion to its width and length. For a two-storey reconstruction, information was obtained from references in ancient literature, from representations of ancient architecture in mural paintings from Pompeii and Herculaneum, from the physical remains of such buildings elsewhere in Italy, as well as from features discovered in excavation trenches (e.g. in Borg, the two extremely small rectangular ‘rooms’ on both sides of the reception hall can hardly be interpreted as being other than stairwells). Excavated features provided information about the position and size of single rooms. Except for the north wing, whose internal order was entirely changed and adapted to modern requirements, the original

20 

Parts of the original half tons were found in the water

clearly that the nymph stood on the edge of the water

basin. The nymph, also found in the basin, was broken into two pieces but could be repaired. Slots at the back

basin and provided it with fresh water. 21

Vitrivius, De arch. 1.2; 6.2-6.

of the nymph and an ashlar with a gap for a tube show



spatial division has largely been kept, although the rooms’ functions sometimes had to be changed. The rooms of today’s tavern, for instance, originally belonged to the baths. We largely relied on data from elsewhere for the reconstruction of doors, windows and furniture. An exception was the choice of half-round windows (cf. fig. 7 ), which were based on one of the walls of the gallery at the back of the reception hall. When the wall collapsed, it fell outwards, leaving the curvilinear window opening intact; this could be documented during the excavation. This window shape was then repeatedly used in other parts of the residence as well. The double windows reconstructed at Borg are based on examples from suburban thermes in Herculaneum,22 while the construction of the windows is based on that of the Roman fort at the Saalburg (D, Hessen).23 They are also well-known from the Vesuvian towns, both in actual reality and through representations on wall paintings or reliefs.24 The doors were designed in keeping with models from Xanten, but here again there are very well-preserved examples from Pompeii. Some of the furniture we reproduced from models in reliefs25 or wall paintings, while other furniture was bought from third parties.26 In order to improve our mises en scène still further, furniture and furnishings will remain a central concern in the years to come.

5 reco n structio n o f si n gle p arts o f the reside n ce the b aths To reconstruct the bath suite we could rely on both characteristic features and specific finds (fig. 7). Among the features preserved are those of the well-known hypocaust system consisting of a heat-resistant floor, pillars of hypocaust tiles piled on top of each other and characteristic box flue tiles (tubuli) still in situ in the remaining brickwork of the walls. They point to the typical floor heating which must have equipped the caldarium and tepidarium. Other features consisting of simple stone foundations indicated the position and size of the large cold-water basin and latrine. Finds may account for each of the different floorings in the reconstructed bath suite. Aside from the usual terrazzo floorings, there were also fragments of marble floors and mosaics, both black-and-white and coloured. Where the presence of tessellations could be demonstrated for almost all the rooms in the bath suite, with the exception of a few well-preserved parts (which today are partly reconstructed in the central part of the residence), they were generally in such poor condition that the designs were no longer recognisable. Often all that remained was just small fragments or single cubes. To reconstruct the wall decoration, we were able to use painted wall-plaster fragments from the excavation.27 The evidence for the half-round windows in the walls has been discussed in the previous section and need not be repeated here. Finally, for some rooms there is strong evidence for what the ceilings must have looked like. Buttresses in the four corners of the tepidarium, for instance, are a firm indicator of the presence of a groined vault

22 

Schalles/Rieche/Precht 1989, 40 f.

27 

Frey 1997. The spaces of today’s tavern are the oldest parts

23 

of the bath to which the other rooms were added one at

24 

a time. However, as a result of previous excavations in this

25 

For instance, the well-known reliefs on the interior of

area by Schneider, a detailed evaluation of the spatial divi-

the sarcophagus from Simpelveld at the National Muse-

sion and its changes through time proved no longer pos-

um of Antiquities in Leiden. Cf. Galestin 2001; Ulrich

sible. The upper layers with the latest construction features

2007, ch. 11, with fig. 11.3.

had been raked up and destroyed at the time, leaving only

Klee 1995, 75, fig. 46. Kühnen 1997, fig. 74.

26 

This includes klinai reconstructed by Dr M. Junkelmann.



the oldest construction phase preserved.

(cf. fig. 3). Similarly, a barrel vault may be assumed for the frigidarium, as is suggested by the discovery of slightly vaulted painted plaster fragments from the ceiling. The blue colour used in the reconstruction (fig. 7) has been clearly identified, as well as the illustration of fish for decoration purposes.

