The Construction of the Saxon Shore Forts 9781841714875, 9781407319896

Table of contents :
Front Cover
Title Page
Copyright
Table of Contents
List of Tables
List of Figures
Abstract
Preface
CHAPTER 1. INTRODUCTION
CHAPTER 2. THE MONUMENTS
CHAPTER 3. THE PROVENANCE STUDY
CHAPTER 4. DESIGN, MATERIALS AND ARCHITECTURE
CHAPTER 5. BUILDING THE SHORE FORTS
CHAPTER 6. CONCLUSIONS
BIBLIOGRAPHY
APPENDIX I. DOCUMENTATION
APPENDIX II. ESTIMATES OF THE RAW MATERIALS
APPENDIX III. TABLE OF LABOUR CONSTANTS
APPENDIX IV. THE LITHOLOGIES

Citation preview

BAR 349 2003 PEARSON

The Construction of the Saxon Shore Forts

THE CONSTRUCTION OF THE SAXON SHORE FORTS

Andrew Pearson

BAR British Series 349 9 781841 714875

B A R

2003

Published in 2016 by BAR Publishing, Oxford BAR British Series 349 The Construction of the Saxon Shore Forts © A Pearson and the Publisher 2003 The author's moral rights under the 1988 UK Copyright, Designs and Patents Act are hereby expressly asserted. All rights reserved. No part of this work may be copied, reproduced, stored, sold, distributed, scanned, saved in any form of digital format or transmitted in any form digitally, without the written permission of the Publisher.

ISBN 9781841714875 paperback ISBN 9781407319896 e-format DOI https://doi.org/10.30861/9781841714875 A catalogue record for this book is available from the British Library BAR Publishing is the trading name of British Archaeological Reports (Oxford) Ltd. British Archaeological Reports was first incorporated in 1974 to publish the BAR Series, International and British. In 1992 Hadrian Books Ltd became part of the BAR group. This volume was originally published by Archaeopress in conjunction with British Archaeological Reports (Oxford) Ltd / Hadrian Books Ltd, the Series principal publisher, in 2003. This present volume is published by BAR Publishing, 2016.

BAR

PUBLISHING BAR titles are available from:

E MAIL P HONE F AX

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

The east wall of the Saxon Shore Fort at Portchester at low tide. On the extreme right of the photograph can be seen the Portsdown Heights, from which many of the raw materials used to build the installation were quarried.

CONTENTS CONTENTS .................................................................................................................................................................1 List of Tables ...........................................................................................................................................................2 List of Figures..........................................................................................................................................................2 ABSTRACT .................................................................................................................................................................4 PREFACE ....................................................................................................................................................................5 CHAPTER 1. INTRODUCTION ...........................................................................................................................6 1.1 Aims and objectives ............................................................................................................................7 1.2 The scope of the study.........................................................................................................................9 1.3 The structure of the monograph ........................................................................................................11 CHAPTER 2. THE MONUMENTS.....................................................................................................................12 2.1 The fabric and archaeology of the Shore Forts..................................................................................12 2.2 The role of the Shore Forts................................................................................................................26 2.3 Time scales for construction..............................................................................................................28 2.4 Conclusions .......................................................................................................................................29 CHAPTER 3. THE PROVENANCE STUDY.....................................................................................................31 3.1 Introduction .......................................................................................................................................31 3.2 Methods.............................................................................................................................................31 3.3 The lithologies...................................................................................................................................35 3.4 CBM, mortar and timber ...................................................................................................................53 3.5 Conclusions .......................................................................................................................................55 CHAPTER 4. DESIGN, MATERIALS AND ARCHITECTURE ....................................................................58 4.1 Introduction .......................................................................................................................................58 4.2 The construction process ...................................................................................................................58 4.3 The choice of materials .....................................................................................................................63 4.4 Architecture and design.....................................................................................................................70 4.5 The order of supply and construction ................................................................................................77 4.6 Conclusions .......................................................................................................................................81 CHAPTER 5. BUILDING THE SHORE FORTS ..............................................................................................84 5.1 Introduction .......................................................................................................................................84 5.2 Raw materials....................................................................................................................................84 5.3 Transport ...........................................................................................................................................91 5.4 Manpower .........................................................................................................................................96 5.5 Human resources: the Roman army ................................................................................................100 5.6 Human resources: the civilian population .......................................................................................102 5.7 Supply .............................................................................................................................................103 5.8 The context of construction .............................................................................................................105 5.9 Conclusions .....................................................................................................................................106 CHAPTER 6. CONCLUSIONS..........................................................................................................................110 6.1 Building materials ...........................................................................................................................110 6.2 The building of the Shore Forts.......................................................................................................110 6.3 Directions for future research..........................................................................................................110 BIBLIOGRAPHY ...................................................................................................................................................112 APPENDIX I. DOCUMENTATION ..................................................................................................................124 APPENDIX II. ESTIMATES OF THE RAW MATERIALS ............................................................................149 APPENDIX III. TABLE OF LABOUR CONSTANTS .......................................................................................153 APPENDIX IV. THE LITHOLOGIES .................................................................................................................154

1

List of Tables Summary of the data used in the identification of building stone in the Shore Forts. .......................34 Table 1. Table 2. Early and Middle Pleistocene Stratigraphy, south-east Norfolk........................................................39 Table 3. Summary of the architecture of the Shore Forts................................................................................83 Table 4. The preservation of the Shore Forts ..................................................................................................85 Table 5. Proposed reconstruction of the profile of the defences. ....................................................................86 Table 6. Distances between quarries and Shore Forts - building stone only. ..................................................89 Table 7. Estimated requirement for transport of raw materials between the quarries and the Shore Forts .....94 Table 8. Estimated labour requirement for the construction of the Shore Fort defences.................................98 Table 9. Average labour forces for the construction of Dover and Pevensey .................................................98 Table 10. Average labour requirements ............................................................................................................99 Table 11. Estimated population of Roman Britain ..........................................................................................101 Table 12. Samples of stone from the Shore Forts, held in the archive of the Postgraduate Research Institute for Sedimentology, University of Reading. ...............................................................................................................155 List of Figures Figure 1. Figure 2. Figure 3. Figure 4. Figure 5. Figure 6. Figure 7. Figure 8. Figure 9. Figure 10. Figure 11. Figure 12. Figure 13. Figure 14. Figure 15. Figure 16. Figure 17. Figure 18. Figure 19. Figure 20. Figure 21. Figure 22. Figure 23. Figure 24. Figure 25. Figure 26. Figure 27. Figure 28. Figure 29. Figure 30. Figure 31. Figure 32. Figure 33. Figure 34. Figure 35. Figure 36. Figure 37. Figure 38. Figure 39. Figure 40. Figure 41. Figure 42.

The Saxon Shore Forts ........................................................................................................................6 Late Roman coastal defences in Britain and Gaul...............................................................................9 The site of the Brancaster Shore Fort ................................................................................................12 Plan of the fort at Brancaster .............................................................................................................13 Caister-on-Sea ...................................................................................................................................14 The fort at Caister..............................................................................................................................14 Burgh Castle......................................................................................................................................15 Plan of Burgh Castle .........................................................................................................................16 Drawing and plan of Walton Castle ..................................................................................................17 St Peter’s chapel, Bradwell ...............................................................................................................18 Plan of Bradwell................................................................................................................................18 Reculver ............................................................................................................................................19 Plan of Reculver ................................................................................................................................19 Ricborough ........................................................................................................................................20 Plan of Richborough..........................................................................................................................21 The Roman forts at Dover .................................................................................................................22 Plan of Dover ....................................................................................................................................22 Stutfall Castle, Lympne.....................................................................................................................23 Plan of Stutfall Castle, Lympne.........................................................................................................24 Pevensey Castle.................................................................................................................................25 Plan of Pevensey Castle ....................................................................................................................25 Portchester Castle ..............................................................................................................................26 The construction dates of the Shore Forts .........................................................................................30 Reused tegulae in the Richborough defences....................................................................................33 Ecclesiastical and other buildings including stone probably from the fort at Brancaster ..................36 Robbed stone from the Brancaster fort in the south chancel of St Mary the Virgin, Brancaster.......37 Sources of stone used in the construction of the Shore Fort defences of Brancaster ........................38 Sources of stone used in the construction of the Shore Fort defences at Burgh Castle .....................40 Possible sources of building stone, Walton Castle ............................................................................42 Sources of stone used in the construction of the Shore Fort defences at Bradwell ...........................43 Ecclesiastical and other buildings including stone probably from the fort at Bradwell ....................44 Sources of stone used in the construction of the Shore Fort defences at Reculver............................46 Sources of stone used in the construction of the Shore Fort defences at Richborough .....................49 Sources of stone used in the construction of the Shore Fort defences at Lympne.............................50 Sources of stone used in the construction of the Shore Fort defences at Pevensey ...........................51 Sources of stone used in the construction of the Shore Fort defences at Portchester ........................53 Doggers of Thanet Sandstone on the foreshore at Reculver Bay ......................................................56 The west wall, Richborough..............................................................................................................61 Portchester: Bembridge Limestone bonding courses on Bastion 5 ...................................................66 Burgh Castle, south wall: brick bonding courses and split flint ashlar..............................................68 Sections through the Shore Fort defences .........................................................................................71 Gate and bastion designs ...................................................................................................................72

2

Figure 43. Richborough: construction sections on the north wall ......................................................................78 Figure 44. Construction sections at Richborough...............................................................................................79 Figure 45. Construction sections at Pevensey Castle..........................................................................................80 Figure 46. Estimated raw materials required for the construction of the Shore Fort perimeter defences. ..........88 Figure 47. Raw material use (percentage by volume): Pevensey .......................................................................88 Figure 48. Estimated volume of raw materials - perimeter defences at Pevensey..............................................89 Figure 49. Distances between quarry and Shore Forts - building stone only......................................................90 Figure 50. Distance between quarries and some selected Shore Fort sites: all raw materials, including mortar components and timber. .........................................................................................................................................91 Figure 51. Flow chart of the construction process ..............................................................................................97 Figure 52. Pevensey Castle: wall section numbers ...........................................................................................145 Figure 53. Portchester Castle: wall section numbers ........................................................................................148

3

ABSTRACT loose rock, was an extremely important activity, accounting for roughly three-quarters of all the raw materials used in the building of the forts.

This is a study of the construction of the Saxon Shore Forts, a series of late Roman coastal installations built on the south and east coasts of Britain during the 3rd century AD. It takes the reader through the generating process involved in the creation of these monuments, from design, through the extraction and transport of the raw materials, to the actual building of the fort defences. The interior buildings are not considered.

A quantitative analysis of the Shore Fort defences enabled the scale of the building programme to be assessed, in terms of the demand for raw materials, transport and manpower. The forts emerge as a relatively modest undertaking, and it is argued that sufficient resources existed to build the eight later installations within the space of a single season. The eleven forts considered in this study were but a small part of a much larger phenomenon of building in Britain and the Continent during the late Roman period, both of a military and civilian nature, and are likely to have drawn on the same sources of labour.

Geoarchaeology has played a major part in this study. A petrological examination of the stone in the defences shows that most was obtained from locations within 20 km of the building site. However, at two sites on the East Anglian coast there is evidence for material having been transported over much greater distances, suggesting the existence of more complex supply arrangements in these instances. Coastal quarrying, including the collection of

4

PREFACE have benefited greatly from the involvement of numerous individuals, particularly those in the Postgraduate Research Institute for Sedimentology at Reading University. In particular I would like to thank John Potter, Bruce Sellwood, Martin Bell, Charlie Bristow (Birkbeck College), Jim Rose (RHUL) and Bernard Worssam. Thanks are also due to Ruth Shaffrey, Cheryl Bishop, Matt Woodman, and to many other individuals within both the amateur and professional branches of archaeology for their advice, assistance and support.

This monograph originated as a doctoral thesis, funded by a grant from the Arts and Humanities Research Board and written at the University of Reading between 1996 and 1999. The monograph text is largely unchanged from that of the PhD thesis, except for a few alterations that take account of new discoveries in the past three years. The illustrations that appear in the monograph are essentially adaptations of the drawings that appeared within the thesis; however, my own efforts have been much improved by the input of Paul Jones of the GlamorganGwent Archaeological Trust.

My work on the Shore Fort sites has only been made possible with the active support of English Heritage; particular thanks are owed to Judith Roebuck of English Heritage South East. The late Harry Margary was also extremely generous in granting me repeated access to Stutfall Castle.

I owe a great deal to my PhD supervisors, Professor Mike Fulford and Professor John Allen, for their many practical contributions to the work and for their encouragement throughout. I would also like to thank Dr Janet DeLaine, without whose input, particularly on the architectural and quantity-surveying aspects, the study would have been much the poorer. The geological aspects of this research

5

CHAPTER 1. INTRODUCTION During the 3rd century AD the Romans built a series of stone forts on the south and east coasts of England. Modern scholarship has come to associate these monuments with the document known as the Notitia Dignitatum, in which they are listed under the command of the comes litoris Saxonici. These so-called ‘Saxon Shore Forts’ constituted a crucial part of the coastal infrastructure during the late Roman period, and the imposing ruins of some of their ancient defences still persist in the present landscape.

their long-term operation. Nevertheless, the question of the installations’ function remains a central question, because this has direct implications for the date, and also possibly for the rapidity, of the construction programme. For this reason, many of the traditional questions that surround these installations will be touched upon. It is now over 25 years since the publication of Stephen Johnson’s Roman Forts of the Saxon Shore (1976), and although more recent commentaries on the sites exist (Maxfield 1989; Pearson 2002a), in many ways his work continues to shape the discussion of the Shore Forts and the current study is no exception. However, it also examines the monuments in other contexts, using sources of data made available by the application of techniques from the fields of geology and quantity surveying.

This monograph addresses the construction of these installations. However, it is not yet possible to write a ‘history’ of the building of the Shore Forts for, as will be seen below, the various dates at which the monuments were created are often only understood in broad terms. Nevertheless, the exercise remains both possible and valid, and indeed by studying the building process it is hoped that this work will shed further light on the context – both chronological and economic – of these important military sites.

As with any study that attempts to recreate some of the activities of an ancient economy, many of the conclusions reached below must be speculative. However, whether the reader accepts or rejects the findings, it is hoped that this work will encourage a necessary broader, interdisciplinary, approach to such monuments in the future.

As this is a study of the very earliest years of the forts, the present work does not venture far into questions of

Figure 1.

The Saxon Shore Forts

6

1.1

built them. In the context of their own time, these installations were an important economic undertaking, and the buildings themselves represent only the end point of a complex generating process. Although the analysis of ancient structures as building projects is a relatively new development in archaeology, there now exist a number of Roman buildings that have been studied in such terms. A significant work of this nature is DeLaine’s study of the Baths of Caracalla in Rome (1997) which developed many of the approaches employed below. Within Britain itself there have been similar analyses of Hadrian’s Wall (Bennett 1990) and the legionary fortress at Inchtuthil (Shirley 1996; 2000; 2001).

Aims and objectives

From the earliest antiquarian works to modern scholarship, most studies of the Shore Forts have been preoccupied with historio-political questions (White 1961; Johnson 1976). When were the forts built? What was their function and for how long were they operational? Without doubt, these are worthy questions, and ones that are central to our understanding of the monuments. However, as we shall see below, this particular debate has not advanced greatly in recent times: the date of the forts’ inception remains unclear, and so too is the history and character of their occupation. A fuller understanding of the Shore Forts depends in large part on future archaeological work, on excavation and field survey. Until, for example, we can date each fort’s construction with greater precision, our explanations for why these installations were built must remain speculation. New evidence of this nature will be slow to emerge, and it is probable that we are many years – perhaps generations – from possessing a satisfactory data set. The destruction of much of the archaeology of the Shore Forts by both man and nature, may indeed lead to some pessimism as to whether we will ever gain definitive answers to such questions. 1.1.1

However, although the basic intention of this monograph is not markedly different from those studies of Roman buildings that have gone before, the Shore Forts themselves are in many respects quite a different proposition. These monuments are ones about which we know remarkably little; their purpose is unclear, and nor do we know the dates, or the order, in which they were built. In many respects, these are enigmatic structures. Various theories exist to explain the reasons for the development of the Shore Forts (Pearson 2002a) and several envisage quite different time scales for the construction process; as will be seen below, suggestions for the duration of the building period of later group of eight monuments range from as little as three years to over four decades. To-date this discussion has taken place with little sense of economic scale, of the implications of building such a series of installations. Chapter Five of the present work is directly concerned with this question; by quantifying the demands for raw materials and human resources we can begin to gain an impression of the economic scale of the Shore Forts. We can, additionally, consider the building of these installations in a wider context, for the 3rd century was a period during which a great many large construction projects were realised, both in Britain and on the near continent.

Building the Shore Forts

Large-scale archaeological projects are not, as this study hopes to demonstrate, the only means by which we can interrogate such sites as the Shore Forts. Nor, for that matter, are questions of date and function the only ones that we can, or indeed should, be asking of these monuments. It has been a feature of the study of the Shore Forts that, as much as their historical importance, it has been the physical remains of the monuments which has attracted the attention of scholars: The site of the old town or castle [of Richborough] is wonderful fair upon an hill. The walls the which remain there yet be in compass almost as much as the Tower of London. They have been high, thick, strong and well embattled. The matter of them is flint, marvellous and long bricks both red and white after the Britons fashion.

It is also necessary to pose the question of whether the Shore Forts represented a building ‘project’ in any meaningful sense. They are, after all, geographically separate, and at least some (if not all) are temporally distinct in terms of their construction date. The building of the installations spanned nearly a century, from Caister in c. 200 to Pevensey in 293 or shortly after; as such it is perhaps erroneous to consider them as a single ‘project’, though the later group of eight forts may have constituted a more coherent, planned programme of construction. By examining the monuments themselves, in particular the broader aspects of the architecture and the building materials, we can perhaps understand this issue rather better. Can, for example, similarities in the methods of construction be detected, which might suggest that two or more forts were the work of a single military unit or building gang? Similarly, is there evidence that certain

(Leland 1964, 61; written around 1535-43) Although the standard of preservation varies widely between the individual monuments, of the forts that remain standing above ground, four – Burgh Castle, Richborough, Pevensey and Portchester – represent some of the finest masonry to survive from Roman Britain. In few other places is such good preservation to be found in such large quantities. Yet these monuments do more than attest to the power of Rome and the skill of those who

7

an insight into the activities of the large-scale building industry in Roman Britain. It becomes possible to examine the management of such projects, from the extraction and supply of raw materials to the provision of the necessary human resources.

quarries were used to supply several forts with raw materials, indicating an overarching organisation underpinning the provisioning of the building projects? If no such evidence is found, it might be concluded that the construction of these installations was a piecemeal affair, and that it was only much later in their history that they came to form part of a unified system. 1.1.2

In respect of quarrying, however, the Shore Forts do provide a rather different view from other structures that might have been chosen for study. Analysis of the building stone shows that most was gathered or quarried on or near the coast, whereas most discussions of Romano-British quarrying have tended to concentrate on extraction from inland locations (see for example Blagg 1990). Erosion by the sea has ensured that many (in fact virtually all) of the actual quarry locations have long since been destroyed, but in the Shore Fort defences themselves we find evidence that the coastal zone was exploited for building materials on a very significant scale.

Construction and quarrying in Roman Britain

The economy as a whole is rather under-represented in the literature on Roman Britain. Large-scale construction and stone extraction are not perhaps viewed by scholars as the most glamorous of all ancient industries, and this fact goes a long way to explaining why they have received even less attention than other economic activities, such as iron or pottery production (see for example Cleere & Crossley 1985; Tyers 1996). Where quarrying in the classical world has been studied, the emphasis has tended to be on fine quality stone: a case in point is the recent investigation of the imperial quarries of Mons Porphyrites, Egypt (Peacock & Maxfield 1997). Whilst there do exist works that have addressed more common building stone, for instance Jones’ study of building in the Ager Capenus (1963), and in Britain, Sellwood’s assessment of the materials in the Silchester town wall (Sellwood 1984), these are relatively rare examples. The only attempt at a general survey of building stone in Roman Britain is that of Williams (1971). Quernstones have also received a limited amount of attention, Peacock’s study of the industry at Lodsworth (West Sussex) being perhaps the most significant work of this type yet published (1987). At present the only major work to specifically address the history of quarrying and building stone in Britain is the volume edited by Parsons (1990), and here the Roman period only features in a few of the 15 chapters.

1.1.3

Geoarchaeology

Geoarchaeology has played an indispensable role in the present study. The scientific analysis of the building stones in the forts is central to our understanding of the patterns of quarrying and supply (Chapter Three), whilst petrological techniques proved a useful tool in the quantitative analysis of the Shore Fort defences (Chapter Five). In the introduction to his 1990 volume, Parsons discussed the study of stone distribution, arguing that: … the prime need is for secure petrological determinations based on scientific observation rather than on the subjective impressions that have tended to bedevil the subject in the past. (Parsons 1990, 13)

Despite its general neglect by scholars, the building industry was of very great importance to the life of Roman Britain, and basic building stone was a vital commodity. The advent of widespread construction in stone is one of the factors that distinguishes the Roman period from preceding eras, and from the ‘Dark Ages’ that followed; stone extraction was an essential part of this phenomenon, providing materials for civilian and military structures, often on a massive scale. As Dark & Dark comment, “quarries might have been among the most striking features of the immediately extra mural zone [of towns]” (1997, 115).

However, in a later chapter within the same work, we find Blagg (1990, 39) repeating Johnson’s (1981) erroneous identification of the building stone at Richborough. It has been one of the primary aims of this monograph to offer a scientific, and above all a testable, study of the building stone in the Shore Fort defences. A rather more general aim of this study has been to demonstrate the usefulness of geology as a tool for broadening our understanding of the past, and in particular, of ancient monuments. The profile of geology in archaeology remains fairly low. One illustration of this fact can be found in Reece’s article British Sites and their Roman Coins (1993) where the author began his discussion by listing find types on Romano-British sites: building stone was entirely omitted, as for that matter were all types of environmental evidence. The present study has benefited greatly from the application of geological techniques, which have complemented the

Whilst the Shore Forts are an exceptional project in some ways, in other respects the choice of these particular monuments as the focus for study is incidental. We could choose any number of town walls, for example, or if we seek a more elaborate project, the bath complex of Roman Bath would be a promising candidate. By analysing any major building project we are able to gain

8

Count of the Saxon Shore in the Notitia Dignitatum. The Notitia itself is a somewhat enigmatic document (Mann 1976), both in terms of its purpose and the date of its composition; moreover, it is clear that the situation it portrays actually bears only partial resemblance to the true circumstances in Roman times. Additionally, there is a need to detach the 3rd century origin of the Shore Forts from the late 4th or early 5th century Notitia listing, which was in actual fact compiled after at least one of the forts (Lympne), and quite possibly several others (for example Reculver), had fallen out of use.

more traditional archaeological data with which many scholars are doubtless more familiar. It is therefore hoped that this exercise will contribute to the wider case for the recognition of geoarchaeology as a means of investigating the past. 1.2 1.2.1

The scope of the study Which monuments?

This work is not a study of all late Roman military coastal installations of the south and east coasts of Britain, but is in fact a much more limited project, one that is restricted to the examination of 11 specific monuments. Listed anticlockwise, these sites are Brancaster, Caister-on-Sea, Burgh Castle, Walton Castle, Bradwell, Reculver, Richborough, Dover, Lympne, Pevensey and Portchester (Figure 1). The scope of this monograph is not particularly defensible on grounds of inclusiveness, and as such requires some justification.

In geographical terms the British sections of the Notitia restrict attention to the coastline between the Solent and the Wash. Excluded, therefore, are major sites on the north-east coast, including Brough-on-Humber (Wacher 1969) and Horncastle (Wilson 1988, 186-187), as well as smaller installations such as the Yorkshire signal stations. Similarly, the relationship between the Shore Forts and the military sites on the Welsh and Lancashire coasts are obscured by the lack of any Notitia listing for this region, perhaps due to the demilitarisation of Wales in the late 4th century (Casey 1994, 124).

The grouping of our monuments by modern scholars is a direct result of their listing under the command of the

Figure 2.

Late Roman coastal defences in Britain and Gaul

Key to site numbers (Britain): 1 Brough-on-Humber, 2 Brancaster, 3 Caistor-by-Norwich, 4 Caister-on-Sea, 5 Burgh Castle, 6 Walton Castle, 7 Bradwell, 8 London, 9 Reculver, 10 Richborough, 11 Dover, 12 Lympne, 13 Pevensey, 14 Chichester, 15 Portchester, 16 Bitterne, 17 Cardiff, 18 Caer Gybi, 19 Lancaster, 20 Caistor, 21 Horncastle Key to site numbers (Continent): 1 The Brittenburg, 2 Oudenburg, 3 Marck, 4 Boulogne, 5 St Valery, 6 Rouen, 7 Lillebonne, 8 Le Havre, 9 Bayeux, 10 Alderney, 11 Coutances, 12 Avranches, 13 Aleth, 14 Rennes, 15 Brest, 16 Vannes, 17 Nantes

9

11 monuments in the study. However, the major datagathering exercise – the analysis of the building stone – has proved a lengthy process, and it was not possible to examine any more sites because of the restraints of time. For this reason the decision was taken to examine only a few monuments, and in this respect the 10 sites traditionally associated with the Saxon Shore offered a logical choice. The addition of Caister to this group ensures that all of the major military installations known between the Solent and the Wash have been included in the study. This having been said, Bitterne might also have been incorporated into the study, but it is not certain whether this site is a military installation or a civil settlement. Furthermore, given the poor preservation of the site and the limited data available from excavation reports, it was not felt that the inclusion of this site would add a great deal to the present study.

It is also necessary to consider the parallel developments on the near continent, where a large number of fortified coastal sites were constructed during the late Roman period, both of a military and civilian nature (Figure 2). Although the dating evidence for many of these sites is poor, the evidence indicates that at least some were provided with defences during the later 3rd century, making them possible contemporaries of the British Shore Forts (Brulet 1989). Apart from the brief period of Allectus’ reign (see 2.2.3) Britain and Gaul were under the same political control throughout the 3rd century, and the development of such sites is likely to have been closely related. Even within the geographical limits of the British Saxon Shore, further omissions can be found. Of primary importance is the fact that the Notitia lists only nine installations, which leaves unaccounted two stone forts on the Solent-Wash littoral. Seven of the eleven monuments can be tied to their Latin titles, although the connection between Othona and Bradwell is still open to some question (Rivet & Smith 1979). With the reassessment of Caister-on-Sea as a military site (Darling & Gurney 1993) the identification of Burgh Castle as Gariannonum is now uncertain. Furthermore, if Portus Adurni applies to Walton Castle rather than to Portchester (Johnson 1976, 67-8), then the ‘Saxon Shore’ can be extended no further west than Pevensey.

Whilst practical limits had to be placed on the scope of this study, it is recognised from the outset that the Roman coastal network did not consist of only 11 stone forts on the southern and eastern shores of Britain. In the chapters below where the historical and economic contexts of the Shore Forts are considered, it will be shown that these monuments were but one element of a wider phenomenon of defensive building in the north-west Empire during the late Roman period. This process embraced not only military sites, but also involved the widespread construction of town walls during the 3rd century, both in Britain and Gaul. This issue is only touched upon within this monograph, but it is argued in the concluding chapter that a broader assessment of the economic relationship between all of these construction projects – both British and continental – is long overdue.

The Notitia is equally silent about a number of other sites on the East Anglian and south coasts of Britain which are likely to be contemporaries of the Shore Forts. A possible military installation is known at Clausentum, Bitterne, Hampshire (Gathercole & Cotton 1958; King 1989; 1990) whilst the site at Shadwell, London, is one of a suspected chain of signal stations on the Thames Estuary (Johnson 1975; Perring 1991).1 Carisbrooke (Isle of Wight) might also be added to this list, but a Roman date for the enclosure under the Norman motte is rather dubious, and in any case its inland location is distinct from that of any of the Shore Forts (Rigold 1969b; Young 1983).

1.2.2

Which parts of the monuments?

The study restricts itself to an analysis of the perimeter walls of the Shore Forts. This is for the simple reason that the external defences of these installations have survived, whilst in most cases the internal buildings appear not to have done so. Whether this apparent absence is a consequence of their failure to survive in the archaeological record, or because there were genuinely very few buildings within the Shore Fort perimeters, is not yet clear. Some stone buildings are known, for example at Brancaster, Caister, Richborough and Lympne, whilst more ephemeral structures have been encountered at every site. However, only at Reculver is there evidence for a regular layout of structures, and this appears to relate to the 3rd century occupation of the site (Johnson 1989; Philp 1996). It is perhaps premature to suggest that many of the installations were virtually devoid of buildings (as argued in Cotterill 1993) because the interior of a number of the forts have yet to be properly explored. However, it is interesting to note that the recent excavations at Portchester encountered no evidence for a regular plan of buildings of Roman date,

A further consideration is that the rapid coastal erosion, which has entirely destroyed Walton Castle and left the Scarborough signal station hanging precariously on the cliff-edge, may have destroyed traces of other Roman installations. Large tracts of the east coast of Britain have been lost, whilst much of the Sussex coastal plain – an important area of Roman settlement – has also disappeared into the sea (see Burnham 1989; Pearson 2002a, Ch. 5). Even from this brief survey it can be seen that the scope of this monograph is very restricted, and it would have clearly been desirable to incorporate many more than just 1

Lakin (2002), however, has recently reassessed the evidence for this structure and suggests a non-military function.

10

In order to present the tentative conclusions presented above, it was necessary to examine ancient and modern brickwork over a wide area and to conduct a programme of geological fieldwork to supplement the published information … It is fair to claim that the modest results recorded in this paper involved a disproportionate investment of research time and resources.

although the find of building materials pointed to there having been at least one substantial structure somewhere within the defences (Cunliffe 1975). The question of internal structures must remain open, but if they had been present in significant numbers then their construction would have been a major task associated with the building of each fort. An assessment of the raw material and manpower demand for internal buildings has been made for Pevensey, based on a theoretical reconstruction, but this was only carried out with the intention of showing what the scale of the demand might have been (Pearson 1999a). At present, the archaeological investigations of this particular fort have only revealed traces of a few timber structures (Salzman 1907; 1908; Lyne unpublished). It would be nonsensical to carry out a quantitative analysis on structures for which there is no evidence, whilst a provenance study would clearly be impossible. Any such assessment must await greater archaeological data in the future. 1.2.3

(Peacock 1977, 246) Given these rather pessimistic conclusions, no attempt has been made to conduct a similar analysis of the Shore Fort CBM. If the material was not of local origin, it could have been imported to the site from any number of areas of production and the published data do not exist to support a provenance study. 1.3

The structure of the monograph

The study as a whole takes the reader through the process of construction, from the design of the forts, through the extraction and transport of the raw materials, to the final building of the perimeter defences. The generating process was extremely complex, and it is impossible to embrace every aspect of the project. However, it is hoped that all of the major activities have been addressed.

