The Rise and Decline of Druce Farm Roman Villa (60-650 CE): Excavations 2012-2018 9781407360010, 9781407360027

Table of contents :
Cover
Title page
Copyright
Archaeology of Roman Britain Subseries
Titles in the Archaeology of Roman Britain Subseries
Of Related Interest
Acknowledgements
Table of contents
List of figures
List of tables
Preface
Summary
Résumé
Zusammenfassung
1. Environment and project background
Location
Landscape and geology
The site
Project background
Geophysical surveys
The excavations
Archaeological methodology
Local community outreach
Funding
Project aims
Summary of Periods
Period 1: Prehistoric up to 43 CE (Fig. 1.7)
Period 2.1: 43-70 CE (Fig. 1.8)
Period 2.2: 70-120 CE (Fig. 1.9)
Period 3.1: 120-200 CE (Fig. 1.10)
Period 3.2: 200-250 CE (Fig.1.11)
Period 4.1: 250-300 CE (Fig. 1.12)
Period 4.2: 300-350/70 CE (Fig. 1.13)
Period 5: 350/70-430 CE
Period 6: 430-650 CE (Fig. 1.14)
Period 7: 650+ CE
Radiocarbon dating
Introduction
Sampling strategy
Results and Calibration
Deposition and further work
2. The Prehistoric background
Early Neolithic activity
The features
Pre-Roman ditches
Ditch 878
Ditch 1291
Ditch 1325
Ditch 1752
Flint and chert
Introduction
Methodology
Raw material
Description of the total assemblage
Cores and core reduction techniques
Flakes and blades
Chips and shatter
Tools
Distribution of the assemblage across the site
Discussion
Illustrated flint and chert artefacts
Prehistoric pottery
Fabric Code
Discussion
Conclusion
3. Enclosures, ditches and middens
The Early Roman enclosure system
The outer enclosure (ditches 1604, 309, 313, 311 and 312)
The inner enclosure (ditches 308, 314 and 1605)
Southern outliers (Ditches 1504 and 310)
Structures associated with a south entrance to the villa
Conclusions
Other ditch systems
Ditches 1641/1644
Ditches 1459/1510
Gullies 372 and 384
Ditches 1260/1305
Eastern enclosure
Ditch 1262
Ditch 1412
Ditch 1410
Ditch 1411
Conclusions
The midden deposits
Midden 315 T2
Midden 1529 T24, T41 and T46 (Fig. 3.15/1)
Midden 1355 T59
Midden 1528 T14 and T52 (Fig. 3.17)
Conclusion
4. Early Roman activity outside the villa complex
Early Roman activity in Lower Limepits field
Pit 1399 T14 (Fig. 4.1/1)
Pit 643 T21 (Fig. 4.1/2)
Activity pre-dating the East Range
Early Roman activity in Eleven Acres
The Ancillary Building
Room AB1 (pits 2015, 1996, 2008, 1991, 2032, 2008 and 1994)
Pit 2015 (Fig 4.7/1)
Pit 1996 (Fig. 4.7/2)
Pit 1991 (Fig. 4.7/3)
Pit 2031 (Fig. 4.7/4)
Pit 2032 (Fig. 4.7/7)
Room AB2 (pits 2000, 2017 and 2036)
Pit 2000 (Fig, 4.7/8)
Pit 2017 (Fig. 4.7/10)
Room AB3 (pits 2295/2020/2038, 2050 and 2052)
Pit 2295 cut by pits 2020 and 2038 (Fig. 4.7/11)
Pit 1735
Occupation Area T82
Pit 1800 (Figs. 4.9/2 and 4.12).
Deposit 1795 (Fig. 4.10)
Pit 1845 (Fig. 4.9/3)
5. The North Range
Phasing
Phase 1: Early Roman, Period 2.2 (c. 70-120 CE) (Fig. 5.4 Phase1)
Room N10
Pit 643 (Fig. 4.1/2)
Pit 644 (Figs 5.10/3 and 5.11)
Conclusions
Phase 2: Middle Roman, Period 3.1 (c. 120-200 CE) (Fig. 5.4 Phase 2)
Room N12
Room N1
Conclusions
Phase 3: Middle Roman Period 3.2 (c. 200-250 CE) (Fig. 5.4 Phase 3)
The west extension (Fig. 5.13)
Room N2
Room N5
Extension of the original building (Figs. 5.4/3 and 5.17)
Room N3 (Fig. 5.12).
Room N6
Room N8
Room N7
Room N10
Corridor N11
Conclusions
Phase 4: Late Roman Period 4.2 (c. 300-350/70 CE) (Fig. 5.4 Phase 4)
The west wing
Room N1
Room N2
Room N4
Paved area and potential building
Room N5
The north wing
Room N3
Room N10
Room N9
Corridor N11
Gully 407
Conclusions
Phase 5: Final Roman Period 5 (c. 350/370-430 CE)
Room N1
Room N3
Room N7
Room N10 (plan Fig. 5.42)
Room N9 (Fig. 5.37/1)
Room N11
Conclusions
Phase 6: Post-Roman Period 6 (c. 430-650 CE) (Fig. 5.4 Phase 5)
Activity south of Room N3
Room N1
Room N3
Room N9
Corridor N11
Collapse south of the North Range
Conclusions
General conclusions
6. The East Range
Phase 1: Early Roman, Period 2.2 (c. 70-120 CE)
The first rectangular building
Features relating to this phase
Conclusion
Phase 2: Middle Roman, Period 3.1 (c. 120-200 CE) (Fig. 6.8)
Extension of the rectangular building and associated features
Tank 778 and associated features
Pit 381 (Figs 6.11/2 and 6.12).
Pit 1149 /1181 (Figs 6.13 and 6.14/1)
Pit 896
Pit 209 (Fig. 6.14/2)
Phase 3a: Late Roman, Period 4.1 (c. 300 CE)
The Aisled Hall
Flint-packed pit 638 (Fig. 6.19/1)
Flint-packed pit 648 (Fig. 6.19/2 and 6.20)
Deposit 1269 (Fig. 6.21)
Pit 1077 (Figs 6.19/3 and 6.22)
Phase 3b: Late Roman, Period 4.1 (c. 350 CE)
Fire damage and site preparation for re-building
Phase 4: Late Roman, Period 4.2 (c. 350/70 CE) (Fig. 6.23)
Extension of the Aisled Hall
The flint-packed pits
Internal walls
Floor surfaces
Deposit 556 (Fig. 6.24)
Oven 207 (Figs 6.23, 6.25-28).
Structure 1203 (Fig. 6.29)
Pit 823 (Fig. 6.30)
Phase 5: Final Roman and Post-Roman, Periods 5 and 6 (c. 430-650+ CE)
Collapse of the building
Material over the collapsed building
Collapse sequence
The roof
Collapse of the north wall 559 (Fig. 6.31) (Period 6 c. 430-650 CE)
Phase 6: Early Medieval, Period 6 (c. 430-650 CE)
Post-collapse activity
Pit 209 (Figs. 6.14/2 and 6.35)
Pit 825 (Figs 6.36 and 6.37/1: the pit is located on section line AB on Fig. 6.26)
Pit 752 (Figs 6.36 and 6.37/1)
Pit 944 (Fig. 6.37/2)
Pit 962 (Figs 6.37/1 and 2: section lines AB and CD shown on Fig 6.26)
Conclusions
General conclusions
7. The West Range
Phase 1: Middle Roman Period 3.1 (c. 120-200 CE) (Fig. 7.3/1)
Phase 2: Period 4.1 Late Roman (c. 250-300 CE) (Figs. 7.1 and 7.5)
Room W1
Room W2
Room W3
Room W4
Conclusions
8. The grain dryer and associated burial
The grain dryer
The flue (Fig. 8.3)
The flue walls
Section 1850 (A) (Fig. 8.5/1)
Section 2075 (B) (Fig. 8.5/2)
Section 1963 (C) (Fig. 8.5/3)
Section 1969 (D) (Fig. 8.5/4)
The stoke pit and burial
The stoke pit 1964 (Figs 8.1, 8.2 and 8.5/5)
The skeleton (Figs 8.1, 8.2/1, 8.5/5 and 8.7)
Conclusions
9. Building materials
Wall construction
Introduction
Non-flint internal walls
Mortar used in the construction of the building
Roofing and ceramic building materials
Introduction
Methodology
Ceramic building material
Stone roof tiles
Roofing slates
Stone building material and architectural fragments
Introduction
Raw material
Column fragments
Jambs?
Post pad
Miscellaneous architectural fragments
Illustrated architectural fragments and building stone
Note on the column capital from context 197
Other occurrences of building stone
Conclusion
10. Internal decoration
Mosaics
Introduction
Room N1: Mosaic 1 (Context 153)
Room N2: Mosaic context 1059 (below Mosaic 2)
Room N2: Mosaic 2 (Context 246)
Other mosaics of the North Range (see Fig. 5.2 for room locations)
Room N3: Mosaic context 210
South of Room N3
Room N4: Mosaic context 98
Room N5
Room N6: Mosaic 3 (Context 357)
Room N7: Mosaic 4 (Context 329)
Room N10
Room N11: Mosaic context 99
The East Range
The West Range
Glass tessera
Discussion
Painted wall plaster
Introduction
The assemblage
Methodology
Room N1
Room N12
Room N2
Room N4
Room N5
Room N3
Room N6
Room N7
Room N9
Room N10
Room N11
Room N3 Infill
Midden - South of Room N3
Midden - East of Room N4
East Range /Aisled Hall
Middens and Other Pits
Discussion
11. Roman and post-Roman pottery
Structure of the report
Imported finewares
Samian
Central Gaulish Terra Nigra (F11) CNG TN
Lyon Colour-coated ware (F18) LYO CC
Argonne Colour-coated ware (Sinzig Roughcast) (F14) ARG CC
Moselkeramic Black-slipped ware (F3) MOS BS
Central Gaulish Black-slipped ware (F2) CNG BS
Overview of amphorae
Mortaria
Early Roman (Period 2 c. 43-120 CE)
Middle Roman (Period 3 c. 120-250 CE)
Late Roman (Period 4 c. 250-350/70 CE)
Late Roman (Periods 4-5 c. 250-400 CE)
Final Roman (Period 5 c. 350/70-420 CE)
Discussion
Catalogue
British finewares
Corfe Mullen ware (F10).
New Forest finewares
Oxfordshire wares
Lower Nene Valley Colour-coated ware
Unsourced finewares
Coarsewares
(South East) Dorset Black Burnished ware (DOR BB1)
South East Dorset Orange Wiped ware (SEDOWW)
South Western Black Burnished ware (SOW BB1)
New Forest greyware
Alice Holt Reduced ware (ALH RE)
Hampshire Grog-tempered ware (HAM GT)
Other coarsewares identified to kiln
Unsourced Coarsewares
Sherds bearing graffiti
Post-Roman pottery
Medieval pottery
12. Pottery assemblages
Introduction
Methodology
The Assemblages
Period 2: c. 43-120 CE
Periods 3 and 4: c. 120-350/70 CE
Period 4.2: c. 300-350/70 CE
Period 5: c. 350/70-430+ CE
Periods 6 and 7: c. 450/500-1000 CE
Discussion
13. Portable objects of stone, shale, glass, bone and fired clay
Stone Objects
Introduction
Raw material
Querns and millstones
Mortars and pestle
Grinding stones
Sharpeners and polishers
Counters
Miscellaneous stone objects
Kimmeridge Shale
Introduction
Methodology
Source of material
Results
Discussion
Catalogue
Roman glass
1st/2nd century vessels (Periods 2.1-3.1)
Colourless cups and bowls of the 2nd to 3rd centuries (Periods 3.1-4.1)
Late Roman vessels (Periods 4.1-5)
Catalogue
Bone objects
Personal adornment
Writing and communication
A note on the bone folding scalprum
Gaming equipment
Domestic utensils
Tool
Tools from the animal bone assemblage
Fired clay objects
Methodology
Conclusions
14. Non-ferrous metal finds
Introduction
Coins
Coin catalogue
Medieval coins (11)
Comments on the Roman coin assemblage
Copper alloy objects
Personal adornment
Bracelets
Finger rings
Ear ring
Necklaces/bracelets
Pins
Toilet implements
Household utensils
Sewing needles
Objects associated with written communication
Fasteners and fittings
Items with a military association
Note on the Late Roman military belt fitting
Miscellaneous and other unidentified fragments
Post-Roman and Early Medieval objects
Lead objects
Pot mends
Weights
Counters
Household utensils
Miscellaneous
Offcuts and waste
Post-Medieval lead
15. Iron objects and metalworking debris
Iron objects
Metalworking tools
Woodworking tools
Possible modelling tool
Leather working tools
Transport
Keys
Domestic equipment
Knives
Structural fittings
Nails
Miscellaneous items
Assessment and analysis of the metallurgical debris
Introduction
Methods and key terms
Results
X-Ray Fluorescence analysis
Discussion
Conclusion
16. Human remains
Perinatal human remains
Introduction
Preservation and taphonomy
Results
Anthropological data
Pathology and non-metric skeletal traits
Discussion
Adult human remains
Introduction
Anthropological data: sex, age, and stature
Pathology
Discussion
Preface to Chapters 17 and 18 (Faunal and environmental remains)
Animal bone
Charred plant macrofossils
Charcoal
Radiocarbon dating programme
17. Faunal remains
Animal bones
Introduction
Recording methods
The hand-collected assemblage
Preservation and taphonomy in the main hand-collected assemblage
Period 1: Prehistoric up to c. 43 CE
Period 2.1 Early Roman c. 43-70 CE
Period 2.2: Early Roman c. 70-120 CE
Period 3: Middle Roman c. 120-250 CE
Period 3.1: Middle Roman c. 120-200 CE
Period 3.2: Middle Roman c. 200-250 CE
Period 4: Late Roman c. 250-350/70 CE
Period 4.1: Late Roman c. 250-300 CE
Period 4.2: Late Roman c. 300-350/70 CE
Period 5: Final Roman c. 350/70-430 CE
Period 5/6: Final Roman/Post-Roman c. 350/70-650 CE
Period 6: Post-Roman c. 430-650 CE
Cattle
Sheep/goat
Pig
Horse
Dog
Cat
Wild mammals
Butchery and carcase processing and disposal
The Associated Bone Groups and skulls
The small vertebrate assemblages
The bulk sample assemblages
Summary of the small vertebrates
Fish
Discussion
Molluscs
Land snails
Marine molluscs
Conclusion
18. Environmental remains
Wood charcoal
Introduction
Methodology
Results
Discussion
Charred and mineralised plant remains
Introduction
Soil processing methodology and selection of samples for analysis
Results
Description of samples by period and feature
Discussion
The integrity of the data
Arable crops
Evidence of crop husbandry recovered from weed taxa
The distribution of cereal, hay and heathland remains around the villa
Other potential crops, orchard fruits and garden plants
Imported foods
Grassland and heathland resources
The Roman plant-based diet at Druce compared to other villas
Summary by period
Conclusions
19. Interpretation and discussion
Introduction
Period 1: Prehistoric up to c. 43 CE (Fig. 1.7)
Early Neolithic
Bronze Age
Field systems
Druce Farm Roman Villa
Introduction
Period 2.1: Early Roman c. 43-70 CE (Fig. 1.8)
Chronology
Settlement and land use
Society
Environment and economy
Period 2.2: Early Roman c. 70-120 CE (Fig. 1.8)
Chronology
Settlement and land use
Society
Environment and economy
Period 3.1: Middle Roman c. 120-200 CE (Fig. 1.10)
Chronology
Settlement and land use
Society
Environment and economy
Period 3.2 Middle Roman c. 200-250 CE (Fig. 1.11)
Chronology
Settlement and land use
Society
Environment and economy
Period 4.1: Late Roman c. 250-300 CE (Fig. 1.11)
Chronology
Settlement and land use
Society
Environment and economy
Period 4.2: Late Roman c. 300-350/70 CE (Fig. 1.13)
Chronology
Settlement and land use
Society
Environment and economy
Period 5: Final Roman c. 350/70-430 CE
Chronology
Settlement and land use
Society
Environment and economy
Period 6: Post-Roman c. 430-650 CE
Chronology
Settlement and land use
Society
Collapse
Post-collapse pit digging
Environment and economy
Period 7: Early medieval c. 650+ CE
Chronology
Settlement, land use, environment and economy
Final Conclusions
Bibliography
Back Cover

Citation preview

BAR BRITISH SERIES 676, 2022

A R C HA E O L O G Y O F R O M A N B R I TA I N VOLUME 8

THE RISE AND DECLINE OF DRUCE FARM ROMAN VILLA (60-650 CE)

Excavations 2012-2018

Lilian Ladle

BAR BRITISH SERIES 676, 2022

A R C HA E O L O G Y O F R O M A N B R I TA I N VOLUME 8

THE RISE AND DECLINE OF DRUCE FARM ROMAN VILLA (60-650 CE)

Excavations 2012-2018

Lilian Ladle

Published in 2022 by BAR Publishing, Oxford, UK BAR British Series 676 Archaeology of Roman Britain, volume 8 The Rise and Decline of Druce Farm Roman Villa (60-650 CE) isbn  978 1 4073 6001 0 paperback isbn  978 1 4073 6002 7 e-format doi  https://doi.org/10.30861/9781407360010 A catalogue record for this book is available from the British Library © Lilian Ladle and East Dorset Antiquarian Society 2022 cover image Reconstruction of Druce Farm Roman Villa c. 350 CE by Gareth Naylor, B.A., B. Arch. 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. Links to third party websites are provided by BAR Publishing in good faith and for information only. BAR Publishing disclaims any responsibility for the materials contained in any third party website referenced in this work.

BAR titles are available from: BAR Publishing 122 Banbury Rd, Oxford, ox2 7bp, uk [email protected] www.barpublishing.com

A R C H A E O L O G Y O F R O M A N B R I TA I N A specialist sub-series of the BAR British Series

SERIES EDITORS Edward Biddulph (Oxford Archaeology), Martin Pitts (Exeter University) Roman Britain presents a dynamic and exciting field of study, with an abundance of data amenable to multi-disciplinary approaches, ‘big data’ studies, the application of theoretical approaches, and a variety of visually stimulating artefacts and reconstructions that speak to our own age in a remarkably direct way. This series promotes research relating to the Roman province of Britannia, spanning a broad period from the late Iron Age to post-Roman Britain (roughly from the 1st century BC to the 5th century AD), as well as encompassing studies that examine the interaction between the British Isles, the nearby Continent, and other parts of the connected Roman empire. If you would like to submit a proposal for the Archaeology of Roman Britain series, please contact [email protected] EDITORIAL ADVISORY BOARD Rob Collins, Newcastle University Andrew Gardner, University College London James Gerrard, Newcastle University Rebecca Gowland, Durham University Birgitta Hoffmann, Independent Researcher John Pearce, King’s College London Alex Smith, Headland Archaeology Stephen Rippon, Exeter University Fiona Seeley, MOLA Ellen Swift, Kent University Lacey Wallace, Lincoln University Philippa Walton, University of Reading Sadie Watson, MOLA Jake Weekes, Canterbury Archaeological Trust

TITLES IN THE ARCHAEOLOGY OF ROMAN BRITAIN SUBSERIES The Clayton Collection An archaeological appraisal of a 19th century collection of Roman artefacts from Hadrian’s Wall Frances McIntosh BAR British Series 646 | 2019 Volume 1 Worcester Magistrates Court Excavation of Romano-British homes and industry at Castle Street Andy Boucher BAR British Series 658 | 2020

Volume 2

London’s Roman Tools Craft, agriculture and experience in an ancient city Owen Humphreys BAR British Series 663 | 2021

Volume 3

Belonging and Belongings Portable artefacts and identity in the civitas of the Iceni Natasha Harlow BAR British Series 664 | 2021

Volume 4

On the Edge of Empire Society in the south-west of England during the first century BC to fifth century AD Siân Alyce Thomas BAR British Series 667 | 2021

Volume 5

Roman Rural Settlement in Wales and the Marches Approaches to settlement and material culture through big data Leah Reynolds BAR British Series 670 | 2022

Volume 6

Dying Young A Bioarchaeological Analysis of Child Health in Roman Britain Anna Rohnbogner BAR British Series 673 | 2022

Volume 7

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Excavations 2012-2018 Lilian Ladle BAR British Series 676 | 2022

Volume 8

For more information, or to purchase these titles, please visit www.barpublishing.com

OF RELATED INTEREST

Sylloge of Defixiones from the Roman West, Volumes I and II A comprehensive collection of curse tablets from the fourth century BCE to the fifth century CE Celia Sánchez Natalías BAR International Series 3077 | 2022 Visitor Experiences and Audiences for the Roman Frontiers Developing good practice in presenting World Heritage Edited by Nigel Mills BAR International Series 3066 | 2021 Drawings of Roman Mosaics in the Topham Collection, Eton College Library Patricia Witts BAR International Series 3064 | 2021 Latin Lexicon of Textiles Clothes, adornments, materials and techniques of Ancient Rome Elena Miramontes Seijas BAR International Series 3051 | 2021 The Resilience of the Roman Empire Regional case studies on the relationship between population and food resources Edited by Dimitri Van Limbergen, Sadi Maréchal and Wim De Clercq BAR International Series 3000 | 2020 Multi-period Occupation at Football Field, Worth Matravers, Dorset Excavations 2006–2011 Lilian Ladle BAR British Series 643 | 2018 Report on the Excavation of a Romano-British Site in Wortley, South Gloucestershire David Wilson, Alan Bagnall, Beryl Taylor BAR British Series 591 | 2014 Rome and the Social Role of Élite Villas in its Suburbs Geoff W. Adams BAR International Series 1760 | 2008 The Horace’s Villa Project 1997-2003, Volumes I and II Report on new fieldwork and research. Volume I: The Reports. Volume II: Documentation Edited by Bernard Frischer, Jane Crawford and Monica de Simone BAR International Series 1588 | 2006 Becoming Consumers: Looking beyond Wealth as an Explanation of Villa Variability Perspectives from the East of England Christopher Barry Martins BAR British Series 403 | 2005

For more information, or to purchase these titles, please visit www.barpublishing.com

Acknowledgements The late Dr Ken Wheatley introduced me to this site in 2009 and suggested the presence of Roman buildings in Lower Limepits field where the Stour Valley Metal Detectorists had unearthed quantities of Roman metalwork and roofing materials. The landowner, the late Mrs Ann Ridout, and her daughter Amanda enthusiastically encouraged and supported the fieldwork undertaken between 2012 and 2018 under the aegis of the East Dorset Antiquarian Society (EDAS). Tenant farmer Bernard Cox is thanked for his patience during those years when dozens of volunteers parked and worked on his very busy farm.

the excavation could not have progressed without this venerable machine. The fieldwork was carried out by 140 individuals who were members of EDAS and other local groups. However, special thanks are extended to the following who committed much of their time for the duration of the project; the archaeological results could not have been achieved without their hard work, enthusiasm and dedication: Janet Bartlet, John Bithell, Sue Cullinane, Alan Dedden, Phil D’Eath, Vanessa Joseph, Len and Pam Norris, Sue Pinyoun, Bryan Popple, Ian Richardson, John Singleton, Finn Stileman, Matthew Tagney, Geoff Taylor, Heather Tidball and Peter Walker.

The work could not have progressed without substantial funding which came from a number of sources. In particular, enormous thanks are extended to the Ridout family for their regular financial contributions, which underpinned the project from start to finish. I am grateful to several members of the excavation team who made significant donations and must particularly mention the generosity of Janet Bartlet and Phil D’ Eath and the estate of the late Graham Adams. We are thankful to countless members of the local community, and to numerous societies from across the region, who made generous contributions after visiting the site. Donations were also received from several local societies, including the Dorchester Association, the Puddletown Society, the Shaftesbury and District Archaeological Society, the Wareham and District Archaeology and Local History Society and the East Dorset Antiquarian Society. Finally, the project is indebted for the support and funding from the Mark Fitch Fund, the Roman Research Trust, the Council for British Archaeology and the Valentine Charitable Trust, who covered the substantial costs required to undertake a comprehensive environmental assessment of the site.

Many thanks go to the metal detectorists, initially members of the Stour Valley Search and Recovery Club. Later, Peter Adcock, Karen Brown, Robert Heaton, Richard Higham and Finn Stileman continued their work, regularly scouring the spoil heaps in and around the villa and the surrounding fields, adding many finds to the material record. The project team was extremely grateful for academic advice and support from Prof. Sir Barry Cunliffe, Prof. Anthony King, Prof. Tim Darvill, Dr Miles Russell, †Anthony Beeson, Paul Cheetham, Stephen Cosh, David Neal and Bryn Walters. Mark Corney is thanked for invaluable advice during the field work, throughout the post-excavation work and during preparation of this volume. I am enormously grateful to Dr Malcolm Lyne, who provided annual pottery identification and spot dates which were used to underpin the phasing of the site. His work was fundamental in dating the archaeological features and many of the material finds. Thanks are extended to all the specialists who contributed towards the artefact studies; their results were vital for the final interpretation of the site. In addition I would like to thank Professor Bill Manning for his advice on identifying some tricky ironwork.

Co-director Andrew Morgan not only helped with the day-to day running of the site, but also managed the crucial financial organisation. His support throughout was unflagging and he is owed an enormous debt of gratitude. Carol O’Hara is thanked for overseeing the planning and recording of this large site; her meticulous work ensured the accurate recording of the villa and its features. Janet Bartlet drew hundreds of finds for the archive and for publication, Mark Corney, Mike Ladle and Gareth Naylor prepared drawings for publication, their contributions have greatly enhanced the text.

Peter Stanier thanks Roger Taylor, consultant geologist, Royal Albert Memorial Museum, Exeter, Tom Cadbury, Royal Albert Memorial Museum Store, Exeter and John Valentin, AC archaeology. Thanks to Jo Crane and Sue Newman for aerial photography and to Robert Heaton for drone photography. Detailed drawings of the mosaics were provided by Luigi Thompson; otherwise all photographs are from the Druce Farm Roman Villa archive.

Grateful thanks are extended to Hannah Simpson and Dave Stewart (Bournemouth University) who undertook the geophysical surveys; their work underpinned much of the site excavation strategy. EDAS member Robert Heaton provided and drove his vintage JCB excavator from 2014;

Matthey Tagney is thanked for the French and German translations of the Summary. vii

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Finally, I am deeply indebted to Geoff Taylor for his thorough and meticulous copy-editing of the text, tables and figures and his wise counsel. The covid 19 pandemic provided an uninterrupted opportunity for writing up the excavation results. However, on the debit side, access to library facilities was severely curtailed. All text not attributed to individual contributors is by the author, who accepts full responsibility for any errors or omissions. Lilian Ladle

viii

Table of contents List of figures.................................................................................................................................................................... xxi List of tables................................................................................................................................................................... xxxi Preface........................................................................................................................................................................... xxxiii Summary........................................................................................................................................................................ xxxv Résumé......................................................................................................................................................................... xxxvii Zusammenfassung....................................................................................................................................................... xxxix 1. Environment and project background...................................................................................................................... 1 Location......................................................................................................................................................................... 1 Landscape and geology – Andrew Morgan................................................................................................................... 1 The site.......................................................................................................................................................................... 3 Project background........................................................................................................................................................ 3 Geophysical surveys................................................................................................................................................. 4 The excavations ....................................................................................................................................................... 4 Archaeological methodology . ................................................................................................................................. 5 Local community outreach ...................................................................................................................................... 6 Funding..................................................................................................................................................................... 6 Project aims.............................................................................................................................................................. 7 Summary of Periods...................................................................................................................................................... 7 Period 1: Prehistoric up to 43 CE (Fig. 1.7)............................................................................................................. 7 Period 2.1: 43-70 CE (Fig. 1.8)................................................................................................................................ 7 Period 2.2: 70-120 CE (Fig. 1.9).............................................................................................................................. 8 Period 3.1: 120-200 CE (Fig. 1.10).......................................................................................................................... 8 Period 3.2: 200-250 CE (Fig.1.11)........................................................................................................................... 8 Period 4.1: 250-300 CE (Fig. 1.12).......................................................................................................................... 8 Period 4.2: 300-350/70 CE (Fig. 1.13)..................................................................................................................... 8 Period 5: 350/70-430 CE.......................................................................................................................................... 9 Period 6: 430-650 CE (Fig. 1.14)............................................................................................................................. 9 Period 7: 650+ CE.................................................................................................................................................... 9 Radiocarbon dating........................................................................................................................................................ 9 Introduction............................................................................................................................................................... 9 Sampling strategy................................................................................................................................................... 10 Results and Calibration........................................................................................................................................... 10 Deposition and further work........................................................................................................................................ 10 2. The Prehistoric background..................................................................................................................................... 13 Early Neolithic activity................................................................................................................................................ 13 The features............................................................................................................................................................ 13 Pre-Roman ditches....................................................................................................................................................... 14 Ditch 878................................................................................................................................................................ 14 Ditch 1291.............................................................................................................................................................. 14 Ditch 1325.............................................................................................................................................................. 14 Ditch 1752.............................................................................................................................................................. 15 Flint and chert – Katherine Walker.............................................................................................................................. 15 Introduction............................................................................................................................................................. 15 Methodology........................................................................................................................................................... 16 Raw material........................................................................................................................................................... 16 Description of the total assemblage........................................................................................................................ 17 Cores and core reduction techniques...................................................................................................................... 17

ix

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Flakes and blades.................................................................................................................................................... 17 Chips and shatter..................................................................................................................................................... 17 Tools....................................................................................................................................................................... 17 Distribution of the assemblage across the site........................................................................................................ 21 Discussion............................................................................................................................................................... 21 Illustrated flint and chert artefacts.......................................................................................................................... 22 Prehistoric pottery....................................................................................................................................................... 22 Fabric Code............................................................................................................................................................. 22 Discussion............................................................................................................................................................... 22 Conclusion................................................................................................................................................................... 23 3. Enclosures, ditches and middens.............................................................................................................................. 25 The Early Roman enclosure system............................................................................................................................ 25 The outer enclosure (ditches 1604, 309, 313, 311 and 312) .................................................................................. 25 The inner enclosure (ditches 308, 314 and 1605)................................................................................................... 29 Southern outliers (Ditches 1504 and 310).............................................................................................................. 32 Structures associated with a south entrance to the villa......................................................................................... 32 Conclusions................................................................................................................................................................. 32 Other ditch systems .................................................................................................................................................... 33 Ditches 1641/1644 ................................................................................................................................................. 33 Ditches 1459/1510.................................................................................................................................................. 35 Gullies 372 and 384................................................................................................................................................ 37 Ditches 1260/1305.................................................................................................................................................. 38 Eastern enclosure......................................................................................................................................................... 39 Ditch 1262 ............................................................................................................................................................. 39 Ditch 1412.............................................................................................................................................................. 39 Ditch 1410.............................................................................................................................................................. 40 Ditch 1411............................................................................................................................................................... 40 Conclusions ................................................................................................................................................................ 40 The midden deposits.................................................................................................................................................... 40 Midden 315 T2 ...................................................................................................................................................... 40 Midden 1529 T24, T41 and T46 (Fig. 3.15/1)........................................................................................................ 40 Midden 1355 T59 .................................................................................................................................................. 42 Midden 1528 T14 and T52 (Fig. 3.17) .................................................................................................................. 43 Conclusion................................................................................................................................................................... 48 4. Early Roman activity outside the villa complex..................................................................................................... 49 Early Roman activity in Lower Limepits field............................................................................................................ 49 Pit 1399 T14 (Fig. 4.1/1) ....................................................................................................................................... 49 Pit 643 T21 (Fig. 4.1/2).......................................................................................................................................... 49 Activity pre-dating the East Range......................................................................................................................... 49 Early Roman activity in Eleven Acres......................................................................................................................... 50 The Ancillary Building .......................................................................................................................................... 50 Room AB1 (pits 2015, 1996, 2008, 1991, 2032, 2008 and 1994) . ............................................................................ 53 Pit 2015 (Fig 4.7/1) . .............................................................................................................................................. 53 Pit 1996 (Fig. 4.7/2)................................................................................................................................................ 53 Pit 1991 (Fig. 4.7/3)................................................................................................................................................ 53 Pit 2031 (Fig. 4.7/4)................................................................................................................................................ 53 Pit 2032 (Fig. 4.7/7)................................................................................................................................................ 53 Room AB2 (pits 2000, 2017 and 2036) . .................................................................................................................... 53 Pit 2000 (Fig, 4.7/8) . ............................................................................................................................................. 53 Pit 2017 (Fig. 4.7/10).............................................................................................................................................. 54 Room AB3 (pits 2295/2020/2038, 2050 and 2052) ................................................................................................... 54 Pit 2295 cut by pits 2020 and 2038 (Fig. 4.7/11)................................................................................................... 54 Pit 1735........................................................................................................................................................................ 55 Occupation Area T82................................................................................................................................................... 55 Pit 1800 (Figs. 4.9/2 and 4.12). ............................................................................................................................. 56 x

Table of contents Deposit 1795 (Fig. 4.10) . ...................................................................................................................................... 56 Pit 1845 (Fig. 4.9/3) . ............................................................................................................................................. 56 5. The North Range....................................................................................................................................................... 59 Phasing........................................................................................................................................................................ 59 Phase 1: Early Roman, Period 2.2 (c. 70-120 CE) (Fig. 5.4 Phase1).......................................................................... 61 Room N10............................................................................................................................................................... 61 Pit 643 (Fig. 4.1/2) .............................................................................................................................................. 63 Pit 644 (Figs 5.10/3 and 5.11).............................................................................................................................. 63 Conclusions................................................................................................................................................................. 65 Phase 2: Middle Roman, Period 3.1 (c. 120-200 CE) (Fig. 5.4 Phase 2).................................................................... 65 Room N12............................................................................................................................................................... 65 Room N1................................................................................................................................................................. 65 Conclusions................................................................................................................................................................. 65 Phase 3: Middle Roman Period 3.2 (c. 200-250 CE) (Fig. 5.4 Phase 3)..................................................................... 65 The west extension (Fig. 5.13) .............................................................................................................................. 65 Room N2................................................................................................................................................................. 67 Room N5................................................................................................................................................................. 68 Extension of the original building (Figs. 5.4/3 and 5.17)....................................................................................... 69 Room N3 (Fig. 5.12)............................................................................................................................................... 71 Room N6................................................................................................................................................................. 72 Room N8................................................................................................................................................................. 72 Room N7................................................................................................................................................................. 73 Room N10............................................................................................................................................................... 73 Corridor N11........................................................................................................................................................... 75 Conclusions................................................................................................................................................................. 75 Phase 4: Late Roman Period 4.2 (c. 300-350/70 CE) (Fig. 5.4 Phase 4).................................................................... 77 The west wing......................................................................................................................................................... 77 Room N1................................................................................................................................................................. 77 Room N2................................................................................................................................................................. 78 Room N4................................................................................................................................................................. 78 Paved area and potential building........................................................................................................................... 79 Room N5................................................................................................................................................................. 79 The north wing........................................................................................................................................................ 79 Room N3................................................................................................................................................................. 79 Room N10............................................................................................................................................................... 84 Room N9................................................................................................................................................................. 84 Pit 798................................................................................................................................................................... 85 Corridor N11........................................................................................................................................................... 85 Gully 407................................................................................................................................................................ 85 Conclusions................................................................................................................................................................. 87 Phase 5: Final Roman Period 5 (c. 350/370-430 CE)................................................................................................. 87 Room N1................................................................................................................................................................. 87 Room N3................................................................................................................................................................. 87 Room N7................................................................................................................................................................. 88 Room N10 (plan Fig. 5.42)..................................................................................................................................... 89 Room N9 (Fig. 5.37/1)............................................................................................................................................ 89 Room N11 . ............................................................................................................................................................ 92 Conclusions................................................................................................................................................................. 93 Phase 6: Post-Roman Period 6 (c. 430-650 CE) (Fig. 5.4 Phase 5)............................................................................ 93 Activity south of Room N3..................................................................................................................................... 94 Room N1................................................................................................................................................................. 94 Room N3................................................................................................................................................................. 95 Room N9................................................................................................................................................................. 95 Corridor N11 . ........................................................................................................................................................ 98 Collapse south of the North Range......................................................................................................................... 98

xi

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Conclusions................................................................................................................................................................. 98 General conclusions................................................................................................................................................... 100 6. The East Range........................................................................................................................................................ 101 Phase 1: Early Roman, Period 2.2 (c. 70-120 CE).................................................................................................... 101 The first rectangular building................................................................................................................................ 101 Features relating to this phase............................................................................................................................... 101 Conclusion................................................................................................................................................................. 104 Phase 2: Middle Roman, Period 3.1 (c. 120-200 CE) (Fig. 6.8) .............................................................................. 104 Extension of the rectangular building and associated features ............................................................................ 104 Tank 778 and associated features.......................................................................................................................... 104 Pit 381 (Figs 6.11/2 and 6.12)............................................................................................................................... 105 Pit 1149 /1181 (Figs 6.13 and 6.14/1) . ................................................................................................................ 106 Pit 896................................................................................................................................................................... 107 Pit 209 (Fig. 6.14/2) . ........................................................................................................................................... 107 Phase 3a: Late Roman, Period 4.1 (c. 300 CE)......................................................................................................... 107 The Aisled Hall..................................................................................................................................................... 107 Flint-packed pit 638 (Fig. 6.19/1)......................................................................................................................... 108 Flint-packed pit 648 (Fig. 6.19/2 and 6.20).......................................................................................................... 110 Deposit 1269 (Fig. 6.21)....................................................................................................................................... 110 Pit 1077 (Figs 6.19/3 and 6.22) ........................................................................................................................... 110 Phase 3b: Late Roman, Period 4.1 (c. 350 CE)......................................................................................................... 110 Fire damage and site preparation for re-building................................................................................................. 110 Phase 4: Late Roman, Period 4.2 (c. 350/70 CE) (Fig. 6.23) ................................................................................... 110 Extension of the Aisled Hall................................................................................................................................. 110 The flint-packed pits............................................................................................................................................. 113 Internal walls........................................................................................................................................................ 113 Floor surfaces........................................................................................................................................................ 114 Deposit 556 (Fig. 6.24)......................................................................................................................................... 114 Oven 207 (Figs 6.23, 6.25-28). . ......................................................................................................................... 115 Structure 1203 (Fig. 6.29)..................................................................................................................................... 115 Pit 823 (Fig. 6.30)................................................................................................................................................. 116 Phase 5: Final Roman and Post-Roman, Periods 5 and 6 (c. 430-650+ CE)............................................................ 116 Collapse of the building . ..................................................................................................................................... 116 Material over the collapsed building ................................................................................................................... 116 Collapse sequence................................................................................................................................................. 118 The roof................................................................................................................................................................ 118 Collapse of the north wall 559 (Fig. 6.31) (Period 6 c. 430-650 CE).................................................................. 120 Phase 6: Early Medieval, Period 6 (c. 430-650 CE).................................................................................................. 122 Post-collapse activity............................................................................................................................................ 122 Pit 209 (Figs. 6.14/2 and 6.35) ............................................................................................................................ 123 Pit 825 (Figs 6.36 and 6.37/1: the pit is located on section line AB on Fig. 6.26) .............................................. 125 Pit 752 (Figs 6.36 and 6.37/1) ............................................................................................................................. 126 Pit 944 (Fig. 6.37/2) . ........................................................................................................................................... 126 Pit 962 (Figs 6.37/1 and 2: section lines AB and CD shown on Fig 6.26) .......................................................... 126 Conclusions............................................................................................................................................................... 128 General conclusions................................................................................................................................................... 128 7. The West Range....................................................................................................................................................... 129 Phase 1: Middle Roman Period 3.1 (c. 120-200 CE) (Fig. 7.3/1)............................................................................. 129 Phase 2: Period 4.1 Late Roman (c. 250-300 CE) (Figs. 7.1 and 7.5) ..................................................................... 130 Room W1.............................................................................................................................................................. 131 Room W2 . ........................................................................................................................................................... 135 Room W3.............................................................................................................................................................. 135 Room W4.............................................................................................................................................................. 138 Conclusions............................................................................................................................................................... 138

xii

Table of contents 8. The grain dryer and associated burial................................................................................................................... 139 The grain dryer.......................................................................................................................................................... 139 The flue (Fig. 8.3)................................................................................................................................................. 139 The flue walls . ..................................................................................................................................................... 139 Section 1850 (A) (Fig. 8.5/1) . ............................................................................................................................. 140 Section 2075 (B) (Fig. 8.5/2)................................................................................................................................ 141 Section 1963 (C) (Fig. 8.5/3)................................................................................................................................ 141 Section 1969 (D) (Fig. 8.5/4)................................................................................................................................ 141 The stoke pit and burial............................................................................................................................................. 143 The stoke pit 1964 (Figs 8.1, 8.2 and 8.5/5)......................................................................................................... 143 The skeleton (Figs 8.1, 8.2/1, 8.5/5 and 8.7)........................................................................................................ 143 Conclusions............................................................................................................................................................... 143 9. Building Materials................................................................................................................................................... 145 Wall construction....................................................................................................................................................... 145 Introduction........................................................................................................................................................... 145 Non-flint internal walls......................................................................................................................................... 145 Mortar used in the construction of the building.................................................................................................... 145 Roofing and ceramic building materials – John Bithell............................................................................................ 147 Introduction........................................................................................................................................................... 147 Methodology . ...................................................................................................................................................... 147 Ceramic building material.................................................................................................................................... 147 Stone roof tiles – John Bithell.............................................................................................................................. 156 Roofing slates – Peter Stanier.............................................................................................................................. 158 Stone building material and architectural fragments – Peter S. Bellamy.................................................................. 160 Introduction........................................................................................................................................................... 160 Raw material......................................................................................................................................................... 161 Column fragments................................................................................................................................................ 161 Jambs?................................................................................................................................................................... 161 Post pad................................................................................................................................................................. 161 Miscellaneous architectural fragments................................................................................................................. 162 Illustrated architectural fragments and building stone.......................................................................................... 163 Note on the column capital from context 197 – †Anthony Beeson........................................................................... 163 Other occurrences of building stone . ....................................................................................................................... 164 Conclusion................................................................................................................................................................. 164 10. Internal decoration.................................................................................................................................................. 165 Mosaics – †Anthony Beeson...................................................................................................................................... 165 Introduction........................................................................................................................................................... 165 Room N1: Mosaic 1 (Context 153) ..................................................................................................................... 165 Room N2: Mosaic context 1059 (below Mosaic 2).............................................................................................. 168 Room N2: Mosaic 2 (Context 246)...................................................................................................................... 169 Other mosaics of the North Range (see Fig. 5.2 for room locations)........................................................................ 171 Room N3: Mosaic context 210 ............................................................................................................................ 171 South of Room N3 ............................................................................................................................................... 171 Room N4: Mosaic context 98............................................................................................................................... 171 Room N5............................................................................................................................................................... 171 Room N6: Mosaic 3 (Context 357)...................................................................................................................... 171 Room N7: Mosaic 4 (Context 329)...................................................................................................................... 172 Room N10 . .......................................................................................................................................................... 172 Room N11: Mosaic context 99............................................................................................................................. 176 The East Range.......................................................................................................................................................... 176 The West Range......................................................................................................................................................... 176 Glass tessera.............................................................................................................................................................. 176 Discussion.................................................................................................................................................................. 177

xiii

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Painted wall plaster – Andrew Morgan...................................................................................................................... 177 Introduction........................................................................................................................................................... 177 The assemblage..................................................................................................................................................... 177 Methodology......................................................................................................................................................... 178 Room N1............................................................................................................................................................... 179 Room N12............................................................................................................................................................. 188 Room N2............................................................................................................................................................... 188 Room N4............................................................................................................................................................... 188 Room N5 . ............................................................................................................................................................ 189 Room N3............................................................................................................................................................... 189 Room N6............................................................................................................................................................... 189 Room N7............................................................................................................................................................... 189 Room N9............................................................................................................................................................... 189 Room N10............................................................................................................................................................. 189 Room N11............................................................................................................................................................. 190 Room N3 Infill...................................................................................................................................................... 190 Midden - South of Room N3................................................................................................................................ 191 Midden - East of Room N4................................................................................................................................... 191 East Range /Aisled Hall........................................................................................................................................ 191 Middens and Other Pits........................................................................................................................................ 191 Discussion............................................................................................................................................................. 192 11. Roman and post-Roman pottery............................................................................................................................ 197 Structure of the report................................................................................................................................................ 197 Imported finewares.................................................................................................................................................... 197 Samian – Mark Corney and Peter Teekamp......................................................................................................... 197 Central Gaulish Terra Nigra (F11) CNG TN........................................................................................................ 200 Lyon Colour-coated ware (F18) LYO CC............................................................................................................ 200 Argonne Colour-coated ware (Sinzig Roughcast) (F14) ARG CC....................................................................... 200 Moselkeramic Black-slipped ware (F3) MOS BS................................................................................................ 200 Central Gaulish Black-slipped ware (F2) CNG BS.............................................................................................. 200 Overview of amphorae – Mark Corney..................................................................................................................... 200 Mortaria..................................................................................................................................................................... 201 Early Roman (Period 2 c. 43-120 CE).................................................................................................................. 201 Middle Roman (Period 3 c. 120-250 CE)............................................................................................................. 202 Late Roman (Period 4 c. 250-350/70 CE)............................................................................................................ 202 Late Roman (Periods 4-5 c. 250-400 CE)............................................................................................................ 202 Final Roman (Period 5 c. 350/70-420 CE)........................................................................................................... 202 Discussion............................................................................................................................................................. 202 Catalogue.............................................................................................................................................................. 202 British finewares........................................................................................................................................................ 206 Corfe Mullen ware (F10)...................................................................................................................................... 206 New Forest finewares ......................................................................................................................................... 206 Oxfordshire wares................................................................................................................................................. 206 Lower Nene Valley Colour-coated ware . ............................................................................................................ 207 Unsourced finewares . .......................................................................................................................................... 207 Coarsewares............................................................................................................................................................... 208 (South East) Dorset Black Burnished ware (DOR BB1)...................................................................................... 208 South East Dorset Orange Wiped ware (SEDOWW)........................................................................................... 210 South Western Black Burnished ware (SOW BB1).............................................................................................. 210 New Forest greyware............................................................................................................................................ 210 Alice Holt Reduced ware (ALH RE) . ................................................................................................................. 210 Hampshire Grog-tempered ware (HAM GT) ...................................................................................................... 210 Other coarsewares identified to kiln .................................................................................................................... 210 Unsourced Coarsewares ...................................................................................................................................... 211 Sherds bearing graffiti – Roger Tomlin...................................................................................................................... 211

xiv

Table of contents Post-Roman pottery................................................................................................................................................... 212 Medieval pottery........................................................................................................................................................ 213 12. Pottery assemblages – Malcolm Lyne...................................................................................................................... 215 Introduction............................................................................................................................................................... 215 Methodology ............................................................................................................................................................ 215 The Assemblages....................................................................................................................................................... 215 Period 2: c. 43-120 CE.......................................................................................................................................... 215 Periods 3 and 4: c. 120-350/70 CE....................................................................................................................... 220 Period 4.2: c. 300-350/70 CE............................................................................................................................... 223 Period 5: c. 350/70-430+ CE................................................................................................................................ 225 Periods 6 and 7: c. 450/500-1000 CE................................................................................................................... 228 Discussion.................................................................................................................................................................. 229 13. Portable objects of stone, shale, glass, bone and fired clay.................................................................................. 231 Stone Objects – Peter S. Bellamy.............................................................................................................................. 231 Introduction........................................................................................................................................................... 231 Raw material......................................................................................................................................................... 231 Querns and millstones........................................................................................................................................... 231 Mortars and pestle................................................................................................................................................. 233 Grinding stones . .................................................................................................................................................. 233 Sharpeners and polishers...................................................................................................................................... 233 Counters................................................................................................................................................................ 234 Miscellaneous stone objects................................................................................................................................. 234 Kimmeridge Shale – Sue Cullinane.......................................................................................................................... 235 Introduction........................................................................................................................................................... 235 Methodology . ...................................................................................................................................................... 235 Source of material................................................................................................................................................. 235 Results................................................................................................................................................................... 235 Discussion............................................................................................................................................................. 238 Catalogue.............................................................................................................................................................. 238 Roman glass – Denise Allen...................................................................................................................................... 239 1st/2nd century vessels (Periods 2.1-3.1)............................................................................................................. 239 Colourless cups and bowls of the 2nd to 3rd centuries (Periods 3.1-4.1)............................................................ 240 Late Roman vessels (Periods 4.1-5)..................................................................................................................... 240 Catalogue.............................................................................................................................................................. 241 Bone objects ............................................................................................................................................................. 248 Personal adornment.............................................................................................................................................. 248 Writing and communication ................................................................................................................................ 249 A note on the bone folding scalprum – †Anthony Beeson......................................................................................... 250 Gaming equipment................................................................................................................................................ 251 Domestic utensils.................................................................................................................................................. 251 Tool....................................................................................................................................................................... 251 Tools from the animal bone assemblage – Clare Randall......................................................................................... 251 Fired clay objects....................................................................................................................................................... 252 Methodology......................................................................................................................................................... 252 Conclusions........................................................................................................................................................... 256 14. Non-ferrous metal finds........................................................................................................................................... 257 Introduction............................................................................................................................................................... 257 Coins – Malcolm Lyne............................................................................................................................................... 257 Coin catalogue...................................................................................................................................................... 258 Medieval coins (11) ........................................................................................................................................... 259 Comments on the Roman coin assemblage – Mark Corney...................................................................................... 260 Copper alloy objects.................................................................................................................................................. 260 Personal adornment.............................................................................................................................................. 261

xv

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Bracelets .............................................................................................................................................................. 264 Finger rings........................................................................................................................................................... 266 Ear ring................................................................................................................................................................. 268 Necklaces/bracelets............................................................................................................................................... 268 Pins....................................................................................................................................................................... 268 Toilet implements................................................................................................................................................. 269 Household utensils................................................................................................................................................ 270 Sewing needles..................................................................................................................................................... 271 Objects associated with written communication.................................................................................................. 272 Fasteners and fittings............................................................................................................................................ 272 Items with a military association ......................................................................................................................... 274 Note on the Late Roman military belt fitting – Mark Corney.............................................................................. 274 Miscellaneous and other unidentified fragments.................................................................................................. 275 Post-Roman and Early Medieval objects . ........................................................................................................... 276 Lead objects . ............................................................................................................................................................ 276 Pot mends ............................................................................................................................................................ 276 Weights................................................................................................................................................................. 278 Counters................................................................................................................................................................ 279 Household utensils................................................................................................................................................ 279 Miscellaneous....................................................................................................................................................... 280 Offcuts and waste.................................................................................................................................................. 281 Post-Medieval lead............................................................................................................................................... 282 15. Iron objects and metalworking debris................................................................................................................... 283 Iron objects................................................................................................................................................................ 283 Metalworking tools . ............................................................................................................................................ 283 Woodworking tools............................................................................................................................................... 284 Possible modelling tool......................................................................................................................................... 284 Leather working tools........................................................................................................................................... 284 Transport............................................................................................................................................................... 285 Keys...................................................................................................................................................................... 287 Domestic equipment............................................................................................................................................. 287 Knives................................................................................................................................................................... 288 Structural fittings.................................................................................................................................................. 289 Nails...................................................................................................................................................................... 292 Miscellaneous items.............................................................................................................................................. 294 Assessment and analysis of the metallurgical debris – Derek Pitman and Stephen Knowles................................... 296 Introduction........................................................................................................................................................... 296 Methods and key terms......................................................................................................................................... 296 Results................................................................................................................................................................... 297 X-Ray Fluorescence analysis................................................................................................................................ 299 Discussion............................................................................................................................................................. 299 Conclusion................................................................................................................................................................. 301 16. Human remains – Clare Randall............................................................................................................................. 303 Perinatal human remains........................................................................................................................................... 303 Introduction........................................................................................................................................................... 303 Preservation and taphonomy................................................................................................................................. 303 Results ................................................................................................................................................................. 303 Anthropological data ........................................................................................................................................... 303 Pathology and non-metric skeletal traits............................................................................................................... 303 Discussion . .......................................................................................................................................................... 303 Adult human remains ............................................................................................................................................... 305 Introduction........................................................................................................................................................... 305 Anthropological data: sex, age, and stature ......................................................................................................... 305 Pathology.............................................................................................................................................................. 306 Discussion............................................................................................................................................................. 307 xvi

Table of contents Preface to Chapters 17 and 18 (Faunal and environmental remains)........................................................................ 309 Animal bone.............................................................................................................................................................. 309 Charred plant macrofossils........................................................................................................................................ 309 Charcoal..................................................................................................................................................................... 309 Radiocarbon dating programme................................................................................................................................ 310 17. Faunal remains.........................................................................................................................................................311 Animal bones – Clare Randall.................................................................................................................................. 311 Introduction ......................................................................................................................................................... 311 Recording methods............................................................................................................................................... 311 The hand-collected assemblage ........................................................................................................................... 311 Preservation and taphonomy in the main hand-collected assemblage.................................................................. 316 Period 1: Prehistoric up to c. 43 CE..................................................................................................................... 318 Period 2.1 Early Roman c. 43-70 CE................................................................................................................... 318 Period 2.2: Early Roman c. 70-120 CE................................................................................................................ 324 Period 3: Middle Roman c. 120-250 CE.............................................................................................................. 324 Period 3.1: Middle Roman c. 120-200 CE........................................................................................................... 324 Period 3.2: Middle Roman c. 200-250 CE........................................................................................................... 325 Period 4: Late Roman c. 250-350/70 CE.............................................................................................................. 325 Period 4.1: Late Roman c. 250-300 CE................................................................................................................ 325 Period 4.2: Late Roman c. 300-350/70 CE........................................................................................................... 326 Period 5: Final Roman c. 350/70-430 CE............................................................................................................. 326 Period 5/6: Final Roman/Post-Roman c. 350/70-650 CE..................................................................................... 326 Period 6: Post-Roman c. 430-650 CE................................................................................................................... 327 Cattle..................................................................................................................................................................... 327 Sheep/goat ........................................................................................................................................................... 330 Pig......................................................................................................................................................................... 335 Horse..................................................................................................................................................................... 336 Dog....................................................................................................................................................................... 337 Cat......................................................................................................................................................................... 337 Wild mammals...................................................................................................................................................... 337 Butchery and carcase processing and disposal..................................................................................................... 338 The Associated Bone Groups and skulls.............................................................................................................. 338 The small vertebrate assemblages........................................................................................................................ 341 The bulk sample assemblages............................................................................................................................... 343 Summary of the small vertebrates........................................................................................................................ 344 Fish – Clare Randall.................................................................................................................................................. 350 Discussion – Clare Randall....................................................................................................................................... 351 Molluscs – Heather Tidball....................................................................................................................................... 356 Land snails............................................................................................................................................................ 356 Marine molluscs.................................................................................................................................................... 356 Conclusion............................................................................................................................................................ 358 18. Environmental remains........................................................................................................................................... 359 Wood charcoal – Dana Challinor.............................................................................................................................. 359 Introduction........................................................................................................................................................... 359 Methodology......................................................................................................................................................... 359 Results................................................................................................................................................................... 359 Discussion............................................................................................................................................................. 365 Charred and mineralised plant remains – Wendy Carruthers.................................................................................... 368 Introduction........................................................................................................................................................... 368 Soil processing methodology and selection of samples for analysis.................................................................... 368 Results................................................................................................................................................................... 368 Description of samples by period and feature...................................................................................................... 376 Discussion............................................................................................................................................................. 388 The integrity of the data . ..................................................................................................................................... 388

xvii

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Arable crops . ....................................................................................................................................................... 389 Evidence of crop husbandry recovered from weed taxa . .................................................................................... 391 The distribution of cereal, hay and heathland remains around the villa............................................................... 391 Other potential crops, orchard fruits and garden plants . ..................................................................................... 393 Imported foods...................................................................................................................................................... 397 Grassland and heathland resources....................................................................................................................... 397 The Roman plant-based diet at Druce compared to other villas . ........................................................................ 397 Summary by period ............................................................................................................................................. 399 Conclusions........................................................................................................................................................... 401 19. Interpretation and discussion................................................................................................................................. 403 Introduction............................................................................................................................................................... 403 Period 1: Prehistoric up to c. 43 CE (Fig. 1.7) ......................................................................................................... 403 Early Neolithic...................................................................................................................................................... 403 Bronze Age........................................................................................................................................................... 403 Field systems........................................................................................................................................................ 403 Druce Farm Roman Villa........................................................................................................................................... 404 Introduction........................................................................................................................................................... 404 Period 2.1: Early Roman c. 43-70 CE (Fig. 1.8)....................................................................................................... 405 Chronology........................................................................................................................................................... 405 Settlement and land use........................................................................................................................................ 405 Society.................................................................................................................................................................. 408 Environment and economy................................................................................................................................... 409 Period 2.2: Early Roman c. 70-120 CE (Fig. 1.8)..................................................................................................... 409 Chronology........................................................................................................................................................... 409 Settlement and land use........................................................................................................................................ 410 Society.................................................................................................................................................................. 411 Environment and economy................................................................................................................................... 411 Period 3.1: Middle Roman c. 120-200 CE (Fig. 1.10).............................................................................................. 412 Chronology........................................................................................................................................................... 412 Settlement and land use........................................................................................................................................ 413 Society.................................................................................................................................................................. 413 Environment and economy................................................................................................................................... 415 Period 3.2 Middle Roman c. 200-250 CE (Fig. 1.11)................................................................................................ 416 Chronology........................................................................................................................................................... 416 Settlement and land use........................................................................................................................................ 416 Society.................................................................................................................................................................. 416 Environment and economy................................................................................................................................... 418 Period 4.1: Late Roman c. 250-300 CE (Fig. 1.11)................................................................................................... 418 Chronology........................................................................................................................................................... 418 Settlement and land use........................................................................................................................................ 419 Society.................................................................................................................................................................. 419 Environment and economy................................................................................................................................... 420 Period 4.2: Late Roman c. 300-350/70 CE (Fig. 1.13).............................................................................................. 421 Chronology........................................................................................................................................................... 421 Settlement and land use........................................................................................................................................ 421 Society.................................................................................................................................................................. 422 Environment and economy................................................................................................................................... 423 Period 5: Final Roman c. 350/70-430 CE ................................................................................................................ 424 Chronology........................................................................................................................................................... 424 Settlement and land use........................................................................................................................................ 424 Society.................................................................................................................................................................. 424 Environment and economy................................................................................................................................... 426 Period 6: Post-Roman c. 430-650 CE ...................................................................................................................... 427 Chronology........................................................................................................................................................... 427 Settlement and land use........................................................................................................................................ 427

xviii

Table of contents Society.................................................................................................................................................................. 427 Collapse................................................................................................................................................................ 428 Post-collapse pit digging....................................................................................................................................... 429 Environment and economy................................................................................................................................... 429 Period 7: Early medieval c. 650+ CE ....................................................................................................................... 429 Chronology........................................................................................................................................................... 429 Settlement, land use, environment and economy................................................................................................. 429 Final Conclusions ..................................................................................................................................................... 429 Bibliography.................................................................................................................................................................... 431

xix

List of figures Fig. 1.1. Key to all archaeological drawings........................................................................................................................ 1 Fig. 1.2. Site location in southern England and Dorset........................................................................................................ 2 Fig. 1.3. Aerial view of the villa site..................................................................................................................................... 3 Fig. 1.4. Contour plot and villa location............................................................................................................................... 4 Fig. 1.5. Geophysical surveys............................................................................................................................................... 5 Fig. 1.6. Location of archaeological trenches T1-T90.......................................................................................................... 6 Fig. 1.7. Phase plan Period 1: prehistoric up to 43 CE......................................................................................................... 7 Fig. 1.8. Phase plan Period 2.1: 43-70 CE............................................................................................................................ 7 Fig. 1.9. Phase plan Period 2.2: 70-120 CE.......................................................................................................................... 8 Fig. 1.10. Phase plan Period 3.1: 120-200 CE...................................................................................................................... 8 Fig. 1.11. Phase plan Period 3.2: 200-250 CE...................................................................................................................... 8 Fig. 1.12. Phase plan Period 4.1: 250-300 CE...................................................................................................................... 9 Fig. 1.13. Phase plan Period 4.2: 300-350/70 CE................................................................................................................. 9 Fig. 1.14. Period 6 distribution of post-Roman metalwork and pottery............................................................................... 9 Fig. 2.1. Neolithic features in Trench 88............................................................................................................................ 13 Fig. 2.2. Plan of Neolithic features in Trench 88................................................................................................................ 14 Fig. 2.3. Sections through ditches 878, 878/1291, 1325 and 1752 (1-4)............................................................................ 15 Fig. 2.4. Illustrated worked flint (1-10)............................................................................................................................... 18 Fig. 2.5. Illustrated worked flint (11-17)............................................................................................................................. 20 Fig. 2.6. Distribution of flint and chert from excavated contexts in Lower Limepits field................................................ 21 Fig. 2.7. Middle Bronze Age pottery (1-3)......................................................................................................................... 22 Fig. 3.1. Location of enclosures, ditches and middens and illustrated sections................................................................. 27 Fig. 3.2. Sections through enclosures 1604, 309, 313, 311 and 312 (1-5).......................................................................... 28 Fig. 3.3. Sections through enclosures 308, 314 and 1605 (1-3).......................................................................................... 30 Fig. 3.4. Ditch 1605 section 1620 during excavation......................................................................................................... 31 Fig. 3.5. Sections through southern ditches 1504 and 310 (1-2)........................................................................................ 33 Fig. 3.6. Ditch 1504 section 1513 with horse skull on ditch base...................................................................................... 34 Fig. 3.7. Foundation Pit 402 .............................................................................................................................................. 34 Fig. 3.8. Features associated with the villa’s southern entrance (T11)............................................................................... 35 Fig. 3.9. Plan of ditches 1641/1644 and pit 1735, and ditch sections................................................................................. 36 Fig. 3.10. Plan of southern site and sections through ditches 1459/1510........................................................................... 37 Fig. 3.11. Ditches 1452 and 1510 with horse skull on ditch base....................................................................................... 38 Fig. 3.12. Sections through gullies 372 and 384, ditches 1260/1305, 1262, 1410, 1411, 1412 (1-8)................................ 39 xxi