ce n tral p art o f the reside n ce The flooring in the large reception hall that functioned as the central focus of this part of the residence could not be clearly identified. It is presumed that it had to be renewed frequently because of the high groundwater level.28 This is suggested by the many levelling and floors that were documented during the excavation. The reception hall was probably at least partly decorated with a floor mosaic. The fountain basin, whose foundation still existed in situ, will have been part of the hall (fig. 8). It was originally hewn out of a sandstone block and then revetted with marble panels. The chamfer in which the water pipe was set is still discernible. The fountain has been completed by a pine cone decoration found in the villa of Merzig (Saarland), which was largely dug away during the construction of the municipal swimming pool. The reconstruction of the walls in the reception hall has to some extent been problematic. In addition to fragments of painted wall plaster, mouldings of limestone pilasters and fragments of capitals have been found, pointing to a vertical partitioning of the wall surface by means of architectural elements. These decorative elements possibly belong to different construction phases, with the walls being painted in one phase and partitioned architecturally in another. Alternatively, we could conceive of a horizontal, bipartite division of the wall, with both systems of decoration present at the same time. When rebuilding the reception hall, we chose the latter option, with the upper part of the walls embellished with architectural components and the lower part with wall paintings. The coffered ceiling of the reception hall is based on finds from the villa of Echternach (Luxembourg). Remains of a figurative wall painting were also found, but the small dimensions of the fragments did not allow a reconstruction. Today, the central part of the residential building functions as the park’s on-site museum.

k itche n The location and size of the building corresponds to the features that were found during the excavation. While ovens could be assumed in this sector, in fact only the hotplates remained. Other features and finds from this building included a brick sewer, which probably belonged to a water basin (which itself could not be identified) and fragments of amphorae and storage vessels, parts of cauldron chains, a fire shovel, etc. Although we could therefore rely on a fair amount of documentation, the picture was not full enough for the complete reconstruction of the model kitchen that we had in mind. The ovens were therefore rebuilt in accordance with better documented examples elsewhere, especially Pompeii. Four different types have been reconstructed: a cupola, an open fireplace, a bank of stoves and a bread oven. In addition, a sandstone basin that served as a sink was installed, as well as a quern to grind cereals.

28 

Unlike most other villa sites, the main building at Borg

may perhaps be explained by the settlement history. Cf.

was situated at the lowest point of the compound. This

Frey 2007, 287.



Fig. 8. Reconstruction of the main building’s reception hall of the villa at Borg with octagonal fountain (top) and original foundation of the fountain basin with chamfer in which the water pipe was set (bottom).

6

f uture p ers p ecti v es

Over the past quarter of a century, the residential part of the villa at Borg has been completely excavated. Not only has it been possible to investigate the building sequence of the villa’s stone phase, but the discovery and documentation of preceding settlement phases of Late La Tène post-built structures and a timber-framed building with portico from the earliest Roman period have made a significant contribution to our understanding of the social development of axial villas in this part of the empire. Another conclusion we can draw is that the uncertain experiment that began with the rebuilding of successive parts of the main residence in the late 1990s has proven successful. The villa of Borg now constitutes a unique example where visitors are offered a spatial experience of the entire residential area of a large Roman villa. And with about 50,000 visitors a year, the Archaeological Park operates on a fairly cost-effective basis.



Once the excavations of the residential part are completed and awaiting final publication, in the years to come fieldwork will continue in the working part of the villa. So far this part of the compound is almost exclusively known from traditional field surveys and geophysics. At the time of writing, the excavation of the first auxiliary building (auxiliary building 1) has just been completed. Its building history consists of one timber construction phase and three stone phases. In every phase, the building was enlarged. Features that are worthy of mention here are a drying kiln and parts of a terrazzo floor (opus signinum). According to Columella, such provisions may be connected with the processing and storage of cereals.29 Examples of constructions for auxiliary buildings are known from other villas, e.g. at Lebach (Saarland)30 and Oberndorf-Bochingen (Baden-Württemberg).31 Further excavations are planned within the walled area of the pars rustica. We hope that the detailed excavation and recording of all finds from the buildings in this part can continue to make a major contribution to our understanding of the economic and social aspects of villa life, i.e. the means of production that formed the basis for the wealth that was conspicuously displayed in the residential part and the organisation of the labour force.32 Finally, in the long term, attempts will be made to find the graveyard that belongs to the villa.

re f ere n ces Birkenhagen, B., 2010: Die Straßenstation „Auf Schiffels“ – eine mansio als Teil des cursus publicus?, Denkmalpflege im Saarland – Arbeitsheft 2, Saarbrücken, 127-138. Brück, J., 1992: Die Villa von Borg, in J. Lichardus/A. Miron (eds), Der Kreis Merzig-Wadern und die Mosel zwischen Nennig und Metz, Stuttgart (Führer zu archäologischen Denkmälern in Deutschland 24), 112-119. Brück, J., 1997: Geschichte und Architektur des römischen Bades von Borg, in A. Miron (ed.), Das Badegebäude der römischen Villa von Borg, Merzig, 48-75. Cüppers, H. (ed.), 1990: Die Römer in Rheinland-Pfalz, Stuttgart. Frey, M., 1997: Die Ausmalung und Möblierung des Villenbades von Borg, in A. Miron (ed.), Das Badegebäude der römischen Villa von Borg, Merzig, 76-83. Frey, M., 2000: Die villa von Borg. Ein reiches Landgut mit vorrömischer Tradition, in A. Haffner/S. von Schnurbein (eds), Kelten, Germanen, Römer im Mittelgebirgsraum zwischen Luxemburg und Thüringen. Akten des Internationalen Kolloquiums zum DFG-Schwerpunkt­programm “Romanisierung” in Trier vom 28. bis 30. September 1998, Bonn (Kolloquien zur Vor- und Frühgeschichte 5), 41-50. Frey, M., 2001: Die Terra Sigillata der gallorömischen Villenanlage bei Borg, Landkreis Merzig-Wadern, Saarbrücken (Bericht der Staatlichen Denkmalpflege im Saarland – Abteilung Bodendenkmalpflege, Beiheft 5). Frey, M., 2007: Die spätkeltische Vorgängersiedlung der Villa von Borg, in R. Echt/R. Gleser/F. Stein (eds), Zwischen Mosel und Morava – Neue Grabungen und Forschungen zur Vor- und Frühgeschichte Mitteleuropas, Bonn (Saarbrücker Studien und Materialien zur Altertumskunde 11), 285-325. Galestin, M.C., 2001. The Simpelveld sarcophagus. A unique monument in a provincial Roman context, in T.A.S.M. Panhuysen (ed.), Die Maastrichter Akten. Typologie, Ikonographie und soziale Hintergründe der provinzialen Grabdenkmäler und Wege der ikonographischen Einwirkung, Maastricht, 63-76.