Which building materials?

Although all the raw materials used in the building of the Shore Forts are included in the quantitative analysis of the defences, only stone was examined in the provenance study (Chapter Three). The origin of the Ceramic Building Material (brick and tile; hereafter CBM), timber and the materials used in the making of mortar are briefly discussed, but their provenance has not been investigated in any great detail. In all instances it was concluded that such materials could have been procured locally, and as none of them lend themselves particularly well to a provenance study, it was decided not to pursue this issue further. In any case, wood survives so seldom in the Shore Fort structures that there was rarely even the data to establish the species of timber that was employed.

Chapter Two serves as an introduction to the history and archaeology of the monuments. Many issues of central importance to this monograph are addressed; particular attention is devoted to a discussion of the dates of the Shore Forts’ construction and to theories regarding their function. Chapter Three contains the study of the stone used to build the Shore Fort defences. The lithologies present at each site are described and a provenance is suggested for each type. The possible sources of the CBM, timber and mortar components are briefly discussed at the end of the chapter. Chapter Four addresses a number of questions relating to the design of the forts. It analyses the methods of construction employed at each site, and then goes on to consider the reasons why the architecture of the each fort may take the form that it does. Various issues are addressed, from the quality of building stones being supplied to each project, to the question of how designs were transmitted in the ancient world. Chapter Five examines the building of the Shore Fort system as a whole. The demand for raw materials, transport and manpower is calculated, and using these results it is possible to demonstrate the scale of the overall building project, and its cost to the Roman State in terms of human and material resources. The remainder of the chapter is devoted to a discussion of how such a building programme could have been achieved in the context of the late Romano-British economy. Chapter Six presents the conclusions of the study and suggests some directions for future research.

A uniform approach to the mortar was not possible, since it cannot now be observed at poorly preserved sites such as Brancaster, Walton Castle and Bradwell. At others of the Shore Forts extensive repointing often obscures the original facing, sometimes making it difficult to determine what was Roman work and what was later repair. At Reculver, recent renovation had completely covered the original mortar. On the other hand, it would certainly have been possible to characterise the petrology of the clays used to produce the CBM in the Shore Forts, as large volumes of brick and tile remain accessible for analysis. A study of this type has been attempted before, which examined the tiles produced by the classis Britannica in south-east Britain and north-west Gaul (Peacock 1977). However, in discussing his methodology the author commented:

11

CHAPTER 2. THE MONUMENTS The following chapter discusses the archaeology, history, and the surviving fabric of the Shore Forts. As such it serves as an essential introduction to many of the themes that will be encountered in later parts of the monograph, particularly those regarding the date of the installations’ construction, and theories about their function.

the monuments can be found elsewhere, the two most recent being in Maxfield (1989) and Pearson (2002a), whilst detailed discussions of the architecture also exist (Johnson 1976; 1989).

2.1

Whilst the fort and extra-mural settlement are shown very clearly in aerial photographs (Edwards & Green 1977), at ground level there is practically nothing to see at Brancaster (Figure 3). Only the eastern ditch of the fort is easily recognisable, preserved as a distinct roll in the ground. Much of the western part of the extra-mural settlement now lies under a housing estate, but the field immediately east of the fort site yields large quantities of building stone, tile and pottery. Ashlar blocks from the fort can be found in the parish church at Brancaster, and elsewhere in the locality (see 3.3.1).

2.1.1

The fabric and archaeology of the Shore Forts

This monograph is very much reliant on data from the surviving remains of the Shore Fort defences; these provide data about the design, the types and use of raw materials, as well as enabling the quantitative analysis. In virtually all senses, this is a monument-based study. What follows below amounts to a brief guide to the Shore Fort defences and their preservation in the modern day. Detailed bibliographic references relating to the sites can be found in Appendix I. More complete descriptions of

Brancaster

Figure 3. The site of the Brancaster Shore Fort View north. The fort’s east ditch can be seen as a slight depression on the right hand side of the photograph The extra mural settlement at Brancaster has been dated to the 2nd century (Hinchliffe & Green 1985), but the fort itself is rather later, and its orientation suggests that it was actually superimposed over the central part of the existing

community. The defences themselves have not been dated but the lowest occupation levels within the fort relate to the mid and late 3rd century (St. Joseph 1936, 451). Coin evidence suggests that the most intense phase

12

of occupation was during the latter parts of the 3rd century. Carausian issues form by far the largest group, but coins of Tetricus are also reasonably represented (Casey 1994, 124). The best dating for the defences is provided by a rubbish deposit overlying the rampart, which contained a coin of Tetricus, thus giving a terminus ante quem of c. AD 270. The pottery in the

deposit is also of late 3rd century character. The early style of Brancaster’s defences (see below) strongly suggests that it should be regarded as a contemporary of the late 2nd or early 3rd century forts of Reculver and Caister. If this is the case, then the construction of Brancaster pre-dated the first period of its intensive use by several decades.

Figure 4. Plan of the fort at Brancaster The internal buildings show as cropmarks on aerial photographs 2.1.2

Caister-on-Sea

Tucked away in the midst of a modern housing estate can be found the remains of the Roman fort at Caister. Only the south-west quarter has been left exposed, including a 15m stretch of the south wall and one of the gatehouses (Figure 5). The defences stand nowhere above 0.5m and in the most westerly portion only the foundations remain. The stub of a wall that was built to retain the tail of the rampart also survives. All sections of the defences have been capped with modern concrete. Within the fort can be seen the remains of two stone buildings, together with a short stretch of a north-south road.

defences of earth and timber were constructed very shortly prior to the erection of the stone perimeter (on the basis of the find of a 'palisade trench'), but this hypothesis has yet to be conclusively proven (Darling & Gurney 1993, 10-11 & 15; Ellison 1962, Fig. 4a). A terminus post quem for the stone defences is provided by pottery in the fill of the palisade trench, which dates to not earlier than the late 2nd century, and is more likely to relate to the early 3rd century (Darling & Gurney 1993, 15). Pottery forming part of a rubbish deposit dumped onto the rampart is of early to mid 3rd century character, and this provides a terminus ante quem for construction (Darling & Gurney 1993, 11-13).

The balance of the evidence suggests that the fort at Caister was built on unoccupied ground at some point early in the 3rd century. There have been suggestions that

13

Figure 5. Caister-on-Sea The south wall of the defences and south gatehouse

Figure 6.

The fort at Caister

14

Figure 7. Burgh Castle East wall

2.1.3

The evidence for the date of this fort is very slight, so much so that in Johnson’s discussion of Green’s excavation data, his primary arguments are based on typology (Johnson 1983, 116). There is no evidence that relates directly to the perimeter defences. Excavation of a small section of the interior between 1958 and 1961 produced no evidence of occupation prior to Constantinian, and the demolished structures adjacent to the perimeter wall dated to the second quarter of the 4th century. As Johnson points out, however, the placing of buildings against perimeter defences is a characteristic of 4th rather than 3rd century military planning (1983, 117), and therefore these structures may post-date the erection of the perimeter. Surface finds suggest that the site was indeed occupied during the pre-Constantinian period (Morris 1947, 68), but allow no more precise an interpretation.

Burgh Castle

Situated on a raised tongue of land on the edge of the Norfolk Broads, Burgh Castle overlooks the muchdiminished ‘Great Estuary’, upon whose shores the fort once stood (see Pearson 2002, Ch. 5). Three sides of the defences remain, standing to heights above 4m, including the entire east wall (Figure 7). The south wall has survived particularly well, albeit leaning at a rather improbable angle in places. The facing of split flints and opus signinum mortar is well preserved on the exterior face (Figure 40) but the inside of the wall has been badly robbed. The west wall once stood at the top of a gentle scarp, but has now tumbled down into the marshes below, where fragments can still be seen in winter.

15

Figure 8.

2.1.4

Plan of Burgh Castle

Walton Castle

remains a few hundred metres out to sea from the present cliff-line, and there seems little doubt that much of this must be fragments of the fort defences (Errington MS; Wall 1937). Practically nothing is known of the chronology of this monument. Field walking on the nearby settlement, 500m from the site of the fort, has found pottery spanning the second to the 4th century (Esmonde Cleary 1995, 35), but how this relates to the Shore Fort is unknown.

The last remnants of the Roman fort of Walton Castle were destroyed by coastal erosion during the 18th century (Fox 1911, 289). Some illustrations of the site exist from the 17th century (Figure 9) together with a few written descriptions dating to the earlier 18th century, and these provide very rudimentary information about the size and appearance of the fort. During the 20th century, several diving expeditions encountered substantial masonry

16

Figure 9. 2.1.5

Drawing and plan of Walton Castle presumably Roman date. Within the area once occupied by the fort there are large scatters of building stone to be found that correspond to the lithologies present in the chapel, especially along the line of the north wall.

Bradwell

Only a very short and extremely unimpressive portion of the Roman defences at Bradwell remains exposed above ground, and even this is obscured by heavy overgrowth. Situated on the edge of the Dengie Marshes, the site itself is dominated by the 7th century Saxon chapel of St. Peter, which occupies the location of the fort's west gate (Parker & Lewin 1867). The chapel is predominantly built of Roman ashlar and CBM, although the central sections of the north and south walls were repaired in the previous century using stone imported to the site (Figure 10). Particularly notable is the reuse of monumental blocks of Lincolnshire Limestone, employed as quoins on the west wall of the chapel, which exhibit lewis holes of

The dating of this fort relies entirely on the interpretation of the very limited coin series from Parker’s 19th century excavations (Casey 1994, 122; Lewin 1867). The 3rd century coins are mostly those of the Gallic Empire, and of Carausius and Allectus, but they are few in number and none has a context. The series is reminiscent of that of Portchester, which is thought to be a Carausian project, but this association alone cannot confirm a similar date for the construction of Bradwell.

17

Figure 10.

St Peter’s chapel, Bradwell View east

Figure 11.

Plan of Bradwell

18

2.1.6

erosion, and the defences can now only be seen on the east and south sides (Figure 12). In most places postRoman stone robbing has left only the core of the wall, although some ashlar can still be seen in some locations, for example at ground level immediately to the south of the east gate.

Reculver

Besieged on three sides by caravan parks, the fort of Reculver is dominated by the remains of St. Mary's church, the majority of which dates to the 12th century. About half of the fort has been destroyed by coastal

Figure 12. Reculver East wall

Figure 13.

Plan of Reculver

19

be seen in the upper registers. By contrast, the north wall is in excellent repair, and represents some of the finest Roman masonry to survive in Britain (Figure 43). This side of the fort, argued below to have faced the water (4.5.1), shows evidence of decoration, as well as the reuse of stone from the 1st century triumphal arch (the so-called Great Monument), which once occupied the site but was demolished to make way for the Shore Fort.

The most impressive feature of the fort is the remains of the monumental south gate, which was left exposed after its excavation during the 1960s (Philp 1996). The east gatehouse, which is far less elaborate, also survives, and here is the only location in the fort defences where brick can be observed. Coins and pottery from the rampart bank excavated in 1957 confirm that the fort defences were built during the early decades of the 3rd century (Philp 1996, 13). The basilica of the fort’s principia has been dated by inscription to the Severan, or immediately post-Severan period (Birley 1981, 173-6; Mann 1977), and this implies that the fort as a whole was begun and completed within the space of only a few years. Most (though not all) of the other internal structures that are known within the fort are also of early to mid 2nd century date. 2.1.7

The dating of the fort is complex, but is bracketed by two principal events. The first is the backfilling of the tripleditched enclosure built around the remains of the Monument, which has a terminus post quem of c.AD 273 (Bushe-Fox 1949, 65). The pit cutting the abandoned east wall foundations was also filled with material of similar date: the uppermost part of this deposit was building rubble, and has been suggested to relate to the clearing of the site when construction was complete (Johnson 1970, 245). In neither of these features were Carausian coins found, with one exception, of uncertain stratification. A terminus ante quem is suggested by occupation evidence, which contains many Carausian coins in primary contexts. The evidence supports a date for construction between AD 273 and the very earliest stages of the Carausian regime, before issues of his coin became common.

Richborough

The fort of Richborough is situated on a low hill that overlooks the now silted Wantsum Channel. The River Stour has eroded the eastern parts of the defences, but significant stretches of the north and south wall survive, together with the entire length of the west wall (Figure 14). The western and southern defences have been extensively robbed, but sizeable areas of facing can still

Figure 14. Ricborough West wall

20

Figure 15. Plan of Richborough Shore Fort phase

2.1.8

adjacent to the York Street bypass. Although the 19881991 excavations on the site of the ‘White Cliffs Experience’ were swiftly written up (Wilkinson 1994), there is much data from the earlier investigations that await detailed publication.

Dover

Although scholars had long suspected the existence of a late Roman military installation at Dover, it was not until rescue excavations in 1970 that the first remains of Roman defences were positively identified. Subsequent rescue work lasting until 1977 revealed the existence of not one, but three forts under the modern town (Philp 1981). The first of these (dated to c. AD 120) was never completed, but was superseded by a second, built on the same site and on the same orientation in approximately AD 130-40. Both installations were strongly associated with the classis Britannica. Overlying the demolished north-east corner of the second fort was the defensive wall of the Shore Fort.

The soil accumulation over the rubble of the demolished classis Britannica fort contains coins that indicate a terminus post quem of AD 253 for construction of the Shore Fort (Wilkinson 1994, 70). A clay layer, interpreted as the tail of the rampart of the south wall contains coins of AD 260–280, whilst the tip layers forming the west wall rampart contained an assemblage dating from AD 275 to the early 4th century (Wilkinson 1994, 71). The excavator suggested a terminus post quem of AD 275 for the entire fort, judging it unlikely that there was a pause of several years between the building of the south wall and work beginning on the western wall of the fort (Wilkinson 1994, 72-73).

More recent work has revealed other masonry, but only a little of the Shore Fort at Dover now remains open to view. Direct study is restricted to the very short section of wall and the accompanying bastion that is preserved

21

Figure 16. The Roman forts at Dover In this excavated area, the wall and bastion of the Shore Fort can be seen cutting across the east gate of the preceding classis Britannica installation

Figure 17.

Plan of Dover

22

2.1.9

which the east gate bastions stood was also recycled. However, despite the presence of significant quantities of reused building material in the defences of the Shore Fort at Lympne (Cunliffe 1980, 285) there is no direct evidence from within the monument to demonstrate any prior occupation of the site (Cunliffe 1980, 281). The date of the fort is mainly based on coin evidence, the bulk of which come from Roach Smith’s excavations and have no context (Reece in Cunliffe 1980, 260-64; Reece 1994, 123; Roach Smith 1852, 32). With one exception the coins are issues of AD 259 and later. Whilst occupation of the site in the 260s could be supported by the evidence, a date during the AD 270s is more likely. A Carausian building date is plausible, although the coin series as a whole perhaps points to a slightly earlier date (Reece in Cunliffe 1980, 263).

Lympne

The remnants of Stutfall Castle lie scattered on an ancient, degraded cliff on the edge of the Wealden Scarp, overlooking the lowland region of Romney Marsh. The extensive landslipping that destroyed the fort has led to a great deal of confusion about the plan of the defences, although a recent geotechnical investigation seems to have resolved this issue (Hutchinson et al. 1985). Fragments of three sides of the fort wall remain, and in some places these stand in excess of 6m: they are enough to show that prior to their destruction the defences were amongst the most massive of the Shore Forts. Plenty of stone remains for study, including small sections of the exterior facing. A great deal of reused material is evident in the defences: most is tile but the raft of blocks on

Figure 18. Stutfall Castle, Lympne A stretch of the ruined north wall

23

Figure 19. Plan of Stutfall Castle, Lympne Left: plan of the present site. Right: reconstructed plan of the Shore Fort defences (after Hutchinson et al. 1985)

2.1.10

of the bastions on the north wall (Bushe-Fox 1932a). However, recent excavations have conclusively shown that the defences were in fact built several decades earlier. Dendrochronological analysis of oak piles from the foundations of the fallen south wall, discovered in combination with coins of Carausius and Allectus, has enabled a confident revision of the construction date to AD 293 or shortly afterwards (Fulford & Tyers 1995). Whilst an explanation of the Constantinian coin is still required, the idea that the north wall was built during the second quarter of the 4th century no longer seems plausible.

Pevensey

The Roman fort of Anderita is an impressive site. Of the 760m oval perimeter roughly two thirds has survived, much of it in extremely good condition. Facing and bonding courses remain at many points on the external face, in particular on the north wall which has almost completely escaped the attentions of stone robbers (Figure 20). The internal face has fared rather less well, and ashlar can only be found on a short section of the north wall. Standing above 8m in places and flanked by projecting bastions, the walls of Pevensey provide a clear illustration of the massive nature of some of the later Shore Fort defences.

The excavations of the interior have produced limited evidence for occupation of the site during the late 3rd century (Lyne, unpublished), and it is difficult to envisage the use of a site whose defences had not been completed.

Until recently, Pevensey was thought to have been built during the fourth century, a conclusion suggested by the find of a coin of Constantine in a cavity underneath one

24

Figure 20. Pevensey Castle The north wall

Figure 21.

Plan of Pevensey Castle

25

2.1.11

bonding courses, which due to their being set deep into the wall, have tended to survive reasonably well in places (Figure 39). However, there has been a tendency for later repairs to adopt a ‘pseudo-Roman’ style (quite probably using fallen Roman materials) and not all bonding courses, particularly those of brick, were built during the Roman period.

Portchester

Portchester Castle has the longest history of occupation of any of the Shore Forts (Figure 22). It remained in use up until the 19th century (Munby 1990), whilst the church built within its perimeter is still active. The extensive post-Roman use of the fort, combined with the erosion of the promontory on which it stands, can clearly be seen in its architecture. Parts of the south wall were entirely rebuilt during the medieval period after they were destroyed by the sea, whilst the insertion of a keep during the Norman period in the north-east quarter led to the complete demolition of a lengthy portion of the Roman defences. Elsewhere, many sections of the perimeter have been extensively repaired, either in stone, or more recently in brick. There is relatively little of the original Roman facing to be seen, the best section being at the lowest levels of wall section 4 (north wall). The most prominent feature of the Roman facing to remain are the

Figure 22. 2.2

Coins of Tetricus I in the early construction phases of the fort walls and internal roads provide a terminus post quem of AD 260-268 for the building of the fort. The recovery of Tetrican and Carausian coins from the very earliest occupation levels suggests a slightly later date for the first use of the installation. The excavator preferred a Carausian date for the fort's construction (Cunliffe 1975, 60) but this interpretation has yet to be proved beyond doubt.

Portchester Castle construction. Other analyses of Roman structures have operated within a time scale that is known in relatively precise terms (e.g. Delaine 1997; Shirley 2000), but this is not a luxury afforded to the Shore Forts.

The role of the Shore Forts

From the above discussion, it is clear that archaeological evidence only helps us to a certain point. It is not certain when many of the forts were begun – particularly those in the later series – and there is even less data to suggest when they were finished. In this crucial respect, therefore, this monograph differs from other studies of Roman

Still more fundamentally, the lack of a precise chronological framework renders uncertain the very reasons why the forts were built (though admittedly even

26

the British and continental coasts, a system that later evolved into the command of the Count of the Saxon Shore which we find listed in the Notitia Dignitatum. In Johnson’s model the forts, lying on the coasts of Britain and Gaul, functioned as fleet bases and also housed garrisons of troops with a high degree of mobility. Because of the primitive nature of the raiders’ boats and navigational technology, they would have been forced to follow a predictable route along the Gallic coast to the Straits of Dover. From this point they could either continue southwards to raid Gallia Belgica or the Tractus Amoricanus, or cross the Channel to southern Britain. The Roman naval patrols acted as a screen, intercepting the raiders before they could attack, whilst those that slipped through the net were dealt with by the land forces. The fleet could capture any who escaped before they reached the North Sea.

if such dates were known, they would not necessarily provide a clear explanation). There are a number of possible factors that might have triggered the building of the Shore Forts and at some stage all have been offered as the reason for their inception. The proposed roles for the forts can be divided into three basic themes; as protection for Britain’s coast from barbarian raiders, as fortified trans-shipment ports, and as a Carausian-Allectan defence against the armies of the Maximian and Constantius. A general failing within the literature is the assumption that these installations could not have served in several roles simultaneously, or that their purpose may have evolved during their lifetime. The authors of the theories outlined below have tended to assume a single function, and only Wood admitted the possibility of a change in the operation of the forts (1990, 95). The post-medieval history of Portchester, during which it served as castle, barracks, prison and stores-base (Munby 1990), however, illustrates the changing uses that a defended perimeter can fulfil. 2.2.1

2.2.2

Trans-shipment ports

Criticisms of this model have led to more recent suggestions of a different role for the Shore Forts. It is argued that the presence of a barbarian threat to Britain in the third century is questionable and that the units stationed at the forts were in any case too inflexible to counter a seaborne threat (Wood 1990). Instead, the installations served as defended trans-shipment harbours, their position at the mouths of navigable waterways not arising from a need to protect the interior but rather to facilitate access for commercial shipping (Cotterill 1993; Milne 1990, 84). These ports would have provided a secure transport link between Britain and the Continent, along which could have travelled both civilian and military traffic.

Defences against seaborne raiding

The idea of our monuments as an anti-pirate defence has provided the most enduring explanation of the Shore Forts and is one that first found expression in Camden’s Britannia: The honorable, Earle or Lieftenant of the Saxon Shore along Britaine, whose office was garisons set upon the shore in places convenient, to represse the depredations and robberies of Barbarians, but of Saxons especially, who grievously infested Britaine (Camden 1637)

2.2.3

A Carausian/Allectan defence

A markedly different theory centres on the period of Carausius and Allectus, whose seizure of power in Britain lasted from the mid-280s until AD 296 (Casey 1994). In the first modern work to address the Shore Forts, White (1961) argued that the forts were the work of Carausius, representing a defensive scheme designed to protect the south and east coasts of Britain from an invasion by Imperial armies. In his view, the scale of the defences were indicative of an enemy with a capacity for siege warfare, and as such the Shore Forts were overengineered for a threat from pirate raiders or bagaudae. Where the Rhine frontier and the Gallic provinces faced the real danger of a repeat of the barbarian invasions of AD 259 and AD 275-6, White argued that no such threat was posed to Britain. He questioned the extent of barbarian raiding in southern England during the 3rd century and reasoned that the only likely threat was from the Roman army itself, during the breakaway regime of Carausius and Allectus. On this basis, it was proposed that all of the Shore Forts, Reculver and Brancaster included, were the work of Carausius (White 1961, Ch.3).

It is an image that persists, little changed, within many modern histories of Roman Britain: [Britain] was under threat of raiding and piracy from peoples along the North Sea littoral outside the Roman frontier on the lower Rhine, the forebears of the Anglo-Saxons. This resulted in the building of the Saxon Shore from the early 3rd century on ... (Esmonde Cleary 1989, 43) The installations are seen as an evolving response to an increasing threat from beyond the Rhine frontier, one which may have been accelerated by the barbarian invasion of Gaul in AD 275 that devastated the region and led to a comprehensive programme of town fortification in the province (Cunliffe 1975, 421). Johnson (1976) has given substance to such ideas, proposing the forts as forming a limes that guarded both

27

Further problems arise because establishing the date when construction started is not the same as determining the point at which it finished. In the earlier forts there is some evidence to help resolve this question, for example at Reculver, where the inscription from the principia strongly suggests that the fort was completed not later than the AD 220s. The evidence from the interior of the fort at Caister is sufficient to show that occupation began very early in the 3rd century (Darling & Gurney 1993). In both of these cases, therefore, we can see that the construction process was completed in the space of a few years, perhaps a decade at most.

White’s ideas have been somewhat unfashionable, particularly following the publication of Stephen Johnson’s theories, but Fulford & Tyers (1995) have recently revived them in the light of the conclusive dating of Pevensey to post-AD 293. In their re-assessment of the origin of the Shore Forts, they have suggested that only with the loss of Boulogne did a seaborne invasion from the Continent become a viable threat. Where White envisaged all the forts as the work of Carausius, Fulford & Tyers have argued for a slightly later inception, seeing the later installations as a distinct group, built late in Carausius’ reign or under Allectus (Fulford & Tyers 1995, 1013).

The evidence for the occupation of the later forts (and by proxy, their completion) is far less good. Despite the absence of archaeological evidence, it is perhaps reasonable to expect that all of the forts were finished by the termination of the reign of Allectus. If these forts were intended to be used as defences against Rome then there was clearly a need for them to be rapidly completed after the loss of Boulogne in AD 293. However, we cannot rule out the possibility that Constantius’ recapture of the province in AD 296 interrupted the building programme before its completion. Allectus’ construction of a ‘palace’ complex in London (Williams 1993) can certainly be taken to suggest that he did not expect his reign to be so short-lived.

The discussion above has concentrated only on the question of function. For some of the proposed roles we can look to certain key events that might have seen the onset of construction, for example the barbarian invasions of Gaul in AD 275 or the Carausian usurpation of power in Britain around AD 285. It is admittedly rather more difficult to suggest any single date when work on a series of trans-shipment ports might have begun. Although any of these historical episodes might suggest why the Shore Forts were built, it is another matter to determine the period over which construction took place. This is one of the main objectives of the present study, to establish the scale of the Shore Forts as a building project and to determine the limits of what might have been possible in terms of speed of construction. As we shall see below (2.3), there are time scales implicit in many of the proposed functions: an evolving anti-pirate system could have developed over several decades, whilst any Carausian-Allectan scheme must have been produced in a rather shorter period. 2.3

If we do accept that all the forts were finished by AD 296 then we can at least place some limits on the length of time that it took to build certain of the Shore Forts. Given that Pevensey’s construction began around AD 293 then we must envisage a project of not more than three years, and perhaps rather less. If Portchester was begun during the reign of Carausius, as has been suggested by Cunliffe (1975, 60), then we have a building programme of not more than 11 years.

Time scales for construction

As described above, the archaeological evidence points to a fairly clear, if basic, chronological pattern for the Shore Forts’ construction (Figure 23). The installations at Reculver and Caister can be placed with reasonable confidence in the earliest years of the 3rd century, and it seems likely that Brancaster was of similar date. The remainder of the forts were built at some later point from the AD 260s, with work on at least one site, Pevensey, beginning during the AD 290s.

Unfortunately, the date of AD 296 as a terminus ante quem is only useful up to a certain point. It tells us little about the length of the construction programme of any of the forts that were started at a date in the middle parts of the 3rd century, as was perhaps the case for Burgh Castle and Dover. It is hard to envisage that a site which was begun in the AD 260s or 270s would have taken decades to reach completion. As we have seen, the archaeological evidence is not sufficient to support any single theory regarding the inception of the Shore Forts, nor does it inform us about the length of any building programme. The coarseness of the archaeological chronology is such that we are free to use (or misuse) the data to fit a number of scenarios.

It is for Pevensey that we have the most precise archaeological evidence for a building date, the foundation piles indicating that construction began in AD 293 or shortly after (Fulford & Tyers 1995). We can place the building of Reculver in the Severan or immediately post-Severan period, whilst Richborough can be dated between AD 273 and AD 285. For many of our other sites the archaeological evidence is less precise, and we often have time-frames of 20 or more years during which construction could have taken place. There is no data to help us suggest a date for Walton Castle.

It is at least possible to dismiss White’s theory of an entirely Carausian project. The data show Brancaster, Caister, Reculver and possibly Richborough to pre-date the usurpation of c. AD 285, whilst the balance of the evidence from Lympne and Dover also suggests an

28

complete each fort (assuming a finish date of AD 296 for the series as a whole) and the entire programme could have been carried out by a single ‘work gang’ and transport fleet.

earlier inception. The Brancaster evidence was available to White, and its omission illustrates his occasional careless or perhaps naïve treatment of the archaeology. A longer evolutionary development, as proposed by Johnson and Cunliffe, can be supported, although there are some problems with the date of the fort at Dover (above, 2.1.8). An early date for Brancaster, Caister and Reculver seems likely, whilst the ‘transitional’ monuments (Burgh, Walton, Bradwell, Richborough, Dover and Lympne) could have been built between the accession of Probus and the early years of the Tetrarchy. The evidence also corroborates a late date for Portchester and Pevensey, which are supposedly the most advanced installations on architectural grounds. Despite the problems outlined above, the time has not yet come to dismiss this model out of hand.

A Carausian or Allectan defensive system In this scenario we find the usurpers having to build eight forts within a period of around 11 years. This assumes an arbitrary date of AD 296 for the completion of the forts, despite the fact that we cannot expect Carausius and Allectus to have known when their regime would end, and as a result the installations may not have been finished by this point. If the building of Richborough can be assigned to the reign of Carausius, then its construction must have taken place in the very earliest months following his seizure of power. This would have left the usurpers having to complete the seven remaining forts built between c. AD 286 and AD 296, a time scale that allows an average of just over a year for the completion of each fort. Sequential construction by a single work unit and fleet remains possible, but it is more reasonable to expect several projects to be undertaken simultaneously. Building may well have been prompted by the new strategic situations that arose at the start of both the Carausian and the Allectan regimes. It is quite possible, therefore, that all seven projects were running simultaneously in the early stages of either man’s reign.

Whilst no fort can be definitely tied to Carausius, the data could support at least seven forts dating to the usurpers’ regime. Although the balance of evidence suggests that Richborough’s construction pre-dated Carausius’ usurpation of power in Britain, it is conceivable that this fort was built extremely early in his reign. Pevensey can be firmly placed in the Allectan period, and there is no archaeological data that preclude the idea that six other forts were built soon after the loss of Boulogne in AD 293.