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Fig. 3.13. Ditch 1412 section 1425 with horse skull.......................................................................................................... 41 Fig. 3.14. Midden 315 with uneven base............................................................................................................................ 41 Fig. 3.15. Midden 1529 plan and sections, midden 1355 section (1-5).............................................................................. 42 Fig. 3.16. Midden 1529 with millstone and box flue tiles.................................................................................................. 43 Fig. 3.17. Midden 1528 initial clear back........................................................................................................................... 43 Fig. 3.18. Midden 1528 infill plans (layers 1-4)................................................................................................................. 46 Fig. 3.19. Midden 1528 context 1422, dump of fired clay, pottery and burnt stone........................................................... 47 Fig. 3.20. Midden 1528 tile deposit 1339........................................................................................................................... 47 Fig. 4.1. Sections through pits 1399 and 643 (1-2)............................................................................................................. 49 Fig. 4.2. Ancillary Building during excavation................................................................................................................... 50 Fig. 4.3. Ancillary Building aerial after excavation completed.......................................................................................... 51 Fig. 4.4. Ancillary Building plan of all excavated features................................................................................................ 51 Fig. 4.5. Ancillary Building Phases 1 and 2 plans.............................................................................................................. 52 Fig. 4.6. Ancillary Building Room AB3 with row of nails from floor planks.................................................................... 52 Fig. 4.7. Sections of pits associated with the Ancillary Building (1-12)............................................................................ 54 Fig. 4.8. Pit 1735 excavated sections 1881, 1746, 1854..................................................................................................... 55 Fig. 4.9. Sections of pits 1735, 1800 and 1845 (1-4).......................................................................................................... 56 Fig. 4.10. Activity Area T82............................................................................................................................................... 57 Fig. 4.11. Activity Area T82 plan with excavated features . ............................................................................................. 57 Fig. 4.12. Pit 1800 half section........................................................................................................................................... 58 Fig. 5.1. Trench 2 revealing structural remains September 2012 ...................................................................................... 59 Fig. 5.2. Plan North Range with Room and wall numbers................................................................................................. 60 Fig. 5.3. Aerial of the excavated North Range September 2015......................................................................................... 60 Fig. 5.4. North Range Phase plans (1-5)............................................................................................................................. 61 Fig. 5.5. Room N3 wall 244................................................................................................................................................ 62 Fig. 5.6. Room N3 wall 244 with quoin 1144 and wall 1228............................................................................................. 62 Fig. 5.7. Room N3 Quoin 147 and walls 1048 and 1046.................................................................................................... 62 Fig. 5.8. Room N3 walls 991 and 667................................................................................................................................ 63 Fig. 5.9. Phase 1 structural elements under Room N10...................................................................................................... 64 Fig. 5.10. Sections of sondage 1194, posthole 952 and pit 644 under Room N10 (1-3) ................................................. 64 Fig. 5.11. Pit 644 half section............................................................................................................................................. 66 Fig. 5.12. Room N3 with all exposed walls . ..................................................................................................................... 66 Fig. 5.13. West extension of the North Range ................................................................................................................... 67 Fig. 5.14. Sections of sondage 1231 (Room N1), scoop 253 and fireplace 264 (Room N2) (1-4)..................................... 68 Fig. 5.15. Room N2 fireplace 264, view east...................................................................................................................... 69 Fig. 5.16. Room N5 oven 992 plan and section (1-2)......................................................................................................... 70 Fig. 5.17. East extension of the North Range .................................................................................................................... 70 Fig. 5.18. Room N7 mortar mixing basin 715.................................................................................................................... 71 xxii

List of figures Fig. 5.19. Room N7 section mortar mixing basin 715, plan fireplace 415 (1-2)................................................................ 72 Fig. 5.20. Phase 4 wall 442 dividing Rooms N3, N6 and N8............................................................................................. 73 Fig. 5.21. Room N6 mosaic 357......................................................................................................................................... 74 Fig. 5.22. Room N6 fireplace 415....................................................................................................................................... 74 Fig. 5.23. Room N7 mosaic 329......................................................................................................................................... 75 Fig. 5.24. Room N10 remnants of upper mosaic 474 and lower mosaic 615..................................................................... 76 Fig. 5.25. Room N10 remnants of lower mosaic 615......................................................................................................... 76 Fig. 5.26. Corridor N11....................................................................................................................................................... 77 Fig. 5.27. Joined tesserae (840) from corridor N11............................................................................................................ 77 Fig. 5.28. Room N2, Phases 3 and 4................................................................................................................................... 78 Fig. 5.29. Room N3 Phase 4 wall 420 with quoin 1227..................................................................................................... 80 Fig. 5.30. Junction of Phase 3 and 4 walls 338 and 420 between Rooms N3 and N6........................................................ 80 Fig. 5.31. Room N3 Phase 4 infill and collapse of chalk floor........................................................................................... 81 Fig. 5.32. Plan, Phases 4-5, Room N3................................................................................................................................ 81 Fig. 5.33. Room N3 infant burial 960................................................................................................................................. 82 Fig. 5.34. Room N3 infant burial 400................................................................................................................................. 82 Fig. 5.35. Room N3, burning against wall 422 and floor collapse..................................................................................... 83 Fig. 5.36. Phases 3 and 4 painted plaster from pit 614....................................................................................................... 84 Fig. 5.37. Room N9 plan, section pits 683, 685, 674 and 798 (1-5)................................................................................... 85 Fig. 5.38. Room N9 chalk floor and possible access to upper floor................................................................................... 86 Fig. 5.39. Room N9, pit 798............................................................................................................................................... 86 Fig. 5.40. Gully 407 behind the North Range..................................................................................................................... 87 Fig. 5.41. Room N3, deposit 577 of cattle skull................................................................................................................. 88 Fig. 5.42. Plan Phase 5 Room N10 .................................................................................................................................... 90 Fig. 5.43. Room N10, sections of pits 818 and 614 (1-2)................................................................................................... 90 Fig. 5.44. Room N10, pit 818 half section.......................................................................................................................... 91 Fig. 5.45. Room N10, deposit of broken roof tiles 595...................................................................................................... 91 Fig. 5.46. Room N10, pit 614 half section.......................................................................................................................... 92 Fig. 5.47. Room N9, sections of pits 590, 632, 633 and 680 (1-4)..................................................................................... 93 Fig. 5.48. Corridor N11, section of scoop 791 and wall 1073 (1-2)................................................................................... 94 Fig. 5.49. Room N3, Phase 6 steps cut into wall 244, infill and accumulation A............................................................... 95 Fig. 5.50. Room N1, collapsed limestone roof tiles in north-east corner........................................................................... 96 Fig. 5.51. Room N1, collapse debris over mosaic 153....................................................................................................... 96 Fig. 5.52. Room N3, collapsed wall and roof..................................................................................................................... 97 Fig. 5.53. Room N3, roof tile deposit................................................................................................................................. 97 Fig. 5.54. Room N9, roof collapse...................................................................................................................................... 98 Fig. 5.55. Corridor N11 and roof collapse over porch . ..................................................................................................... 99 Fig. 5.56. South of Room N3, east of Rooms N2 and N4, fallen wall 1073...................................................................... 99 xxiii

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Fig. 6.1. East Range building Phases 1-4.......................................................................................................................... 102 Fig. 6.2. Plan Phase 1 rectangular building; section through walls 1174 and 1202 (1-2)................................................ 103 Fig. 6.3. Posthole 1187 with limestone pad 1191............................................................................................................. 104 Fig. 6.4. Oven 1045.......................................................................................................................................................... 105 Fig. 6.5. Oven cluster 1045, 1091 and 1081 plan and section, plan oven 985 (1-3)........................................................ 106 Fig. 6.6. Pit 1182 section................................................................................................................................................... 106 Fig. 6.7. Pit 1182 as excavated.......................................................................................................................................... 107 Fig. 6.8. Plan Phase 2 rectangular building...................................................................................................................... 108 Fig. 6.9. Tank 778, fallen apex 827, flint pit 649 and wall 1210...................................................................................... 109 Fig. 6.10. Tank 778 half sectioned with fill of limestone roof tiles.................................................................................. 109 Fig. 6.11. Sections tank 778 and pit 381 (1-2).................................................................................................................. 110 Fig. 6.12. Pit 381 as excavated..........................................................................................................................................111 Fig. 6.13. Pits 1149/1181 during excavation......................................................................................................................111 Fig. 6.14. Pits 1149/1181 and 209 sections (1-2).............................................................................................................. 112 Fig. 6.15. Aisled Hall after excavation showing collapsed wall 559................................................................................ 113 Fig. 6.16. Plan Phase 3 Aisled Hall ................................................................................................................................. 114 Fig. 6.17. Re-used limestone quoins 538 and 539............................................................................................................ 114 Fig. 6.18. Fallen wall 559 and limestone plinth 158......................................................................................................... 115 Fig. 6.19. Flint pits 638, 648, 1113 and pit 1077 sections (1-4)....................................................................................... 116 Fig. 6.20. Flint pit 648 and Phases 1 and 4 chalk floors................................................................................................... 117 Fig. 6.21. Pottery deposit 1269......................................................................................................................................... 117 Fig. 6.22. Pit 1077 with nave hub in fill........................................................................................................................... 118 Fig. 6.23. Plan Phase 4 Aisled Hall................................................................................................................................... 118 Fig. 6.24. Pottery deposit 556........................................................................................................................................... 119 Fig. 6.25. North-west corner of the Aisled Hall, oven 207, plinth 158............................................................................. 119 Fig. 6.26. North-west corner of the Aisled Hall plan, and profile across features 432 to 1203 (1-2)............................... 120 Fig. 6.27. Oven 207 view into chamber............................................................................................................................ 121 Fig. 6.28. Oven 207 section ............................................................................................................................................. 121 Fig. 6.29. Structural remnant 1203 and pit 209................................................................................................................ 122 Fig. 6.30. Pit 823 with large shale slab............................................................................................................................. 123 Fig. 6.31. Aisled Hall wall and roof collapse.................................................................................................................... 124 Fig. 6.32. Reconstructed profile of Aisled Hall................................................................................................................ 125 Fig. 6.33. Slate and limestone roof tiles under fallen wall............................................................................................... 125 Fig. 6.34. Slate deposit 161 and postholes 773, 739 sections (1-3).................................................................................. 126 Fig. 6.35. Pit 209 half section........................................................................................................................................... 127 Fig. 6.36. Section through pits 896, 752 and 825............................................................................................................. 127 Fig. 6.37. Sections of pits 896, 962, 752, 825 and 962 and 944 (1-2).............................................................................. 128 Fig. 7.1. The West Range after excavation....................................................................................................................... 129 xxiv

List of figures Fig. 7.2. Room W1 with collapse debris........................................................................................................................... 130 Fig. 7.3. Phase 1 building and ditch 308 plan, pit 504 and ditch section 478 section (1-2) ........................................... 131 Fig. 7.4. Room W1, threshold 542 in wall 142................................................................................................................. 131 Fig. 7.5. Plan Phase 2........................................................................................................................................................ 132 Fig. 7.6. Room W1, complex 706 with oven 558, flue 780 and tile structure 1129......................................................... 132 Fig. 7.7. Room W1, oven complex 706 plan.................................................................................................................... 133 Fig. 7.8. Sections through oven complex (1-3)................................................................................................................. 134 Fig. 7.9. Sections through oven 219, Room W1, oven 1118 Room W2 (1-2).................................................................. 135 Fig. 7.10. Room W3, ovens 691 and 692, wall 171.......................................................................................................... 136 Fig. 7.11. Room W3, plan ovens 691 and 692.................................................................................................................. 137 Fig. 7.12. Sections through ovens 691 and 692 and section locations (1-6)..................................................................... 137 Fig. 7.13. Ovens 691 and 692 after excavation................................................................................................................. 138 Fig. 8.1. Dryer 1758, stoke pit 1964 and skeleton 1957................................................................................................... 139 Fig. 8.2. Dryer 1758 plan and locations of excavated sections (1-2)................................................................................ 140 Fig. 8.3. Cross flue of dryer.............................................................................................................................................. 141 Fig, 8.4. Flue wall profiles 1896, 1899 and 1895 (1-3).................................................................................................... 142 Fig. 8.5. Dryer sections 1850, 1963, 2075, 1969 and stoke pit 1964 (1-5)....................................................................... 142 Fig. 8.6. Stoke pit 1952 with cattle bone deposit.............................................................................................................. 143 Fig. 8.7. Skeleton 1957 on the base of stoke pit 1952...................................................................................................... 144 Fig. 9.1. Ancillary building terraced into the slope........................................................................................................... 146 Fig. 9.2. West Range, construction of wall 142................................................................................................................ 146 Fig. 9.3. Mortar from the inside of an imbrex................................................................................................................... 147 Fig. 9.4. Defining characteristics of tegulae..................................................................................................................... 148 Fig. 9.5. Imbrex with internal mortar wash....................................................................................................................... 148 Fig. 9.6. Lengths of upper cutaways ................................................................................................................................ 149 Fig. 9.7. Lengths of lower cutaways................................................................................................................................. 149 Fig. 9.8. Tegula with rounded end of lower cutaway, context 494, Room W1................................................................. 150 Fig. 9.9. Tegula pierced after firing, context 1155, Room W3.......................................................................................... 150 Fig. 9.10. Mammata fragment, context 1370, midden 1528............................................................................................. 151 Fig. 9.11. Signature types from Druce villa...................................................................................................................... 151 Fig. 9.12. Groups of Type E signature sizes..................................................................................................................... 152 Fig. 9.13. Score marks for location of a pre-fired nail hole, context 587, Room N3........................................................ 152 Fig. 9.14. Tile with ’ghost’ impressions of flint nodules, context 827.............................................................................. 152 Fig. 9.15. Evidence of hammer strikes to remove tile flange, context 945, Pit 944......................................................... 152 Fig. 9.16. Tegula with pre-fired score marks, context 180 outside Room N1.................................................................. 153 Fig. 9.17. Tegula with pre-fired inscribed lines from corridor N11.................................................................................. 153 Fig. 9.18. Complete imbrex from phase 4.2 infill, Room N3............................................................................................ 153 Fig. 9.19. Box flue tile with ‘former’ seam-impression, context 716, Aisled Hall........................................................... 154 xxv

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Fig. 9.20. Box flue tile with handling indentations, context 1298, ditch 1260................................................................. 154 Fig. 9.21. Box flue tile scored with a damaged comb, context 137, ditch 313................................................................. 154 Fig. 9.22. Range of Lydion thicknesses .......................................................................................................................... 155 Fig. 9.23. Lydion fragment with a chamfered edge, context 1946, Room AB2................................................................ 155 Fig. 9.24. Lydion fragment with large pre-fired hole, context 735, ditch 313.................................................................. 155 Fig. 9.25. Lydion fragment with cross grooves, context 1723, ditch 1641....................................................................... 155 Fig. 9.26. Lydion fragment with cross grooves and perforation, context 1690, ditch 1641............................................. 156 Fig. 9.27. Stone roof tile types 1-4.................................................................................................................................... 156 Fig. 9.28. Display of limestone roof tiles from Druce Villa............................................................................................. 157 Fig. 9.29. Leaf-shaped sandstone tile, context 417, south of Room N3........................................................................... 158 Fig. 9.30. Locations of possible slate sources................................................................................................................... 158 Fig. 9.31. Column, context 197, Room N1....................................................................................................................... 161 Fig. 9.32. Cable-decorated dressed stone, context 812, Room N11................................................................................. 161 Fig. 9.33. Stone architrave, context 197, Room N1.......................................................................................................... 162 Fig. 9.34. Limestone post pad, context 158, Aisled Hall.................................................................................................. 162 Fig. 9.35. Column and architrave in Room N1 collapse................................................................................................... 163 Fig 10.1. Room N1 mosaic 1 (153) and Room N2 mosaic 2 (246).................................................................................. 165 Fig.10.2. Room N1 mosaic 1 (153) (painting Luigi Thompson)...................................................................................... 166 Fig. 10.3. Room N2 mosaic 2 (246) (painting Luigi Thompson)..................................................................................... 169 Fig. 10.4. Room N7 mosaic 4 (329) (painting Luigi Thompson)..................................................................................... 170 Fig. 10.5. Room N10 mosaic 5 (474) (painting Luigi Thompson)................................................................................... 173 Fig. 10.6. Room N10 schematic floor layers.................................................................................................................... 174 Fig. 10.7. Tesserae from mosaic 615 in pit 614................................................................................................................ 174 Fig. 10.8. Tesserae in situ in pit 614................................................................................................................................. 175 Fig. 10.9. Tesserae in situ in pit 614................................................................................................................................. 175 Fig. 10.10. Painted plaster, examples of imitation marble designs (1-4).......................................................................... 178 Fig. 10.11. Painted plaster, examples of linear designs (1-5)........................................................................................... 179 Fig. 10.12. Painted plaster, examples of circles and geometric designs (1-6).................................................................. 180 Fig. 10.13. Painted plaster, examples of vegetal motif designs (1-5)............................................................................... 181 Fig. 10.14. Painted plaster, examples of designs using irregular shapes (1-4)................................................................. 182 Fig. 10.15. Painted plaster, examples of Phase 3 decorative designs (1-5)...................................................................... 183 Fig. 10.16. Schematic decorative scheme for Room N1, Phase 4.................................................................................... 186 Fig. 10.17. Comparison of the uses of colour between rooms.......................................................................................... 193 Fig. 10.18. Comparison of decorative types used in each room....................................................................................... 193 Fig. 10.19. Comparison of linear design types used in each room................................................................................... 194 Fig. 11.1. Samian potters stamps (1-2)............................................................................................................................. 199 Fig. 11.2. Mortaria (1-4)................................................................................................................................................... 203 Fig. 11.3. Mortaria with X graffito, context 1946............................................................................................................. 203 xxvi

List of figures Fig. 11.4. Comparison of New Forest and Oxford forms by sherd number..................................................................... 207 Fig. 11.5. Oxford F5 modified bowl, context 417............................................................................................................. 208 Fig. 11.6. Modified F5 Oxford bowl, context 559............................................................................................................ 208 Fig. 11.7. Black Burnished ware pottery of intrinsic interest (1-5).................................................................................. 209 Fig. 11.8. Other pottery of intrinsic interest (1-3)............................................................................................................. 210 Fig. 11.9. Sherds bearing graffiti (1-4).............................................................................................................................. 212 Fig. 12.1. Assemblage 1 (1-13)......................................................................................................................................... 217 Fig. 12.2. Assemblage 2 (14-29), Assemblage 3 (30), Assemblage 6 (31)....................................................................... 219 Fig. 12.3. Assemblage 7 (32-40), Assemblage 9 (41-42), Assemblage 10 (43-46).......................................................... 221 Fig. 12.4. Assemblage 11 (47), Assemblage 15 (48-49), Assemblage 17 (50-57), Assemblage 18 (58-63).................. 224 Fig. 12.5. Assemblage 19 (64-65), Assemblage 23 (66-67).............................................................................................. 228 Fig. 13.1. Millstones (1-3) and quern (4).......................................................................................................................... 232 Fig. 13.2. Mortar, pestle, whet and sharpening stones (5-12)........................................................................................... 233 Fig. 13.3. Stone counters and miscellaneous items (13-18)............................................................................................. 235 Fig. 13.4. Frequency of shale objects by century ............................................................................................................ 236 Fig. 13.5. Shale objects (1-10).......................................................................................................................................... 237 Fig. 13.6. Shale objects (11-16)........................................................................................................................................ 237 Fig. 13.7. Glass bowl and cups (1-19).............................................................................................................................. 242 Fig. 13.8. Glass beakers, bowl, jar, jug (20-25)................................................................................................................ 244 Fig. 13.9. Glass jugs, flasks and miscellaneous (26-46)................................................................................................... 245 Fig. 13.10. Glass beads and counter (47-52).................................................................................................................... 247 Fig. 13.11. Glass tessera from ditch 313.......................................................................................................................... 248 Fig. 13.12. Bone pins, pegs, stylus and counter (1-8)....................................................................................................... 249 Fig. 13.13. Knife with bone handle (9) SF 483 context 979, ditch 311............................................................................ 250 Fig. 13.14. Knife with bone handle (9) SF 483 context 979, ditch 311............................................................................ 250 Fig. 13.15. Bone spoon and point (10-11)........................................................................................................................ 251 Fig. 13.16. Fired clay bars, container and object (1-6)..................................................................................................... 253 Fig. 13.17. Fired clay jar with slag adhering, context 1421, ditch 310............................................................................. 254 Fig. 13.18. Fired clay crucible and jar (1-2)..................................................................................................................... 254 Fig. 13.19. Fired clay spindle whorls and loom weight (1-7)........................................................................................... 255 Fig. 13.20. Fired clay slingshot from context 197, ditch 313 and 103, clearance Aisled Hall......................................... 256 Fig. 13.21. Fired clay objects of unknown function (1-2)................................................................................................ 256 Fig. 14.1. Druce coins expressed as per mills................................................................................................................... 260 Fig. 14.2. Brooches (1-9).................................................................................................................................................. 263 Fig. 14.3. Millefiori brooch, context 1703, ditch 1605..................................................................................................... 264 Fig. 14.4. Bracelets (1-8)

........................................................................................................................................... 265

Fig. 14.5. Finger rings, ear ring, bead, bell and pendant (1-14) ..................................................................................... 267 Fig. 14.6. Coin reused as pendant SF 371......................................................................................................................... 268 xxvii

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Fig. 14.7. Pins and toilet implements (1-9)....................................................................................................................... 269 Fig. 14.8. Spoon, vessels and sheet fragments (1-12)....................................................................................................... 270 Fig. 14.9. Needle and seal boxes (13-16).......................................................................................................................... 271 Fig. 14.10. Fasteners and fittings (1-14) .......................................................................................................................... 273 Fig. 14.11. Military fittings (1-3)...................................................................................................................................... 275 Fig. 14.12. Belt tube (context 841) from above and below.............................................................................................. 275 Fig. 14.13. Rings, counter, pin and bar (1-6).................................................................................................................... 276 Fig. 14.14. Early medieval ring, buckle and hooked tag (1-3)......................................................................................... 277 Fig. 14.15. Pot mend SF 35 on Black Burnished ware dish sherd.................................................................................... 277 Fig. 14.16. Lead pot mends (1-9)...................................................................................................................................... 278 Fig. 14.17. Lead weights and counters (1-8).................................................................................................................... 279 Fig. 14.18. Lead vessel fragments (1-7)........................................................................................................................... 280 Fig. 14.19. Lead items of unknown function, offcuts and waste (1-10)........................................................................... 281 Fig. 15.1. Tools (1-8)........................................................................................................................................................ 283 Fig. 15.2. Tools and objects associated with transport (9-19)........................................................................................... 285 Fig. 15.3. Wheel fittings from context 1117 (20-23)........................................................................................................ 286 Fig. 15.4. Keys (24-27)..................................................................................................................................................... 287 Fig. 15.5. Domestic fittings and knives (25-32)................................................................................................................ 288 Fig. 15.6. Structural fittings (34-46)................................................................................................................................. 289 Fig. 15.7. Structural fittings and cleats (47-62)................................................................................................................ 291 Fig. 15.8. Nails Types 1-8 (63-77).................................................................................................................................... 293 Fig. 15.9. Shoe plate, nail Types 9-12, chain, ferrules, binding (78-92)........................................................................... 295 Fig. 15.10. Miscellaneous objects (93-103)...................................................................................................................... 297 Fig. 15.11. Smithing hearth bottom fragment................................................................................................................... 298 Fig. 15.12. Furnace conglomerate fragment..................................................................................................................... 298 Fig. 15.13. Copper smelting slag fragment....................................................................................................................... 299 Fig. 15.14. Hammerscale from pit 962............................................................................................................................. 299 Fig. 15.15. Copper droplet fragment................................................................................................................................. 299 Fig. 17.1. Percentage of number of bone fragments by period......................................................................................... 316 Fig. 17.2. Bone condition: percentage of fragments by period......................................................................................... 316 Fig. 17.3. Loose teeth by period and species.................................................................................................................... 317 Fig. 17.4. Taphonomic markers by period........................................................................................................................ 317 Fig. 17.5. Taphonomic markers by period, type and species............................................................................................ 317 Fig. 17.6. Butchery marks by period and species............................................................................................................. 318 Fig. 17.7. Butchery cut marks by period........................................................................................................................... 318 Fig. 17.8. Butchery marks by period and species showing light and heavy cuts.............................................................. 326 Fig. 17.9. Wear stages of cattle mandibles........................................................................................................................ 326 Fig. 17.10. Fused elements in cattle.................................................................................................................................. 328 xxviii

List of figures Fig. 17.11. Element frequency MNE in cattle.................................................................................................................. 330 Fig. 17.12. Percentage of sheep/goat surviving by period................................................................................................ 333 Fig. 17.13. Fused element percentage in sheep/goat........................................................................................................ 333 Fig. 17.14. Element frequency by MNE sheep/goat......................................................................................................... 334 Fig. 17.15. Element frequency by MNE pig..................................................................................................................... 336 Fig. 17.16. Red deer skull with antlers removed.............................................................................................................. 337 Fig. 17.17. Associated Bone Groups by period and species............................................................................................. 338 Fig. 17.18. Number of Associated Bone Groups by feature type and period................................................................... 338 Fig. 17.19. Associated Bone Groups, deposition context by species................................................................................ 339 Fig. 17.20. Associated Bone Groups animal size compared to feature type..................................................................... 339 Fig. 17.21. Associated Bone Groups composition by species.......................................................................................... 339 Fig. 17.22. Species in small vertebrate deposit 847 Period 5/6........................................................................................ 342 Fig. 17.23. Species in small vertebrate deposit 197 Period 5/6........................................................................................ 342 Fig. 17.24. Species in small vertebrate deposit 230 Period 6........................................................................................... 342 Fig. 17.25. Species in small vertebrate deposits from Period 4.2 to 5/6 contexts............................................................ 342 Fig. 17.26. Bird types by period....................................................................................................................................... 346 Fig. 17.27. Relative abundance of livestock species by period........................................................................................ 353 Fig. 17.28. Comparison of chicken and sheep/goat by period.......................................................................................... 354 Fig. 17.29. Comparison of cattle and horse by period...................................................................................................... 355 Fig. 18.1. Charred oak peg transverse section ................................................................................................................. 364 Fig. 18.2. Charred oak peg showing tapering point.......................................................................................................... 364 Fig. 18.3. Charred ash worked fragment transverse section............................................................................................. 364 Fig. 18.4. Charred ash worked fragment showing cut surfaces and tapering morphology............................................... 365 Fig. 18.5. Percentage frequency of taxa by number of samples....................................................................................... 365 Fig. 18.6. Taxa abundance based on fragment count........................................................................................................ 365 Fig. 18.7. Charcoal data by simplified habitat type.......................................................................................................... 367 Fig. 18.8. Period 2: most productive features................................................................................................................... 368 Fig. 18.9. Period 3: most productive features................................................................................................................... 375 Fig. 18.10. Period 4: most productive features................................................................................................................. 375 Fig. 18.11. Period 5: most productive features................................................................................................................. 376 Fig. 18.12. Cereal types using grain counts...................................................................................................................... 376 Fig. 18.13. Concentrations of cereal remains................................................................................................................... 376 Fig. 18.14. Samples from ovens 1045, 1128, 992, 764 and 1522, furnace 764, grain dryer 1758................................... 376 Fig. 18.15. Samples from most productive layers ........................................................................................................... 377 Fig. 18.16. Samples from ditch sections........................................................................................................................... 377 Fig. 18.17. Samples from most productive deposit, pits and postholes............................................................................ 377 Fig. 18.18. Grain dryer 1758 percentage remains............................................................................................................. 377 Fig. 19.1. Druce Farm boundaries and field names noted in the text............................................................................... 403 xxix

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Fig. 19.2. Druce villa and Durotrigian sites other than villas.......................................................................................... 404 Fig. 19.3. Southern Durotrigian Roman villas................................................................................................................. 405 Fig. 19.4. Comparison of Druce apsed room with similar at Fishbourne and Southwick................................................ 413

xxx

List of tables Table 1.1. Periods and phasing of the Druce Villa buildings................................................................................................ 7 Table 1.2. Radiocarbon determinations.............................................................................................................................. 10 Table 2.1. Summary of the total flint and chert assemblage composition.......................................................................... 15 Table 2.2. List of combination tools by context.................................................................................................................. 19 Table 3.1. Early Roman enclosures. Morphological information and total numbers of finds............................................ 26 Table 3.2. Midden 1528: total number of finds from layers............................................................................................... 44 Table 3.3. Midden 1528: coins from midden layers........................................................................................................... 46 Table 9.1. CBM and stone tile categories: numbers of recorded items............................................................................ 147 Table 9.2. Numbers of lower cutaway CBM roof tiles by Warry type and production date............................................. 151 Table 9.3. Tegulae signatures identified by Warry (2006)................................................................................................ 151 Table 9.4. Comparative measurements of stone tile types................................................................................................ 157 Table 9.5. Architectural stone objects by phase................................................................................................................ 160 Table 10.1. Phase 4 decorative style catalogue for Room N1.......................................................................................... 184 Table 11.1. Druce pottery fabrics with NRFC equivalents............................................................................................... 198 Table 11.2. Mortaria by fabric and phase.......................................................................................................................... 201 Table 11.3. Post-Roman/Saxon fabrics with number of sherds........................................................................................ 213 Table 12.1. Estimated vessel equivalents for pottery in Assemblage 1............................................................................ 215 Table 12.2. Estimated vessel equivalents for pottery in Assemblage 2............................................................................ 218 Table 12.3. Estimated vessel equivalents for pottery in Assemblage 7............................................................................ 220 Table 12.4. Estimated vessel equivalents for pottery in Assemblage 10.......................................................................... 223 Table 12.5. Estimated vessel equivalents for pottery in Assemblage 12.......................................................................... 225 Table 12.6. Estimated vessel equivalents for pottery in Assemblage 17.......................................................................... 226 Table 12.7. Estimated vessel equivalents for pottery in Assemblage 18.......................................................................... 227 Table 13.1. Stone objects by period.................................................................................................................................. 231 Table 13.2. Quantification of stone object types............................................................................................................... 231 Table 14.1. Identifiable Roman coins by Reece period.................................................................................................... 257 Table 14.2. Dorset villa coins against the British mean.................................................................................................... 261 Table 14.3. Summary of identifiable Roman copper alloy objects................................................................................... 262 Table 15.1. Summary of identified iron objects................................................................................................................ 284 Table 15.2. Iron nail typology and numbers..................................................................................................................... 292 Table 15.3. Summary of the metalworking debris............................................................................................................ 298 Table 15.4. Results of the XRF analysis........................................................................................................................... 300 Table 15.5. Metalworking debris by period...................................................................................................................... 300 Table 16.1. Summary of perinatal remains....................................................................................................................... 304 xxxi

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Table 17.1. Species abundance......................................................................................................................................... 312 Table 17.2. Associated Bone Groups................................................................................................................................ 313 Table 17.3. Number of bone fragments by feature type................................................................................................... 315 Table 17.4. Bone condition by phase................................................................................................................................ 316 Table 17.5. Taphonomy by period, percentage of total number of fragments.................................................................. 317 Table 17.6. Butchery by period and types of cut.............................................................................................................. 318 Table 17.7. Fragmentation all periods............................................................................................................................... 319 Table 17.8. Metrical information, mammals..................................................................................................................... 319 Table 17.9. Cattle toothwear............................................................................................................................................. 328 Table 17.10. Cattle sex from metacarpals......................................................................................................................... 329 Table 17.11. Cattle estimated withers height.................................................................................................................... 329 Table 17.12. Identified sheep/goats................................................................................................................................... 331 Table 17.13. Sex data, sheep/goats, pigs and domestic fowl............................................................................................ 331 Table 17.14. Tooth wear for sheep/goat............................................................................................................................ 331 Table 17.15. Sheep/goat estimated withers height............................................................................................................ 334 Table 17.16. Pig tooth wear.............................................................................................................................................. 335 Table 17.17. Horse estimated withers height.................................................................................................................... 336 Table 17.18. Small vertebrates from hand-collected concentrations................................................................................ 341 Table 17.19. Small mammals from context 197, Room N1.............................................................................................. 343 Table 17.20. Bone from soil samples................................................................................................................................ 344 Table 17.21. Wild mammal natural history....................................................................................................................... 345 Table 17.22. Bird NISP..................................................................................................................................................... 347 Table 17.23. Wild bird natural history.............................................................................................................................. 348 Table 17.24. Fish from hand-collected samples................................................................................................................ 350 Table 17.25. Fish from sieved samples............................................................................................................................. 351 Table 17.26. Fish natural history....................................................................................................................................... 352 Table 17.27. Contexts providing largest numbers of land snail species........................................................................... 356 Table 17.28. Contexts providing largest numbers of marine mollusc species.................................................................. 356 Table 17.29. Oyster shell infestation................................................................................................................................. 357 Table 18.1. Feature types and periods of analysed charcoal samples............................................................................... 359 Table 18.2. Analysis of charcoal fragments from Period 2-2.2/3.2 features..................................................................... 360 Table 18.3. Analysis of charcoal fragments from Period 3.1-3.2 features........................................................................ 360 Table 18.4. Analysis of charcoal fragments from Period 4.1-4.2 features........................................................................ 361 Table 18.5. Analysis of charcoal fragments from Period 5 features................................................................................. 362 Table 18.6. Summary of charred and mineralised plant remains...................................................................................... 369 Table 18.7. Comparison of stored grain from Lullingstone and Druce villas................................................................... 398 Table 19.1. Villas within the southern Durotrigian territory............................................................................................ 406 Table 19.2. Comparison of Dorset villas with aisled buildings........................................................................................ 419 xxxii

Preface This report presents in great detail the results of seven years of excavation on a villa complex sited some 6km to the north-east of Roman Durnovaria, modern Dorchester. The excavations at Druce will change our perception of villa development in central southern and south-western Britain. Druce is founded in the first century, soon after the establishment of nearby Dorchester. Such an early foundation date is unique in the region, with the majority of early villas in Britain occurring in the south and east, focusing on Sussex, Kent, Essex and Hertfordshire. The quality of the artefactual and ecofactual assemblages allows a detailed narrative of the status, economy and development of the site over the Roman period. In addition to the early foundation date, the careful excavation has also given a remarkable early post-Roman sequence which includes the latest Romano-British ceramics and Late Roman amphora imports. Additionally, the coin assemblage is well above the regional average for Reece period 21; a pattern in close accord with the neighbouring villa at Dewlish only 4km to the northeast. This pattern of fifth century – and perhaps later – activity has great implications for the continuing functioning of a villa estate in the environs of Durnovaria. This report will, I am certain, become a landmark publication in our understanding of the rise and decline of Romano-British villas and will become essential reading for scholars of the archaeology of the North-Western provinces of Rome. Mark Corney June 2022

xxxiii

Summary It was the wish of the landowner, Mrs Ann Ridout, to explore and excavate the part of Druce Farm where Roman material had been located by metal detecting and fieldwalking; subsequent fieldwork was organised by the East Dorset Antiquarian Society (EDAS). Post-excavation work was funded by the landowner, grants from the Mark Fitch Fund, the Roman Research Trust, the Valentine Charitable Trust, the Council for British Archaeology, and contributions from individuals and visiting groups. The excavation provided an opportunity to examine in detail the development of a small, rural Roman villa which was excavated to modern standards. A small area dated to the Early Neolithic period (c. 3941-3648 BCE) and consisted of a complex C-shaped ditch-like feature with associated pottery, flintwork and stone. Other contemporary pits and postholes were also recorded. Flintwork dating from the Mesolithic to the Late Bronze Age was scattered across the whole site, and small numbers of Bronze Age and Iron Age pottery fragments suggest activity in the area. Remnants of undated, but preRoman field systems were identified. The earliest Roman activity (c. 43-70 CE) consisted of a major ditch system defining two rectangular enclosures bounded by a narrower L-shaped enclosure. Contemporary features included a small flint-walled, three-roomed ‘Ancillary Building’ and an associated ‘Occupation Area’ which were located in the larger western enclosure; ironworking was located on its eastern side. Three pits were recorded in the eastern enclosure.

built opposite the East Range, replacing the Ancillary Building. Two areas to the east of the North and East Ranges, which may initially have been quarry pits, were used for rubbish disposal. Major building work took place on the North Range in the first half of the 3rd century. The apsed extension was demolished and replaced with two or three smaller rooms. The semi-cellar was retained, but the original rooms were extended north and were fronted by a new corridor. The room renovations included mosaic floors and painted walls. A stone and flint-walled grain dryer was constructed to the west of the villa, suggesting the processing of substantial quantities of cereals. In the second half of the century, the East Range was demolished, and an Aisled Hall with two aisles and six pier bases was constructed. Significantly, the waterproof tank was retained. The West Range was rebuilt with four large rooms, two of which contained stone-built ovens. The ditches around the villa continued to be used for rubbish disposal, but midden areas to the east and south were increasingly used. A small rectangular Eastern Enclosure, possibly used as a paddock, was set out to the east of the villa, and a north-south ‘hooked’ ditch was dug to the west of this.

The enclosure ditches saw further infilling, which continued throughout the life of the villa.

The most radical, extensive and probably expensive building works were carried out c. 300-350/70 CE. On the North Range, the cellar was filled in and the room was extended north. The mosaic floor of the easternmost reception room was replaced and a new service room was built to the east of this. Rooms in the western part of the range were re-modelled, new mosaics were laid and a porch constructed. A damaging fire in the Aisled Hall necessitated major rebuilding, resulting in a longer building with two more pier bases. The floor level was raised, and an oven was built in the north-west corner. All ranges were re-roofed with Purbeck limestone tiles; the Aisled Hall roof was decorated with imported slate. The buildings formed a courtyard villa, with a formal entrance which was constructed over the infilled ditches to the south of the buildings. The grain dryer was abandoned, and an adult human male, covered by large parts of a cow carcass, was buried in the stoke pit, reflecting non-normal or “abnormal”? burial practices.

At some time between c. 120-200 CE an imposing apsed extension, which has parallels in palatial buildings on the south coast, was built onto the North Range. The East Range was extended north, and a small waterproof tank installed. A series of pits were located outside this building. A small single-room, probably a workshop, was

Occupation continued through the final years of Roman control and into the early part of the 5th century (c. 350/70-430 CE). Crude repairs to the deteriorating North Range mosaic floors were undertaken but, by the end of this period, a series of pits were dug through the floors. The Aisled Hall and West Range continued in use, but by

Between c. 70-120 CE, maintenance of the ditch system had ceased, and infilling with domestic rubbish and building material commenced. The rooms within the Ancillary Building were reconfigured and activities in the Occupation Area continued. Towards the end of this period, a small enclosure was set out to the south of the Ancillary Building. The first flint-walled ranges were constructed in the eastern enclosure. The North Range comprised a strip building of four rooms, the western-most being a semicellar. The East Range was rectangular in form; a deep pit to the south-east was probably dug as a well.

xxxv

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) the late 4th/early 5th century, the large oven in the West Range was abandoned. Household rubbish and building debris was dumped on the top of long-abandoned ditches and in midden areas, suggesting that living areas were kept clean. Two large deposits of Barn Owl pellets confirm that parts of the North Range and Aisled Hall had been abandoned as living quarters. Eventually, the three ranges of buildings deteriorated and collapsed. The precise date of these events is uncertain, but a sequence of falling roof tiles followed by collapsing flint walls is evident on all ranges. Substantial pits were dug through the fallen roof in the north-west corner of the Aisled Hall; their fills contained large amounts of building debris and domestic rubbish relating to earlier phases of occupation, but also contained pottery dating the pit-digging and latest occupation to the 7th century. Similarly dated pottery confirms activity in the other decaying ranges. Sometime after c. 650 CE the villa buildings were abandoned and the site reverted to agriculture, probably grassland. The very small amounts of pottery and coins possibly derived from the settlement at Waterston on the south side of the river. The detailed chronological phasing of the villa, together with evidence from material finds, faunal and human remains and charred plants, has allowed the site to be placed within a broad economic, social, agricultural and landscape framework. Comparison with other similar sites in Dorset and beyond has highlighted its significance.

xxxvi

Résumé C’était le souhait de la propriétaire foncière, Mme Ann Ridout, d’explorer et de fouiller la partie de Druce Farm où des matériaux romains avaient été localisés par détection de métaux et promenades sur le terrain; des travaux de terrain ultérieurs ont été organisés par la East Dorset Antiquarian Society (EDAS). Les travaux de postexcavation ont été financés par le propriétaire foncier, des subventions du Mark Fitch Fund, du Roman Research Trust, du Valentine Charitable Trust, du Council for British Archaeology, de l’Association for Roman Archaeology et des contributions de particuliers et de groupes de visiteurs. La fouille a permis d’examiner en détail le développement d’une petite villa romaine rurale qui a été fouillée selon les normes modernes. Une petite zone datait de la période néolithique précoce (vers 3941-3648 av. J.-C.) et consistait en un élément complexe qui ressemble à un fossé en forme de C avec de la poterie, du silex et de la pierre associés. D’autres fosses et trous de poteaux contemporains ont également été enregistrés. Des silex datant du mésolithique à l’âge du bronze tardif étaient dispersés sur l’ensemble du site, et un petit nombre de fragments de poterie de l’âge du bronze et de l’âge du fer suggèrent une activité dans la région. Des vestiges de systèmes de terrain non datés mais préromains ont été identifiés. La première activité romaine (vers 43-70 après JC) consistait en un important système de fossés définissant deux enceintes rectangulaires délimitées par une enceinte en forme de L plus étroite. Les caractéristiques contemporaines comprenaient un petit «bâtiment auxiliaire» à trois pièces aux murs de silex et une «zone d’occupation» associée qui étaient situés dans la plus grande enceinte ouest; la ferronnerie était située sur son côté est. Trois fosses ont été enregistrées dans l’enceinte orientale. Entre env. 70-120 après JC, l’entretien du système de fossés avait cessé et le remplissage avec des ordures ménagères et des matériaux de construction a commencé. Les pièces du bâtiment auxiliaire ont été reconfigurées et les activités dans la zone d’occupation se sont poursuivies. Vers la fin de cette période, une petite enceinte est aménagée au sud du bâtiment auxiliaire. Les premiers chaînes aux parois de silex sont construits dans l’enceinte orientale. La Chaîne Nord comprenait un bâtiment en bande de quatre pièces, la plus à l’ouest étant une demi-cave. La Chaîne Est était de forme rectangulaire; une fosse profonde au sud-est a probablement été creusée en guise de puits. Les fossés de l’enceinte ont connu un nouveau comblement, qui s’est poursuivi tout au long de la vie de

la villa. À un certain moment entre c. 120-200 après JC, une extension imposante de l’abside, qui a des parallèles dans les bâtiments palatiaux de la côte sud, a été construite sur la Chaîne Nord. La Chaîne Est a été étendue au nord et un petit réservoir étanche a été installé. Une série de fosses étaient situées à l’extérieur de ce bâtiment. Une petite pièce unique, probablement un atelier, a été construite en face de la Chaîne Est, en remplacement du bâtiment auxiliare. Deux zones à l’est des Chaînes Nord et Est, qui étaient peut-être initialement des fosses de carrière, ont été utilisées pour l’élimination des ordures. D’importants travaux de construction ont eu lieu sur la Chaîne Nord dans la première moitié du IIIe siècle. L’extension de l’abside a été démolie et remplacée par deux ou trois pièces plus petites. La demi-cave a été conservée, mais les pièces d’origine ont été agrandies vers le nord et ont été précédées d’un nouveau couloir. Les rénovations des chambres comprenaient des sols en mosaïque et des murs peints. Un séchoir à grains aux murs de pierre et de silex a été construit à l’ouest de la villa, suggérant le traitement de quantités importantes de céréales. Dans la seconde moitié du siècle, la Chaîne Est a été démolie et une salle à allées avec deux allées et six bases de piliers a été construite. De manière significative, le réservoir étanche a été conservé. La Chaîne Ouest a été reconstruite avec quatre grandes salles, dont deux contenaient des fours en pierre. Les fossés autour de la villa ont continué à être utilisés pour l’élimination des ordures, mais les zones de dépotoir à l’est et au sud étaient de plus en plus utilisées. Une petite enceinte orientale rectangulaire, peut-être utilisée comme enclos, a été aménagée à l’est de la villa, et un fossé nordsud « en crochet » a été creusé à l’ouest de celle-ci. Les travaux de construction les plus radicaux, les plus étendus et probablement les plus coûteux ont été réalisés env. 300-350/70 après JC. Sur la Chaîne Nord, la cave a été comblée et la salle a été agrandie vers le nord. Le sol en mosaïque de la salle de réception la plus à l’est a été remplacé et une nouvelle salle de service a été construite à l’est de celle-ci. Les chambres de la partie ouest de la chaîne ont été remodelées, de nouvelles mosaïques ont été posées et un porche a été construit. Un incendie dévastateur dans la salle à allées a nécessité une reconstruction majeure, entraînant un bâtiment plus long avec deux bases de piliers supplémentaires. Le niveau du sol a été surélevé et un four a été construit dans l’angle nord-ouest. Toutes les chaînes ont été refaites avec des tuiles de calcaire Purbeck; le toit de la salle à allées était décoré d’ardoise importée. Les bâtiments formaient une villa avec cour, avec une entrée

xxxvii

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) formelle qui a été construite au-dessus des fossés remplis au sud des bâtiments. Le séchoir à grains a été abandonné, et un mâle humain adulte, couvert de grandes parties d’une carcasse de vache, a été enterré dans la fosse de chauffage, reflétant non normal ou “anormal” ? pratiques funéraires. L’occupation s’est poursuivie pendant les dernières années du contrôle romain et au début du Ve siècle (vers 350/70-430 après JC). Des réparations grossières des sols en mosaïque de la Chaîne Nord en détérioration ont été entreprises mais, à la fin de cette période, une série de fosses ont été creusées dans les sols. La salle à allées et la Chaîne Ouest ont continué à être utilisés, mais à la fin du 4ème / début du 5ème siècle, le grand four de la Chaîne Ouest a été abandonné. Des ordures ménagères et des débris de construction ont été déversés au sommet de fossés abandonnés depuis longtemps et dans des zones de dépotoir, suggérant que les zones de vie étaient maintenues propres. Deux grands gisements de boulettes d’Effraie des Clochers confirment que des parties de la Chaîne Nord et salle à allées avaient été abandonnées comme quartiers d’habitation. Finalement, les trois chaînes de bâtiments se sont détériorées et se sont effondrées. La date précise de ces événements est incertaine, mais une séquence de chutes de tuiles suivies d’effondrements de murs en silex est évidente sur toutes les chaînes. Des fosses importantes ont été creusées à travers le toit effondré dans le coin nordouest de la salle à allées; leurs remplissages contenaient de grandes quantités de débris de construction et d’ordures ménagères liés aux phases antérieures d’occupation, mais contenaient également de la poterie datant du creusement de la fosse et de la dernière occupation du 7ème siècle. Des poteries de date comparable confirment l’activité dans les autres chaînes en décomposition. Quelque temps après env. 650 après JC les bâtiments de la villa ont été abandonnés et le site est revenu à l’agriculture, probablement des prairies. Les très petites quantités de poterie et de pièces de monnaie provenaient peut-être de la colonie de Waterston, du côté sud de la rivière. Le phasage chronologique détaillé de la villa, ainsi que les preuves de découvertes matérielles, de restes fauniques et humains et de plantes carbonisées, ont permis de placer le site dans un large cadre économique, social, agricole et paysager. La comparaison avec d’autres sites similaires dans le Dorset et au-delà a mis en évidence son importance.

xxxviii

Zusammenfassung Es war der Wunsch der Landbesitzerin, Frau Ann Ridout, den Teil der Druce Farm zu erkunden und auszugraben, wo römisches Material durch Metalldetektion und Feldwanderungen gefunden worden war; die anschließende Feldforschung wurde von der East Dorset Antiquarian Society (EDAS) organisiert. Die Arbeiten nach der Ausgrabung wurden vom Landbesitzer, Zuschüssen des Mark Fitch Fund, des Roman Research Trust, des Valentine Charitable Trust, des Council for British Archaeology, der Association for Roman Archaeology und Beiträgen von Einzelpersonen und Besuchergruppen finanziert. Die Ausgrabung bot die Gelegenheit, die Entwicklung einer kleinen ländlichen römischen Villa, die nach modernen Standards ausgegraben wurde, im Detail zu untersuchen. Ein kleines Gebiet wurde auf die frühe Jungsteinzeit (ca. 3941-3648 v. Chr.) datiert und bestand aus einem komplexen C-förmigen grabenähnlichen Merkmal mit dazugehöriger Keramik, Feuerstein und Stein. Andere zeitgenössische Gruben und Pfostenlöcher wurden ebenfalls aufgezeichnet. Feuersteinarbeiten aus dem Mesolithikum bis zur späten Bronzezeit waren über die gesamte Stätte verstreut, und eine kleine Anzahl von Keramikfragmenten aus der Bronze- und Eisenzeit deutet auf Aktivitäten in der Gegend hin. Überreste von undatierten, aber vorrömischen Feldsystemen wurden identifiziert. Die früheste römische Aktivität (ca. 43-70 n. Chr.) bestand aus einem großen Grabensystem, das zwei rechteckige Gehege definierte, die von einem schmaleren L-förmigen Gehege begrenzt wurden. Zu den zeitgenössischen Merkmalen gehörten ein kleines dreiräumiges „Nebengebäude“ mit Feuersteinwänden und ein zugehöriger „Besatzungsbereich“, die sich im größeren westlichen Gehege befanden; auf seiner Ostseite befand sich eine Eisenhütte. Im östlichen Gehege wurden drei Gruben registriert. Zwischen c. 70-120 n. Chr. wurde die Wartung des Grabensystems eingestellt und mit dem Auffüllen mit Hausmüll und Baumaterial begonnen. Die Räume im Nebengebäude wurden umgestaltet und die Aktivitäten im Besatzungsbereich fortgesetzt. Gegen Ende dieser Zeit wurde südlich des Neben-gebäudes eine kleine Einfriedung angelegt. Die ersten Reihen mit Feuersteinwänden wurden im östlichen Gehege errichtet. Die Nordreihe bestand aus einem Streifengebäude mit vier Räumen, von denen der westlichste ein Halbkeller war. Die Ostreihe hatte eine rechteckige Form; eine tiefe Grube im Südosten wurde wahrscheinlich als Ziehbrunnen gegraben. Die Einfassungsgräben wurden weiter verfüllt, was sich während der gesamten Lebensdauer der Villa fortsetzte.

Irgendwann zwischen c. 120-200 n. Chr. wurde eine imposante apsidische Erweiterung, die Parallelen in palastartigen Gebäuden an der Südküste hat, an der Nordkette errichtet. Die Ostreihe wurde nach Norden verlängert und ein kleiner wasserdichter Tank installiert. Außerhalb dieses Gebäudes befanden sich eine Reihe von Gruben. Ein kleines Einzelzimmer, wahrscheinlich eine Werkstatt, wurde gegenüber der Ostreihe gebaut und ersetzte das Nebengebäude. Zwei Gebiete östlich der Nord- und Ostreihen, bei denen es sich möglicherweise ursprünglich um Steinbruchgruben gehandelt hat, wurden für die Müllentsorgung genutzt. In der ersten Hälfte des 3. Jahrhunderts fanden an der Nordreihe große Bauarbeiten statt. Der Anbau mit der Apsis wurde abgerissen und durch zwei oder drei kleinere Räume ersetzt. Der Halbkeller blieb erhalten, aber die ursprünglichen Räume wurden nach Norden erweitert und erhielten einen neuen Korridor. Die Renovierung der Zimmer umfasste Mosaikböden und gestrichene Wände. Westlich der Villa wurde ein Getreidetrockner aus Stein und Feuerstein gebaut, was darauf hindeutet, dass beträchtliche Mengen Getreide verarbeitet wurden. In der zweiten Hälfte des Jahrhunderts wurde die East Range abgerissen und eine Ganghalle mit zwei Gängen und sechs Pfeilerbasen errichtet. Bezeichnenderweise wurde der wasserdichte Tank beibehalten. Die Westreihe wurde mit vier großen Räumen wieder aufgebaut, von denen zwei aus Stein gebaute Öfen enthielten. Die Gräben rund um die Villa wurden weiterhin für die Müllentsorgung genutzt, aber zunehmend auch die Misthaufenflächen im Osten und Süden. Östlich der Villa wurde ein kleines rechteckiges östliches Gehege angelegt, das möglicherweise als Koppel genutzt wurde, und westlich davon wurde ein von Nord nach Süd verlaufender „hakenförmiger“ Graben ausgehoben. Die radikalsten, umfangreichsten und wahrscheinlich teuersten Bauarbeiten wurden ca. 300-350/70 n. Chr. durchgeführt. Auf der Nordreihe wurde der Keller zugeschüttet und der Raum nach Norden erweitert. Der Mosaikboden des östlichsten Empfangsraums wurde ersetzt und östlich davon ein neuer Dienstzimmer errichtet. Räume im westlichen Teil des Bereichs wurden umgestaltet, neue Mosaike wurden verlegt und ein Vorbau errichtet. Ein verheerendes Brand in der Ganghalle machte einen umfassenden Umbau erforderlich, was zu einem längeren Gebäude mit zwei weiteren Pfeilerbasen führte. Das Bodenniveau wurde erhöht und in der nordwestlichen Ecke ein Ofen eingebaut. Alle Bereiche wurden mit Purbeck-Kalksteinfliesen neu überdacht; das Dach der Ganghalle wurde mit importiertem Schiefer geschmückt.

xxxix

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Die Gebäude bildeten eine Hofvilla mit einem formellen Eingang, der über den zugeschütteten Gräben südlich der Gebäude errichtet wurde. Der Getreidetrockner wurde aufgegeben, und ein erwachsener Mann, bedeckt von großen Teilen eines Kuhkadavers, wurde in der Schürgrube begraben, was nicht normale oder „abnormale“ ? Bestattungspraktiken andeutet. Die Besetzung dauerte in den letzten Jahren der römischen Kontrolle und bis in den frühen Teil des 5. Jahrhunderts (ca. 350/70-430 n. Chr.). Es wurden grobe Reparaturen an den sich verschlechternden Mosaikböden der Nordreihe durchgeführt, aber am Ende dieser Zeit wurden eine Reihe von Gruben durch die Böden gegraben. Die Ganghalle und die Westreihe wurden weiter genutzt, aber im späten 4./frühen 5. Jahrhundert wurde der große Ofen in der Westreihe aufgegeben. Hausmüll und Bauschutt wurden auf lange verlassene Gräben und in Misthaufen-Bereiche geworfen, was darauf hindeutet, dass die Wohnbereiche sauber gehalten wurden. Zwei große Lagerstätten von Schleiereule-gewöllen bestätigen, das Teile der Nordreihe und der Ganghalle als Wohnquartiere aufgegeben wurden. Schließlich verschlechterten sich die drei Gebäudereihen und stürzten ein. Das genaue Datum dieser Ereignisse ist ungewiss, aber auf allen Ebenen ist eine Abfolge fallender Dachziegel, gefolgt von einstürzenden Feuersteinwänden, zu erkennen. Erhebliche Gruben wurden durch das eingestürzte Dach in der nordwestlichen Ecke der Ganghalle gegraben; ihre Füllungen enthielten große Mengen Bauschutt und Hausmüll aus früheren Besatzungsphasen, aber auch Keramik, die den Grubenbau und die letzte Besetzung auf das 7. Jahrhundert datierte. Ähnlich datierte Keramik bestätigt die Aktivität in den anderen verfallenden Bereichen. Irgendwann nach ca. 650 n. Chr. wurden die Villengebäude aufgegeben und das Gelände wieder landwirtschaftlich genutzt, wahrscheinlich als Grünland. Die sehr geringen Mengen an Töpferwaren und Münzen stammen möglicherweise aus der Siedlung Waterston auf der Südseite des Flusses. Die detaillierte chronologische Phasenlage der Villa, zusammen mit Beweisen aus Materialfunden, tierischen und menschlichen Überresten und verkohlten Pflanzen, hat es ermöglicht, die Stätte in einen breiten wirtschaftlichen, sozialen, landwirtschaftlichen und landschaftlichen Rahmen zu platzieren. Der Vergleich mit anderen ähnlichen Stätten in Dorset und darüber hinaus hat seine Bedeutung hervorgehoben.

xl

1 Environment and project background This report presents the results of the archaeological work associated with the excavation of the Roman villa complex at Druce Farm which took place between 2012 and 2018. Figure 1.1 is the key for all archaeological sections and plans.

between 90m and 70m above Ordnance Datum (OD). The aerial photograph (Fig. 1.3) shows the excavation areas in both fields in 2017; their location within the boundaries of Druce Farm is illustrated on Figure 19.1.