29 

landes, a decision has been taken not to reconstruct the

30 

pars rustica or its buildings. There visitors will be shown

31 

the original features.

Columella, De re rustica 1.11. Miron 1990. Sommer 2005, 282-285.

32 

In collaboration with the Landesdenkmalamt des Saar-



Haffner, A./S. von Schnurbein (eds), 2000: Kelten, Germanen, Römer im Mittelgebirgsraum zwischen Luxemburg und Thüringen. Akten des Internationalen Kolloquiums zum DFG-Schwerpunkt­programm “Romanisierung” in Trier vom 28. bis 30. September 1998, Bonn (Kolloquien zur Vor- und Frühgeschichte 5). Hettner, F., 1901: Museographie über das Jahr 1900 – Provinzialmuseum Trier, Westdeutsche Zeitschrift für Geschichte und Kunst 20, 362-367. Hettner, F., 1902: Museographie über das Jahr 1901 – Provinzialmuseum Trier, Westdeutsche Zeitschrift für Geschichte und Kunst 21, 439-443. Klee, M., 1995: Die Saalburg, Stuttgart (Führer zur hessischen Vor- und Frühgeschichte 5). Kühnen, K., 1997: Zur Rekonstruktion des Badetraktes der Villa von Borg, in A. Miron (ed.), Das Badegebäude der römischen Villa von Borg, Merzig, 84-93. Miron, A., 1990: Vertikale Architektur in horizontaler Fundlage, Archäologie in Deutschland 6 (Heft 3), 44-45. Miron, A., 1997: Das Badegebäude der römischen Villa von Borg, Merzig. Schalles, H-J./A. Rieche/G. Precht, 1989: Die römischen Bäder. Colonia Ulpia Traiana – Coriovallum, Köln (Führer und Schriften des Archäologischen Parks Xanten 11). Sommer, C.S., 2005: Römische Häuser: 12 Meter bis zum First. Die villa rustica von Oberndorf-Bochingen, in S. Schmidt/M. Kempa/A. Wais (eds), Imperium Romanum. Roms Provinzen an Neckar, Rhein und Donau. Katalog zur großen Landesausstellung Baden-Württemberg 2005, Esslingen am Neckar, 282-285. Ulrich, R.B., 2007: Roman woodworking, New Haven. Zinn, F., 2001: Die Klinenfigur am Sarkophag von Simpelveld, in T.A.S.M. Panhuysen (ed.), Die Maastrichter Akten. Typologie, Ikonographie und soziale Hintergründe der provinzialen Grabdenkmäler und Wege der ikonographischen Einwirkung, Maastricht, 97-105.



List of contributors Bettina Birkenhagen Archäologiepark Römische Villa Borg [email protected] Wim de Clercq Department of Archaeology, Ghent University [email protected] Laura Crowley Archeological Centre, VU University Amsterdam; Research Institute CLUE [email protected] Ton Derks Archeological Centre, VU University Amsterdam; Research Institute CLUE [email protected] Wolfgang Gaitzsch Landesamt für Bodendenkmalpflege, Bonn [email protected] Diederick Habermehl VU-Hendrik Brunsting Stichting, VU University Amsterdam [email protected] Colin Haselgrove School of Archaeology & History, University of Leicester [email protected] Peter Henrich Deutsche Limeskommission, Römerkastell Saalburg, Bad Homburg vor der Höhe [email protected] Karen Jeneson VU University Amsterdam / Thermenmuseum, Heerlen [email protected] Jean Krier Musée National d’Histoire et d’Art, Luxembourg [email protected]



Nico Roymans Archeological Centre, VU University Amsterdam; Research Institute CLUE [email protected] Florian Sărăţeanu-Müller Stiftung Europäischer Kulturpark Bliesbruck Reinheim [email protected] Caty Schucany Independent researcher [email protected] Jeremy Tailor School of Archaeology & History, University of Leicester [email protected] Marenne Zandstra VU University Amsterdam / Auxilia [email protected]