An Allectan defensive system

Because of the difficulties posed by the archaeological data, the findings of the later chapters will be discussed in the context of several plausible scenarios. In all cases, however, the first phase is the same, since the evidence points to the building of Caister and Reculver having taken place during the early 3rd century. Even if one does not accept the arguments for a strict typological evolution, the style of the defences at Brancaster would seem to suggest that the date of its construction also lay in the first half of the 3rd century, and therefore the building of this site can also be seen as distinct from the later series of forts. Beyond this point, however, we are faced with the prospect of three quite different, but equally plausible, time scales:

The Allectan scenario is perhaps the most interesting, in that it required the most demanding schedule and would have imposed the greatest strain on the economy. Given that Allectus’ regime lasted between AD 293 and AD 296, there was a maximum of only three years to build seven forts. Furthermore, since the precise months in which he assumed power in AD 293 and was deposed in AD 296 are unknown, it is quite possible that the period of his rule lasted little more than two years (Casey 1994, 45). Such a time scale requires up to four forts to be built every year and crucially for the first time a multiple labour and transport force would definitely have been needed. In this scenario, probably above all others, the question of completion must be regarded as uncertain.

The building of eight installations beginning around AD 260

2.4 Such a scenario is most likely if we envisage our installations as an anti-pirate defence, or as transshipment ports. Although there is nothing to preclude the idea that each fort was built relatively rapidly, a longer term and evolutionary development of the Shore Fort scheme as a whole seems most likely. Certainly the system could not have been completed until the AD 290s, when Pevensey was built, suggesting that the programme could have stretched over a period of 35 years. Construction may have been entirely sequential, but an overlap between some of the projects is possible. The time scale allows an average of four or five years to

Conclusions

This chapter has examined a number of subjects that are crucial to the present study of the Shore Forts. Whilst it is not the stated aim of this monograph to directly address the questions of date and function, the findings of this work do have a significant bearing on the matter. Although a judgement is not made about the intended role of the installations, in establishing the scale of the building project we are at least able to more fully understand the implications of a short or long building programme.

29

Figure 23. The construction dates of the Shore Forts Each bar represents the range of time over which the archaeological evidence suggests each fort was built. At precisely what point within this range construction actually took place is not known, nor is the duration of each project. An arbitrary cut-off date for construction is taken as AD 300.

30

CHAPTER 3. THE PROVENANCE STUDY Until recently, therefore, the state of knowledge about the Shore Fort building materials has been fragmentary. Moreover, where identifications have been suggested, many – especially those in older texts – were made by archaeologists who had little knowledge of geology. As a result, many of the descriptions were vague, of suspect accuracy, or simply incorrect. It is particularly unfortunate that Stephen Johnson, the leading author on the Shore Forts in recent years, tended to pass over geological issues in his studies.

The identification of the lithologies used in the building of the Shore Forts is a central aim of this monograph. The purpose of this chapter is to outline the geological survey of the monuments – its methods, conclusions, and its limitations. The building stones are described, and, where possible, a provenance is suggested. The lithological descriptions on which the provenance study is based can be found in Appendix IV. What follows below is essentially an outline of the data; the conclusions drawn from the survey are discussed at length within the later chapters. 3.1

The one work to specifically examine the building stones in the Shore Forts is that of Allen and Fulford (1999). This paper was concerned with a reconnaissance-level identification of the lithologies employed in the construction of the 3rd century coastal forts of East Anglia, Essex and north Kent; it did not extend to any installations westward from Dover. On the basis of these data, Allen & Fulford were able to discuss the supply of building materials to the east coast forts, and further, more detailed studies of the facing stone used at Brancaster have followed (Allen & Fulford 1999; Allen, Fulford & Pearson 2001).

Introduction

Interest in the geology of the Shore Forts can be traced back well into the 19th century. In his discussion of Richborough Castle, the famous antiquary Charles Roach Smith not only described the stone present in the defences, but also commented that: ... it would be a curious question to know whence all these materials, foreign to the locality, came; and to ascertain if there are any springs or rivulets depositing travertino or calcarious tuffa in the neighbourhood.

In the present chapter, the investigation has been widened to include all of the ‘Shore Forts’ on the East Anglian and south coasts of Britain. It also attempts to improve the precision of some of the provenances suggested by previous authors, and in cases entirely different conclusions about the source areas have been reached.

(Roach Smith 1850, 38-39) A curious question perhaps, but not one that inspired much interest amongst later generations. Most 19th century works on the Shore Forts made little or no reference to the building materials, being far more concerned with descriptions of the architecture, or with archaeological excavation. Even individuals such as George Dowker – a leading figure in Kentish geology and archaeology during the late 19th century – paid very little attention to the building stone in his works on the Shore Forts at Reculver and Richborough (e.g. Dowker 1885).

3.2

Methods

From the outset, the intention of this study was to examine, first hand, as many types of building stone from the Shore Forts that were accessible at that time. In this way a uniform approach could be brought to all of the lithologies, whilst the descriptions of the material could be presented within a single text for the first time. Whilst the existing literature has sometimes provided a useful starting point, the conclusions of this study have, wherever possible, been made independently of all previous identifications.

In more recent times the subject has received somewhat better treatment. Excavation reports from Portchester (Cunliffe 1975), Lympne (Cunliffe 1980a) and Dover (Wilkinson 1994) have included some details about the building materials, whilst scattered notes regarding the stone in the forts can occasionally be found in other archaeological literature. However, such references only form part of the general narrative about the forts’ architecture; little attempt has been made to describe the stones in any detail, and only rarely is a provenance suggested. A rare exception to this trend is the exemplary treatment of the Marquise oolite and Calcaire Grossiere employed at Richborough, although these lithologies are only present as trace quantities within the defences (Worssam & Tatton-Brown 1990).

3.2.1

Sources of data

A number of sources of data were used to establish the lithologies employed in the Shore Fort perimeters. The survey operated with a hierarchy of types of evidence, which are detailed below, in descending order of reliability.

31

examine, first hand, some materials that might have originated from the fort. Similarly, the presence of Roman facing stones in the chancel of St Mary’s Church, Brancaster, enabled the identification of the accompanying flint as being originally from the nearby Shore Fort. Even when some standing Roman masonry survives, as at Bradwell, robbed stone can be of value. Although a short stretch of Shore Fort wall still survives, religious buildings on the Dengie peninsula exhibited a far larger quantity, and, significantly, a wider variety of stone types robbed from the Roman perimeter.

The monuments The primary source of data was the Shore Forts themselves, a self-evident fact given that the monuments were partly chosen for study because of their preservation. Standing masonry survives at eight of the eleven sites, with very large amounts in situ at Burgh Castle, Richborough, Pevensey and Portchester. This represented the most desirable evidence; the stone was in its original setting, its context was clear, and, where significant amounts survived, it was possible to estimate the relative quantities of each lithology present. Field walking of the site of Brancaster also yielded significant quantities of building stone from the Roman fort, which, although lacking a context, at least enabled first hand examination of the construction materials.

It is true that the identification of Roman stone in later buildings is, at least in part, based on personal experience in the field rather than on a check-list of objective indicators (Whitworth 1994, 19). However, in his study of the robbing and reuse of Roman building materials, Eaton comments that:

Significant post-Roman reuse of the Shore Fort sites is known at Burgh Castle, Walton, Bradwell, Reculver, Pevensey and Portchester. At the first four sites listed, occupation only resulted in the partial demolition of the Roman perimeter; however, at Pevensey and Portchester significant alterations were made to the fabric in terms of repair and addition. Pevensey retains much of its ashlar facing, and the repairs were undertaken in a markedly different style – either in ‘herringbone’ pattern, or in rubble – and as such they were easily identified. Portchester represented a more difficult challenge, since both the original masonry and medieval repairs were of coursed flint.

the very fact that identifying reuse is not a science … makes it all the more essential for those working on the subject to make absolutely clear the characteristics of the masonry that have led them to identify reused Roman stone Eaton (1999) Eaton proposed a number of criteria for the identification of Roman stone in later buildings. Not all are applicable to the Shore Forts, but several are of direct relevance and are summarised as follow:

Post Roman reuse Petrology. A jumble of stone types present within a single building phase indicates that the material may have been robbed from an earlier structure. If new stone was being quarried it is logical to expect that it would be of a uniform lithology, particularly if relatively small quantities were required. The presence of a lithology not used locally at any time other than the Roman period also indicates robbing.

All of the Shore Forts were quarried for building stone during the post-Roman period, and a proportion of this material remains identifiable in later structures. Robbing of the forts can be shown to have begun at least as early as the 7th century, for it is at this time that the chapel at Bradwell was built. The practice persisted at a number of the sites well into the 18th century. Much Shore Fort stone was quarried for the construction of local churches during the medieval period, and it has largely been in such structures that Roman masonry has been located during this study. Brancaster provides a good example of the reuse of Roman stone in secular buildings; this was doubtless a widespread phenomenon at the other forts, but one whose traces are more difficult to identify.

Ceramic Building Material. After the Roman period there is little evidence for brick and tile manufacture in Britain before the 12th century and therefore the presence of CBM in structures earlier than this date almost certainly indicates the robbing of a Roman structure. Identifying Roman CBM in post 12th century structures is more problematic: tegula flanges are distinctively Roman (see Figure 24) whilst any glazed fragments will be of medieval date. The presence of CBM suggests that the accompanying rubble may also be Roman, but cannot prove this to be the case.

Where the Shore Fort from which the stone was robbed is well preserved, these younger buildings tend only to mirror the lithologies present in the fort itself. The interest of these buildings lies in other aspects, providing an insight into the robbing process itself, for example the types of stone that were preferred by the medieval builder, and the order in which demolition occurred. Where little or nothing of the parent structure survives, reused stone becomes an important source of data. For Walton Castle, robbed stone offered the only chance to

Lewis holes. The use of lewises for the lifting of large blocks is not confined to the Roman period, but does have a strong affiliation with Roman construction (Adam 1994, 49-50). Where lewis holes are visible on the exposed face of a wall, the block can be said to be reused.

32

Block form. Identification of Roman stone by size and shape is largely confined to monumental blocks (opus quadratum). However, comparison of the dimensions of

Figure 24.

reused stone to blocks still in situ in a Shore Fort may suggest whether it was likely to have been robbed from the Roman fort.

Reused tegulae in the Richborough defences

In some cases, robbing only involved the intra-site movement of stone, examples being the construction of St Mary’s church, Reculver, and St Peter’s chapel, Bradwell, both of which were built within the original area of the Roman defences. Similarly, the construction of parts of the medieval keep at Portchester was achieved by the thinning of the inner face of the Roman perimeter (Cunliffe 1975, 11), clearly apparent today. In these cases, it is self-evident that the stone was quarried from the Shore Fort.

some 20 km – for the construction of a barn (Rose 1985, 188). Even where a specific connection could be made between the Shore Fort and a later building, the original context of most robbed stone was lost. At Brancaster, and more crucially at Bradwell, it was impossible to determine which (if any) of the ashlar was from the perimeter defences, and which from internal buildings. Textual sources

At a distance from the parent structure, difficulties emerge in proving that robbed Roman building material had been taken from a Shore Fort. Eaton’s recent study of the robbing and reuse of Roman masonry in Monmouthshire, Kent and Northumberland has shown that stone was generally transported no more than five kilometres from the source building. Even when the material was moved by water, journeys of more than 10 km were generally avoided (Eaton 1999; 2000).

Where building materials were no longer accessible for study, there had to be some reliance on textual sources – antiquarian records or drawings, and excavation reports. None of the foundations of the forts were exposed at the time of the study, and so excavation reports provided the only source of data for this aspect of the monuments. As noted in Chapter One, stone is a neglected find-type on Romano-British sites, and such reports were often not as detailed as might have been desired. However, in most cases it was possible to gain a reasonable impression of the material types employed, particularly at sites such as Dover, where the stone was distinctive, and easily identifiable (Amos & Wheeler 1929; Philp 1981).

However, in at least one case Shore Fort masonry can be shown to have travelled considerably greater distances overland, 19th century documents recording the transport of ashlar from Brancaster to Thorplands, Fakenham –

33

Table 1.

Summary of the data used in the identification of building stone in the Shore Forts.

Brancaster 1 Burgh Castle Caister-on-Sea Walton Castle 2 Bradwell Reculver Richborough Dover Lympne Pevensey Portchester 1 2

Fabric in situ

Robbed or reused stone

Textual sources

Y Y Y Y Y Y Y Y Y

Y Y Y -

Y Y Y -

No standing masonry. Building stone from the site obtained from field walking. Reused Roman stone at St Mary’s church, Walton, of dubious authenticity.

3.2.2

3.2.3

Lithological analysis

Provenance

Any provenance study is reliant on the quality of data available about the lithology, primarily in terms of geological information. This quality can be assessed in the following terms:

The majority of the lithologies were analysed only at hand specimen level. Where in situ, whether in the Roman fort or reused in a later structure, the stone was examined with a hand lens with a magnification of x10. Where samples were brought back to the laboratory, it was possible to study them under a binocular microscope with magnification of x20.

Accessibility. Of central importance is the lithology itself – how much remains to be examined, how well it has survived exposure to the elements, and the extent to which fresh surfaces are accessible for study. In broad terms, the more material that is available, the better the impression gained about its appearance and overall properties. Gaining access to the higher parts of the walls was often impossible, and therefore most of the lithological examination was carried out close to ground level. Observations were made at height, less effectively, through binoculars.

Examination of the material in situ avoided the lengthy process of obtaining the permission to take samples, and, practical difficulties aside, was normally considered to be adequate for the purposes of this study. Hand specimen analysis enabled the identification of a rock’s basic macroscopic properties (colour, texture and sorting) the major mineral and fossil components, including cements, and sedimentary structures. The descriptions of each rock followed the guidelines suggested in Tuckers’ Description of Sedimentary Rocks in the Field (1996).

Distinctiveness. The distinctiveness of a rock is also a factor in determining how precise a provenance may be suggested. Some lithologies encountered in this study, for example the Lincolnshire Limestone present at Bradwell, are quite distinctive, in this case because of its colour and fossil assemblage. In contrast, ‘blander’ materials, such as the sandstone facing at Brancaster, give far fewer clues about their origin from only a hand specimen analysis.

Thin section or chemical analysis of some stone types present in the Shore Forts may in the future refine, or perhaps even disprove, some of the conclusions reached below about the provenance of certain lithologies. However, with so many different stone types used in the construction of these monuments, a more detailed scientific analysis of each was beyond the scope of this study. The thin-sectioning of each sample from the fort would have had to be matched by a number of others from the likely quarry areas and it was clearly not feasible to examine more than a few of the lithologies in such terms. Thin section analysis was only employed in the case of the Quarr stone from Portchester Castle, where the present study coincided with an ongoing project that is investigating the use of this lithology in Roman and medieval construction (Bishop, forthcoming).

With the characteristics of the rock described, the conclusions about a likely provenance were reached by a combination of the following methods: Use of geological texts, in particular the memoirs that accompany the British Geological Survey map series. Comparison of samples from the Shore Forts to specimens in reference collections of building stones. This was of particular use for lithologies that vary

34

is not possible to suggest a precise provenance for any given cementstone used for building.

considerably according to their precise provenance (e.g. Kentish Ragstone) and for stone where several origins had been suggested (e.g. the facing material at Brancaster). The extensive archive held at the Natural History Museum (London) was used for this purpose, and to a lesser extent, the Sedgwick Museum (Cambridge).

Where a provenance can been suggested for the bulk items needed for the forts’ construction (the foundations and core rubble), it is readily apparent that the Romans tended to rely on local, coastal resources, although not necessarily those nearest to the fort.2 Where a provenance could not be suggested with certainty – in particular the flint in the southern forts – local sources have been assumed. Such conclusions are supported by economic logic, but the possibility of the long-distance transport of large quantities of stone cannot be totally discounted.

Field visits to the locality from which the stone was suspected to have been quarried. Modern geological exposures were visited to determine whether similar facies to those that the Romans might have exploited were observable in the modern landscape. Study of older structures such as churches, castles, domestic houses and even garden or sea walls, were all studied in order to gain an impression of the stone types from the locality that had been exploited in the past.

3.3

The lithologies

In the following sections the building stones present at each fort are discussed in turn. The reader is referred to Appendix IV for a more lengthy description of each lithology and relevant references. Because of its late inclusion in the study, and the lengthy process of obtaining permission to take samples, Caister-on-Sea has not been treated in the same detail as the other forts. An on-site analysis was sufficient to demonstrate that the stone was obtained from glacial deposits, though not the same as those exploited for the later building of Burgh Castle, but more detailed work is required on this issue in the future.

Historical data. A provenance suggested on purely geological grounds could on occasion be enhanced by reference to historical sources. The Roman quarry industry has many successors, many of which have utilised similar resources and have operated within the same basic economic parameters. Thus, while Roman quarrying is neither visible in the modern landscape, nor attested in textual sources, later industries may indicate plausible sources as well as information about methods of extraction and transport. Occasional documentary evidence points to very specific conclusions, but more often it indicates only a general locality from which the stone might have been procured. Historical analogy is not equivalent to scientific proof, however, and such ethnographic data must be used with caution.

3.3.1

Brancaster

The thorough demolition of the fort has left no material in situ for examination (Rose 1985). A little masonry survives below modern ground level, and the building materials were described in brief terms by St Joseph in his report on the 1935 excavations of the fort (St Joseph 1936).

Finally, in both the initial identification of some of the stone types, and in the suggestion of a provenance, I am indebted to a number of geologists who I have acknowledged at the beginning of this thesis. Their advice has been invaluable to this project, and without their expertise the conclusions of this study would have been much the poorer.

Field walking of the land around the fort provides a good impression of the materials used, and confirms St Joseph’s observations. The actual field in which the fort was situated was given over to pasture for the duration of this study; only a few scatters of flint were observed on the land surface, whilst limited amounts of building materials were retrieved from the surrounding hedges. However, significant quantities of stone and CBM were found in the ploughed field immediately to the east of the fort (its western margins about 60m from the site of the fort wall). Here, beach flint was abundant, together with rather fewer nodular examples. Some Red Chalk and white Lower Chalk was evident, together with small amounts of Carstone, and a pink lava. Previous field walking of the same site had also turned up fragments of a “hard, dull yellowish-grey slightly micaceous, very fine

The closest source? It is unfortunate that the lithologies employed in the greatest quantities in the Shore Forts do not lend themselves particularly well to a provenance study. Whilst the morphology of flint can say much about its recent environment, for example whether it was collected from a beach or freshly dug from the chalk, it cannot indicate a precise provenance. Flints occurs almost ubiquitously in the Upper Chalk formation of southern England and only in the cases of Burgh Castle, Caisteron-Sea and Brancaster, where a number of distinctive glacial erratics were present alongside the flint, was it possible to suggest a locality for the quarries with any certainty. Septarian cementstones present similar problems; whilst they are more abundant in some units of the London Clay than in others (Potter 1999, 289-294), it

2

For the purposes of this study the term ‘local’ is arbitrarily used to denote an area within a radius of 30 km of the building site. ‘Regional’ is used to describe an area 30-100 km from the Shore Fort.

35

Robbed stone from the fort has also been observed in later buildings in the region (Figure 25; Rose 1985). Blocks of white, grey-weathering sandstone have been found in the churches at Brancaster, Burnham Deepdale and Titchwell, as well as in domestic structures such as the barn behind Staithe House (Brancaster Staithe). Further afield, similar blocks can be seen in the rebuilt barn gable of the farm at Thorplands Lodge, near Fakenham, a structure built for Rev. J. Lee-Warner, the first excavator of the Brancaster fort (Rose 1985).

grained, flaggy quartz sandstone”. This lithology was interpreted as a Coal Measures sandstone, perhaps from the southernmost part of the East Pennine Coalfield (Allen & Fulford 1999). Significantly, a number of large blocks of white sugary sandstone were also discovered, many of which were roughly dressed. No shaped stones of any other lithology were discovered, suggesting that this sandstone was the only type of stone used in the facing of the fort wall.

Figure 25.

Ecclesiastical and other buildings including stone probably from the fort at Brancaster After Allen & Fulford (1999, Fig. 1)

This stone is a precise match to that from the fort site, both in terms of its lithology and the shape and dimensions of the blocks. However, the exact part of the fort from which the material was robbed is less certain. In all probability the stone was indeed taken from the perimeter, which would have been the major source of ashlar blocks. However, St Joseph did record the walls of an internal building in the south-west corner of the fort, commenting that it was ‘formed of sandstone blocks, similar to those of the facing of the fort-wall’ (1936, 450). Whether or not from the perimeter, the blocks remain representative of the stone employed in the facing.

including part of a moulded Romanesque arch, a blocked window with quoins (built of a cream coloured chalk rock) and Lincolnshire Limestone (Figure 26). Several dozen squared blocks of the sandstone facing from the fort are also present, built in rough courses. The majority of the church fabric (also 14th century) comprised flint, mostly battered beach cobbles but also some occasional nodules. Amongst the flint were dispersed a number of basic igneous rocks, including examples of the pink lava which is found on the fort site. It therefore seems highly likely, therefore, that the bulk of the church was constructed with flint and other robbed stone from the perimeter of the fort. There is a great deal of flint used elsewhere in Brancaster and the villages to the east, and it must be suspected that much of this material also originated from the fort.

The church of St Mary the Virgin, Brancaster, contains a great deal of Roman stone. The south wall of the 14th century chancel (Pevsner 1962, 94) incorporates a wide variety of reused material, not all of Roman date,

36

Figure 26.

Robbed stone from the Brancaster fort in the south chancel of St Mary the Virgin, Brancaster of Brancaster, where it is found as a formation near the base of the cliff, and also as a reef on the foreshore (Gallois 1994, 100-116). Numerous examples of Carstone buildings can be seen in the older parts of Hunstanton. Overlying the Carstone are thin strata of Hunstanton Red Rock (also known as Red Chalk) and above this is white (Lower) Chalk (Chatwin 1961, 25-6). Both the white, and particularly the Red Chalk have been employed in local construction, although only in low status structures and garden walls. The Lower Chalk almost certainly provided the lime for the “hard white mortar” described by St Joseph, and is noted in other functions within the fort, principally as pitching for the intra-vallum road and for flooring (St Joseph 1936, 448). The sequence of Carstone and Chalk is exposed at a number of locations between Hunstanton and Snettisham, and in the past the Carstone has been quarried across the whole region. The chalks were commonly gathered directly from the beach at Hunstanton, but have also been worked in a number of pits around Hunstanton and the area to the south (Gallois 1994, 122).

The core materials The textual and field walking evidence combines to suggest that the core of the Roman defensive wall was formed of flint, Carstone, erratics and hard chalk rubble set in concrete, commonly with an irregular footing course of flints. Flint of this character is abundant on the beaches of the north-east Norfolk coast. That at Weybourne is somewhat too small to provide a match, but further to the east the beaches between West Runton and Cromer yield highly battered cobbles and nodules of an acceptable size. Cobbles of basic igneous erratics are also present, including lava. It has to be suspected that the Coal Measures Sandstone described by Allen & Fulford (1999) near to the fort was also glacially introduced, although none was observed on any of these beaches during the present study. The other stone types used in the core of the wall were also of local origin and form part of the Cretaceous sequences that outcrop in north-west Norfolk (Chatwin 1961, 24). Carstone is a ferruginous, often pebbly sandstone, found within the Lower Greensand. It has been described as the chief building stone of Norfolk (Woodward 1901, 5) and indeed represents one of the few litholgies of any quality in the region. A major typesite for the Carstone is the cliff at Hunstanton, 7 km west

The facing St Joseph described the facing of the wall at Brancaster as composed of a rather soft sandstone, with large flints and lumps of Red Chalk used below ground surface (1936,

37

the village of Castle Rising, including the medieval castle, church and hospital, and Wakeield House, a postmedieval structure on the south-western outskirts. A few other examples decorate lawns or occur in hedgerows beside the east-west lane at the foot of the slope on the north side of Castle Rising village. Geological field methods have failed satisfactorily to position the rock stratigraphically, and nor is it clear where any quarries would have been located. Gallois, the most recent investigator, suggests that the stone from the church and castle is a ‘secondary quartzite’ from the Lower Cretaceous Leziate Beds of the immediate area, but gave no indication of where it could be seen in context (Gallois 1994, 77, 79 & 176).

446-447). There have been several previous suggestions made for the identity of this sandstone; these have included Kentish Ragstone, Aislaby Sandstone from east Yorkshire (Rose 1985, 189), Ashdown sandstone from the Wealden Beds of Kent, and a mature sandstone from northern England, perhaps from the Jurassic Estuarine or Deltaic series (Allen & Fulford 1999). Recent work has now resolved this question, however, and demonstrates a local origin for the material (Allen, Fulford & Pearson 2001). Heavy mineral analysis of samples from the fort have shown that the material originates in the Castle Rising area, to the south-west of Brancaster and 35 km distant by water. Blocks of identical stone can be observed in buildings in and around

Figure 27. 3.3.2

Sources of stone used in the construction of the Shore Fort defences of Brancaster fossiliferous sandstones, shelly and oolitic limestones, calcite-cemented claystones, and, rarely, Hertfordshire Puddingstone and quartzites.

Burgh Castle

The fort at Burgh Castle exhibits a great deal of uniformity throughout its structure, both in its architecture, and in the range of materials employed in its construction (Figure 28). The face consists of flint cobbles, bonded at intervals with triple courses of brick. The core is built from flint cobbles and some small flint boulders, amongst which are evenly distributed minor-totrace lithologies of pebble to cobble size. A wide variety of these erratic lithologies are represented, and the list in Appendix IV is not exhaustive. Present are pure and

Recent drift deposits, dating from the early Pleistocene to the close of the Anglian Glaciation (Chatwin 1961) cover much of the western part of Norfolk and Suffolk. The composition of these deposits varies according to the formation processes, and to the route of the glaciers which carried the material to its present locations. The sequence is summarised in Table 2.

38

Table 2. Early and Middle Pleistocene Stratigraphy, south-east Norfolk Based on McBridge & Hopson (1985) Lithostratigraphy (Group/Formation) (Member) Lowestoft Formation Aldeby Sands and Gravels Pleasure Gardens Tills Oulton Beds Lowestoft Till North Sea Drift Formation Corton Sands Lee Hill Sands and Gravels

Sediment

Environment

Sand and gravel

Scottish Ice source Proglacial outwash

Chalky, clayey diamicton Sand,silt and clay

Flowtill

Chalky, clayey diamicton

Lodgement Till

ChronoStratigraphy

ANGLIAN

Lake sedimentation

Scandinavian Ice Source Distal proglacial outwash

Fine sands Sands and gravels with igneous erratics Sandy, brown diamicton

Proximal proglacial outwash

Silt, sand and quartzose gravel

Fluviatile/ marine

Rootlet Bed

Clay with calcareous nodules

Alluvial

Un-named member

Silt, sand and quavels

Fluviatile/ marine

Quartzose sands and gravels

Fluviatile

EARLY PLEISTOCENE

Shelly sand, silt and clay

Marine

EARLY PLEISTOCENE

Starston Till Cromer Forest-bed Formation Un-named member

Kesgrave/ Bytham Group Kirby Cane sands and gravels

Crag Group

The erratic lithologies found in the fabric of Burgh Castle tend to suggest that much of the building material was derived from the Lowestoft Till, the Lowestoft Till outwash, and other associated deposits.3 Most are present in the Lowestoft Till, some exclusively so. The majority are not found in earlier drift deposits such as the North Sea Drift formations, the Crag Group or the prePleistocene gravels. Most of the erratic lithologies have a north-eastern English provenance, and many (although not all) are of Jurassic age. A provenance in the Lowestoft Till is supported by the near total absence of metamorphic lithologies, the gneisses, schists and quartzites that are abundant in the North Sea Drift formations. A few lithologies were not from the Lowestoft Till, for example Hertfordshire Puddingstone (Kesgrave Group, sands and gravels) and a quartzite (North Sea Drift formations).

ANGLIAN

Till clasts

CROMERIAN COMPLEX

erratic lithologies (10%). Given the volumes of matrix that would have had to be dug in order to procure stone from the Till, it is more likely that building material was collected from the outwash, either on the coast or on river margins. The largest exposures of the Lowestoft Till and its outwash are found on the coast, where it forms part of the cliffs between Corton and Lowestoft (McCBridge & Hopson 1985). Its presence inland is less well understood, and without a clear stratigraphic sequence, difficulties still remain in distinguishing between the various drift deposits. Only in quarry pits, dug for the extraction of clay for the cement industry, are the tills readily distinguishable from each other. Lowestoft Till, for example, can be seen in section in the quarries at Great Blakenham, Suffolk. However, extensive spreads of outwash from the Lowestoft Till, consisting of coarse sand, and gravels up to cobble grade, have been proven on the flanks of the Waveney valley between Burgh St Peter and Beccles (McCBridge and Hopson 1985).

Lowestoft Till itself is a diamicton, of which around 93% is a chalky-clayey matrix. The remaining part is comprised of flint (50%), chalk (40%) and the various

Given the even distribution of erratic lithologies amongst the flint in the structure of Burgh Castle, it is apparent that all the stone in the fort was gathered from the same source. Flint of a similar size to that in the fort – nodules

3

I am grateful to Professor Jim Rose (RHUL) for his assistance in identifying these lithologies.

39

lithologies such as the oolitic limestones and septaria would not have survived long in the outwash, a fact that perhaps explains why they are present in smaller quantities at Burgh Castle than the harder sandstones and shelly limestones.

from the Lowestoft Till and more rounded cobbles from the Kesgrave sand and gravels – can be obtained from the beaches around Covehithe. At both of these locations, the soft matrix of the till has been eroded away by the sea, leaving the flint and other glacial erratics on the beach below the cliffs. From here it could have been collected and shipped to Burgh Castle via the Great Estuary. Soft

Figure 28. 3.3.3

Sources of stone used in the construction of the Shore Fort defences at Burgh Castle schist and gneiss. Preliminary examination of these erratic lithologies shows that, as a group, they are rather different to those present at Burgh Castle. The presence of igneous and metamorphic rocks is notable, whilst soft lithologies such as oolitic limestones and cementstones are absent. It can therefore be tentatively suggested that the stone for the defences was derived from the North Sea Drift formation. In geographical terms, there may be little separation between the sources exploited for the building of Caister and Burgh Castle. The North Sea Drift is also exposed on the east Norfolk coast, and at locations such as the cliffs at Corton it is present immediately below the Lowestoft Till (McBridge & Hopson 1985).