Location

Landscape and geology Andrew Morgan

The site on Druce Farm is located approximately 6.5km to the north-east of Dorchester in the parish of Puddletown, Dorset (Fig. 1.2). The villa and its environs are set within an area of land measuring 190m by 220m (c. 4.2 hectares) on the south-west corner of the farm (centred on SY 3733 9540). The complex overlooks the River Piddle to the south and occupies south-facing sloping land on two fields, Lower Limepits and Eleven Acres, which lie

The villa site is situated on the south-facing slope of the River Piddle valley; Fig. 1.4. illustrates its location and the contours. The slope has a gradient of approximately 1:12 and terminates in a narrow flat terrace and slight drop where it meets the river floodplain. Immediately to the east of the villa is a shallow dry valley running north-south. The Piddle (a chalk stream) rises at Alton Pancras and flows roughly south-east into Poole Harbour over 20 kilometres

Fig. 1.1. Key to section and plan drawings

1

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 1.2. Location of Druce Farm (upper) in Southern England in relation to major Roman sites noted in the text (lower), in Dorset

away. A spring is located just beyond the south-west corner of Lower Limepits field. By the early 19th century its flow had been redirected along a leat to feed cress beds further down river. The land throughout the wider area is now used

primarily for arable farming. Some of the field boundaries, including those between Lower Limepits and Eleven Acres, are approximately 8m to 10m wide, are wooded and were planted for game-shooting purposes. 2

Environment and project background

Fig. 1.3. Aerial view of the Druce site in 2017 with the extent of the excavations outlined in red. The Early Neolithic area and Early Roman features are located in Eleven Acres (to the left) and the villa complex is in Lower Limepits (to the right), view from the south.

and in the middle of Eleven Acres field that are recorded as abandoned pits (see Figs 1.3 and 1.5). The clay-withflints is a plastic material that has been used to remodel the landscape and create terraces, particularly for building purposes. These terraces provide a strong base that can support load-bearing walls without additional foundations.

The underlying bedrock is chalk, forming part of the chalk downlands that extend over much of central Dorset. This is overlain by superficial deposits of clay-with-flints. The surface of the chalk is irregular due to erosion and dissolution, with evidence of palaeo-channels (pit-like sink-holes) and vertical dissolution pipes of varying depth and width. In places the chalk is exposed after the shallow topsoil has been removed, whilst elsewhere it is several metres below the surface. The clay-with-flints is a heterogeneous and un-bedded formation comprising orange-brown or red-brown sandy clay with abundant small and medium-sized angular flint pebbles. In several places there is a basal layer, up to 100mm thick, comprising dark reddish-brown stiff clay containing fresh flint nodules stained black or dark green with manganese. The river floodplain is filled with a fertile mix of alluvium consisting of clay, silt, sand and gravel.

The villa is located in the exact centre of the county of Dorset. A circle with a 40-kilometre radius will encompass the whole county, touching the River Lim in the west, the river Avon to the east and extending beyond the Vale of Blackmore to the north. The site The field work and management of the site was undertaken by members of the East Dorset Antiquarian Society (EDAS) supported by members of other local societies, with input and advice from professional colleagues.

The chalk is an important resource, providing an endless supply of flint nodules, and is also used to make lime that can be mixed with sand to create mortar and plaster. The lime can also be applied to the soil as a fertiliser. Numerous pits have been quarried throughout the ages for the extraction of flint and chalk. There are large steepsided depressions to the north of Lower Limepits field

Project background In the early 2000s, the farm owner Mrs Ann Ridout gave permission for the Stour Valley Search and Recovery Club to metal detect on the farm. This resulted in several areas 3

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 1.4. Contour plot and location of the villa buildings and spring line

where Roman material was located, including the western side of the extensive Lower Limepits field. Detectorist, the late Dr Ken Wheatley, invited the author to view this field in 2009 due to the presence of large fragments of Roman ceramic and limestone roof tiles. After consulting the owner, a walkover survey was conducted in 2011. The collected finds included large amounts of building material and occupation debris confirming high potential for the location of a Roman settlement.

metre apart. The one metre vertical separation of sensors within each gradiometer on this instrument gives a theoretical detection depth of approximately one metre but may accentuate surface readings. A Geoscan FM256 gradiometer, which employs two fluxgate magnetometers 0.5m apart, was used in the 2011 survey. The vertical separation of sensors on this instrument gives a theoretical detection depth of approximately half a metre. Magnetometry measures the effect of buried objects on the Earth’s magnetic field. It is most effective on negative features, provided that the topsoil and sub-soil contain differing proportions of iron oxide, and on hearths and objects containing iron or nickel. By measuring the difference in readings from two sensors mounted vertically, a gradiometer eliminates interference from local objects such as power lines and deep geology.

Geophysical surveys An initial geophysical survey of an area measuring 120m by 120m comprising magnetometry and earth resistivity was conducted by Hannah Simpson of Bournemouth University in 2011. The results suggested the presence of three poorly preserved buildings situated on the east, west and north of a central courtyard, with associated enclosures and field systems (Ladle and Simpson 2012, 108). As the fieldwork progressed, further surveys were conducted in 2014 and 2016 by Dave Stewart (Bournemouth University and EDAS) indicating the full extent of the site (Fig. 1.5).

The excavations Mrs Ridout and her family were eager to find out as much as possible about this potential Roman site and, with this in mind, approached EDAS to arrange further work which took place between 2012 and 2018. A total of 153 weeks were worked Monday to Friday (10am – 4pm) by a core group of 20 volunteers; a further 90 people and supervised groups of school children also spent time on the project.

Equipment The survey equipment comprised a Bartington 601 gradiometer which employs twin gradiometers one 4

Environment and project background

Fig. 1.5. Composite geophysical surveys in Lower Limepits, Eleven Acres and Watery Mead.

Archaeological methodology

were noted. All features were recorded by section at a scale of 1:10. Site plans were produced at scales of 1:10, 1:20; 1:50 and 1:100. A site record was maintained using forms derived from the AC Archaeology recording system. A record of digital images consisting of over 5,000 photographs was compiled. Finds were washed, marked where appropriate, and catalogued by material type. Material was appraised annually and involved the identification, cataloguing and spot dating of all finds. A comprehensive environmental sampling strategy was undertaken consisting of both bulk and hand-retrieved samples.

No previous work had been undertaken on this site and strategies for investigation were developed on an annual basis dependent on the excavation results and advice from academic colleagues. In 2012, three trenches (T1, T2 and T3) were opened; a depth of approximately 0.2m of topsoil was removed by the farm bulldozer and the exposed areas cleaned by hand. In 2013, a small back-acting excavator was hired and used to expose a further seven trenches (T4 -T10). This machine was used to clear further areas at the beginning of the season in 2014, but in May that year EDAS member Robert Heaton provided a small JCB which subsequently cleared areas as and when needed. A total of 90 trenches were opened and numbered T1 to T90 (Fig. 1.6).

Regular on-site metal-detecting concentrated initially in and around the villa complex and was later extended to surveys of the adjacent fields. Excavation areas and their related spoil heaps were specifically targeted.

During the seven years of excavation a total of 2,296 contexts were recorded comprising clearances, features, layers, walls, pits, deposits, middens and large and small enclosures. A continuous series of context numbers were allocated, and stratigraphic and inter-context relationships

The Dorset County Museum provided access to a drone in 2014 and regular flights in a fixed-wing microlight aircraft by EDAS members Jo Crane and Sue Newman produced invaluable aerial photographs. In 2016, towards 5

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 1.6. Location of archaeological trenches T1-T90

the conclusion of the excavation in Lower Limepits field, Robert Heaton brought a 12-metre aerial work platform (cherry picker) to the site for photographs. Following a BBC South Today broadcast, Simon Dawson of SIAD Ltd., volunteered to undertake a terrestrial 3D laser scan of the site and the results were made available as a point cloud. The large quantity of data has not yet been processed for lack of the specialised technology, although 2D photographic images were made available. Over two weekends, Ceri Lambdin and a colleague undertook a topographical survey of the villa buildings using a Total Station. John Oswin undertook a topographical survey of each field using a Dumpy Level. Results from these surveys are held in archive.

The work and results have been widely disseminated through lectures to archaeology societies, amenity groups and to conferences. An article written by Andrew Selkirk was published in Current Archaeology (2017, 28-33). Regular interim reports were published in the Proceedings of the Dorset Natural History and Archaeological Society, Association for Roman Archaeology News and the Council for British Archaeology’s newsletters. Funding Funding for the project was received from four main sources: • The landowners Mr and Mrs Ridout and their daughter Amanda provided the initial funds and have continued to provide funding throughout the project. • The Environmental Research Programme was funded by the Mark Fitch Fund, the Roman Research Trust and the Valentine Charitable Trust. • Members of the general public made donations after visits to the site on Open Days or as part of an organised group, or after attending a series of talks about the excavation. • Several members of the excavation team and EDAS members made significant private contributions, with special recognition of large donations from Janet Bartlet, Phil D’Eath and the estate of the late Graham Adams.

Local community outreach From 2013 regular site tours were arranged for individual groups, together with a series of public ‘Open Days’ when the archaeology and artefacts were on view. This work generated valuable publicity for the project and donations towards post-excavation costs. The villa featured in two BBC South Today broadcasts. In 2014 and 2015, in liaison with Dorset County Museum, a Schools Programme was devised by Sue Cullinane and Bryan Popple. Over 250 pupils from 17 schools across Dorset, including all the pupils from the local Piddle Valley Primary School and members of the Dorchester branch of the Young Archaeologists’ Club, were able to spend time on site experiencing hands-on archaeology.

Over 87% of the funds were used to engage specialists to undertake various post-excavation tasks. A further 11% 6

Environment and project background covered other professional services required throughout the project and the remainder was spent on essential materials. The volunteers spent over 4,889 days on site, receiving no payment and covering their own expenses. This figure does not include the time spent by volunteers on ad hoc tasks and specialist post-excavation work. Full details of all receipts, payments and expenses are available in the Druce Roman Villa Project Financial Plan which is part of the project archive. Project aims • Determine the chronology of the villa comparing and contrasting with local and non-local examples. • Examine changes in settlement status and function. • Examine cultural associations, human remains and non-domestic activities. • Examine evidence for environmental change, agricultural production, food processing, craft and trade activities.

Fig. 1.7. Summary plan Period 1: prehistoric up to 43 CE

Summary of Periods In this volume ‘period’ refers to relative time scales, and ‘phase’ refers to building development. Table 1.1 details the chronological periods and the building and occupation phases of the villa complex, and a series of phase plans (Figs 1.7-14) illustrates the development of the site. The periods, which in some cases have been sub-divided, correspond to: • • • • • • •

Prehistoric up to c. 43 CE (Period 1) Early Roman c. 43-120 CE (Periods 2.1 and 2.2) Middle Roman c.120-250 (Periods 3.1 and 3.2) Late Roman c. 250-350/70 CE (Periods 4.1 and 4.2) Final Roman c. 350/70-430 CE (Period 5) Post-Roman c. 430-650 CE (Period 6) Early Medieval c. 650+ CE (Period 7)

Fig. 1.8. Summary plan Period 2.1: 43-70 CE

between 3941-3705 and 3775-3648 cal BCE. Small numbers of Bronze Age and Iron Age pottery suggest activity in the vicinity. Remnants of undated but preRoman field systems were identified.

Period 1: Prehistoric up to 43 CE (Fig. 1.7)

Period 2.1: 43-70 CE (Fig. 1.8)

Flint artefacts dating from the Mesolithic to the Late Bronze Age are scattered across the site. Early Neolithic occupation in Trench 88 was dated by radiocarbon to

A major ditch system delineating two rectangular enclosures was set out. Linear anomalies on the geophysics plot (Fig. 1.5) suggest that this system extended outside the site to

Table 1.1 Periods and phasing of the Druce Villa buildings ER (Early Roman), MR (Middle Roman), LR (Late Roman), FR (Final Roman) PR (Post-Roman) Period

Date (CE)

Ancillary Building (phase)

North Range (phase)

East Range (phase)

2.1 (ER)

43-70

1

2.2 (ER)

70-120

3.1 (MR)

120-200

West Range (phase)

2

1

1

3

2

2

1

3

2

T82 Occupation (phase) 1

3.2 (MR)

200-250

4.1 (LR)

250-300

4.2 (LR)

300-350/70

4

4

5 (FR)

350/70-420

5

5

6 (PR)

420-650

6

6

2

3

7

3

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) the south. A small, square flint-walled ‘Ancillary Building’ was constructed to the west in the larger inner enclosure together with a contemporary ‘Occupation Area’ to the north and a large pit (1735) to the south. Ironworking was located on the eastern side of this enclosure. Three pits were recorded in the eastern enclosure. Period 2.2: 70-120 CE (Fig. 1.9) The enclosure system ceased to be maintained and infilling with domestic rubbish commenced. Pit 604 was dug into ditch 308. The Ancillary Building was reconfigured and pit 1735 and the Occupation Area continued in use. Towards the end of this period a small enclosure was set out south of the Ancillary Building. The first phases of the flint-walled North and East Ranges were constructed; the North Range comprising a strip building of four rooms, the western-most (N3) built at a lower level. The East Range was rectangular in form. Pit 1182 on the south-east corner of the East Range was probably dug as a well.

Fig. 1.11. Summary plan Period 3.2: 200-250 CE

continued. An apsed extension was constructed to the west of the North Range. The East Range was extended north, and a small waterproof tank installed. Pit 1182 continued in use and a similar deep pit 1181/1149 was dug beside this. A probable cess pit 381 was dug to the north-east of the building and two other pits (209 and 896) were located to the west of this. To the south of the North Range, a flintwalled, single-roomed building comprised the first phase of the West Range. Two midden deposits (1355 and 1529) were located to the east of the North and East Ranges.

Period 3.1: 120-200 CE (Fig. 1.10) There was further infilling of the enclosure system and occupation of the Ancillary Building and Occupation Area

Period 3.2: 200-250 CE (Fig.1.11) Major work on the North Range consisted of demolishing the apsed extension and constructing a probable threeroom west extension over this and abutting the original range. Rooms N6, N7 and N10 were extended north and a corridor N11 gave access to these. The first mosaics were laid in Rooms N2, N10 and N11. Two parallel ditches (1459 and 1510) were dug inside the largely infilled inner enclosure south of the villa. A stone and flint-walled grain dryer was constructed to the west of the villa. Period 4.1: 250-300 CE (Fig. 1.12)

Fig. 1.9. Summary plan Period 2.2: 70-120 CE

The East Range was demolished and an Aisled Hall (which utilised the west wall of the previous building) with six pier bases was constructed. The waterproof tank was retained. The West Range was rebuilt forming four large rooms, two of which contained stone-built ovens. A new midden (1528) on the south side of the site was begun and the midden areas to the east of the complex continued in use. The enclosure ditches around the villa continued to be used for rubbish disposal. A small rectangular Eastern Enclosure was set out to the east of the villa complex and a north-south ‘hooked’ linear was dug to the west of this. The grain dryer continued in use. Period 4.2: 300-350/70 CE (Fig. 1.13) In the North Range, Room N3 was extended north. The mosaic in Room N10 was replaced and a further room

Fig. 1.10. Summary plan Period 3.1: 120-200 CE

8

Environment and project background

Fig. 1.12. Summary plan Period 4.1: 250-300 CE Fig. 1.14. Distribution of Period 6 metalwork and pottery indicating activity in and around the villa in the 5th and 6th centuries

Period 6: 430-650 CE (Fig. 1.14) Collapse of the three ranges of buildings with the fall of roofs (limestone and ceramic tiles) and flint walls. Pit digging occurred through this material in the north-west corner of the Aisled Hall, the pit fills contained large amounts of building debris and domestic rubbish relating to earlier phases of occupation, but also contained pottery dating the pit-digging to this later post-Roman period. Similarly dated pottery and two metal objects confirm activity in the decaying villa buildings. Fig. 1.13. Summary plan Period 4.2: 300-350/70 CE into the early 5th century

Period 7: 650+ CE Total abandonment of the villa; the site reverts to agriculture, probably grassland. Very small amounts of pottery and coins attest to occasional activity on the two fields probably from the Waterston settlement on the south side of the river.

(N9) was built to the east of this. Rooms N1 and N2 to the west were re-modelled, new mosaics laid and a porch (or lobby) N4 constructed. The Aisled Hall was extended north with two more pier bases. The floor level was raised, and an oven built in the north-west corner. A new midden area (315) was created in addition to those already in use. A formal entrance was constructed over ditches 311 and 312 and there was major re-cutting of enclosure 313. Infilling was noted in the easternmost ditches set out in the previous period. The grain dryer was abandoned, and an adult male (X) was buried in the stoke pit.

Radiocarbon dating Introduction In total 10 radiocarbon samples from Druce Farm were processed and dated at the Scottish Universities Environmental Research Centre (SUERC) at East Kilbride. The samples comprised four carbonised macrofossils, five animal bone and one human bone.

Period 5: 350/70-430 CE

The dating programme was designed to address the following objectives:

Crude repairs to the mosaic floors in Rooms N1 and N7 were undertaken. At the end of this period a series of small pits were dug in Rooms N9, N10 and N11. The Aisled Hall continued in use as did the West Range but by the late 4th/early 5th century the large oven in Room W1 was abandoned. Three of the four middens continued in use; enclosure 313 was re-cut and late Roman pottery, probably from the Aisled Hall, was discarded on top of this ditch and ditches 308 and 311 in the vicinity of the West Range. Smaller amounts of rubbish were deposited in ditches in the north-east corner of the site.

• Date the Neolithic feature • Date the earliest activity in the East Range • Confirm the date of an oven in order to correctly phase Room N5 • Date the collapse of the roof over Room N1 in the North Range • Date the skeletal remains from the stoke pit of the grain dryer 9

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) blank and the random machine error. The calibrated age ranges are determined from the University of Oxford Radiocarbon Accelerator Unit calibration programme (OxCal4). The calibration plots were calculated using methods derived from Bronk Ramsey et al. (2013) and Reimer et al. (2013). Date ranges were calculated using the IntCal20 atmospheric calibration curve (Reimer et al. 2020). All certificates and calibration curves are held in archive.

• Date the latest activity in Room N3 • Date the pit containing four juvenile sheep in Room N9 in the North Range • Date the pit containing two adult sheep in Room N10 in the North Range Sampling strategy Financial constraints limited the number of samples which could be submitted for radiocarbon dating. Material submitted included charred grains, animal and human bone. Samples were selected where pottery or coins were not available.

The results confirmed that eight of the ten dates returned from the samples were compatible and in accordance with the suggested phasing for the site. The dates from the small mammals from owl pellets associated with the abandonment of use of Room N1 in the North Range however are anomalous given the weight of other evidence.

Results and Calibration The results are presented on Table 1.2 and are conventional radiocarbon ages (Stuiver and Pollach 1977) which are cited according to Trondheim convention standards (Stuiver and Kra 1986). The 14C ages are quoted in conventional years BP (before 1950 CE) and require calibration to the calendar timescale. The error expressed at the one sigma level of confidence includes components from the counting statistics on the sample, modern reference standard and

Deposition and further work The site archive will be deposited in the Dorset County Museum. Due to limited storage, only finds from features noted in this publication will be deposited. The remaining material finds are retained by the landowner. There

Table 1.2. Radiocarbon determinations from selected features Laboratory Code

Sample

δ¹³C δ¹⁵N C/N Radiocarbon relative relative ratio Age (BP) to VPDB to air (Molar)

Calibrated date Range (95%) confidence

SUERC-77831 (GU46699)

DF 1880 Wheat grain Tritium sp from layer of feature 1860

-25.3 ‰

4928 ± 34

cal 3775-3648 BCE

SUERC-77832 (GU46700)

DF 1877 Hazelnut shell Colylus avellana from layer of feature 1860

-25.0 ‰

5008 ± 30

cal 3941-3858 BCE cal 3816-3705 BCE

1719 ± 29

cal 129-254 CE cal 289-321 CE

1831 ± 29

cal 126-253 CE cal 290-320 CE

SUERC-100062 DF1060 -22.3 ‰ (GU58584) Emmer/spelt from oven 1045 in the East Range

2.9 ‰

3.3

SUERC-100061 DF1009 (GU58583) Emmer/spelt from oven 992 in Room N5

-20.2 ‰

SUERC-100059 DF 822 (GU58581) Sheep/goat femur from pit 818 in Room N10

-21.3 ‰

5.3 ‰

3.3

1776 ± 29

cal 216-362 CE

SUERC-100055 DF 799 (GU58580) Sheep/goat tibia from pit 798 in Room N9

-21.0 ‰

5.2 ‰

3.2

1746 ± 29

cal 241-401 CE

SUERC-100060 DF 577 (GU58582) Cattle skull fragment from floor of Room N3

-21.1 ‰

5.0 ‰

3.3

1692 ± 29

cal 255-285 CE cal 326-421 CE

SUERC-84712 (GU50205)

DF 1957 -19.6 ‰ Human bone from stoke pit 1964 of grain dryer

8.9 ‰

3.3

1740 ± 26

cal 240-381 CE

SUERC-62809 (GU38738)

-22.3 ‰ DF 197 Water vole left mandible (1) from deposit over mosaic 1 (153), Room N1

2.9 ‰

3.3

1719 ± 30

cal 249-391 CE

SUERC-62810 (GU38739)

-23.3 ‰ DF 197 Water vole left mandible (2) from deposit over mosaic 1 (153), Room N1

6.1 ‰

3.3

1768 ± 30

cal 140-197 CE cal 208-346 CE

10

Environment and project background is potential for further work on a number of the finds categories, in particular, analysis of the paint pigments used on the wall plaster, analysis of the mortar and fabric analysis of the ceramic tiles. The small collection of human remains could yield DNA information. The biggest potential of course is the villa site itself: developing technology and additional excavation may answer some of the questions which have been highlighted in the following pages.

11

2 The Prehistoric background Early Neolithic activity

varying from 1.1m to 1.7m. The base was uneven and depths varied from 0.5m to 1.1m. The stratigraphy was complex and at least five phases were apparent. The first phase comprised a deep posthole 2160 which was succeeded a little distance away by a silty loam-filled pit 2294 containing small fragments of black and burnt flint. The third phase comprised a series of five small pits, set out in a crescent shape; the bases of four of these survived and the fifth 1781 was located outside the crescent. The loam fills of the pits contained worked and burnt flint and copious small quartz pebbles. In the fourth phase, the crescent-shaped ditch-type feature 1860 succeeded the pits. A fifth phase consisted of slumping or re-digging in the central, upper part (the grey area on Fig. 2.2), and it was this area which produced large numbers of carinated bowl pottery sherds, worked flint and charred material, two samples of which provided the radiocarbon dates.

In 2017, an area measuring 21m by 32m was stripped in Eleven Acres field and designated T88 (see Fig. 1.6) where the geophysical plot had suggested Roman activity. Initial investigation revealed features potentially associated with the Early Neolithic period. Two samples were submitted for radiocarbon dating. A fragment of hazelnut shell returned a date of 3941-3705 cal BCE (SUERC-77832: 5008±30 BP) and a wheat grain returned a calibrated date of 3775-3648 cal BCE (SUERC-77831: 4928±34 BP). As a result, an excavation programme was designed and implemented between July and November 2018. The work was undertaken by the EDAS team in consultation with Professor Timothy Darvill (Bournemouth University). The results will be published as a separate paper (Darvill et al. forthcoming), therefore only a summary is presented here. This small site has produced evidence for some of the earliest occupation activity in Dorset with associated pottery and flintwork. The site and location of the features are illustrated on Figures 2.1 and 2.2. The features

In addition, three groups of irregularly shaped pits were recorded; their loam fills were similar to that of the previous feature and produced worked and burnt flint. They may have been small sinkholes which infilled during the Early Neolithic period.

Excavation revealed a crescent-shaped feature 1860 with rounded terminals measuring 5.8m long and widths

Seventeen postholes in three rows were located and may have been associated with a possible timber-framed

Fig. 2.1. Neolithic features in Trench 88; feature 1860 at top left hand corner, view to the west.

13

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Ditch 878 Running north-south, this 50m length of ditch was located in the centre of the villa enclosure and did not extend beyond it; the northern extent lay underneath the North Range of the villa. The east and west ditches of the Roman enclosure (308 and 313) are parallel to 878 and may have been laid out when the ditch was still visible. Four sections were excavated with widths varying from 1.4m to 1.2m and depths from 0.3m to 0.45m. The profile had consistent gently sloping sides and a slightly rounded base. This ditch meets east west ditch 1291 at a T-junction half-way down its length. Section 879 (Fig. 2.3/1) Located on the southern extent of the North Range of the villa. This was covered by fallen limestone and sandstone roof tiles (884) from the porch of the building. The fill (880) comprised mid-brown loam, the base consisted of a 0.1m layer of grey silt. Sections 1282 and 1292 (Fig. 2.3/2); junction of 878 and 1291 The bases of these two sections were cut into chalk. The fill (1283) of section 1282 was clay loam with occasional charcoal fragments and moderate quantities of mediumsized flints. The fill (1293) of section 1292 which was excavated along the length of the ditch, was a dark brown, slightly sticky, silty loam.

Fig. 2.2. Plan of the Neolithic features in Trench 88

Ditch 1291

building; its width would have been approximately 7m and its length more than 21m. In addition, 16 postholes were located to the north and east of the structure and may have been associated with it.

This 120m long, east-west aligned ditch appeared to terminate in the west in the vicinity of north-south enclosure 1604 and was cut by later enclosures 1605 and 308. Only one section was excavated at the junction with 879. The section suggested that this ditch was earlier than ditch 878. It was probably about 1.2m wide and 0.5m deep with very steep sides; its profile may resemble that of ditch 1325. Due to the location of the trench, only a small area was available for investigation.

A sampling programme confirmed the presence of wheat (Tritium sp), barley (Hordeum vugare), bramble (Rubus sect. Glandolusus), hazelnut (Colylus avellana) and orache (Atriple sp). Further work on the charred seeds and charcoal is currently on-going. Pre-Roman ditches

Ditch 1325

Earlier field systems (see Fig. 1.7) underlay the Roman site and were apparent as linear features aligned slightly north-east/south-west and south-east/north-west. These are visible on the geophysical survey (Fig. 1.5) and were all difficult to locate on the stripped surfaces. In total, 15 ditch segments were excavated, four of which are illustrated. The location of the ditches and the excavated sections are illustrated on Figs 3.1 and 2.3. There were no diagnostic finds from any of the fills. The north-south ditches 878 and 1325 were approximately 60m apart and the east-west ditches 1291 and 1752 were approximately 45m apart. Both conform to the generally rectangular pattern of ‘Celtic Fields’ recorded by the RCHM (1970, 322); remnants of such field systems survive in the elevated north-east corner of the farm (Fig. 19.1).

A north-south ditch approximately 70m long having a possible junction with east-west ditch 1291. Three sections were cut on its southern extent. The widths varied from 1.6m to 1.4m and the depths from 0.75m to 0.7m. The steep sides and narrow, rounded bases were consistent, and the lower sides of the ditch were cut into chalk. Section 1506 (Fig. 2.3/3) The mid-brown loam fill (1507) contained moderate quantities of medium-sized flints, some burnt flint and rare charcoal fragments.

14

The Prehistoric background Ditch 1752

varied from 0.86m to 0.6m and depths varied from 0.5m to 0.16m; the sides were moderately steep and the base was rounded. The sections were cut into the clay-loam subsoil.

This 50m length of east-west ditch was only evident in Eleven Acres. It was 6m distant from, and parallel to, Roman enclosure 309. Narrower than the other preRoman ditches, the widths of the three excavated sections

Section 1793 (Fig. 2.3/4) The dark brown silty loam fill (1794) contained small amounts of medium-sized flint and rare charcoal fragments. Flint and chert Katharine Walker Introduction The flint and chert assemblage from the 2012 to 2016 Roman villa excavations in Lower Limepits field is described and discussed in the context of other lithic assemblages from the region. Flintwork from Eleven Acres and the Early Neolithic site will be published separately (Darvill et al. forthcoming). A total of 2,241 pieces of worked flint and chert were recovered from the excavation. These are broadly residual and redeposited in later contexts. Of these, 789 or 35% are tools, of which 439 are formal tools and 350 have been classified as expedient tools comprising utilised flakes and those with minor retouch; however, the distinction between these two categories is somewhat subjective. The assemblage is characterised by high numbers of expedient tools and poorly made formal tools on good quality flint from the local clay-with-flints deposits. Smaller numbers were made from Portland and greensand cherts. Against this background are finely produced formal tools of Neolithic to Early Bronze Age date made from both local flint and Portland chert. A particularly striking aspect of the total assemblage is that many of the objects are or have been produced on large thick flakes. Evidence for core rejuvenation is relatively sparse, which is perhaps unsurprising given the ready availability of local raw material. The majority of pieces to which a period can be attributed cluster around the Late Mesolithic and the Late Neolithic to Early Bronze Age. There are smaller quantities of Early Neolithic and later prehistoric material, and an even smaller number of pieces which may be Late Palaeolithic and Early Mesolithic in date, although these are incomplete and therefore insufficiently diagnostic to identify with certainty. The assemblage is recorded in detail in Appendix 1 (held in archive) and summarized in Table 2.1.

Fig. 2.3. Sections through pre-Roman ditches 878, 878/1291, 1325 and 1752 (1-4)

Table 2.1. Summary of the total flint and chert assemblage composition Raw material

Flakes and broken flakes

Blades and broken blades

Cores

Tools

Shatter

Spalls

Totals

Flint

1087

247

64

771

30

5

2204

Greensand chert

   6

  2

 2

   5

 0

0

   15

Portland chert

   4

  4

 1

  13

 0

0

  22

Totals

1097

253

67

789

30

5

2241

15

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Methodology

were recorded. Primary flakes are those with completely cortical dorsal surfaces, secondary with partly cortical dorsal surfaces and tertiary with uncorticated dorsal surfaces. Following Bellamy and Montague (2009), neither platform nor striking angle were recorded due to reasonable doubts raised by Whittaker (1994, 91) about the reliability of the index.

The flint and chert assemblage comprises material recovered through excavation except for four surface finds. Its analysis involved nine broad stages which both inform and provide structure for this report. 1. A detailed catalogue was made of the complete assemblage where the raw material, morphology and metrics of each piece were recorded. 2. The raw materials and their uses were described and discussed at assemblage level. 3. 3. The total assemblage was described with reference to the archaeological features excavated. 4. Cores represented in the assemblage and their reduction techniques were addressed. 5. Flakes and blades were considered in terms of assemblage composition. 6. Chips and shatter were analysed and interpreted. 7. The tools were examined. 8. The distribution of the total lithic assemblage was plotted and interpreted. 9. Conclusions and discussion.

Raw material Three main raw material types are represented; the largest component at just over 98% being flint, with less than 1% each of Portland chert and greensand chert. Most of the assemblage is moderately patinated to a mid-grey colour, but some less affected pieces indicate a buffcoloured flint. Several pieces display banding within the flint and a number contain geodes or small clusters of quartz crystals. Some pieces display a classic smoky ‘blue’ patina, whereas others are more heavily patinated white with occasional iron spotting along the ridges. Some occurrences of differential patination suggest that already fractured pieces, both natural and humanly struck, were also collected and worked. Where cortex is present, it is generally moderately thick and white, and ranges from a little worn to very fresh with some examples of thinner pebble cortex.

For the catalogue, each piece was recorded in detail. It was then counted by context and classified into a series of broad general categories for describing lithics (Andrefsky 1998; Ballin 2000; Butler 2005 and Inizian et al. 1999). A detailed description was given of any retouch, including location, angle, type, direction, and extent of removals (following Inizian et al. 1999). A range of other diagnostically significant features were also noted for each object (Pitts 1978 and Harding 1991, table 16). These helped with inferring function and attributing periods where possible. Raw material, condition, and the type of blank used were recorded. Patination, burning and postpatination damage were noted, along with anything else of importance.

The probable source of most of the raw material is the clay-with-flints which is associated with the local chalk bedrock. Abundant and relatively fresh angular flint nodules and rounded pebbles of flint feature heavily in the surface geology. These surface deposits provided a source of large workable nodules of sufficiently good quality, even for core tool production, though fresh flint was undoubtedly present in the underlying chalk. Greensand chert also occurs as a raw material for worked objects within the assemblage, albeit comprising only 15 pieces (0.67% of the total assemblage). Greensand chert displays abraded sugary sand grains and occurs in a range of colours including browns and oranges. It is widespread across southern England with coastal and inland outcrops from Devon to the Isle of Wight, including some close to the site at Druce. It also occurs as drift in residual deposits on the land surface, as glacial outwash and on beaches and in river gravels (Stewart 2019, 133 and fig. 9.1).

None of the flint and chert artefacts considered in this report were securely stratified. A small number were recovered from pre-Roman ditches, although the only dating evidence for these features was relative; the remainder were found in Roman contexts or in the clearance layers after topsoil was removed. Chronological characterisation was based largely on diagnostic tool types, production techniques and comparative data from flint assemblages elsewhere.

Portland chert occurs within the assemblage. It is a bluegrey lead colour (Rankine 1951, 93) but it varies in intensity, although compared with other types of chert its appearance is broadly consistent. Portland chert is confined to a relatively small region (Stewart 2019, 139 and fig. 9.1) and is geographically restricted to southern Dorset and some river gravels in the Vale of Wardour. It occurs on and around the Isle of Portland in the south, from Durlston Head in the east to Portesham in the west. Some pieces within this assemblage have a worn pebble cortex which has also been observed on items from the flanks of Maiden Castle and at Shroton near Hambledon Hill (Stewart 2019, 141). Stewart interprets this raw

Morphological and metrical analyses were based on Andrefsky (1998), while tool types and object attributes were based largely on Butler (2005) with reference to other publications (Edmonds and Bellamy 1991; Pitts 1996; Saville 1980). The length, breadth and thickness of all flakes and blades were measured and the weight of each piece recorded. Ballin (2017, 11) chooses not to weigh each item due to the time taken to do this, although it can be useful for identifying individual pieces should it be necessary to revisit the record later in the assessment. Condition, amount of cortex (i.e., primary, secondary or tertiary following Bradley (1970), and hammer mode 16

The Prehistoric background Flakes and blades

material as deriving from the local River Nadder that cuts through an exposure of Portland chert (Stewart 2019, 141). The material within the Druce assemblage may have been collected directly from Portland or its environs, from an outcrop or secondary source further inland, or both. Stewart noted that Portland chert never makes up a dominant or even significant percentage of an assemblage (ibid 2019, 139). The assemblage from Druce Lower Limepits field is no exception to this, with only 22 pieces representing 0.98% of the total. Some interpretations see Portland chert as an expediently gathered raw material (e.g., Sharples 1991); however, the abundance of naturally occurring flint in the Druce assemblage suggests its use was a more deliberate choice. Some of the Portland chert pieces display evidence of heat treatment, seen as an increase in lustre (Price et al. 1982), which makes it easier to work (Stewart 2017, 62). Domanski and Webb (2007) describe this as a ‘greasy shine’. If people in prehistory had ventured further afield to collect Portland chert, it is possible that among the otherwise local flint is some imported from elsewhere.

A total of 1,350 flakes and blades were recovered during excavation. Of these, 1,097 were flakes and comprise 48.95% of the total assemblage. Blades and bladelets total 253 pieces or 11.29% of the assemblage. Overall, 93 are primary flakes, 887 are secondary flakes, and 1,201 are tertiary flakes. There tend to be more primary and secondary flakes in later prehistoric assemblages as fewer flakes were removed. These came from the outer, cortical part of the nodule. Earlier prehistoric assemblages have higher numbers of tertiary flakes (Butler 2005, 181) as seen here. Pieces with a pebble cortex are typically, though not exclusively Mesolithic in date. Hard hammer mode (Ohnuma and Bergman 1982) predominated, probably using direct percussion, with some soft hammer mode. It is probable that production of flint artefacts continued well into the Bronze Age at Druce, with high numbers of squat flakes present. While there are no refitting pieces, there are some flakes that have clearly been struck from the same cores or nodules. These blades are likely to date from the Late Mesolithic and Early Neolithic. The continued production of blades is consistent with many Early Neolithic assemblages from southern England (Wainwright and Longworth 1971; Pitts 1978; Harding 2010).

Description of the total assemblage The flint and chert assemblage from the villa excavations was widespread across Lower Limepits field, and all came from Roman contexts except for a small number of pieces from pre-Roman ditches (878, 1291, 1325 and 1752). The character of the assemblage is mixed, with diagnostic pieces from the Mesolithic to the Late Bronze Age. The material from the features is likely to be residual and probably represents accidental incorporation rather than deliberate deposition by later communities. This is typical of the general ‘background scatter’ seen across the fields in the area; for this reason, the material is all considered together.

Axe-head production debris was present in very small quantities. The general lack of these flakes may be attributed to the fact that they are almost indistinguishable from debitage produced during flake or blade core preparation (Newcomer 1971; Harding 1990; 2010). Similar flakes are known from Maiden Castle (Edmonds and Bellamy 1991) Middle Farm, Dorchester (Harding 2004) and Poundbury Farm, Dorchester (Harding 2010). Chips and shatter

Cores and core reduction techniques

Shatter comprises angular pieces that cannot be identified as either flake or core debitage. Any flakes smaller than 10mm in diameter are classified as chips (Butler 2005, 41) and display all the usual attributes of flakes. The minimal numbers of chips and shatter within this assemblage is surprising given the presence of a significant amount of later prehistoric material with which larger amounts of shatter are commonly associated. The small size of this category may indicate that knapping took place elsewhere on the site, though a more likely explanation is perhaps a selective bias in collection.

Cores represent 2.99% of the total assemblage. This is slightly lower than assemblages from the South Dorset Ridgeway and around Maiden Castle, which had between 3% and 7% cores (Woodward 1991, table 1; Sharples 1991, table 5). A total of 67 cores were represented (Fig. 2.4/1-4). One of these was made from Portland chert (Fig. 2.4/4), two were greensand chert, and the majority were made from flint (Fig. 2.4/1-3). Of these, 54 had two or more platforms and almost all could be dated to the Neolithic and Bronze Age. The remaining 13 were unidirectional bladelet and flake cores dating from the Late Mesolithic; all single platform pyramidal (previously termed coned) cores of small size with overhangs left from the final removals (Butler 2005, 86). There is some evidence of platform abrasion to prepare blades and flakes before removal, but little evidence for raw material testing within the assemblage, which is typically seen as nodules with only one, two or three flake removals. One flake core of late Neolithic to Early Bronze Age date shows battering, indicative of subsequent use as a hammerstone.

Tools Scrapers A wide range of scrapers, 157 in total, were retrieved during this excavation (Fig. 2.4/5-7). These range from those with carefully retouched edges to those with short lengths of retouch. Among them are some Neolithic horseshoe scrapers of exceptional quality (Fig. 2.4/6 and 7). Of these, 73 are end scrapers, 45 are side scrapers, 16 are side and 17

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 2.4. Illustrated worked flint (1-10)

end scrapers, 13 are concave, three are discoidal, three are shouldered, two are nosed, and one is an end scraper with retouched shoulder. There is one which approximates a Y-shaped piece. This range is unsurprising given that scrapers are perhaps the most common implement in almost all periods of prehistory and are often found in greater numbers on a site than any other tool type (Butler 2005, 49). The size of the scrapers in this assemblage, except for those of Late Mesolithic date, is generally large, which probably reflects the availability of raw material.

Knives and cutting flakes A total of 93 knives were recovered from the site (Fig. 2.4/8). All were made from flint, and most were made on long flakes or broad blades. They fall into five different types. The first and most numerous types are simple unretouched knives made on flakes or blades; in total there are 90, some of which are a distinctive D shape indicative of an Early Neolithic date (Butler 2005, 129). The second largest category comprises long flakes with 18

The Prehistoric background an unretouched straight cutting edge which are naturally ‘backed’ with cortex. These have been termed ‘cutting flakes’, of which there are 21. There are nine backed knives, one of which is serrated. The more elaborate discoidal and ovate types of Neolithic knife are absent from this assemblage; however, there are two Early Bronze Age plano-convex knives.

Table 2.2. List of combination tools by context Context Trench

Burins, awls and piercers This assemblage yielded six burins or engraving tools. Burins date from the Upper Palaeolithic to Early Neolithic (Butler 2005, 51). While awls and piercers or borers are often grouped together due to a superficially similar appearance, there are notable differences between them (Butler ibid, 53). Both are present within this assemblage, totalling 54 pieces which display considerable variety. There are four awls which are characterised by a thinner cross-section retouched on either one edge of the point or alternate edges to perform a cutting motion when the tool is rotated (Fig. 2.4/9). The 50 piercers have abrupt retouch on two lateral edges and a thicker cross-section than the awls. Many of these are made on large hard hammer-struck flakes. One is made on a core. These are likely to date from the Late Neolithic to Early Bronze Age. Notched and combination pieces While notched pieces (Figs 2.4/10 and 2.5/11) occur in most periods of prehistory, they are never a common tool type. They occur frequently on late Neolithic and Early Bronze Age sites (Butler 2005, 170). It is striking that this assemblage has such a high number (54). Care has to be taken not to confuse them with plough-damaged flakes. The notches occur on a variety of flake and blade types and in a range of sizes up to approximately 10mm; 44 are on flakes, nine on blades, and one is on a bladelet. The notches are abruptly retouched on one of the lateral edges to create a deep, curved incision. All these pieces have a single notch. The assemblage contains 35 combination pieces (Table 2.2): 22 have scrapers; 18 have knives; 12 have borers or piercers; 12 have notches (Fig. 2.5/12). These first appear in the archaeological record in the Neolithic and are most commonly associated with the Late Neolithic, though they occasionally appear in Early Bronze Age assemblages (Butler 2005, 168). Arrowheads The assemblage contains four arrowheads. Two leafshaped forms (Fig. 2.5/13) are typically considered to be early Neolithic in date; however, some have been found in association with Middle Neolithic transverse arrowheads and Peterborough ware (Green 1984, 32). One is a Late Neolithic to Early Bronze Age transverse arrowhead made from flint (Fig. 2.5/14). Another is a transverse arrowhead made from Portland chert (Fig.

Feature 1

Feature 2

1

1

knife

notch

1

1

notch

side scraper

5

2

notch

scraper

16

2

cutting flake/knife side scraper

130

6

end scraper

knife

152

10

piercer

scraper

204

10

notch

knife

316

11

borer/piercer

side scraper

317

12

borer

scraper

322

17

side scraper

D-shaped knife

322

17

notch

side scraper

323

18

end scraper

knife

325

20

side & end scraper

notch

325

20

knife

piercer

325

20

knife

notch

325

20

knife

scraper

325

20

knife

fabricator (?)

325

20

side & end scraper

notch

325

20

knife

piercer

326

21

piercer

notch

347

11

knife

side scraper

355

20

notched piece

burin

360

11

side & end scraper

notch

386

14

borer/piercer

side scraper

848

31

piercer

scraper

1010

40

backed knife

piercer

1010

40

knife

notch

1033

39

concave scraper

knife

1037

40

end scraper

piercer

1140

N/A

notched blade

piercer

1278

52

cutting flake/knife piercer

1364

14

notch

end scraper

1364

14

knife

piercer

1408

14

backed knife

piercer

1439

60

scraper

notch

Feature 3

notch

scraper

notch

broken piercer

end scraper

2.5/15); this aligns most closely with Nicholas’s (2017, fig. 11) number 24. Portland chert was used for making a range of tools in prehistory but was particularly favoured for arrowheads (Green 1980; Keene 1999; Stewart 2017). Characteristically it is softer, lighter and more elastic than flint, making it ideal for this purpose (Stewart 2017, 64). 19

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) considered informal tools and are likely to have been expedient pieces made and used for a specific task before being discarded.

The quality of the invasive retouch on the Portland chert arrowhead is high. Miscellaneous ‘expedient’ flake tools and retouched flakes

Picks, adze-heads and axe-heads Two picks, three axe-heads (Fig.2.5/16), two complete and one fragment (Fig. 2.5/17) comprise the core tools within this assemblage. Both the complete axe-heads show evidence of partial polish. This, combined with their

A total of 349 otherwise unclassified pieces displayed miscellaneous retouch or evidence of use. These vary considerably in terms of the blank chosen, as well as the type, location and amount of retouch. These are

Fig. 2.5. Illustrated worked flint (11-17)

20

The Prehistoric background broadly lenticular shape, is indicative of a probable Early Neolithic date. The picks are likely to be of Mesolithic or Early Neolithic date. Gardiner (1990) identified a Late Neolithic relationship between axe-head manufacture and clay-with-flints on Cranborne Chase; however work at Poundbury Farm, also on clay-with-flints, suggests that axe-head production on a large scale was largely restricted to the Early Neolithic period (Harding 2010). The same was true at Maiden Castle (Wheeler 1943).

features in these areas which contributed to the matrix of later fills. An alternative interpretation might be that the fills of these later features included material brought in from elsewhere that incorporated flints. Discussion The lithic assemblage from Lower Limepits field represents a collection of material ranging from the Mesolithic to the Middle/Late Bronze Age. Elsewhere in Britain (Adkins and Adkins 1985, 69-75) and in Gaul (de Vesly 1909, 84-92), some Roman sites show evidence of prehistoric flint artefacts, in particular axe-heads of Palaeolithic and Neolithic date, being deliberately incorporated into buildings or collected as curiosities and deposited in local features (Walker 2018, 27). There is no clear evidence that this was the case at Druce, as much of the assemblage is rather ordinary, rather than high quality or prestigious, and it is unlikely that it would have attracted particular attention from the RomanoBritish community. The re-occupation of a site wellused in earlier times seems more likely. Generally, the artefacts were extremely fresh, showing little sign of postdeposition damage, suggesting that they had not moved far; however, they are interpreted as residual having made their way into archaeological features accidentally. The large size of many of the flakes and tools reflects the availability of the local raw material collected from the clay-with-flints. Materials including Portland chert were acquired from further afield. The assemblage from Lower

Distribution of the assemblage across the site A table of the occurrence of each artefact type was prepared for each context across the excavation site (Appendix in archive) and the results were then combined to provide an overview of the distribution of the total worked flint assemblage (Fig. 2.6). It should be noted however, that surface collection would typically provide a more accurate picture of the general distribution of flint across a site as the method of retrieval would not be distorted by the varying number, size and complexity of archaeological features excavated in different parts of the site. Not to include a distribution plot would, however, be an omission, and some notable concentrations can be observed on the plan. There is a particularly high concentration in the southwest corner of the site, with over 500 pieces of worked flint coming from Trench 52. A relatively high number of pieces can be seen in features diagonally across the centre of Lower Limepits Field from southwest to northeast. Perhaps there are destroyed

Fig. 2.6. Distribution of flint and chert from excavated contexts in Lower Limepits field

21

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Limepits Field is representative of similar activity across neighbouring fields at Druce Farm; significant Early Neolithic activity in Eleven Acres (see above) will be the subject of a future report. The present assemblage is also likely to be broadly typical of flint assemblages across large parts of southern England, with high numbers of expedient tools made for a range of purposes. The presence of finely made horse-shoe scrapers and leafshaped and oblique arrowheads may suggest a particular focus of activity in the Neolithic. Further work at the site may shed greater light on this; however, Lower Limepits Field at Druce was clearly a place of significance to people throughout much of prehistory as well as the Romano-British period and beyond.

F4: Sand F5: Sand + flint F6: Sand + flint + grog

F10: Sand + rock + iron ore F11: Grog + sand F12: Grog + sand + iron ore

Discussion All the prehistoric pottery discussed here derived from Roman deposits. Twenty-two Early Neolithic sherds weighing 82g were retrieved from the vicinity of the features in Trench 88. There were no rims or bases, but the small size of the sherds (average weight 3.7g) and fabric types (F1, F2, F3) compare with similar fabrics from the T88 activity area. Nine sherds (98g, average weight 10.8g) derived from Middle Bronze Age vessels, six from the upper fill of midden 1528 (T52) and three from clearance around Early Roman pit 1735 (T73). Five fabric types (F1, F2, F3, F10 and F12) were represented; three sherds bore typical Middle Bronze Age decorative motives (Fig. 2.7/1-3). All sherds were either unstratified or residual. Comparisons can be found in the exceptionally large assemblage from Bestwall Quarry (Woodward 2009, 213-244). Five sherds were identified as broadly Iron Age (Fabrics F6, F7, F11) and a further three sherds were unattributed but the fabrics (F3, F4, F5) suggest a pre-Roman date.

Illustrated flint and chert artefacts Some Special Find (SF) numbers were allocated during excavations and are noted after the trench (T) number. Fig. 2.4 1. Bladelet core: context 416, T16 2. Bladelet and flake core: context 1278, T52 3. Multi-directional flake core: context 361, T11 4. Portland chert core: context 25, T2 5. End scraper: context 349, T22 6. Horseshoe scraper: context 225), T5 SF115 7. Horseshoe scraper: context: 949, T11 8. Knife: context 1408, T14 9. Awl: context 997, T35 10. Notched piece: context 808, T32

The small assemblage confirms that the Neolithic activity in T88 probably extended out of the excavated area, but neverthe-less, was confined to Eleven Acres. The Middle Bronze Age sherds confirm activity in the southern part of both fields and the Iron Age sherds are distributed across the site.

Fig. 2.5 11. Notched piece: context 1, T1 12. Combination piece: awl and scraper: context 1, T1 13. Leaf-shaped arrowhead: context 429, T20, SF 217 14. Oblique arrowhead: context 1403, T52 15. Portland chert oblique arrowhead: context 143, T10 16. Axe-head, partially polished: context 322, T17, SF 317 17. Axe-head fragment, partially polished: context 893, T36 Prehistoric pottery The Early Neolithic site in Trench 88 produced 432 pottery sherds weighing 1,327g, from carinated bowls and are discussed in Darvill et al. (forthcoming). A total of 45 sherds (279g) of other prehistoric pottery was recorded from 40 contexts across the remainder of the site. The sherds were generally small (average weight 6.2g) and most were moderately abraded. In total, 12 fabric types were identified and correlate well with other local contemporary fabrics. Fabric Code F1: Grog F2: Grog + flint F3: Flint

F7: Sand + flint + rock F8: Sand + limestone F9: Grog + black iron ore

Fig. 2.7. Middle Bronze Age pottery from contexts 1631 and 1278 (1-3)

22

The Prehistoric background Fig. 2.7 1. Context 1631; clearance over pit 1735 and ditch 1641, T73 Wall sherd with row of small fingertip impressions, from the belly of a biconical vessel. Red-brown F12 fabric with chunks of cream-coloured grog up to 4mm, coarse sand and angular ‘slaggy’ ironstone. 2. Context 1278; upper fill of midden 1528, T52 Wall sherd with row of fingernail impressions, probably executed by a left hand. From the belly of a biconical vessel. Red-brown fabric F2 with pale brown grog and sparse small flint. 3. Context 1278; upper fill of midden 1528, T52 Wall sherd with plain raised cordon from the upper part of a probable straight-sided bucket urn. F1 Black fabric with beige exterior. Conclusion There is clear evidence for Early Neolithic activity on the site with rare, dated features. Flintwork collected during excavation of the Roman site suggests the presence of people utilising the landscape over a long period and the small assemblage of pottery supports this. There is no doubt that underlying prehistoric features are present in this landscape.

23

3 Enclosures, ditches and middens The villa complex lies within the inner enclosure which measures approximately 85m by 50m. To the west, the outer enclosure measures approximately 120m by 75m and contains an ancillary building, an occupation area, a grain dryer and a small ditch system.

The enclosures, ditch systems and middens associated with the villa were initially identified on the geophysical survey (Fig. 1.5) and were located in Lower Limepits and Eleven Acres fields. A 15m wide north-south band of trees and scrub separated the fields. The excavation results suggested that these features were dug and subsequently infilled over several centuries. They are considered by type and phase. The survey extended outside the excavation area to the south and east but the ditches here were not investigated. In addition to the major ditches which were associated with the villa complex and the area south and west of this, a short ‘hooked’ linear and a small enclosure were located on the eastern side of the excavation area. Midden areas used for the disposal of rubbish were located; the geophysics suggests other potential areas to the north-east of the excavated site. The ditches and middens are plotted on Fig. 3.1 with the ditch and midden numbers shown and the illustrated sections highlighted.

The ditches were cut into the natural clay-loam and chalk and their morphology is particularly regular. The sections are generally fairly steep sided with gently rounded or flattish bases. Widths varied from 2.8m to 1.3m and depths from 1.4m to 0.6m. The fillings generally comprised midbrown clay loam with varying quantities of flint-gravel. The enclosures and ditches are considered in groups and appropriate sections have been selected to illustrate their profiles, infilling and development, with the relevant trench (T) number also noted. The outer enclosure (ditches 1604, 309, 313, 311 and 312)

The Early Roman enclosure system

Ditches 1604 and 309 form the western and northern extent of the villa complex and ditches 313, 311 and 312 comprise the southern and eastern part of this circuit.

Now located in two modern fields, a rectilinear enclosure system was laid out early in the Roman period with evidence of infilling starting in the mid-1st century CE; the date of its cutting, however, must precede that of the first infilling. The total length of ditches exposed and recorded in plan was 758m and 75 segments were excavated, so that approximately 12% of the system was sampled. The discrete lengths of ditch were designated unique context numbers (308, 309, 310, 311, 312, 313, 314, 1504, 1604 and 1605) to aid excavation, although they were all part of a deliberately planned single system which the morphology suggests were likely to have been laid out at the same time. There is no evidence to indicate where the spoil was deposited. Table 3.1 details the ditch morphologies and finds.

Ditch 1604 Ten sections were cut through this 140m long ditch on the western extremity of the site. All the sections were moderately steep sided with slightly rounded bases. The width of the ditch varied from 1.73m to 2.1m and the depths from 0.6m to 1.36m. The lack of artefactual material and homogeneity of the fills was notable. A thin layer of primary fill was generally visible. The pottery suggested final infilling during the 2nd century CE, although the section at the junction with ditch 309 produced material dating to the later 4th century. There was no evidence for re-cutting.

The ditches overlay the earlier ‘Celtic field system’ and were constructed on a similar alignment. The layout comprised two north-south, western ditches (1604 and 1605) 25m apart which turn at a sharp right angle and continue eastwards as 309 and 314. The latter turns south as 308. Ditch 309 continues south as 313, then sharply turns west as 312/311. Originally one ditch, this was infilled in its central portion and became a southern entrance to the villa in the 4th century. The western end of Ditch 311 formed a junction with the north-south 308. Ditch 310 was 17m south of 312/311 and parallel to it; its length could not be determined as the eastern extent lay under the southern boundary of the site. The western terminal of ditch 310 appeared to join north-south ditch 1504, and pottery evidence suggests that they are of the same date. The ditches are considered as outer and inner enclosures with southern outliers.

Fig. 3.2/1 Section 1634 (T67) Width 2.1m, depth 1.88m with moderately steep sloping sides and a narrow, rounded base. The upper fill (1635) was mid-brown loam with flints, mostly derived from building debris. Two body sherds of pottery (12g) were identified as Early Roman. The lower fill (1708) was dark brown clay-loam with charcoal flecks. Ditch 309 Seven sections were cut through this 151m long ditch on the northern part of the site. The ditch merged with 1604 on the west and 313 on the east. The profiles varied from steep to moderately steep and the bases from flat to round. The 25

Ditch

No Cuts

Length

av width

av depth

pot(no)

pot (wt)

Fe (no)

Cu/Pb (no)

slag (wt)

AB (wt)

Shell (no)

Sh/st obj

F clay (wt)

308

11

70

2.2

1.2

1762

19878

169

9

2566

9201

104

6

201

309

7

151

2.2

1.1

259

3236

93

299

956

7

1

310

6

21

1.5

0.8

903

1078

58

3

3010

1328

7

21

2

587

26

311

5

24

2.2

0.96

701

6518

30

665

1539

26

1

312

3

28

1.5

0.94

796

6423

35

548

1234

22

2

8

313

9

86

2

0.97

1889

20339

96

606

8372

81

8

303

314

3

8

1.7

0.8

53

451

6

56 1123

1270

69

3

409

36

68 1610

15

2 44

1504

4

115

1.83

0.88

265

3725

29

1604

10

140

1.86

0.96

146

796

127

1605

17

115

1.7

1.14

1566

13882

29

1

Totals

75

758

1.86

0.97

8340

76326

672

19

8909

25578

331

Ditch

Bt St (wt)

Hst (wt)

LBS (wt)

SRT (wt)

CRT (wt)

FT (wt)

Tess (no)

Bx (wt)

PP(wt)

Mort (wt)

308

1816

1421

39034

51781

1610

76677

3049

104

52

64

25291

95

1131

10891

948

35528

185

6156

309

4

59

310

2022

883

26536

15221

30

18357

1804

2

75

87

311

701

2103

20271

11752

258

58091

278

39

37

3117

312

259

1045

3560

3189

4811

795

1

135

120

363

313

8110

404

10667

80334

217950

2106

69

280

1134

33407

314

18

1245

301

1643

1504

432

44

7275

204

9841

1604

33

2050

322

96

1492

3237

1962

126

14831

578

1605 Totals

13373

10025

13372

758

114571

181192

4391

438336

160

8192

89

274

579

1503

73968

1508

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Table 3.1. Early Roman enclosures: morphological information and total numbers of finds (All measurements in metres, all weights in grams) Abbreviations: Wt (weight), no (number), Fe (iron), Cu/Pb (copper alloy and lead), AB (animal bone), Sh/St obj (shale/stone object), F clay (fired clay), Bt St (burnt stone), Hst (heathstone), LBS (limestone building stone), LRT (Limestone roof tile), SRT (Sandstone roof tile), CRT (ceramic roof tile), FT (floor tile), Bx (box flue tile), Tess (tessera), PP (painted plaster), Mort (mortar)

Enclosures, ditches and middens

Fig. 3.1. Location of enclosures, ditches and middens with allocated number and illustrated sections highlighted. For detailed plan of T52 see Fig. 3.10/1

for recutting; the profiles of the initial cut continued to be moderately steep-sided with rounded/flat bases, those of the recuts were wider and shallower. The width of the ditch varied from 1.64m to 2.2m and the depths from 0.58m to 1.2m. All sections were artefact-rich, the lowest fills generally providing Middle Roman dates and the upper fills Late Roman. A single fragment of human infant bone was recorded. The largest quantities of building material came from this ditch and probably relate to re-building and refurbishments of the Eastern Range. The earliest rectangular building lay 6m to the west of the ditch while the later Aisled Hall was only 3m away; the re-cut sections were all in the vicinity of this building.

width of the ditch varied from 1.3m to 2.4m and the depths from 0.7m to 1.4m. Only the sections nearest to the North Range produced artefactual material, which dated from c. 200-400+ CE. The lack of earlier material may suggest that this section of the enclosure was scoured regularly. It may also account for its irregularly shaped base which, on several segments, retained the original profile. There was no evidence for re-cutting, but the pottery is associated with large quantities of building material which is likely to have derived from re-modelling the North Range in the early 3rd century and again in the 4th. Fig. 3.2/2 Section 57 (T3) Width 2.25m, depth 1.37m with steep sides and a very narrow base. Three fills were recorded. The upper (61) of dark brown loam contained quantities of building debris (flint and ceramic, limestone and sandstone roof tiles). The middle fill (58) of dark brown charcoal-rich loam contained large amounts of ceramic roof tile. Associated pottery (21 sherds weighing 405g) comprise Black Burnished ware (BB1) jar, bowl and dish fragments dating c. 300-400+ CE. The lower fill (62) of mid-brown loam had rare charcoal and gravel at the base. Pottery (33 sherds weighing 499g) included BB1 jars, bowls and dishes and sherds from New Forest and Oxfordshire beakers and a flagon. Deposits of animal bone indicate butchery waste.