Caister-on-Sea

The fort at Caister was not studied in the same detail as the other Shore Forts. What follows is a brief summary; more work, including sampling of the erratic lithologies, is required at this site. The majority of the south wall is built from flint. Most of the flints are very well rounded and extremely abraded, and are clearly from a marine environment. The size of these flints ranges from large pebbles to small cobbles. There are a lesser number of relatively fresh nodules (some slightly battered), mostly of cobble grade but also a few small boulders. As at Portchester there is a separation of these two forms of flint, with the rounded beach flint having been used for the core, whilst the more elongated nodular flint was reserved for the facing.

There is a potential problem with this analysis, however, which relates to the consolidation of the masonry of the fort after its excavation by Charles Green during the early 1950s (Darling & Gurney 1993). The defences have quite clearly been capped with concrete, and there is a possibility that stone was imported to the site during the repairs. Certainly this was the case for the reconstructed road in the fort interior (D. Gurney, pers. comm.). For

Amongst the cobble flint in the core there are a significant number of erratic lithologies, accounting for around 10% of the total stone present in the defences. These include quartz sandstones, calcareous sandstones, shelly limestones, quartzites, lava, and the occasional

40

century date and are brick-built, for older structures the principal building materials employed are indeed flint, and calcareous cementstones (septaria). Flint could have been quarried from the chalk formations to the north and west of Ipswich (Boswell 1927, 94; 1928, 64). However, given the quantities required and the ease of coastal transport, drift deposits on the Suffolk and east Norfolk coast might have been exploited. It is possible that in Walton Castle we find a twin to Burgh Castle, in date, style, and quarries. In the literature the calcareous concretions derived from the London Clay formations are variously described as ‘septarian cementstones’, ‘septarian nodules’ or simply as ‘septaria’. Large accumulations of such cementstones used to be present on the Suffolk, Essex and north Kent coast wherever the London Clay formed low cliffs bordering the shore. Weathered from the cliffs by the sea, the cementstones were collected to supply lime for the ‘Roman’ cement industry during the late 18th and 19th centuries (George 1984; Hewitt 1932; Lamplugh 1908; Woodward 1903). It is reasonable to assume that very great quantities of cementstones were present on the foreshore in Roman times and that immediately local sources were exploited.

this reason, we may be examining the quarrying practices of the Ministry of Works rather than of the Romans. 3.3.4

Walton Castle

Although the ruins of Walton Castle have long since fallen into the sea, sufficient evidence exists to identify – in basic terms – the building materials used in the fort’s construction. Many of these data are detailed in Fox (1911, 289) and Fairclough & Plunkett (2000), and can be summarised as follow: • An 18th century copy of a drawing made in c. 1623, which shows a fort not dissimilar to the present-day remains of Burgh Castle. The defences are depicted with bastions, and with what appear to be brick bonding courses. • An account of the site made by a certain Dr Knight in 1722 in a communication to the Society of Antiquaries of London. This letter stated that the Roman wall was ‘composed of pepple and Roman brick in three courses’ (S.A.L Minutes Vol. 1, 72-3)

The drawing of the site noted above clearly shows that the defences of Walton Castle had been extensively robbed by the 17th century. Allen and Fulford (1999) have claimed to have identified some of this plundered material in two local churches. The first of these, St Peter and St Paul, Old Felixstowe, is perhaps best dismissed. Only a small amount of CBM was present, none of it patently Roman, whilst the cementstones could have been quarried in any era from the Romans onwards and are far larger than any cementstone ashlar seen in surviving Shore Fort architecture. The church of St Mary, Walton, contains material that is more likely to be Roman (and incorporates all of the possibly Roman lithologies identified in St Peter and St Paul, Old Felixstowe). In addition to CBM were blocks of cementstone from the London Clay, tufa, shelly oolite reminiscent of Lincolnshire Limestone, and a lava. This latter rock was examined in hand specimen, and subjected to both thinsection and chemical analysis. Allen & Fulford made the following observations:

• A description of an 18th century cottage known as Felixstowe Lodge, a part of which was ‘entirely paved with Roman brick, taken from an old ruinous castle [at Walton] half a mile distant’ (Raw 1808, Barnes 1809). • At extreme low tide rocks are visible at a distance from the beach beneath the modern cliff line at Walton. Some of these are natural outcrops of septaria, but others have proved to be the remnants of the Roman defences. In 1933 ‘a visit by boat … showed it to be composed of lumps of stone (septaria), Roman bricks, etc. held strongly together with cement or mortar’ (Wall 1937). • An expedition by members of the Ipswich branch of the British Sub-Aqua Club encountered a substantial fragment of wall, including ‘part of a large circular column’ (quite possibly a bastion). It was reported that ‘the wall consisted of pieces of flint like stone joined together by septaria’ together with ‘many red brick layers’ (Errington MS).

This rock is light-to-dark-bluish-grey in colour and friable to hard, depending on the extent to which it is honeycombed by small, irregular, gas vesicles in a sub-parallel, flow-determined alignment. In thin-section, and under chemical analysis, the rocks prove to be finely crystalline, soda-rich tephrites, with occasional, small phenocrysts of green-cored clinopyroxene … The lavas are not British and appear to be a fartravelled import. They compare closely in all respects with Pleistocene lavas (especially the Niedermendig mülstein) from the Eifel in Germany. Allen & Fulford (1999)

The evidence – both antiquarian and modern – indicates that the Roman fort at Walton Castle comprised septarian cementstones and flint, with triple bonding courses of brick, and with semi-circular bastions. As such, the defences of Walton Castle would have resembled those of its flint-built neighbour at Burgh Castle, or the predominantly septaria-built fort to the south at Bradwell. Both flint and septarian cementstones are available in the district, and have been widely used for building during more recent eras. Although the majority of buildings in and around the town of Felixstowe are of 19th and 20th

41

Bradwell (below) demonstrates that it is possible that lithologies such as Lincolnshire Limestone and tufa could have been deliberately imported for the building of the Walton Castle. Even if such stone was employed, however, it was probably only in limited quantities. It is to be expected that the vast majority of the fabric of the fort comprised a combination of flint, cementstones and brick. It is conceivable that a wider survey of the region, perhaps of medieval and early-modern flint-built churches, might turn up more Roman CBM from Walton Castle. However, any association between such material and the Shore Fort would be almost certainly beyond proof.

On the basis of the stone in these two churches Allen & Fulford proposed that sources of stone from an array of far-flung sources were exploited for the building of Walton Castle, both British and continental. Whether any of this material is of Roman origin is far from clear, however. Although Niedermendig lava was imported to Britain during Roman times, it was also widely used during the period AD 700 – 850 (Hill 1981, 117-119). It is certainly doubtful that it would have found use in large quantities as a building material in the Roman defences, except as a chance inclusion. Though the use of lava is improbable, the example of

Figure 29.

Possible sources of building stone, Walton Castle

42

3.3.5

St Peter's Chapel, Bradwell-on-Sea (Bradwell JuxtaMare)

Bradwell

The Roman Fort Greater quantities of stone, and a wider range of lithologies, are evident in the 7th century Saxon chapel of St Peter, built on the site of the Roman fort’s west gate. Nearly all of the fabric of the chapel is Roman brick or ashlar, with the exception of the central sections of the north and south walls, which were rebuilt in the last century using newly quarried stone (RCHME 1923, 1316).

A section of the south wall, less than 2m in length and heavily overgrown, is all that remains of the fort perimeter above ground. The exterior facing is comprised of septarian cementstones from the London Clay and brick, whilst the core was also built from cementstone rubble and mortar. Lewin’s report on Oxley Parker’s excavations of portions of the south wall also described ‘several rows of stones’ interspersed with tile (Lewin 1867, 441-442). The accompanying plan labelled the lithologies in the wall as ‘rubble and septaria’.

Figure 30.

Sources of stone used in the construction of the Shore Fort defences at Bradwell

43

Natural History Museum from the Furfield quarries to the south-west of the town of Maidstone, and to stone from numerous other quarries in this district. Present in lesser quantities is a pale grey glauconitic sandy limestone, some with ferruginous nodules evident. This lithology, also Kentish Ragstone, is likely to originate from the Hythe formation of southern Kent, on the Sandgate foreshore. A few blocks can also be seen of a highly glauconitic calcareous sandstone with a speckled grey appearance. It is much sandier than Kentish Ragstone from Maidstone and north Kent, and contains considerably less calcite. It has been identified as being from basal doggers of the Sandgate Formation, probably from the Sandgate-Folkestone foreshore (B. Worssam, pers. comm.). This lithology is also present at both Richborough and Lympne, and seems likely to have been quarried at the same location.

Present in large quantities was the same buff-coloured calcareous cementstone observed in the surviving Roman wall. Such stone could have been obtained on the coast from Orford Ness to The Naze, around Clacton and Mersea Island, and to the south, between Shoebury Ness and Canvey Island. Sources on the far shore of the Thames Estuary, at Whitstable and Sheppey in north Kent are also plausible (Laver 1907, 408). Three types of Lower Cretaceous glauconitic stone are built into St Peter’s chapel. There are significant amounts of a pale grey glauconitic limestone, composed mainly of crystalline calcite. Some examples are cherty, and large bivalve fossils (Aetostreon) are also evident. This material is highly reminiscent of Kentish Ragstone from the Hythe Beds in the Maidstone vicinity (Worssam & Tatton-Brown 1993). It is a good match to samples in the

Figure 31.

Ecclesiastical and other buildings including stone probably from the fort at Bradwell oolitic limestone, to a nearly pure oolite. In some blocks both the oolitic and the shelly facies can be seen. The stone is Lincolnshire Limestone. In hand specimen the pure oolite compares well to Ketton Stone whilst the shelly oolite is a very good match with Clipsham “Pink Stone”, both in terms of its colour, grain size and shell composition (Judd 1875; Woodward 1894). The modern altar (c. 1960) within the chapel is also of Clipsham Stone, and thus provides a useful comparison. Although these two modern quarries are separated by a distance of

Another glauconitic sandstone is to be found in the chapel, namely Upper Greensand. However, this lithology is only present in the central section of the north wall, and was imported to the site during repairs during the twentieth century (RHCME 1923). Significant quantities of cream or pink tinted limestone are present in the chapel, both as ashlar and as dressed monumental blocks. Various facies are represented, from a micritic limestone, a medium to medium-course shelly

44

this distance from Bradwell robbing from an alternative Roman building is a very distinct possibility. These structures exhibited a similar range of lithologies to those used in St Peter's Chapel, with one exception, a lava, present in the churches at Bradwell and Tillingham. It was described in similar terms to the Niedermendig Lava found in St Mary’s church, Walton, described above, which was suggested to be robbed from Walton Castle. The occurrence of the lava in conjunction with CBM in churches close to Bradwell led Allen and Fulford to assume a Roman origin for the lava. As noted above, however, there could well be other explanations for its presence in medieval structures.

some 12 km it is possible that beds yielding the two facies were accessible in a single location, but it is equally plausible for the stone to have been quarried at several locations. In either case, a provenance immediately to the north-west of the Roman town of Water Newton seems likely. Tufa, a popular building material in Roman times, is also found in the chapel walls. It was used in others of the south-eastern Shore Forts, in significant amounts at Richborough and Dover, and as a trace material at Lympne and Pevensey. Calcareous tufa occurs in a number of locations in south-east Britain, but the harder tufa that is suitable for building is only found in the extreme south of the region, in Kent and Sussex (Pentecost 1993). Of note are the deposits in the Dour Valley, which, as discussed below, were exploited by the Romans for military building work in Dover.

3.3.6

Reculver

Examination of the standing remains of Reculver fort identified a limited range of lithologies (Figure 32; Pearson 2002b). The core of the wall is largely composed of flints, the majority of which are well-worn and heavily percussion-marked cobbles. Although some of the flints retain something of their original nodular form, all show some sign of abrasion and water-wear. The flint is likely to have come from the storm beaches of the Isle of Thanet, between the eastern end of Pegwell Bay and Kingsgate Bay, north of Broadstairs (Shephard-Thorn 1988, 20).

An iron oxide cemented sandstone is present in small quantities, which Allen & Fulford (1999) have suggested is from the Oldhaven Beds that immediately underlie the London Clay in north Kent (Holmes 1981, 39-44). However, examination of the only exposed section of these beds, at Bishopstone, shows the bed to be only a few centimetres thick and for this reason they do not seem a promising source of building stone. The lithology in the chapel at Bradwell (and also in the fort at Richborough) is reminiscent of some less pebbly facies of the Carstone of north-west Norfolk. If a closer provenance is sought, there are localised deposits of well developed ferruginous sandstone (described as carstone) in the Folkestone Beds; however, these are only recorded at inland locations to the north-west of Ashford. No coastal exposures of this lithology are listed in the region (Smart et al. 1966, 54).

A number of flint strata are contained within the chalk cliffs, the most prominent being the “Whittaker 3 inch band”. Eroded in large quantities from the cliffs, the flint could have been conveniently gathered from the beaches at low tide. It is possible that the flints were collected from other storm beaches beneath the chalk cliffs from Kingsdown southwards to Dover. However, it has to be questioned whether it would have been necessary to look even this far afield for flint, and, as detailed below, there is little in the fabric of the Reculver perimeter to suggest much movement of raw materials from southerly sources.

Also present in the chapel is a very small quantity of flint, too battered to determine whether it was nodular or beach worn material.

Occasional water worn chalk cobbles were also present in the core, which would have been gathered from the same beaches. Some unworn nodules of flint were observed, which were probably a by-product of the extraction of chalk for mortar from inland pits. There is a chalk quarry of Iron Age or Roman date at Birchington, Thanet (Fulford, Champion & Long 1997, 166), but whether it was exploited for the building of either Reculver or nearby Richborough, is entirely beyond proof.

Religious buildings on the Dengie Peninsula Allen and Fulford (1999) have identified Roman material in the churches of Bradwell-on-Sea, Tillingham, Dengie, Asheldham, Southminster, and in Stanesgate Priory, Steeple (Figure 31). Further afield, Roman brick is present in the church of St Mary the Virgin, Maldon (Bristow 1985, 92-93), but at

45

Figure 32.

Sources of stone used in the construction of the Shore Fort defences at Reculver the closest source is the eastern foreshore of the Isle of Sheppey, but large quantities would also have been available on the Essex coastline.

In addition to flint and chalk, the core of the defensive wall also contains Thanet Sandstone, a grey, buff or occasionally mauve-coloured glauconitic silty sandstone of Palaeocene age. The formation is exposed in westward-dipping strata between Reculver and Bishopstone Glen, and also in the cliffs of Pegwell Bay (Shephard-Thorn 1988, 26-28; Sumbler 1996, 95-99). The vast majority of the Thanet Sandstone formation is unconsolidated and entirely unsuitable for building. However, in the cliffs to the west of Reculver and also at Pegwell Bay, calcareous doggers of hardened stone can be found near the base of the formation, parallel to the bedding. Such doggers are a relatively rare resource, as their exploitation relies on the erosion of the softer overlying sands (often up to 10m in thickness) by the sea. For this reason, only small quantities of material are available from a single location at any given time.

Of the facing that has survived, virtually all is a hard, occasionally cherty facies of Kentish Ragstone, which would have been quarried from the Maidstone-Sevenoaks district. There was one block of tufa on the south wall, roughly shaped, which was probably also once part of the facing. The monumental blocks used to build the South Gate of the fort were of glauconitic calcareous sandstone, also Kentish Ragstone, which could either have been quarried from the Sevenoaks, or the Folkestone district. Allen and Fulford’s examination of the Saxo-medieval church of St Mary built within the fort, identified flint, Thanet Sandstone and Greensand. In addition, occasional blocks of hard tufa, and an iron oxide-cemented sandstone were seen, the latter suggested to be from the Oldhaven Beds. Whether these additional lithologies were employed in the perimeter wall is uncertain – the presence of tile clearly indicated that some of the fabric of the church was not robbed from the fort perimeter. Much would have been taken from internal buildings, although Peacock has also raised the possibility that the tile may have been robbed from an earlier naval installation in the locality (1977, 238-239).

Modern exposures of these doggers can be seen at Pegwell Bay, and such material was employed in Roman structures at Richborough, and the nearby Abbey Farm villa. It is also found in Jutish and Anglo-Saxon burials, where the slabs readily adapted to serve as an outer coffin, or as a frieze or fence around the grave (D.Perkins, pers.comm.). Calcarious cementstones are found in small quantities in the core of the east wall. Derived from the London Clay,

46

3.3.7

Marquise Limestone, an oolitic stone quarried from an area 10 km north-west of Boulogne (Worssam & TattonBrown 1990). It is found as small rubble lumps in the core of the perimeter wall, whilst much may have been burned for lime.

Richborough

It is established in Chapter Four that recycled masonry from earlier Roman structures on the site could have accounted for as much as 70% of the fabric of the Shore Fort’s perimeter (5.3). The principal source of re-usable stone on the site was the Great Monument, which by the later 3rd century was already in a state of ruin. Although the structure was demolished to foundation level to make way for the Shore Fort’s internal buildings, enough fragments have survived to understand the types of stone from which it was built (Strong 1968). The excavation of the fort has also given a reliable picture of the types of stone used in other buildings within the interior (BusheFox 1926; 1928; 1932b; 1949; Cunliffe 1968).

Ditrupa Limestone, a richly foraminiferal limestone from the Valois-Soissons region, some 25 km to the northeast of Paris (Worssam & Tatton-Brown 1990). This lithology is present in small amounts as petit appareil on the exterior face of the Shore Fort’s north wall. Caen stone, a cream-coloured, very fine-grained limestone from Normandy. Only a few blocks of this material can be observed, the vast majority being on the exterior face of the north wall, to the east of the north postern gate. It is used, along with the ferruginous sandstone described below, to create a striking decorative course along this short section of wall (Figure 43). Caen stone was a much favoured building material in Britain during the Norman period and after, but the author is aware of no other instances of its use in Romano-British architecture. Other contemporary examples may exist but this is, at the very least, a rare occurrence.

The lithologies present in the Shore Fort defences can be divided into three categories. It is important to distinguish which material types fall into which category, if the supply situation in the late 3rd century is to be properly understood. Some of the stone types discussed below under Headings 1 and 2 may have been robbed from a variety of existing buildings; however, all were present in the Great Monument and some were exclusive to this structure.

Marble. The Great Monument was covered with a veneer of white marble, mostly imported to Britain from the Luna-Carrara quarries, together with some Pentelic marble from Mount Pentelikon, Attica (Strong 1968, 64). Small amounts of marble are visible in the core, particularly around the west gate, but it is to be suspected that much of this valuable veneer was recycled in finer buildings further afield than Richborough. A more mundane alternative is that, along with Marquise Limestone and other limestones, most was burnt for lime.

Lithologies obtained only from the robbing of existing structures on the site of the fort. The geological origin of these lithologies, though of interest, is not of central importance to this study. At the time of the Shore Fort’s construction they were effectively quarried from the building site itself, and required practically no transportation. It is possible that some of these stone types might fall into the second category, the robbed material having been augmented by the quarrying of new supplies. However, the quantities involved are easily small enough for the Monument to account for all of the local materials. The lithologies present are as follow:

A ferruginous sandstone, similar to that found in small quantities at St Peter’s chapel, Bradwell, is used in the north wall at Richborough as part of a decorative band. Allen & Fulford (1999) suggest that this lithology is from the Oldhaven Beds of the north Kent cliffs to the west of Reculver. However, as discussed above (3.3.5), it is perhaps more likely to be a facies of the Lower Greensand around Folkestone.

Lower Greensand, thought to have been quarried from basal Sandgate Beds doggers in the vicinity of Mill Point, Folkestone (B. Worssam, pers. comm.). Although these doggers are now rarely exposed above the tide at this location, a similar lithology can be observed in the sea wall above the beach. A number of monumental blocks of this material were reused in the foundations of the west gate at Richborough (see Bushe-Fox 1926, 30 & Pl. xxxv), whilst rubble of a similar lithology is also found in the perimeter core.

Lithologies robbed from existing structures on the site, augmented by quantities of newly quarried stone of the same type. In this category are the two lithologies that form the bulk of the perimeter wall – flint and chalk. Although calculations make it clear that reused material cannot account for the entire core of the Shore Fort defences, it is impossible to determine how much is robbed stone, and how much was newly quarried material.

Tufa. Present in limited quantities on the interior face of the north wall. The tufa has been carefully shaped, blocks which probably directly reflect their form in the monument. The provenance was perhaps in the Dover area, but a number of sources in Kent or Sussex are plausible (Pentecost 1993).

The flint assemblage at Richborough is comparable to that at Reculver, in that it comprises mostly well waterworn cobbles together with material that shows a lesser

47

A considerable amount of the material present was recycled, much of it doubtless derived from the preceding classis Britannica structures on the site. For this reason, no source map is offered here for the building materials used in the Dover Shore Fort.

degree of abrasion. As for Reculver, the most plausible source of such beach flint is the Thanet coastline. Examination of the mortar at Richborough shows that much of the chalk quarried for the Shore Fort was burned for lime, but a considerable quantity is found as rubble in the core of the wall. The stone is Upper Chalk, and the Thanet cliffs are likely to have been the source. Much of the rubble displays ‘Pholas’ and Polydora borings (see Yonge 1966), indicating that it was collected from a beach. Some chalk rubble and fresh flint nodules and are also occasionally found in the core, suggesting that some chalk was dug from pits, perhaps immediately inland from the coast.

The sections of wall encountered by recent excavations were mostly comprised of chalk, tufa and flint, all of which was present in the earlier CL BR II fort (Philp 1981, 20-23). Amos & Wheeler also recorded ‘concrete, tufa, flint, tiles, iron slag and discarded sculpture’ in the core-fill of the Shore Fort wall (1929, 49). A green sandstone, probably from the Hythe or Folkestone beds, was also present in the core, a large block of which (originally part of the ‘Painted House’) is incorporated into ‘Bastion F' of the Shore Fort (Philp 1989). Unfortunately it was not possible to examine this lithology in any detail.

Lithologies newly introduced to the site specifically for the building of the Shore Fort. A number of raw materials that had not been employed in preceding structures at Richborough were quarried during the late 3rd century, specifically for use in the Shore Fort defences. In addition to the lithologies detailed below, we might also add the beach stone used as the aggregate in the core mortar, and the CBM that was employed as the medium for construction of the bonding courses.

The large quantities of chalk could have been easily obtained from the nearby cliffs, which represent a vast quarry; the stone could have been dug out from pits, or gathered from the beach. None of the limited amount of stone viewed showed any indications of having lain on a beach for any significant time, so the former option is preferable. Chalk was also used for the lime mortar. The aggregate was coarse, and included flint fragments and water worn pebbles up to 2cm in diameter (Wilkinson 1994, 73).

The newly introduced lithologies seem to be those present in the facing of the south and west walls. The lowest two courses of these walls, rather larger than those above and forming a plinth, are built from Kentish Ragstone from the Maidstone district. As for the building of Reculver, calcareous doggers of Thanet Sandstone were exploited, being used for most of the facing of the west and south walls. No Upper Greensand was identified (contra. Allen & Fulford 1999).

Tufa was also available locally, though only in limited amounts. It was a favoured building material, which was widely employed in the vicinity during Roman times, most notably in the Pharos now encompassed within the medieval castle of Dover (Wheeler 1929). The use of tufa continued into the Norman period (Amos & Wheeler 1929, 53). In addition to deposits in the Dour Valley and elsewhere in Kent and Sussex (Pentecost 1993; ShephardThorn 1988, 33) tufa has also been exposed during excavations within the town of Dover itself, one location being the Victoria Hospital. The deposits are only a few feet below the modern ground surface, allowing it to be quarried out from relatively shallow pits. These are limited in extent, however, and their discovery would be more a matter of chance than of design. Once encountered, a deposit would probably be completely quarried out; the precise provenance of the material for the Shore Fort would depend on which sources were known at the time of construction. Most if not all of this valuable stone present in the Shore Fort was probably recycled from the earlier classis Britannica fort, demolished to make way for the new installation.

Septarian cementstones were also present, presumably quarried from the London Clay formations on the eastern foreshore of the Isle of Sheppey, or a little further to the west at Whitstable. However, it is conceivable that such cementstones could have been brought from further afield, from Essex or Suffolk although this would have entailed a considerably longer coastal journey, or the use of ships suitable for short crossings of open sea. 3.3.8

Dover

The defences of the Dover Shore Fort are principally composed of chalk, flint and tufa, with some ceramic material used for the bonding courses in a number of the bastions (Philp 1981; Wilkinson 1994). With the exception of the CBM, all of the raw materials could have been obtained from sources in the immediate vicinity.

48

Figure 33.

Sources of stone used in the construction of the Shore Fort defences at Richborough

Of all the raw materials in Shore Fort, the flint is the most likely to have been freshly quarried at the time of building. It seems to have been present only in small amounts in the CL BR II fort wall (Philp 1981, 20-23) although it could have been obtained from other structures within the old fort. The flint is nodular and unabraded, and there are no indications that it was ever in a beach environment. Flint is present inland, in the Head on the Roman margins of the River Dour, but more concentrated amounts could be found in Dry Valley and Nailbourne deposits, a number of which open onto the Dour (Shephard-Thorn 1988, 33-36). In either case the quarried material could have been brought down-river by boat to the site of the Shore Fort. The fact that the flint was not procured from the beach is somewhat curious, since it was present in large amounts on the local beaches. Records of the building of the medieval castle at Dover in AD 1354 documented the transport costs for no less than 3464 cartloads of flint from the shore (Clifton Taylor 1972, 191). 3.3.9

The principal stone types present in the defences are glauconitic sandstones; these are almost the only lithologies that are employed in the facing, and also account for most of the rubble core. All of this stone has previously been assumed to be Kentish ragstone, quarried from immediately local sources, a logical conclusion given that there are numerous outcrops of the Hythe Beds in the vicinity of Lympne (Worssam & Tatton-Brown 1993). Geotechnical investigations of the degraded cliff on which the fort was built revealed the presence of a level plateau of Atherfield Clay, immediately below the crest of the escarpment (Hutchinson et al. 1985, 213214). This was interpreted as a quarry bench, which the authors suggested was created by exploitation of the overlying Hythe Beds for the building of the Shore Fort. The present examination of the fort has reached similar conclusions, with some qualifications. Most of the stone was indeed Kentish Ragstone, a hard, grey, limestone that was very sparsely glauconitic. As such, it could have been quarried from any part of the Hythe Beds between Lympne and the coast at Folkestone (B.Worssam, pers. comm.). It nevertheless seems very likely that most was obtained very close to the fort, for there are outcrops yielding an identical lithology only a few hundred metres distant from the ruins of the defences. Interestingly, however, the find of several water-worn boulders in the core of the wall indicates that at least some Kentish Ragstone was quarried on the coast; a few of these boulders displayed borings of the modern sponge Cliona

Lympne

The Shore Fort at Lympne exhibits a limited range of building materials, all of which appear to have been derived from local sources (Figure 34). A considerable amount of reused stone and CBM have been employed, and it has often been assumed that it was recycled from an earlier coastal fort near to the late 3rd century installation (e.g. Cunliffe 1980a, 285).

49

(Yonge 1966), whilst one example still had barnacles attached. This water worn material was seen in the eastern portions of the perimeter wall, particularly around

Figure 34.

the east gate, and also on the eastern parts of the north wall.

Sources of stone used in the construction of the Shore Fort defences at Lympne The Roman Landscape is based on Cunliffe (1980, Fig. 6.1) have been used in the Folkestone villa (Williams 1971, 174) and at Pevensey (3.3.10).4

Also present in the wall were limited quantities of two other lithologies that were quarried from the Folkestone coast. The first of these is Folkestone Stone, most of which is a grey, fine-grained calcareous sandstone, with occasional inclusions of quartz grains up to 1 mm. A rather flaggy, pale grey medium grained glauconitic calcareous sandstone is found as large slabs in the core of the surviving sections of the wall on the north-west. This lithology has also been identified as Folkestone Stone, quarried from a similar location, and is particularly evident in the core of the north-west portion of the fort.

3.3.10

Pevensey

The geology of East Sussex is dominated by rocks of Cretaceous age, which together make up the Wealden formations (Gallois 1965). It is from these strata that many of the building stones originate (Figure 35). As at many of the other Shore Forts the most important material in terms of volume is flint, although at Pevensey this material is used only in the core. The vast majority of flints in the fort wall are virgin nodules that show little or no sign of erosion or impact; only around 10 per cent are abraded beach cobbles, distributed evenly amongst the nodular material within the core. Plausible sources for the flint are the Dry Valley and Nailbourne deposits of the South Downs, for example those visible in the cliff section at Birling Gap and found immediately inland around Eastdean and Westdean (Gallois 1965, 62-63). Within these extinct water courses, flints of a comparable size to those in the core at Pevensey are found in concentrated deposits, making extraction a more economical proposition than had they been quarried directly from the chalk, where flints make up only a small proportion of the total volume of material. The presence of water-worn cobbles in the perimeter wall suggests that the quarried flint was brought to the coast where it was dumped on the beach to await collection by ship. The

The second lithology is mainly present around the east gate, in part as core rubble but most notably as large blocks forming a platform underneath the flanking bastion. It is a highly glauconitic, fine-grained calcareous sandstone, with rare 1mm quartz grain inclusions. This lithology has previously been identified as Upper Greensand (Cunliffe 1980a, 236) but is in fact Sandgate Sandstone, from the basal Sandgate Formation, probably from doggers on the Sandgate-Folkestone foreshore. It bears a very strong resemblance to some of the ashlar in St Peter's chapel, Bradwell, as well as to the reused stone in the foundations of the west gate at Richborough. The slabs at Lympne are also clearly reused (Cunliffe 1980a, 236-237). Three blocks of hard tufa were present in the south-west parts of the perimeter wall. This find adds to the evidence for the movement of small quantities of tufa along the south-eastern coast of Britain, which is also known to

4

I am grateful to Dr. Bernard Worssam for his identification of a number of samples from Stutfall Castle.