Fig. 3.2/3 Section 128 and recut 177 (T6) Width 2.3m, depth 0.95m with steep sides and a rounded base; recut 1.85m wide and 0.5m deep with moderately sloping sides down to the base. Fill 129 covered the complete section and consisted of building debris in midbrown loam and was dated c. 370-430 CE by two BB1 bowls. The upper fills of re-cut 177 (130 and 137) were dark brown loam; the former contained a higher density of charcoal fragments. The pottery (49 sherds weighing 450g) from 130 was dated c. 330-430 CE, and that from 137 (147 sherds weighing 1,714g) dated c. 270-350 CE. Most of this assemblage was BB1 jars, bowls and dishes together with New Forest and Oxfordshire beaker, bottle and bowl sherds. A single sherd from a Corfe Mullen flagon (c. 43-100 CE) was recorded. Fill 169, a charcoal-rich brown-black loam contained pottery (82 sherds weighing 721g) dating c. 120-300 CE. The coarsewares were BB1 apart from a single sherd of New Forest greyware. New

Ditch 313 Nine sections were cut through this 86m stretch of ditch on the eastern side of the villa. The ditch merged with 309 on the north and 312 to the south. Five sections had evidence 27

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 3.2. (1-5) Sections through enclosures 1604, 309, 313, 311 and 312

Forest bottle and beaker sherds were also present, but the earliest sherds were from a Corfe Mullen flagon and an East Gaulish samian beaker. The base fill (176) of the recut contained decaying chalk, probably from the ditch sides. The small pottery assemblage (18 sherds weighing 178g) dated c. 200-300 CE. Butchery waste was recorded from all layers. This re-cut dates to the Late Roman period 4.1. The original ditch lay slightly to the west; its profile of rounded base and steep sides conforms to the unaltered profiles seen on ditches 1604 and 1605. The orange/brown clay/loam fill (168) produced large quantities of pottery (441 sherds

weighing 5,422g) dated c. 70/120-200 CE, suggesting that this section had infilled by the Middle Roman period 3.1. The assemblage included early BB1 and South-Western Black Burnished ware coarsewares, fragments of South and Central Gaulish samian, a Corfe Mullen flagon and Dressel 20 amphora. Other finds included part of a stone mortar. Ditches 311 and 312 The geophysical survey suggested two separate ditches which were numbered accordingly. However, excavation 28

Enclosures, ditches and middens proved that they were originally one entity, with the central section being filled in during the Later Roman period 4.2 forming a southern entrance to the villa. The two lengths are considered together and had a total length of c. 50m. On the east, ditch 312 merged with the north-south ditch 313 and on the west, 311 merged with north-south ditch 308. A total of seven sections were cut with widths varying from 1.07m to 2.8m and depths of 0.85m to 1.15m. The profiles were moderately steep-sided with rounded/flat bases and displayed evidence for infilling from the late Early Roman to the late Middle Roman periods. There was no evidence for re-cutting. The animal bone deposited in the ditch fills included sheep/goat, cattle, pig, horse, dog and cat. An entire cattle skull came from the top fill (979) of section 978.

were built close to the ditch edges suggesting that the ditches had filled in by this time. Ditches 314 and 1606 were aligned in east-west and north-south directions respectively. Ditch 308 Eleven sections were cut through this 66m stretch of ditch on the western side of the villa. The ditch merged with 314 to the north, had a junction with 311 on the east and had a defined south termination. Two sections had been recut. The profiles were all relatively consistent, being steep sided with gently rounded bases. The widths of the ditch varied from 1.6m to 2.7m and the depths from 1.08m to 1.4m; the recuts were all 2.3m wide and varied from 0.5m to 0.8m in depth. There was a gap of 3.5m between the terminal and ditch 1504, which appeared to be a southern continuation of the enclosure system; this gap may have been an original entrance into the enclosure.

Fig. 3.2/4 Section 333 (T11, ditch 311) Width 1.75m, depth 0.97m with steep sides and a narrow, rounded base. Six fills were recorded including a surface layer (332) of substantially intact tegulae and imbrices probably from an entrance structure, together with BB1 pottery dated c. 180/200-270/300 CE which must be associated with the latest episode of infilling. The second and third fills (334 and 340) comprised dark brown loam with varying amounts of charcoal. Fill 334 contained building material, particularly tegulae. The pottery was overwhelmingly BB1 (250 sherds weighing 2,603g) dating c. 180-300 CE. A small amount of slag (549g) was recorded. The two upper fills had probably been inserted into the slumped fill 340. Lesser amounts of building material and pottery were present, whilst the pottery (49 sherds weighing 460g) dated c. 200-300 CE. Charcoal continued to be a feature of the mid-brown loam fills of layers 342, 347 and 348, with decreasing quantities towards the base; associated pottery gave dates of c. 90-200 CE suggesting initial infilling in the early Middle Roman period 3.1.

At the northern extremity of the ditch, the southern part of the apsidal wall (1042) of Room N12 of the North Range was constructed over the final infill; the pottery dated from c. 43-250 CE. The walls of Rooms N2 and N5 were only 2m from the edge of the ditch, suggesting that infill was complete when this part of the building was constructed. The two-phase West Range was built even closer to the ditch. The first phase building was aligned on the ditch with a 3m gap between the ditch and its west wall. When the range was re-built in the late 3rd century, the gap between the building and the ditch was only 0.5m, suggesting that infilling was completed by this time. The pottery evidence supports this; Early Roman pottery was generally associated with the primary fills, most sherds were Middle Roman, fewer were Late Roman and very few Final Roman. The animal bone assemblages included sheep/goat, pig and cattle including a whole baby pig and two cattle skulls, red deer and dog skulls and a relatively large number of horse bones.

Fig. 3.2/5 Section 358 (T11, ditch 312) Width 1.65m, depth 0.9m with steep sides and a narrow, rounded base. Four layers were recorded and, like section 333 from ditch 311, the two uppermost layers appear to have been placed into a slumped filling. The top fill of charcoal-rich, dark-brown loam (359) contained large chunks of building flint. Pottery (13 sherds weighing 69g) dated this deposit to c. 90-200 CE. The second layer (360) was mid-brown loam with lesser amounts of charcoal and building material; the pottery (126 sherds weighing 820g) was dated c. 120-200 CE. The pottery (64 sherds weighing 626g) from the third fill (361) of dark-brown loam dated c. 160-200 CE. There was no dating material from the base fill. The tight date range of the pottery, which was BB1 except for five sherds of South-Western Black Burnished ware, suggests that this ditch was also infilled by the early Middle Roman period 3.1.

Fig. 3.3/1 Section 165 (T10) Width 2.22m, depth 1.55m with steep sides and a narrow, flat base. Located immediately to the west of Room W2 in the West Range, the section comprised ten discrete contexts. The upper 0.75m consisted of dark-brown loam (fills 166, 189 and 204) with three deposits (188, 196 and 231) of knapped gravel nodules in loam. On the eastern side of the ditch, three courses of gravel (234) were set on a clay base; this is likely to have been an attempt to consolidate the edge of the ditch and was probably associated with the construction of the second phase of the West Range. The pottery from the three upper fills dated to c. 200-350 CE, although the flint-rich fill 196 produced six sherds (40g) of post-Roman/early Saxon pottery dating c. 450-650 CE, suggesting that layers 186, 189 and 166 were midden deposits which were brought in during Period 6. Fill 204 was artefact-rich and was dated by the 434 (4,336g) sherds of pottery to c. 120-270 CE. BB1 forms included

The inner enclosure (ditches 308, 314 and 1605) Ditch 308 formed the western boundary of this ditch complex; the later phases of the North and West Ranges 29

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) jars, bowls, dishes, beakers, flagons and colanders. Small sherds of samian and Central Gaulish black colour-coated wares were also present. Copper alloy finds included a ring (Fig. 14.5/7), a leather-working needle (Fig. 14.9/13) and a hair pin (Fig. 14.7/1). A large fragment from an iron wool-comb (Fig. 15.1/8) was recorded together with four ceramic spindle whorls (Fig. 13.19/1-4). In addition, very large quantities of animal bone were recorded. Fill 213 yielded 106 sherds of BB1 pottery (1,077g) dating c. 120230 CE, suggesting that this fill accumulated during the Middle Roman period. There were no dateable finds in the two lowest layers 252 and 271.

Ditch 314 Three sections were cut through this 80m length of ditch on the northern part of the site. The ditch merged with 1605 on the west and 308 on the east and was parallel to ditch 309 which lay 25m to the north. The profiles of the sections were moderately steep, and the bases were flat or rounded. The width of the ditch varied from 1.35m to 2.01m and the depths from 0.9m to 1.22m. Pottery was only recovered from one section (1139) giving spot dates of c. 70-150 CE. The paucity of finds suggests minimal activity along the length of this ditch.

Fig. 3.3. (1-3) Sections through enclosures 308, 314 and 1605

30

Enclosures, ditches and middens of the material was located near the base of the ditch, implying probable infilling before c. 100 CE. Of note was an assemblage of glass vessels including conical jugs, bottles and a bowl (see Chapter 13).

Fig. 3.3/2 Section 1139 (no trench allocated) Width 2.15m, depth 1.04m with moderately steep sides and a narrow, rounded base. A single mid-brown loam fill (1140) was recorded, suggesting a single infill event; 53 sherds of pottery (451g) dated c. 70-150 CE, were early/ mid-Roman in date and included a greyware copy of a samian bowl (Fig. 11.8/3).

Fig. 3.3/3 Section 1620 (T71) Width 1.95m, depth 1m with moderately steep sides and a wide, flattish base. Six fills were recorded. In the centre, fill 1621contained much knapped gravel, which may have originated from building activities; there was no pottery in this layer but five fragments from a small globular glass jar dated to the late 1st/early 2nd century. On either side of this, fill 1669 comprised a compacted hoggin-type clay loam, initially thought to be natural but the presence of charcoal and a fragment of quern indicated fill. On the lower edge of this layer, fill 1678, a charcoal-rich, sticky clay, contained BB1 (45 sherds weighing 448g), large fragments from a Gauloise 5 amphora (5 sherds weighing 994g) dating c. 50-100 CE (Fig. 3.4). The clay/loam fill 1683 extended across the width of the section. A large mortaria base sherd was recorded; presently the production site of its F20 fabric is unknown. Underlying this, fill 1684 was similar to 1683 but more clay and charcoal were noted. The 88 sherds of pottery (1,150g) included early BB1 fabrics and South Gaulish samian with a date range c. 45-70/100 CE.

Ditch 1605 Seventeen sections were cut through this 80m length of north-south ditch on the western part of the site. The ditch merged with 314 in the north and was parallel to, and 25m from 1604 on the west. The profiles of the sections were moderately steep, and the bases varied from flat to rounded. The width of the ditch varied from 1.6m to 2.5m and the depths from 0.7m to 1.15m; there were no re-cuts. Eleven consecutive sections (totalling 10.8m in length) were cut in the central part of the ditch in Trenches 70 and 71 where initial work had suggested the presence of substantial amounts of early material. This was confirmed by excavation and in total 1,390 sherds of pottery weighing 13,204g was retrieved with dates of c. 43-100 CE. By comparison, the remaining six cuts produced only 176 sherds (weighing 678g) of similarly dated pottery. Much

Fig. 3.4. Ditch 1605 section 1620 during excavation. Lilian Ladle with large fragment of Gauloise 5 amphora, view south

31

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) enclosure system. The infilling sequence implies that the ditch was completely filled between c. 250 and 300 CE. There were no recuts. A fragment of Lyon ware pottery came from the lower fill of section 1396 and has possible implications regarding a military presence as this pottery type is pre-Flavian and often associated with the army. Two infant bones were recorded with pottery giving a date of c. 120-220 CE.

The primary silty loam fill (1694) contained 57 sherds of pottery (241g), including fragments from BB1 amphora and jars, and Corfe Mullen flagon sherds dating c. 43-70 CE. Southern outliers (Ditches 1504 and 310) Two ditches on the southern side of the site were also part of the main enclosure system set out in the earliest Roman period 2.1. The full extent of these ditches could not be followed as both continued out of the excavation area and into the water meadow to the south.

Fig. 3.5/2 Section 1368 (T52) Width 1.65m, depth 0.76m with moderately steep sides and a narrow, flat base. Two fills were recorded, the upper (1369) a mid-brown loam contained building debris and pottery dating c. 200-270 CE. The charcoal-rich dark brown loam lower fill (1387) produced fewer finds, although the pottery was significant, with jars providing a date range c. 50 BCE to 150 CE. Five sherds of Exeter Fortress ware were recorded.

Ditch 1504 Three sections were cut through the available 7.6m length of north-south ditch which was on the same alignment as 308 and separated from it by a gap of just over 3.5m. The ditch continued southwards beyond the field boundary. The profiles of each section varied from moderately steep to gently sloping, and the bases varied from flat to rounded. The width of the ditch varied from 1.4m to 2.1m and the depths from 0.7m to 0.9m. The terminal 1517 had a similar profile to the terminal of 308, the next section 1513 had evidence for re-cutting and the third section 1505 was similar in profile to the recut, being wider and shallower. A dump of oyster shells (64 weighing 3,505g) was placed in the terminal 1517.

Structures associated with a south entrance to the villa Features associated with this are illustrated on Fig. 3.8/1. Two large pits (402 and 619) about 2m apart, were located south of ditches 311/312 in Trench 11 and were probably associated with an elaborate southern entrance to the villa. Pit 402 measured 1.55m by 1.43m with a depth of 0.83m and was half-sectioned (Figs 3.7 and 3.8/2). The similarly sized pit 619 was not excavated. The pit sides sloped steeply, and the base was flat. The fill of dark brown loam (403) was packed with large to medium sized flints which lessened in density towards the base on which lay two flat pieces of limestone (Fig. 3.7). There was no dating material; however similar pits were recorded in the Aisled Hall (Chapter 6). There, flint-packed pits supported stone plinths and timber uprights forming the structural basis for roof supports. It is likely that the similar flint-packed pits here supported structural uprights related to an elaborate entrance. Large numbers of tegulae and limestone pieces were recorded on the surface of ditch 311 and are likely to have derived from such a structure.

Fig. 3.5/1 Section 1513 (T52) Width 2.08m, depth 0.8m; recut width 1.95m, depth 0.55m. Original cut with moderately steep sides and a wide, flattish base, recut with gently sloping sides and wide undulating base. Initially, a single fill 1462 was recorded, however weathering of the section revealed that the lower part of the fill (1462a) was the original infilling. The fills differed in that the upper, re-cut part contained building flint, limestone building stone and stone and ceramic roof tiles. This material was absent from the lower fill. A horse skull had been placed on the base of the ditch (Fig. 3.6) and associated with this was a deposit of pottery (95 sherds weighing 1502g) including Black Burnished and greyware, storage jars, South Gaulish samian and Corfe Mullen flagons with a date range of c. 43-70 CE. Infant remains comprising 20% of the skeleton were recorded in this section.

Two small scoops (521 and 523) were located to the east of 619 (Fig. 3.8/3-4) and may have been associated with the entrance. There were no finds from these features. Conclusions

Ditch 310

This rectilinear ditch system was constructed in the earliest Roman period 2.1, most likely between c. 50-70 CE. The precise regularity suggests that surveyors may have been used to plan and execute the layout. It is not inconceivable that there may have been a military involvement, which is strongly suggested by the presence of South Gaulish samian, Lyon ware, Corfe Mullen fineware and early Black Burnished wares.

Six sections were cut through this 21m length of eastwest ditch which was aligned at right angles to ditch 1504. Excavation, however, was unable to detect the relationship because the fills were similar and merged. The ditch is parallel to ditch 311 and separated from it by a gap of approximately 20m. The profiles of each section were generally moderately steep, and the bases varied from flat to rounded. The width of the ditch varied from 1.3m to 1.75m and the depths from 0.7m to 0.9m. The ditch profile suggests that it was cut at the same time as the rest of the

Geophysical survey suggests that the outer and inner enclosures 1604 and 1605 peter out on the southern side 32

Enclosures, ditches and middens

Fig. 3.5. (1-2) Sections through southern ditches 1504 and 310

311/312 and 313 were re-cut between 200 and 300 CE, with final fills laid down in the late 4th/early 5th century.

of Eleven Acres; but it is possible that they continue southwards into the water meadows. Currently, there is no evidence for the ditches joining or creating an entrance but there must have been such a feature. In the early 4th century, the top surface of ditches 311 and 312 was consolidated to create a gateway to the villa complex. Two deep, flint-packed pits probably supported heavy pillars and large amounts of limestone building stone and ceramic roof tile fragments retrieved in the near-vicinity suggest an imposing structure framing this entrance.

Other ditch systems Four smaller ditch systems were identified and are discussed by period. Ditches 1641/1644 The westernmost extent of a probably rectangular enclosure was recorded in the extreme south-east corner of Eleven Acres (Fig. 3.9/1); the geophysical survey suggests that this system may continue into the water meadow to the southeast. The 38m long north-south section of ditch 1641 had an average width of 1.6m and average depth of 0.72m and is parallel to ditch 1605. The northern section 1644 was 10m in length with an average width of 1.14m and average depth of 0.43m. In total, 15 sections were investigated and although ditch 1641 was wider and deeper than ditch 1644, they appeared to be part of the same system. Pottery evidence implies that this feature was dug between c. 120 and 180 CE and cut the Early Roman pit 1735.

Ditches 308, 311/312, 313 and part of 309 enclosed the villa buildings. The Ancillary building, grain dryer and occupation area T82 were located directly to the east of ditch 1605; the geophysics plot suggests that there may have been other features between these and ditch 308 (time and manpower precluded investigation here). No Roman features were recorded between ditches 1604 and 1605. Pottery evidence suggests that the ditch construction was part of the first phase of activity on the site. Subsequently the ditches were used as convenient refuse dumps for occupants of the villa over several centuries. Ditches

33

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 3.6. Ditch 1504 section 1513 with horse skull placed on the on the base of the ditch, view west

Fig. 3.7. Flint packed foundation Pit 402, half section, view west

34

Enclosures, ditches and middens

Fig. 3.8. Features associated with the southern entrance of the villa. Plan of Trench 11 and sections of pit 402, scoops 521 and 523 (1-4)

Fig. 3.9/2 Section 1786 (T 87, ditch 1641)

a South West Black Burnished ware jar dated c. 60-100 CE. The primary fill (1679) yielded five undated pottery sherds.

Width 1.6m, depth 0.65m with moderately steep sides and a gently rounded base. The single fine, dark loam fill (1787) had charcoal fragments near the base. Pottery was only located near the surface (nine sherds weighing 56g) and included Black Burnished ware and a chunk of Baetican DR 20 amphora dating c. 43-200 CE.

Ditches 1459/1510 These parallel ditches aligned north-south are located on the south-west corner of Limepits Field and are considered together. They are part of a complex of ditches in what was designated the ‘Southern Site’ (Fig. 3.10/1). They cut and are later than the Early Roman ditch 310. The ditches are parallel to the earlier ditch 1504. Ditch 1510 continued southwards beyond the field boundary. Three sections were cut through the available length of 7.5m. The sides were moderately sloping, the base was rounded and well-defined; the cuts had an average width

Fig. 3.9/3 Section 1664 (T73, ditch 1644) Recut width 1.32m, depth 0.23m with shallow sloping sides and a flat base over the original cut, which was must have been approximately 0.8m wide and 0.32m deep. Despite the re-cut, the fills are a similar mid-brown loam, the upper (1665) containing more gravel and sherds from 35

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 3.9. Plan of ditches 1641/1644, pit 1735 and grain dryer 1758 and ditch sections of 1641 and 1644 (1-3)

of 0.6m and average depth of 0.5m. Ditch 1459 was dug immediately west, and in part cut ditch 1510. Three sections were excavated having average widths of 1.65m and average depths of 0.76m and were notable for the amount of mortar dumped in the upper fill. Five infant bones were recorded. Pottery suggests that infilling started in the early 3rd century and that this process continued into the early 5th century.

Fig. 3.10/2 Sections 1452 (T52, ditch 1459) and 1511 (T52, ditch 1510) The section incorporated the cuts of both ditches. The two upper fills (1435 and 1437) had accumulated over both ditches. Dark brown loam layer 1435 contained numerous large building flints and pottery (29 sherds weighing 291g), including New Forest indented beakers dating c. 260-370 CE. Animal bone and a quern fragment were recorded. The pottery from the chalky loam of 1437 dated c. 37036

Enclosures, ditches and middens

Fig. 3.10. Plan of southern site and sections through ditches 1459/1510 (1-2)

all mid-brown loam. Pottery sherds were sparse and small, and all were retrieved from near the section surfaces. After considering similarly filled features on other parts of the site, it is likely that clay from naturally occurring fissures may have been extracted and the resulting voids filled with soil from nearby areas. It is debateable whether the gully had any functional use.

430 CE; this included 30 large sherds from a South East Dorset Orange Wiped ware storage jar. A shale bracelet fragment was also noted. A single fill of dark brown sticky loam (1450) was present in the section of ditch 1510. A horse skull had been placed on the base of the ditch (Fig. 3.11) and associated pottery dated this deposit to c. 290350 CE. Three fills from ditch 1452 were recorded (1445, 1456 and 1457); finds were only located in the clay and charcoal-rich fill 1445 which was dated c. 180-280 CE by BB1 everted rimmed jars.

Fig. 3.12/1 Section 1322 (T52) Width 0.8m, depth 0.48m with steep, irregular sides and a pointed base and cut into natural chalk. The fill (1323) contained many gravel nodules, some of which were embedded into the side of the cut. There were no finds.

Gullies 372 and 384 Two narrow gully-type features were recorded on the Southern Site (Fig. 3.10/1). Gully 384 was parallel to ditch 1504 and gully 372 was aligned north-east/southwest and located north of ditches 1505 and 1459.

Gully 384 Seven sections were cut through this 13.5m long linear, although it was not possible to determine its full extent. The profile was generally a wide ‘U’-shape with an average width of 0.71m and depth of 0.31m. Small numbers of finds came from all sections and included ceramic and limestone roof tile fragments and pottery (29 sherds weighing 127g)

Gully 372 Five sections were cut through this 5.5m long feature which had a consistent ‘V’-shaped profile with an average width of 0.74m and average depth of 0.45m. The single fills were 37

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 3.11. Section through ditches 1452 and 1510 with horse skull on the base of ditch 1510, view north.

material is likely to have derived from nearby ironworking activities. Large quantities of building material, including building stone fragments, sandstone roof tile, box flue tile and tesserae were dumped on the surface of the section fills. The sandstone tile suggests that this material came from the North Range. Ten fragments of infant bone were recovered from the cleaned surface of the ditch. In total 510 sherds of pottery (weighing 4,653g) dating c. 270-400+ CE suggests in-filling during the Late Roman period. Two coins, an antoninianus of Claudius II (268-270) and a minim of Valentinian (268-270), support this dating.

with broad date ranges of c. 43-100 to 240-450+ CE; this included South Gaulish samian, New Forest greyware and two sub-Roman sherds. The function of this shallow linear is unclear; it may have been associated with possible features outside the excavated area. Fig. 3.12/2 Section 1330 (T52) Width 0.75m, depth 0.22m with moderately sloping sides and a flat base cut into clay/loam. Building material from the compacted clay/loam fill 1331 included roof tiles and a piece of floor tile. In addition, the base of a ceramic crucible (Fig. 13.18/1) was recorded. The fill was above a spread of gravel (1324) which may have been the western extent of midden 1528 (see below).

Fig. 3.12/3 Sections 1274 (T50, ditch 1260) and 1295 (T47, ditch 1305) This section extended 3.55m across both ditches, as did the top layer (1275) of loosely packed gravels. There was little surface definition between the two sections, but the fills differed markedly. Fill 1276 was a consolidated layer of compacted flints which appeared to have been laid across ditch 1260 and extended into the fill 1296 of ditch 1305, suggesting that this deposit was laid down later. The lowest fill 1281was a very dark brown, charcoal-rich loam. Fill 1296 of ditch 1305 was characterised by the presence of many large flints. A large fragment of hearth bottom was recorded. Associated pottery dated all fills to c. 300-430 CE.

Ditches 1260/1305 The geophysical survey suggested a single north-south aligned linear feature ‘hooked’ at the north end, located outside the south-east corner of the villa complex. Excavation revealed a 43m long north-south shallow ditch 1260 which turned east (designated 1305 and 7.5m in length). Seven sections were cut through ditch 1260 with average widths of 0.85m and average depths of 0.3m, although it became deeper towards the south. The shorter ditch 1305 was parallel to 1260 with a gap of approximately 1m to 1.5m between them. Three sections were cut through this showing it was wider (average width 2m) and slightly deeper (average depth 0.33m) than 1260. Both ditches were notable for the presence of ironwork and slag; in total 159 iron objects, 8,692g of slag and fired clay were recorded. This

Fig. 3.12/4 Section 1286 (T50, ditch 1260) Width 1.2m, depth 0.47m with moderately sloping sides and a rounded base. The section was located under 38

Enclosures, ditches and middens

Fig. 3.12. Sections through gullies 372 and 384, ditches 1260/1305 and Eastern Enclosure ditches 1262, 1412, 1410 and 1411(1-8)

Ditch 1262

clearance 1266. The single fill 1287 comprised dark brown charcoal-rich loam which contained significant quantities of slag and ironwork. Pottery (94 sherds weighing 627g) dating c. 320-350 CE comprised fragments from BB1 jars, bowls and dishes and New Forest beakers and a bowl.

Approximately 30m of the westernmost ditch of the enclosure was revealed. Five sections were cut having an average width of 1.26m and depth of 0.4m. Fig. 3.12/5 Section 1466 (T60)

Eastern enclosure

Width 1.1m, depth 0.4m with moderately sloping sides and a rounded base with a single fill 1467. The largest number of pottery sherds (193 weighing 1,707g) was retrieved from this section and dated c. 43-230 CE; the fresh condition suggests that they were dumped shortly after breaking.

Located on the south-eastern extremity of the site and set out approximately north-south/east-west, this group of ditches formed a rectangular enclosure measuring approximately 33m by 29m (Fig. 3.1). A gap in the northwest corner may have been an entrance or gateway. The outer ditches comprised 1262 on the west, 1412 on the north and 1410 on the east. There was no ditch on the south, but this could lie under the wide hedge of the field boundary. The enclosure was divided into two unequal parts by east-west ditch 1411, parallel to 1412, separated from it by about 9m and intersecting with 1262 and 1410. Sixteen sections were cut through this ditch system; the profiles were generally consistent with moderately steep sloping sides and flat/gently rounded bases. The average width was 1.34m and average depth 0.65m. Large amounts of building debris (limestone and ceramic roof tiles) were present over ditch 1412 and in its upper fills, indicating the dumping of such material well outside the villa complex. Deposits of animal bone were abundant. Pottery from this enclosure suggests that it started filling in c. 270 CE and material was still being added at the end of the 4th century. Three coins, two radiates of Carinus (282-285) and an antoninianus of Carausius (282-285) fit this dating. Representative sections from each ditch are described below.

Ditch 1412 This 20m long ditch formed the northern extent of the enclosure and it is possible that an entrance was located on the extreme western side. Five sections were excavated with an average width of 1.43m and average depth of 0.67m. Fig. 3.12/6 Section 1425 (T60) Width 1.4m, depth 0.6m with moderately sloping sides and a narrow flat base with two fills. The upper fill (1426) comprised charcoal-flecked dark brown loam dated by pottery (174 sherds weighing 1347g) to c. 210-400 CE and by a coin of Carausius (286-293). In addition, there was a fragment of copper alloy Hod Hill type brooch dated to the 1st century CE. The upper and lower elements of a horse skull had been placed on the base of this layer (Fig. 3.13). The lower fill (1439) was dated c. 120-370 CE by pottery (41 sherds weighing 312g). 39

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Ditch 1410

The midden deposits

The easternmost ditch was visible on the geophysics plot for about 25m; two sections were excavated with an average width of 1.75m and depth of 0.98m.

A number of features associated with rubbish disposal were located to the west and south of the villa. Large possible quarry pits and natural hollows were utilised, as were deliberately cut pits associated with the buildings, which were filled in when they no longer fulfilled their original use. These latter features are considered in the relevant Range chapters.

Fig. 3.12/7 Section 1473 (T61) Width 1.7m, depth 0.9m with a very steep side to the east, a moderate slope to the west and a narrow almost pointed base with two fills. The upper (1474) contained numbers of flints, some of which appeared to have been tipped in deliberately. Large bone elements were present, and pottery (20 sherds weighing 200g) dated c. 270-330 CE. The only find in the lower silty fill 1475 was a large bovine leg bone.

The first midden area (315) was encountered in 2013 when its function and use was not fully recognised. Subsequently, scrutiny of the geophysical survey (Fig. 1.5) suggested that a number of areas to the east of the villa may also have been used for the disposal of rubbish. Five of these were aligned in a north-south direction approximately 10m east of the enclosure ditch 313, and three of them (315, 1355 and 1529) were sampled. The northern-most in this alignment was outside the excavation area as were a further three, sited to the north in the then arable field. A midden to the south of the villa (1528) was completely excavated. The middens are located on Fig. 3.1 and are characterised by their artefactrich fills which include pottery, glass, coins, metalwork, metalworking debris and animal bone, as well as building material comprising limestone, sandstone, slate, ceramic roof tiles, box flue tiles, tesserae and building flint which no doubt originated from villa alterations. The site mechanical excavator was used to locate the middens and to excavate east-west trenches across them.

Ditch 1411 This 28m long east-west ditch divided the enclosure into two unequal parts. The geophysical survey suggested that there was no break in this ditch, and therefore the two small plots were entirely separate. Four sections were cut; average width was 1.25m and depth 0.6m. Fig. 3.12/8 Section 1429 (T60) Width 1.35m, depth 0.64m with moderately steep sloping sides and a rounded base. Two fills were recorded, the upper (1430), filled with abundant small gravels, produced 45 sherds of pottery weighing 335g and dated c. 270370 CE. The lower fill (1433) was a lighter brown with numerous larger gravel nodules but no finds.

Midden 315 T2 This midden, located about 22m east of the North Range, was excavated in 2012 and its function was not recognised at the time. Its estimated extent is c. 9m wide and 7m long, and its uneven sides suggest that this may have been a natural hollow. An east-west section (53) measuring 1.8m long by 1.5m wide, with a maximum depth of 0.7m was cut across the western side of the feature (Fig. 3.14). The eastern side was apparent but not excavated. The feature was cut into chalk, with the sides sloping gently down to a very uneven base. Finds were located throughout the dark brown loam fill (48) and much building debris was noted on the surface. Infilling occurred in the Final Roman period 5. The pottery assemblage (116 sherds weighing 1,869g) was dated c. 350-430 CE.

Conclusions The minor ditch systems were established after the large enclosures were in place and were infilling and were most probably dug to fulfil specific purposes. The enclosure 1641/1644 located south of the Ancillary Building, may have had a function related to this structure. Unfortunately, most of its putative extent lay outside the excavation area. Ditches 1459/1510 were constructed over two phases. The short ditch 1459 was earlier and was replaced by 1510. This latter ditch may have replaced the Period 2.1 ditch 1504. These two ditches were ultimately covered by midden deposit 1528 in periods 4.2-5. The pair of ditches 1260/1305 were filled with material associated with ironworking; the focus of this activity has not been located. The Eastern Enclosure 1262/1410/1411/1412 was contemporary and the small size may point to its use as a paddock.

Midden 1529 T24, T41 and T46 (Fig. 3.15/1) This midden cut into natural chalk and may have been another quarry. It measured approximately 11m by 15m and was located 15m to the east of the East Range. Three sections (470 and 1026 in T41 (Fig. 3.15/3), 1254 in T46 (Fig. 3.15/4) were dug through the northern half of this feature. The profile sloped gently on the west and east, and the base appeared to deepen towards the north, the depths varying from 0.33m to 1.7m. A number of fills with discrete layers were apparent, particularly in cut 1026 where tip layers were evident. Finds were abundant and included 322 sherds of pottery (12,1545g) with date

The construction dates of the ditches cannot be recovered, nor can the time-period of their use when they were scoured and kept clean. Like all the Roman ditches at Druce, they were ultimately used as convenient dumping places for household and low-grade industrial waste, as well as for building debris.

40

Enclosures, ditches and middens

Fig. 3.13. Ditch 1412 section 1425 with upper and lower elements of a horse skull on the ditch base, view east

Fig. 3.14. Midden 315, section 53 with uneven base, limestone slabs, excavated building flint (behind section), view north

41

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 3.15. Midden 1529 plan and sections (1-4) 470, 126 and 1254. Midden 1355, section 1362 (5)

east. Finds were less numerous; the 46 sherds of pottery (557g) dated c. 200-430 CE, were dominated by late South East Dorset Orange Wiped ware storage jars. Under this, a discrete deposit (642) of charcoal-rich, very dark brown loam was dated c. 260-400 CE by 16 sherds of pottery (83g).

ranges of c. 200 to 430 CE. Three coins, an antoninianus of Victorinus (268-270), an antoninianus of Claudius II (270) and a nummus of the House of Valentinian (364378), strongly suggest that infilling continued into Period 5. Building material was abundant. Large amounts of animal bone were present but perhaps most significant was the presence of human infant remains.

Midden 1355 T59

Fig. 3.15/2 Section 470 (T41)

This midden on the extreme eastern edge of the site was difficult to locate and only a 1.1m by 1m section (1362) was available for excavation (Fig. 3.15/5). The geophysical survey suggests that it was approximately the same size as middens 315 and 1529. The moderately steep western profile, flat base, and depth of 0.6m hint at its large size. The finds were of a similar nature to those found in the other two middens, the pottery dates ranging from c. 220-330 CE. One of the two coins was a nummus of Constans dated 337347, the other too worn for identification. The finds suggest that this is the earliest midden in the sequence.

A 3.35m section was excavated through the central western part of the midden. The depth of the undulating base varied from 0.3m to 0.5m. Four fills were recorded. To the west, fill 471 produced 127 sherds of pottery (weighing 1,629g) with a date range c. 150-430 CE. A broken millstone (Fig. 3.16) had been placed on the base of the layer. The remains of a tiny infant (673) were laid within a depression in the base of 471; 60% of the skeleton was present and further infant bones were recorded from fill 471 above. Fill 471 was later than the mid-brown clay-rich fill (511) to the 42

Enclosures, ditches and middens

Fig. 3.16. Midden 1529, section 470 with broken millstone and box flue tiles, view west

Midden 1528 T14 and T52 (Fig. 3.17)

was visible after 0.4m of topsoil was removed. Feature 1324, a thin linear spread of building flint 4.5m long and 0.4m wide overlay the midden on the eastern side. The midden was excavated in four layers which generally comprised very dark brown loam; this method ensured that the position of artefactual material was accurately recorded. The finds are recorded on Table 3.2, and Table 3.3 lists the coins. A series of plans (Fig. 3.18) illustrate the infilling sequence.

Located on the Southern Site this midden was completely excavated and, as a result, there is a full understanding of the sequence and nature of the deposits. The midden was subcircular in shape measuring approximately 10.5m east-west and 8.5m north-south, the sides sloped to a gently undulating base and the maximum depth of the feature was 0.55m. It

Fig. 3.17. Trench 52 with initial clear back of Midden 1528 in progress, view east

43

44

Layer

Context

Coins (no)

Cu/Pb(no)

Fe (no)

slag (wt)

pot (no)

pot (wt)

AB (wt)

St Obj (no)

FC (wt)

spot date

Period

layer 1

1278

9

5

38

346

172

1336

73

2

18

200-400

layer 1

1364

1

1

layer 1

1314

1

3

40

layer 2

1365

2

4

5

layer 2

1335

1

2

12

layer 2

1373

layer 2

5 5 5 4.2 5 4.2 5 4.2 4.1 5 5 4.1 5 4.1 5 5 4.1 4.2 5

none 178

1715

61

394

119

1256

38

7

11

70

14

1339

5

26

230

44

layer 2

1340

1

16

162

layer 3

1380

5

9

269

12

layer 3

1346

187

1874

1232

layer 3

1359

9

73

49

layer 3

1370

149

3574

237

layer 3

1400

9

172

569

layer 3

1379

5

1510

20

layer 4

1381

layer 4

1405

layer 4

1349

layer 4

1404

29

layer 4

1422

2

Totals

1

1

1

16

39 1

7

3

27

150

164

793

90

51

1 1

564

16 7

43-300 300/50-43 325-400

4

210-270/3 345-400 64

95

1252

950

1

1

33

14

96

997

403

1

17

260

10

44

254

6

93

1211

8

1296

16609

3716

1961

240-400

250-350

35

281

210-370

370-430

30

21

62

270-400

1

369

200-430

43-300 270-430

1141

370-430 120-350

4

260-350 820

21

2685

370-430

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Table 3.2. Midden 1528: total number of finds from layers, artefacts above, building materials below (abbreviations as Table 3.1)

Table 3.2. Continued

45

Context

LBS (wt)

LRT (wt)

SRT (wt)

layer 1

1278

3359

3750

15

layer 1

1364

layer.1

1314

layer 2

1365

layer 2

1335

10180

layer 2

1373

57

layer 2

1339

44

layer 2

1340

layer 3

1380

layer 3

1346

layer 3

1359

layer 3

1370

layer 3

1400

345-400

layer 3

1379

43-300

layer 4

1381

98152

14579

layer 4

1405

4765

15297

layer 4

1349

layer 4

1404

layer 4

1422

Totals

295

slate (wt)

2368 915 2162

2 204

3025

23903

698

26

323 3996

Tess (no)

spot date

Period

5 5 5 4.2 5 4.2 5 4.2 4.1 5 5 4.1 5 5 5 5 4.1 4.2 5

1830

5

none

6413

6

200-430

5333

5

210-370

833

270-400 1

240-400

30845

370-430

171

250-350

669

43-300

15142

166

2

2

34

8

6049

205

Mort (wt)

200-400

395 322

PP (wt)

32

118

386

362

FT (wt)

14014

5997

54

7816

CRT (wt)

325-400

4

3949

160

8612

99

300/50-43

210-270/3

6

53

270-430 370-430

101

120-350

82

158

260-350

3886

3004

954

128916

73790

1498

28

100750

256 1258

61

10

343

370-430

Enclosures, ditches and middens

Layer

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) or layers, layer 4 being the first and layer 1 the last and are discussed by layer.

Table 3.3. Coins from midden 1528 layers Context Layer SF

Type

Ruler

Date CE

1278

1

706 nummus Constantine

323-4

1278

1

707 sestertius illegible

43-100

1278

1

708 as

Claudius I

41-54

1278

1

711 short X

Henry III

1216-1242

1278

1

714 as

Antoninus Pius 138-161

1278

1

717 minim

Ho. of Theodosius

388-395

1278

1

721 coin

illegible

4C

1278

1

722 coin

illegible

4C

1278

1

740 radiate

Carinus

282-285

1314

1

715 minim

Gratian

367-375

1365

2

755 minim

Gratian

367-383

1365

2

752 minim

Valens

364-367

1335

2

775 nummus Julian

355-363

1346

3

761 coin

illegible

4C

1370

3

750 denarius

Severus Alexander

226

1381

4

777 nummus illegible

Layer 4 (Fig. 3.18/1) The initial 0.1m deep layer 4 consisted of five contexts (1349, 1381, 1404, 1405 and 1422). Context 1422 on the south edge of the midden (Fig. 3.21) comprised fired clay and fire-reddened limestone, together with BB1 jars, a beaker and a South East Dorset Orange Wiped ware storage jar which dates the assemblage to after 370 CE. Very large quantities of building stone and limestone roof tile were associated with contexts 1381 and 1405. Layer 3 (Fig. 3.18/2) Six contexts made up layer 3 with depth varying from 0.1m to 0.2m. The area was initially divided into two (1346 and 1370) and, as the work progressed, two further areas (1359 and 1380) were added. The fill of dark brown loam continued, but the soil was lighter in 1380 on the south-west edge; there were also fewer finds. Area 1359 was also defined by fewer finds, though one of these was part of a copper alloy bracelet (Fig. 14.4/4). Two deposits containing only pottery and animal bone were noted (1379 and 1400). The pottery from layer 3 is earlier that that generally found in the midden and broadly dates c. 210-300 CE. Two pots are notable: a Late Iron Age style bowl with bead rim, straight sides curving to a foot-ring (see Fig. 11.9/2) and large fresh sherds from a Dressel 20 amphora. Contexts 1346 and 1370 were artefact rich and finds included a denarius of Severus Alexander (226 CE), a complete copper alloy spoon (Fig. 14.8/1), quern/ millstone fragments, and a broken whetstone (SF 762). The pottery assemblage was dominated by Black Burnished ware (461 sherds weighing 6,018g), and fragments from bottles, beakers, jugs and bowls from the New Forest and

4C

The midden accumulated in a probable natural hollow. Early Roman ditches 310 and 1504 had infilled as had Middle Roman ditch 1459; this latter was replaced by ditch 1510 in the earliest Late Roman period, and it was at this time that the first deposits were laid down, possibly after a decision was taken to level-up the ground in this area. Most of the infill comprised dark brown loam with discrete areas of dumping. Post-excavation scrutiny of the records allowed a sequence of infill events to be determined. Nineteen contexts were allocated to four filling sequences

Fig. 3.18 Midden 1528 infill sequence layers 1-4

46

Enclosures, ditches and middens

Fig. 3.19 Midden 1528 context 1422, layer 4: dump of fired clay, pottery and burnt stone, view west

Fig. 3.20. Midden 1528, layer 2 ceramic roof tile deposit 1339, view north

47

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Oxfordshire potteries. The earliest vessel was a Corfe Mullen flask dating to c. 43-70 CE. Large amounts of animal bone were also recorded. The pottery dates this layer to the Middle Roman/early Late Roman periods.

episodes, but otherwise the material had probably been stored in rubbish heaps and then moved to this area. The presence of early material in what is generally a Late and Final Roman deposition may suggest disturbance of early phase material in the later period. Episodes of dumping large quantities of rubbish outside the villa in the Late and Final Roman periods may have been the result of the major enclosure ditches closest to the building being filled to capacity, necessitating the opening of new areas for rubbish disposal.

Layer 2 (Fig. 3.18/3) This 0.1m to 0.2m deep layer comprised five contexts, two of which (1339 and 1340) consisted of dumps of re-deposited chalk, and ceramic and limestone roof tiles which possibly derived from a re-roofing episode (Fig. 3.20); associated pottery suggests a Late Roman date. The three other layer 2 contexts (1335, 1365 and 1373) were located on the southern side of the midden and all comprised dark brown loam. Finds included three coins, minims of Gratian (367-383) and Valens (364-367) and a nummus of Julian (355-363). Copper alloy items included the base of a seal box (Fig. 14.9/16). A lead rod (Fig. 14.19/2) may have military parallels. Local Black Burnished wares dominated the pottery assemblages, an Oxfordshire whiteware mortarium and a sherd from a Gauloise 4 wine amphora were suggest dates between c. 210 CE to the late 4th century for the infill material. Layer 1 (Fig. 3.18/4) This undulating layer varied in depth from 0.1m to 0.4m and comprised three contexts; the initial clearback 1278 which was extended to encompass 1364 in the north-west and context 1314 which was visible on the stripped surface as a spread of building rubble. Figure 3.17 shows the site at the beginning of the excavation of the area. The soil from this layer was a mid-brown loam. There were few finds from 1364. The major clearance 1278 yielded nine coins including a cut short cross silver penny of Henry III dated to 1216-1242. The Roman coins span the entire Roman period from an as of Claudius (41-54) to a coin of the House of Theodosius (388-395). There was also an illegible sestertius (43-100), an as of Antoninus Pius (138-161), a radiate of Carinus (282-285), a nummus of Constantine (323-4), a minim of the House of Theodosius (388-395) and two illegible 4th century coins. A further coin, a minim of Gratian (367-375) came from context 1314. Copper alloy items included two brooch fragments (Figs 14.2/6-7) dating to the mid-1st/2nd century. The lead items included a short rod with two protuberances (Fig. 14.19/1). Building material in the form of limestone chunks, limestone roof tiles and ceramic roof tiles was well-represented. The pottery was dominated by Black Burnished ware with forms dating from c. 200- 430 CE. Mortaria from the Oxfordshire and New Forest potteries was recorded. Conclusion The midden material is likely to have derived from activities centred on the villa and was mainly made up of household rubbish and building material, probably resulting from alterations and repairs to the villa itself. Specific deposits in the midden relate to discrete dumping 48

4 Early Roman activity outside the villa complex An early Roman presence (Periods 2.1-2.2) confirmed by pottery and coins appears to be concentrated on the western side of the site in Eleven Acres Field, with only two outlying features to the east in Lower Limepits Field (see Figs 1.8 and 1.9). It is likely that further features of this date lie to the west of enclosure 308 and within the north-south copse which divides Limepits Field from Eleven Acres, but it was not possible to excavate within these areas.

was a single sherd from a Corfe Mullen flagon (c. 43-70 CE) and three sherds from a Lyon ware Type 1 roughcast cup (c. 43-70 CE). Corfe Mullen and Lyon wares are often linked to pre-Flavian military sites. The pit must have been associated with ironworking activities (see Chapter 15). Pit 643 T21 (Fig. 4.1/2) The isolated pit 643 pre-dated the North Range and was located under the floor on the south side of Room N10 (see Fig. 5.42) and is the earliest evidence for activity in this area. Identified by burnt limestone and charcoal on the cleared surface, the pit was figure-of-eight shaped with shallow sloping sides and a flat base. It measured 1.4m long and varied in width from 0.25m to 0.5m and was 0.22m deep. A roughcast Lyon ware cup base dated c. 4370 CE came from the upper fill (1087) of charcoal-rich loam as did animal bone fragments. A substantial deposit of yellow/red clay (1090) underlay 1087 at the eastern end of the pit. Below this, layer 1094 comprised fine black ash. Burnt limestone, including three large chunks of building stone were present in contexts 1087 and 1094

Early Roman activity in Lower Limepits field Pit 1399 T14 (Fig. 4.1/1) This small pit was recorded 3m to the west of the terminal of enclosure 308. It lay under an undated deposit of building flint; although this resembled part of a wall, it could not be related to any structure. The pit was cut into chalk and measured 0.7m by 0.5m and was 0.35m deep. It was vertically sided but irregular in outline, with a flat base. The upper layers (1385 and 1386) of clay loam and gravel nodules capped the pit. The charcoal-rich fill (1395) contained slag (1,647g), a piece of oven lining, 28 iron fragments and pieces of ceramic roof tile. Hammerscale indicating smithing was retrieved from an environmental sample. A small assemblage of 60 pottery sherds (508g) was recorded. This contained bead-rimmed, necked and lid-seated jars in the Late Iron Age Durotrigian tradition with a date range c. 50 BCE to 90 CE. In addition there

Activity pre-dating the East Range A small assemblage of pottery dating to c. 43-120 CE was located under flint-packed pit 638 (fill 850) (see Chapter 12, Assemblage 6). The pit itself was associated with the Period 4.1 construction of the Aisled Hall but this early

Fig. 4.1. Sections through pits 1399 and 643 (1-2)

49

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) northernmost terrace. The south-west quadrant of the building was badly plough-damaged and the southern extent of wall 1908 (2045 and 1888 only existed as consolidated hoggin foundations, as did the southern wall 2044. The walls comprised courses of large, knapped flints with a rubble core, and incorporated occasional ceramic building material and, unusually, fragments of pottery. The walls were best-preserved in the north-east corner where four courses survived; the average width was 0.5m and the foundation layers were about 0.8m wide. Little wall flint was evident in the five excavated layers over the building, suggesting that dwarf walls may have supported a less substantial superstructure. However, these layers yielded 43,226g of ceramic roof tile fragments which probably derived from the roof. In the second phase of the building, there was evidence for a wooden floor (see below), but otherwise the floors were likely to have been beaten earth or part-tiled.

pottery is associated with ground levelling for the first phase of the East Range building of Period 2.2. Early Roman activity in Eleven Acres The earliest Roman occupation of the site was located to the east of enclosure ditch 1605 and consisted of an Ancillary Building, an associated pit (1735) and an Occupation Area (T82) – see Fig. 1.6. The Ancillary Building This small square building in T86 was located 8m east of enclosure ditch 1605, 40m south of occupation area T82 and 10m north of pit 1735 in T73 and the contemporary ditch system 1641/1644 in T73, T85 and T87 (Fig. 1.6). Pottery evidence suggests that it was constructed before c. 70 CE and had gone out of use by the mid-to-late-3rd century (Chapter 12, Assemblage 5). Figures 4.2 and 4.3 show the building during and at the end of the excavation; Fig. 4.4 is a plan of the building showing all excavated features.

A probable eaves-drip gully (1837) was recorded north of wall 1778, and similar gullies (1867 and 1910) ran south against walls 1888 and 1908. The overwhelming majority of pottery sherds which accumulated in the gullies dated c. 43-100 CE; also present were fragments of ceramic roof tile, nails, slag and a rim from a glass bottle. The presence of an everted rim jar (c. 280-370 CE) suggests that the gullies were not cleaned out on a regular basis and that occupation may have continued into the 3rd century.

The site lies on a north-south slope which was terraced to accommodate the building. Externally, the structure measures approximately 10m by 10m with walls 1888, 1778 and 1908 forming the east, north and west sides. The north wall was constructed directly beneath the

Fig. 4.2. Ancillary Building during excavation, view south-east

50

Early Roman activity outside the villa complex Building sequence and internal organisation Phase 1 (Period 2.1 c. 43-70 CE) (Fig. 4.5/1) Two east-west terraces were created and this was followed by the building of the outer flint walls. Internally, the building was divided on the line of the lower terrace by a similarly constructed east-west wall 2022, and a northsouth wall 2023/2024. These were fragmentary and visible as a 0.6m wide deposit of flint nodules set into hoggin-type clay loam. East-west wall 2022 was bonded to the external wall 1888 confirming that that it was part of the first phase of building. The structure consisted of a western room 9m long and 3m wide, a north-east room measuring 2.4m long by 5m wide and a south-eastern room measuring 6m long and 2.5m wide. There was tentative evidence for a doorway which may have been located in wall 1908; this was difficult to confirm as only one course of flints survived. Phase 2 (Period 2.2 c. 70-100 CE) (Fig. 4.5/2) Lyne (Chapter 12) suggests that internal walls 1929 and 1931 were part of the initial construction phase, although excavation strongly suggested that they were part of the second phase of the building. Only one course of flints

Fig. 4.3. Aerial view of the Ancillary Building after completion of the excavation

Fig. 4.4. Ancillary Building plan of all excavated features

51

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) survived, the construction was not as fine as the outer walls, with haphazardly placed, smaller flints laid on clay. The east-west wall 1929 was built against the lower terrace and it appeared to be bonded to external wall 1908, which may imply a re-build for this part of the wall. The walls had an average width of 0.5m. The room configuration now consisted of a north room (AB1) measuring 3m long and 7m wide, an eastern room (AB2) measuring c. 6m long and 2.6m wide and a western room (AB3) measuring 5.6m long and 3.2m wide. A consolidated deposit of flint nodules on the eastern side of the building (1993) may have marked an entrance. Abutting the southern edge of wall 1929 in room AB3 was a layer of dark brown silty loam which extended along the length of the wall. It was 0.25m wide with a maximum depth of 0.15m. Within this and against the wall, 35 large, mostly complete, Type 1B nails were recorded. The nails were mostly in a vertical position and were from 10mm to 15mm apart (Figs 4.4 and 4.6). The silty loam was probably the decayed remains of a large plank (or planks) through which the nails were driven, providing compelling evidence for a sprung wooden floor. At the northern end of Room AB2, a layer of charcoal-rich loam (1946) with a maximum depth of 0.1m covered its entire width and extended south for about 4m. It was also present underneath wall 1888, suggesting the location of a possible doorway. Finds from this layer included ceramic roof tile and floor tile fragments (2,035g and 1,867g), 25 nails and a small piece of mortar (63g). Glass

Fig. 4.5. Ancillary Building plans of Phases 1 and 2 with wall numbers and room designations

Fig. 4.6. Ancillary Building Room AB3 with row of nails (between the flint wall and scale) from floor planks on the north side of the room. View north

52

Early Roman activity outside the villa complex fragments derived from a cylindrical bottle. The pottery assemblage dated c. 70-100 CE and contained sherds from a BB1 jar and beaker, a mortarium, a Dressel 20 amphora and fragments from a South Gaulish samian plate and bowl.

a Dressel 20 amphora (c. 50-180 CE). A graffito MART was inscribed on the body of one of the beakers (Fig 11.9/3). The presence of a flanged bowl suggests a date after c. 120 CE for this feature. Also present were fragments from the rim and neck of a glass bottle. Two pieces of animal bone were recorded (a dog rib and a cow horn). Other finds included large pieces of burnt limestone and ceramic roof and floor tile. Charred grains included emmer and spelt wheat glume bases. A quern fragment suggests cereal processing nearby.

It would seem that a fire broke out in this room, resulting in extensive burning. A spread of building flint (1928), interpreted as part of a fallen wall, lay over 1946 in the north-east corner of the room and where the flint touched 1946 it was covered in soot/charcoal. After this event, the southern extent of wall 1888 was rebuilt. Any conflagration did not go beyond Room AB2 and may have been contained within its walls, suggesting that the first phase wall 2022 was still in place.

Pit 2031 (Fig. 4.7/4) Shallow oval pit measuring 1m by 0.7m and 0.16m deep with gently sloping sides and an undulating base. The mid-brown gravelly clay-loam fill (1952) yielded pottery fragments from BB1 jars and an unidentified amphora (c. 43-160 CE).

Phase 3 (Periods 2.2-3.2. c. 100-220 CE) That the building was maintained and in use is implicit in material dating to after c. 100 CE. It would seem that no further constructional alterations were made and, by the mid to late 3rd century, the building had gone out of use and was abandoned.

Posthole 1994 (Fig. 4.7/5) Partially under East-west wall 1929 and must pre-date the construction of the wall. Oval in shape measuring 0.35m by 0.2m, straight-sided with a flat base; the undated fill (1995) was of mid-brown silty clay/loam.

Pits and postholes associated with the Ancillary Building (Figs 4.4 and 4.7)

Posthole 2008 (Fig. 4.7/6)

Thirteen pits/postholes (of which nine could be dated by pottery) were recorded and are considered by room. The reason for the pit digging is beyond recovery but is likely to have been associated with activities taking place in the building.

Posthole measuring 0.4m by 0.3m and 0.2m deep with steep sides and a flat base. The undated dark brown loam fill (2009) contained medium-sized flints on the base and sides.

Room AB1 (pits 2015, 1996, 2008, 1991, 2032, 2008 and 1994)

Pit 2032 (Fig. 4.7/7) Underlying the north wall 1778 of the building, therefore pre-dating it, this undated and probably circular pit had a diameter of 0.7m, was steep-sided on the west with a gentle slope on the east. The fill (2033) comprised midbrown silty loam with sparse small gravels.

Pit 2015 (Fig 4.7/1) Sub-circular pit measuring 0.9m by 0.7m and 0.35m deep with steep sloping sides and a flat base. Finds from the fill of fine, mid-brown silty loam (2016) included 15 sherds of BB1 pottery (c. 120-200 CE), two nails and fragments of ceramic roof and floor tile.

Room AB2 (pits 2000, 2017 and 2036) Pit 2000 (Fig, 4.7/8)

Pit 1996 (Fig. 4.7/2) Circular pit with a diameter of 0.8m and a depth of 0.35m with moderately steep sloping sides and a narrow, flat base. There were no finds in the gravelly upper fill (1997), but the lower fill (2003) produced 10 sherds of BB1 pottery and two large chunks of Gauloise 4 and Dressel 20 amphora (c. LIA-140 CE), eight nails and fragments of ceramic roof and floor tile.

Underlying the charcoal deposit 1946, this oval pit with almost vertical sides and a flat base measured 0.85m by 0.95m and 0.4m deep. The primary dark brown clay-loam fill (2002) was sterile. Fill 2001, a similar dark brown clay-loam, produced sherds from BB1 bead rim beakers and a cooking jar (c. 43-100 CE), and fragments from a thick-walled glass bottle. Pieces of ceramic roof tile and floor tile were also recorded.

Pit 1991 (Fig. 4.7/3)

Posthole 2036 (Fig. 4.7/9)

Oval pit measuring 0.8m by 0.65m and 0.35m deep with almost vertical sides and a flat base. Finds were located all through the mid-brown silty loam fill (1992) and included BB1 sherds from bead rim beakers and an everted jar from

With a diameter of 0.39m, depth of 0.30m, vertical sides and a flat base, this undated feature was likely to have been a posthole. The dark brown clay-loam fill (2037) was charcoal-rich and contained large flints, ceramic roof tile 53

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Room AB3 (pits 2295/2020/2038, 2050 and 2052)

fragments and decayed animal bone. Charred spelt grains were abundant.

Pit 2295 cut by pits 2020 and 2038 (Fig. 4.7/11)

Pit 2017 (Fig. 4.7/10)

Primary pit 2295 infilled and was subsequently cut by two smaller pits (2020 and 2038). Pit 2295 measured 0.9m by 0.7m with a depth of 0.24m and had very steep sides and a flat base. The mid-brown loam fill (2078) contained BB1 and Central Gaulish samian sherds (c. 120-200 CE). The

Underlying the charcoal deposit 1946, this oval pit measured 0.46m by 0.34m and was 0.2m deep. The charcoal-rich, very dark brown clay-loam fill (2018) contained four very large smooth pebbles.

Fig. 4.7. Sections of pits associated with the Ancillary Building (1-12)

54

Early Roman activity outside the villa complex excavated (Fig. 4.8); section 1854 is illustrated (Fig. 4.9/1). Four discrete fills were noted, the upper two (1810 and 1843) were variations of mid-brown loam; the third (1849) was dark brown loam with a high clay content and concentrations of charcoal and cess-type material. The base fill (1855) varied from an orange/brown clay loam at either end with dark brown humic loam in the centre. Finds included prismatic glass bottle fragments (c. 43-200 CE), a small fragment from a copper alloy vessel, a knife blade and an iron bucket handle mount (Fig.15.5/28) which had been placed on the on base of the pit. Animal bone from cattle and sheep occurred throughout, but there was a particular concentration of bone from the lowest fill 1803 in section 1746 in the centre of the pit. The 25 fragments of sheep/goat included butchery/kitchen waste. The infilling is dated by pottery to c. 70-100 CE (Chapter 12, Assemblage 3), but the feature must have been constructed before this date and is likely to be contemporary with the cutting of the major enclosures, the Occupation Area T82 and the first phase of the Ancillary Building. The pit had a fairly short period of use since it was cut by ditch 1641, which had filled in by about 150 CE.

upper part of this layer comprised packed flints which had probably filled the top of the pit and survived in the centre of it. The flint layer was cut by two smaller, similar-sized, steep-sided and flat bottomed features with mid-brown loam fills. To the north, undated pit 2038 contained one nail and immediately south, pit 2020 yielded 17 BB1 pottery sherds and a fragment from a South Western Black Burnished ware bowl (c. 100-200 CE), a nail and ceramic roof tile fragments. This pit cluster belongs to the latest occupational phase of the building. Undated pits 2050 and 2052 (Fig. 4.7/12) The profiles of coeval undated pits 2050 and 2052 were similar with moderately steep sloping sides, narrow slightly rounded bases and fills (2051 and 2053) of midbrown silty loam. The pits were overlain by layer 2043 which consisted of small (c. 5mm) packed flints in red clay/loam. Pit 2050 measured 0.7m by 0.2m and was 0.32m deep. Pit 2052 was 0.5m in diameter with a depth of 0.6m; two nails were retrieved from its fill 2053. Pit 1735

Occupation Area T82

Located about 10m south of the Ancillary Building and cut into the natural chalk, this large, sub-circular, vertically-sided and flat-bottomed pit measured 3m by 2.7m, was 1.6m deep and was cut by ditch 1641 (see Fig. 3.9/1). Three sections (1746, 1854 and 1881) were

Geophysical survey indicated activity to the west of the villa complex, south and east of the inner enclosure ditches 309 and 1605 (see Figs 1.5 and 1.6) and, as a

Fig. 4.8. Pit 1735 excavated sections (from left to right) 1881, 1746 and 1854. View east

55

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Deposit 1795 (Fig. 4.10)

result, an area measuring approximately 15m long and 9m wide was cleared by mechanical excavator to a depth of c. 0.30m. Three areas were targeted for work and the exposed surfaces (1748, 1797 and 1774, Fig. 4.11) were trowelled by hand revealing mid-brown gravelly loam. Finds included 81 nails, fragments of glass, animal bone and building material including pieces of stone, ceramic roof and floor tile. Pottery (343 sherds weighing 1,849g) dated occupation to c. 43-200 CE. A T-shaped brooch dated to the 1st/2nd century (Fig. 14.2/4). The numbers of finds noticeably diminished towards the edges of the cleaned area. Three discrete features were recorded: pit 1800, deposit 1795 and pit 1845.

A dark brown silty loam deposit was defined by ash and charcoal fragments, and measured 2m by 2.5m and was 0.12m deep. Excavation suggested that it had originally been mounded. Finds included 33 sherds of pottery weighing 152g comprising BB1fragments from a lid, flagon and jars with a date range c. 43-100 CE. Five pieces of glass from a cylindrical bottle were contemporary. Pit 1845 (Fig. 4.9/3) This pit was approximately 1m in diameter and up to 0.7m deep, with almost vertical sides and a slight step to the south. Finds from the upper fill (1846) included fragments from a BB1necked jar which dated the infilling to c. 43120 CE. The base respected two geological north-south fissures filled with orange/brown clay loam and large flints.

Pit 1800 (Figs. 4.9/2 and 4.12). The edge of this small vertical-sided pit was defined by small cobbles. It measured 0.5m in diameter and 0.20m deep with a charcoal-rich fill (1801) which contained emmer grains and spelt wheat. Fragments from a beadrimmed jar with a date range of Late Iron Age to c. 70 CE suggest that this is the earliest feature in this area. The pit was only visible when deposit 1795 had been removed and it is possible that it may have been truncated by later activity.

It is likely that this area was occupied from the earliest Roman period until the end of the 2nd century (Periods 2.1, 2.2 and 3.1) and may have been an outside workplace. It is possible that occupation extended both west and north but the lack of finds on the east and south suggest that those edges of the excavation define the occupied area.

Fig. 4.9. Sections of pits 1735, 1800 and 1845 (1-3)

56

Early Roman activity outside the villa complex

Fig. 4.10. Occupation Area in T82 with deposit 1795 and pit 1800 to the right of the ranging rod, pit 1845 to the right. View east.

Fig. 4.11. Plan of Occupation Area T82 with excavated features

57

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 4.12. Pit 1800 half section, view east

58

5 The North Range Parts of the North Range were exposed in the initial trenching programme of 2012, when work on the 98m long by 1.5m wide Trench 2 was undertaken (see Fig. 1.6 for trench locations). The area was cleared by the farm bulldozer, when approximately 0.1m to 0.15m of topsoil was removed. At the west end of the trench, cleaning revealed evidence for surviving walls and collapsed structural material (Fig. 5.1). The bulldozer was used again in 2013, when Trench 5 was extended to the north in the area where walls had been located the previous year.

visible. The building was terraced into the slope and was altered and reconstructed over three centuries and continued to be occupied for a further two centuries. An aerial photograph taken in 2015 (Fig. 5.3) shows the range at the end of the excavation. Phasing There was great difficulty in allocating constructional phases due to the length of occupation associated with rebuilding and alterations over a period of three centuries. The strategy, which was developed over the period of excavation, determined that areas of mosaic should be left intact with as little intervention as possible. This policy restricted exploration of earlier building phases, particularly under Rooms N1and N6. However, extensive excavation in Room N3 highlighted phasing possibilities which were applied to other parts of the range; stratigraphy, pottery and coins aided this process. A series of phase plans (Fig. 5.4) illustrate the building sequence. Information regarding the construction of the external and internal walls (details in archive) was used to aid the phasing; the wall construction is discussed in Chapter 9.

As the work progressed new trench numbers were allocated; T19, T20, T21, T21b, T22, T23, T28, T30, T32, T36 and T38. The stripped surfaces were cleaned by hand and clearance numbers allocated. Large amounts of building debris, particularly roofing material and mortar relating to the collapse of the building, were recorded. Five coins were retrieved from this; an antoninianus of Tetricus I (271-273), a minim of Gratian (367-375), two nummi of the ouse of Theodosius (388-395HHHouse of Theodosius (388-395) and an unidentified issue. It was noted that there was less overburden over the southern and eastern parts of the building. Figure 5.2 illustrates the room designations and wall numbers, which were allocated as they were uncovered. Plough damage was evident in Rooms N5, N7, N8 and N10. In 2013, five rooms were exposed and numbered N1 to N5; a complete mosaic floor was uncovered in Room N1 and part of a floor in Room N2. In 2014 a mechanical excavator was hired and trenches to the south and east were set out. That year, a 50 year-old JCB mechanical excavator, belonging to volunteer Robert Heaton, was brought onto the site enabling targeted stripping. At the end of the season evidence for seven further rooms (N6 to N12) had been uncovered, with mosaic floors recorded in three of these. Work was completed during the 2015 season. Fieldwork consisted of excavating features within the building and removing part of the deep infill sequence in Room N3. Scrutiny of the plans, sections and photographs enabled phasing of this range; however, caution has to be exercised regarding this, due to limited excavation in a number of areas. Care was taken not to compromise or damage any of the surviving mosaic floors or walls. The building faced south and was constructed on land which sloped south and east. The first building phase was located 16m south of enclosure ditch 309 and 8m northeast of the junction of ditches 314 and 308 (see Fig. 3.1). When excavations were complete, an L-shaped building was revealed consisting of a final phase of nine rooms plus two narrow passages. In addition, semi-circular walls of an apse, and walls and features from other phases were

Fig. 5.1. September 2012, Trench 2 revealing structural remains of walls 30, 31 and 32, view east

59

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.2. Plan of the North Range with Room and wall numbers. Inset shows walls and phasing sequence of Room N3

Fig. 5.3. The excavated North Range 2015, the north-west corner of the Aisled Hall is centre top. Aerial view to the south-east.

60

The North Range Phase 1: Early Roman, Period 2.2 (c. 70-120 CE) (Fig. 5.4 Phase1)

tooled, mortared limestone blocks surviving to a height of 0.86m formed the north-west corner of the range (Fig. 5.7). The north-south wall 1046, of which it was part, could be seen underneath wall 126. Similarly, the eastwest, northern wall of the range (1048) could be discerned under wall 119 and its presence was confirmed by a small exploratory excavation in Room N3. Four courses of mortared flint from a north-south wall 991 survived (Fig. 5.8), creating a room approximately 4.2m wide by 3.2m long. Significantly, this room was approximately 1m deeper than the rest of the range.

The ground was levelled before building work proceeded. Approximately 0.7m of chalk was removed at the west end of this ‘terrace’; there was little need for terracing to the east where the ground was lower. An east-west orientated strip building measuring possibly 24 m long (externally) and approximately 4.5m wide, with an internal width of 3.5m, was set out and was identified by walls located in Room N3 and stonework and a posthole under Room N10. It is likely that there were four rooms but, apart from Room N3, it was not possible to identify their dimensions.

Room N10 A number of features were excavated in Room N10 at the eastern end of the building and, although three were undated, in all likelihood belong to this phase. All were located below the Phase 3 floor. Two comprised structural elements (952 and 1166) and two were pits (643 and 644); for locations see Fig. 5.42).

The first structural element to be identified was east-west wall 244 in 2013 when a 1.8m length was uncovered (Fig. 5.5). The 0.5m wide wall comprised seven courses of mortared flint surviving to a height of 0.7m. A shallow bedding trench for the wall had been cut into the natural chalk and measured approximately 0.7m wide and 0.18m deep. In total, 3.5m of wall was exposed; its eastern extent either did not survive or there may have been a doorway. At the western end, the corner was strengthened with a quoin (1144) consisting of large mortared limestone blocks, five of which survived (Fig. 5.6).

Feature 1166 (Figs 5.9 and 5.10/1) Located at the western side of the room, this feature was revealed in the base of the Phase 5 posthole 1163 as a deposit of mortar and limestone blocks. A sondage (1194) measuring 1m by 0.6m and 0.38m deep was subsequently opened, revealing stonework which related to an earlier building sequence. The stratigraphy is illustrated on Figure 5.10/1. The stonework survived under the mortar layer

Quoin 147 was identified early in 2013 but its structural significance was not realised at the time. Seven large,

Fig. 5.4. North Range Phase plans (1-5)

61

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.5. Inside Room N3, wall 244 in bedding trench, view south

Fig. 5.6. South-west external corner of Room N3, wall 244 and quoin 1144 with remnants of wall 1228 above, view north-west

Fig. 5.7. North-west external corner of Room N3. Quoin 147 with wall 1048 visible by the ranging rod and wall 1046 visible next to the paper tag, view south-east

62

The North Range

Fig. 5.8. Room N3; remnants of Phase 1 north-south wall 991 (right) under east-west Phase 3 wall 667, view north

1197, which related to the Phase 3 (Period 3.2) mosaic floor 615 and comprised at least three layers of mortared limestone blocks. It was not possible to follow the feature northwards as this would have destroyed the small amount of surviving mosaic pavement. Due to the size of the intervention, it was not possible to determine the function of this feature, although its alignment suggests that it was structural and may have been part of a north wall of the range. The levels were raised by two infills; clay loam 1199, and above this black ashy layer 1198. The only finds were three fresh sherds of Central Gaulish samian pottery (33g) from three different vessels dating c. 120-200 CE, suggesting that structural element 1166 had gone out of use some time in the 2nd century.

779. The undated lower fill (953) was sticky clay loam and probably represented infilling after the removal of a structural upright.

Posthole 952 (Fig. 5.10 Phase 2; see Fig. 5.42 for location)

Located on the western side of the room, this pit underlay the opus signinum layer 625 and was identified by an area of burnt stone and gravel. The oval pit measured 0.6m by 0.4m, had steeply sloping sides, a flat base and was 0.18m deep. The single fill (874) of dark brown/black ashy loam contained 26 large pieces of burnt limestone. Fifteen fragments of slag were scattered over the surface. The presence of burnt stone in this pit and in pit 643 suggests that they could have been contemporary. If so, then it is possible that small-scale ironworking was taking place in this part of the range.

Pit 643 (Fig. 4.1/2) This pit is described in Chapter 4 and is dated by pottery to c. 43-70 CE, suggesting that its digging could have predated the building. It is possible, however, that the feature could have been contemporary with the initial occupation of the range. It was characterised by the presence of large amounts of burnt limestone. Pit 644 (Figs 5.10/3 and 5.11)

The top of posthole 952 was 0.55m below the latest floor level on the eastern side of the room and was recorded in the base of the Phase 5 pit 614. The flat-based posthole measured 0.45m by 0.4m at its surface, and halfway down narrowed to a diameter of 0.21m. Two fills were recorded, the upper (943) contained debris from the fill of pit 614 including flint nodules, mortar and painted plaster which probably derived from the collapsing wall 63

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.9. Room N10; Phase 1 structural element 1166 (right) underneath the later floors, view west

Layers associated with sondage: 474 – mosaic laid in white mortar 623 – pink/red opus signinum 1195 – grey mortar with small stones 1196 – pink opus signinum 1197 – yellow mortar/sand 1198 – black ash with red/grey clay 1199 – mid-brown loam Fig. 5.10. Phase 1 features associated with Room N10: sondage 1194, posthole 952. Pit 644 (1-3)

64

The North Range Conclusions

a strengthening addition to the Phase 1 wall 1046. It was sloppily constructed with large flints in mortar (similar to those in the apse wall), resulting in an untidy rough internal face (see Fig. 5.12).

The general lack of dating material is consistent with a thorough clean-out of the building prior to further alterations and rebuilding. A small amount of pottery was retrieved. The earliest came from pit 643 in Room N10, where the presence of a sherd from a Lyon ware roughcast beaker (c. 43-70 CE) hints at a military presence. Sherds from a similarly dated Corfe Mullen whiteware flagon came from layer 355 in Room N6. A deposit of burnt material (369) in the same room produced fragments from a BB1 jar dated c. 90-200 CE. This material came from the south-east corner of the room where the mosaic pavement (357) was not preserved, and thus pre-dates its construction. The enclosure ditches 314 and 308 to the west of the villa were open and used for rubbish disposal. The fill of sections 1139 and 363 nearest to the North Range produced BB1 jars and bowls and a C17 Dr 37 copy bowl dating c. 70-150 CE. It is likely that this material derived from activities associated with the Phase 1 building.

Room N1 A small sondage (1231) measuring 0.7m by 0.2m and 0.6m deep was excavated on the extreme south-east corner of the room (Fig. 5.14/1), where earlier activity was suspected and where the eastern edge of the mosaic (153) petered out. The upper fill 983 of mid-brown loam contained small fragments of limestone and ceramic roof tile and four white limestone tesserae. The similar middle fill 987 differed in that mortar and painted plaster were present. Two sherds (134g) from a BB1 cavetto-rimmed jar dated c. 160-200 CE suggested that this fill accumulated in the late 2nd century. The lowest fill 989 was a chalky loam containing chunks of chalk and occasional charcoal flecks; the base of this layer was not excavated. It is possible that 989 comprised remnants of a floor belonging to Phase 2. The fills were located behind and underneath remains of wall 1014.

The Phase 1 building was probably constructed at the end of Period 2.2 and the beginning of Period 3.1 in the early years of the 2nd century CE, most likely at the same time that the first phase of the East Range was built. The dimensions of the building suggest that it was of a simple strip type possibly with four rooms, the floor of the westernmost (N3) being at least 1m deeper than the others (Fig. 5.12). The use of stone implies access to local sources of limestone and to craftsmen who could work such material, as well as the necessary finance to accomplish this. It is not possible to determine the status and function of the building, although its construction suggests that it was of some significance.

Conclusions There is sparse but compelling evidence for the construction of what was probably a very grand, spacious reception space built on to the range of rooms to the east, at a very early date in the villa’s history. Its function is irretrievable, but such spaces were often reception areas or audience chambers. The putative size suggests a building of high status and the presence of tesserae and painted wall plaster from sondage 1231 hints at wall and floor decoration. The rough internal face of wall 1229 visible on the west side of Room N3 suggests that activities there did not require high standards of decoration.

Phase 2: Middle Roman, Period 3.1 (c. 120-200 CE) (Fig. 5.4 Phase 2)

Phase 3: Middle Roman Period 3.2 (c. 200-250 CE) (Fig. 5.4 Phase 3)

Room N12 The north part of a substantial curved wall (1042/51) underlay the later Phase 4 west wall (30/149) suggesting the presence of a large apsidal-ended reception room placed at an angle to the Phase 1 range (see Fig. 5.13). The 0.6m wide wall was constructed of large flints in mortar and appeared to have been rebuilt or re-aligned. There was no ceramic material or stone in the exposed section, indicating an early build. No external doorway was located but it was likely to have been opposite the apse. Finds recorded between wall 149 and the apse wall included sherds from a BB1 jar (c. 90-160 CE), from a South West Black Burnished ware jar (c. 70-250 CE) and glass fragments from a 2nd/early 3rd century colourless cup. A 1m length of a substantial wall (1146) was visible to the south of Room N3 on the eastern side, creating a room with a possible east-west width of about 8m (but unknown north-south dimensions). This wall appeared to continue over wall 244 into Room N3, was visible for 0.9m and was recorded as 1228. Later, it formed a base for the construction of phase 4 wall 119 and must have been

The west extension (Fig. 5.13) The apsed wall was demolished and a north-south extension was set out, not quite at right angles to the Phase 1 east-west range (approximately 5 ͦ west of south) on the west side of Room N3, forming an ‘L-shaped’ building comprising west and north wings. The new range consisted of two, possibly three, rooms and was defined by east wall 46/255, west wall 32/1076 and south wall 863. An east-west wall (256) separated the two southern rooms. The external width of the extension was approximately 5.5m; its length could not be estimated but it was in excess of 10m, possibly in the region of 20m. The extant walls were between 0.5m to 0.6m wide and constructed of welllaid flints in mortar, although only one to three courses survived. Fragments of ceramic and box flue tiles were incorporated into the flint rubble core suggesting re-use of this material. The rooms underlay the later Rooms N1 and N2. 65

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.11. Room N10, pit 644 half section with burnt stone in situ, view west

Fig. 5.12. Room N3 with all exposed walls. In the foreground Phase 1 wall 244, to the right truncated Phase 2 wall 991 with Phase 4 wall 422 above. To the left Phase 2 wall 1228 with Phase 4 wall 119 above. Centre, Phase 3 wall 667 and behind, Phase 4 wall 420, view north

66

The North Range

Fig. 5.13. West extension of the North Range comprising Rooms N1, N2, N5 and corridor N4; remnants of the apsed room N12 to the left, quoin 147 (Phase 1) bottom left, view south

of burning. Fill 266 was a gravelly loam, burnt bright red. The mid-brown loam layer 269 contained decomposing mortar, ceramic and limestone tile fragments and six tesserae, and was present around the fireplace to the west and north.

Room N2 The Phase 4 mosaic 246 overlay east-west wall 256 which divided this room from Room N5 to the south. Evidence from three small interventions revealed contemporary activity.

Sondage 1057

Scoop 253 (Fig. 5.14/2)

A small sondage (0.45m by 0.35m by 0.2m deep) was cut on the south side of wall 235 to investigate the possibility of wall 1076 continuing northwards. However, it was not present, but below the edge of the Phase 4 mosaic (246), 25 tesserae were recorded, several of which were mortared together. These were from an earlier floor (context 1059) and, together with similar tesserae recovered from fireplace 264, are strong evidence for a mosaic relating to this phase. The sondage is visible between the mosaic and wall in Figure 5.13.

Located in the south-east corner of the room, this shallow, quadrant-shaped scoop was dug between the junction of west wall 255 and south wall 256. The fill (254) of dark brown loam flecked with chalk and charcoal contained burnt flint and broken ceramic roof tile near the flat base. Fragments (5/64g) from a BB1 bowl dated c. 210-280 CE. Fireplace 264 (Figs 5.14/3-4 and 5.15) Immediately north of the scoop and constructed against the west wall 255, this feature measured 0.8m long, had a depth of 0.2m and extended 0.45m into the room. It was defined to the north and south by semi-domed, white clay ‘cheeks’, 0.12m high and 0.15m wide. The northernmost (267) was shorter at 0.23m, the southern cheek (268) was 0.3m long. The clay was not burnt due to the presence of flints and stone which lined the fireplace. An arrangement of flints defined the front of the fireplace. The upper fill (265) of white clay and chalk must have been inserted after the fireplace went out of use as there was no evidence

Make-up layers Between 0.18 and 0.3m of material was removed at the south-east end of the room (257/296/1015). This consisted of gravelly loam containing quantities of flints, limestone building stone and roof tile, mortar, painted plaster, chalk fragments and 75 tesserae; the material was interpreted as infill. Fragments from a BB1bowl dated to c. 200-300 CE suggesting a 3rd century timeframe for this event which must be related to the Phase 4 building work. 67

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.14. Features associated with Rooms N1, N2 and N5; sections of sondage 1231 and pit 253 (1-2), plans and sections of fireplace 264 (3-4)

gravelly loam with relatively small amounts of building material and no datable pottery. Underneath was a hoggintype layer (849). The base of an oven (992) was located in the southern part of the room and cut through the hoggin (Figs 5.16/1-2). The oval structure measured 1m by 0.8m, had a surviving depth of 0.2m and was defined by burnt clay, large flints and burnt limestone (1007). The oven chamber (1008) was 0.4m in diameter, had almost vertical

Room N5 This room, bounded by walls 256 to the north, 46 to the east, 863 to the south and 32/1076 to the west was 4.2m wide and 5.8m long. There was a 1.5m break in wall 32 (see Fig. 5.13) which may indicate the presence of an external doorway. This part of the building had suffered badly from plough damage. Trowelling back revealed 68

The North Range

Fig. 5.15. Room N2 fireplace 264, view east

sides and a flat clay-covered base. Two fills were recorded; the north (1009) was dark brown and charcoal-flecked, the south (1011) was mid-brown with clay and chalk. No finds were recovered, but a sample of emmer/spelt wheat was submitted for radiocarbon dating and returned dates of 129-254 cal CE (SUERC-100062) at 95% confidence. Activities involving heat are indicated. It is similar in form to the early Period 2.2 ovens in the East Range.

in use until the building was abandoned. Mosaic floors were laid in Rooms N6, N7, N10 and N11, together with plastered and painted walls. Ploughing had removed any evidence for such decoration in Room N8. Architectural details of the mosaics and wall plaster are considered in Chapter 10.

Extension of the original building (Figs. 5.4/3 and 5.17)

A box trench (496) measuring 1.5m by 1m by 0.48m deep was cut into the north-east corner of Room N7 to determine the make-up of mosaic (329) and its relationship with walls 338 and 472. The complex stratigraphy, however, illustrated a sequence predating the laying of the pavement and of relevance to this building phase.

Mortar mixing basin 715 (Figs 5.18 and 5.19/1)

With the exception of Room N3, Rooms N6, N7 and N10 were extended northwards; prior to the rebuilding, the ground was terraced to accommodate to accommodate this. A narrow corridor N11, was constructed in front of the range. The building was the same length but now measured 8m wide. This extension involved the laying of a thick deposit of hoggin-like material (1048) to the east of N3. Further terracing took place on the north side of the house to facilitate the construction of the new north wall (338) and as a result, gulley (460) was formed behind this wall. This feature is discussed after consideration of the rooms in this range. The re-build included a new south wall (750/895). Apart from the west wall of Room N3, all north-south walls were extended north and they continued

Excavation revealed an oval mortar mixing basin (715), 1.08m long by 0.82m wide with gently sloping sides and a depth of 0.15m with walls approximately 0.1m thick. It was located 0.2m from the north and east walls of the room and was formed from an extremely hard grey, concrete-type material (529). It was completely filled with thin grey limestone roof tiles (22,225g), building stone (19,300g) and ceramic roof tile fragments (2,956g). Two complete limestone tiles were cemented to the base (see 69

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.16. Plan and section of oven 992 (1-2)

Fig. 5.17. Extension of the North Range comprising Rooms N3, N6, N7, N8, N10, N9 and corridor N11; East Range to the top right, view east.

70

The North Range

Fig. 5.18. Mortar mixing basin in the north-east corner of Room N7. Limestone roof tiles were cemented to the floor of the basin and were used to level up before the feature was covered by the construction of the floor, view north

Fig. 5.18), suggesting that its function had been to mix large quantities of mortar, some of which was present and still wet when the tiles were thrown in. The infill material is probably debris from building work; in this case the tiles were likely to have been removed from the Phase 1 building to facilitate the northern extension of Phase 2. The dark grey limestone was very different to the later stone roof tiles but was similar to those found in the opus signinum-lined tank 778 in the Period 3.1 East Range. It is likely that the Phase 3 building roofs were clad with this limestone.

opus signinum (330), onto which the mosaic pavement (329) had been laid. The rooms in this wing are considered from west to east. Room N3 (Fig. 5.12). The north wall (1048) of Room N3 was demolished and replaced by the narrower (0.45m wide) wall 667 which partially overlaid it. Seven courses of this exceptionally well-built, east-west wall survived and comprised carefully knapped and mortared flint in a herringbonepattern and laid directly onto chalk bedrock. Two 1.2m wide sections were exposed on the east and west of the room. The make-up of the courses on the east side differed in that the bottom 0.15m consisted of large pieces of stone and occasional flint crudely mortared together. This course butted against the demolished earlier wall 991, and the upper flint layers were constructed across it. The two remaining courses of wall 991 may have been retained as a low bench or shelf. The reasons for the very different construction of the lower course of wall 667 on the east may indicate the position of a former doorway through the first phase wall 1048. A new north-south wall (422) divided this room from Rooms N6 and N8. This appeared to be constructed on introduced compacted hoggin and incorporated rough limestone blocks in the lowest course at the junction of the dividing wall 337

The layers around and above the basin indicate the sequence for the floor make-up. This was aided by the fact that much of the room had been badly plough-damaged, and investigation did not impinge on the area of surviving mosaic. The lowest layer (704), only recorded on the south side of the feature, was sticky clay loam which was probably used as consolidation. To the north, layers 531 and 532 against the north wall (338) were at a similar level and comprised yellow/orange sandy loam and a similar darker brown loam with possibly decomposed mortar. A levelling layer of chalk (530) varying from 0.05m to 0.1m in thickness was recorded over these layers, the fill (512) of the basin, and survived over three quarters of the floor of the room. Over this was a c. 0.1m layer of yellow mortar (497) which supported compacted flints (495). The final layer comprised 0.05m of cream 71

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) maximum depth of 0.1m and extended approximately 0.4m into the room. The north and south sides were defined by cheeks of opus signinum (1136). The 80mm wide north cheek was better preserved and merged with the quarter round moulding against the east wall. The fireplace was filled with a charcoal-flecked loam (1134), but there was little evidence for great heat, so it is likely that this fill was introduced at a later date.

(Fig 5.20). Natural chalk was utilised as the floor of the room, now measuring approximately 4m by 3m, and still at a lower level than the rest of the building. Room N6 Internally the room measured 4.5m by 3m. New external walls were constructed to the west (422) and north (338) and an internal wall to the south (337). The dividing wall (401) between Rooms N6 and N7 was extended north. Walls 402 and 401 were butted against the north wall 338. The make up for the floor consisted of flints laid on a levelling layer of chalk, then a further layer of chalk overlain by yellowish opus signinum. Over this was a pavement (mosaic 357) of red ceramic and occasional white limestone tesserae (Fig. 5.21), most of which were probably cut from roof tiles, although there were large number of pieces of box flue tiles. The walls were plastered and painted, and a quarter round moulding or skirting board of red opus signinum (399) was laid around the edges of the room to protect the walls. This was rounded at the top, 60mm high with a width of 30mm at the base.

Room N8 The small room south of N6 measured 4.5m long, was only 2m wide and had been severely damaged by ploughing; the effects can be clearly seen on Fig. 5.17. Wall 337 separated this room from Room N6 to the north. North-south wall 401 in Room N6 appeared to peter out in Room N8; scrutiny of the site photographs suggests that there was never a wall in this position during this and later phases. However, the shadow of the demolished wall can be seen in Figure 5.3. The narrow width of the room hints that the area probably served as an ante-room for Room N7 to the east. The south wall 441 belonged to the first constructional phase of the building and a chalk floor (1039) was recorded in the extreme south-west corner of the room, a layer of compacted flints separating this from the later floor (370) (Fig. 5. 20). Both floors were 0.1m thick. It is likely that the room would have had a

Fireplace 415 (Figs 5.19/2, 5.21 and 5.22) Located slightly off-centre, the fireplace was constructed against east wall 401. It measured 0.8m wide, had a

Fig. 5.19. Room N7, section of mortar mixing bowl 715; Room N6, plan of fireplace 415 (1-2)

72

The North Range

Fig. 5.20. Phase 4 wall 442 separating Rooms N3 (left), N6 and N8. The early chalk floor 1039 is visible under the wall, and the later floor 370 is above a layer of compacted chalk, view north

tessellated pavement but no evidence of this or of a mortar bedding layer survived. Wall 422 was constructed on top of the earliest chalk floor.

northern third of the mosaic pavement (329) survived (Fig. 5.23). The mosaic comprised a central checkerboard of red and white squares, each square measuring about 0.3m by 0.3m, which was bordered by a simple meander also in red and white and would have measured 5m by 4.3m. The red tesserae were ceramic and the white were two types of limestone. When dry, the limestone appeared to be the same shade of white, but when wet it could be seen that two very different sources of stone had been used. A rounded quarter moulding of red opus signinum (366) separated the wall and floor as in Room N6. Large surviving fragments of wall plaster indicated that the room had a decorative scheme of cream with a pink stripe.

Room N7 This rectangular room measured 6.5m by 5m; the southern half had been extremely damaged by ploughing (see Fig. 5.17). Wall 338 formed the north side of the room. As noted in the descriptions of Rooms N6 and N8, the west wall (401) was absent on the eastern side of Room N8, and it is probable that doors or a screen divided the two spaces. The eastern internal wall 472 could be traced for its complete length although it was not well preserved, with only one course of scattered, horizontally laid flints in mortar. The southern wall 441 survived as one course of flints in mortar; it is likely that a door with access to corridor N11was located in this wall. The complex stratigraphy of the floor was noted during the excavation of the mortar mixing basin. Only the

Room N10 This extensively plough-damaged room measured 6.2m by 5m. It shared the west wall (472) with Room N7 and in this phase, wall 606 was the eastern external wall. At 73

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.21. Room N6 mosaic 357 with fireplace 415 at the top of the photo. Quarter moulding 399 which separated the floor from the wall survives on the north and east edges of the room, view east

Fig. 5.22. Room N6 fireplace 415 against wall 401, view east

74

The North Range

Fig. 5.23. Room N7, surviving northern third of mosaic 329, view west

N8, N7 and N10 (Fig. 5.26). Wall 441 became an internal feature and the corridor floor was 0.4m below the level of the rooms to the north. The new external south walls comprised 750 to the west and 895 to the east; these were separated by a gap of 1.5m possibly indicating double doors. A large, rounded, rectangular limestone slab measuring 0.6m by 0.25m lay in the corridor outside Room N7 and may have been part of a doorstep. The outer walls at 0.5m wide were narrower than the other exterior walls, hinting that they were probably only one storey high. The Phase 6 roof collapse, and an underlying chalk floor (981) suggest the presence of a porch. There may also have been east and west doorways. There was no evidence for the corridor extending beyond Room N10.

0.55m wide, it was the same width as the north wall (338) of the building (much of which was re-built as 779 in Phase 4). A 2m length of the south wall (441) survived in the south-east corner. Access to the room is likely to have been though a doorway in the south wall. Because of the plough damage it was possible to identify the make-up of the floors. The sequence of hoggin, compacted flint, chalk, mortar and mosaic was identical to that in Room N7. There were, however, further mortar layers and a second, later mosaic floor which is discussed in the Phase 4 section. A very small amount of the mosaic (615) which was laid in this phase survived as a triangular patch in the north-west corner of the room (Fig. 5.24). Most of this was under later mosaic 475 but cleaning back located its presence further south and a section taken to determine relative levels (Fig. 5.25) revealed the mosaic had been laid on hard pink opus signinum. The site notes record that considerable plough damage had displaced at least 1,880 small, white limestone tesserae. These must have comprised the central element of the floor, as small red and brown tesserae were present on what would have been the pavement’s borders. Examination of the Phase 4 painted plaster indicated two decorative schemes separated by a plaster skim, indicating painted walls in this phase (Fig. 5.36).

Fragments of faded painted plaster were located on the north wall 441and were left in situ. Pieces weighing 2,109g, including decorated fragments, were recorded in the collapse debris and were probably part of this room’s decoration. The lowest floor encountered was recorded as beaten chalk, although 2,626 tesserae, including several joined pieces (Fig. 5.27) and chunks of opus signinum were recorded in the layer above this, strongly suggesting the presence of a tessellated pavement. The painted plaster and a tessellated floor indicate that this was an enclosed space, not open to the elements.

Corridor N11

Conclusions

A very narrow corridor with an internal width of 1.3m and measuring in excess of 15m long (the eastern extent had been ploughed away), was constructed in front of Rooms

Major building works at this time saw the demolition of the short-lived apsidal Room N12 and the conversion of the 75

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.24. Room N10 remnants of upper mosaic 474, lower mosaic 615 and underlying bedding layers, view west

Fig. 5.25. Room N10 with remnants of lower mosaic 615 embedded in pink opus signinum. Upper mosaic 474 visible at top right corner, view west

76

The North Range

Fig. 5.26. Corridor N11 constructed at a lower level to the rooms of the North Range, view east

original strip building into a winged, ‘portico-type’ villa consisting of north and west wings. Rooms N6, N7, N8 and N10 in the north wing had mosaic floors and painted plaster walls and were probably used as reception rooms with access to them through the extremely narrow corridor N11. This also had a mosaic floor and painted walls and appeared to be the main entrance into the courtyard to the south of the building. Room N3, a ‘half-cellar’, was

probably a service room. The function of the west wing is less certain but may also have contained decorated reception rooms (N1 and N2) and a service room (N5). Phase 4: Late Roman Period 4.2 (c. 300-350/70 CE) (Fig. 5.4 Phase 4) Major building work was undertaken during this phase and included the extension and re-positioning of rooms in the west wing, a further room constructed on the eastern end of the north wing, as well as extension and major changes to the sunken Room N3. The decorative schemes are considered in Chapter 10. The west wing The work in this area comprised the demolition of internal walls 32, 1076 and 256, the construction of a new west wall (30/149) and the raising of floor levels in Rooms N1 and N2, creating a building measuring 13.7m long and 6m wide. A narrow, corridor-type extension (N4) was built to the east of Room N5. The rooms are considered from north to south and the layout is apparent on Figure 5.13. Room N1 The north wall (120) of the Phase 3 building was retained and extended west by 1.2m where it turned south as the new west wall 149, with the corner consolidated by a quoin of

Fig. 5.27. Joined tesserae (840) from the layer above the chalk floor in corridor N11

77

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) limestone blocks. It is probable that a biforium window was constructed in this wall as evidenced by the Tuscan column and architrave fragment located in the collapse debris (Fig. 9.35). To the east a new wall (126) was constructed against wall 119 of Room N3. An internal wall (235) butted the west wall 149 and separated this room from N2 to the south. The walls were all of mortared flint. Phase 6 collapse evidence suggests that the room was vaulted. Internally, the room measured 5.25m by 4.5m with a 1.7m wide doorway in the south-east corner. Mosaic 153, Fig. 5.13) was laid on hard yellow mortar. The floor make-up was not investigated due to the almost-complete survival of the pavement which is considered in detail in Chapter 10. The major element of an expanding rectilinear central swastika in blue lias is surrounded by poised white squares and a panel of ashlar blocks. The wide border comprised larger limestone and ceramic tesserae, whilst the doorway was marked by three rows of white tesserae. A quarter round moulding of red/ pink opus signinum survived on all but the southern wall. Large quantities of painted wall plaster were preserved underneath the Phase 6 roof collapse. The back of some plaster fragments had imprints of wooden laths, suggestive of wattle and daub, possibly from the non-load bearing internal wall 235. An area of red discolouration just inside the doorway may have been caused by heat, probably from a portable brazier. The figured floor and the decorated walls suggest that this was a major reception room.

Room N2 The Phase 3 floor was raised and a new room measuring 5.5m by 2.6m with access to Room N1 was refigured (Fig. 5.28). The Phase 3 wall 253 was retained on the east, and new walls comprised the Phase 4 west wall 149, the internal wall 335, which was shared with Room N1, and a new south wall (31), which divided Room N2 from Room N5. A figured mosaic (246) was laid, and quarter round moulding protected the painted walls. Only the western part of this floor survived due to extensive plough damage. It is likely that this was an anteroom to Room N1. Room N4 This small enigmatic area to the east of Room N5 (Figs 5.3 and 5.17) appeared to form a lobby or porch with a width of only 1.1m. The west wall utilised wall 46 of Room N5 and part of wall 255 of Room N2. The 1.2m length of north wall (60) appeared to be butted on to this and survived as a single layer of flints in mortar. The north part of the wall incorporated thin limestone pieces which were possibly part of a doorstep. The east wall (44) survived for only 2m at the north; its southern extent was ploughed out. It was the narrowest wall (0.33m wide) recorded on any villa building at Druce Farm, its width suggesting that it was not load-bearing. A rammed chalk floor was recorded; the

Fig. 5.28. Room N2 Phases 3 and 4. The earlier east-west wall 256 is below mosaic 246. Note the change in levels between Rooms N1 and N2 and the plough damage to mosaic 246, view north

78

The North Range stratigraphical sequence in this room was the bestpreserved and most complex on the site.

post-abandonment layer over this produced loose tesserae and large amounts of painted plaster, strongly suggesting that this ‘room’ was decorated to a high standard. Pottery (44 sherds weighing 353g) broadly dated c. 240-430 CE; a coin of Gratian (367-375) and a Late Roman globular bead support the dating for use of the area. Because of its size, its function is problematical; it may have been an entrance porch to the reception suite of Rooms N1 and N2 or even an external shrine and, as such, requiring a high standard of decoration.

It is probable that the new walls were constructed first. A decision was taken to build the west wall (119) on the remnants of the Phase 2 wall (1228) (see Fig. 5.12), which butted the east wall (126) of Room N1. No effort was taken to key this in (see Figs 5.13 and 5.17), perhaps relating to the fact that the west wing was set out at a slight angle to the north wing and possibly suggesting that the roof levels of Rooms N1 and N3 were different. The base of the wall was offset and seven courses of flint laid in a herringbone fashion survived. The north wall (420) was constructed in a similar fashion; the north-west corner was strengthened with large, well-laid limestone blocks forming quoin 1227 (Fig. 5.29). The junction between wall 420 and the earlier north and west walls (338 and 422) was visible in the wall construction (Fig. 5.30). Wall 442 was extended north and retained as the internal east wall. Wall 244 continued as the external south wall; the infill discussed below was apparent to the north of this wall but not to the south, confirming that the wall was extant when the infill was deposited. The extension created a room measuring 4.2m wide and 5.5m long.

Paved area and potential building Evidence for a paved area (947) and a collapsed wall (1073) outside and immediately east of Rooms N2 and N4 can be seen on Fig. 5.17. This consisted of deliberately laid limestone and ceramic roofing tiles set on a chalk surface and was probably associated with the putative entrance of Room N4. To the south of this was a 0.3m deep deposit of building material consisting of mortared flint and roofing tiles, suggestive of a collapsed building. Painted plaster and tesserae were also recorded. Pottery (17 sherds weighing 148g) indicated that it was occupied c. 370-430 CE. Two coins, minims of Valentinian (364367) and Jovian (363-364) help date the use of this area. It was not possible to explore the extent of these features due to the proximity of a major spoil heap, but the building debris hints at the possible presence of a further building to the east of the west wing.

When the new walls were in place, the old exterior wall 667 was demolished to a height of 0.8m; beyond this to the north, hoggin was used for levelling. To the south, the ‘half-cellar’ was infilled with building debris (Figs 5.12 and 5.31) measuring approximately 1m depth in the south, to 0.8m where it was placed against wall 677. The infill was excavated in two blocks on the east and west sides of the room (Fig. 5.12) and consisted of interleaving layers of deliberately dumped material comprising limestone and ceramic roof tiles, building flint, very large amounts of mortar, small amounts of tesserae and painted plaster, and layers of ash, as well as domestic rubbish. The layers appear to have been deposited from the south, suggesting the presence of a doorway in wall 244, possibly in the south-east corner.

Room N5 This room was extended by the construction of a new west wall. This was numbered as 30 in 2012, but as excavation progressed it was apparent that it was the same as wall 149 of Rooms N1 and N2. It shared internal walls with Room N2 to the north (31) and Room N4 to the east (46). Wall 863, which was the south wall of this wing, was poorly preserved with only the foundation layer of flints extant. The wall was absent at its extreme western end, and it is possible that there was a 1.5m entrance here. This Phase 4 room measured approximately 5.2m by 5m internally. The small amount of pottery (38 sherds weighing 361g) suggests occupation in the 4th century CE. A total of about 1,000 loose tesserae were recorded, but there was no evidence for a chalk and mortar under-surface so that these items were probably ploughed in from Room N2.

A number of significant finds were recorded from these layers and included a complete pene hammer (Fig. 15.1/1), a drill bit (Fig. 15.1/5), awl (Fig. 15.2/11) and part of a hub lining. Complete nails and nail fragments were common. Two complete and two fragments of bone pins were recorded. A cuboid-headed example (Fig. 13.12/3) dated to c. 250+ CE and a conical-headed pin (Fig. 13.12/4) broadly dated to c. 200-400 CE and the fragments dating c. 300-370 CE. In total, 1,057 sherds of pottery (9,397g) were retrieved and 97 vessels were identified. Of these, 68 were BB1vessels with date ranges from c. 200-400 CE, including 36 examples of jars, 17 dishes, nine bowls, and sherds from fish dishes, lids, beakers and jugs. Fineware mostly derived from the New Forest potteries and included sherds from 14 vessels (beakers, bowls and a bottle). In addition there were two sherds of Oxford colour coated ware (a bowl and a mortaria) and two sherds from Moselkeramic beakers. The earliest pottery, a single abraded fragment (18g) from a Dr 37 amphora (c. 120-200

The north wing The building was extended east with the addition of a new room (N9) and Room N3 was extended north, as was the gully behind the building. The rooms are considered from west to east. Room N3 Major building work took place here, including the construction of new north and west walls, a probable new roof and substantial raising of the floor level. The 79

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.29. Room N3, Phase 4 wall 420 with limestone quoin 1227, view south

Fig. 5.30. Junction of Phase 3 wall 338 and Phase 4 wall 420 between Rooms N3 and N6, view south

80

The North Range

Fig. 5.31. Room N3, Phase 4 infill layers at bottom of photo. Collapse of chalk floor visible next to top ranging rod. Elevated chalk platform 659 in north-east corner of the room, view north

CE) was located at the base of the infill and the latest, a sherd from a LR1 amphora (5g), came from near the top. It is not possible to determine where this material originated, but it was likely to have derived from building works taking place on the villa at this time. The date range of the finds suggest that the infill was deposited between c. 300 and 370 CE. Three infant burials (229, 666 and 960) were placed in the infill, a further infant (400) was buried in the hoggin layer and a number of discrete skeletal elements were also recorded. These are discussed in Chapter 16. All infants were new-born and were buried around the sides of the room (Fig. 5.32). In the south-west corner, 40% of infant 299 survived; in the south-east corner 70% of infant 666 survived; neither of these inhumations had apparent grave cuts. Infant 960 (Fig. 5.33), however, was placed against the east wall 422 within a well-defined oval grave cut measuring 0.47m long, 0.26m wide and 0.08m deep. The head of the infant was to the north and two flat fragments of limestone appeared to be associated with the burial; 98% of the skeletal elements were present. In the northwest corner, infant 400 (Fig. 5.34) was placed on flint nodules in a shallow (un-numbered) cut with the child’s skull to the east; 90% of this infant survived. There was no dating material with any of the inhumations, which appear to have been carefully placed and were probably buried during Phases 4 and 5. The infill layers were not consolidated as the material was often ‘loose’ when excavated. A thin layer of hoggin (30-

Fig. 5.32. Room N3 plan, Phases 4 and 5

81

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.33. Room N3 infant burial 960 by wall 422, view east

Fig. 5.34. Room N3 infant burial 400 by wall 420, view north

82

The North Range

Fig. 5.35. Room N3, evidence for burning against wall 422 and floor collapse south of wall 667, view east

of the floor described below slumped (Phase 5), part of this deposit (604 and 669) slumped with it.

70mm) was applied over the top layer of infill to marry up with that laid in the northern part of the room. Before the final floor was laid, however, there was evidence for a severe fire on the east side of the room (Fig. 5.35). A burnt area (907) measuring approximately 2m long and 1.2m wide was recorded between the east wall 422 and the stub of the east-west wall 667 the latter showed slight evidence of burning, but wall 422 had been affected and the limestone, flint and hoggin in this wall was burnt bright red Excavation revealed up to 0.2m of burnt material. It would seem that an accidental fire in this area was severe enough to burn wall 422 but caused minimal damage and was quickly dealt with. Deposit 605/700

A chalk floor was then laid over the hoggin and this survived in patches (Fig. 5.31), varying in thickness from 0.05m to 0.08m. Small amounts of degraded opus signinum were noted together with 98 small red tesserae. A total of 5,238 small red tesserae were recorded from various layers and features in this room; they may have derived from a much-degraded plain red mosaic floor, possibly from a room above, or could have been stored here. The presence of small amounts (1,785g) of painted plaster are curious given that this room had a service function; it is possible that they could have decorated a domestic shrine which was commonly placed in such rooms.

After this event, a deposit of grey clay (Fig. 5.32), probably a residue of material used during the building works, was dumped on top of the burnt area 907; the base of the deposit lay on the surviving surface of wall 667. A layer of dark brown ashy loam (604/670) covered the clay and was probably used as a levelling layer. Finds included quantities of building material and BB1 pottery fragments (36 sherds weighing 637g) from jars and dishes, but mostly large C2 storage jars dating c. 370-430 CE, dating this feature to the late 4th century. When the southern part

A platform (659 measuring 1.8m by 1.8m was constructed in the north-east corner of the room and butted up to walls 420 and 422 (Figs 5.31 and 5.32). The edges were defined by well-laid flints set in a cob/chalk mix which raised the structure 0.1m above the floor of the room. The south-west corner was accentuated by the placement of two large, long flints and a slab of burnt heathstone. The rammed chalk floor of the platform may have supported a structure, perhaps an oven; deposits of charcoal were recorded on the surface. The presence of the raised platform may 83

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) room. Plough damage had completely obliterated any evidence for the southern wall and also the southern extent of wall 564. The north-west corner of the walls appeared to be rounded rather than strengthened with limestone as evidenced on the other corners of the building, but it is possible that such building stone could have moved by plough action or robbed-out at a later date. The room was 3.7m wide and would probably have been 5m long, matching other rooms in this wing. The plan (Fig. 5.37/1) illustrates the Phases 4 and 5 features. A 1.5m length of internal wall (789), comprising a 0.4m wide foundation course of flint nodules and limestone pieces, was positioned parallel to the west wall 564. Associated with this was a 0.4m square of yellow mortar (790), embedded in which were occasional small red tesserae and chalk pieces. These features created a space approximately 2m long and 0.7m wide and may have formed the base for a stair to an upper storey or loft. Large quantities of mortar (48,294g), together with 1,597 tesserae, small amounts of opus signinum and painted plaster were recorded in the collapse layers over this room, perhaps pointing to material which had derived from an upper storey.

Fig. 5.36. Painted plaster from pit 614 in Room N10 showing two phases of decoration divided by a plaster skim, the lower green colour is associated with Phase 3 work and the upper yellow with Phase 4. Note the difference in colours with the same piece in Fig. 10.15/2 due to six years in store.

The floor of this room was composed of rammed chalk varying in thickness from 0.15m to 0.15m and was apparent across the width of the room (Fig. 5.38). In places, broken limestone slabs embedded in the chalk suggested that they had been used as flooring. The chalk floor was approximately 0.3m below the Phase 4 floor level in Room N10 and had been constructed over compacted hoggin which was not excavated. The finds were located on, or just above, the floor. The pottery was generally BB1, including sherds from dishes, bowls and a flagon and at least 13 jars. There were substantial numbers of large jar sherds in the late fabric C2; these vessels were not manufactured until c. 370 CE. Two coins were recorded; a minim of Jovian (363-364), and another too corroded to identify.

have been associated with cooking activities suggesting a service room, possibly a kitchen; infants were often buried in such places (Perring 2002, 198). Rooms N6, N7 and N8 were unaltered during this phase. Room N10 A new figured mosaic (474) was laid on opus signinum (see Fig. 5.24), presumably because the bedding for the original mosaic (615) had deteriorated and the floor was too damaged to mend. Only a small part of the mosaic survived in the north-west corner of the room. The north wall 338 was partially rebuilt and merged with a new wall (779) which was associated with construction of Room N9 to the east. The walls were re-plastered and painted in a scheme where yellow was a major colour component (Fig. 5.36), and a new opus signinum quarter round moulding was set between the floor and wall. This refurbished room was likely to have been a major reception area.

Stakeholes and pits A number of features were dug through the chalk floor and are illustrated on Figure 5.38) were located in the central part of the room. All were defined by edgings of small flints, with flints within the dark brown loam fill. They varied in width/diameter from 0.18m to 0.15m and in depth from 0.12m to 0.26m, but were consistent in having steep, almost vertical, sides and narrow rounded bases (Fig. 5.37/2-3). They are likely to have held wooden posts which may have supported a structure or partition.

Room N9

Two pits (674 and 798) assigned to this phase were very different in size but both contained deposits of animal bone (Chapter 17).

The range was extended east by the addition of this room. A new north wall (779) was constructed on hoggin and survived patchily as a single course of flints. It was widest (0.7m) at the north-east corner and narrowed to 0.55m where it merged with wall 338 in Room N10. A new east wall (564) was constructed, but only the foundation layer of flint survived. What had been the exterior east wall (606) of the North Range became the west wall of this

Pit 674 Located near the southern side of the room, this small, bowl-shaped pit measured 0.4m in diameter and was 84

The North Range

Fig. 5.37. Room N9 plan and sections of pits 685, 631, 674 and 798 (1-5)

0.13m deep (Fig. 5.37/4). The mid-brown loam fill (634) contained quantities of animal bone comprising parts of a pig skull, fragments of pig, cattle and sheep/goat. A single sherd of pale green glass in a late Roman fabric was of indeterminate form.

The excavated evidence suggests that this room was built as a service area to Room N10. The pits containing animal bone may have been deliberate deposits similar to the infant burials in Room N3. Corridor N11

Pit 798

There was no evidence that the corridor N11 extended to front Rooms N3 and N9. It gave entrance to Rooms N7, N8 and N10.

This relatively large, vertically sided, pit in the centre of the room was cut into hoggin and measured 0.65m by 0.56m, had a flat base and was 0.35m deep (Fig. 5.37/5). The fill (799) was a consistent dark brown loam with charcoal fragments and occasional lumps of chalk. Two whole sheep and part of a third had been placed on the pit base; some leg bones were arranged vertically against the pit side (Fig. 5.39). A bone sample returned a date of 241-410 cal CE (SUERC- 100055). Prior to excavation it was noted that the pit surface was defined by chunks of creamy-white mortar which may have been a sealing layer. Five small sherds of pottery (13g) broadly dated c. 240400 CE. The size and shape of the pit suggests that it may have had a function relative to the use of the room prior to its final infilling.

Gully 407 Terracing at the rear of the north wing resulted in a gully (407) outside the north wall of the building which can be seen in Figs. 5.3 and 5.4/4; it varied in width from 0.9m to 0.7m and in depth from 0.7m at the west end to 0.25m in the east. The base was flat, and the northern side sloped steeply (Fig. 5.40). It would seem that the gully was not cleaned out regularly, as building debris and domestic rubbish accumulated over the lifetime of the building. The pottery dated from c. 200-430 CE and was composed mainly of BB1 jars, bowls and dishes. 85

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.38. Room N9 with chalk floor, remnants of limestone slabs and post pad 790. Plough damage to wall 564 and southern extent of floor apparent, view north

Fig. 5.39. Room N9, sheep carcases placed on the base and against the sides of pit 798

86

The North Range

Fig. 5.40. Gully 407 behind the North Range. Photograph taken in September 2014 during a BBC1 TV broadcast, view east

Conclusions

In the north-east corner of the room, a quadrant-shaped area (155) had been dug into the mosaic (Fig. 5.13). A 2mm layer of fine charcoal and ash appeared to be the remnants of a fire, as heat from this had reddened the walls above. A nummus (388-395) of the House of Theodosius was recovered, suggesting that this activity took place at the very end of the 4th or early in the 5th century.

Major building work on this range resulted in a twowinged, L-shaped building. Each wing may have had its own entrance; Porch N4 for the West Wing and Corridor N11 for the North Wing. The presence of three reception rooms, two anterooms and two, or even three, service areas possibly suggest differing functions for each wing. Positive evidence for mosaic floors was recorded in all rooms except N3, N5 and N9, which may have been related to their function as service rooms.

Relatively little pottery was recovered in the layer above the mosaic (94 sherds weighing 1,491g) and comprised BB1 fragments from seven jars and two dishes (c. 300430 CE). Two pieces of glass from the neck of a blue/ green bottle and part of a plain copper alloy bracelet (Fig. 14.4/2) were recorded.

Phase 5: Final Roman Period 5 (c. 350/370-430 CE) No further building work was undertaken, but occupation is implicit in small-scale alterations and other interventions as well as in the material finds, particularly pottery and coins. The building is considered as a whole and rooms are only discussed if late occupation is evident.

The room appeared to have been abandoned for some time as a layer of owl pellets consisting of small animal bones was present in the north-west corner (Chapter 17). The birds were probably barn owls roosting on cross beams of a vaulted roof; their presence implies that people had largely abandoned this space. Two bones were submitted for radiocarbon dating and returned dates (SUERC-62809 and 62810) in the mid-4th century which were inconsistent with the finds evidence.

Room N1 At least two phases of mosaic repairs were undertaken on the extreme western side of the room (Fig. 5.13). The earliest attempted to follow the original blue lias band, but either the stone type was unavailable or it was deemed not necessary to match the design. Four rows of small white and brown tesserae were used. Behind these, further areas were patched using much larger tesserae, the majority of white limestone with occasional ceramic tesserae. It is not possible to determine the timescale for these repairs.

Room N3 In the last quarter of the 4th century there was a dramatic slump of the floor over the infill layers. This was due to the ‘loose’ nature of the infill materials which were not consolidated as they were being deposited, resulting in 87

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) its edges. Immediately to the south of this, deposit 515 was similar, measuring 0.5m by 0.4m. Associated pottery comprised 27 sherds weighing 294g from a BB1 dish, a C2 store jar and fragments from a South Western Black Burnished ware bowl dating this activity to c. 370-430 CE. A similar deposit (588) was located centrally in the room and produced sherds from a BB1 bowl (c. 370-400 CE).

a 0.2m drop in floor level in the south part of the room (see Fig. 5.31). This was remedied by adding a layer of building material and ash, and re-flooring with chalk. Animal head deposit (Fig. 5.41) A cow skull (577) was placed on the chalk floor over a small deposit of red tesserae at the north end of the room, halfway between the east and west walls. The horns had been broken off, a large fragment of ceramic roof tile had been inserted into the skull cavity and two further pieces were placed by the skull (suggesting a mouth). A large Type 2 iron nail lay beside the ‘mouth’ and an exceptionally large cockle shell had been placed on the skull in the eye socket. A sherd of BB1 cooking jar dates this deposit to c. 300-430 CE, with the likelihood is that it belongs to the end of this date range. A bone sample returned a date of 241-410 cal CE (SUERC-100060).

This room continued in use until the building collapsed. Material finds in the layers below this and above the chalk floor included 478 sherds of pottery weighing 6,957g comprising BB1 jars, dishes, bowls, a lid and tankard. Large storage jars in fabric C2 were well-represented and there were occasional sherds of New Forest and Oxfordshire beakers and bowls. The pottery dated c. 370 to 430+ CE. A minim of Valentinian I (364-375), a copper alloy ring (Fig. 14.5/4) and a bracelet fragment (Fig. 14.4/6) broadly dated c. 240-410 CE. The dates suggest that occupation continued into the first third of the 5th century.

Charcoal deposits

Room N7

Two discrete areas of charcoal and ash with a depth of 40mm were located in the north-east corner of the room directly above the chalk floor (Fig. 5.32). Tight against the corner, deposit 483 measured about 0.5m by 0.5m and was defined by a loose arrangement of flint nodules around

The mosaic pavement in this room displayed evidence of crude repairs on the north edge consisting of fragments of limestone roof tile randomly placed where the tesserae had failed (Fig. 5.23), although this cannot be dated precisely.