50

is also evidence for quarrying of this stone a little inland from the coast. In 1878 a bell pit dug for the extraction of the greensand was discovered, close to the site of the Eastbourne Roman Villa (Holden 1975; Sutton 1952). Although unlikely to be of Roman date, the presence of this pit does give some impression of the manner in which quarrying has taken place in the past.

cobble flint would have been gathered up and transported along with the quarried material. The most widespread lithology in the facing is Upper Greensand, a calcareous glauconitic sandstone. Predominantly pale green in colour, it accounts for the majority of material found on the inner, and the exterior face, including the plinth. An outcrop of Upper Greensand can be found in a narrow outcrop on the Eastbourne coastline and, although there is no evidence of Roman quarrying, this material has been exploited for building stone in more recent times (Lake et al. 1987, 93). 13th century building accounts of the medieval keep at Pevensey record the quarrying of a great deal of this stone, which can also be found in local medieval churches such as Westham and Beddingham (White 1926, 84). In the present day the outcrop near Eastbourne is only exposed at low tide, a fact that perhaps does much to explain the limited use this stone has found even within the local area. The extent to which the quarry would have been accessible during the Roman period is uncertain but the slightly lower sea levels of the time (see Devoy 1979; Waddelove & Waddelove 1990) might have made exploitation of the outcrop somewhat easier. There

Figure 35.

At least some of the Upper Greensand used in Pevensey Roman fort is from the coast; many of the plinth blocks are water-rounded boulders, whilst a few petit appareil in the exterior face have recent marine borings. The large slabs employed in the bonding courses were probably too large to have been quarried from bell pits; the aperture at the surface of such a pit is normally very small, and the lifting of such heavy slabs would have been very difficult. It is more likely that the bonding course slabs were levered up from the Eastbourne foreshore; some may well have marine borings on their upper surface, but these would not be apparent once the slab was embedded into the fort wall. There is no reason to assume that much of the petit appareil could not have been dug from inland pits, however.

Sources of stone used in the construction of the Shore Fort defences at Pevensey stone was quarried at nearby Chilley which in Roman and medieval times would have been a small island, slightly elevated above the water level (Mantell 1833, 172-73). Whether Roman quarrying did actually take place at Chilley has to be regarded as uncertain, since Mantell’s

Wealden sandstones from the Tunbridge Wells beds have also been used as a facing material. These rocks, ranging from fine-grained buff sands to much coarser ferruginous examples, form a band of colour in the middle courses of the exterior face. It was once suggested that some of this

51

the area, and visible to the north-west of the fort. A number of old workings are present in this locality, which Cunliffe thought might be associated with quarrying during the Roman period (1975, 20). Chalk has been dug in numerous pits across the Hampshire chalk downs in the past, and many would have had to be opened to supply the vast amounts of flint needed for the building of Portchester.

description of ancient finds made at this location is very brief and of dubious accuracy. A local origin for these sandstones remains probable, however, since similar facies can be found at other outcrops within a 5 km radius, for example at Garden Cottages and Lusteds (Lake et al. 1987, 32). Quarrying has taken place immediately to the west of Pevensey, at Westham, but this cannot have been the source of the Roman material since the area would have been under water until the Levels were drained.

A smaller, though significant proportion of flint was brown or orange-stained nodules, many heavily frostshattered, indicating that they had lain on the land surface for a considerable length of time. Many of these flints could have been derived from deposits of clay-with-flints, perhaps ploughed up during agricultural operations and cleared to the field boundaries. Over time, ‘plough cairns’ built up by this process can become quite large, and judging by the quantities present at Portchester Castle, collection of these deposits accounted for a good proportion of the flint required. English Heritage still uses plough cairns as a source of brown flints for restoration work (F. Powell, pers. comm.).

Chalk was required in large amounts, both for use in the foundation trench and also as the source of lime required for the production of mortar. The chalk used was white, with a compact structure that contained no shells or fossils. Exploitation of identical facies from the Middle and Upper Chalk formations of the South Downs has been widespread in the past and this area was undoubtedly the source of the material for Pevensey. Numerous pits have been quarried across the South Downs, a list of those still operational in the 1830s including Holywell, Alfriston, Piddinghoe, Beddingham, Lewes, Offham, Preston, Falmer and Southerham (Mantell 1833, 408).

In addition to flint and chalk, a number of other lithologies were present in the perimeter structure, mostly, although not exclusively, in the plinth- and bonding courses. A large number of these lithologies were of Tertiary age, and prominent among this group were facies of the Bembridge Limestone. The Bembridge Limestone is exposed only on the Isle of Wight, mostly along the north-eastern coast, and consists mainly of freshwater deposits (fluviatile, marsh and lacustrine) together with two deposits representing brief marine incursions (Insole et al. 1998; Melville & Freshney 1982, 105). A wide range of facies of the Bembridge Limestone had been used in the Roman phases of the fort walls, including significant quantities of a grey ripple-laminated micritic limestone, found in the bonding courses of the south wall. Various packstones and peloidal limestones were also found, described individually in Appendix IV.

Horsham Stone, a thinly-splitting, laminated calcareous sandstone, is used as a bonding course along a small stretch of the north wall. Greenish-grey in colour, this lithology has been widely used in the region for paving, roofing and roadstone (Gallois & Worssam 1993, 111). There are a number of outcrops around the town of Horsham, from which it derives its name, the majority of which have been exploited in the past. Although no Roman road runs directly past any potential quarry site, the material would only have had to be transported a few kilometres before reaching a major highway. Nevertheless, its presence in the Pevensey defences is curious, as its provenance contrasts with the other stone types present, being from a somewhat more distant, inland source. 3.3.11

Quarr stone, another member of the Bembridge formation was also present, a creamy white packstone, weathering towards grey (Bishop, forthcoming; Jope 1964). The lithology is composed of comminuted shells, and has been described as a “shell-brash” or “featherbed stone” on account of its texture. It is also characterised by clam moulds, between 1 mm and 6 mm in length, and is sometimes bioturbated. The coarseness of the lithology, the mould size and the percentage of micrite often vary within a single ashlar- or bonding course block. In occasional blocks the Quarr stone could be observed grading into micritic facies, more typical of Bembridge Limestone proper (also termed Binstead Limestone). Quarr stone outcrops in a limited area in the north-eastern parts of the Isle of Wight, particularly in the inter-tidal zone around Binstead, Ryde and Quarr. Many of the blocks used in the plinth were heavily bored, indicating a beach quarry. Quarr stone was in great demand during the

Portchester

Flint and chalk represent the dominant lithologies in the Roman phases of Portchester Castle (Figure 36). In contrast to every other Shore Fort, all of the flint appears to be from inland sources. The majority of flint was virgin, white-cortexed, some still with chalk filling the cavities in the nodules. Most of the chalk quarried for Portchester was burned for the production of lime mortar, but rough lumps can also be seen in the core of the wall, particularly on the south wall. Neither the flints nor the chalk showed any sign of water wear or abrasion. The source of both materials was almost certainly local. Upper Chalk is the dominant geological formation in the region, in which numerous bands of flint are present. The closest source of such flints would have been the chalk of uplands above Portsdown, a major east-west feature of

52

Limestone, it does seem likely that all the Tertiary limestones were quarried from the Isle of Wight. Upper Greensand was employed in the Landgate, Watergate and both posterns (Cunliffe 1975, 11 & 2936). It is also found as rubble in the core, particularly on the south side of the fort. An iron oxide-cemented sandstone, also from the Upper Greensand, was used in the Watergate and a course of rubble facing, possibly Roman, is also present on the south wall between bastions 14 and 15 (Cunliffe 1975, 20; see Appendix I for location plan). Upper Greensand outcrops on the Isle Wight as a narrow exposure running east-west across the centre of the island (Insole et al. 1998). Quarrying of this fine sandstone in the past has taken place along the coast, from the undercliff between Bonchurch and Blackgang, and inland at St Catherine’s Down, St Martin’s Down, and at Gatcliff (Hewitt & Vellacott 1912, 463).

late Saxon and Norman period for fine architectural work, with the result that by the 1120s most quarries and coastal exposures had been exhausted (Jope 1964). Quarr stone has been neglected in the geological and archaeological literature. Some architectural studies have noted its presence in medieval church building (Jope 1964) but its occurrence at Portchester represents the only known use of this lithology for building during the Roman period. Other stone examined in the perimeter was also of probable mid-Tertiary age, and could also have formed part of the Bembridge Limestone. However, the interpretation of these lithologies was less certain in hand specimen, and only a broader provenance within the Hampshire Basin can be proposed (B.Sellwood, pers. comm.). Given that so much of the stone used in the bonding courses can be identified with the Bembridge

Figure 36. 3.4

Sources of stone used in the construction of the Shore Fort defences at Portchester

CBM, mortar and timber few shiploads at most. As a consequence, it is quite possible to entertain the notion that brick could have been supplied from production centres that were distant from the forts.

Although a detailed analysis of building materials other than stone has not been undertaken in this study, a brief discussion of the likely provenance is possible. This will be of particular relevance to Chapter Five, where the transport and supply process supporting the Shore Forts’ construction is examined (5.3). 3.4.1

Although the army turned increasingly to civilian producers for its pottery from the 2nd century onwards, tile manufacture, at least in the military zone, appears to have remained in the hands of the State (Jones & Mattingley 1990, 217). Certainly at Brancaster and Reculver there is evidence for military tileries having supplied the forts with some CBM, albeit mostly for the

Ceramic building material

As will be seen in 5.2.3 the quantities of CBM used in the building of the Shore Fort defences was never great, a

53

production might have been carried out in the district of Brampton, where a major group of pottery kilns has been detected: it would have been an easy prospect to ship the brick down the River Bure to Burgh Castle (a distance of some 40 km).

internal buildings, since neither site employs brick in the defences elsewhere than the gates. In the south-east of Britain, the association of the classis Britannica with brick and tile production is well known (Cleere 1977). Peacock’s (1977) study of stamped tiles of the classis Britannica demonstrated that there was likely to have been an important production centre in the Fairlight district (‘fabric 2’). These tiles were present at a number of sites in the Weald and along the Sussex and Kent coast, their distribution extending as far as the fort at Richborough some 70 km distant (Peacock 1977, Fig. 2). However, the disappearance of the classis Britannica from the archaeological record around the mid 3rd century (coincident with the demise of local iron manufacture on a significant scale) must lead to doubts as to whether brick production in the Weald continued on any great scale. Certainly ‘fabric 2’ was only found in the walls of Pevensey in very small quantities, which led Peacock to suggest that these few bricks had been reused from an earlier classis Britannica base on or near the site of the Shore Fort (1977, 246).

The above discussion is far from satisfactory. However, given the difficulty of establishing a provenance of CBM (see 1.2.3), it is probably the best that can be hoped for without extensive petrological analysis. We can conclude that in many cases brick could have been produced within a few kilometres of the Shore Forts, if not on the site itself, but the distances could equally have been much greater – it is, for example, 125 km from Burgh Castle to Colchester. Although Peacock’s study only dealt with a limited number of tiles and fabrics, it may at least suggest the normal distances over which CBM might have been transported. The distribution of ‘fabric 2’ CL BR tiles does not appear to extend to the north Kent coast, nor is it found on the south coast to the west of Pevensey. However, this is only one example and it can be demonstrated that other commodities, including pottery (Tyers 1996) and in the present study, building stone, could be transported much greater distances. We must not rule out the possibility that CBM was being supplied from distant sources, which could have included the near continent. Boulogne, for example, is known to have been a centre for classis Britannica tile production (Peacock 1977).

Finds of actual tile kilns in Roman Britain are quite rare: McWhirr (1979) listed some 51 sites, and the identification of some of these as tile- rather than pottery kilns is not certain. This is doubtless only a small proportion of the total number in the province, and the overall distribution as known from the archaeological record may well not be representative. However, as noted by Jones & Mattingley (1990, 217) the requirements and constraints of tile production were essentially the same as those for pottery manufacture, principally the presence of suitable clays. A number of tile kilns occur on known pottery-producing sites, and it is likely that this association was quite common. It has been suggested for at least one Shore Fort, Pevensey, that the brick was produced on the building site itself (Salzman 1907b, 251).

3.4.2

Timber

It would have been possible for much of the timber needed for the Shore Forts’ construction to have been procured close to the building site: certainly in the modern day there is extensive woodland close to many of the forts. That this was the case in the Roman period is supported by the material evidence from our sites: excavations at Bradwell, for example, turned up many bones of woodland creatures such as deer and boar (Lewin 1867, 443). The study of pollen sequences suggest that southern Roman Britain was more heavily wooded than at present (Dark & Dark 1997, 31-33), and therefore it is reasonable to expect that procuring timber from the local area would not have posed too great a problem.

The Victoria County Histories shows that brick production was a very important industry on the south coast of Britain during the 19th century, particularly in Hampshire (Hewitt 1912; 1932; Salzman 1907). Essex was another large-scale producer of brick, much of it destined for London (Christy 1907). It is in these counties that many of the known Roman tile kilns have been found: there are, for example, a cluster of kilns some 15 km to the north-west of Portchester. In Roman times Colchester was a major centre for brick manufacture, and if civilian production sites were being used to supply the Shore Forts, then this particular town might well have been involved.

It would, however, be wrong to entirely rule out the possibility of more distant supply. Whilst the timber could have been procured from the clear-felling of isolated local coppices, densely forested areas might have offered a more efficient option for the Roman commander. The exploitation of the Weald during the medieval period is well documented: timber was a major export of the Cinque Ports (including Pevensey), particularly during the 14th century (Pelham 1928). It is quite possible that the Weald was used to supply timber to Lympne and Pevensey, and perhaps others of the south-eastern Shore Forts.

The county of Norfolk is rather less associated with brick manufacture, both in the Roman period and in more recent eras. The forts at Brancaster and Caister used little or no brick in their defences, but the source of the brick for Burgh Castle is an interesting question. CBM

54

3.4.3

stone from more distant sources. Although we shall see below that the volumes of these longer-travelled lithologies were not great (6.2.3), their presence is nonetheless extremely significant. Here we have evidence of complex patterns of supply: in the following chapters an attempt will be made to explain the nature of these arrangements, and it will be argued that this stone is a proxy for the movement of other commodities during the building programme. In addressing this question, we must also seek to understand why there is no evidence for similar long distance supply at our other sites. The Shore Forts westward from Dover appear to have been built from nothing but local stone, but we do not have a simple East Coast-West Coast divide, for Brancaster, Caister and Burgh Castle also lack any far-travelled stone.

Mortar

The principal components of non-hydraulic Roman mortar were limestone (burned in kilns for lime), an aggregate, and sand (Adam 1994, 74). Examination of the mortar in the Shore Forts shows that various limestones were exploited. At Pevensey, Portchester, Dover and Richborough we find chalk being used, and it seems likely that it was also employed in the white mortar recorded by St Joseph at Brancaster (1936, 450). Inspection of the surviving stretch of wall at Bradwell suggests that septarian cementstones provided the lime for the mortar, whilst Kentish Ragstone was used at Lympne, almost certainly quarried from the outcrop above the fort. The burning of all of these limestones in more recent times is well-documented (George 1984; Mantell 1833; Worssam & Tatton-Brown 1993). All of the lithologies being exploited were potentially available within 20 km of the building site.

3.5.2

In the Shore Forts, we also have a rather different view of quarrying practices to that normally found in the literature on Roman Britain, where what little has been written about the stone industry has tended to concentrate on extraction from inland sources (see Blagg 1990). However, only at Portchester and Lympne do we have evidence of large-scale exploitation of inland deposits: for the remainder of the Shore Forts it was the coast that provided the vast majority of raw materials.

The aggregate in the mortar normally consisted of small beach pebbles, and the coastal location of the Shore Forts would have allowed them easy access to large deposits of such material. Shingle barriers would have offered concentrated deposits of just such material, and certainly it seems that the pebbles used at Richborough were collected from the bank blocking the southern outflow of the Wantsum Channel (Robinson & Cloet 1953, 77). Sand too would have been available locally on the coast. Although Vitruvius (II.v.7) cautions against the use of sea sand for mortar, locations that were above the normal tidal reach and which were exposed to rainfall (dunes, for example) would have yielded sand with an acceptably low salt content. 3.5 3.5.1

Coastal quarrying

Much stone was procured directly from loose deposits on the beaches themselves, such as flint and septarian cementstones. These did not require quarrying, but could simply be picked up and loaded into the awaiting transport. This is evidence, therefore, of an extremely opportunistic – and highly efficient – practice of gathering stone. Beach platforms were another important resource, providing materials such as Thanet Sandstone, Upper Greensand and Bembridge Limestone. These flaggy lithologies often proved useful in specialised functions, particularly as bonding courses. Often they may not have required cutting from the rock face: rather, they could be levered up from the beach platform, once again a very efficient quarrying method.

Conclusions The pattern of supply

The findings of this chapter indicate that most of the building stones present in the Shore Fort defences were procured from local sources. In some cases quarrying was carried out on the building site itself, in the form of the demolition of existing buildings, whilst natural outcrops very close to the forts were also exploited, as was notably the case at Lympne. More often, however, the Romans had to cast a little further afield, but most of the stone appears to have been obtained no more than 30 km of the fort (see 6.2.3). These findings are supported by our brief examination of the sources of CBM, timber and the mortar components, which demonstrates that these raw materials could also have been procured relatively close to the building site.

Most other lithologies, even if not procured directly on the coast, were present in outcrops that were close to a navigable water source. For example, the Kentish Ragstone quarries that supplied stone to Bradwell, Reculver and Richborough all lay within easy reach of the River Medway. In more recent times it has been the practice at locations such as Allington for freshly cut stone to be loaded directly into barges drawn up on the foreshore, only a few tens of metres from the quarry face (Worssam & Tatton-Brown 1993). We can envisage a similar scenario for chalk extraction: the ancient pit at Birchington, Thanet, for example (see 3.3.6), lay within 0.5 km of the margins of the Wantsum Channel, directly across the water from Reculver. Mantell’s list of 19th century chalk pits in Sussex (1833) shows many to be located on the coast, and most others on navigable rivers.

These conclusions should not come as a surprise: considerations of economic efficiency doubtless ensured that an attempt was made to gather stone from locations close to each fort. Of more interest, however, is the presence at Bradwell (and possibly at Walton Castle) of

55

found on the Folkestone foreshore, where it appears that quern production took place on a significant scale (Keller 1989). Coincidentally, it is from Folkestone that a number of lithologies present in the Shore Forts at Lympne, Richborough, Reculver and Bradwell were derived, namely Folkestone Stone, Kentish Ragstone and Sandgate Sandstone.

The nature of most Roman coastal stone extraction will mean that it has left little identifiable trace in the modern landscape. The exploitation of loose deposits or beach platforms does not produce quarry faces, whilst in any case erosion will often have destroyed the actual locations from which the stone was gathered. A little evidence does survive, and some of the best has been

Figure 37. Doggers of Thanet Sandstone on the foreshore at Reculver Bay These have been left exposed after coastal erosion has removed the unconsolidated sands that form the overlying cliffs. Oppotunist exploitation of such deposits accounted for a great deal of the quarried stone supplied to the Shore Forts, in this case to Reculver and Richborough When we come to quantify the amount of building stone present in the Shore Forts (Chapter Five), it is clear that the coast was a resource that was capable of yielding raw materials on a massive scale. Coastal quarrying has been a widespread activity in more recent times, as is evident in the early series of British Geological Society memoirs. Examples can be cited from all around Britain, such as in Devon, South Wales and Scotland, to name only a few instances (Cantrill et al. 1916; Dixon et al. 1921; Woodward & Ussher 1911). Often, as was the case at Aberthaw (South Wales), the industry only declined when the advent of railways made the exploitation of inland sources a viable economic prospect (Strahan & Cantrill 1904).

Harwich and Felixstowe alone (Laver 1907, 410). Roman practices were likely to have been very similar to those of the 19th century, and by way of illustration, we can turn to John Fowles’ account of the history of Lyme Regis: .. a new local industry did arise in the 1820s. This was sea-quarrying … The toughness of the workmen, known locally as stoneboatmen, has become legendary. They developed a special double-hulled boat, rowed by two long sweeps, called the stoneboat, and akin to the Portland lerret. Having broken the ledge-stone at low water, they then rowed it back to the Cobb … Notoriously independent, these titans worked in all conditions, often up to the waist in water, and of course needed great physical strength. One did not lightly start a quarrel with a stoneboatman.

One only has to examine the 19th century ‘Roman Cement’ industry to appreciate its economic importance: an Admiralty report of 1843 estimated that since 1812 over one million tons had been taken from the beaches of

Fowles (1991, 48-49)

56

Roman structures in the coastal zone exploited similar resources of stone. Coastal quarrying, if perhaps lacking the glamour of other ancient industries, can nevertheless be shown to be an activity of major economic importance, and one worthy of more emphasis in future literature on the Romano-British economy.

Stone at Lyme Regis was also collected from tumbled masses on the beach (Woodward & Ussher 1911, 83). As was the case with the ‘Roman Cement’ industry, only when beach deposits were exhausted, or collection from the beach was prohibited due to the amount of damage being done to the cliffs by quarrying, were stones dredged from the shallow coastal waters (Laver 1907, 410). It seems probable that the Romans were able to gather their stone directly from the beaches, but we cannot rule out the possibility of dredging.

Such a discussion, however, is for the future. Quarrying was only one part of the process that led to the creation of the Shore Forts and although we have established the sources of the stone used to build the forts, we have yet to put them in context. To do so we must try to assess the quality of each stone type, and the quantity in which they were supplied to each building project. It is to these, and to many other issues relating to the building of the Shore Forts, that our study now turns.

In the light of the findings of this chapter, and other work on the physical settings of the monuments (see Pearson 2002a), it is possible to see that the coastal locations of the Shore Forts were utilised to the full during the building process. We must also expect that many other

57

CHAPTER 4. DESIGN, MATERIALS AND ARCHITECTURE 4.1

circumstance, materials, supply or even to the individual builder. Section 4.4.3 investigates how architectural designs were transmitted in the ancient world, and asks whether the differences between the monuments’ defences may be the result of differing interpretations of the same contemporary instructions. It then looks at two specific features of the Shore Forts’ architecture, the ‘bonding courses’ and opus signinum mortar, examining the reasons for their presence and the building traditions from which they may have sprung.

Introduction

The generating process, by which any structure is created, can be broadly divided into three parts: design, the supply of raw materials, and construction. Our monuments represent the finished product, and their architecture has the potential to inform us about every stage of the generating process, and the many factors that altered or constrained the final form of the structure. Distilling such information from the remaining fabric of the forts can be a rather uncertain exercise, only in part due to their ruinous state. The evidence of the architecture is often somewhat ambiguous, and it is important not to try to make too much of such data. Some questions asked of the defences remain unanswered, whilst other clues that they contain have doubtless been overlooked.

In addition to preserving individual aspects of the design, the surviving remains of some of the Shore Forts also contain some clues as to the order in which raw materials were supplied, and in which sections of the fort were constructed. It is to this issue that the final part of the chapter is devoted. The structure of this chapter has presented some difficulties. It progresses from methods and materials through to design, which means that certain subjects are addressed at several points. Bonding courses, for example, are discussed in the context of the construction process, aesthetics, and in their own right as an aspect of the design. The reader is therefore directed to section 4.4.4 for the main discussion of the bonding courses, to 4.4.5 for opus signinum mortar and to 4.5 for construction sections.

Within these restrictions, this chapter is essentially concerned with the details of the construction process, in a sense with the relationship between the pure design and the practicalities of realising each installation. A complete description of the building process is not possible, nor is an attempt made to do this. Rather, by examination of a variety of subjects, this chapter tries to build up a qualitative image of the building project. A particular attempt is made to establish the balance that was struck between quality and expediency, between care and haste.

4.1.1

The nature of the evidence

As is evident from the brief discussion of the monuments in Chapter Two, the preservation is rather variable between the individual sites. As a result, we know far more about the defences at Pevensey and Richborough, for example, than we do about those of Brancaster or Bradwell. This presents particular problems when discussing the presence or absence of certain features of the design, for instance opus signinum mortar, internal offsets and bonding courses. In the discussion that follows an attempt has been made to distinguish between the ‘positive’ absence of such features, and where the data are simply not sufficient to indicate presence or absence.

Only by studying the fine details can we produce such an image. These provide a necessary background to Chapter Five, which offers a broad overview of the construction process, addressing the large-scale questions of logistics, and of a possible time-scale for construction in a political and economic context. The present chapter begins by describing the construction process, and this section serves both as an introduction to many of the topics discussed below, and as a study of the many differences in detail, both within and between, the forts. A series of issues is then discussed, the first of which concerns the raw materials, in particular the building stones that were identified in the previous chapter. The implications of using materials that were mostly derived from the locality of the construction site are investigated, examining the physical and aesthetic qualities of each and looking at the practicalities of building with such materials.

4.2

The construction process

The following sections are concerned with the details of the last of the generating processes, that of construction. A summary of the data is made in Appendix I. For this reason, and so as not to continually interrupt the narrative, the details described below are not exhaustively referenced.

The description of the construction process shows that, within a basic ‘blueprint’, it is possible to see the adaptation of that design, perhaps according to

58

4.2.1

In virtually all cases there is evidence from excavations to show that the Shore Fort perimeters rested on a trench foundation, which was normally roughly vertical-sided and flat-bottomed. The only instance where this may not have been the case is the East Gate bastion at Lympne, which was built on a substantial platform of large stone slabs (Cunliffe 1980a; Roach-Smith 1852, 10), but these slabs may themselves have rested on prepared footings. Vitruvius makes the comment that the footings should be proportionate to the size of the work, and wider than the superstructure above (III.iv.i). Where there are data, the depth of the trenches can be seen to be somewhat variable, but normally falls between 0.7m and 1.5m. This is often rather less than the depth advised by Palladius, who writes that foundations in solid clay should be 1/6th to 1/5th of the height of the superstructure (I.viii.2); at Pevensey, for example, the footings are approximately 1/9th of the height of the standing wall.

Site clearance and levelling

A number of the Shore Forts were built on previously occupied locations, and before construction could begin the site had first to be cleared. This must have been a particular concern at Dover and Richborough, where the sites had been extensively used in the past. Where prior occupation had been less intense, or the buildings less substantial, the task would have been of less significance. As intimated in the previous chapter, and discussed further below (5.2.3), the demolition of earlier stone structures provided a very useful source of raw materials for the new fort. At Richborough a number of structures on the line of the perimeter wall were levelled, for example the courtyard building on Site III in the north-east corner of the surviving area of the fort (Bushe-Fox 1926). The ditches surrounding the mid 3rd century earth fortlet were filled in with clay from the fort’s rampart and other rubbish. What remained of the Great Monument was also demolished, much of the stone being recycled in the perimeter wall, most obviously in the north wall and west gate (see 4.5.1).

Burgh Castle provides a useful illustration of the manner in which the style of the foundations could vary within a single site. Johnson has shown that the western parts of the perimeter defences, built close to the top of a scarp, were considerably less substantial than those on the east side (1983a, 13-14 & Fig. 4). The wider east wall rested on a foundation trench some 0.6m deep, filled entirely with concrete. The width of the superstructure of the north and south walls was progressively narrowed towards the west, and here the foundations were far less substantial, comprising only a shallow trench ‘a few inches’ deep, filled with rammed chalk resting on a layer of clay and flints (Johnson 1983a, 13). Closer to the scarp, at the north-west angle of the fort, the foundation trench deepened sharply to allow for the weight of the angle bastion, and for the potential instability of the sloping ground to the west. The west wall itself has also been shown to have had footings of rammed chalk and clay (Johnson 1983a, 14).

Where the Shore Fort at Dover has been excavated, it has also been shown to overlie the remains of numerous earlier structures. The line of the perimeter cuts across the defences of the classis Britannica fort, which were demolished along with other structures, including prestigious buildings such as the Painted House (Philp 1989). The general practice appears to have been to level earlier structures down to the contemporary ground level, so as not to interfere with the superstructure of the new perimeter. There would have been little point in a more thorough demolition. Sections of the foundations at both Richborough and Dover show stubs of wall in situ, the remaining trench being packed with various combinations of clay, chalk, flints and mortar (see for example BusheFox 1928, Pl. XLV; Philp 1981 Fig. 24 sections 6, 10 & 11). Traces of floor, walling and heating ducts project under the bastion that stands on the Painted House at Dover. 4.2.2

A similar situation may exist at Lympne, although only the northern defences have been extensively excavated. The section of the perimeter between Bastion 6 and Wall 4-5 employs foundations some 1.5m deep, with timber piles driven into the base of the trench (for locations and plan see Hutchinson et al.1985, 227-230). Such piles are absent beneath Bastion 3, where the trench is also less deep (0.8m). It is not known whether these lighter foundations continued under the west wall. Vitruvius (III.iv.1) recommends the use of timber piles when building on unstable ground and it may well have been decided that the north wall, running across the hillside, required additional underpinning. The west wall, perpendicular to the slope, would have been more stable and therefore needed no extra measures. Piles are also present in other sizeable Romano-British structures, including the Shore Forts at Pevensey (Fulford & Tyers

Foundations

The foundations were a vital first stage in the actual construction process, a fact of which the Romans were well aware (DeLaine 1997, 145-149). Both Vitruvius (I.v.i; III.iv) and Palladius (I.viii.2) demonstrate an appreciation of the importance of well-constructed footings, whilst Tacitus gives a vivid illustration of the consequences of poorly built foundations in his description of the collapse of the amphitheatre of Atilius at Fidernae (Annals iv.60).

59

1995) and Richborough (Bushe-Fox 1928, 22-25),5 the London town wall (Salway 1993, 192) and the Allectan ‘palace complex’ at Peter’s Hill, London (Williams 1993).

4.2.3

Superstructure

Once the foundations had been completed, attention could turn to the superstructure. However, where concrete had been used to fill the trench, a pause in construction would have been necessary until the foundations had become solid. This would have been less necessary had ‘dry’ materials been used in the fill of the trench, although even these have a tendency to ‘settle’. This difference has some bearing on the question of how rapidly the Shore Forts could have been built, although it is difficult to precisely quantify the likely duration of this pause in construction. It is unlikely to have been particularly long, and in modern buildings where Portland Cement is employed, it is normally a matter of weeks (J. DeLaine, pers. comm.). When dry foundations were being employed, this break in building work could have been dispensed with altogether; this was certainly true in the case of the Fortezza da Basso, a 16th century Florentine fortress that is discussed in more detail below (5.4.5).