Fig. 5.41. Room N3, Period 5 deposit 577 of cattle skull adorned with a cockle shell, iron nail and ceramic tile fragments, view north

88

The North Range Room N10 (plan Fig. 5.42)

To the north and south were two similar fills (936 and 942) of sticky ginger-brown clay/loam, similar to hoggin, but differing by the presence of building material, particularly mortar. The base of the pit consisted of a 0.08-0.1m layer (937) of soft ginger-brown clay/loam containing frequent small gravels and patches of mortar, tesserae and painted plaster, which continued under fill 936 to the north. The pit overlay the Phase 1 posthole 952 (see Fig. 5.10/2), and within its upper fill (943) were quantities of tesserae, opus signinum and mortar capped by a layer of flint, which are likely to have been deliberately inserted for consolidation.

The devastation caused by ploughing removed evidence for any possible repairs to the mosaic floor 475. A series of pits and deposits were recorded, and material from the fills strongly suggests that they were in use during the final phase of occupation. Pits from the southern part of the room had been truncated by ploughing. Pit 818 (Fig. 5.43/1 and 5.44) This feature comprised an area of burnt flint and loam in the south-east corner of the room; excavation revealed a moderately steep-sided oval pit with a flat base measuring 1m long, 0.5m wide and 0.16m deep. Much of the surface was covered with a clay/chalk mix. The charcoal-rich loam fill (822) contained burnt and fractured flint (4,550g). A large piece of burnt limestone appeared to divide the pit into two. Probable butchery waste including pig, sheep and sheep/goat bone was recorded in the southern half; a radiocarbon date of 216-362 cal CE (SUERC-100059) was returned from a sheep femur. Disturbed opus signinum and tesserae were present around the northern edges where the pit had cut through the floor. Three small sherds of pottery dated c. 340-370 CE. The base of a colourless glass cylindrical cup dated to the 2nd/early 3rd century, and the animal bone may have been associated with earlier activities in this room.

The reasons for digging this large pit are irretrievable, but it is likely that the room was intact when the event occurred. It contained debris from a collapsed or demolished wall, tesserae from two mosaic pavements and painted plaster from two schemes of decoration. It is possible that the mortar-rich layer 877 sealed the pit, which appeared to be the final episode of occupation in this room. Room N9 (Fig. 5.37/1) A layer of fine loam accumulated below the collapsed roof in this room and was recorded as two contexts, 622 to the north and 626 to the south. Pottery fragments, which may have been related to contemporary activities, were recorded (40 sherds weighing 350g) dated c. 340-430 CE. Four pits were dug through the base of this accumulated soil.

Deposit 595 (Fig. 5.45)

Pit 590 (Fig. 5.47/1)

Consisting of broken and decomposing ceramic roof tiles embedded in red clay and measuring 0.9m by 0.65m, the deposit may have formed a base for an undefined structure above.

This shallow pit was dug into hoggin against the west wall (606) of the room. It measured 0.8m in diameter, was 0.1m deep, with edges defined by burnt limestone, heathstone and flint. The flat base and gently sloping sides had been affected by heat and were burnt bright red. The fill (613) of charcoal-specked, slightly sticky dark brown loam contained large fragments of burnt stone and shattered burnt flint. Two sherds from a BB1 cooking pot broadly dated c. 250-430 CE. It is possible that the feature had been used as a hearth.

Pit 614 (Figs 5.43/2 and 5.46) Located in the north-east corner of the room, this large pit was covered by a deposit of building flint and ceramic roof tile spread over an area approximately 1.5m wide and 2m long. A section (C-D) was excavated through this, revealing a vertically sided pit 0.5m deep with five discrete fills, which had been dug through the upper and lower mosaics. The phase 1 posthole (952) was sealed by the base of the pit. Much of the pit surface was covered with a 0.1m layer of mortar in brown loam (877) which may have derived from adjacent decaying walls. The major fill (903) in the centre of the pit was a dense concentration of building material (flint, mortar and limestone and ceramic roof tile fragments) in dark brown loam. Large numbers of tesserae (641), some joined together, and painted plaster (7,846g) were recorded, as well as fragments of quarter round moulding and opus signinum.

Pit 632 (Fig. 5.47/2) Located by the north wall 799, this small, oval pit measured 0.7m by 0.43m, was 0.22m deep with steeply sloping sides and an uneven base and was completely covered with limestone tiles from the collapsed roof. The dark brown loam fill (640) contained three roofing nails, two tiny animal bone fragments and sherds from two jars, one in BB1fabric, the other in New Forest Greyware dating c. 260-400 CE. Pit 633 (Fig. 5.47/3)

A small pottery assemblage (10 sherds weighing 145g) included fragments from a BB1 cooking pot, an Oxfordshire type C99 mortarium and a sherd (18g) of post-Roman handmade ES1fabric. The pit fill is dated c. 450-650 CE by this latter sherd.

This small bowl-shaped, oval pit was dug though the chalk floor 631, measured 0.7m by 0.45m and was 0.12m deep. The fill (641) of dark brown, charcoal-rich loam contained large numbers of limestone roof tile fragments. A small 89

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.42. Room N10 plan

Fig. 5.43. Room N10 pits 818 and 614 (1-2)

90

The North Range

Fig. 5.44. Room N10, pit 818 half section with deposit of burnt stone, view south

Fig. 5.45. Room N10, deposit 595 of broken ceramic roof tiles. Note loose tesserae in trays from cleaning back, view north

91

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.46. Room N10, pit 614 half section during excavation, showing dump of building material, view west

assemblage of ironwork included a barb-spring padlock bolt, part of a small pair of cloth-working shears (Fig. 15.2/9), binding and plate fragments and roofing nails, which were all located near the top of the pit. Pottery (22 sherds weighing 178g) comprised fragments from two BB1 jars dated c. 200-430 CE.

recorded, together with a ceramic spindle whorl (Fig. 13.19/5), and pottery (68 sherds weighing 879g) which included BB1 jars, bowls and dishes and an Oxfordshire flagon fragment dating c. 350-400 CE. A nummus of the House of Theodosius dated 388-402 is one of the latest coins on the site.

Pit 680 (Fig. 5.47/4)

Animal bone deposits

Located in the centre of the room, this undated bowlshaped pit cut through the chalk floor and measured 0.6m by 0.5m and was 0.23m deep. The fill (681) of grey, silty loam contained occasional small gravels, chalk fragments, some decomposing mortar, and small pieces of limestone.

Two small deposits of cattle skull bones were recorded: deposit 853 was placed against the south wall (750) about 2m from the corridor’s main entrance and deposit 981 was placed centrally in the entrance. When the corridor roof collapsed a layer of loam accumulated over the floor, within this were material finds dating to the final phase of occupation. The layer was consistent along its 15m length, which was excavated in four sections. One of these sections included layer 812 above scoop 791, which produced the highest numbers of finds and the largest amounts of building material. Finds however, were strewn along the length of the corridor, the majority located at the west end and lessening in quantity towards the east. A total of 2,479 assorted tesserae and lumps of opus signinum were recorded. suggesting that the floor was disintegrating at this time. Animal bone fragments (pig, cattle, sheep/goat and horse teeth) were present. Glass from a pale green conical beaker and a yellow/green cup were Late Roman in date. The pottery

Room N11 Three features dating to this phase were associated with the corridor. Scoop 791 (Fig. 5.48/1) This extremely shallow feature measured 1.3m by 1.1m by 0.08m and was located at the extreme west end of the corridor directly under the rubble of the roof collapse and between north wall 441 and south wall 750. The fill (792) of dark brown, charcoal-rich loam contained pieces of limestone roof tile and painted plaster, probably derived from the walls. A small deposit of sheep/goat bone was 92

The North Range

Fig. 5.47. Room N9 sections of Phase 5 pits 590, 632, 633, 680 (1-4)

Phase 6: Post-Roman Period 6 (c. 430-650 CE) (Fig. 5.4 Phase 5)

assemblage (529 sherds weighing 7,485g) was dominated by BB1 jars, dishes and bowls. The presence of C2 storage jars confirms that the material dates c. 370-430 CE. Bowls and beakers from Oxfordshire and New Forest kilns were also recorded. A nummus of Constantine II (317-340) was retrieved from layer 812. A fragment of architectural moulding (Fig. 9.32) came from the west end of the corridor.

A final phase of activity was evident in front of Room N3, after which the building was abandoned. The heavy structural roof timbers would have rotted, resulting in catastrophic collapse of the limestone tiles. This would have been followed by the disintegration and fall of the mortared flint walls. Where it was visible, the sequence is discussed below. The presence of collapsed roof and flint walls was recorded in the earliest excavations in Trench 2 in 2012; the true significance of this, however, was not realised until the following year when the remains of walls and rooms were uncovered. The North Range rooms were not consistently covered with collapse debris, which appeared to depend on the height of wall survival,

Conclusions The final years of occupation of the North Range consisted of basic maintenance with the building eventually sliding into decrepitude, resulting in hugely reduced living standards. It is probable that, at this time, only part of the range was occupied. 93

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.48. Corridor N11 section of scoop 791 under collapse layer and fallen wall 1073, east of Room N4 (1-2)

produced re-deposited material. The pottery assemblage broadly dated c. 370-430 CE and was dominated by C2 jars. However, two sherds of LR1 amphora (39g) dating c. 400/450-550/650 CE from layers 2 and 3 suggest that the material was moved during the Phase 6 period. Two coins were recorded in layer 2 (a minim of Constantine II (317-340) and another too corroded for identification. Two further minims came from layer 3. One of Constantine II and one of Valentinian I (364-375). The coins were probably within the rubble used to infill the area and not losses associated with its deposition.

the room floor levels and subsequent plough damage; the latter can be seen in Figures 5.3 and 5.17. Activity south of Room N3 The east-west wall 244 in this room was the only wall to survive from the original Phase 1 building. It would appear that most of this was demolished to a height of approximately 0.7m, but the eastern 2m was cut in a series of five steps (Fig. 5.49); the upper steps were about 0.1m in height, but the lowest step was only 0.03m high and underneath this was the chalk of the original terracing. The steps varied in length from 0.65m to 0.18m. That the removal of this wall was a very late action is confirmed by the presence of the Phase 4 infill north of the wall and within Room N3 but absent over the wall and to the south of it.

The re-deposited material may have been used to level-up or could have been dumped in an already hollow area; it was similar in composition to that in the late-occupation pits in the Aisled Hall (Chapter 6) and could have derived from the same source.

An area measuring approximately 5m wide and 3m long immediately south of Room N3 and east of corridor N4 was investigated in spits between 2013 and 2015. Very large quantities of building material accumulated to a depth of approximately 0.60m over and to the south of the wall remnants (A on Fig. 5.49); some of this spilled into Room N3 where the wall had been removed. To aid removal of the material, the area was excavated in three spits, all of which

Room N1 This was the best-preserved room in the range, largely because the external walls (120 and 149) survived to heights of 0.42m and 0.39m, despite there being only 0.2m of topsoil over the archaeological layers. The accumulated material over mosaic 153 was excavated in three layers. The upper layer which had been affected by ploughing 94

The North Range

Fig. 5.49. Room N3, Phase 6 steps cut into wall 244, infill layers north of the steps and accumulation A (south of Room N3) to the south, view west

and 119), approximately 0.5m above the final floor level. The sequence was similar to that in Room N1, with flint and mortar from the walls covering limestone roof tiles (Fig. 5.52). There were up to ten layers of roof tile against the north wall (Fig. 5.53); 23 of the tiles were unbroken. Associated pottery was dominated by C2 storage jars dating c. 370-430 CE. The section of gully 407 to the north of Room N3 also contained very large amounts of roofing material.

had a depth of 0.1 to 0.15m and largely consisted of flints from the collapsed wall, but limestone building stone and roof tiles were also present. The 0.1m to 0.2m deep middle layer of mid-brown loam contained very large amounts of roofing material. There was approximately 428kg of limestone tiles, many of which were intact with some containing their fixing nails, and 43kg of ceramic tile. In addition, 17kg of mortar and 34kg of painted plaster which derived from the walls was recorded. The lowest layer was 0.2m deep and also contained collapse debris. Even more roofing material was recorded in this layer: 735.5kg of limestone tiles and 103.5kg of ceramic tile fragments. Mortar (156kg) and painted plaster (22kg) were also present, as well as fragments of sandstone (2,320g) and slate roof tile (1,096g). In the north-east corner of the room, Figure 5.50 shows the fallen limestone tiles on the top of the second layer and Figure 5.51 shows mosaic 153 as it was uncovered, with the lowest layer of collapse lying directly on its surface. Mortar, painted plaster and flint from the fallen walls are visible.

Room N9 This floor of this room at the east end of the range was at a lower level than the reception rooms in the centre of the range resulting in preservation of the collapse, which was covered with a 0.15m clearance layer of dark brown loam and building debris. Intriguingly, 1,583 tesserae were recorded from this layer, possibly moved across from Room N10 by plough action. A 0.2m to 0.3m layer of limestone roof tiles (566) covered the northern two-thirds of the room (Fig. 5.55). Over this, on the western side of the room and underneath the clearance, was a 1.3m wide strip of fallen wall flint and associated mortar. A single small sherd (3g) of post-Roman pottery from the clearance dates the collapse to c. 450-650 CE.

Room N3 Preservation of the final collapse in this room was due to the surviving height of the west and north walls (420 95

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 5.50. Room N1, north-east corner: collapsed limestone roof tiles after removal of building flint, view south

Fig. 5.51. Room N1, collapse debris over mosaic 153. Note fallen painted plaster and the limestone roof tile with in situ nail at top of photo. Wall 120 in the foreground, view south

96

The North Range

Fig. 5.52 North-east corner of Room N3, collapsed wall flint and limestone roof tiles, view south

Fig. 5.53. Room N3, limestone roof tile deposit 513, view north

97

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Corridor N11

(Fig. 5.47/2) revealed three layers of flint in decomposing yellow mortar; underneath was a deposit of limestone roof tiles (1143), associated pottery dated c. 370-430 CE. The location of the east-west section through the wall can be seen on Figure 5.16. This feature may have represented another room or small building, but the presence of a large spoil heap located south of the North Range prevented further exploration.

The corridor floor was lower than that of the reception rooms to the north resulting in preservation of the collapse, particularly its western extent. The eastern end of the corridor had been destroyed by ploughing. In total, approximately 450kg of limestone roof tile, 195.5kg of ceramic roof tile and 1kg of sandstone roof tile was removed during excavation. The collapse was also recorded 1.5m in front of the corridor’s central entrance and extended 1m either side of the 2m doorway (Fig. 5.56). Large amounts of building material were removed including 143kg of limestone tile, 1.8kg of ceramic tile and 4.5kg of sandstone tile. The pottery assemblages dated c. 370-430 CE.

South of Corridor N11 A 1.1m wide trench was cleared immediately south of external wall 750. Building and roofing material was evident along its length commensurate with collapse of the corridor wall and roof, while associated pottery was dated c. 260-430 CE. Five coins were retrieved, one of which was too corroded to date. The earliest was an antoninianus of Tetricus I (271-274), two minims of the House of Valentinian (364-367) and of Gratian (367-375) and the latest coin was a nummus of the House of Theodosius (395). A fragment of decorated stone column (SF 438) is discussed in Chapter 9. Small finds included a flat lead weight (SF 17) and a slate gaming counter (SF 419). A copper alloy military belt fitting (Figs 14.11/3 and 14.12) dates to the late 4th/early 5th century.

Collapse south of the North Range Initial excavation of Trench 2 in 2012 noted collapse sequences extending for almost 18m from wall 44 on the eastern side of Room N4; subsequent work confirmed that these were located approximately 2m in front of the North range. The collapse consisted of areas of fallen wall and roofing which extended about 5m further east of the corridor central entrance (Fig 5.4 Phase 5). Fallen wall 1073

Conclusions

A 2.8m by 3m clearance to the east of Room N4 revealed a further area of collapse (Fig. 5.56); this was a southern extension of similar material recorded in 2012. A section

That occupation continued from the 5th, and possibly into the 7th, century is implicit in the small assemblages of very

Fig. 5.54. Room N9, limestone roof tile collapse, wall 606 left, view southwest

98

The North Range

Fig. 5.55. Corridor N11 with collapse of limestone, sandstone and ceramic roof tiles over the porch. Section through preRoman ditch 878 to the right of the wheelbarrow, view west

Fig. 5.56. South of Room N3, east of Rooms N2 and N4, fallen wall 1073 to the left, view north-west

99

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) late Roman pottery vessels, Late Roman 1 amphorae, postRoman coarse pottery and metalwork. Living conditions were greatly reduced but the acquisition of wine and luxury goods was still possible. The beginning and end of this phase are not easy to define, but the collapse of walls and roofs ultimately rendered the building uninhabitable. Over the centuries, good accessible building stone would have been removed. The eventual assimilation of the site into agriculture resulted in the remains of the building being covered in soil. General conclusions The development of a relatively high-class domestic building over three or four centuries displayed characteristics integral to a Romano-British lifestyle, by a series of high-status occupants who conformed to the fashions and conventions of the time and who, importantly, had the finances to undertake such changes. The phasing of this building, however, has to be ‘best-guess’, based on the excavated structural remains and the associated material finds.

100

6 The East Range to have been part of an internal wall which may have incorporated a doorway. This would have created a narrow room, possibly a vestibule, at the south end of the building. The main entrance could have been here, but severe plough damage removed evidence for such a feature. The levelled natural chalk bedrock appeared to have been utilised as a floor (852, 1043, 1206 and 1392), and in places charcoal and clay had been trampled in.

Excavation in this area began in 2013 and was completed in 2016. In 2013, at an early stage in the work and after discussions with academic colleagues, it was decided that excavation in the northern part of the structure would be minimal due to the extent of fallen wall which survived there. As a result, information regarding earlier structural elements in this area was limited. The buildings were constructed on a sloping site with a 1m difference in elevation from the north to the south and were uncovered in stages by opening up and extending Trenches 6, 7, 16, 17, 18, 26, 31, 34 and 43 (see Fig. 1.6 for trench locations). Between 0.1m and 0.15m of topsoil was removed by mechanical excavator and the exposed areas were trowelled by hand until stratified deposits were encountered. Insights gained during excavation, and in scrutiny of the records and plans during post-excavation work, revealed four major building phases and evidence for the ultimate collapse of the structure. These phases (1-4) are illustrated on Figure 6.1. In addition, two later phases of activity (5-6) are discussed.

Underneath wall 1220 were 27 sherds of pottery (130g) comprising a BB1 lid and jar and fragments of Exeter Fortress ware with dates of c. 43-120 CE, suggesting a construction date sometime around 100 CE. Features relating to this phase Oven cluster 1045, 1081 and 1091 and oven 985 Remnants of three similar ovens were located in the northern half of the building (Fig. 6.2/1); these had been almost obliterated by the construction of the Phase 3 flint pit 648 and only the lowest elements survived.

Phase 1: Early Roman, Period 2.2 (c. 70-120 CE)

Oven cluster 1045, 1081 and 1091 (Figs 6.5/1-2)

The first rectangular building

Sub-circular oven 1045 was set in a pit cut into chalk and was the best preserved of the three (Fig. 6.3) and was probably the latest in the sequence. The feature had gently sloping sides, a slightly rounded base, an external diameter of 0.7m and a surviving depth of 0.22m. Seven discrete fills were recorded. The oven structure of chalk with clay (1063) formed a bowl, lined with a thin layer of heat-affected red clay (1064). The base of the oven was filled with ashy silt and charcoal (1072) to a depth of 0.04m, probably the remnants of the last firing. Over this was a very thin layer of re-deposited chalk (1071). The major fill of the oven (1065) consisted of broken tegulae and imbrices and a fragment of quern (SF 530). Covering this deposit was a layer of ashy loam and charcoal (1060). Emmer/spelt wheat was present; a grain was submitted for radiocarbon dating and returned dates of 129-254 cal CE (SUERC-100062). The presence of the quern confirms the processing of grain at this time. Charred weld and flax seeds indicate activities associated with cloth dyeing.

The building (Fig. 6.2/1) was constructed c. 100 CE, on a site which was levelled to create a flat platform. This was visible behind the north wall 1202 where the natural chalk had been quarried out to a depth of 0.4m. Externally, the building measured 17.5m long by 9m wide and internally 16.2m by 7m. The western wall 1210 survived as seven courses of flint in mortar to a height of 0.5m, was c. 0.8m wide and constructed on a 1m wide foundation course. Only 2m of the south wall 1219 survived where it was bonded to the west wall. Fragments of the north wall 1202 were visible in a single intervention (Fig. 6.2/2 A-B) and survived as a 0.5m wide layer of flints in pale grey mortar laid on chalk. This wall appeared to have been dismantled for the northern extension of the Phase 2 building. Two small, vertically sided, square postholes were located on the line of the north wall. Posthole 914 was 0.11m by 0.11m and 0.15m deep, with a complete ceramic roof tile on its base, and posthole 1187 was 0.2m by 0.2m and 0.18m deep, with a dressed limestone padstone (1119) on its base (Fig. 6.3). These postholes may have contained timber uprights which related to wall 1202, and possibly mark a 2m wide doorway. The east wall 1220 was visible where a 1m wide single course of flints underlay the Phase 3 east-west wall 713, and remnants survived in slot 1310 and in context 820. Fragments of 0.66m wide flint bases for two internal partitions (1209 and 1193) are likely

Oven 1081 was only partially excavated but had a similar form to 1045 with chalk and clay walls lined with red, heat-affected, clay. Its external diameter was 0.58m. The fill (1083) consisted of ashy loam and charcoal which contained fragments of mortar, limestone and sandstone roof tiles and roofing nails.

101

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 6.1. East Range/Aisled Hall building Phases 1-4

Oven 1091 was the earliest and was severely compromised by the construction of ovens 1045 and 1081. It was not fully excavated but resembled the later structures in form and fill.

Pit 1182 (Figs 6.2/1, 6.6-7)

Located half-way down the building towards the east wall, only the base of oven 985 survived with a scant 0.13m of structural material remaining. The remains comprised a horseshoe-shaped structure (986) of burnt cob, clay, flint and limestone measuring 0.8m long and 0.75m wide.

Constructed immediately outside the south-east corner of the building, this sub-circular, almost vertically-sided, pit cut into the natural chalk measured approximately 1.6m in diameter on the surface and at a depth of 3m the diameter was 0.96m. The depth was in excess of 3.2m (Fig. 6.7) when excavation was abandoned; probing with metal rods confirmed at least a further 1m of filling. It is possible that the pit was initially dug as a well and was later used as a convenient rubbish dump.

The oven cluster (1045, 1081 and 1091) was destroyed by the insertion of the Phase 3 flint pit 648. Five undated sherds (27g) of pottery were recovered, but the radiocarbon result places use of the ovens within the Phase 1 building. There was no dating evidence for oven 985, but its size and form suggest that it is of similar date to the oven cluster. It is likely that all the ovens had a dual domestic/ semi-industrial function. In form they resemble oven 992 in Room 5 of the North Range, which had a similar date.

Ten discrete fills were recorded (Fig. 6.6), with evidence for a series of re-cutting episodes. Three fills (1226, 1225 and 1224) continued into the unexcavated depth and had been compromised by later re-cutting. Fill 1226, visible to a depth of 2.21m on the pit side, was a green/grey silty loam, indicative of cess; its surface was originally 0.85m below the Roman ground surface. Next to this, fill 1225, a layer of red clay, may have been lining for a re-cut. Over this was a sterile layer of chalk with small amounts of clay (1224)

Oven 985 (Fig. 6.5/3)

102

The East Range

Fig. 6.2. Plan Phase 1 rectangular building and section through walls 1174 and 1202

which may also relate to the re-cut lining. Pottery from the loam (1223) of the re-cut had similar dates (c. 180-200 CE) to material from 1225 and 1226 suggesting that infilling took place over a relatively short period of time. Fill 1215, which was recorded on either side of the pit, must have originally formed a layer 1.6m deep and comprised sandy clay loam; pottery dated the fill to c. 210-250 CE. Later fill 1222 was dug into 1215 and contained moderate amounts of building debris and one small fragment of window glass. Layers 1215 and 1222 were capped with a compacted layer of chalk (1221) which appeared to have slumped in the centre. The upper three fills (1214, 1213 and 1212)

consisted of very dark brown crumbly loam and comprised separate dumping events. Layer 1214 was characterised by the presence of at least 50% charcoal; the pottery dated to c. 120-300 CE. The composition of the two top fills (1212 and 1213) was similar but the latter lacked the high density of building flint. Animal bone was recorded in eight of the ten fills, mostly cattle, sheep/goat and pig, but roe and red deer were also noted. Birds included crow, woodcock, songbirds and duck as well as chicken. Of note was the presence of sandstone roof tile fragments; this material generally related to the North Range roof but is otherwise rare on the site. 103

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 6.3. Posthole 1187 with limestone pad 1191 on the base, view north

The west, south and east walls were retained, but the north wall was demolished and replaced by wall 1174 directly to the north of wall 1202. The base was set 0.15m higher than the earlier wall; it was 0.6m wide and composed of tightly packed flint nodules in grey/yellow mortar (Fig. 6.2/2). It is possible that this wall was constructed to strengthen wall 1202. It is likely that both potential entrances continued in use. Patches of opus signinum floor (1205 and 1257) were recorded.

Pottery from the upper fills dates to c. 260-300 CE and from the lowest excavated fill to c. 180-220 CE. However, the earliest pottery (South-West Black Burnished ware) dates from c. 43-250 CE. A fragment of window glass is 1st/2nd century in date. It is likely that the unexcavated lower layers would provide earlier dates commensurate with Phase 1 activity. Conclusion Contemporary with first phase of the North Range, this building may have had dual semi-domestic and industrial functions, suggested by the presence of ovens and the putative well.

Tank 778 and associated features A rectangular tank (778) measuring 2.6m by 1.75m was constructed towards the southern end of the building, centred between the east and west walls, and stood proud of the floor (Fig. 6.9). The 0.30m wide walls were built of welllaid flint nodules, two wide, cemented with sandy mortar. The base of the tank sloped, the internal depth varying from 0.46m on the west to 0.58m on the east. Externally the top of the tank was 0.34m above the floor. It was lined with a thin layer (0.08m) of opus signinum which was probably applied as waterproofing. There was no evidence for an outlet, but the tank was probably constructed to hold water; its location directly in front of the potential south entrance may have been significant. To the east side of the tank, a segment of compacted chalk (932) may have supported a stone step. The tank was filled in when the building was

Phase 2: Middle Roman, Period 3.1 (c. 120-200 CE) (Fig. 6.8) Extension of the rectangular building and associated features The building was extended northwards; an opus signinum floor was laid and a tank-like feature (778) was installed. Outside the building, pit 1182 continued in use; a similar pit (1149/1181 was dug just north of this. A further deep pit (381) was dug to the north-east of the building and two other pits (897 and 209) were located west of this. 104

The East Range

Fig. 6.4. Oven 1045 half sectioned, view north

fully excavated and measured 1.8m by 1.66m by 1.8m deep. A gradual infilling sequence was suggested with domestic rubbish and building material (not necessarily from this range) thrown in from either side of the pit. Eight fills were recorded. The undated primary fill (451 and 452) was grey/brown silty loam. Overlying this, the greenish clay fill 446 was indicative of cess. The intrusion of fills (445 and 438) into 446 may be evidence for partial cleaning out. Fill 445 was also cess-like and was notable for the quantity of animal bone which included cattle mandibles, sheep/goat skulls and feet, chicken, curlew, two wood pigeons and a rib from a neonate infant. A discrete deposit (444) on the base of the pit included a cattle skull, sheep/goat feet, puppy bones and small bones from a song bird, crow and pigeon. Three tiny perinatal phalanges were recorded and may have been associated with the infant from fill 445. The sole of a shoe re-enforced with iron rivets (Fig. 15.9/78) was associated with the bones. Pottery dated the fill to c. 150-200 CE. Above this, clay loam fill 438 appeared to be of the same date. Fills 396 and 416 comprised grey silty loam containing large amounts of building material spread across the entire pit. Metal finds included the base of a seal box (Fig. 14.9/14) dating to the 2nd/3rd century and copper alloy scrap. The pottery broadly dated c. 200-

altered for the Phase 4 work which was probably instigated by extensive fire damage. The tank was half-sectioned (Figs 6.10 and 6.11/1). The fill (627) of mid/dark brown loam contained large amounts of roofing material including 159 limestone roof tile fragments (120,617g), 4,573g of ceramic roof tile plus hundreds of highly fragmented pieces, 25 roofing nails, 17,253g of mortar, large chunks of building stone (some of which were burnt) and seven tesserae. This material must have derived from the demolished Phase 3 building. Animal bone including elements from at least four sheep, a few cattle bones and a single woodcock bone were found in the south-west corner of the tank. Sixteen sherds of pottery from a BB1 beaded and flanged bowl, dated c. 300+ CE were located near the top of the fill suggesting an early 4th century date for the filling of the tank, which was probably due to the raising of the floor level for the Phase 4 building and the need for a solid surface. A circular chalk feature 933, 0.42m in diameter and located to the northwest, was set into floor 984 and was probably associated with the tank. Pit 381 (Figs 6.11/2 and 6.12). Cut into chalk and located 5.5m north of the building, this rectangular, vertically sided, flat-bottomed pit was 105

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 6.5. Plan of oven cluster 985, 1091 and 1081 (1), section through oven 1045 (2) and plan of oven 985 (3)

Fig. 6.6. Pit 1182 section

excavation revealed that this was cut into an earlier pit (1181). The profile and fills suggest that this may have been dug as a cess pit but was ultimately used for the dumping of household rubbish and building debris.

300 CE. The upper fills appear to have been dug through layer 396/416; the lowest of these (395) consisted of 75% yellow sandy mortar, probably dumped from building work. Within this was a Gillam Type 15 jar (c. 90-150 CE). The top-most fills (382 and 383) in the centre of the pit were similar dark brown clay loam; 383 contained quantities of charcoal. Large amounts of building material were present and the pottery was closely dated to c. 260-280/90 CE. Eight fragments of infant bone from fills 383, 395 and 445 (which also contained large amounts of building material) suggest disturbed burials from renovation works on the villa. A thin spread of clay (380) may have been a capping layer. Above this, a clay/ gravel surface (379) was laid after the pit went out of use. Pottery and a seal box from near the base of the pit gave dates of c. 200-300 CE, which accords with the probable timeframe for the Phase 2 building. The pottery (Assemblage 10) is discussed in Chapter 12.

The earlier, vertically sided pit 1181 measured 2m by 1.3m and was excavated to a depth of 1.3m; auguring confirmed that the fills continued for at least a further 1m. The mid-brown loam fill 1175 contained building debris including roofing and building material. Animal bone (42 fragments) comprised sheep/goat, pig, chicken, duck and four fragments of red deer bone, including a limb element with cut marks. BB1 pottery (50 sherds weighing 476g) included jars, bowls, dishes and fragments of jug and beaker dating c. 270-300 CE. To the western side, fill 1180 comprised consolidated mortar with occasional small lumps of raw clay. The upper fill (1184) was a sterile mid-brown loam, which may have been a capping.

Pit 1149 /1181 (Figs 6.13 and 6.14/1)

The steep-sided, round-bottomed pit 1149 was dug into fill 1184. It measured 0.69m in diameter with a depth of 0.7m and was composed of two layers. The lower charcoalrich fill (1170) contained building debris and one sherd

Located at the very end of the 2015 season and only halfsectioned, this pit was sited outside the south-east corner of the building. Initially identified as small pit 1149, 106

The East Range

Fig. 6.7. Pit 1182 at the end of excavation, view south

pit 896. Eleven layers were recorded (not all are located on the recorded section). It was disturbed by later features, including the Phase 4 oven 207 and Phase 6 pit digging and, as a result, the profile only survived on the western side. However, only the lowest fill (287) belonged to the original pit; this consisted of weathered chalk on a flat base measuring c. 1.3m in diameter. Pottery comprised 10 sherds (68g) of BB1 dating c. 200-350 CE. Pottery from fills above this correlate with the Phase 6 pit digging which is discussed under that phase. The pit would appear to have been similar in size to pit 381.

from a BB1 bowl dated c. 290-370 CE. Building debris (including eight red tesserae) was also present in the dark brown loam upper fill (1150). Links from an iron chain (Fig. 15.9/87) are likely to have derived from a domestic utensil. The animal bone assemblage included sheep/goat, pig and duck as well as dog and a single crow bone. Pottery (79 sherds weighing 855g) included BB1 jars, bowls and dishes (c. 270-300 CE) and a New Forest indented beaker (c. 260-400 CE); the dates suggest that the pit was in use in the final years of the Phase 2 building. Pit 896

Phase 3a: Late Roman, Period 4.1 (c. 300 CE)

Located about 6m directly west of pit 381 this pit was compromised by Phase 6 pit digging (see Fig. 6.26/1). It measured approximately 1.2m by 0.6m and was 0.85m deep with almost vertical sides, cut into chalk on the west, north and east. Two fills were recorded; the upper (897), a dark brown loam containing building debris was dated by pottery (43 sherds weighing 468g) to c. 180-300 CE, the undated lower fill (1000) comprised crumbly chalk and clay speckled with charcoal. The section is illustrated on Fig. 6.37/1.

The Aisled Hall

Pit 209 (Fig. 6.14/2)

Around 300 CE, the Phase 2 building was demolished and replaced by a rectangular building in the form of an Aisled Hall measuring externally 13.5m long by 11.5m wide, and internally 11.8m by 10.2m. There was no excavated evidence for any entrances through the east, west or south walls; it is therefore likely that a doorway was located in the north wall where only a potential vestige survived. Figure 6.15 shows the Aisled Hall after excavation and Figure 6.16 illustrates the Phase 3 plan.

This 1.40m deep, vertically-sided pit measured 2m by 2.2m on the excavated surface and was located south of

The walls consisted of courses of flint nodules cemented in a white sandy mortar which incorporated fragments of 107

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 6.8. Plan Phase 2 rectangular building

ceramic roof tile. The line of the 0.8m wide west wall was retained but rebuilt (1211), and four to six courses of flint survived for much of its length. The east wall (224) was not so well-preserved, but up to three courses survived in places. A new 1m wide south wall (713) was bonded to the east and west walls. Two dressed limestones (538 and 539) utilised as quoins, were located on the south-east corner (Fig. 6.17). The smaller rectangular block (538) had vertical tooling on two faces whilst the larger, thinner, stone (539) with vertical tooling on the edge was broken; Bellamy (Chapter 13) suggests that it may have originally been part of a lid. The wall was built against tank 778, which remained in use. Only a very small remnant of the north wall (1230) survived at the north-west where it bonded with the west wall, which at this point narrows where the Phase 4 extension starts.

to Phase 4 survived (see below and Fig. 6.18). There was a central space of approximately 4m between the plinths, which were approximately 3m apart. The plinths stood on flint-packed foundation pits (636, 637 and 638 to the east and 647, 648 and 649 to the west); the densely packed flint would have spread the load from the weight of the roof. The eastern row of pits was located on the line of the demolished east wall (1220) of the Phase 1 and 2 building. Two of the pits were half-sectioned and displayed evidence for re-modelling associated with later Phase 4 works (discussed below). Flint-packed pit 638 (Fig. 6.19/1) This pit with moderately steep sides and a flat base was cut into natural chalk; its excavated width was 0.9m and its depth was 0.28m. The fill (850) comprised tightly packed flint nodules in chalky loam. Pottery from near the base (27 sherds weighing 130g) dated to the late 1st century and was associated with Early Roman activity in this area. At

The building was divided into a central nave with two side aisles delimited by three pairs of timber uprights located on stone bases (plinths); only one of these (158) belonging 108

The East Range

Fig. 6.9. Central part of the East Range with wall 1210 in the foreground, flint pit 649 and Tank 778 in the centre. The fallen apex 827 is visible behind the pit. Trench 41 is centre top with midden 1529 showing as dark soil, view east

Fig. 6.10. Tank 778 half-sectioned showing the fill of limestone roof tiles, view north-east

109

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Pit 1077 (Figs 6.19/3 and 6.22) Located outside the south-west corner of the building, this small vertically sided pit with a diameter of 0.65m and depth of 0.45m was cut through the floor remnants of the Phase 2 building and into natural chalk. It is possible that, when the pit was dug, the Phase1/2 north-south wall (1210) was partially intact, providing shelter from the prevailing south-west winds. Three fills were recorded; the upper and lower (1078 and 1117) were both charcoal-rich, while the mid-brown loam middle fill (1114) had frequent lumps and fragments of mortar. Burnt flint was recorded in the two lower fills. Of significance were two sets of iron hub linings, nave rings and fragments of lynch pins from a wheeled vehicle (Fig.15.3/20-23). Pottery sherds (18 weighing 300g) from all the fills derived from a Type 8/12 BB1 straight-sided dish (c. 290-370 CE). Sherds from a re-fired, beaded and flanged shallow BB1 dish (c. 200-400 CE) were noted in fill 1078. Charred hay was recorded, suggesting that animal bedding had been dumped here. No deep pits are associated with either phase of the Aisled Hall, and evidence from middens and ditch fills suggests that household rubbish was disposed of well away from the building. Phase 3b: Late Roman, Period 4.1 (c. 350 CE) Fire damage and site preparation for re-building It would appear that about 350 CE a fire damaged the Aisled Hall necessitating a major rebuild. The northern half of the central apex (827) of the range collapsed onto the floor resulting in an 8.5m length of mortar adhering to ceramic tiles which had formed the ridge of the building. This material was absent on the southern part of the site (Figs 6.15-16). The collapse was between 1m and 1.27m wide and approximately 0.32m thick and was slightly offcentre to the walls of the building. A deposit of ceramic roof tiles (1106), (Fig 6.2/2) probably relates to this episode.

Fig. 6.11. Tank 778 section (1) and pit 381 section (2)

the time of excavation it was not realised that the pit was of two phases and that the upper 0.35m of fill belonged to the Phase 4. Flint-packed pit 648 (Fig. 6.19/2 and 6.20) This pit had four fills and the two lowest (971 and 968) were associated with the Phase 3 building. The original chalk-cut pit was approximately 1.3 in diameter and 0.25m deep with steep sides and a flat base. The primary fill consisted of tightly packed flint nodules in ash and crushed chalk with occasional fragments of ceramic roof tile. Above this, the flints in layer 968 were embedded in a fine ashy soil. Fill 968 produced 77 pottery sherds (627g) from at least five BB1 vessels with a date range c. 340-370 CE.

Tank 778 was deliberately filled with limestone roof tiles (Fig. 6.10), most likely from the roof of the building. The floor level was raised to the top edge of the tank and consisted of a charcoal-rich, ashy loam (contexts 300, 716, 866, 911, 922, 1017, 1105 and 1344) which contained large amounts of building debris, including roofing material, building stone and mortar. In total 978 sherds of pottery (9,340g) was recorded, most of which dated c. 200-370 CE. Perhaps significant were 20 fragments from a 1st/2nd century blue/green square glass bottle, some of which were cracked and distorted by fire.

Deposit 1269 (Fig. 6.21) Four nested pots were placed against the east wall 224. The outer vessel, a large substantially intact Type 1/2 BB1 cooking jar, contained fragments from a similar, smaller vessel, together with sherds from a Type 6/2 flanged bowl and large sherds from a DR20 amphora The location of this material hints that it may have been a deliberately placed foundation deposit. The date for this assemblage (c. 43-290 CE) suggests the use of curated material.

Phase 4: Late Roman, Period 4.2 (c. 350/70 CE) (Fig. 6.23) Extension of the Aisled Hall The south wall (731) was retained, the east and west walls were extended north, this was particularly noticeable 110

The East Range

Fig. 6.12. Pit 381, half-sectioned, view north

Fig. 6.13. Pit 1149/1181 half-sectioned, base not excavated, view north

111

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 6.14. Sections of pits 1149/1181 and 209 (1-2)

on the west where the new build (432) was narrower at 0.6m wide. A doorway may have been located here. The building was now 17.5m long. The east wall (224) was the least well-preserved and it was not possible to discern any change of build here. The new north wall (569) was 0.6m wide and set on a flint offset. Parallel to this was a 2.5m length of wall (754) constructed at a slightly lower level, of which four courses of flint in mortar survived. This short length could have been an abandoned foundation prior to the construction of wall 569 or an attempt at strengthening. There may have been an entrance near the north-west corner, but later pit digging destroyed all stratigraphy here.

Features associated with the construction of the north wall Two postholes (739 and 773) dug into a clay loam surface were located outside the north wall of the building (Fig. 6.23). It is likely that they held scaffolding timbers erected to facilitate access to the upper levels as the wall was being built. Both postholes were circular, with posthole 739, located on the centre line of the building. This was larger at 0.45m diameter and 0.9m deep. One side was vertical, the other sloped slightly and the base was flat. Small quantities of building debris were noted in the upper fill (739). Posthole 773 was 0.3m in diameter and 0.3m deep, with vertical sides and an almost flat base.

112

The East Range

Fig. 6.15. Aisled Hall after excavation with collapsed wall 559 in situ, view south

The flint-packed pits

Flint and limestone packing on the edges suggested a square post.

The six original flint-packed pits were raised by mounding flint nodules over the original pits (see Figs 6.19/1-2 and 6.20). Two further flint pits (635 and 1113) were constructed to the north. Pit 1113 was excavated (Fig. 6.19/4), however Phase 5 pit digging had undermined the ground, resulting in the pit slumping north. The depth of the compacted flint (684) was c. 0.2m and a complete limestone pier base (158, Fig. 6.18) rested on this (see Chapter 9). Slumping, and later roof fall which covered the stone, ensured its survival. It is probable that similar plinths were removed from the site.

Six features (162, 163, 164, 199, 200 and 223) were interpreted as putlogs. These comprised small rectangular deposits of tegulae varying from 0.42m to 0.23m by 0.13m to 0.04m and were visible in the collapsed wall. As the building work progressed upwards, scaffolding posts were inserted to enable safe access. On completion, the scaffolds were removed and the voids were filled with ceramic fragments. The putlogs were 1.5m apart in two rows which were also 1.5m apart. Co-incidentally, scaffolding posthole 739 was in a direct line with the middle putlogs.

Internal walls Fragments of two internal walls were recorded (Fig. 6.23). The base of an east-west wall (775), between the east wall 432 and pier base 158, was 0.35m wide and consisted of flints set in a cob/mortar mix and constructed on a

The floor level was raised by approximately 0.35m to 0.5m in the south part of the building; the floor of the northern extension utilised the natural chalk and two new supports for aisle piers (635 and 1113) were installed.

113

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) probable chalk floor. North-south wall 856 was 0.3m wide and comprised flints, set into clay/cob. The fallen wall concealed its extent. Material recorded during the cleaning of this feature and incorporated into its matrix included 17 sherds of pottery (150g), mostly BB1 jars, and two New Forest beakers dating c. 260-350 CE. Floor surfaces As the excavation progressed, areas of chalk and opus signinum were recorded which were interpreted as floor surfaces. These did not survive in the southern part of the building due to thinner layers of overburden and subsequent plough damage. Six separate areas of chalk floor were recorded and were visible to the north of flint pits 637 and 648; their location underneath collapse debris across the width of the building strongly suggests bedding for an opus signinum floor. Fragments from BB1 cooking and storage jars (91 sherds weighing 406g) dated c. 240430 CE were recovered from the floor contexts. The following features were associated with the Phase 4 building: Deposit 556 (Fig. 6.24) Placed against the west wall of the building, this possible foundation deposit consisted of a large Oxford Red-

Fig. 6.16. Plan Phase 3 Aisled Hall

Fig. 6.17. Quoins 538 and 539 of re-used limestone, view north

114

The East Range

Fig. 6.18. Collapse debris with stone plinth 158 (foreground), limestone tiles of roof overlain by flints from the fallen wall 559. Slate from the roof visible top left, view south-east

the Phase 6 pit 209. The throat of the oven was defined by coursed limestone blocks (291) set into clay/loam; this material was also used to construct the oven chamber and subsequently was heavily burnt. Remnants of a heatreddened clay/cob lining survived in the oven chamber. Approximately half-way up at the back of the chamber, the natural chalk had been cut to form a shelf approximately 0.15m wide. Two layers of collapse/demolition debris (332 and 447) covered the oven. A layer of mortar (570) overlaid the structure and contained 42 sherds of pottery (328g) from BB1 jars and bowls and an Oxford beaker dated c. 390-430 CE, probably relating to the final years of the building before the roof collapsed. The upper layer (599) contained quantities of slightly burnt limestone roof tiles which may have derived from a superstructure. Below this, fill 600 comprised charcoal-flecked loam containing large amounts of limestone roof tile and a substantial block of dressed stone. Fill 607 was similar but with a smaller amount of building debris. Pottery from these three fills consisted of BB1 jars and bowls dating c. 340-430 CE. The primary fill (612) of grey ash and charcoal probably relates to the last firing of the oven.

slipped ware type C31 indented beaker dating c. 340-400 CE. Within the pot were three sherds from a BB1 jar and three oyster shells. Oven 207 (Figs 6.23, 6.25-28). Located in the extreme north-west corner of the building and contained within the small room created by the cross-wall 775, oven 207 comprised a raised rectangular structure (581) and a circular oven chamber (598). The feature was located in 2013 but not identified until the following year when further topsoil was removed. Flints from the collapsed north wall (160) (numbered 569 in this area) covered the top of the oven. The upper surface of the structure (581) was set against the west wall (432) of the building. It measured 2m long and 1.12m wide and comprised small limestone and flint pieces set in a cream/pink clay/cob/chalk mix; the northern extent did not survive. Initially, a pit must have been excavated into the natural chalk (219) to receive the chamber structure, utilising the chalk as a floor. The top of the oven chamber (598) was delineated by small, thin rectangular blocks of limestone creating a well-defined edge. The oven was circular on the surface, having a diameter of 0.66m and a depth of 0.78m. The sides were near-vertical and narrowed towards the base (Fig. 6.28); the north side of the chamber had been compromised by the digging of

Structure 1203 (Fig. 6.29) Scrutiny of the associated archaeological sections strongly suggest that the chalk sub-floor extended further east by 2.5m and had been deliberately dug out to form an 115

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Pit 823 (Fig. 6.30) A small, shallow oval pit was dug outside the southwest corner of the building. Measuring 0.94m by 0.75m and 0.18m deep and cut into clay loam, the sides sloped to a rounded base. Two very dark brown charcoal-rich fills were recorded (805 and 806) and between these was a large unworked slab of shale. Animal bone included cattle, sheep/goat and duck fragments. Pottery (46 sherds weighing 408g) had a broad date range of c. 260-400 CE. Phase 5: Final Roman and Post-Roman, Periods 5 and 6 (c. 430-650+ CE) Collapse of the building In 2013, Trenches 6 and 7 were opened to determine the location of the East Range, followed in 2014 and 2015 by the gradual uncovering of the building. Overburden (clearance layers) accumulated after the building collapsed and consisted of homogenous mid-brown loam varying in depth from 0.3m to 0.05m, the lesser depths occurring at the south end of the building where ploughing had removed archaeological layers. Material over the collapsed building Very large amounts of building material were present, particularly over the north part of the building. This included limestone building stone, building flint and very large quantities of roofing material. Limestone roof tiles (383,235g), mostly broken but some intact, and ceramic roof tiles (549,988g) formed the largest group and are likely to have been the main roofing materials. Fragments of slate (6,819g) and sandstone roof tiles (2,751g) were also present, suggesting that the roof was decorated with bands or blocks of contrasting material. There were large numbers of roofing nails. Building material, probably from the interior of the range, included fragments of ceramic floor tile (1,207g), opus signinum (1,539g) and 642 tesserae in various colours and sizes. Very large amounts of mortar (58,567g) are likely to have derived from wall surfaces, but chunks of mortar which had cemented the ceramic roof tiles together were also recorded. Box flue tile fragments (13) were noted; their original location is unknown. Artefactual material was scattered throughout, but most finds were located at the north end of the building where soil deposits were deeper. Pottery was the commonest find (1,229 sherds weighing 9,466g) with BB1 dominating. Also present were fragments of late samian vessels, Moselkeramic, New Forest and Oxford wares. The pottery has a broad date range of c. 120-430 CE. Ten coins were recorded of which eight could be dated. The earliest was an antoninianus of Tetricus I (271-274) and the latest, a nummus of Constantius Gallus (351-354). Also present were fragments of quernstone, glass, copper alloy and lead and iron items. Three jar sherds (34g) of post-Roman pottery (c. 450-650 CE) and four medieval sherds (33g) were recorded. Infant bones representing probably two infants were present within the clearance

Fig. 6.19. Sections flint pit 638 (1), flint pit 648 (2), pit 1077 (3), flint pit 1113 (4)

access area to the oven chamber. Its north-south extent is irretrievable, but the northern limit was confirmed during the excavation of pit 209. On the final day of the 2014 season, structural remnant 1203 was identified. This wall-like feature consisted of seven courses of limestone blocks and flint in a clay/chalk matrix and appeared to be built against the chalk bedrock to the north and intruding into pit 209 on the south and east (Fig. 6.14/2). Located almost opposite the oven, it survived to a height of 0.72m with a north to south width of 0.63m; it was not possible to determine its eastern extent but a 0.4m length was recorded and it is possible that more of the feature survives. A profile (Fig. 6.26/2) from the oven to the structure 1203 demonstrates the relationship between these two features. The function of the wall fragment is presently unknown; it may have been part of a wall around the sunken area or associated with other potentially, still-buried features. 116

The East Range

Fig. 6.20. Flint pit 648 showing ‘mounding’ of the flint to raise the floor level for the Phase 4 work. Earliest Phase 1 chalk floor is in the foreground, Phase 4 chalk floor top right, view north

Fig. 6.21. Black Burnished ware pottery (jar) deposit 1269 after removal of amphora and flanged bowl sherds, placed against east wall 224, view south

117

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 6.22. Pit 1077 with nave hub in fill 1117, view north

layers. The date, quantity and variety of the material may suggest that nearby midden deposits were incorporated into the layer by ploughing action, possibly in the early medieval period. Collapse sequence The removal of the clearance layer over Trench 6 revealed a collapse sequence approximately 0.2m beneath the present ground surface (Fig. 6.31). This consisted of part of the north wall (569) which overlay roof elements of limestone and slate. It would seem that the roof collapsed first, followed by the north wall. The collapse debris extended across the complete width of the building, but only extended south for about 6m. The roof Most of the roof was constructed of Purbeck limestone tiles with bands or panels of slate; the ridge is likely to have been formed from ceramic imbrices (see Chapter 9 for discussion of the roofing material). Limestone roof tiles The quantity and density of tiles on the north-west side of the building can be seen in the foreground of Figure 6.31; fewer tiles were present on the east side where they underlay the fallen north wall. The major tile deposit (157) was uncovered in 2013, and other areas (580, 909 and 916) were designated as the work progressed; all belong to the collapse phase of the building. It is likely

Fig. 6.23. Plan Phase 4 Aisled Hall

118

The East Range

Fig. 6.24. Pottery deposit 556 (large Oxford C31 beaker) placed against west wall 432, view west

Fig. 6.25. North-west corner of the Aisled Hall during excavation, with oven 207 in the centre, plinth 158 top left. The break in the north wall 569 is apparent, as is the loam fill of the later pit cluster, view south

119

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 6.26. Plan of features in the north-west corner and profile across the features

Collapse of the north wall 559 (Fig. 6.31) (Period 6 c. 430-650 CE)

that a central roof covered the ‘nave’ with single storey ‘aisle’ roofs on each side (Fig. 6.32). The enormous weight of tiles would have needed external load-bearing walls and internal pier bases to support substantial roof timbers.

It would seem that once the roof fell, the north wall (559) followed soon afterwards. The inside face of the collapsed wall (context 160) survived, leaving several courses of wall in situ. This was about 0.2m thick, with the assumption that c. 0.4m, or about one third of the wall thickness, has been ploughed away. Approximately 6m of wall had fallen from the north and consisted of horizontal rows of flint nodules set into creamy/white mortar and extending east-west for 4.5m. The south edge of the collapse captures the pitch of the gable and the clerestory level, but the gable height cannot be estimated as it is not known where the fracture in the north wall occurred. Comparison with other similar buildings and events is considered in Chapter 19.

Slate roof tiles Deposits of slate tiles (contexts 161, 486, 913 and 994) were recorded in the centre of the north-south building line. The use of this material on Roman sites is not common. The width of the deposit was approximately 4m and it is probable that the slate was part of a decorative scheme with a wide band of the material near the apex on both sides of the roof, but below at least several courses of limestone tiles. Figure 6.33 shows the relationship between the slate and limestone tiles, and the north-south section on Figure 6.34/1 shows the slate and the deposits above and below this. 120

The East Range

Fig. 6.27. Oven 207 after excavation with limestone blocks 291 forming the neck of the firing chamber; at the rear the shelf is visible. Intense head has reddened all surfaces, view west

Other material associated with the wall collapse The material in layer 836 (Fig. 6.34) lying above the slate deposit (161) and below the flint fabric of the wall consisted of very large amounts of building debris, which included 24 tesserae and large lumps of degraded mortar. A coin of Postumus (260-269) came from this layer. Pottery (93 sherds weighing 557g) dated c. 200-400 CE and two similarly dated fragments of glass must have accumulated before the wall fell. This material is likely to have derived from the very latest activities in the building. The layer of slate (161) lay directly under this and was above the thin 0.03 to 0.06m deposit 851which consisted of fine soft grey/brown loam containing numerous small animal bones. This material must have built up just before the roof fall and can be compared to a similar layer from Room N1 in the North Range. The lowest fill (847) comprised crumbly mortar and fragments of painted plaster, as well as small animal bones. The material is likely to have fallen from walls or ceiling.

Fig. 6.28. Oven 207, section

121

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 6.29. Structural remnant 1203 against natural chalk to the left and pit 209 to the right, view west

Phase 6: Early Medieval, Period 6 (c. 430-650 CE)

962) had been dug in the building after its final collapse, removing part of Phase 2 pit 896. Figure 6.26/1 illustrates their relationships. The pottery from the Phase 6 pit fills was overwhelmingly BB1; vessel types and non-local pottery are noted in the fill descriptions below. The pottery from pits 209, 752, 944 and 962 is considered in Chapter 12 (Assemblages 21-23). The stone and metal finds are discussed in Chapters 13-15.

Post-collapse activity This area in the north-west corner of the building was defined by soft dark brown loam which was surrounded by building debris. As excavation progressed it was apparent that a series of intercutting pits (209, 752, 825, 944 and 122

The East Range

Fig. 6.30. Removal of large shale slab from pit 823, view north

It is probable that the lower floor level associated with the Phase 4 oven 207 was utilised as a convenient dumping area with a successive series of pits filling the void. Small quantities of post-Roman and early medieval pottery were retrieved from the upper fills of pits 825, 944 and 962, suggesting that they were the final infillings in this area. The finds, however, are overwhelmingly Roman and are in secondary contexts, but must have derived from ‘clearing’ episodes around the villa.

recorded (not all are on the illustrated section) and are considered from the base to the top of the pit. The lowest (286) consisted of a grey/brown ashy loam with profuse charcoal and was characterised by large amounts of burnt limestone. The BB1 pottery dated c. 300-400 CE and included jars and a flanged bowl. An un-numbered deposit of chalk intruded into this layer on the west side and was probably collapsed natural chalk from the side of the pit which fell as the layer accumulated. Above this, layer 285 comprised a scorched, pink-tinged chalk/ clay mix. A single BB1 incised beaker base sherd was recovered (Fig. 11.9/4). The grey/brown ashy loam (240) extended across the pit; very large numbers of broken burnt limestones were recorded. The pottery dated c. 370430+ CE and included BB1 jars and bowls and a single

Pit 209 (Figs. 6.14/2 and 6.35) The pit was first dug during the occupation of the Phase 2 building (c. 200-300 CE), with the lowest fill (287) dating to this period. Eleven subsequent fills were 123

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 6.31. Phase 5 Aisled Hall wall and roof collapse as excavated in 2014, view south

0.2m thick layer of charcoal-rich, black, ashy loam and burnt building material which lay between layers 239 and 227. An illegible corroded coin is likely to be of 4th century date. Fill 227, a dark brown crumbly loam, contained large building flints. The pottery included BB1 jars and a bowl, and fragments of a New Forest greyware jar and an Oxford C23 beaker dating c. 350-430+ CE. The following three fills, 218, 217 and 216, were recorded only on the north side of the pit. Fill 218 was located above layers 227 and 228 and below layer 217; its dark

New Forest greyware jar. Fill 239 comprised a reddish sandy loam with a high percentage of raw red clay and degraded chalk. It extended up the west side of the pit, suggesting that the layers above had cut into it. Building material was sparse; pottery included BB1 jars and an Oxford C23 beaker fragment dating c. 270-430+ CE. Above this, fill 230 comprised a hard compact layer of sandy mortar with small chunks of painted plaster. There was no dating material. Fill 228 (not on the section), was confined to the north side of the pit and consisted of a 124

The East Range brown loam fill was flecked with patches of white/grey ash. The small assemblage of BB1 pottery included jars and a bowl and fragments from a C97 Oxford mortarium (c. 370-430 CE). Also on the pit’s north side, black ashy loam layer 217 contained large amounts of broken limestone roof tiles. The pottery, dating c. 350-430+ CE, included BB1 bowls and a jar, a New Forest greyware jar and an Oxford C51 bowl. One fragment of infant bone was recorded. Fill 216, above fill 217, comprised a grey/ brown loam; a single BB1 jar dated to c. 260-400+ CE.

The topmost layer of mid-brown loam (194) contained building material, including roof tiles and mortar, which may have been associated with the collapse sequence. Finds included a single BB1 jar dated c. 340-430 CE and the end of a bone pin. The layers of this pit suggest that it was filled with material which may not necessarily have derived from the Aisled Hall. The pottery mostly dated c. 400+ CE, hints that occupation of the building perhaps continued beyond 430+ CE. Pit 825 (Figs 6.36 and 6.37/1: the pit is located on section line AB on Fig. 6.26) Excavation revealed that this pit was earlier than, and was cut by, pits 752 and 962 and cut pit 209. Its full dimensions could not be ascertained, but the undulating base was cut into chalk and the two recorded fills suggested that it was at least 0.7m deep. The dark brown loam lower fill (826) contained burnt material which probably derived from the collapsed oven 207. Pottery included BB1 jars, bowls and dishes and New Forest beakers dating c. 300-430 CE. A single sherd of post-Roman pottery was dated c. 430450 CE. An upper fill (201), was restricted to a small area (removed to draw the section) and comprised a very dark brown loam with high charcoal content; large amounts of limestone roof tiles and mortar were present. BB1 jars and bowls and New Forest greyware and colour coated bowls and beakers dated c. 240-430 CE.