The majority of the foundations were built using ‘dry’ materials, rather than with concrete. The normal materials used to fill the trench were chalk, flint or other locally available stones, and occasionally clay. Thin spreads of concrete were employed at Brancaster and Pevensey, respectively as the lowest and highest components of the foundations. Only under Burgh Castle’s east wall was concrete used to fill the entire foundation trench. The complexity of the foundations differed considerably. Those at Reculver consisted of a trench that was apparently entirely filled with small flint cobbles (Dowker 1878, 8). By contrast, those at Pevensey were far more elaborate (Bushe-Fox 1932a; Salzman 1907). Above the piles driven into the trench base were laid successive layers of flint and rammed chalk and resting on the highest chalk layer, just below the Roman ground level, was a frame of timber baulks. Chalk was packed in the spaces between the timbers and a layer of hard mortar spread over the top, sealing the foundations and providing a level platform for the superstructure. The complexity of the foundations – the timber frame in particular – indicates great attention to detail and a considerable investment of time and manpower. Similar foundation frames were also employed at Portchester (Cunliffe 1975, 17) and Richborough (Bushe-Fox 1928, 23-24), although in neither case have the timbers themselves survived, only the cavities left behind after they have rotted.

At the sites of Richborough, Pevensey, Portchester and possibly at Lympne, the superstructure can be seen to have been built in a series of distinct sections, a topic addressed in 4.5. These ‘construction sections’, normally assumed to be the work of separate gangs, can be recognised by changes in materials or building styles, or in differing heights of the bonding courses or putlog holes. Of the better preserved Shore Forts, only Burgh Castle appears to lack these horizontal breaks in the defences, and here the superstructure appears to be of one build. Due to their poor state of preservation it has proved impossible to establish whether the other monuments were constructed in one or several sections.

Timber also served various other functions during the construction of the foundations and substructures.6 At Bradwell, both faces of the masonry below the Roman ground surface exhibit a recess of some 0.20m, which may be the impression left by timber used to shutter the foundations (Pugh 1963, 53-54 & Fig. 14). However, the section drawing is rather difficult to interpret, and this interpretation is somewhat uncertain. At Burgh Castle, timber baulks were laid at right angles to the direction of the wall, but appear to serve no structural purpose, as they are unevenly spaced and placed well below ground. They have been interpreted by Johnson (1983a, 13) as levels, to indicate the height to which the chalk in the foundation trench was to be laid. An alternative explanation is that they were employed as braces for shuttering, which might have been used to retain the sides of the foundation trench.

At Portchester and Lympne masonry construction began within the excavated trench, significantly below the Roman ground surface. At the other forts, construction began roughly at ground level. The lowest component was normally a plinth, wider than the masonry above, stepped out on one or both faces. Only at Brancaster was the plinth course absent, at least so it would seem from the excavated sections of the west wall (St. Joseph 1936). Above the plinth, the outer face of the wall was vertical, but in many cases the inner face was tapered, or progressively thinned by a series of offsets. This had the effect of creating a more stable structure, one that did not require a rampart against the interior face, unlike the thin, rectangular-profile walls of earlier style Roman defences. Cunliffe (1975, Figs. 7 & 8) depicts the walls at Pevensey and Portchester without offsets, continuing to their full height at a width of 3.8m and 3.1m respectively. In the case of Pevensey he is mistaken, as examination of the excavated section of the inner face of the north wall reveals a series of offsets, at 4m, 6.2m and 7.6m above the plinth. The robbing of the inner face at Portchester makes it difficult to determine whether offsets were also present at this site, but had they been absent the wall

5

These were found in the foundation trench of the abandoned east wall. 6 For an illustration of timber in Roman foundation construction see DeLaine (1997, Fig. 56).

60

level, in order to take account of the terrain. The fort was built on roughly level ground, but the western wall was built close to a scarp, down which it has since fallen. The surviving superstructure near to the scarp is significantly thinner than the eastern parts of the wall, 2.14m wide as opposed to the 3.2m of the more substantial east wall.

would have been less stable, and its construction would have required considerably more stone. The same uncertainty surrounds Brancaster; St. Joseph (1936, 447) thought the wall to be carried up without offsets, but the surviving wall stands only around 1m high, and it is difficult to make a judgement on this issue. In the case of Burgh Castle, the width of the wall varied at ground

Figure 38. The west wall, Richborough Here can be seen many of the characteristic features of late Roman defensive architecture. At the base of the wall are two courses of large stone blocks forming the plinth. Immediately above, stone robbing has left the core rubble exposed; a layer of chalk rubble set amidst the flint and septaria illustrates how the core material was laid down in successive courses. At higher levels the small facing stones remain in situ, bonded to the core of the wall by periodic brick bonding courses. Two horizontal lines of square putlog holes, left open when the scaffold was removed, can be seen between the third and fifth bonding courses. At this location the Roman wall presently stands 7m high, but it would originally have been somewhat taller

61

signinum mortar and reasons for its use are discussed below (4.4.5).

As the wall was built up, courses of brick, tile, or stone slabs were employed at vertical intervals, extending more deeply into the core than the standard facing blocks. In the literature these are generally described as ‘bonding courses’ and for the sake of consistency I have continued to use this term. However, the purpose of such courses can be interpreted in a number of ways, and it is argued below (4.4.4) that in the Shore Forts their primary role was as a levelling device, and that they had only a limited bonding function.

Once the wall reached heights of around 1.5m and above, higher than could be comfortably reached by a standing man, scaffold would have become necessary. The different types of scaffold employed in the Roman world have been discussed by Adam (1994, 81-87) and DeLaine (1997, 145-149), and early 20th century texts show that practices remained little changed in England until very recent times (Thatcher 1904). At the lowest level it is likely that free-standing trestles were used, but as the wall rose higher some form of engaged scaffold would have become necessary. Evidence for such scaffold at Pevensey and Richborough is present in the form of putlog holes, at heights of 3.4m and above: below this level, robbing tends to obscure any trace of the putlog holes.7 There are no traces of Roman putlog holes at Portchester (although some of later date are present), but this is unsurprising since the Roman facing rarely survives, and never higher than about 1m above the Roman ground surface.

Examination of the better preserved superstructures enables a detailed understanding of the actual construction methods. A course of stone was laid on the outer and inner face of the wall, creating a trough that was then filled with a mixture of rubble and mortar and rammed to remove any interstices. The next course of facing stones was then laid and the process repeated. Evidence for this method can be seen at a number of locations. At Pevensey the pink opus signinum mortar used in the setting of the facing squeezed out behind the blocks to lie over the core fill: where the facing has been robbed these are now preserved as thin horizontal stripes in the core. Elsewhere, particularly at Portchester and Richborough, horizontal courses of chalk rubble are prominent in places where the core is exposed.

The absence of putlog holes at Burgh Castle is somewhat more curious, since the facing survives in very good condition to heights of 4m and above. It has to be assumed that free-standing scaffold was employed, in what represents a very interesting departure from the practice at the other well-preserved Shore Forts. Engaged scaffold only requires a single set of vertical standards, as the wall itself helps to support and stabilise the arrangement (see Adam 1994, Fig. 180). It is quite possible to use free-standing scaffold to a considerable height, but to do so effectively doubles the number of tall timbers needed, since two rows of standards are required. Additional props may also be needed to provide lateral stability (Adam 1994, Fig. 180). There is no reason why engaged scaffold could not have been used at Burgh Castle, and given that the whole perimeter is argued to have been built at the same time (4.5.5) the amount of additional timber required for free-standing scaffold was very large.

Only at Lympne was this method was not applied. Here, many interstices are evident within the core, and the rubble seems to be less carefully laid. Instead, the method was to build several courses of the retaining face, before tipping rubble and mortar into the cavity. Although not a crucial failing by any means, the resulting product is not of as high a quality as that of the other sites. This departure from what otherwise appears to have been the standard practice is perhaps informative about the conditions under which the fort at Lympne was built, and the apparent lack of attention to detail will be returned to in the conclusions of this chapter. The mortar used in the majority of situations was nonhydraulic, using chalk as the source of the lime and sand and pebbles as the aggregate. The use of opus signinum mortar is a peculiar feature of Shore Fort architecture, and appears to have been employed only in the facing. This is essentially a standard non-hydraulic lime mortar, into which crushed ceramic had been added, giving it a number of important physical properties, as well as a distinctive pink colour. It is known to be present at a number of sites (Burgh Castle, Bradwell, Richborough, Pevensey and Lympne) but there is no evidence for its use at the other Shore Forts. This would appear to be a genuine absence, rather than a consequence of preservation. St Joseph (1936) made no mention of any such mortar at Brancaster, and field-walking on the site of the fort has only turned up fragments of a white, chalky mortar. Enough of the superstructure survives at Caister, Reculver and Dover to be reasonably certain that it was also absent at these sites. The properties of opus

The absolute height of the Shore Fort defences is not known in certain terms, since at no location do they stand to their original level. Vitruvius (I.v.1) recommended that the wall walk should ‘be so made that armed men meeting one another … can pass without hindrance’, and thus thickness of the highest surviving section of wall at Burgh Castle (1.5m) suggests that here the defences stand close to their original height. Adding a parapet of 1.6m, comparable to that excavated at Wörth, Germany (Johnson 1984, Fig. 43) suggests that the total height of the defences was around 6m. Roman masonry survives above 8m at Richborough and Pevensey and it is evident 7

At Pevensey the regular row of round holes, bored into the bonding course 1m above the plinth on Elevation 15, date to the 19th century, when animal pens were fastened to the wall.

62

4.3.1

that at these two sites, and at Portchester, the full height of the wall was probably in excess of 9m. Although the breastwork has not survived, Roman defences are known to have been crenellated; a reconstruction of the west gate and parapet of Pevensey is offered by Bidwell (1997, Fig. 33).

How may we judge the choice of materials, in particular the stone, which was chosen for the building of the Shore Forts? A useful starting point for the discussion is offered by the Building Research Establishment Digest 269 (1983) which gives guidelines for the selection of natural building stone in modern buildings. The recommendations can be summarised under three basic headings:

It was nearly universal practice for stone forts of the early Empire to have a substantial earth rampart behind the masonry wall, a practice that continued in Britain elsewhere than the Saxon Shore well into the 4th century. In the case of many forts and towns in Britain, a masonry wall was added on to the front of an existing earth rampart. Silchester (Fulford 1984), Cirencester and Caister-by-Norwich (Wacher 1995, 74-76) are just a few examples of this practice.

Economic considerations – availability (stockpiles); proximity to the building site; amount of wastage during quarrying; cost. Suitability – ease of working the stone; durability; ‘applicability’ of the stone to a given function.

Ramparts are present at a number of Shore Forts, and under the model for a typological evolution of our sites, these are often assumed to be ‘early’ (Brancaster, Caister, Reculver) or ‘transitional’ (Bradwell, Dover). The reality, however, is perhaps rather more complex. Pevensey, probably the last of the forts to be built, has a rampart, albeit one that reaches only a small way up the wall (see Salzman 1907, 17). That it was put in place prior to the completion of the inner face is attested by Upper Greensand chippings from the mason’s dressing of the facing blocks, which lie on the top of the clay rampart, but below later occupation debris (Salzman 1907, 22). 4.3

Criteria

Aesthetics – colour; texture. Other texts offer specific tests by which the physical properties of a building stone can be measured (e.g. Knight & Knight 1955, Ch. 5). These are designed to investigate characteristics such as load-bearing ability and durability, looking specifically at properties such as porosity, the response to freezing, thawing and crystallisation of salts within the pore spaces, and at the reaction to acid in rainwater. Vitruvius’ De Architectura, written during the 1st century BC, shows a clear awareness of the need to assess many of the above properties (II.vii). Amongst many other issues, it offers advice on the selection of stone for building, which, as well as discussing aesthetics, lists criteria that we would recognise today as hardness, porosity, weathering characteristics and resistance to exposure. A short extract concerning limestones provides an illustration of the detailed understanding of stone selection shown in the text:

The choice of materials

The narrative above presents a picture of slight variations around a basic theme. However, one important area of difference not yet touched upon concerns the types of building materials that were employed in each individual structure. Chapter Three demonstrated that the majority of building stones were quarried from sources close to the construction site, and as a result the fort defences are themselves a reflection of the local geology. An obvious conclusion to be drawn is that the materials were chosen because of the efficiencies that they offered, both in terms of the organisation of manpower, and of the transport burden, an issue that is discussed more fully in the following chapter. However, there is not necessarily a conflict between efficiency and other possible considerations. The following sections investigate why these particular lithologies might have been chosen, whether merely for their proximity to the building site, for their quality, appearance, or out of a combination of these factors. The balance between these various considerations has the potential to be extremely informative about the rationale behind the project as a whole, in particular about the balance struck between quality and expediency.

… all these quarries which are of soft stone have this advantage: when stones are taken from these quarries they are easily handled in working, and if they are in covered places they sustain their burden, but if they are in open and exposed places, they combine with ice and hoar frost, are turned to powder and dissolved: along the sea coast, also, being weathered by the brine, they crumble and do not endure the heat. (Vitruvius II.vii.1) Later Roman building manuals, namely those of Faventius and Palladius, written c. AD 300 and c. AD 400 respectively, also demonstrate a clear awareness of the need for the careful choice of building materials (Plommer 1973). In contrast to De Architectura, however, these manuals are concerned solely with domestic architecture, and their emphasis is on brick and concrete. That stone is not addressed directly is a

63

In traditional masonry construction, even the weakest stone may be expected to withstand the loads imposed upon it in normal use … (BRE 1983, 2)

reflection of the importance of opus testaceum as the medium of construction in private buildings of the late Empire. Nevertheless, Vitruvius was still clearly considered valid by these later authors, who draw heavily on his work and differ for the most part only in the minor details.8

Nevertheless, it is evident that for some particular aspects of the structure, certain types of building materials were preferable, if not essential.

Vitruvius offers advice as to the sorts of stone that were suitable in a particular role (between, for example, hard, soft and porous rocks) but in the main he is concerned with discussing the properties of particular Italian lithologies. It is clear from his writings that knowledge of a stone’s quality was based on experience of its use, in large part on trial and error. Thus, the relevance of De Architectura to the British builder would have been limited, except in terms of the basic guidelines. Knowledge of the geology of south and east England would probably have been of more relevance to those choosing materials for the Shore Forts, expertise that could only gained by practical experience of building in the region. 4.3.2

Bonding courses Where bonding courses were employed, brick (or, more rarely, tile) appears to have been the preferred medium of construction. The reason probably lay in the small size of the individual bricks and tiles (each weighing around 6 kg), which allowed them to be carried up the wall on the workmen’s backs, several at a time (Adam 1994, 43). At Pevensey and Portchester, thick slabs of stone were substituted for CBM in the bonding courses, probably due to shortages of ceramic material (see 3.5.2). The Upper Greensand blocks used at Pevensey commonly measure 0.8m x 0.11m x 0.5m, and would have weighed around 100 kg. Although able to fulfil the same structural function as CBM, such blocks must have been difficult to quarry and transport, whilst hoisting them up the wall would also have been an awkward proposition.

The requirements of the perimeter defences

The Shore Fort defences are uncomplicated structures and, whilst it is impossible to read the intentions of those who built them, it seems likely that they were also not intended to be prestige buildings in the same sense as major public buildings. As such, the requirements for most raw materials were not particularly demanding. Vitruvius (II.viii.8) tells us that opus reticulatum masonry in domestic buildings had a lifetime of approximately 80 years, but it is not known how long defences such as those of the Shore Forts were expected to endure. Whatever the Romans expected, it is evident from inspection of the remaining fabric that all of the materials employed were at least serviceable, since extensive stretches of wall survive today that were built from even the poorest stone.

Monumental masonry Monumental masonry is rarely found in the Shore Forts, only in the gatehouses at Reculver, Richborough and Pevensey. The large blocks of Lincolnshire Limestone in St. Peter’s chapel, Bradwell, are also probably robbed from the west gate of the Shore Fort. Monumental masonry requires a material that has no pronounced jointing, allowing large blocks to be cut from the quarry face in a single piece. Freestones are also desirable, since they are easier to shape.9 Such lithologies could be obtained from sources reasonably near to the southern Shore Forts (such as Sandgate Sandstone and Upper Greensand), even though the coastal sources exploited for the rest of the facing and core of the defences were not always suitable for this purpose. East Anglia, by contrast, is devoid of quality freestones, and this may account for the apparent absence of monumental masonry gateways at Caister and Burgh Castle, and for the importation of Lincolnshire Limestone to Bradwell. The absence of freestone at these sites, and at Lympne and Portchester, did not mean that gatehouses or arches could not be built, merely that different techniques were necessary, and that the finished product was perhaps less impressive. The Balkerne Gate, Colchester, is a good example of a major gateway that was built using brick rather than stone.

The core and foundations, which accounted for over 90% of all raw materials in the wall (see 5.2.3), could be built with practically anything, as is shown by sites such as Dover, where Wheeler observed rubble, re-used sculpture, and even iron slag (Wheeler 1929). The requirements for the facing were also undemanding, since no individual component fulfilled a load-bearing function. Moreover, as is commented in the Building Research Establishment guidelines, strength measurements of stone are rarely required, even in the present day:

9

Freestones have been defined as ‘a stone (either sandstone or limestone) that can be cut easily in any direction, being comparatively free from obvious planes of lamination.’ (North 1930, 206).

8

For a summary of the additions and changes made by Faventius to the substance of Vitruvius, see Plommer (1973, 111).

64

4.3.3

Bidwell also records other examples later in history where walls have been whitewashed to produce a brilliant effect (1996, 28).

Aesthetics

A concern for aesthetics is evident in the defences of the Shore Forts and in many other late Roman fortifications in the Western Empire. The decoration of RomanoBritish buildings, including military sites, has been recently discussed by Bidwell (1996). Here it is demonstrated that a number of techniques were used, including whitewashing, rendering, decorative pointing and the use of different-coloured stone. The decoration of the Shore Forts themselves appears to takes a number of forms, principally the use of different coloured building materials, and the pointing of the joints between the facing stones with opus signinum mortar.

Although the bonding courses were essentially functional (see 4.4.4) these too provided a means of decoration, creating bands of colour within the wall. The use of brick in this capacity is particularly evident at Pevensey, where it was used to adorn the prominent features of the circuit, namely the gates and bastions. Bastion 5 at Portchester, which flanks the north postern gate, provides another example of this practice (see Appendix I for location plan). This particular bastion has a profusion of thin courses of stone, far more than are present on other bastions around the circuit, once again for decorative effect at an important part of the perimeter (Figure 39).

Decoration employing stone and other materials of contrasting colours is widely evident in surviving Roman structures. Some schemes were quite elaborate and involved complex designs, notable examples being the walls of Le Mans and the Römerturm, a tower on the Augustan city wall at Cologne (Bidwell 1996, 24-25). Others were simpler, including that at the second classis Britannica fort at Dover, which used alternating chalk and tufa blocks to produce a basic geometric pattern. Of the Shore Forts, only Richborough shows evidence of an attempt to create a similar effect. Here, on the north wall, Caen Stone and a brown ferruginous sandstone were used alternately to create a striking band mid way up the wall. The implications of the decoration of this section are discussed below (4.5.1).

It is concluded below that in nearly all cases opus signinum mortar was employed solely as a means of decoration (5.4.3). The north wall of Pevensey shows the effect of using such mortar for pointing, pinkish red lines very clearly emphasising the joints between the facing blocks. Here, an interesting parallel can be drawn between this practice and that of ‘false jointing’, which is evident on some watchtowers of the German limes. The latter technique involved the painting of vertical and horizontal lines on rendered or whitewashed surfaces, creating the impression of ashlar construction. One example of this type of decoration is known in Roman Britain at present, on a structure at New Weir, Swainshill, near Kenchester (Bidwell 1996, 19). A somewhat different method, but one that achieves the same effect, has been observed in the amphitheatre at Caerleon, where paint was applied to the flush pointing, further emphasising the joints in the stonework (Wheeler & Wheeler 1928, 121-122, Pl. XXIV, 1-2). In all cases the paint was red, and it may be that using a pink-coloured mortar for the facing of certain of the Shore Forts was an attempt to produce a similar effect.

At the other sites, the argument for stone colour as a means of decoration is less clear, and too little survives at many of the forts to reach any conclusions. However, two sites – Pevensey and Brancaster – may also have been decorated in this fashion. The facing of the fort at Pevensey is primarily of pale green Upper Greensand, but in the mid-registers there are significant quantities of orange to brown Weald Sandstone. This may relate to the order in which stone from different sources was supplied to the project (see 4.5.2), but may equally have been a deliberate form of decoration.

It has on occasion been suggested that the Shore Forts might have been rendered with lime plaster, but the only instance where this practice is known is at Burgh Castle, where plaster was found on the inner face of the north postern (Johnson 1983a, 11). There is no indication that lime plaster was used elsewhere in Burgh Castle’s defences, and the evidence of the surviving remains of the other forts seems to argue against the possibility that rendering was a widespread practice. The decoration of the north wall of Richborough, the presence of opus signinum mortar, and the use of brick or stone courses as a means of emphasising particular parts of the circuit, all seem to suggest that the Shore Forts were neither rendered nor whitewashed.

The use of the white Leziate Beds sandstone at Brancaster may also owe much to its visual properties. Immediately local materials such as flint and Red Chalk were apparently rejected, and it may be that the Leziate Beds stone was chosen simply in order to create a white fort. Bidwell notes that: Aelius Aritides, writing in the second century AD, refers to walls which ‘stand gleaming with stucco’ and to the defensive walls at the limits of the empire ‘which gleam more brilliantly than bronze.

It is clear from this overview that the decoration of the Shore Forts was at a fairly basic level, but this is not to say that the appearance of the forts was entirely unimportant to their builders. The question of the facing

(Bidwell 1996, 28)

65

shortages of CBM for the bonding courses (4.5.2).

stone for Brancaster remains open, and if used purely for its colour it would certainly demonstrate that aesthetics was in some cases an extremely important factor in the choice of materials.

Despite this, it can shown that the elaboration of the Shore Fort defences was generally achieved within the existing range of materials that were being made available for the rest of the fort structure. Many of the lithologies used to decorate the north wall at Richborough, for example, appear to have been robbed from the defunct monumental arch (4.3.4) and thus did not need to be specially quarried.

Similarly, there is the possibility that the Wealden Sandstone employed at Pevensey was deliberately quarried for use as a decorative band in the mid-registers of the exterior face. There is also the interesting point that considerable quantities of brick were pounded up to make the opus signinum mortar at Pevensey, despite apparent

Figure 39.

Portchester: Bembridge Limestone bonding courses on Bastion 5

66

4.3.4

has questioned this view, it remains the case that in square-cornered structures, flint is invariably used within a framework of freestone or brick. It is therefore no surprise to find that the bastions of flint- or septaria-built Shore Forts are semi-circular, rather then rectangular.

Practicalities

It has been established above that it would have been possible to build the Shore Fort defences from raw materials, which, although fit for the purpose, were by no means fine architectural stone. However, within the range of materials that was actually used, it is possible to see significant variations in quality, and some were probably more desirable to the mason than others. Certain building stones offered particular challenges to the mason, and as will be discussed below, it is occasionally possible to see cases where these may have influenced the overall design of the structure.

Despite the practical difficulties, it is clear that the Romans were adept at building in flint. Its use was almost inevitable in East Anglia due to the local geology, but it is also notable that it was also chosen for the facing of Portchester, where sandstones of good quality were available from local sources, and were indeed used to build parts of the gates. Various solutions were found to the problems of working with flint. Burgh Castle was built with flints that had been knapped into a roughly square shape, which could then be built as coursed rubble. Portchester demonstrates a different approach: here elongated nodules were laid with their long axis at right angles to the direction of the core, in a ‘herring bone’ fashion that is reminiscent of the technique of opus spicatum (Adam 1994, 144-145; Cunliffe 1975, 19). It is assumed in the provenance study that all flint was from the same source (3.3.11), and it is likely that the elongated flints were deliberately selected from the mass of nodules supplied to the site. The differences in technique between Burgh Castle and Portchester may well be a consequence of the shape of the flints themselves, those at Portchester being too thin to make squaring them a practical option.

Many lithologies were employed in the fort defences, and only a few of these are outlined in the following paragraphs. These include the most common stone types used in the facing (the quality of material used for the core rubble being irrelevant), namely flint, septarian cementstones and Kentish Ragstone. Two other lithologies are also detailed, since their characteristics are of especial interest (Lincolnshire Limestone and Chalk). This selection can be used to illustrate the varying quality of the stone supplied for the building of the fort defences. Flint Flint is by far the most significant lithology present in the Shore Forts in terms of quantity. Only at Lympne does it appear to be absent, and at all other sites apart from Bradwell (and perhaps Walton Castle) it probably formed the overwhelming proportion of building material (see 5.2.3). It is extremely hard, and is impervious to the weather, and in this respect it is a very useful building material. However, the use of flint in facing work demands a high level of skill on the part of the mason, due to the irregular shape of the nodules, particularly when building the wall up in courses as opposed to uncoursed rubble. This problem also applies to septarian nodules, although the soft nature of these stones does mean that they are more easily shaped into rough block form. Building in such materials is also more timeconsuming, since the facing mortar must cure before the wall can continue upwards. If this is not done the flints will slip out of the face under the vertical pressure from the courses above. As mortar cures extremely slowly in cold conditions (Moore 1995, 185) it is impossible to build with flint during the winter months (Clifton-Taylor 1972, 196). Slightly more mortar is required for the setting of flint than is needed for regular ashlar or brick.

Lincolnshire Limestone The Lincolnshire Limestone employed at Bradwell (and possibly at Walton Castle) is probably the best building stone to be found in the Shore Forts. The oolitic freestones of the Lincolnshire, Rutlandshire and Northamptonshire regions have been much sought after, and have been exported widely for use in many fine buildings in the recent historic period. Although probably only present at Bradwell in small quantities, the size of the blocks in St. Peter’s chapel indicate that it was employed in monumental masonry. This is the only instance where we can be certain that stone was transported over long distances, presumably to meet the requirements for an elaborate gate complex. Oolitic limestones tend to be soft when buried and are therefore much easier to cut from the quarry face than lithologies such as Kentish Ragstone. However this softness means that it should not be used immediately, and it is standard practice in the modern day to leave freshly quarried oolitic stone to season before use (Elsden & Howe 1923, 65-69). This corresponds in some ways to the recommendations of Vitruvius, who writes that:

In addition to these difficulties, it is also awkward (although not impossible) to build square corners with unknapped flint. This is immediately evident from examination of more recent structures in East Anglia, notably the churches. Norfolk and Suffolk contain the vast majority of ‘round-towered’ churches in England, and the form of the towers is attributed to the use of flint and the absence of freestone. Although Heywood (1988)

When we have to build, let the stone be got out two years before, not in winter but in summer, and

67

In this manner the quality of stone could be assessed in the absence of scientific testing. Lincolnshire Limestone is probably the only lithology employed in the Shore Forts for which seasoning was a necessity, and therefore is the only one that could not have been procured immediately had stockpiles been unavailable. Once hardened by exposure to the elements, however, Lincolnshire Limestone proves a relatively durable building stone, quite suitable for external use but still relatively easy to shape or cut.

let it lie and stay in exposed places. Those stones which in the two years suffer damage by weathering, are to be thrown into the foundations. Those which are not faulty are tested by Nature and can endure when used in building above ground Vitruvius (II.vii.5)

Figure 40.

Burgh Castle, south wall: brick bonding courses and split flint ashlar

68

Kentish Ragstone

Septarian cementstones

Despite its widespread use in the south-east of England from the Roman period to the present century, Kentish Ragstone is not a particularly easy material with which to work.10 Most facies are too rough, brittle and hard to be readily dressed and the limited dimensions of the blocks that can be quarried render them unsuitable for monumental stonework (Elsden & Howe 1923, 65-69). Only a few of the sandier facies (termed ‘hassock’) can be used in this capacity, as in the south gate at Reculver. In general, the harder facies of Kentish Ragstone, including those that incorporate chert, weather reasonably well, but the hassock has not proved particularly durable in exposed situations.

Described by Clifton-Taylor (1972, 207) as ‘scraping the barrel of English building stones, little better than unfired mud brick’ septarian cementstones were nevertheless an important component of some of the East Anglian forts and also at Richborough. The use of septaria as a building stone in East Anglia and Kent persisted to recent times, albeit only as a consequence of a dearth of quality local stone. Notable examples of its use in military architecture are the Norman castles of Colchester and Orford, whilst it is also evident in many of the region’s churches (Potter 1999). Potter’s study demonstrated that septaria were rarely transported far from the coastal exposures from which they were gathered; most are found in churches that are no more than 0.5 km from a navigable waterway (Potter 1999, 132). It seems likely that the septaria used in the construction of the Shore Forts also travelled no great distance.

Chalk Certain hard chalks have found favour in the past as building stones, but that which is used in the facing of the Shore Forts is Upper Chalk, which is a far softer material. Williams’ survey of Roman building materials shows that chalk was rarely used in south or east Britain (Williams 1971, 166) and Philp comments that the two CLBR forts at Dover ‘represent the most extensive use of chalk on any Roman site in Britain’ (1981, 176). Philp estimated chalk to have accounted for around 70% of the CLBR fort structures, and the Dover Shore Fort seems to have used an equally large quantity.

Reused building materials Examination of the Shore Forts shows many instances of the reuse of building materials from earlier structures. Identification of this material is a rather uncertain exercise, although in a few cases the evidence for reuse is unequivocal. This is true for the blocks used to build the west gate at Richborough, and also the raft of stone under the east gate bastion at Lympne. In both of these instances the slots cut on the edge of each monumental block to house iron clamps do not align with each other, clearly demonstrating that the stones are not in their original position. Tegulae are also easily identifiable, but only if they are positioned in the wall in such a way that the flanges can be seen (Figure 24). However, if placed with the flange flush to the direction of the wall, such tile can simply be mistaken for thick bonding brick.

Although easy to cut and to shape, chalk has extremely poor weathering characteristics. Tests conducted by Philp showed that exposed chalk blocks shatter during the winter months due to frost action (Philp 1973, 81; 1981, 176). This poor durability may account for the distribution of chalk and tufa in the perimeter defences of the Shore Fort, where the external face was built from a combination of tufa and some chalk, but the inner face seems to have been almost entirely built of chalk. It may well be that the much more durable tufa was reserved for use on the exposed outer face. Additional measures to protect the inside of the wall might have been thought necessary, and it has been suggested that the rampart was primarily intended to shelter the soft chalk from the worst of the elements (Wilkinson 1994). This is certainly plausible, as at 2.5m wide, the wall of the Dover Shore Fort certainly did not require a rampart for stability.11

Beyond this point, the identification of reused stone is largely a matter of conjecture. It is arguable that there is recycled material in the Shore Forts at Dover and Richborough, if only because identical lithologies were already present in existing structures on the construction site. In the case of Richborough, it is perhaps too great a coincidence that the same lithologies described as present in the 1st century Great Monument (Strong 1968) also turn up in the north wall of the fort. This perhaps only an argument for those willing to be convinced, however, as none of the lithologies comes from too distant a location (see 4.5.1). The use of stone from older buildings made a virtue out of necessity, effectively combining site clearance with quarrying. All material could be utilised, whether employed as ashlar, bonding courses, core rubble, or burnt for lime. The extent of this practice varies between sites, almost certainly reflecting the amount of material that was available for robbing in the vicinity of the construction site. Far from being a second-rate product,

10

The term ‘rag’ or ‘ragstone’ is used to denote any stone of hard or coarse texture that is not a freestone. 11 The villa at Brading (Isle of Wight) provides an example of the way in which the Romans normally used chalk in buildings. Here, the exterior walls were built using Bembridge Limestone, Upper Greensand and flint: only the internal walls were of chalk (Tomalin 1987, 23-25).

69

Walls became thicker and higher than had previously been the norm, probably reflecting as much as anything else, the insecurity of the times. Increases in scale were accompanied by architectural innovation. Bastions or towers were built at intervals around the new defensive circuits: projecting from the outside of the walls, these provided platforms for archers or artillery, thus giving the perimeters greater protection from attackers. Entrances also became more heavily defended, often with flanking bastions on either side of the gate and gatehouse designs also became more complex. Other more minor, but nonetheless significant, changes can also be seen, foremost of which was the introduction of bonding courses.

reused material was often a highly desirable commodity for the builder. The tufa used at Dover is a case in point, since geological deposits of this quality stone are small, and not easy to locate (see 3.3.8). The abandoned CLBR fort must have provided a useful quarry of tufa, which was a far superior building material to the local chalk. In the case of Richborough, the demolition of the Great Monument provided not only a large quantity of stone (see 5.2.3), but also a variety of materials, which lent themselves to the decoration of the exterior face of the north wall. Summary Ironically, the destruction of the Roman defences at Bradwell provides one of the best illustrations of the relative quality of the various building stones. When the Saxon builders plundered the site in order to build their chapel, they chose the higher quality lithologies, principally Lincolnshire Limestone, Kentish Ragstone and tufa, whilst rejecting the septarian cementstones that had formed the bulk of the Roman defences. It remains the case, however, that it was perfectly acceptable for the Shore Forts to be built using local materials, and this situation raises certain issues. The first is that the longdistance transport of stone to the site cannot be explained purely by the demands of the structure, and where it occurs we must look for alternative explanations. Second, although all stone used in the forts was serviceable, the quality was variable and certainly not all was ideal. 4.4 4.4.1

In Italy the emergence of the new architecture can be seen in the walls of Rome, begun under the auspices of Aurelian in 271. Its adoption in Gaul can be observed in early-dated town walls such as Dijon, whose defences were also attributed to Aurelian and which were described by St Geoffrey of Tours (Butler 1959). These defensive improvements are also apparent in Britain within the later series of Shore Forts. Gone were the comparatively simple designs seen in the earlier coastal forts, and in their place were much more substantial, modern defences. In certain cases the regular rectangular plan, which had typified (with a few exceptions) the forts of the early Empire, was superseded by a variety of polygonal arrangements (e.g. Burgh Castle, Bradwell, Dover) and in the case of Pevensey, an irregular oval. Not all were irregularly planned, however: Portchester and Richborough were designed as rectangles, although the latter fort suffered from minor surveying errors on its north and west walls.

Architecture and design The architecture of the Shore Forts

Although the methods of construction were broadly similar, the architecture of the forts is actually far from uniform. Particular differences can be observed between the installations built early in the 3rd century (Caister, Reculver, and probably Brancaster) and the group of eight constructed after c. AD 260. The early Shore Forts adhere to what might be termed a ‘traditional’ design, one that continued in the military building traditions of the 2nd century. In plan the defences were rectangular and roundcornered, with walls backed by an earth rampart which extended to the base of the parapet. Internal towers were present at some or all of the angles, whilst gates tended to be relatively simple in their design and only lightly defended. Inside the forts was a regular layout of roads and buildings, with the principia located within the central range. Such a design directly compares with that of the vast majority of 2nd century forts in the British province, and indeed elsewhere in the Empire.

The architectural innovation seen in the later Shore Forts is in marked contrast to the innate conservatism found in most contemporary building within the British province. On the northern frontier, reconstruction of existing forts from AD 270 onwards was undertaken along 2nd century lines, whilst the new forts at Piercebridge (Co. Durham) and Newton Kyme (Yorkshire) were also built according to traditional design. Only at a relatively small number of sites – notably in coastal locations – does the architecture demonstrate late-continental influences. A pertinent example is that of Cardiff, which in plan is strongly reminiscent of Portchester, whilst the late 3rd century addition of bastions to the estuarine site of Brough-onHumber also brought this latter site up to the most modern standards. Contemporary town fortifications in Britain were also largely unaffected by continental trends, continuing to employ narrow, rampart-backed walls without external bastions. An interesting example is that of Canterbury, which was fortified in approximately 270. Unlike many other urban centres where the stone defences evolved from earlier earthen ramparts, there was nothing at

During the later 3rd century defensive architecture in the western Roman world as a whole was to change. Drawing directly on influences from the eastern Empire, new defences in Gaul, Italy and elsewhere – both military and urban – were built on an altogether more massive scale.

70

characteristic designs, must be viewed as a unique development within the British province, somehow divorced from the other defence-building programmes of the period.

Canterbury to influence the form of the new perimeter wall, but here, as elsewhere, the defences were built as a narrow-gauge wall with an internal rampart. It would therefore seem that the Shore Forts (and perhaps coastal installations over a wider geographical area), with their

Figure 41. Sections through the Shore Fort defences The traditional style of narrow, rampart-backed fortifications at Reculver (top) contrasts with the more massive freestanding wall of the later fort at Pevensey (below) defensive features were designed and incorporated into each structure was also markedly different, the bastions being a case in point. Normally the bastions were semicircular or rounded, though at Burgh Castle they are better described as ‘pear-shaped’. In most instances they were solid (as for example at Pevensey, Lympne and Burgh Castle) but in rare cases they were hollow (e.g.

Each of the later group of eight Shore Forts was of somewhat differing design. Even the most basic aspect, the thickness of the wall, varied very considerably between sites, from 2.3m in parts of the circuit at Dover to 4.2m at Pevensey and Lympne. Given these differences, the final height of the defences was probably also very variable. The manner in which the new

71

defences. There is a considerable contrast, for example, between the simple west gate of Richborough and the more elaborate bastion-flanked east gate at Lympne, or the complex inset Landgate at Portchester. These three forts were broadly contemporary, and yet the detail of the design was markedly different at each (Figure 42).

Portchester). Usually there is consistency within an individual monument, but at Richborough this is not the case, the corner bastions being round and solid, with those on straight sections of wall being rectangular and hollow. Gate designs, too, were quite distinct at each site, both in their arrangement, and in the massiveness of their

Figure 42.

Gate and bastion designs the rampart may in fact have been used as a means of protecting the chalk facing of the interior from the weather. However, at Bradwell a rampart was present behind a wall 4.2m thick at the base. So too at Pevensey, although here the rampart extends only a very short way up the wall.

Crucially too, there appears to have been some confusion as to how many of the new architectural concepts were to be incorporated into the design. The bastions once again provide a case in point. Those at Burgh Castle were only joined to the wall in the upper courses, giving the appearance of having been ‘tacked on’, perhaps as an afterthought to the original design. Moreover, at this particular site the corner bastions are imperfectly designed, for they do not project far enough to have been able to provide cover for the stretches of wall on either side. At Dover, several of the known bastions were built as an integral, original part of the perimeter wall, but others were added at a later stage, as if it was realised that the spacing between the original bastions was too great. Other inconsistencies can be observed. For example, there sometimes seems to have been a failure to appreciate the fact that the new, thicker defences had rendered the rampart behind the wall unnecessary. This is certainly the case at Dover, although as suggested above,

4.4.2

A typological evolution?

As has been shown above, the raw materials themselves may be responsible for some of the minor variations, whilst others may have resulted from the differing practices of the masons working on the forts. In general terms, however, such factors would have only had a limited impact on the typology of the defences, and the gross variations in design require a different, broader explanation. Why, for example, do the scale of the defences, or the form and complexity of the gateways,

72

The arbitrary examination of architectural criteria as a means of demonstrating the evolution of building styles is also risky. The thickness of perimeter walls is a case in point: whilst there is a demonstrable trend towards more massive defences during the 3rd century, the western parts of Burgh Castle’s defences show a clear appreciation of the need for a lighter superstructure where the ground was less stable (Johnson 1983, 10-11). Similarly, and as discussed above, the internal rampart at Dover, deemed an ‘early’ feature, might in fact have been present only to protect the chalk face of the interior wall from exposure to the weather.

vary so considerably? Why is a rampart present at some sites and not others? The most widely accepted explanation was first proposed by Cunliffe (1975, 419-22; 1977b), and has been further elaborated upon by Johnson (1976, 94-113). The essential plank of their argument is the idea that the design of the Shore Forts evolved during the course of the century, beginning with simple structures that continued the traditions of 2nd century design, and ending with the most massive and complex defences. According to this model, Reculver, built along traditional lines and dated by archaeology to the first decades of the 3rd century, was seen as the first of the series. Although the archaeological evidence for the date of Brancaster is less conclusive, the ‘playing card’ plan of its defences, internal rampart and the absence of external bastions also led to the fort being assigned to the early stages of the 3rd century. Caister, still thought to be a civilian port, was not considered as part of the scheme, but would be similarly placed within this initial group of installations. The second phase of Shore Fort construction was seen to begin with Burgh Castle and Dover, built no earlier than the reign of Gallienus (253-68). These forts, which rather imperfectly incorporate external bastions, are said to represent a transitional phase in fort design. It is suggested that the remaining forts were built in close succession, in a phase of construction that began with the erection of stone defences at Richborough from 275. Completion of the series culminated in the building of Pevensey, which at the time of Johnson and Cunliffe’s writing was thought to have occurred around the reign of Constans in the mid 4th century.

The notion that ancient architecture did not evolve in linear fashion is supported by examination of other, wider, aspects of Roman building. One example that can be cited is the Pompeian house. Here, studies have shown that ‘traditional’ design aspects persisted up until the destruction of the town in AD 79, alongside the supposedly ‘later’ features that were once thought to have superseded them in the mid and late 1st century AD (Laurence & Wallace-Hadrill 1997). It is quite possible, therefore, that improved dating evidence for the forts will dispel, rather than confirm, any typological scheme. 4.4.3

The transmission of the design

An alternative solution to the question of architectural variance within the Shore Fort defences lies in the transmission of the design, and thus in the relationship between the architect and the builder. It has been commented that the architecture of the later group of Shore Forts was so different that they ‘did not emerge from a single stereotyped blueprint from central authority’ (Wilson 1980, 74). However, the very lack of standardisation might in fact argue for many, if not all, having been virtual contemporaries. Stephen Johnson, architect of the typological scheme, has commented that, ‘if not actually planned together, they show all the hallmarks of different sets of military architects and surveyors shown an original blueprint and told to go off and apply the new methods to what was built’ (Johnson 1989, 43).

Any such scheme awaits proof in the form of conclusive dating evidence for the individual sites, though a temporal separation between the three early-style forts, and the eight more complex monuments built after c. 260, seems reasonably clear. At present, however, the archaeological data are simply not sufficiently precise to confirm or deny the proposed sequence for the later group’s construction. Problems also exist because the idea of a basic, linear typological evolution, from simple to complex, may itself be fundamentally flawed. In his study of Roman fortifications, von Petrikovits concluded that,

The transmission of plans in the ancient world is a subject that has received a certain degree of attention in recent years, although studies have tended to concentrate on the buildings of Rome.

… different methods of fortification were employed side by side, simultaneously and in the same areas, so that we should guard against any tendency to date late Roman fortifications on typological grounds. This method, if cautiously employed, is useful for dating the forts of the Principate, but (with few exceptions) is worse than useless in the late Roman period.

Vitruvius (I.i.3) wrote that the architect should be both a man of letters and a skilful draftsman, and one of the many finds from Pompeii is a set of draughting tools including dividers, a folding foot-rule and callipers (Sear 1982, 70). However little evidence has survived of the products of such men, beyond the actual buildings themselves. Haselberger (1997) has summarised the main examples of architectural drawings from the classical world of which we are aware. Probably best known are

von Petrokovits (1971, 203)

73

Such massive defences, projecting bastions and complex gateways were rather novel innovations in Britain, and were likely to be unfamiliar to any unit drafted from Wales or the northern frontier for the construction of the forts. Each unit would have had to find its own solution for the incorporation of the bastions, some better than others. A learning process almost certainly took place, and Richborough, where the south west bastion is ‘tacked on’ to the wall but where all the others are built integrally, may represent just such an example. At Burgh Castle the angle bastions were poorly designed, with a restricted field of view that would have meant that they could not provide covering fire for the adjacent wall sections (Johnson 1983a). Each unit built the gates and gatehouses in its own way, whilst some persisted with the rampart, perhaps not appreciating that it had been rendered obsolete by the thick, tapering wall.

the marble fragments of a plan of Severan Rome, the ‘Forma Urbis’, whilst on a number of structures, including the temples of Didyma and Pergamon, very large scale drawings have been found inscribed onto the walls, depicting that particular structure. More humble buildings are also drawn, such as a private house of the 2nd century A.D from Oxyrhynchos, drawn to scale and complete with written dimensions of each room. A number of models are known, for example the late Republican temple model from Ostia, which are carved or cast with sufficient detail for some perhaps to be a planning or building model. Some or all, however, may be funerary artefacts. Beyond proving an ability to draw and model to scale, the relevance of such examples is probably rather limited. The Forma Urbis is not even precisely to scale, exaggerating the size of imperial residences, and whilst the in situ inscriptions may have served as a guide on the site, none was in any way portable and it is extremely unlikely that these were the original design drawings. The town house drawing is thought to be a survey of an existing structure, perhaps for legal reasons, and thus an actual design drawing has yet to be discovered.

We need not assume, however, that all workers on the fort were from the British province, an idea pursued in the following chapter (5.9.2). Any craftsmen recruited from Gaul could well have had direct experience of defences that incorporated bastions and bonding courses, as these were widely present in the Gallic town walls. It should be noted that the most advanced of the Shore Forts (that is to say those which incorporate the new architectural features most perfectly) are those on the south coast, namely Lympne, Pevensey and Portchester. Although it perhaps takes the argument too far, it is possible to entertain the idea that these forts were built using labour from Gaul, whilst Richborough and Burgh Castle were constructed by those who lacked direct experience of the new architectural styles.12

In the absence of direct evidence, insights into this question are largely provided by detailed study of individual Roman structures, and by experience of building in the modern era. The Baths of Caracalla, for example, are of such complexity that they could not have been easily realised without some design drawings. DeLaine has written that: While for a simple structure, verbal or written instructions for elevations and details would have sufficed, the same information might be transmitted more easily by a visual image with measurements added. (DeLaine1997, 66)

In this hypothesis, it is not necessary for all of the later eight Shore Forts to be contemporary, as particular units may have been assigned to the construction of two or more sites. Indeed, the drawing of Walton Castle is so similar to the remains of Burgh Castle that it suggests that the two sites may have been the work of the same hand, although obviously this is now beyond proof. It is argued in the conclusions, however, that the other forts were probably the work of separate groups.

Gros (1985; 1996) takes this view rather further, arguing that section drawings or elevations were not used at all, the necessary information being transmitted verbally to the builders.

There may well be evidence for a similar situation having arisen during the construction of Hadrian’s Wall. There exist appreciable differences in the style of the Wall itself, independent of the broad and narrow gauge sections, whilst the precise design of milecastles and interval turrets also varies. These differences are thought to be a reflection of slightly different work practices of the three legions responsible for building the wall, each assigned certain stretches (Breeze & Dobson 1987, 7079).

The Shore Fort defences are, by contrast to such great buildings of the Empire, extremely simple structures. It is quite plausible, therefore, that written instructions were issued for the construction of a series of forts, perhaps without an accompanying plan and almost certainly without elevation drawings. Leaving aside the three early sites (Brancaster, Caister and Reculver), all the other Shore Forts possess the basic characteristics of massive defences and projecting bastions. If individual architects or builders were left to work out the details of how this was to be achieved, the architecture of the Shore Forts may in fact reflect a number of contemporary solutions to the same design brief.

12

The manner in which the bastions were incorporated at Bradwell and Walton Castle is uncertain due to a lack of data.

74

assessed individually at every monument. Sear (1982, 77) suggests that the use of brick was becoming common in late Republican Rome, and opus mixtum construction was introduced into Gallo-Roman architecture by the beginning of the 2nd century (Adam 1994, 143). Early examples of brick bonding courses in Romano-British structures can be found in the Jewry Wall, Leicester (c. AD 125) and the Dover pharos (assigned a date in the 2nd century). Adam (1994, 143) argues that in Italian architecture, the brick in opus mixtum construction did not serve a bonding function, but was merely an element of the facing; instead, he suggests that the levelling function was paramount. DeLaine expands on this argument, suggesting that when they were originally introduced into Roman architecture they were intended as a levelling device, and that the other advantages they offered (such as the level base for scaffold) were recognised at a later stage (1997, 143-145).

The transmission of the design offers a plausible alternative to the theories of a typological evolution of the defences put forward by Johnson and Cunliffe However, it remains only a hypothesis, and only improved evidence from the forts, particularly in terms of the dating of the defences, will determine if this explanation is indeed viable. 4.4.4

Aspects of the design: bonding courses

A notable characteristic of some Shore Forts is the presence in the facing at vertical intervals of deeper courses of brick, stone slabs, or, more rarely, ceramic tile. As we have already seen (4.3.2), these courses did have an aesthetic value, but it is argued below that they also served a structural function. These courses are generally referred to as ‘bonding courses’, a term which has been retained in this monograph for simplicity. However, as DeLaine (1997, 143-145) amongst others have demonstrated, the role of such courses is quite varied and, depending on their precise design, they can fulfil a number of functions in Roman architecture.

In some Gallo-Roman defensive architecture these courses do serve a bonding function. The walls of Bourges and Beauvais (both of the later 3rd century) exhibit bonding courses which extend through the wall (Adam 1994, 126).13 This does not seem to be the case for most defences in the province, however. Where they have been sectioned, the courses at Jublains, Sens, Bordeaux and Nantes do not penetrate more than about 0.5m into the core (Butler 1959, 41). The same is true of the London town wall, built around AD 200 (Maloney 1983). Butler suggested that these courses did fulfil a bonding function, keying the shallow facing blocks to the core; he also argued that they were used in a levelling role and as support for the scaffold. 14

The bonding function can take two take two forms. Where the course extends through the wall it serves as a tie between the two faces, truly ‘bonding’ the wall. Where the courses only extend part way into the core, their bonding function is more restricted, keying the shallow facing stones into the mass of the core. The brick in opus mixtum construction can also serve as a means of levelling the structure. Petit appareil blocks are rarely of precisely the same height, and the use of flat brick courses at short vertical intervals enabled the builders to even out these slight differences. They also provided a level, reliable platform on which to build the next stage of the wall.

The Shore Fort bonding courses clearly do not bond the inner and outer face, since they only extend part way into the core, normally about 0.5m. Furthermore, where those on the inner face have survived, as at Burgh Castle, they can be seen to be at a different height to the bonding courses on the external face. As in much Gallo-Roman architecture, the bonding function they serve is between the face and the core, rather than between the two faces.

It has been suggested that in domestic Roman architecture, the bonding course was built at the end of a ‘lift’, perhaps representing a day’s work (Sear 1982). In large-scale construction, lifts would have taken far longer, and there is probably no hard and fast rule about the length of time to which they relate. Nevertheless, the bonding courses may indicate pauses in construction, particularly when the next level of scaffold needed to be erected. There are many examples in Roman architecture where putlog holes can be seen in the first course above the bonding course, and the brick would have provided a level surface on which to position the putlogs. As DeLaine has pointed out, the erection of scaffold would bring about a natural pause in the construction of the wall, whilst the putlogs were inserted, ledgers, walkways and braces put in place, and the standards extended where necessary (1997, 143).

There is a clear connection that can be made between flint and the use of bonding courses. Such courses were employed at nearly every Shore Fort that was built of flint, with the exception of Caister. The courses would have helped to bond the irregular flint facing to the core, and, probably of even greater importance, levelled the structure at vertical intervals. The septaria used at Bradwell would have created the same kind of difficulties as flint, and this fort, as well as that at Walton (built almost certainly of flint or septaria), employed bonding 13 This is also the case at Silchester, where the thick slabs of stone extend through the entire width of the wall. 14 Knight (1999, 26) states that the use of bonding courses “would have speeded up construction by enabling the mortar to dry more quickly”, but this assertion makes little sense.

From these varying interpretations of the function of the ‘bonding course’, it is clear that their role must be

75

4.4.5

courses. Where petit appareil were used, bonding courses were probably rather less necessary, though they were still used at Pevensey, Richborough and Lympne. The example of Dover is perhaps quite informative, since bonding courses were absent on the main wall and its contemporary bastions, built from squared chalk and tufa, but were present in the later flint bastions (see 4.5.5).

Aspects of the design: opus signinum mortar

Opus signinum mortar was used widely in Roman Britain in capacities such as flooring and cistern linings, but one unusual feature of the Shore Forts is its use in the facinng of the perimeter walls. This is the only instance of which the author is presently aware where it is used in defensive architecture, although there are likely to be other examples. At Pevensey and Burgh Castle opus signinum mortar was used for the setting of the facing blocks on the outer face, not just as decorative pointing of the joints. In neither case was it present on the inner face, where the blocks were set in ordinary mortar. At Richborough it was also present, but the amount of repair and repointing makes it rather difficult to ascertain precisely how it was used (though it was probably only employed for the outer face). It can be observed in the outer face of the wall at Bradwell, but the inner face of the wall is not presently exposed. At Lympne it is used more sparingly, in the facing courses below ground level and between the bonding courses of the outer face. Once again the form of the inner face is unknown.

There is no discernible pattern that can be detected in the vertical spacing of the bonding courses at any of the forts. The intervals do not correspond to Roman measurements, nor to meaningful fractions of the pes or cubit. They do tend to be roughly evenly spaced, however. At Burgh Castle the vertical interval is between 0.44m-0.61m, but it is markedly wider at Pevensey (1.3m-1.6m), Richborough (0.9m-1.12m) and Portchester (1.5m). There are instances at Pevensey and Portchester where the interval is extremely small, but this is where the bonding courses are primarily present as decoration 4.3.3). At the latter sites, the intervals correspond to manageable scaffold heights, and to comfortable working heights for the mason. The shorter intervals between bonding courses at Burgh Castle are less easy to explain, however. Putlog holes are entirely absent (see 4.2.3) and it is thus unlikely that there is any relationship between bonding courses and scaffold. The short intervals may merely be a reflection of the sheer amount of brick being supplied to the site: certainly there was no shortage of flat brick, which could serve few useful functions if left over after construction of the wall was complete.

Opus signinum mortar has certain physical properties that set it apart from the ‘standard’ lime mortar employed in the bulk of the Shore Fort structures (Moore 1995). Firstly, it attains its initial strength at a faster rate, which might have enabled more rapid building, since the two ashlar faces retaining the core rubble would have become solid in a shorter time (Moore 1995, 158). However, there is no advantage if opus signinum mortar is employed only the outer face, as seems the case at most of the Shore Forts where it is found.

There may well be some link between the bonding courses and pauses in construction. Where sections of wall have fallen, the bonding courses appear to be a plane of weakness, as is the case on the north wall at Pevensey and the tumbled bastion on Burgh Castle’s south wall. The implication is that in these instances brick courses were built as the last stage of a lift, before halting construction of that particular section of wall for a period. Even during a short pause the core mortar would have begun to cure, and when building recommenced it did not fully bond with the new mortar being laid above. Many centuries later when these sections of wall fell, the masonry broke at these weak points in the structure.

A second advantage is that opus signinum mortars will cure in damp conditions, as a result of the presence of clay minerals. The active ingredients within the clay are amorphous and vitreous silicates and aluminates, which combine with lime to form hydrated silicate of calcium and other aluminate/silicate complexes. These reactions mean that such mortar does not need to lose water by evaporation – adding clay to non-hydraulic lime effectively converts it to hydraulic lime. Its properties are similar to pozzolanic mortar, used widely in Roman Italy and beyond, and described in some detail by Vitruvius (II.vi.1). The hydraulic properties of pozzolanic mortars were notably exploited at sites such as Ostia and Cosa, where it was used in the construction of underwater harbour-works (Potter 1987).

The placing of brick courses at the end of a lift may offer an explanation for the short intervals between the bonding courses at Burgh Castle, which was possibly built in smaller lifts than some of the other forts. Similar short lifts can be observed at the highest surviving levels of the north walls of Pevensey and Richborough. These perhaps represent the completion of the superstructure, which was finished with a course of brick when it reached wall-walk height.

An alternative explanation for its use in the Shore Forts may lie in the decorative properties of opus signinum, which served to define the facing blocks by pinkish-red lines, creating a striking appearance. Parallels can be drawn with the technique of false jointing, and with the painting of flush pointing, which have already been described above (4.3.3).

76

Although practical considerations were clearly important, there is still the question of why this rather peculiar form of decoration was used only at these forts. After all, it would have been easier to paint the mortar joints. This technique may have represented a new fashion, but might also suggest that there was a common influence behind these particular forts. The dating evidence for the construction of the forts can be stretched to allow for the possibility of a single group of builders having been responsible for each site in turn, but the hypothesis is beyond any proof.

The last property seems to provide the most plausible explanation for its use at Burgh Castle, Pevensey, Richborough and Bradwell. Although some doubt remains as to whether opus signinum mortar was present on the inner face of Bradwell’s walls, it is likely that it was only used on the outer face of the other sites’ defences. Although it would have created a waterproof wall, the absence of any measures to protect the interior face at Richborough and Pevensey argue against this as the main reason for its use. The presence of opus signinum mortar at Lympne arises for rather more practical reasons. Here its use was more specifically targeted, below the Roman ground surface and in the joints between the brick and tile of the bonding courses. The superstructure of the wall at Lympne began below ground, within the sloping sided foundation trench. Before the wall could reach any height this trench would need to be filled, partly for stability, and also because of the need for a level surface on which to rest the scaffold standards. Thus, soon after its construction the base of the wall would have to be covered, and the use of opus signinum would ensure that the buried facing would set in such closed, damp conditions. It seems highly likely that opus signinum will be found below ground on the inner face of the defences, if they are ever excavated.

4.5

The order of supply and construction

There are too many unknown variables to determine the precise order in which each fort’s defences were built, or in which the raw materials were supplied to the site. However, the surviving fabric of some of the Shore Forts does enable a number of observations to be made. There is a certain amount of evidence for a number of the Shore Forts’ defences having been built in several parts, divisions which are normally termed ‘construction sections’. These are generally understood to be the work of individual gangs of workers, and can be detected by differences in techniques and materials. They can also be detected from slightly differing levels of the bonding courses and putlog holes (see for example Johnson 1981). Often at the junction of two construction sections, the bonding courses on the two pieces of wall can be seen to be mis-aligned.

The use of the opus signinum mortar to cement the bonding courses at Lympne also exploited the same physical properties. The bonding courses were primarily levelling devices and the use of this special mortar would have created a solid, reliable platform upon which to continue building. If the bonding courses also represent a pause in construction, then the presence of a layer that was effectively waterproof would also have been extremely desirable.

4.5.1

Richborough

The defences of Richborough provide perhaps the best illustration of this issue, and have been addressed in some detail by Johnson (1981). Although there is relatively little to add to his description of the architecture of the fort, it is possible to be more specific about the lithologies used at various points around the circuit. Johnson used the term ‘ragstone’ to describe virtually every lithology except flint, so in this respect some further discussion is justified.

Aside from Lympne, the use of opus signinum mortar is a somewhat idiosyncratic and strictly unnecessary trait, and there remains the question of why is it present at some sites and not others. There is quite a clear chronological pattern, in that it is absent from the earlier sites of Brancaster, Caister-on-Sea and Reculver. This, however, seems to be a reflection of the major reason for its presence or absence, which is the availability of ceramic material on each individual site. There is a direct connection between those forts where CBM is widely used and those that employ opus signinum mortar. There would doubtless have been a significant amount of broken brick at these sites, damaged during transit or in the construction process and it is notable that brick fragments are rarely found in the core of the forts, suggesting that most was directed towards making the mortar. Conversely, where brick was absent, in short supply, or as was the case at Dover, only used at the end of the construction process, opus signinum mortar is apparently absent.

The defences survive on the south, west and north walls, and can be divided into two distinctly different parts. The first of these is the surviving portion of the south wall and the entire west wall, a length of some 230m, which is apparently of one build (see Figure 14; Figure 44). In this portion the first two courses, forming a plinth, are of Kentish Rag from the Maidstone district. On the south wall the facing above the plinth is predominantly septarian cementstone, whilst on the west wall there is much Thanet Sandstone and rather fewer septaria. These building materials were almost certainly freshly quarried at the time of the fort’s construction. Except for a very few tegulae on the west wall, the bonding courses were also built with new brick. The only location where a large amount of recycled material is found is in the foundations

77

and ‘C’. Eastwards of ‘C’ there is a greyish lithology that may well be a facies of Kentish Ragstone, but as it was only present in the higher registers of the wall it was not possible to examine this material. These are the same lithologies that Strong (1968) described as having been used to build the 1st century monumental arch, and it seems virtually certain that those we can observe in the north wall of the Shore Fort have been recycled from the same structure. The different altitude of the putlog holes in sections B-C, C-D, and eastwards from D to the point at which the wall has fallen, indicate that each of these pieces of wall were built separately to one other. Additionally, plans of the fort show that there is a slight change in the direction of the wall at break ‘D’. The bonding courses appear to be quite carefully aligned between these sections, except at break ‘C’, where the first bonding course above the plinth is fractionally misaligned.

of the west gate, which was built from large blocks of Sandgate Sandstone robbed from the Great Monument. The north wall is rather different, for the surviving portion was built in four distinct sections of between 30m and 40m in length (Figure 43). The heavy lichen cover makes it difficult to map these stone types onto elevation drawings, and no attempt has been made to do so here: however, the general picture can be described as follows. Thanet Sandstone is abundant in all sections of the external face, but eastwards of ‘B’ several other lithologies can be identified, which are present nowhere else in the surviving circuit. Significant quantities of tufa are extant, together with a Ditrupa Limestone from the Calcaire Grossiere formation. Flints and septarian cementstones are also used in limited amounts. Caen Stone and a ferruginous sandstone are employed in the higher registers to form a decorative band. A very few blocks of Marquise Oolite could be observed between ‘B’

Figure 43.