Fig. 6.32. Reconstructed profile of Aisled Hall

Fig. 6.33. Slate 161 and limestone roof tiles 916 under fallen wall 160, view south

125

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 6.34. Slate deposit 161 (1) and postholes 773, 739, sections (2-3)

Pit 752 (Figs 6.36 and 6.37/1)

was cut by pits 209 and 962 and had a diameter of 1.26m and a surviving depth of 0.86m. A single dark brown loam fill (945) was recorded; mortar and flint had possibly been dumped against the north side of the pit. Building material was abundant. Copper alloy objects included an indeterminate scrap fragment, a 1st century broken crenelated bracelet (Fig. 14.4/8), a complete bead (Fig. 14.5/12) and 2nd/3rd century military-type pelta-shaped stud (Fig. 14.11/1). Animal bone included cattle, sheep/ goat, pig, dog and cat as well as bat. The large quantities of pottery dated c. 150-400 CE, although there were sherds from two earlier vessels: a BB1 jar with countersunk handles and a Corfe Mullen flagon. The extended time frame for artefacts suggests that some of this material may have been redeposited.

This oval pit measured 1m by 1.9m with a surviving depth of 0.55m. The north-west and north-east sides of the pit were directly against the remains of the north wall 569 and its possible buttress 754. The possible doorway in the north wall at this point may have resulted in ground which was not too difficult to dig through. The pit cut earlier pits 209 and 825 and was separated from the Phase 2 pit 896 by a baulk of natural chalk and clay with flints. A layer of red clay (807) varying in thickness between 0.05m and 0.1m was applied to the south and southwest sides of the pit and may have formed a lining or consolidation. Due to its position, this is not on the section drawing. BB1 sherds dated c. 370-430+ CE. Three other fills were recorded. The primary fill, 810, a fine black ashy silty loam contained fragments from BB1 jars, bowls and dishes, sherds from a South West Black-burnished ware jar and a large chunk from a New Forest mortarium, suggesting a date range of c. 270-370 CE. Fill 802 was 95% chalk, only 0.08m thick and may have been the slumped and eroded sides of the pit. It is possible, however, that these two layers were associated with activities involving the Phase 4 oven 207, and the 12 sherds of pottery (c. 340-370 CE) suggest that this might have been the case. The upper fill (795) contained building debris. The pottery included BB1 jars and bowls and Oxford and New Forest finewares dating c. 300-430 CE. An antoninianus of Carausius dated 286-287. Post-Roman pottery sherds (14 weighing 90g) suggest a 5th/6th century date for infilling.

Pit 962 (Figs 6.37/1 and 2: section lines AB and CD shown on Fig 6.26) This bowl-shaped pit, with gently sloping slightly stepped sides and an undulating base, measured 2.7m in diameter with a maximum depth of 0.55m. It cut into the soft fills of pits 752, 825, 896 and 944 and was the latest feature in the complex. Two finds-rich, very dark brown, loam fills were recorded. The lower (705) lacked the charcoal noted in the upper fill (533 to the north and 624 to the south). Very large quantities of building material were present, including limestone building stone and roofing tiles, mortar, ceramic roof and floor tile. Smaller amounts of box flue tile, opus signinum, architectural mortar and painted plaster were also present. Finds included a single ear ring (Fig. 14.5/11), a complete bone hair pin, a sharpening stone, and indeterminate glass fragments. Animal bone included cattle, sheep/goat, pig and dog plus domestic fowl, duck, wood pigeon and crow bones; many loose teeth indicated that this material was re-deposited.

Pit 944 (Fig. 6.37/2) Cut into natural chalk and through walls 569 and 754, this circular, vertically sided pit with a gently rounded base 126

The East Range

Fig. 6.35. Pit 209 half section, view south-east

Fig. 6.36. Section through pits 896, 752 and 825, view south-east

127

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 6.37. Phase 6 pits 896, 962, 752 and 825 and 962 and 944, sections (1-2)

Six infant bones were recorded. The pottery assemblage (1,465 sherds) was dominated by BB1 (1,266 sherds), but other types included late samian, Moselkeramic, Nene Valley, New Forest and Oxford wares with dates varying from c. 150 to 430+ CE. The upper fill yielded 20 sherds (122g) from post-Roman hand-made jars dated c. 450-650 CE; their presence suggesting that this fill could have been deposited between the 5th-7th centuries.

not possible to state with certainty the use of this building; the architecture and finds suggest both domestic and light industrial use, which no doubt fluctuated over time.

Conclusions The reasons for the digging and then infilling of the pit cluster in the north-west corner of the collapsed building are unclear. Very large amounts of building material and finds were recorded and could have derived from clearance episodes relating to 5th/6th century activity, possibly in the North Range where evidence for occupation at this time is strongest. General conclusions The development of the East Range from a small rectangular building to an Aisled Hall of some pretention implies not only continued use, but the need to demonstrate status and quality, particularly in the choice of building materials. It is 128

7 The West Range The West Range, located on the western side of the courtyard, was built on a gentle south-facing slope. It was constructed in two phases and was terraced into the slope Fig. 7.1). The topsoil was cleared by a mechanical excavator in east-west strips comprising Trenches 8, 9, 10, 12, 29, 33 and 37 (Fig. 1.6). The building was excavated from north to south and as internal walls were revealed, room numbers (W1, W2, W3 and W4) were allocated. A layer of building debris comprising building flints overlying limestone, and to a lesser degree ceramic roof tiles, covered the northern half of the building (Fig. 7.2).

roomed building measured 9.7m long by 5.5m wide and had an entrance on the eastern side. The walls were composed of courses of mortared flints with no foundations but had an offset base layer of large flints. Three courses of flint laid in a herringbone pattern survived in parts of the north wall 184 (see Fig. 7.6). Four courses of flint survived in east wall 142. A doorway giving access to the courtyard consisted of a 1.3m wide threshold (542) composed of broken ceramic roof tiles (Fig. 7.4). Only one course of flints survived in walls 1005 and 1006 which were demolished when the building was extended. The wall widths varied between 0.5m and 0.63m. Remnants of a chalk floor, which may have utilised the underlying chalk bedrock, survived. There were no features or finds related to this phase of the building, which must have been completely cleared prior to rebuilding. It possibly replaced the Ancillary Building which was located away from the villa complex (see Chapter 4).

Phase 1: Middle Roman Period 3.1 (c. 120-200 CE) (Fig. 7.3/1) Probably constructed in the late 2nd century and located about 2.5m east of ditch 308, the small rectangular, single

Fig. 7.1. The West Range with Rooms W1, W2, W3 and W4 after excavation. Phase 1 walls are visible under Rooms W1 and W2, view south

129

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 7.2. Collapse debris covering Room W1. Ceramic floor tiles are utilised as quoins on the north-west corner of the room, view south-east

where the variation in wall thickness is evident. The new west wall 171 extended the length of the building but was vestigial at its south end. This wall was only one metre away from the edge of ditch 308, which must have been filled in when the building was constructed. However, it seemed that the builders were aware of the ditch and the possibility slumping and reinforced its eastern edge with three courses of closely packed flints (234). Quantities of ceramic tiles including roof and box flue tiles were scattered on the ground surface between the infilled ditch and the building (113, 152 and 772), creating a hard surface. Pottery dated this layer to after c. 260 CE. Only the south-west stub of the south wall survived. Other building materials in the wall matrix included ceramic roof and flue tiles, fragments of limestone and slate roof tiles.

Pit 504, which cut into the west side of ditch 308 (Figs. 7.3/1-2), was probably contemporary with this phase. Measuring 1.5m in diameter and 0.9m deep, the homogenous mid-brown loam fill (506) was flecked with charcoal and chalk; 15 sherds from a BB1 necked jar and a beaker date the fill to c. 90-200 CE. The pit was cut by a later re-cut of ditch 308, apparent in a tipped fill (481) of charcoal and ash dated by pottery to c. 160-270 CE. It is possible that this fill equates with the re-building of the West Range. The upper fill (480) of the ditch was chalkrich, with pottery dating to c. 240-300 CE representing the final filling of the ditch. Layer 479 which covers the pit and the ditch dates to c. 370-430 CE and indicates final activities in this area. Phase 2: Period 4.1 Late Roman (c. 250-300 CE) (Figs. 7.1 and 7.5)

Three east-west walls (167, 135 and 690) divided the building into four rooms (W1, W2, W3 and W4). The walls were about 0.5m wide and butted the north-south walls. All were well-constructed with courses of mortared flints, the outer ones set lengthways and the interior ones tightly packed with smaller flints. The north-facing aspect of wall 167 displayed herringbone pattern construction. The floor surfaces were chalk (which probably utilised the natural bedrock), with imported material brought in to level up. Slabs of limestone were also used. The rooms and their features are considered below and are illustrated in plan on Figure 7.5.

The Phase 1 building was extended, creating a structure which measured 7m wide and at least 26m long composed of four rooms. The east wall 142 was extended south and recorded as 136. Only one course survived; this was the base layer which was up to 1m wide and comprised tightly packed flints laid on compacted loam. The southern extent of this wall had been removed by plough action. The original doorway was retained. The north wall 184 was extended west, continuing a similar building method of outer layers of tightly packed mortared flints with a rubble core. The wall extension is apparent on Fig. 7.1, 130

The West Range Room W1 Internally this room measured c. 5.5m east-west by 5.8m north-south, and contained features probably associated with semi-industrial processes, as well as infant burials. Deposits Three shallow deposits were located over the earlier north-south wall 1006; the smaller undated features may represent patches of flooring. Deposit 804 This consisted of a thin spread of charcoal and ash. Within it were 13 sherds from a BB1 bowl dating c. 270/300-370 CE; conjoining sherds came from context 525, an upper layer of the stoke pit 781. Deposits 797 and 811 These consisted of a spread of burnt opus signinum (797); underneath this (811) were 35 limestone tesserae and one of shale. Animal bone deposit 733 Located in the north-west corner of the room, at the junction of walls 184 and 171, a number of animal bones were buried under layer 505. Comprising parts of young sheep/goat, cow and pig, as well as some chicken bone,

Fig. 7.3. Phase 1 building plan and associated stretch of ditch 308 (1), pit 504 and ditch section 478 (2)

Fig. 7.4. Room W1, door threshold 542 composed of broken ceramic tiles in wall 142, view east

131

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) this collection may have been votive and deliberately placed. Oven Complex 706 Located centrally at the north end of Room W1, this large feature comprised an oven (558), flue (780), stoke pit (781) and ceramic tile structure (1129). It was completely excavated and is illustrated on Figure 7.6, with the plan and relevant sections on Figures 7.7 and 7.8. The oven and associated features were constructed below the chalk floor surface (557), suggesting that an area measuring approximately 3.6m by 3.2m with a depth of c. 0.6m was cleared to make space for the complex. On abandonment, the oven and its immediate area was covered by a 0.1m layer of mid-brown loam (494) containing large amounts of building material derived from the collapsed building and dated by pottery to c. 270-370 CE; a nummus (Gloria Excercitus) dated 337-347. The oven (706) was dome-shaped, built of mortared, roughly squared limestone and heathstone blocks, and measured 2.56m long, 1.54m wide with a surviving height of approximately 0.7m (Figs 7.8/1-2). The eastern side of the dome was strengthened with large chunks of tegulae. The oven was coeval with the Phase 2 work, since the stones at the back of the oven chamber were located underneath wall 184 (Fig. 7.6) and excavation on the north side of the wall confirmed that the structure was confined to the wall space. The stonework was heavily burnt, indicating very high temperatures during the lifetime of the oven.

Fig. 7.5. Plan Phase 2 building

Fig. 7.6. Structure 706 composed of oven 558, flue780 and tile structure 1129, view north

132

The West Range

Fig. 7.7. Plan of oven complex 706. Inset shows section lines for Figure 7.8

The circular oven chamber (558) was well constructed with blocks of limestone and measured 0.4m in diameter at the top, narrowing towards a 0.2m diameter base. The oven was covered by collapse layers 494 and 505 which accumulated after the oven had gone out of use (Fig. 7.8/1).

those in the oven. Three tegulae acted as a flat base on the western side (Fig. 7.6). The tiles are Warry Type C5, manufactured c. 160-260, suggesting a construction date c. 240-260 CE. Attached to the oven on its eastern side was a ceramic structure (1129), measuring 0.71m by 0.67m and 0.35m high, consisting of seven courses of mortared tegulae laid on natural chalk (Fig. 7.6). The top of the structure utilised the tile flanges as a lipped edge. The function of this remains speculative but must have been associated with activities centred on the oven.

Layer (559) comprised a light brown soil with chalk, mortar and building material, which may have derived from the disintegrating top of the oven. Pottery (30 sherds weighing 429g) included BB1 jug, jar and bowl fragments, the latest material dating to c. 370-430 CE, which must equate with the final infilling of the feature. Fill 639 was a grey ash and chalk with lesser amounts of building debris. Under this fill 657 consisted of a heat-affected orange sand. The black charcoal-ash base 658 was probably residue from the last oven firing. A coin of the House of Constantine (337-340) dates this event to the mid-4th century at the earliest.

Initially identified as a stoke pit (781), a deliberately dug, sunken area to the south and west of the oven (Fig. 7.7) may have been a working space associated with the oven. This area measured 1.8m from north to south and 1.1m east to west, with a narrow 0.3m gap on the west side of the oven and a 1m area in front of the flue. The base was natural chalk which contained trampled charcoal. Four fills were excavated (Fig. 7.8/3). The uppermost loam

The flue 780 (Fig 7.8/2) was approximately 0.4m long and 0.2m wide with a surviving height of 0.4m and was carefully constructed with limestone blocks similar to 133

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 7.8. East-west A-B (1) and north-south sections C-D (2), through flue 780 and north-south section D-E through stoke pit 781 (3)

fill (494) extended across the oven and contained very large amounts of building debris including limestone and ceramic roof tiles. Pottery (73 sherds weighing 853g) dating c. 270-370 CE, included BB1 jars, bowls, dishes and a jug, as well as fragments from an Oxford C100 mortarium and a New Forest type 27 beaker. A House of Constantine coin was dated 337-347. Fill 525 contained charcoal, ash and lesser amounts of building material, and was only present over the ‘stoke-pit’. Large amounts of pottery were retrieved (181 sherds weighing 2,724g), including BB1 sherds from jars, bowls and beakers and two Oxford mortaria (C100 and M22), with a date range of c. 260-430 CE. Ironwork included roofing nails, a rove, a bolt and a needle fragment. Of particular significance was the presence of 15 coins which were found in a group (see Fig. 7.8/3), suggesting a scattered hoard, which included 11identifiable coins (four minims and seven nummi with a date range 330-383 CE). Fill 505 was 90% mortar and building debris. Fragments from BB1 jars, beaker and a bowl (27 sherds weighing 251g) date the layer to c. 350430 CE. The base fill 561 contained ash, charcoal and relatively large amounts of building debris. Pottery (112 sherds weighing 1,783g) included BB1 jars, bowls and beakers, a New Forest Type 67 bowl and an Oxford C45 dish, dating c. 370-430 CE. A fragment of Purbeck stone mortar (Fig. 13.2/5) was recorded.

The presence of so much building debris and household rubbish suggest that the two lowest fills (505 and 561) are likely to have accumulated after the oven was abandoned and may have been used to level up the floor surface. Layer 525 may also have been used as levelling but could also have been associated with activity within the room. The extensive surface layer (494) contained very large amounts of roof and wall material, probably associated with the collapse of the building. Oven 219 (Fig. 7.9/1). Identified as an oven and located in the south-east corner of the room, this oval feature measured 2.2m by 0.8m with a depth of 0.4m. The sides were generally steep, and the immediate area around the feature was heavily burnt and discoloured red. A layer (170) of building debris covered the oven, which was cut into the remnants of a floor surface (507) of chalk with clay in loam. Three fills were recorded. The upper (220) was a 0.05m deposit of chalk which more or less covered the oven; similar small deposits were noted around the edge of the feature. This chalk appeared to cap layer 221, an ashy, charcoal-rich loam which contained undiagnostic iron fragments, slag and 14 pottery sherds broadly dated c. 134

The West Range

Fig. 7.9. Sections of oven 219, Room W1(1), and oven 1118 (2), Room W2

a compacted layer (970) of gravelly loam, into which two small features had been cut.

200-400 CE. Perhaps significantly among the animal bones were infant limb and rib bones. The presence of slag hints that this oven may have been associated with ironworking. The primary fill (222), a clay-rich loam, contained very large quantities of limestone (156 pieces weighing 39,950g), some of which was burnt. The stone may have derived from collapsed structural elements of the oven.

Oven 1118 and pit 1084 Located on the western side of the room, measuring 1.2m by 0.6m and 0.22m deep, the remnants of this degraded feature 1118 survived as small blocks of burnt heathstone on the surface of a charcoal-filled pit (Fig. 7.9/2). The keyhole shape suggested that it may have been an oven. A small pit (1084) with a charcoal-rich fill, measuring 0.18m by 0.15m and 0.07 deep, may have been associated with the oven. These features were undated.

Infant burials In addition to the infant remains from oven 219, the poorly preserved remains of three infants (Chapter 16) were located in small pits against the east and south walls of the room. Oval pit 609, dug against the east wall 142, measured 0.65m by 0.4m with steep sides, a flat base and was 0.2m deep. Infant remains (610) and sherds from a cooking jar dated the burial to c. 340-430 CE. Pit 963, against the southern wall 167, was 0.5m by 0.35m and 0.17m deep, with sloping sides and a flat base. Fragmented remains (964) from two infants were recorded

Room W3 This was the largest room, internally measuring 5.8m eastwest by 8.0m north-south. Only 2m of the foundation course of the east wall (136) survived in the north-east corner of the room. The one surviving course of the dividing east-west wall (690) comprised flint nodules set in chalky mortar. Floor surfaces

Outside the north wall of this room, a spread (141) of 134 unused ceramic tesserae cut from roof and box flue tiles was recorded, possibly indicating tile cutting activity in this area.

Approximately 0.1m of mid-brown loam covered the room. This layer contained limestone building stone, limestone and ceramic roof tiles, which were probably associated with the collapse of the building. Finds included jar sherds (c. 370-420 CE) and two nummi of Constantine II dated 317-340. A fragment from a colourless glass cup was 2nd/early 3rd century in date. Beneath, were remnants of a rammed chalk floor.

Room W2 Internally, this measured c. 5.8m east-west by 5.5m northsouth. Initial clearing revealed limestone tiles from the collapsed roof in the north-east part of the room. The presence of building material in the clearance layers lessened towards the south side of the room. A coin of Valens (364-378) and associated pottery indicate an accumulation date of c. 350-400 CE. Beneath this (Fig. 7.9/2), a chalk layer (969) survived in patches and a further coin of Valens was retrieved from this surface, which was interpreted as a floor. Underlying the chalk was

Two conjoining features (691 and 692, Figs 7.10 and 7.11) were interpreted as a pair of ovens which were likely to have been used together. The pottery assemblage suggests that the complex had gone out of use by the end of the 3rd century. The plan of section locations is on Figure 7.12/1, sections (Fig. 7.12/2-6) and Figure 7.13 shows the features after excavation. 135

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 7.10. Room W3 ovens 691 and 692 prior to excavation. Wall 171 in the foreground, wall 690 to the right, view east

accumulated on a probably natural chalk surface. Four lumps of burnt limestone must have been part of the wall structure. Two BB1 bowl sherds (c. 210-290 CE) date the abandonment to the 3rd century. Layer 794 comprised a clay/mortar mix, very similar to the top fill (764) of oven 691, with ceramic floor and roof tile fragments and further quantities of burnt limestone. Pottery included fresh sherds from two BB1 jars and a rim sherd from a samian Dr 31 bowl. The thin layer 771 above this was similar to 794 but contained large building flints. The top fill consisted of a chalk-flecked, light brown loam (768) containing limestone blocks and limestone and roof tiles tile fragments, but there was no dating material. A layer of mortar (723) appeared to deliberately cap the structure.

Oven 691 (Fig. 7.12/2) This undated pear-shaped oven was 0.8m long, had a maximum width of 0.85m and a depth varying from 0.3m to 0.12m; the edges were consolidated with large flints. The feature had steep sloping sides and a rounded base to the west and a much shallower profile to the east. The homogenous fill of orange/brown clay (679) contained small fragments of building material. Above this was a chalk and clay mixture (764) which had been burnt pink in places. Oven 692 (Fig. 7.12/3) Seemingly attached to oven 691 on the north side, this circular, vertically sided structure had an external diameter of 1.3m, an internal diameter of 0.75m and a depth of 0.72m. The inside of the chamber was lined with blocks of limestone (763) set into a chalky mortar; where these survived, they were burnt grey/blue. A square opening (approximately 0.3m by 0.3m) was located on the east side of the feature, the top of which was defined by large limestone blocks (Fig. 7.13). The construction of this feature suggests that it could have been a furnace associated with ironworking.

Section E-F was cut immediately east of the oven (Fig. 7.12/4) to investigate the construction. The flue opening was revealed with intact structural limestone blocks in situ on the southern side. However, on the northern side these elements (1155), which were contained within the chalk and clay fill 900, appeared to have collapsed and lay above a clay/loam layer 1133 which extended into stoke pit 766. A thin basal fill of charcoal (1151) was similar to layer 796 in the oven. Pit 766 (Figs 7.12/5: H-J-K and 7.12/6: F-G)

Excavation revealed five layers which must have accumulated after the oven went out of use. The thin base layer (796) comprised charcoal and ash which had

Measuring approximately 1.7m by 1.5m and 0.7m deep, this feature probably functioned as a stoke pit to oven 136

The West Range

Fig. 7.11. Room W3, plan of ovens 691 and 692

692. Sections through the pit reveal a complex infilling sequence. Fill 900, which covered the eastern part of the oven, continued into the stoke pit. Around and underneath this, fill 765 contained very large amounts of building material. A small assemblage of animal bone (34 fragments) included cow, sheep/goat and a single hare bone. There were 97 sherds of pottery (618g) including BB1 jars and bowls and four sherds of fresh samian (Dr 31, Dr 37 and Dr 38), suggesting an infilling date of c. 200-270/300 CE. A possible clay lining (770) was present on the sides of the pit but did not extent as far as the oven. A thin spread of charcoal (901) was recorded in this area. A further section (Fig. 7.12/6) was cut in an attempt to understand the infilling sequence. Fill 1128, a dark brown loam, contained building material and 47 sherds (626g) of BB1 pottery comprising jars and bowls dating c. 180-250 CE; 15 fragments of sheep/goat bone were also recorded. Below this, clay fill 1133 was similar to fill 770; five fragments of cow and sheep/goat bone were recorded. At the eastern end, these two fills appeared to be cut by a deposit (1141) of clay loam with gravel, chalk lumps and remnants of ceramic. This was possibly the wall of a further feature which was not investigated. Pit 954 Located to the north of the oven complex, and probably associated with it, a small steep-sided pit (0.35m diameter

Fig. 7.12. Ovens 691 and 692 plan and Sections (1-6)

137

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 7.13. Ovens 691 and 692 after excavation, view east

and 0.16m deep) was filled with a charcoal-rich loam. One nail was recorded.

weight implies the use of large structural timber elements for support. Ceramic tiles were also used, but in smaller quantities. The original doorway continued in use, but there may have been others both externally and internally which were not apparent in the excavations. The building probably functioned as workshops.

Room W4 The internal measurements were approximately 6.2m east-west by 4.5m north-south. Extensive plough-damage was apparent; none of the east wall (136) survived and only a suggestion of an east-west wall was present at the extreme southern extent of the west wall 171. No features were recorded, and the dark brown gravelly loam fill over the room produced only very small fragments of roofing material and no pottery.

The presence of five oven-type structures strongly suggests semi-industrial processes taking place on a domestic scale. The large stone-built oven in Room W1 may have had a domestic use for cooking or baking or could have been a heat source for forging activities associated with the small ironworking hearth in the same room. The function of the oven in Room W2 is not known, but the form of the ovens in Room W3 makes it likely that these structures were also associated with ironworking. It is probable that iron fittings used in the villa buildings were manufactured here. A deposit of loose, clean tesserae found to the north of the building suggests the preparation of ceramic material for tesserae took place here.

Conclusions It is likely that the single-roomed Phase 1 building was constructed in the late 2nd-century (Period 3.1), when evidence suggests that the non-domestic Ancillary Building in Eleven Acres (Chapter 4) was going out of use. Major re-building probably occurred in the mid-3rd century (Period 4.1), when the small rectangular structure was replaced by a four-roomed building on the same alignment using the north and east walls of the earlier building. Constructional techniques of flint mortared walls matched the East and North Ranges and formed a complex of three buildings arranged around a courtyard. The main roofing material was limestone tiles; their putative

The remains of at least four infants were recovered from Room W1. Three were buried near wall edges and loose bones from a fourth baby came from the nearby ironworking pit. The small hoard of 15 coins dating to the mid-4th century could have been hidden in a loft space, only to fall when this collapsed in the late 4th/early 5th centuries.

138

8 The grain dryer and associated burial The dryer was located in the extreme south-east corner of Eleven Acres and was excavated during 2017. The structure in Trench 87 was sited about 75m from the south end of the West Range, some distance from the villa complex (see Fig. 1.6). The eastern edge of the stoke pit was adjacent to ditch 1641 which infilled during the 2nd century CE. This well-made structure (Fig. 8.1) comprised a T-shaped flue (1758) and associated stoke pit (1964) aligned west-east; the latter contained a skeleton (1957) and the implications of this are considered in the stoke pit description.

level floor and was 2.95m long with an internal width of 0.5m. Prior to construction, excavation confirmed that the ground had been prepared by digging a T-shaped trench into which the dryer was built. It is likely that the stoke pit was dug at this time. Cream chalky mortar (2297) was laid against the construction trench edges; on the surface this was between 0.3m to 0.55m wide. This material was also used as mortar between the courses. The walls of the structure were given separate context numbers to aid excavation (1896, 1897, 1898, 1895, 1925, 1926 and 1899). The fills of the dryer were removed in sections (1850, 1759, 1969, 1969, 1963, 2074 and 2075) which are located on Figure 8.2/2.

The grain dryer Constructed of limestone blocks, flint nodules and occasional ceramic tile, the dryer was completely excavated. There did not appear to be any other associated structures, although investigations only took place in its immediate vicinity. The structural components, wall numbers and section locations are illustrated on Figures 8.2/1-2.

The flue walls These very well-built walls survived almost intact. The west wall of the cross flue (1895) comprised five to eight courses of neatly laid, knapped flint nodules and limestone blocks (Fig. 8.4/3). Small limestone slabs capped the wall and one third of these remained in situ. Both ends of the wall were constructed with limestone blocks, as was the central portion, and there was no evidence for burning or scorching. Small amounts of ceramic roof tile and heathstone were also incorporated. The short walls 1898 and 1925 had a greater proportion of flint with eight and five courses respectively surviving. The west-facing walls

The flue (Fig. 8.3) The long arm of the flue was 3.4m with an internal width of 0.5-0.6m and a maximum surviving height of 0.6m; the floor sloped from west to east with a 0.2m change in level. At right angles to this, the cross flue was 0.75 high, had a

Fig. 8.1. Dryer 1758, stoke pit 1964 and skeleton 1957, view east

139

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 8.2. Dryer 1758 plans with context numbers and excavated section locations (1-2)

of the cross flue (1897 and 1926) were characterised by precisely laid limestone corner blocks which were bonded to walls 1896 and 1899 (Fig. 8.3). The east-west flue walls (1896 and 1899) were constructed in a slightly different manner (Fig. 8.4/1-2). The use of limestone was evident at the mouth of the flue; much of this was burnt, particularly on the side of the north wall 1896. An area of burnt flint nodules was noted on the south wall 1899. After the dryer was built, a layer of red/brown clay (1932) between 0.05 and 0.08m thick was applied to the flue base.

and infilling sequence; their locations (A-D) are noted on Figure 8.3. Section 1850 (A) (Fig. 8.5/1) Located at the junction of the cross flue and the main flue, three fills (excluding clay 1932) were recorded. A thin layer of charcoal and clay (1930) had accumulated over 1932. Over this, 1922 comprised chalky clay which contained building material dominated by large quantities of flint nodules. One Bidwell Type 5.1 beaker sherd dated c. 50-100 CE could have derived from earlier activities associated with ditch 1641. The upper fill 1851 was a

The dryer was excavated in sections, four of which are illustrated (Fig. 8.5/1-4) and demonstrate the collapse 140

The grain dryer and associated burial

Fig. 8.3. Cross flue of dryer showing construction method and building materials, view north

raking out. Above 1932, six fills were recorded. A sherd from a BB1 flanged bowl dating c. 240-300 CE came from charcoal-rich layer 1941. The chalky loam fill 1947 contained a limestone block and a burnt limestone slab probably from the side of the flue. Above this, the grey chalky fill 1934 contained limestone blocks and fragments of ceramic roof and floor tile. Fill 1921 was similar to 1934 but differed by the presence of many flint nodules. The upper fill 1851was also present (and described) in section 1850.

chalky mortar with quantities of building debris. A single iron nail and bones from cattle and a dog were recorded. Section 2075 (B) (Fig. 8.5/2) This south-facing section had a flat base characteristic of the cross flue. The base layer of clay (1932) was up to 0.08m thick. Three fills were recorded. The lowest (1972) was dark brown loam with a high percentage of charcoal, occasional lumps of chalky clay and fragments of ceramic roof tile. Above this layer 1958, a chalky loam with occasional charcoal pieces, contained building material including limestone building stone and roof tiles, as well as building flint and tegula fragments. The upper fill (1892) comprised loose chalky mortar within which were many flint nodules and stone and ceramic building material, most likely derived from the collapse of the structure. Also present were three pieces of opus signinum (1450g), animal bone fragments from cattle and sheep/ goat, and a BB1 jar rim dating c. 280-370 CE.

Section 1969 (D) (Fig. 8.5/4) The context numbers 1932, 1934, 1941 and 1947 related to layers accumulated throughout the length of the flue and are described above. On the base of this section, fill 1971 comprised charcoal-rich loam which contained three sherds from the same vessel noted in fill 1941. Over this was a band of chalky clay (1951); this collapsed layer may have formed the underside of the floor of the flue. Within layers 1947 and 1934 was a jumbled deposit of limestone blocks and building flint, probably from both the walls and floor of the structure; large chunks of ceramic floor tile were also recorded.

Section 1963 (C) (Fig. 8.5/3) Located about 0.5m from the flue entrance, this section profile was rounded at the base, possibly by the action of 141

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 8.4. Flue wall profiles sections 1896, 1899 and 1895 (1-3)

Fig. 8.5. Dryer 1758 sections 1850, 1963, 2075, 1969 and stoke pit 1964 (1-5)

142

The grain dryer and associated burial The stoke pit and burial

end of the pit. The remains were those of an adult male who had suffered healed trauma injuries prior to death. The body was on its side, facing west. The legs were slightly flexed, the left leg uppermost. The arms were flexed, the open hand of the right arm was near the face and the left hand was bent back at a right angle to the arm. The pit was too small for the remains, resulting in the rather squashed position of the body. It is likely that the body was placed in the pit shortly after death before rigor mortis had set in. There were no grave goods. Chalky layer 1806 was likely to be a deliberate infill after the burial. A femur was submitted for radio carbon dating and returned dates of 240-381 cal CE: 95.4% probability (SUERC-84712), strongly suggesting that use of the dryer had ceased by the mid-4th century CE.

The stoke pit 1964 (Figs 8.1, 8.2 and 8.5/5) Located at the east end of the flue, the stoke pit was cut into clay/loam, measured 2m by 1.5m, was 0.95m deep with vertical sides for half of its depth and had a slightly rounded profile sloping towards the flat base. Three fills were recorded; the upper (1760), a clay loam, contained small amounts of building material together with large base fragments from a C2 storage jar (c. 370-430+ CE) and four small, abraded sherds from a vessel dated c. 450-650 CE. The undated middle layer 1806 was 70% chalk in loam with small amounts of building material. The base layer (1907) was 0.55m deep and comprised mid brown loam with chalk flecks and occasional lumps; the upper 0.3m of this was almost finds-free. Below this were large quantities of limestone and ceramic building material derived from the dryer. Underneath this were a large number of cattle bones, some of which were articulated, spread across the feature (Fig. 8.6). As the bones were removed, it was apparent that they covered human remains. These human and animal bone deposits are considered in Chapters 16 and 17. A deposit of flint and ceramic roof tile (1954) edged the pit near the base (Fig. 8.1) and may have been part of the original lining of the stoke pit.

Conclusions The dryer was located outside the villa complex and, unfortunately, it is not possible to either date it or to link it with any of the constructional phases of the villa. Such features were relatively common on villa sites. The dryer was built with care, incorporating squared blocks of limestone; this material was rare on the villa itself and the question remains open as to where it originated. Its function was to artificially dry grains. Charred material from the dryer is discussed in Chapter 18. It is not possible to determine the period of use of the structure, although pottery from the base of the flue dated c. 240-300 CE and may relate to a final firing just before abandonment. Pottery recorded from the upper fills indicates material deposited after collapse.

The skeleton (Figs 8.1, 8.2/1, 8.5/5 and 8.7) A complete skeleton (1957) was located on the base of the stoke pit, orientated north-south with the head at the north

Fig. 8.6. Stoke pit 1952 during excavation with cattle bone deposit from layer 1907

143

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 8.7. Skeleton 1957 on the base of stoke pit 1952, view east

The dating of the burial in the stoke pit confirms that this small complex had gone out of use by the mid-4th century at the latest. The placing of a body in an outlying pit at this time, rather than in a regulated cemetery, hints at a ‘deviant’ burial. The placement of cattle bones over the corpse perhaps suggests ritualistic behaviour. The presence of large quantities of heavy material over the two sets of remains and the chalk layer above this, strongly point to a deliberate filling-in of the grave site.

144

9 Building materials This chapter considers materials used in the construction and fabric of the Druce buildings.

the building do not appear to have been strengthened; the north-east corner was slightly rounded (Fig. 9.1).

Wall construction

The Phase 1 domestic North and East Ranges were contemporary and constructed during Period 2.2. These buildings were altered and extended up until Period 5 and continued to use the coursed flint in mortar construction. Where survival was best (Room N3 in the North Range, see Chapter 5, Fig. 5.12), the flints were laid in herringbone pattern, and it is probable that most of the villa buildings were constructed in this way. The North Range was notable for the use (and survival) of large Purbeck limestone blocks used as quoins.

Introduction All building phases of the villa used flints bedded in mortar for walls with occasional limestone blocks or ceramic roof tiles for quoins. The flint was probably extracted locally from the underlying clay-with-flints. The limestone derived from the coastal Purbeck/Portland beds. The origin of the ceramic tiles is not known. The archaeological details of the Ancillary Building and the North, East and West Ranges are considered in Chapters 4-7; relevant aerial shots (Figs 4.2, 4.3, 5.3, 5.13, 5.17, 6.15 and 7.1) show these buildings during their final stages of excavation.

When the fabric of the Period 5 Aisled Hall deteriorated, the north wall collapsed southwards (Fig. 6.15). Subsequent ploughing removed much of the flint, but what had been the internal wall surface survived where it fell. Densely packed flint was also used in pits to support aisle posts (Fig. 6.20).

As the buildings were excavated, each wall was numbered. The widths, surviving height, lengths and orientations were recorded together with joins and butts. The presence or absence of mortar and matrix inclusions was noted. This information is held on Excel sheets in archive. Table 1.1 records the relative phasing and periodisation of the buildings. Early in the excavation, a decision was taken to uncover the walls and to leave them intact. However, to help elucidate construction methods, a number of sections were cut through walls in the Ancillary Building and in the West Range. There was no evidence that the exterior surfaces of the walls were rendered or painted.

The West Range had two building phases (Fig. 7.1) and may have replaced the Ancillary Building, which went out of use as the Phase 1 building was constructed. Investigations on the east wall confirmed that there was no bedding trench and that the closely packed flints in thick white mortar were laid directly onto the clay/loam (Fig. 9.2). Non-flint internal walls All internal walls of the North and West Ranges were built using flint in mortar. It is possible that the upper parts of some of these walls were of wattle and daub comprising thin, flexible, woven wooden poles, most likely of willow or hazel, which was then covered with a daub of clay, chalk and straw (Rook, 2013, 47). Within the large assemblage of fallen wall plaster from Room N1, such material was recorded adhering to the back of the painted plaster (see Morgan, Chapter 10). Some of this probably derived from the internal wall 235 which divided rooms N1 and N2.

The Ancillary Building constructed in the mid-1st century CE (Period 2.1) was the earliest. Coursed flint in mortar was used throughout, both for external and internal walls. The building was constructed on a slope, resulting in a gully at the north end to facilitate the construction of the north wall. Shallower gullies were associated with the east and west walls. In contrast, the foundation courses for the south part of the west and south walls were built up with hoggin to compensate for the decrease in slope. The first course of flintwork was set directly onto the natural clay/ loam. The external walls were 0.45m to 0.5m wide, with larger flints laid lengthwise on the inner and outer edges. It was noted that the outer flints slightly sloped downwards. Between these was a matrix of tightly packed smaller flints with occasional pieces of broken ceramic roof tile, limestone, heathstone and conglomerate. Fragments of pottery were also noted, which included sherds from amphorae, a flagon and various jars all with a broad late 1st century date. This material provided bulk and strength (Pearson 2006, 75), although pottery fragments were not present in the walls of the villa complex. The corners of

In the Phase 4, Period 4.2 Aisled Hall, two internal walls of flint in a cob/clay/mortar mix were recorded at the north end of the building. These were small lengths of an east west wall 775 and a north-south wall 856 (Fig. 6.23). There is a strong likelihood that other internal walls are preserved under the fallen wall. Mortar used in the construction of the building All mortar which had fallen from ceilings and walls, together with deposits in pits and ditches, totalled 4.13 145

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 9.1. Ancillary Building terraced into the slope with north and east gullies visible. Note the rounded north-east corner of the building, view south-west

Fig. 9.2. West Range, Room W1; wall 142 constructed directly onto natural clay/loam with the collapsed wall to the right, view south-west

146

Building Materials Roofing and ceramic building materials John Bithell

tonnes. In 2014, 2.72 tonnes were removed for analysis, but unfortunately the material was not returned and the ‘specialist’ was unavailable for further contact. Samples of the 2015 and 2016 mortars were retained but have not been further investigated. However, basic information on colour and texture was recorded on relevant context record sheets.

Introduction Ceramic Building Material (CBM), and stone roofing tiles from several stone sources, was recovered from all phases of buildings, pits, ditches and middens on the villa site, and indicates a variety of roofing styles through time.

Two specific mortar types are considered, that used as a bonding agent between the wall flints and that used on the roof to secure the ceramic imbrices. The wall mortar was recorded in all the buildings and varied in quality, preservation, and colour. A number of walls appeared to be bonded with clay loam, but it is likely that in these instances the mortar had decayed. A lime mortar was used which would have been approximately three parts sand to one of lime (Rook 2013, 53). The lime was probably produced in the vicinity of the villa by carefully burning chalk and/or limestone. Chalk was probably the preferred option as it was locally abundant. Indeed, the field where the three major villa ranges are located is known as ‘Lower Limepits’ and a very large pit, probably for the extraction of chalk, is located west of the Ancillary Building in Eleven Acres. Both are visible on Figs 1.3 and 1.5. All mortars were pale in colour and were recorded as white, grey, cream, yellow and, in one instance, pink; the coloration may have been affected by depositional circumstances. No chemical analysis was undertaken.

Methodology Items were weighed and recorded by context on an annual basis. Identifying and categorising the CBM was based on Warry (2006). All fragments with recordable data were separated into groups which comprised: • tegula (upper right and left corners, lower right and left corners, plus any fragments with pierced holes or signatures) • imbrex • box flue tile • brick and tile • stone tile Items with recordable data were marked with context numbers and a discrete CBM or stone tile number. These details were recorded on site and transferred to Excel spread sheets (held in archive); Table 9.1 records the categories by number. All CBM fragments with recordable features were photographed. Non-identifiable fragments with less than two recordable features were discarded. No fabric analysis was undertaken.

Mortar used to attach the ceramic roof tiles was always white and was described as ‘architectural mortar’; c. 18.6kg was recorded from the villa buildings. That which was associated with the imbrices (Fig. 9.3) must have been applied with the tile held upside down, as the mortar completely filled the semi-circular voids.

Ceramic building material Ceramic roof tiles In total, 9.9 tonnes of ceramic roof tiles were recovered from the site. The assemblage comprised tegulae and imbrices; including all pieces containing nail holes, signatures, plus other unidentifiable pieces which were Table 9.1. CBM and stone tile categories: numbers of recorded items Type

Fig. 9.3. Mortar from the inside of an imbrex showing tegula impressions (above) and the ‘domed’ internal shape (below). From the Phase 3 Aisled Hall fallen apex 827

147

No. recorded

Imbrex

   35

Large tegula

   15

Lower cutaway

 276

Upper cutaway

  139

Signature

   48

Lydion/other CBM

   61

Tegula nail hole dimensions

  64

Complete stone tiles

  235

Stone tile fragments with nail holes

  621

Totals

1494

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) recorded as fragments. Four aspects of tegulae comprising upper and lower cutaways, flanges and makers signatures were examined and recorded.

lower tegula. The lower end of the roof tile must therefore be narrower than the upper end to form this overlap, and the interlocking system of two tegulae must be matched for a width-fit, usually from the same mould. The overlap is enhanced by a series of cutaways. Upper cutaways allow the lower cutaway of the tile above to slide into them, and thus hold the tile in place. The lower end of the tile also has a cutaway; this was modified over time, allowing an assessment of differing styles and fashions (Warry 2006, 4). The amount of overlap is dictated by the lengths of the upper and lower cutaways. An overlap spread varying between 120mm and 100mm is evidenced by the examination of the mortar wash on the insides and outsides of imbrices (Fig. 9.5). This is substantiated by the length of upper and lower cutaways, allowing for the greater variety of lower cutaway types (Figs 9.6-7).

No complete tegulae were present in collapsed roof contexts, but two were utilised as the floor of an oven flue (780) and a stack of seven (1129) were attached to the flue (see Fig. 7.6) in Room W1 of the West Range. In this room, fragments of shattered tegulae and imbrex were found under threshold (542) where a door sill would have been positioned (Fig. 7.4). Complete imbrex survive in greater numbers than tegulae and examples were found as part of a roof fall (827) relating to Phase 3b fire-damage of the first Aisled Hall. Many of the imbrices retained mortar, which may have contributed to their survival as complete units (Fig. 9.3). The presence of early, whitecoloured tegulae (Type C5, c. 160-260 CE, Warry 2006, 4) and white-coloured imbrex fragments suggest that a patterned application of red and white tiles was present on the south facing slope of the roof over corridor N11 on the North Range.

The flanges forming the tegula sides vary in thickness, being wider at the bottom. Flanges from different kiln sites often have different shapes, which can be square, rounded or fluted. A small number (14.2%) of Druce lower flange ends were not square ended, but had rounded ends (Fig. 9.8), or were concave, convex or with a straight slope. The upper ends of some tegulae had pre-fired holes for the insertion of iron nails used to secure the tiles to wall plates and planks on the rafters. Lime mortar wash, and adhering mortar on the surfaces and the flanges, indicate that these tiles were mortared into position, with imbrices covering adjacent tile flanges. This combination of interlocking tegula, nails, mortar and imbrex ensured a watertight fit and a roof solidly secured to the structure. A single fragment of a lower cutaway tegula from the internal rubble fill of oven 692 in Room W3 in the West Range had a post-firing hole close to the base (Fig. 9.9); the hole had been worked from the upper surface and during the process had spalled around the lower edge.

Tegulae The basic tegula shape is trapezoidal, with the upper part being wider than the lower, and with raised flanges along the long sides. These items are usually made in a mould. Figure 9.4 illustrates the defining characteristics. The method of manufacture is covered elsewhere, but notably in Cunliffe (2013b) and in greater detail in Warry (2006). Tegulae need to overlap and interlock vertically to form a solid roof for the prevention of weather ingress (rain, snow and wind). It follows that the lower end of the tile, when fitted, must sit on and above the upper level of the

Fig. 9.4. Defining characteristics of a tegula A= Upper cutaway, B= Flange, C= Lower cutaway, D= Signature

Fig. 9.5. Imbrex with internal mortar wash

148

Building Materials

Fig. 9.6. Lengths of upper cutaways in mm

Fig. 9.7. Lengths of lower cutaways in mm

149

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 9.8. Tegula with rounded end of lower cutaway, collapse layer 494, Room W1

The allocation of tegula fragments to types C5 (c. 160 to 260 CE) and D15 (c. 240 to 360 CE) by Warry (2006) is frequently a subjective assessment rather than a distinct differentiation of cutaway and, as such (on this site), provides an ambiguity of dating. However, clear phases of use are evident; the first probably centred on c. 200 CE, a second phase at about c. 300 CE followed by a final phase of new roofs (principally of limestone) in the later stages of the villa’s occupation. The breakdown of cutaway types by period is detailed in Table 9.2.

this. The roofs of the Phase 3 North Range and Phase 3 East Range (first Aisled Hall) appear to have been roofed with calcareous mudstone. Dumps of these were located in the Phase 3 (Period 3.1) mortar mixing basin under Room N7, associated with the extension work in the North Range, and in tank 778 under the Phase 4 (Period 4.2) extension of the Aisled Hall. At Druce, ceramic roof tiles are likely to have been used for ridges and perhaps for decoration. A single mammata tegula tile fragment (Fig. 9.10), from midden 1528, appears to be a broken example of Brodribb’s (1987) Type A. This tile type has between one and four mammae (or bosses), formed from lumps of clay attached to the tile surface, and was used on floors, or in levelling courses in walls where the mammae assisted bonding with the underlying cement (Brodribb 1987, 60-2). These tiles do not have a smooth finish.

Only the Period 2.2-3.1Ancillary Building provided evidence for a wholly ceramic roof. It is possible that the roofs of the Phase 1 North Range and the Phase 1 and 2 East Range were the same, but there is no absolute evidence for

Signatures Crude ‘signatures’ on the upper surfaces of tegulae were impressed into the tile by finger-tip when the clay was probably still in the mould. Nine were identified at Druce (Fig. 9.11). Some 80 out of 276 (28%) lower cutaway fragments had identifiable signatures (Table 9.3). The greater percentage (62%) was from signature group E (Warry 2006, 149), and this group was considered to have potential for further study. It was noted that the top loop size varied from relatively small to quite large, and this may have indicated a different hand within the tilery, or a different period of manufacture. Of the 80 lower cutaway fragments with a type E signature, only 20 had complete loops, and of these, only 14 had lower cutaways enabling allocation of a tegula type. Measurements were taken of these loop sizes (Fig. 9.12), and three sets of distinct dimensions were noted at 50 to 58mm, 40 to 45mm and

Fig. 9.9. Tegula pierced after firing, from collapse over oven 692, Room W3

150

Building Materials Table 9.2. Numbers of lower cutaway ceramic roof tiles by Warry type and production date Warry (2006) cutaway Type

Type 4

Type 5

Type 6

Type 15

Type 16

Fragments

Date (CE)

160-260

160-260

100-180

240-380

240-380

160-380

Numbers

18

18

4

188

3

45

Table 9.3. Tegulae signatures identified by Warry (2006). If followed by a number, this indicates the number of curves/repeats of the signature. L = a large signature Warry pattern

A

B

B2

B3

BL2

E

Q or N2

P

R

S

S2

S3

SL1

SL2

SL3

T

Totals

2

1

4

1

3

49

1

1

1

8

3

1

2

1

1

1

Perforations Prior to firing nearly half of tegula upper fragments (45%) were pierced with a single hole. Unusually, two of the 65 perforated tegula fragments were pierced with equally spaced double holes. The holes allowed the tegula to be secured to roof beams below and created a firm position for the lowest layer of roof tiles on a wooden wall plate. The hole diameters varied between 5.6mm and 11.4mm, and all were pierced from the upper side of the tile. A tile from room N3 of the North Range had a pre-fired score line and remnants of an iron nail embedded in the hole (Fig 9.13). The assemblage of perforated tiles was spread across the full area of the villa site. Alternative uses of tegulae Tegulae were used for purposes other than as roof tiles. Two were placed upside down as a level surface on the floor of the flue of the oven 706 in Room W1 of the West Range (Fig. 7.6). This created a smooth surface to act as a base for shovelling out ash from the oven. Alongside and to the east of the oven, seven layers of tegulae had been stacked to build a work surface. The top layer contained a tegula that had been cut to form a circular depression, where perhaps a container may have stood.

Fig. 9.10. Mammata fragment with applied boss, context 1370, midden 1528

Tile evidence suggests that two varieties of string course construction may have been used, although no evidence of this building technique was encountered on extant or fallen wall surfaces. A tile from the apex collapse of the first Aisled Hall (context 827, Fig. 9.14), has a thick mortar layer on the upper surface, and contains the ghost of a flint nodule where it sat on the flint and mortar fill of a wall. Another tile, from a layer of pit 944, shows at least three clean blows to remove the flange (Fig. 9.15), to facilitate its use in a string course. Broken tiles were used to plug putlog holes from scaffolding as evidenced in the fallen wall of the Aisled Hall (Fig. 6.31)

Fig. 9.11. Pre-fired signature types from Druce villa (after Warry 2006, with permission)

Manufacturing, handling and stacking marks

15 to 33mm. Examination of the cutaway types available for comparison, failed to reveal a precise pattern within the three groups of loop size. There is, however, clear indications of three different hands associated with the use of this signature type.

The upper half of a tegula from the roof collapse over Room N1 in the North Range, was reconstructed from fragments and shows inscribed lines, set out to allow the positioning of two holes which were pierced prior to firing 151

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 9.12. Grouping of Type E signature sizes

Fig. 9.14. Tegula with ’ghost’ impressions of flint nodules in the mortar from the collapsed apex 827 in the Aisled Hall

Fig. 9.13. Tegula from Room N3 with score marks for the location of a pre-fired nail hole and an iron nail in situ.

Fig. 9.15. Evidence of hammer strikes to remove the tile flange for secondary use in a string course from pit 944 in the northwest corner of the Aisled Hall

152

Building Materials (Fig. 9.16). Slight indentations on the sides and tops of some tegulae may have been caused by handling when wet. Based on the absence of animal or other marks, it is likely that the tiles were dried vertically in racks. There are no tile stamps within the assemblage. Graffito Only one possible graffito was noted. This occurred on a tile from the collapsed roof over Corridor N11 in the North Range. The lines were inscribed before firing (Fig. 9.17), but their use and function are not known. Examination under oblique light suggests that they could be an impression from a rectilinear object, possibly an apron strap or hinge. Imbrices An imbrex is a semi-circular roof tile (Fig. 9.18), narrower at the upper end to allow overlap to aid the shedding of rainwater from the top to bottom of the roof. The tiles are manufactured by laying the clay over a wooden ‘former’, prior to drying and firing in a kiln, thereby creating a standard size for length and width at the upper and lower ends. Variations of these parameters indicate different formers and may point to separate manufacturing kiln groups. No signatures were found on Druce imbrices. When fitted to a roof, the tiles are positioned to sit over the outer flanges of adjacent tegulae and are held in place by lime mortar so that the imbrex and mortar provide a solid fill (see Fig. 9.3/2), giving rigidity, and producing watertight integrity to the slope of the roof. Mortar wash on the inside of imbrices was noted on complete or large fragments. Measurements were taken from tiles recovered from the roof fall (827) of the upper ridge of the Aisled Hall (Figs. 6.9 and 6.16). The survival rate of imbrices within this context was enhanced by the inclusion of mortar, which was still in place within the arch of some of the tiles and adhering to large fragments. As expected, the upper width of individual imbrex was smaller than the

Fig. 9.17. Tegula from corridor N11 with pre-fired rectangular inscribed lines

lower width, and all imbrices would have fitted to the roofs with overlaps of approximately 100mm. The upper widths varied between 130mm and 95mm, while lower widths were between 180mm and 135mm. Imbrex lengths varied from 375mm to 360mm. Finishes A number of tegula and imbrex were fired with a white or cream finish. The tegulae proportion of cream/white to normal firing was found to be 15 of 462 fragments from upper and lower ends, or 3%, and that for imbrices was eight of 36 or 22%. The disparity in percentages is probably caused by the discard of tegula fragments with no recordable features. Differing firing temperatures produced a variety of shades from pink and red to almost

Fig. 9.16. Tegula from Room N1 with pre-fired score marks for two nail holes

Fig. 9.18. Complete imbrex from phase 4.2 infill, Room N3

153

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) brown; this may have been utilised by the builders to provide decorative variation on the roof. Box flue tiles The presence of box flue tiles is enigmatic as no evidence for heated rooms or a bath suite was located. The earliest fragment was retrieved from a fill of ditch 1641 in Eleven Acres; the associated pottery dating c. 90-200 CE, suggesting a Period 2.2 to 3.1 for use of the tile. Box flue fragments were found embedded in the fabric of flint walls and in ditch, pit and midden fills, but there was no evidence in any building for a hypocaust. The buildings saw major reconstruction over three centuries and such a feature could have been dismantled. It is possible that an early bath house may have been located to the south in Watery Mead, where geophysics highlighted anomalies (see Figs 1.5 and 19.1). The 46 fragments with recordable features included three joined fragments but no complete box flue or tubuli. Corner fragments and end pieces were recovered, and indications of cut-outs in side faces that had no comb marks were occasionally identified. A box flue fragment (Fig. 9.19) from a layer in the central part of the East Range contained a horizontal casting seam caused by a two-piece former being used. This is very unusual, and has not been seen before, as withdrawing two pieces of former from different ends of the box flue whilst the clay is wet would require greater skill than simply pulling out a single box-shaped former. Occasional handling impressions in the form of finger indentations were observed (Fig. 9.20). Tile-combing was achieved with a 6-tine comb on all fragments except one from a section of ditch 313. Only three tines are visible (Fig. 9.21); two were 8mm apart and a third at nearly 18mm, suggesting the use of a damaged 4-tine comb.

Fig. 9.20. Box flue tile with handling indentations from ditch 1260

differ from tiles in that they tend to be rectangular and flat with no flanges. Round bricks were produced but are absent from Druce. None of the Druce Farm assemblage matches the standard dimensions of lydion type (440mm by 290mm by 140mm) suggested by Brodribb (1987, 37). These bricks were generally used as bonding/levelling courses in walls and could be used for flooring. Two lydiontype bricks formed the corner of the north-west corner of Room W1 of the West Range and measured 320mm by 300m by 40mm, and 300mm by 240mm by 40mm (see Fig 7.2). A third, similarly sized, near-complete brick was excavated from the fill of Ditch 310 on the south side of the site. A plot of thicknesses of the bricks showed that 81% of the assemblage was within a very narrow band (Fig. 9.22). A small number of bricks had a very distinct chamfered

Bricks Fragments from a total of 60 bricks (lydions) were recorded, including three complete or near-complete items. Bricks

Fig. 9.21. Box flue tile scored with a damaged 4-tine comb from ditch 313

Fig. 9.19. Box flue tile with ‘former’ seam impression (deposit 716, Phase 4.2 Aisled Hall)

154

Building Materials

Fig. 9.22. Range of lydion thicknesses in mm

edge. An example (Fig. 9.23) came from the charcoal layer 1946 in Room AB2 of the Ancillary Building; associated pottery dates this layer to c. 70-100 CE. Brodribb suggests that two such sloping edges placed together aided bonding and is rare (ibid., 40). Several fragments had pre-fired, wide perforations (Fig. 9.24), and a few fragments had a series of longitudinal semi-circular grooves (Figs 9.25 and 26). One of these also displayed a perforation. The function of these pre-fired features is not known but McComish (2015, 16) terms tiles displaying such features ‘parietalis’. The holes were employed to fix them upright to walls with nails or clamps and the grooves as a key for plaster. It is significant, however, that the fragments came from the earliest occupation in the Ancillary Building, and

Fig. 9.24. Lydion fragment with large pre-fired hole from ditch 313

Fig. 9.23. Lydion fragment with a chamfered edge from Room AB2, Ancillary Building

Fig. 9.25. Lydion fragment with cross grooves from ditch 1641

155

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Stone roof tiles John Bithell Introduction A total of 52 tonnes of roofing stone was excavated and recorded. Four different stone types were recognised and comprise limestone, calcareous mudstone, sandstone and slate. The locations for the extraction of the stone types have not been determined, although the limestone is likely to have been sourced locally from the Purbeck Beds, the Ridgeway or Portland Bill, with the calcareous mudstone coming from the local Lias or Mercia Groups and the sandstone from the Jurassic or Cretaceous beds. The slate is considered by Stanier below. Stone tile typology Fig. 9.26. Lydion fragment with cross grooves and perforation from ditch 1641

The Druce stone roof tiles are of four types, each having defining features and specific uses (Fig. 9.27). 1. Standard tiles are elongated, near pentagonal in shape, with a 90° chamfered 2. end, and comprised the main element of the roof covering. 3. Ridge tiles are not elongated, have much shorter sides, are chamfered and are the 4. final tiles positioned at the ridge of the roof. The ridge is then provided with a water-tight seal, for example, by mortaring to the ceramic imbrex. 5. Eaves tiles are standard tiles but have no point and are chamfered along the lower 6. edge: they form the first row of tiles on a wooden wall plate at the base of the roof. 7. Gable tiles are effectively half-standard tiles and form the end edges of each layer of ascending tile that forms the roof.

from the fills of ditches 1641 and 313, which all dated to the 1st/2nd centuries CE. Uses for surplus and broken CBM Surplus and broken CBM were utilised as tesserae, which were cut from roof and box flue tiles. Two discrete deposits were located north of Room W1 outside the West Range (context 141, Fig. 7.5), as if newly cut and ready for removal and use. Deposits of much smaller tesserae were located in the Period 4.2 activity in Room N3. This material was used extensively as tesserae in all the mosaics on the site. The red floor of Room N6 in the North Range (Fig. 5.20) was overwhelmingly composed of tesserae from roof and box flue tiles, with occasional limestone pieces. CBM fragments were included as ‘rubble’ within the lime mortar mix of several walls, their presence indicating later building work. They could also be crushed and added to mortar to produce opus signinum, which was used as a waterproof cement and floor surfaces.

A display of tiles was set up and used on site Open Days and illustrates how these tiles would have been located on the roofs (Fig. 9.28).

Conclusions

The production of stone tiles was a skilled craft, requiring expertise to ensure that the finished tile was a standard shape, with chamfered edges and a perforation for a nail

Tegula and imbrex were used as roofing in all phases of the villa buildings. No complete tegulae survived the collapse of the villa other than those used in the oven in room W1. However, a small number of complete imbrex did survive due to the fact that they were supported by the underlying mortar which fixed them over the flanges of the tegulae. This was visible in the Aisled Hall, where the height of the fall when the roof collapsed was greatest and where the raising of the floor associated with the Phase 4 building work secured their survival. CBM was cut to form tesserae which were incorporated into every mosaic floor on the site. The use of complete ceramic tiles in oven 706 in the West Range suggests that this material was used for a variety of functions, of which only these examples remain.