Richborough: construction sections on the north wall watergate, presenting an impressive front to those arriving by sea. If true, it is a comment on the relative importance of land and sea communications to the fort, since the west wall is comparatively plain, even though the main road from Canterbury entered on this side.

These observations raise two main questions, firstly about the distribution of the different lithologies around the circuit, and second about the shortness of the sections of the north wall. A large part of the explanation lies in the location of a harbour or beaching point, which, although physical remains have not been found, is thought likely to be to the north of the fort. It is there that Boys claimed to have identified the remains of a ‘landing place (Rolfe 1843, 24). Although the east wall has been lost and its form is unknown, the fine appearance of the north wall strongly suggests that the north postern was in fact the

The short construction sections of the north wall would also tend to support the idea that the landing point was to the north. Given the coastal resources exploited, it is almost certain that the raw materials for the building of the fort were transported by water. This is further

78

confirmed by the find of a series of lime kilns of late 3rd century date some 100m north of the north wall (BusheFox 1932b, 36-38 & Pl. XLVI), which may well be part of a depot for the storage and preliminary working of raw materials during the building of the fort. The north postern is an extremely constricted entrance and it would have been practically impossible to carry bulky items such as the timber balks for the foundations through this gate. A larger gap would have been necessary to facilitate the easy movement of materials between the harbour and the fort.

buildings. Raw materials were unloaded at the harbour and stored, to be moved through the open north side of the perimeter as and when required. It would be logical for the north wall to be built last, not only for logistical reasons but also because the decorative facing would have been somewhat more time-consuming to build than a plain stretch of wall. Materials that in large part were already present inside the perimeter were used to close the circuit. The short sections probably resulted from the concentration of manpower on a comparatively small length of the wall, but the high quality of the masonry strongly argues against the haste and lack of care suggested by Johnson (1981).

In this scenario, the west, east and south walls were constructed first, as well perhaps as some of the internal

Figure 44. 4.5.2

Construction sections at Richborough After Johnson (1981, Fig. 2) bonding courses) there are significant quantities of orange-to-brown Wealden sandstone, which, although not arranged in any decorative pattern, nevertheless forms a striking band of colour between the green-coloured Upper Greensand. The Wealden sandstone appears to be concentrated on the south wall, the north-east quarter and the north wall, but is absent from elevations 12b and 12c and the north-west quarter of the circuit.

Pevensey

The fort at Pevensey was also constructed in a series of sections, but the loss of much of the south wall and the poor preservation of the north west part of the circuit makes it difficult to determine how many were present. The remains of the perimeter exhibit at least seven sections of varying length (Figure 45). The north east quarter of the circuit appears to be a single unit of 100m, which contrasts with the immediately adjoining units on the north wall (Elevations 12b and 12c) which are both roughly 15m long. (Elevation numbers are shown in Figure 52 in Appendix I).

Two possibilities arise to explain its presence and distribution: first that a new quarry was opened up at the mid-stage of the construction of the superstructure; second that it was quarried at an early stage and was stockpiled specifically for use as a decorative band.

The core of the wall was built almost entirely of flint rubble and mortar, and the majority of the facing was Upper Greensand, which appears to have been quarried throughout the project, since it is present at all points of the circuit and at all heights. In the mid-registers of the external face (normally between the second and fourth

The explanation for the short sections of the north wall at Pevensey is probably similar to that outlined for the north wall of Richborough. They lie roughly mid-way between the north postern and the main west gate, and their construction may well have been left until last to provide

79

above, was the preferred medium for the bonding courses. On the north wall the placing of the bricks for each bonding course seems to have begun at the bastion, working outwards onto the adjacent sections of wall. However, at certain points these brick courses are terminated mid-way along a wall section, to be replaced by large slabs of sandstone. The implication is that supply of brick was not equal to demand, and that it was necessary to find substitute materials. One such substitute was Horsham Stone, present only in limited amounts in the bonding courses of Elevations 12b and 12c. The scarcity of brick is slightly puzzling in view of the number of suitable clays for brick and tile making in the Wealden district.

easier access between the water and the interior of the fort. The height of the bonding courses and putlog holes show that these sections bear little relation to each other, nor to the stretches of wall on either side. The vertical cracks, extending through the entire width of the wall, also tend to suggest that the core of each section was imperfectly bonded to the next, probably because each unit was built at a separate time. The absence of Wealden sandstone from these sections might also suggest that they were built at a stage when supplies of this lithology were exhausted. The walls of Pevensey also show evidence of an insufficient supply of brick, which, for reasons given

Figure 45. 4.5.3

Construction sections at Pevensey Castle supply, since only the lower parts of the wall could be examined and the preservation was rather patchy. However, clusters of certain lithologies can be detected, for example of a coarse facies of Quarr Stone (PORT 8B), which is found on Wall 14 and again on Wall 17. Similarly, there are concentrations of a fine-grained sandstone (PORT 8D) on Wall 18, Wall 5 and Bastion 5. These clusters are probably a reflection of the arrival of individual boatloads of stone from the quarries on the Isle of Wight.

Portchester

The poor survival of the Roman facing at Portchester makes it rather difficult to say too much about the order of construction. Close examination of the walls suggests that they were built in sections of roughly equal length, each gang being responsible being for a section of wall, a bastion, and stub of the next wall (Cunliffe 1975, 19). Examination of the junction between work gangs on the west wall (Wall 19) indicates that at this point the process of construction was proceeding from north to south (see Cunliffe 1975 pl. III). (Elevation numbers are shown in Figure 53 in Appendix I).

Brick appears to have been in even shorter supply here than at Pevensey. Although brick was present on some of the bastions and wall sections in the south-west quarter of the circuit, the majority of bonding courses were built using sandstones and limestones. As at Pevensey, these

Mapping of the stone types present in the bonding courses does not give too great an insight into the order of

80

unit. The workforce was probably spread fairly evenly around the circuit, construction was at a uniform rate, and the heights of the bonding courses were carefully coordinated. It may be that a gap was left in the west wall, which was probably the point of the circuit closest to the water, to be completed once construction of the rest of the fort was complete.

blocks were difficult to quarry and transport, and unwieldy to install. They would probably have been an unwelcome substitute for brick. 4.5.4

Dover

The evidence for changing supply patterns during the construction of the Shore Fort at Dover has already been touched upon above (4.3.4). The excavated sections of the perimeter wall show it to have been built with a rubble-mortar core, enclosed by a chalk and tufa outer face and chalk inner face. There is as yet too little evidence to determine whether the wall itself was built in sections, but differences between the bastions suggest that construction took place in two distinct phases. A number of the bastions are integral with the wall and are built using the same materials as the wall. Alternating with these are a second series of bastions that were added to the wall at some point after its construction, albeit perhaps only marginally later (Wilkinson 1994). These latter bastions were built with a flint facing, and with brick bonding courses. Presumably by the time they came to be added, the tufa supplies had been exhausted and it was necessary to exploit new sources of raw materials. 4.5.5

4.6

Conclusions

In his recent study of Romano-British architecture, De La Bedoyere has argued that “successful projects were as much a result of luck as design” and is dismissive of the image of Roman building as arising from competent planning and techniques (1991, 15). Without wishing to comment more widely on the issues he raises, it is clear that the Shore Forts were built by men who were both expert in their trade and adaptable to circumstance. Within what are comparatively simple structures there can be seen adjustments of technique to incorporate new design features, and to work with materials which, if not always of particularly high quality, were always serviceable. Mistakes can be detected, most of which, it should be noted, are at Richborough. Here, the north wall is not straight, the western ditch had to be re-aligned, and a foundation for the east wall was constructed and then abandoned (Johnson 1981). None of these mistakes were critical in any way, however, and there is little sign of haste or shoddy workmanship in the surviving remains of any of the Shore Forts. The only possible exception is Lympne, where the core rubble was tipped rather carelessly into the cavity between the facing walls.

Burgh Castle

The example of Burgh Castle offers a very curious contrast with the sites of Richborough, Pevensey and Dover. Here, the defences display a remarkable homogeneity, being built with stone that was probably from a single source area, and apparently as a single unit. It is possible, although perhaps unlikely, that this situation may be illusory, as careful co-ordination of the heights of the bonding courses may disguise the joins between work gangs. There are many vertical and near vertical cracks in the wall which may indicate such divisions, but the spacing between them is often extremely short and rather random, and they are more likely to relate to the uneven settling of the foundations.

The preference for ‘dry’ footings over those employing concrete might be taken as an attempt to speed up the construction process, since it eliminated the need for a pause between the completion of the foundations and the start of work on the superstructure. However, footings employing dry materials can be found in other contemporary structures, one example being the Allectan ‘palace complex’ in London, which was intended to be a far larger structure than the Shore Forts (Williams 1993). Dry foundations were clearly considered adequate for major buildings, and probably offered a saving of effort that had no detrimental effect on the structure. The quality of the foundations at Pevensey suggests that even this may not have been the case, since construction of the timber frame must have been a lengthy exercise requiring skilled craftsmen.

Measurement of the bricks in the bonding courses also indicates that the wall was rising at a uniform rate at all points around the circuit. A cursory study shows that particular sizes of brick can be found at the same altitude on all three surviving walls. One example can be found in the large number of bricks of dimensions 24cm x 24cm (bessales), which are widely present in the 3rd bonding course above the modern ground level. Other brick units have a more localised distribution, such as those measuring 23cm x 37cm, which are concentrated on the west end of the south wall.15

Most of the evidence from the Shore Forts gives the impression of a project to which considerable care was devoted. As well as the foundations at Pevensey, Portchester and Richborough, the attempts at decoration albeit rudimentary - imply that effort was expended over and above that which was strictly necessary for the completion of the defences. Although supplies of raw

On balance the evidence does tend to suggest that the surviving walls of Burgh Castle were built as a single 15

On the classification of brick and tile types see Brodribb (1987, 3 & 151-152).

81

the trench is vertical-sided, in others the sides are sloping. The manner of filling the trench also differs, even when the same types of raw material were being employed. Sometimes the superstructure began within the trench, in other cases construction began at or above the Roman ground surface. If the same group were responsible for two or more of the forts, we should expect to find some evidence of a common methodology, but this appears not to be the case. If a learning process was occurring – and here we return to the idea of an evolutionary development – it was not a single architect or single military unit that was doing the learning. It is quite plausible that several separate projects were taking place around the south and East Anglian coasts of Britain, each assigned its own labour force. Some building gangs may have had more experience of the new architectural styles than others, perhaps because they had previously worked on Gallic projects: alternatively, some groups may simply have been more successful in the manner in which they attempted to incorporate these unfamiliar features. Early features such as the rampart were retained by some groups, whilst it was dispensed with by others in favour of a thicker, free-standing wall. The mix of design features that we can observe in the Shore Forts’ architecture, as summarised overleaf in Table 3, may well be a result of this situation.

materials were not always constant (witness the shortage of brick at both Pevensey and Portchester) the choice of free-standing scaffold at Burgh Castle argues against any attempt to economise on raw materials. The fact that local materials were adequate for the task of the building the Shore Forts, as demonstrated in 4.3, goes a long way to explaining the distribution patterns that were observed in Chapter Three. However, the low quality of locally available stone in certain areas, particularly in East Anglia, offers us an explanation as to why better construction materials such as Lincolnshire Limestone were being transported over considerable distances. It does not, however, answer the question as to why these particular lithologies are present at certain forts, why Lincolnshire Limestone, for example, is present at Bradwell, and not at Burgh Castle or Caister. This topic will be addressed in the following chapter (5.2.3). When we consider the architecture of the Shore Forts, it is evident that the new style defensive features, particularly bastions, were incorporated into the design with varying degrees of success. Only at Burgh Castle, however, was there a genuine failure to get to grips with this novel concept in defensive architecture; here, the field of view of the angle bastions was too restricted to fully protect the adjacent walls (see Johnson 1983a).

To conclude, whether some or all of the forts are contemporary, the details of their design and the differing construction methods would tend to imply that they were built by several independent work groups or military units. This is a reasonable conclusion, although the loss of Walton Castle prevents us from seeing the full picture As stated earlier, the drawings of this particular fort are sufficient to show that it is superficially similar to Burgh Castle. Perhaps the two forts were built in identical style by the same gang of workers, but this is beyond proof. In Chapter Five we will turn to consider the building of the Shore Forts, and the economic and political implications of their construction. The findings of the present chapter suggest that we must take forward a view of a project in which our forts are being built, if not necessarily contemporaneously, then certainly in comparative isolation from one another.

The detailed analysis of the construction methods across the Shore Forts as a whole, however, tends not to support Johnson and Cunliffe’s scheme for a typological evolution of the monuments’ defences. Burgh Castle need not necessarily be seen as a ‘transitional’ design, an interim stage in a learning process about the means of incorporating bastions into a defensive perimeter. Most of the evidence points toward the forts having been built by separate groups, although the use of opus signinum mortar as a form of decoration is a curious trait that is common to five of the eight later sites. Some differences in the details of the architecture can be explained in terms of adaptation to the raw materials, but most cannot be justified in this way. The foundations are particularly informative about this subject. On some sites

82

Table 3.

Summary of the architecture of the Shore Forts Internal towers

Brancaster Caister Burgh Castle1 Walton Castle Bradwell Reculver Richborough Dover2 Lympne Pevensey Portchester

Y ? ? Y -

Rampart

External bastions Y Y Y Y Y Y Y Y

Y Y ? Y Y Y -

1

Bonding courses Y Y Y Y Y Y Y Y

Opus signinum mortar Y ? Y Y ? Y Y Y

Burgh Castle: inconclusive evidence for a tower in the north-east angle, and for an interval tower on the south wall, close to the south-west angle. 2 Dover: bonding courses present only on certain of the bastions, those which are built of flint or brick. Opus signinum is not recorded anywhere in the structure, but given the association between the presence of brick and the presence of opus signinum, it would be no surprise to find it on the bastions which have bonding courses.

83

CHAPTER 5. BUILDING THE SHORE FORTS

5.1

However, a close inspection of the surviving superstructures was necessary in order to gather information about the finer details, which were rarely recorded elsewhere. This examination was largely concerned with establishing the relative proportions of each raw material, particularly within the inner and outer faces.

Introduction

The discussions of the Shore Forts in Chapters Three and Four were largely of a qualitative nature. In the present chapter, an effort is made to add a quantitative dimension to the study. An analysis is made of the scale of the building programme, and leading from this an attempt is made to show how the demands imposed by such a construction project could have been met.

In order to establish these proportions, a modal analysis of the fort walls was undertaken. Point counting is a wellestablished method in sedimentology, where it is used to assess the modal composition of sedimentary rock in thin section under the microscope (Chayes 1956; Tucker 1988). In the present study it was used to determine the relative proportions of:

The initial parts of the chapter are concerned with establishing the overall requirements of the Shore Fort defences, which can be divided into three basic categories: raw materials, transport, and manpower. Once general estimates have been made, the chapter moves on to discuss the management of the Shore Forts’ construction. As discussed in Chapter Two, the time scale over which the installations were built – particularly the eight later forts – is uncertain. Of especial interest, therefore, is an analysis of how these varying periods of construction might have impacted on the organisation of the project, and on the lengths to which the Roman State16 would have had to go in order to bring it to completion. 5.2

• • •

At every fort where Roman masonry was well preserved, a series of elevations, each 1m wide, were selected for analysis. When analysing the face, one elevation was chosen from each distinct unit or ‘construction section’. The precise location of each elevation was chosen on the basis of which section in that particular area of wall was the best preserved. Each elevation was photographed, and the resulting print could then be examined in much the same manner as could a thin section of a rock. A grid of points was placed over each photograph, and the material type (whether a particular lithology, brick or mortar) lying under each point was recorded. This counting process was carried out at the fort itself. At the lowest levels of the wall the material could be directly examined: higher up, the different lithologies were identified (rather less effectively) with the aid of binoculars.

Raw materials

The first stage in quantifying the building project is an assessment of the demand for raw materials. Estimates are made below of the total amounts of stone, mortar, brick and timber required by the Shore Fort defences, as well as the relative proportions of each type. In this way we can gain an impression not only of the total scale of the raw material demand, but can also understand which lithologies assume the greatest importance, in terms of quantity, within each Shore Fort. The results are discussed below, whilst the estimates of the raw materials, listed by fort, can be found in Appendix II. 5.2.1

stone to mortar in the core stone to mortar in the face the different types of stone in the face

The result of the point-count was an estimate of the relative proportion of each material type in that particular elevation. Using these proportions, the final step was to calculate the total volume of each material type within the structure as a whole.

Methods of quantification

Much of the analysis took the form of a desk-based study. The gross dimensions of the forts, such as the length and thickness of the perimeter walls, were available in a number of published sources, whilst many other data could also be gleaned from excavation reports, summarised in Appendix 1.

At the better-preserved forts, this methodology enables a reasonable impression to be gained of the relative volume of each type of raw material within the structure. However, it does have certain limitations, of which the reader should be aware: •

16

The term ‘state’ is used in the most general sense to denote the political authority in Britain. As well as the emperor in Rome, it could equally apply to the rulers of the Gallic Empire (AD 259-274) or the British Empire (AD 285-296).

84

Values are being computed in two dimensions only. This was only a problem for the analysis of the core, since the face is effectively only one ‘grain’ deep. To compensate for this problem, a series of

division of the monuments into a number of categories, shown below as Table 4.

elevations of the core of each fort were analysed: the results proved consistent to within 5%. Practical problems were encountered in identifying stone types, quite possibly leading to recording errors. This was a particular difficulty at Richborough, where the north wall is heavily covered by lichen. In every case, the results of the analysis of surviving wall sections had to be used as the basis for calculations of the amount and proportion of materials in fallen sections. The length of such sections is often quite large, and is over 200m at Pevensey and Richborough (in both cases roughly one-third of the entire circuit). However carefully the modal analysis is undertaken, the form of these missing sections remains conjectural.

•

•

5.2.2

For the best preserved of the monuments, the reconstruction can be proposed with considerable certainty, even down to the relative proportions of the materials used in the facing. Excavation reports tend to provide significant data about both the foundations and superstructure, whilst the latter is readily accessible for direct study. Even for these forts, however, some assumptions have to be made, but these tend to relate to minor details and do not influence the overall results to any great extent. Where the preservation of a site is less good, the task of reconstruction is more difficult, and the balance between those calculations based on data, and those based on assumption, begins to tip in favour of the latter. In many cases the exercise of reconstruction emphasises just how little we know about the defences of these forts. In the case of Walton Castle, a site about which we know practically nothing, the estimates are no more than guesswork.

Assumptions

The preservation of the Shore Forts has a significant influence on the accuracy of the quantification process. The variable survival of the defences has resulted in the

Table 4.

The preservation of the Shore Forts

Site

Preservation, quality of data, assumptions

Pevensey Burgh Castle

The preservation of the Roman fabric, including the facing is extremely good. The majority of the superstructure survives for first-hand study, whilst excavation reports provide clear data about the form of the foundations. It is possible to estimate the overall quantity of raw materials used in the defences with considerable certainty. Good, if not absolutely precise, estimates can be made of the relative proportions of each material Moderate preservation of the defences. It is possible to estimate the overall volume of raw materials needed, but post-Roman repair or the loss of significant parts of the defences makes it difficult to assess the precise quantities of each type of material present The overall preservation of the defences is poor, or limited to isolated sections of the perimeter. Only small areas of the facing survive. Whilst the plan of the defences is reasonably well understood, the wall stands nowhere to its original height. Estimates of the total volume of raw materials rely in some part on assumptions as to the height and profile of the wall Whilst the plan of the defences is reasonably well established only a limited amount of the superstructure survives to the modern day. The original height and profile of the wall are unknown, and estimates of the total volume of raw material rely heavily on assumptions about the form of the wall

Richborough Portchester Lympne Reculver

Brancaster Caister-on-Sea Bradwell Dover Walton Castle

The destruction of the fort leaves no data more recent than the 18th century. The plan of the defences and the fabric are virtually unknown. The reconstruction is entirely hypothetical. best-preserved sites, namely Pevensey, Burgh Castle, Richborough and Portchester:

At every fort, certain assumptions had to be made to compensate for the limits of the data. As well as solutions relating to individual forts, which are listed in Appendix II, there were a number of ‘standard’ assumptions that were applied to the monuments as a whole. These solutions were based on a combination of data from the

The height and profile of the defences This was by far the most crucial estimation that had to be made, and was one required for all monuments apart from

85

narrow, rampart-backed wall (Dover). These sites are assumed to have walls rising to 5m (6.5m when the parapet is included), in keeping with ‘traditional’ 2nd century defences elsewhere in the north west Empire (Johnson 1984). Unless, as at Reculver, there is evidence to suggest otherwise, it is also assumed that the walls rose to their full height without internal offsets: certainly this is true of Dover, where none are present up to 4.5m.

Pevensey and Burgh Castle. At these latter sites the thickness of the highest surviving parts of the superstructure (1.5m - 2m) suggests that both stand close to their original height. This evidence indicates that the defences of Burgh Castle were 5m to wall-walk height, above which was a parapet that would have been around 1.5m tall. Pevensey was considerably larger, probably 8m or more to the wall-walk and perhaps 9.5m at the top of the parapet. In both cases the inner face of the wall was thinned as the wall rose higher: distinct offsets can be seen at Pevensey, whilst the wall of Burgh Castle has an even taper.

The second group of forts comprises Bradwell, Richborough, Lympne, Pevensey and Portchester. With the exceptions of Bradwell and Lympne, the defences of these sites survive above 6m: all are characterised by a significantly wider superstructure at ground level than the forts in the first group. All were probably comparable to Pevensey in terms of scale, and are assumed to have walls rising to 8m (9.5m to the top of the parapet). Internal offsets are known at Pevensey and Lympne, whilst the inner face at Richborough also exhibits a distinct taper. It is assumed they were also present in the defences of Portchester and Bradwell.

Of the other Shore Fort defences we can be less certain, since much of the superstructure has been lost. Richborough (7m) and Portchester (8.3m) probably stand close to their original height, although the uppermost parts of Portchester’s defences have been rebuilt in the post-Roman period. Elsewhere, with the possible exception of Dover, where sections of the wall 4.5m high have been excavated, at no other of our monuments does the wall stand close to its original height. The walls of Reculver and Lympne survive to 2.4m and 3.9m respectively, whilst only stubs of the superstructure, less than 1m high, remain at Brancaster, Caister and Bradwell. In all of these cases the original height and the profile is uncertain, and in the absence of data, the monuments have been divided into two groups. These have been summarised in Table 5.

Such generalisations must be treated with some caution given the lack of standardisation in Shore Fort architecture, as outlined in Chapter Four. As we shall see below, these assumptions have a very significant bearing on the way we will view the overall size of each fort as a building project. However, they are probably justified: it is hard to imagine, for example, that the defences of sites such as Bradwell, which were equal in width to those at Pevensey, were not of comparable height.

In our first group are sites of early, or suspected early, date (Brancaster, Caister, Reculver), and those which demonstrate the same early sylistic characteristics of a Table 5.

Proposed reconstruction of the profile of the defences.

Site Brancaster Caister-on-Sea Burgh Castle Walton Castle Bradwell Reculver Richborough Dover Lympne Pevensey Portchester

Thickness of the wall at base 2.7m 2.9m 2.1m - 3.2m ? 4.2m 3m 3.3m 2.6m 3.9m 4.2m 3.7m

Surviving height 0.6m 0.3m 4.0m 0m 0.5m 2.4m 7.0m 4.5m 3.2m 8.2m 8.3m3

Assumed original height1 6.5m 6.5m 6.5m ? 9.5m 6.5m 9.5m 6.5m 9.5m 9.5m 9.5m

1

Internal offsets: present/ absent/ assumed? assumed absent assumed absent Present2 ? assumed present present assumed present absent present present assumed present

The figure given includes the parapet, which is assumed to have been 1.5m in height. The inner face is gradually tapered, rather than offset at vertical intervals. 3 The upper parts of the superstructure have been rebuilt in the post-Roman period. Recognisably Roman masonry does not survive to this height. 2

86

The relative proportion of the raw materials The presence of construction sections (detailed in 4.5) may well suggest that not every part of the perimeters at Richborough, Lympne, Pevensey and Portchester were built contemporaneously. If true, then the scaffold timber used in one part of the circuit could have been dismantled and re-used when a new section came to be built. To take account of this possibility it has been assumed that during construction only half of the wall was covered by scaffold at any one time. The exception is once again Burgh Castle, where the evidence suggests that the whole perimeter was built as a single unit, thus preventing any recycling of scaffold timbers (see 4.5.5).

The point-count analysis of Richborough, Lympne and Pevensey demonstrated that the ratio of stone to mortar in ashlar facing was around 75:25. At Portchester and Burgh Castle the irregular morphology of the flints in the facing produced a ratio of 60:40. At all of the five sites, analysis suggested that the ratio of rubble to mortar in the core fill of the wall was around 60:40. This analysis enables reasonably good estimates to be made of the proportion of stone to mortar in the remaining Shore Forts. It cannot, however, say anything about the relative proportions of the different stone types present in a given structure. In the absence of clear data on this matter, what information there existed (often very little) was used as the basis of rough estimates of the proportion of each material type. These assumptions are noted in Appendix III, and the reader should be aware of the speculative nature of many of the estimations that result.

5.2.3

The demand for raw materials

The calculations suggest that the total volume of raw materials demanded for the construction of the perimeter defences of the eleven Shore Forts was of the order of 200,000 cubic metres. Adopting an arbitrary average of 2500 kg/m3 this equates to 500,000 tonnes of stone. By comparison, calculations made for Hadrian’s Wall have suggested that the construction of the curtain wall, milecastles and turrets required some 1,178,000m3 of material. This figure excludes estimates for the turf wall at the western end of the Wall, and the nearby forts. (Bennett 1990).

Mortar Mortar analysis was not within the scope of this study (see 1.2.3) and thus the components are not known in any detail. In estimating the relative proportion of the materials used in the making of the mortar, the guidelines laid down by Vitruvius have been followed (II.v.5-9). It is assumed that the ‘standard’ non-hydraulic mortar, such as was used in the core, comprised two parts sand, one part aggregate (which in the case of the Shore Forts was normally beach pebbles) and one part lime. Opus signinum mortar comprised two parts sand, one part crushed ceramic (brick or tile) and one part lime. To both of these mortars was added water, normally 15-20% of the total volume of the other components.

Finally, then, we are able to begin to assess the scale of the Shore Fort project. Large though the demand for raw materials was, it is immediately apparent that the volumes which we are discussing for our forts are considerably smaller than were required for Hadrian’s Wall. As can be seen in Figure 46, the volume of raw materials required for each fort differed very considerably. Sites with the lower, narrower walls generally required between 12,000 and 14,000m3, whilst those with more substantial defences consumed far greater volumes. In part this may be a consequence of the rigid division of the forts into two categories (Table 4). However, the contrast between Burgh Castle (estimated at 14,300m3) and Pevensey, which has the longest and widest superstructure of any Shore Fort (estimated at 33,710m3) is a genuine one, and demonstrates the very different scales of the various defences. Cotterill (1993) has argued that the new, larger style of the later defences was a case of Britain imitating the latest continental ‘fashion’, but the massive increase in raw materials required by forts such as Pevensey would tend to suggest that this is unlikely. Whilst fashion always comes at a cost, a nearly threefold increase in the requirement for material and manpower seems too extravagant a price to have been justified.

Scaffold As was noted in 4.2.3, there is evidence for two types of scaffold being used during the construction of the Shore Forts. Putlog holes in the walls of Pevensey and Richborough indicate that engaged scaffold had been used, whilst in the case of Burgh Castle the absence of such holes infers that free-standing scaffold was employed. Elsewhere, the partial or total destruction of the superstructure makes it impossible to determine which type of scaffold was used. In the interests of maintaining a minimum estimate, where there is no evidence to the contrary, the calculations of the demand for timber assume that engaged scaffold was being used. (This practice required half the number of vertical standards needed for free-standing scaffold).

87

Figure 46.

Estimated raw materials required for the construction of the Shore Fort perimeter defences

Figure 47.

Raw material use (percentage by volume): Pevensey Within the core itself, rubble stone was the dominant material, followed by the components of the mortar (sand, beach pebbles and lime). By contrast, the facing stones – which the provenance study suggests were the only raw materials travelling any great distance – account for only 5% of the overall demand. So too timber, which constituted only 6% at Pevensey, and still less at sites such as Caister (2%) where it was not used in the foundations, but only for the scaffold.

On the level of the individual fort, the estimates are informative about the significance of the various types of material used in the structure. This can be illustrated by examination of Pevensey Castle, which is typical of the Shore Forts as a whole (shown as Figure 47 and Figure 48). The rubble core consumed the overwhelming proportion of raw materials, accounting for 74% of the total volume.

88

Figure 48.

Estimated volume of raw materials - perimeter defences at Pevensey building stone - although the difficulties of establishing a provenance for flint and septaria means that the results must be treated with an (albeit very small) degree of caution (see Chapter Three). The reader should also be aware of the assumptions made for Richborough and Lynpne, which significantly affect the results (Table 6, notes 1 & 2).

The introduction of bonding courses, although a striking aspect of the design, did not create a great demand for brick: within the defences at Pevensey, only 80m3 of brick had been used (0.0023%). The innovation of bastions, however, was moderately significant, increasing the overall material requirement by 7%. Table 6 and Figure 49 illustrate how sources within a few tens of kilometres of the forts provided the bulk of Table 6.

Distances between quarries and Shore Forts - building stone only

Site Brancaster Caister-on-Sea Burgh Castle Bradwell Reculver Richborough 1 Dover Lympne 2 Pevensey Portchester

0 - 1 km (%) 0 0 0 0 0 17 81 93 0 0

1 - 10 km (%) 29 0 0 98 76 0 19 7