Fig. 9.27. Stone roof tile types 1-4 L = length, W = width, BH = distance from central line to shoulder (after Warry 2006 with permission)

156

Building Materials 8.42m with a roof pitch at 37°, and 37.6 tonnes at a roof pitch of 34°. The calculations and choice of pitch for this are held in archive. Because of the spread of nail positions, it would seem that the tiles were nailed to wider planks rather than narrow rafters. Calcareous mudstone tiles A total of 12.8 tonnes of fragments of calcareous mudstone tile was recorded across all ranges. Two complete standard roof tiles (Fig. 5.17) were set into the base of a mortar mixing basin (715) in the north-west corner of Room N7. One had been placed with the upper surface face up; the other was face down. These two roof tiles are the only complete examples of this stone type. They are similar to other roof tiles in that the upper surface has chamfered edges to aid rain water movement downwards towards the eaves of the roof. A large amount of shattered and smashed calcareous mudstone was deposited on top of the two tiles. Similarly, tank 778 in the Period 3 Aisled Hall was filled with this material. Both of these features were covered by later floors, suggesting that early roofs were clad in this material.

Fig. 9.28. Display of limestone roof tiles using all stone tile types, set up as an on-site display

which secured the tile to planks or rafters. It is likely that the tiles were manufactured as blanks at the quarry and the holes inserted on site. This single hole was bored from upper side, leaving a marked spalling on the underside of the tile. Occasionally this process failed and a second hole was bored; by necessity this tended to be more centrally placed.

Sandstone tiles

Limestone tiles

A total of 1231kg of sandstone tile fragments were recovered, the vast majority deriving from the North Range and most from contexts associated with collapsed roofs at the front of the building. They were generally fragmented and degraded, although a single leaf shaped tile with no securing hole (Fig. 9.29) was recovered from the area south of Room N3. Its function is unclear, but it could have been used as a decorative feature, possibly as a ridge tile insert between tegulae immediately below the ridge capping.

A total of 239 complete limestone roof tiles were recovered from all three ranges of the villa and includes all tile types. This is a particularly high number and is perhaps second only to the number recorded at Frocester (Gracie 1970,19). The standard roof tile would have produced a diamond patterned roof (Fig. 9.28). In addition to the complete tiles, 621 perforated stone tile fragments were recorded. The roof tile survival rate, when falling into rooms as the villa collapsed, is high in comparison with the almost total failure of CBM to survive as complete elements. Falling stone roof tiles damaged the pavement in Room N1, causing deep indentations and lacunae in the mosaic.

Conclusions The roofing material from Druce villa illustrates changing styles and taste over a period of three hundred or so years. Constructed in the mid-1st century, the Ancillary Building was the earliest structure on site; ceramic tiles were used as roofing. The early phases of the North and East Ranges dating to the late 1st/early 2nd century used calcareous mudstone with CBM for ridges. The early phase of the West Range may have been roofed with CBM; many such tile fragments were recorded in and around this building. The later phases of the North, East and West Ranges were covered with limestone roof tiles; the pent corridor N11 roof and porch of the North Range was probably decorated with sandstone tiles, and the East Range with slate (see Stanier below).

The stone roof tiles were recorded using Warry (pers. comm.): the shapes and average dimensions were calculated (Table 9.4); there was very little variation of tile width across the site. Using these figures, a roof weight of 39.4 tonnes was estimated for the main North Range based on the range length of c. 28.62m and width of c. Table 9.4. Average dimensions and weights of stone tile types. The hole height was measured from the pointed base of the tile. All holes on standard tiles were pierced to the left of the centre line. Tile type Width Length Base Hole Weight No. of (mm) (mm) Height Height (g) tiles (mm) (mm) Standard

263

399

131

362

3610

199

Left gable

185

323

119

215

2109

  17

Right gable

181

307

115

286

2224

  13

Ridge

199

201

131

170

1503

  6

Eave

206

211

200

none

2391

  4

As the buildings were renovated and rebuilt, defunct roofing materials were re-used. This is evident in the construction of the grain dryer 1758 (Chapter 8). Burnt limestone roof tiles were recorded in the final fill of oven 207 in the East Range and may have been part of a collapsed superstructure (Chapter 6). The use of stone indicates access to sources and transport, availability of craftsmen, knowledge of building techniques and the 157

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) (see Fig. 6.33) of the Aisled Hall. A total of 1231kg was recorded, mostly deriving from this collapsed roof. Their arrangement in patches among Purbeck stone tiles suggests they were used as an ‘exotic’ decoration on the roof. They are a pale grey slate of Devonian age, with a likely origin in either South Devon or West Somerset (Fig. 9.30). Discoveries of true roofing slate in a Roman context in Dorset have been rare, so those from the Druce Farm villa are of particular interest. Metamorphic slate does not always appear in the archaeological record, and excavation reports of Roman sites using the terms ‘roofing slate’ or ‘stone slate’ can be confusing when it is unclear if these are true slates. ‘Stone roof tile’ is the best term for those sedimentary limestones or sandstones which can be split to serve a similar purpose. In Dorset these include limestones from the Purbeck series, the Lias and even Kimmeridge shale. Precedents in Devon The use of roofing slate in late 2nd to 4th century Roman Exeter is well established from excavations in the city. At least eight different types from sources in South Devon have been identified, among which the Kate Brook slate is the most common (Holbrook and Bidwell 1991, 2824). Topsham, considered the port of entry for the city, has also yielded good evidence for roofing slate, most recently from the collapsed roof of a probable mansio. The best examples were virtually undamaged, of the typical Roman shape, and often still retaining the original nail head. These too were mostly from the Kate Brook Formation, with two other sources in the South Hams (Farnell and Rainbird 2017, 47). Further east along the coast, ‘Devon slate’ was recovered from the Honeyditches villa or possible mansio at Seaton near Axmouth (Miles 1977, 140). All

Fig. 9.29. Leaf-shaped sandstone tile with no fixing hole, indicating that it was secured to the roof by mortar, from accumulation south of Room N3

necessary finance to achieve results. Then, as now, roofing these buildings would have been a heavy financial outlay. Roofing slates Peter Stanier Background Broken fragments of roofing slate (i.e. true metamorphic slate) were found within the collapsed Phase 5 roof

Fig. 9.30. Locations of possible slate quarry sources and sites where roofing slate was used in Roman buildings. The River Parret is highlighted as a possible transport route

158

Building Materials this supports the existence of a viable local coastal trade in slate from South Devon (Holbrook, 2001, 153).

in the bedrock has been found at Ipplepen near Totnes (Farnell 2010, 85, grid ref: SX 84296653). These sites were in easy reach of the navigable heads of the Dart and Teign estuaries, from which their slate could be shipped coastwise more economically than by using difficult roads through hilly country. Another possible Devon source is the softer slate from the Meadfoot Group (Lower Devonian) near Kingsbridge, where there was direct access to shipping from the coast at Beesands (SX 822415) and the shores of the sheltered Frogmore Creek on the Kingsbridge estuary, where Geese Quarry (SX 755415) at Charleton was recorded at work at least in medieval times (Born 1988). This must remain speculation while there is no convincing archaeological evidence of a Roman presence in this area, although it may be worth noting that there is a rectilinear enclosure ‘of uncertain date’ near the latter quarry site (Devon HER, MDV 66945).

Precedents in Somerset Some slate roofing materials are known in RomanoBritish contexts beyond Devon. Unsurprisingly, a Roman site overlying a Morte slate bedrock in the Quantocks had a roof ‘using a combination of slate and ceramic tile’ (Wilkinson et al. 2003, 192). ‘Devon’ slate has been recorded in Somerset at Churchie Bushes, Kingsdon, Low Ham, High Ham, and Tatworth, and perhaps elsewhere (Williams 1971, 105-9). Roman roof tiles of ‘Devon’ slate were found during fieldwork at Churchie Bushes (Dewar et al. 1956, 94-5), ‘roofing slates’ were discovered at Low Ham in the 1940s and again in 2018 (Radford 1946, 26; 1947, 6 and 1948, 141; pers comm. David Roberts, Historic England, 27 November 2018), and ‘Devon slate probably from the Quantocks’ was noted at High Ham and Kingsdon (Radford 1951, 49-51). These four were certainly within access of potential slate sources in the south Quantock Hills via the River Parrett and overland trackways. ‘Devonian’ roofing slates were also recorded at Tatworth (Taylor 1966, 281). Interestingly, this site is in the Axe valley near Chard, further inland than Honeyditches, but with potential access to coastal traffic from Devon.

In Somerset, the Devonian Morte Beds have been a source of roofing slates in the south Quantock Hills. These are likely to have supplied the slates found at some Somerset villas, although no actual quarry site has been identified. More is known of slate quarrying in medieval times, particularly at Rooks Castle, Broomfield (ST 253323). Two slaters were listed in Tax Rolls for 1327: John le Helier at ‘Bromfelde’ and Hugone le Helier (Hugo the slater) at ‘Castrum’, presumably Rooks Castle (Dickinson 1889, 162). Slate tiles from here were used at Bridgwater church in 1393-4 and the castle (Dilks 1938, 231; 1940, 103). Excavated roofing slate fragments at Taunton Castle, were also thought to have come from Rooks Castle (Radford and Hallam, 1955, 80). The earthworks of quarry pits survive in the wooded valley side at Rooks Castle (Riley 2004, 7). While it is uncertain that Rooks Castle supplied the Romans, it does show that slate of a reasonable quality did exist and was available at least for medieval builders.

Precedents in Dorset It has been acknowledged that ‘Devon slate has a limited use for roofing from the Roman period in Dorset’, fragments occurring in and around Dorchester at Walls Field, Charminster, and Poundbury (Green 1997, 101). The former site was the trial excavation of a possible villa, which identified ‘Devon-Somerset’ slates among fallen roofing materials which also included stone and pottery tiles (Dewar 1960, 86). Some of the slates were comparable in quality and size to examples from the Low Ham villa in Somerset, and it was considered they were ‘probably derived from the Quantock Hills’ (Dewar 1961, 8-9). There are some references to roofing ‘slate’ in the excavation reports for Poundbury (Green 1987; Farwell and Molleson 1993). The late Roman settlement at Poundbury yielded roof tiles of Purbeck limestone and ‘fragments of Devonian or Cornish slate.’ (Anderson et al. 1987, Microfiche 2, E5, 2). Debris within Mausoleum R3 of the extensive late Roman cemetery included a fine grey soil containing ‘small fragments of soft grey slate.’ (Farwell and Molleson 1993, Microfiche 1, C12).

Transport Sea transport is proven by large quantities of South Devon slate roofs from excavated buildings in Roman Exeter and the city’s port at Topsham on the Exe estuary, with perhaps more continuing eastwards to Axmouth in the case of the Honeyditches site. A voyage from the head of the Teign estuary near present-day Newton Abbot was 17 miles (27 km) to Topsham, and 28 miles (45 km) to Axmouth. It is not unreasonable to suggest that trading vessels could have made a more exposed journey of 58 miles (93 km) across Lyme Bay and around Portland Bill to enter the presumed Roman port somewhere on the Backwater at Weymouth (Putnam 2007, 143). From there it was only 11 miles (17.5 km) by road via Dorchester to the Druce Farm villa, although this part of the journey was perhaps as expensive as the sea voyage from the south Devon estuaries.

Slate sources in Devon and Somerset There is little doubt that metamorphic slate was sourced as a Roman roofing material from different exposures of the Kate Brook and other beds in the Buckfastleigh area of Devon. The rougher examples may have been collected from outcrops rather than quarries (Farnell and Rainbird 2017, 47). Dateable quarry sites are always difficult to identify, but a possible small-scale roofing slate quarry

Beyond Devon, the Devonian slates would have been in competition with the more locally available stone tiles which are commonly found at Roman buildings in Dorset and Somerset, such as at Halstock villa where only limestone and sandstone roofing tiles were recorded (Lucas 159

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) 1993, 107). Proof that overland transport of heavy stone tiles was practicable is evidenced, for example, by finds of Purbeck and Ham stone tiles in and around Dorchester, a distance of 22 miles (35 km) from Corfe, and 23 miles (37 km) from Ham Hill. This sort of distance must have been the maximum achievable in a long day. Packhorses were a likely means of transport through countryside with an unsophisticated local road network. Indeed, packhorse trains were still a familiar sight in the early 19th century in parts of Cornwall, where roads for wheeled traffic were poor, carrying copper ore in the Redruth district, sea sand on the north coast, and slates from the St Neot quarries in the east. Most of these covered relatively short distances to or from shipping places.

slates, identified from the Kate Brook formation (Upper Devonian age) around Buckfastleigh in south Devon. During fieldwork, slate samples were collected from sites in the Lower Devonian Meadfoot Group at Beesands Quarry and Geese Quarry at Charleton near Kingsbridge. Some tentative comparisons were made, especially the slate from Beesands, but as stressed above, any possibility of Roman quarrying here is speculation. Fieldwork also collected Upper Devonian Morte slate samples from quarry sites at Rooks Castle and King’s Cliff near North Petherton in the south Quantocks, the nearest potential source for Druce Farm. Some of the grey slate gathered here could match the Druce Farm slates, although many of the samples are poorly cleaved.

Slate sources were considerably more distant, and in the case of Druce Farm the nearest potential Quantock source was approximately 45 miles (72 km), either overland via Crewkerne or a combination of land and the Parrett and Yeo rivers via Ilchester. This sort of overland distance starts to make a sea voyage from South Devon a little more attractive. The greater expense of transport from either source may have restricted the rarer slate to decorative purposes.

Conclusion The Roman use of Devonian slate for roofing was not as rare as perhaps originally thought in Dorset or elsewhere in the region. Although it has not been possible to pinpoint an exact source for the Druce Farm slate, there are good cases for quarrying in either South Devon or the Quantocks of Somerset. It is possible that the presence of scattered slate fragments on other sites may have gone unrecognised as a roofing material, especially where a site has suffered from intense ploughing. The significance, therefore, of the Druce Farm villa excavations is in the survival and recording of a large part of a collapsed roof containing Devonian slates in context.

Identification No complete roof slate survived at Druce Farm. The largest broken piece examined was irregularly shaped 165mm x 160mm x 4mm thick. The small fragments have a thin slatey cleavage, well-defined and of good quality, suggesting these delicate slates were deliberately selected for their light weight, which was seen as both ‘exotic’ and easier to transport longer distances.

Stone building material and architectural fragments Peter S. Bellamy Introduction

The Druce Farm villa slates are a rather soft pale grey slate of Devonian age, and two possible quarry sources were investigated: South Devon and the south Quantock Hills of Somerset, the nearest sources by overland or sea transport. Small slate samples from the villa site were considered to be a close visual match when compared with roofing slates found in excavations in Roman Exeter and now stored by the Royal Albert Memorial Museum. These are grey-green

This report includes all stone building materials and architectural fragments retained from the excavation, with the exception of the stone roof tile (see Bithell above). It does not include the structural stone which formed part of the in situ building remains, which are described as part of the main stratigraphic report. The recovered stone building materials are tabulated in Table 9.5 by period.

Table 9.5. Architectural stone objects by phase Period

Column

Jambs

Post Pad

Cover

Misc. frags.

Dressed blocks

Tesserae

1 2 3.1

1

3.2

1

4

1

4.1

1

4.2

1

5

2

6

1

1 1

2

1

1

Unphased Total

6 1

4

2

1

1

160

2

9

2

Building Materials Raw material

probably from dwarf columns: one small fragment (SF438, context 841) preserved part of a fillet above a torus and the other was a fragment of column shaft with the edge of a fillet preserved (SF340, context 322). The column capital and one of the column base fragments were recovered from the North Range. The other two fragments were from the East Range. All the column pieces were in secondary contexts dated to Periods 4.2 and 5. It is unclear from which building they derive, but the column capital was found in Room N1 and may have been part of a biforium window.

Petrological identification was undertaken by visual examination using a hand lens. The stone building raw materials derive mainly from the Purbeck and Portland beds, together with one piece of Ham Hill stone and one of greensand. Column fragments Four column pieces were recovered; all were of Purbeck/ Portland limestone. The most complete fragment was a Tuscan dwarf column capital with two upright cavetto mouldings and a series of fillets, with a column shaft diameter of 195mm (Fig. 9. 31). This is of Blagg type IVA (Blagg 2002, 129-132) and is similar in type, but more elaborate than, the dwarf column capitals in the Dorchester Roman Town House (Durham and Fulford 2014, 89). The dating of the Druce capital is uncertain but may be similar to that of the Colliton Park examples (mid-4th century CE). One other column shaft fragment was recovered (SF530, context 1075), also from a dwarf column, and this had a reconstructed diameter of approximately 250mm. The remaining two fragments were of column bases, again

Jambs? Two fragments of chamfered dressed stone may have come from door or window surrounds. One of Purbeck/Portland limestone is decorated with a cable moulding bordered by fillets (Fig. 9.32) and the other, of Ham Hill Stone, is plain (Fig.9.33). Both were found in the collapse over the North Range, and it is assumed that they originally derived from this building. Post pad A large block of limestone with dressed faces and diagonal chisel toolmarks (c. 50mm wide) evident on all edges and the top surface (base not viewed) (Fig. 9.34). The upper face is worn with a square recess worked flat with pecking rather than chiselling. The recess measures 312mm by 296mm across and is 8mm deep at one end shallowing to 3mm at other end. The shallower end is less flat and finished. This block is likely to have been used as a post pad to support a large square post. It was found in the Phase 4 Aisled Hall (SF1008, context 158, Period 4.1), set on top of flint-packed pit 1113 at the north end of the west aisle. It appears to be in its original position. The slightly irregular finish of the recess and its shallowing at one end may indicate that it was finished on site to fit an existing post, perhaps suggesting it was a secondary addition to counteract the problems with possible slumping of the original flint post pad 1113.

Fig. 9.31. Column capital (SF 93) of Portland/Purbeck limestone from collapse over Room N1

Fig. 9.32. Architectural fragment (SF 409) of dressed Portland/Purbeck limestone decorated with cable moulding from collapse in corridor N11

161

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 9.33. Architrave fragment (SF 104) of Ham Hill stone from collapse over Room N1

Miscellaneous architectural fragments

Cover?

Two small stone pieces may be part of unidentified architectural fragments. One was a small broken piece of oolitic limestone (probably Portland Whit Bed) that had two near parallel, slightly curving incised lines on one dressed surface and some mortar adhering on other surfaces. This piece was recovered from just south of the North Range (SF1009, context 904, Period 5). The other piece is a fragment of Portland Freestone with one curved dressed surface and one flat dressed surface at an oblique angle; the remainder is broken. This was found within Pit 636 in the East Range Aisled Hall (SF979, context 1165, Period 4.2).

One large broken slab of Portland/Purbeck limestone, with a worn upper surface and roughly chipped lower surface and a slightly irregular worn curving tooled edge, was recovered from the wall in the south-east corner of the Phase 3a Aisled Hall (context 539; Period 4.1, Fig. 6.17). This slab was 59mm thick and appears to have originally been circular with a diameter of about 850-900mm. It may have originally served as a cover for a well or pit but was recovered from a clearly secondary context. Dressed blocks Nine dressed limestone blocks were recovered from the site. These were all of Portland/Purbeck limestone. The blocks all had one or more worked faces, but only two blocks had evidence of toolmarks in the form of diagonal chisel marks. Three blocks retained traces of lime mortar. All the blocks came from the North Range. One dressed block of greensand may have originally been part of a quern or millstone, though no original quern surface could be identified, so has been classified as building stone. Also, at least three identifiable quern or millstone fragments were reused as building material, evidenced by the presence of mortar on at least one face. One was recovered from pit 206 in the Aisled Hall (SF1006, context 227, Period 5), one from wall 690 in the West Range (SF510, context 761, Period 4.1), and one from ditch 1459 (SF793, context 1435, Period 4.2). Further fragments of quern or millstone may have also been reused as building material, but precisely which pieces cannot be determined with any certainty. Tesserae Two irregular cubes of mudstone were recovered from the site. They measured 39 x 39 x 33mm and 38 x 35 x 35mm. One piece was found in the North Range (context 931, Period 4) and the other in Pit 752 in the East Range (SF752, context 533, Period 6). Mudstone of this type, possibly from the Purbeck Beds, was used for the

Fig. 9.34. Post pad 158 (SF 1008) of Portland/Purbeck limestone from the Aisled Hall

162

Building Materials Note on the column capital from context 197 †Anthony Beeson

production of tesserae and inlays (Sunter 1987, 35-67). However, when dry, this material tends to split cubically and no trace of working was evident on either piece, so it is unclear whether either was intended to be used as part of a tessellated pavement or inlay.

A substantial provincial Tuscan-style lathe-turned capital (Fig. 9.31) made from Portland/Purbeck limestone (context 197, SF93) was discovered in the north-west corner of Room N1 (Fig. 9.35). It has a deep, square abacus complete with a central pivot slot for fixing on a lathe. The capital itself is extremely ornate in the number of turned mouldings, and an exact match has not been traced. Unfortunately, the recent publication of the Colliton Park excavations at Dorchester all but omits the important series of columns found on site although, to judge by a photograph issued at the time of the original excavations, only one of those seemingly matches this, in its substance but not ornamentation (Durham and Fulford 2014, 47, fig 1; 88-9, fig 93). From the underside of the abacus to the moulding above the start of the necking, the capital measures approximately 280mm in depth.

Illustrated architectural fragments and building stone Fig. 9.31 Column capital of Blagg Type IVA (Blagg 2002, 129-132) with two upright cavetto mouldings. Almost square abacus, 320mm by 295mm with a central square fixing hole 30mm across and 49mm deep. PortlandPurbeck limestone. Shaft diameter 195mm. SF93; context 197; Period 5. Fig. 9.32 Chamfered dressed architectural fragment decorated with a cable moulding with a narrow fillet on either side. Portland-Purbeck limestone. SF409, Context 812; Period 5. Fig. 9.33 Chamfered dressed architectural fragment. Ham Hill stone. SF104; context 197; Period 5.

Below the abacus the area of the echinus is unusually deep and is occupied by three stepped fillets, the upper being the boldest. Below this are two, steep-sided, upright mouldings with the curve of the cavetto only forming at the top of each. The upper one again is the boldest and

Fig. 9.34 Post pad. Portland/Purbeck limestone. Length 565mm, width 508mm, thickness 275mm. SF1008; context 158; Period 4.1.

Fig. 9.35. Collapse debris in Room N1 as found: column (SF 93), architrave (SF 104) and fallen limestone roof tiles in situ in decomposing wall mortar by the west wall 149, view north-east

163

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Conclusion

deepest and is footed by a narrow but prominent fillet. These mouldings may just about fit the capital into Blagg’s type IV (Blagg 2002, 129-131, fig 34, plate xcviii). The lower cavetto moulding is footed by a recessed fillet, beyond which are the remains of the necking of the column. No doubt this would have been terminated by an astragal or neck ring.

Large limestone blocks and occasional heathstone was used both in the buildings and in other features. It is interesting to note that, apart from quoins on the North Range, limestone was not used for walling on the villa buildings. However, it appeared to be the material of choice for use in the construction of the ovens and dryer. The limestone was not only durable and strong but could also have been chosen for its heat-absorbing qualities.

Unfortunately rules of proportion do not apply to provincial columns, so one is unable to calculate the original height of this column. If it were a dwarf column then it may well have been used to support an abacus or side table. However, its broken condition might suggest a fall from some height and it therefore could have been part of a biforium widow, or one of several forming an arcade on the western wall of the room. Other occurrences of building stone Dressed limestones and occasional heathstones were recorded in the villa buildings and other features. Tooled limestone blocks were used as quoins on the North Range to strengthen the corners of the building. They survive on the extreme north-west side, and it is probable that the other corners of the building may have been similar but have not survived. This building technique was apparent in the earliest phase, a strip building dating to Period 2.2, when very large Purbeck limestone blocks were used for the west corners (Figs 5.6 and 5.7). Such blocks were also used on the Period 4 west extension of Room N1 and the north extension of Room N3. A number of features which were constructed within the buildings used limestone as a major component. These are described in Chapters 5, 6 and 7 relating to the North, East and West Ranges. In the North Range, this included (burnt) limestone bases of a Period 3.2 oven 992in Room N5 (Fig. 5.16) and a Period 5 pit 590 (Fig. 5.47/1) in Room N9. At the north-west end of the Period 4.2 Aisled Hall, substantial remains of oven 207 survived. Limestone blocks were a major element of its construction (Fig. 6.27). Four ovens, all dating to Period 4.1, were located in three of the four rooms in the nondomestic West Range. In Room W1, the largest recorded oven (706, Fig. 7.6) was built from large blocks of limestone and occasional heathstone; the stonework was heavily burnt. In Room W2, the fragmentary remains of oven 1118 (Fig. 7.9/2) were recorded. Burnt heathstone and charcoal pointed to its function as an oven of some sort. Two conjoined ovens (691) and (692) were located in Room W3 (Figs 7.10-13), There was no stone associated with the smaller keyhole shaped oven 691, although the circular, vertically-sided oven 692 was constructed with limestone blocks. Located away from the villa in Eleven Acres, the construction date of the very well-built grain dryer (1758) is unknown, but it had gone out of use by the mid-4th century. Much of the material used in its build comprised limestone blocks (see Chapter 8 and Figs 8.1, 8.4-5). 164

10 Internal decoration Mosaics †Anthony Beeson

as is obvious from some that still retain their polished surface. The in situ mosaics discussed below are numbered 1-6; this is followed by the archaeological context number which was allocated during excavation. All in situ mosaics were located in the North Range; room numbers are illustrated on Figure 5.2. Loose tesserae from rooms with no in situ pavements are defined by their context numbers.

Introduction Providing a building with mosaics needed to be part of a general building programme or refurbishment, and it is likely that mosaicists worked in conjunction with other specialists in a builder-decorator’s firm. Although mosaics would be the first element of the scheme to be installed, they had to dry out properly, and then the plasterers and painters would attend to the walls and ceilings and add a quarter round moulding around the room as a skirting. Only when the mess of the plastering and painting had been removed could the mosaic be washed and the surface ground to a polish with sharp sand, water and a grinding stone. A quartzite polishing stone was recovered from Littlecote villa (Johnson 1982, 9). This burnishing process both enhanced the colour of the tesserae and imparted a gloss to the floor, whilst levelling the mosaic and removing any proud grouting. The stones used for tesserae at Druce Villa would have been highly polished, resembling marble,

Room N1: Mosaic 1 (Context 153) The adjoining and interconnecting rooms (N1 and N2), probably a private domestic suite for the owner or important guests, were constructed during Period 4.2 building work and formed the western wing of a remodelled range. The mosaic in this room was protected by the lower courses of its surrounding walls and the roof collapse and survives remarkably intact (Fig. 10.1). The phasing regarding N1 and N2 is complex. An apsidal room (N12), perhaps an audience chamber, dated c. 120200 CE, was replaced in the 3rd century by a range probably consisting of three rooms, one or two of which were below

Fig 10.1. Room N1 mosaic 1 (153) in foreground and Room N2 mosaic 2 (246) behind, view south

165

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

The coarse border

laid in a hard cream-coloured mortar of unknown depth. The large white tesserae of the border are made from indurated grey chalk from the bonding layers between the Chalk and Greensand beds. The border also contains a scattered medley of local stone including Purbeck limestone, sandstone and the same limestone used for roofing tiles. Furniture in antiquity, and until recent centuries, was always placed around the walls of a room and, thus, the tesserae of mosaic borders were generally made of larger stones to cope with this. As would be expected, the offwhite tesserae employed in the mosaic’s main design are not only smaller than those of the coarse borders but, being made of fine indurated chalk, appear whiter. The contrast thus separates the mosaic proper from its surrounding tessellation. Observations made by this author during the excavation at Boxford’s Mud Hole Villa has suggested that there, the coarse border was laid first to form a working platform and frame into which the great figured pavement was fitted. Whether this was the case at Druce Villa or was a peculiarity of the Boxford mosaicist is unclear, although the treatment of the floor in Room N6, Mosaic 3 (357), might suggest that it was.

Room N1 measures 5.25m east-west by 4.5m north-south. Its mosaic was executed in off-white and blue-grey tesserae

The north and western borders are wider than the other two, suggesting they were designed to accommodate large

N1 and N2, and another (N5) to the south. A later phase saw the building of a new west wall and the re-modelling of Rooms N1, N2 and N5. The development of Room N1 is problematic as, owing to the survival of its mosaic, there was only a 1m section on its eastern side where it was possible to investigate the underlying levels. The clearing of the vertical edges of this sondage under the south-east corner of N1 revealed three layers (Fig. 5.14/1). Four white stone tesserae were recorded in the upper layer 983, and probably originated from Mosaic 1 (153). Underneath this, layer 987 produced painted plaster and two sherds  from a Black Burnished ware 1 jar dated c. 160-200 CE. Beneath this was a chalk layer which seems sensible to interpret as the remains of a floor. The latter apparently had a mosaic, as excavation within N2 also disclosed wall plaster and loose tesserae belonging to it. Unfortunately there were no finds in the lowest fill that equated with the apsidal room N12. Structural remains on the west side of N3, and also located in front of the section made in N1, may be those of the east wall of the apsidal room.

Fig. 10.2. Room N1 mosaic 1 (153) (painting Luigi Thompson)

166

Internal decoration pieces of furniture. As there is no apparent damage to the north border, it is suggested that any furniture placed there must have been large and immovable. Luigi Thompson’s careful scrutiny of the pavement’s individual tesserae for his recording has noted several irregularities in the choice of stones used for the coarse borders (Beeson et al. 2015-16, 79). In many cases, when the stone was freshly cut, the contrasts in colour would not have been as noticeable as they are today, or when recorded in Thompson’s painting (Fig. 10.2). It is a common visual trick, employed by mosaicists to enliven what would otherwise be visually large bland areas of a single colour, to drop the occasional speckle of a different shade into the ensemble. Interestingly, only the north-eastern section of Mosaic 1’s coarse border contains a substantial scattering of red ceramic tesserae, which begs the question whether they were used on purpose for this particular section of the border. Were they used to designate the placing of a window, door or piece of furniture, such as an abacus or stone side table, or were they perhaps the result of a lack of chalk tesserae, or an earlier superior repair to the border that is now not apparent? The terracotta tesserae on this section of the border include ribbed pieces from reused flue-tiles. The southern section of the coarse border again has an interesting collection of tesserae. There, apart from the hard chalk, the mosaic contains much dark and salmon-brown stone, lias and even some fine white chalk. To a lesser extent, this continues to the eastern border. Were these stones part of the original border and used for artistic effect, through an insufficiency of the grey chalk tesserae, or the result of a superior restoration using other materials, all of which are to be found on the neighbouring pavement in Room N2? It is likely that a doorway opened into N2 in the south-eastern corner of N1, and a mosaic panel of small fine white chalk tesserae either paved an entrance step or abutted a threshold slab. A three-tesserae wide band of dark brown ran southward towards Room N2 and crossed the white panel. A parallel band may have joined it, and together they would have framed the threshold panel.

illustrated by the 2nd century design on the 4th century mosaic at Yarford, Somerset (Cosh and Neal 2010, 406-7) that, although in itself polychromatic, would be more at home in the early houses of Verulamium or Camulodunum. On its off-white ground, Room N1’s fine central rectangular panel is bounded by a three-tesserae-wide fillet border of blue-grey lias tesserae. Within this, a wide blue-grey band frames the entire mosaic and is enlivened with a row of poised white squares. Another such band, independent of the first, spans the north-south width of the pavement on its western side, dividing the central field into two panels of unequal size. The main area of decoration occupies perhaps two thirds of the pavement, and consists of a square holding a swastika, the symbol of good fortune. This expanding swastika design, with multiple returns, resembles a labyrinth to the casual viewer. The panel would have been the visual and artistic focal point of the floor, to be seen both from the room’s ‘place of honour’ on the western side and also through its entrance from the southern vestibule. The remaining third of the mosaic on the western side, within the grey and white poised squares framing borders, is occupied by a long rectangular panel. Here a three-tesserae-wide, bluegrey fillet border surrounds a design of two courses of ashlar masonry, formed from two-tesserae-wide blue-grey lines. Such decorative panels, that are separate from the main design of the mosaic, generally denote the position of social importance, from whence the occupant viewed it and any visitors. This place of honour was generally the position of an important piece of furniture within a chamber, and often indicates where the chief occupant would have reclined or sat. The use of the ashlar pattern on Romano-British mosaics is unusual, and particularly so in the South-West. It did, however, occur together with poised squares (used as a decoration for square panels) and swastika-meanders on a lost mosaic from High Ham in Somerset (Beeson 2015-16, 80, fig. 2; Cosh, 2010, 30, fig. 4). There, two end ashlar panels of three courses framed the main square central panel that held four swastikas and five squares. These, positioned in the centre and corners of the design, bore poised squares. A similar panel to that at High Ham was found in 2009 at Butleigh, Somerset (Cosh 2010, 28-31; Cosh and Neal 2010, 409). That fragment however, lacked any ashlar work. Fullerton Villa in Hampshire had an early to mid-4th century mosaic with a bold ashlar pattern in a purplish-brown stone on white decorating one of its transverse corridors (Neal and Cosh 2009, 172-173, figs 119 and 121). Ashlar pattern, sometimes reversed as white on a dark ground, may be found elsewhere in Britain, but generally on 2nd century mosaics such as at Eastgate Street, Gloucester (Cosh and Neal 2010, 149), suggesting that the pattern’s use was a revival of an earlier fashion. It features in the very ‘first century style’ black and white forum mosaics in Cirencester (Beeson, 2015-6, 80, fig. 3; Cosh and Neal 2010, 74), dated by the excavator to the mid-4th century and remarked upon as representing ‘an interesting continuity of design’. Both High Ham and

The central panel Within its coarse border, the cool and elegant ‘black and white’ floor of Room N1 is one of a series of 4th century West Country mosaics that hark back to designs of earlier centuries. The Italian black and white style, so popular in the Mediterranean provinces and Gaul from Republic times, and admirably illustrated by the mosaics from Fishbourne Palace, seemingly went out of fashion in Britain in the 2nd century CE in favour of polychromatic designs. Only imperially sanctioned commissions, such as the exercise basilica mosaics of the Wroxeter Baths, seem to have employed black and white thereafter. Mosaic 1 is an example of what would today be termed ‘retro design’. Fashion, and the collecting of antique works of art (and their reproduction) led to several revivals in Roman sculpture, architecture and painting, so it would not be surprising if this trend is occasionally reflected in mosaics whose designs appear to reference an earlier era. This is 167

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Butleigh (Cosh 2010, 29-30), are believed to be the work of mosaicists based at Ilchester (Lindinis). The similarity of design motifs to those at Druce suggest that the same mosaicists also worked there on Mosaic 1 (153), Mosaic 2 (246) and Mosaic 5 (474). The cool colours of N1’s mosaic, so reminiscent of 1st century pavements, seem to be a feature of some of those accredited to the Lindinis group. Mosaic dating, however, is still highly speculative. The room overlies earlier walls of two periods suggesting that, notwithstanding its design, the mosaic is a late one. Steve Cosh (Cosh 2013) suggested that it was laid after 350 CE as one of the Butleigh mosaics sealed a coin of Magnentius (350-353). If so, the series of repairs to the mosaic suggest that the room was long used for domestic purposes, and that Romanitas continued at Druce far into the 5th century at least.

south-west corner. A smattering of terracotta and reused materials were incorporated. The size and shape difference of these tesserae give them the illusionary appearance of standing slightly proud of the rest of the floor. A roughly crescent-shaped area defined by these large tesserae was filled with smaller hard white chalk, grey chalk, and a smattering of terracotta tessellation and other stones, may also date from this period. Finally, it seems that when all else failed, stone roof tiles were inserted into lacunae to provide a firm surface. These rather ham-fisted repairs hint that, as at most sites, a functional mosaic was more important to the occupiers than artistic finesse. In this room, a couch complete with covers would have masked the floor beneath. Other causes of repairs could be subsidence into earlier features or the failure of mortar. There is a slight undulation in the centre of the pavement and in the area of one of the repairs. As could be seen in the sondage on the east of the room, the make-up material to heighten the floor was not solid, which could have affected the mosaic surface.

The area of the ashlar mosaic panel, and the tessellated border surrounding it, presumably housed a couch in antiquity. Whereas cupboards, side-tables and chests placed on coarse borders were of little threat to a mosaic’s stability, the constant moving pressure exerted on the tessellation by a couch, especially if it had legs terminating in points, could eventually cause considerable damage, necessitating repair. Damage could also occur when such furniture was moved. Subsidence into recently filled areas or earlier pits was a common problem, especially for late mosaics that often had little in the way of deep and solid foundations to rely on. Differences in the tesserae employed in the repairs, and the manner and direction in which the tessellation was laid in the damaged areas, suggest that restoration occurred on at least three occasions, terminating in a rough patching of roof tiles. This further suggests the longevity of use in this chamber. There is similar evidence for late domestic occupation from nearby Dewlish villa (Cosh and Neal 2005, 76). It would seem that the later jobbing mosaicist called in to effect the repair totally disregarded the complications of the existing patterns and used available material he had at hand (see Fig. 10.1).

Although room usage is notoriously difficult to define in ancient domestic buildings, it seems most likely that Room N1was either a bedchamber, a private dining room or a reception room. An area of burning is evident on the swastika returns on the south-east side; this may have been where a brazier warmed the chamber in antiquity. Wall plaster from the room shows that, in contrast with the cool palette employed in the floor, the walls were richly decorated in a panelled scheme that employed an ochre yellow, Pompeian red, purple, slate-blue and white (see Morgan below). Room N2: Mosaic context 1059 (below Mosaic 2) Three sections were dug into the underlying floor of N2, and a presumed fireplace (Fig. 5.15) or brazier stand belonging to the second building phase (following the demolition of the apsidal room) was found at a depth of around 0.1m-0.15m below the later Mosaic 2 (246). Associated loose tesserae and painted plaster discovered below this mosaic probably belong to the earlier phase. Tesserae (numbered 1059) included dark brown examples (only 10mm square), as well as slightly larger white and grey stones. Large coarse examples in terracotta and a light grey stone were recorded, and a grey slab approximately 4cm square that may be an opus sectile element, suggest that an earlier floor in the underlying room had included at least a panel of such decoration. In colour, shape and size the piece equates to the detached elements that Bryn Walters suggested came from a lost opus sectile floor at Halstock villa, some 13 miles from Dorchester. David Tomalin also recorded similar pieces at Banwell villa in Somerset (Walters 2011, 315-318; 2012, 12-13). Opus sectile pavements were luxury items that enjoyed a vogue in the 1st and 2nd centuries, and a revival in the 4th. The loose tesserae belonging to 1059 were above a solid layer of ‘hoggin’ (natural clay/loam/small gravels) which was used for levelling and consolidation.

Repairs The first repairs possibly neatly employed the original white and grey tesserae in the ashlar panel, and followed the existing lines of tessellation, but not the pattern. In the area to the west of the ashlar pattern, hard white chalk, terracotta and dark brown stone (all also used in the floor of Room N2), joined re-used original tesserae. Either at the same time or later, a neat area of coarse tessellation incorporating ceramic tesserae spread from these first repairs, over and obliterating the pattern of the western poised-square border. These tesserae were larger than those of the original coarse tessellation. Attempts to follow the size and lines of tessellation weakened as the repair spread westwards. It is assumed that further repairs were undertaken, as two large darker grey areas of lias tesserae (much larger than those of the original borders or the repairs, and often oblong in shape), were used on the ashlar panel and poised square border, and also in the 168

Internal decoration Room N2: Mosaic 2 (Context 246)

of the chamber. More square or rectangular panels were probably spaced along the long sides of the room (Ladle and Bithell 2013, 4, fig. 3; 2015-16, 74, fig. 12; Beeson, 2015-16, fig. 6). There was a deep red border of tessellation on the short western side, and possibly also on the eastern side. It seems highly likely that the rather strange western area of red tessellation may have held a couch, or a large and important piece of furniture, and was not intended to play a part in the main decoration of the floor, being mostly hidden apart from the westernmost corners that were enlivened with white squares. Bed chambers often had a separate section, beyond the main decoration of the floor, on which the bed stood, and this room tallies well with those that survive in other areas of the Empire. One is suspected in Room 1 at Newton St Loe villa, Somerset, where a wide area of plain white tessellation, at the end of chamber and beyond the mosaic proper, is decorated with two wide parallel lines of red tesserae (Cosh and Neal 2005, 273-274, figs 271 and 272). If this were the case in Room N2, then the main design of the mosaic would start from the eastern edge of the large area of red tessellation and stretch eastwards for an unknown distance. A modern analogy might be of a rug laid on parquet flooring. It is notable that, unlike any other room in the villa, there are no areas of coarse tessellation along the sides of the chamber for furniture, and so the focus would be entirely on whatever occupied the tessellated area. This account will therefore treat the two areas of the central mosaic and the tessellated area as being separate.

The rectangular Room N2 measured 5.5m by 2.6m. It was joined to Room N1 by a doorway at the north east corner. For practical considerations, it is assumed that Mosaic 2 (246) was laid at the same time as Mosaic 1 and by the same mosaicists, although some large commissions are believed to have employed more than one firm of mosaicists on a site. The badly plough-damaged Mosaic 2 (246) stretched across and overlay earlier walls (Figs 5.13 and 10.1). It was laid on a bed of yellow mortar about 10mm thick, which was disintegrating and may have been problematic in antiquity. Unlike the mosaics of the 1st to 3rd centuries, those of the later Roman period were generally not laid on firm foundations but might simply be laid directly over compacted soil. On excavating the Boxford mosaic in 2019, the author was astonished to find that this remarkable art work was laid on a 30mm bed of ginger coloured sand but, apart from ancient damage, had survived remarkably well (Beeson 2019, 76). Presumably because of the villa’s hillside location, Mosaic 2 lay some 100mm below that in Room N1, resulting in a probable stepped entrance near the north-east corner of the room. The surviving fragments of Mosaic 2 (246) in the north-west corner of the room (Fig. 10.3) are difficult to interpret but suggest that the main field of the mosaic was white but with white corner squares repeated on a red tessellated ground at the western end

Fig. 10.3. Room N2 mosaic 2 (246) (painting Luigi Thompson)

169

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) panel, being bounded by the red tessellated area on the western side. North of the surviving portion of this red frame, the remains of a square white border panel abuts the south wall of Room N1. Framed on the east by a fine blue lias tesserae and to the west by red tessellation, this white square is speckled with dark brown tesserae and formed the corner panel of the main mosaic. It complemented the surviving white panel enlivening the area of red tessellation at the mosaic’s north-western corner. The damaged area of blue lias tesserae is sprinkled with the occasional white tessera. It is not possible to determine this blue panel’s original width. Was it simply one of a series of bands separating square and rectangular white border panels, or was it once squared and part of a repetitive blue and white chequered border?

The mosaic Nothing remains of the eastern part of the mosaic or of any tessellated threshold panel opposite the entrance into Room N1. It is thus unknown whether the broad red western area would have been duplicated at the eastern end, or whether the white mosaic panel itself would have stretched that far. As a result, we cannot know if the visitor entering the room was first confronted with white or red tesserae. This author would favour white to tie in with the room’s main mosaic and also that of Room N1, with possibly a simple northsouth linear threshold panel, perhaps of poised squares, or a band of contrasting coloured tesserae. One occasional feature of mosaics of the Lindinis mosaicists is that some floors, such as that from South Street, Dorchester (Cosh and Neal 2005, 115, fig. 101) or Preston (ibid., 165, fig. 159), incorporate a red line to surround the composition. Such a framing line, albeit within the border rather than outside, existed here. The main colours of the mosaic were white, red, blue and very dark blackish-brown, with the main ground again being white as in Room N1.

It is most likely that the central panel was rectangular. A mosaic based around a single square panel would have barely stretched eastwards to reach the centre of the chamber. If, like Room N1’s pavement, it was the work of mosaicists from Lindinis, then several of their attributed mosaics such as those at Spaxton (Cosh and Neal 2005, fig. 292), and High Ham (Cosh and Neal ibid., 289-290, 211, fig. 207) incorporate long rectangular central designs. Spaxton’s incorporates a design of double latchkey meander which, considering the known use of the meander at Druce, might seem very suitable for here. Certainly, the remaining white tesserae in the central panel are very fine and might suggest figured work, but there is no evidence for this due to extensive plough damage.

The centre of the mosaic was occupied by a rectangular or square panel of indeterminate length made with very fine and slightly larger tesserae. A central single rectangular panel, or perhaps three spaced square panels, could have extended the design as far to the east as was considered necessary. The central area of the panel may have been entirely composed of very fine tesserae, as three rows of tiny hard white chalk tesserae remain. Whether these once entirely filled the panel and were part of some lost elaborate central motif, or simply are all that is left of a fine border fillet, is now impossible to determine. This area or fillet was bordered on its outside by three rows of slightly larger dark brown tesserae that were enclosed by a wide frame composed of thirteen rows of medium-sized white tessellation. A fillet composed of two rows of red terracotta tesserae bordered the above and survives on the northern side. It presumably originally framed the entire wide white band all the way around the mosaic’s central

The mosaic’s overall design is so far unmatched. With its corner boxes, and possibly rectangular side panels, it has very faint echoes of the 2nd century nine-panel schemes such as found in one variant on the 4th century version at Yarford, but, unlike those, the usual grid is only visually inferred by change of colour in the border panels and is missing rectangular panels at the west and (presumably) also to the east of the central area. It possibly may be a more elaborate rectangular version of an unusual concentric

Fig. 10.4. Room N7 mosaic 4 (329) (painting Luigi Thompson)

170

Internal decoration squares mosaic from Low Ham villa in Somerset, also attributed to the same Lindinis firm believed responsible for Mosaic 1 in Room N1. Low Ham produced another swastika meander pavement allied to their mosaics at High Ham and Butleigh. The Low Ham ‘squares’ mosaic is believed to have floored a vestibule in that villa, and simply consisted of a series of four grey square frames concentrically placed and spaced, one inside the other on a white ground (Cosh and Neal 2005, 257-8, 260, figs 253 and 256).

wall would have provided a good foundation. Strangely, north of the bust, disjointed white tesserae appear to form the disjointed letters I, A and V. Dating of this mosaic is particularly difficult, but presumably it was also laid at the same time as that in Room N1 and by the same mosaicists. Other mosaics of the North Range (see Fig. 5.2 for room locations) Room N3: Mosaic context 210

Tessellation

The original North Range comprised Rooms N3, N6/8, N7 and N10, with N3 constructed at a lower level forming a semi-sunken room. This was extended north in Period 4.2: an internal flange at skirting level might suggest a wooden floor. However, a secondary floor layer of compacted chalk and mortar may hint at a mosaic foundation. A total of 5,238 small red tesserae (which would have covered approximately one square metre) were found scattered throughout the room. There was a concentration in the north-western corner, a number of which had mortar adhering. Their spread was such (Fig. 5.32), perhaps suggesting a mosaic laid on a light foundation, maybe as in an upper chamber, or their use as a wall decoration, perhaps a dado, especially as some were found in the demolition or collapse layer. Pottery associated with the phase is dated to c. 370-430 CE.

White corner squares visually enlivened the large area of red, white-streaked, tessellated border at the west end of the room, of which perhaps about half survives. A stone roofing tile was neatly used to fill a lacuna in the mosaic and abuts the surviving corner square; that this was a late repair is supported by the fact that no tesserae were found beneath it. Remnants of a decayed quarter round moulding survived against the north wall. The very odd, red and white-streaked area of tessellation south of this may be a repair, and perhaps coeval with the first neat attempt at repair found in Room N1 unless it was intended to enliven a plain surface. Just south of what would have been the centre of this west side, a patch of tiny white, and a few grey tesserae give way to a shaped area crudely infilled with larger white tesserae that (with some imagination) resembles the featureless outline of the head and shoulders of a bust. The existing shape may just be a chance survival as the southern half of this feature is missing but, strangely, an isolated cross-shape of black tesserae marks the position of an eye just where one might expect it, while a red tessera places a nose on the ‘face’. Unfortunately, the area of the mouth (which is often accomplished by two tesserae on mosaics) and the other ‘eye’ are lost. The group of very fine white tesserae appear below the ‘arms’ of the ‘bust’. There are five rows which might be interpreted as fingers. Was this, perhaps, a hamfisted attempt to copy a lost and finer bust situated elsewhere on a threshold panel at the eastern end of the chamber, or in the centre of the mosaic, or was there originally, as on the great floor at Brading (Henig 2013, fig. 13.8), some apotropaic symbol here to guard the occupant? Brading has a swastika below the chair of the Dominus; could this be a very poor Medusa to guard someone as they slept? Perhaps the mosaic structurally failed and we only see the result of the restoration, then perhaps whoever repaired it could only approximate the shape of its original design. Alternatively, some other apotropaic device may have been used, and the repairer ignored the original design and created this odd bust instead with the reused tesserae. What is one to make of this seemingly arbitrary juxtaposition of fine and coarse tesserae in the ‘hand’ feature? A more prosaic interpretation might account for the bust in a repair to the south-western corner and the jumbled reuse of the white tesserae of that panel in with the red, although there appears to be more white present in the surviving area than might be expected from this source. This must have been a stable section of the floor, as the underlying earlier

South of Room N3 A deposit yielded unstratified tesserae of all sizes and colours and is noted in the excavation record as resembling those recorded in Room N10. Room N4: Mosaic context 98 This small enigmatic room, damaged by ploughing, seems to have been a vestibule or short corridor to gain entrance to the suite of rooms in the remodelled North Wing. Although only 1.1m wide, it was decorated with painted wall plaster and its floor of rammed chalk once had a coating of opus signinum and a mosaic or tessellated floor. Many tesserae of all sizes and colours were discovered in the overlaying layers, with terracotta and limestone predominating. Room N5 It might be expected that, as part of the suite of rooms that Room N4 probably gave access to, this room may have had a tessellated floor. Unfortunately, ploughing had caused considerable damage, however its status as a probable service room is consistent with the lack of pavement. A number of large red and limestone tesserae were found in the covering layers and probably originated from Room N2. Room N6: Mosaic 3 (Context 357) Room N6 is 4.60m in width east-west by 3.10m northsouth. Room N8, an antechamber, lay south but its floor 171

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) has been lost. The room is floored with plain, mediumsized red terracotta tessellation, composed of straight bands laid in east-west and north-south directions (Fig. 5.21). It is noteworthy, as the mosaicists appear to have visually enlivened and treated this tessellated floor as though it were a mosaic with a decorative central panel and laid the tesserae so that the central rectangular area was visually demarcated by these bands as a panel of slightly smaller, ‘finer’, but plain, tessellation. Additional visual enlivenment of the otherwise plain tessellated was achieved with a scatter of white tesserae on the central panel. It was suggested during excavation that this might represent constellations. However, although two groups might just be interpreted as Ursa Major and Minor, they may probably be there just to lift the monochrome monotony of the central panel. This does, however, suggest that again the mosaicists laid the coarser borders first before working on the central panel (Beeson 201516, 83, fig. 8.9). The floor has survived well apart from its northern border, where ploughing has removed most of the tessellation and wrenched an island of tesserae from its base.

The mosaic was set into opus signinum on a base of chalk and hard packed flint laid over a levelling layer of hoggin. Almost the entire length of the north side of the mosaic survives, but just under a half of the total as the southern area of the pavement has been ploughed out (Beeson 201516, 82-83, figs 7-10; Selkirk 2017, 31-32). What appears to be the remains of Room N7’s southern wall is, in fact, the mosaic’s flint foundation, exposed by ploughing. The highest part of the surviving mosaic was found to be only 0.2m below the present ground surface. While the tesserae of the borders are large and those of the chequer-work somewhat smaller, the meander employs the smallest. Notwithstanding the large size of tesserae, the mosaic is both attractive and beautifully laid out, with sharp edges to the chequer-work and meander. The meander uses a darker stone than that used for the chequer-work. The surviving border does not employ flue tile tesserae, suggesting a more public function and a purpose to impress. Chequerboard mosaics in red and white are a feature of floors found at nearby Dewlish villa and at sites in Dorchester. The former are far inferior in technique to this, however (Cosh and Neal 2005). The chequerboard design was a favourite throughout Britain, being traditionally based on designs used in opus sectile flooring. Simple meanders with double returns also appear at Dewlish, albeit more elaborately designed and used alone as a central motif in corridors. The same elegant quarter round moulding that is found in Room N6 also survives in Room N7 at the junction of floor and wall. It seems highly likely that mosaics 3 and 4 both date from the late 2nd to 3rd century, Period 3.2 northward extension and refurbishment of the North Wing.

At the western end of the room the tessellation includes many reused box flue fragments with comb patterns. A skirting of very fine and elegant opus signinum quarter round moulding survives on its eastern border and also partially surrounds the footings of what may have been a fireplace or, perhaps more likely, a permanent brazier emplacement that may have been needed in this northfacing chamber. The emplacement lies approximately midway along the eastern wall. An area of burning is apparent on the tessellation close to this spot. Several areas on the central panel show evidence of heat, perhaps caused by a movable brazier to warm diners (if this is interpreted as a winter dining room). This section of the villa (Rooms N6 and N8) conveys a more private and domestic feeling than the grander chambers to the east, and one wonders if Room N7 might have been the private chamber of the Dominus. Painted plaster fragments suggest that N6 was decorated in pale brown and red.

A quantity of wall plaster survived, covering the coarse tesserae of the northern border, where it had fallen from the room’s walls, and affords some evidence for the room’s decorative scheme. Here, in contrast to the rich red floor, the walls were simply decorated in red, white and brown.

Rooms N7 and N10 have yielded mosaics, but the area has been so badly ploughed that only the northern segments survive. Two are almost certainly earlier than the existing mosaics 1 and 2, previously discussed, and their overwhelming colour palette is terracotta.

Room N8 probably served as a vestibule via an arch or a pilaster-framed opening to Room N7, and also gave entrance to Room N6. The floor had been completely destroyed by ploughing (see Fig. 5.17), with only rammed chalk surviving. An attractive but speculative proposal might place the house’s lararium or an impressive piece of furniture, such as an arca (strongbox/chest), here.

Room N7: Mosaic 4 (Context 329)

Room N10

Room N7 measured 6.5m east-west by 5m north-south. It was the central chamber of the North Range and obviously a main reception room. It may have been L-shaped and included Room N8. The pavement is a well-made, neatly laid and a most attractive red and offwhite chequer-board carpet surrounded by a triple row of red tesserae and bordered with a simple, double row, red meander with double returns on a white ground (Figs 5.23 and 10.4)

This room contained two mosaics, a lower one (615) which was replaced by a later floor (474). Mosaic 5 (Context 474) Mosaic 5 in Room N10 was only 0.2m below the surface and has mostly been lost to ploughing, which is most unfortunate as it was probably the most impressive in the villa. At the north-west corner of the room, a border of coarsely shaped off-white and terracotta tesserae surrounds 172

Internal decoration

Fig. 10.5. Room N10 mosaic 5 (474) (painting Luigi Thompson)

a finer panel of white and dark brownish-grey fillets (Fig. 10.5). The tesserae are set into a cream-coloured mortar. Just a corner of the design survives, revealing that the fine central mosaic was bordered by a three tesserae band of white tesserae. Within this is a parallel three-tesseraewide fillet of dark-brown. Another parallel band of brown tesserae joins the western brown border, and yet another appears to its right and forms a corner within the surviving fragment. The most likely solution for the lost design is that the parallel fillet, and that to its right, form the arms of a swastika and that either the mosaic had end panels filled with swastika meanders or, more likely, that it featured an all-over swastika meander that included square boxes within the design, as occurs in a simplified form on the small mosaics from High Ham and Butleigh (Beeson 2015-16, 80, figs 11 and 12: Cosh 2014, 3, fig. 2). Unlike the mosaic of Room N1, the arm of the swastika here joins the outer border. Later pits 614, 644 and 1163, which were dug into the floor of the room, contained mosaic fragments and wall plaster from both decorative periods. Finds from pit 614 include a number of fragments with small white tesserae bordering red, both measuring perhaps just over 10mm square, perhaps suggesting that, as at High Ham, the boxes were filled with similar devices as the poised squares found there, but here in red. Mosaic 5 (474) appears to have coarse borders composed of the same stones as Mosaic 1(153), although here employing

many red tesserae, and the suggestion must be that both were laid at the same time in the mid-4th century. It even features a surviving and similar area of repair on its west side to those affected on Mosaic 1. Some of the tesserae from pit 614 included large (30mm) examples, possibly of Kimmeridge shale, now in a state of lamination. Mosaic 6 (Context 615) The author was fascinated to hear that Mosaic 5 (474) was not the first to floor the chamber. When viewed from the south it was obvious that the several layers of mortar underlying the remaining fragment included the remains of an earlier mosaic founded in opus signinum. The use of this medium, and smaller coarse red tesserae, suggest a date in the late 2nd or early 3rd century for mosaic 6 (615), suggesting that, like the mosaics in Rooms N6 and N7, it dated from the time when the villa was expanded northwards. Small tesserae from the main panel averaging 10mm square in red, white, grey and dark, somewhat brownish, grey were located below the final layer bearing the later mosaic. These are not only found in situ but are mixed into the mortar holding the later mosaic. They include some narrow, white, lozenge- shaped examples. An excavated section (Fig. 5.24) through the floors revealed many white, grey and red tesserae: two small groups around 100mm apart were still in situ. One featured 173

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) the room was redecorated in the 4th century when the replacement mosaic was laid. A photograph of a tray of fragments from the pit (Fig. 10.7) shows neat tessellation in straight lines with double fillets in dark grey, with either double red fillets or perhaps red boxes, presumably on a white ground. A large fragment displayed a row of red tesserae apparently joined to five rows of white; all are small and some of the white are oblong in shape. One fragment measuring approximately 120mm by 100mm features a row of 10mm square white tesserae bordering an equally sized three tesserae wide dark grey fillet, which itself is bordered by a fillet of white three tesserae wide. This fillet differs from the other two in having narrower oblong tesserae, and possibly indicates the edge of a box encompassed by a meander or other shaped frame (Figs 10.8 and 10.9). Unfortunately, the piece disintegrated soon after excavation and so is only now preserved by photograph.

Fig. 10.6. Room N10 schematic floor layers 1) Mosaic 1 (153), 2) yellow mortar, 3) Mosaic 2 (615), 4) opus signinum, 5) flint embedded in chalk, 6) hoggin

two white and two grey joined tesserae, and perhaps was part of a grey fillet angled towards the corner of the room, whilst the other may have been part of a rather square guilloche curl featuring a dark grey eye and a row of red, light grey, white and dark grey tesserae. The mosaic’s underlying mortar retained imprints of tesserae, but no recognisable pattern was discerned by the excavators. Below the opus signinum was a layer of chalk above flints and below that a layer of hoggin. A schematic drawing illustrates the composition of the floor layers in Room N10 (Fig. 10.6) and relates to the constructional methods used in other tessellated floors in this range.

To judge by the surviving fragments, Mosaic 6 (615) appears to have been purely geometric as no potential figural fragments survived. The later mosaic may have replicated its design, although a simple grid scheme possibly featuring octagons with guilloche knot centres, might be a possibility. Its replacement suggests that something fairly drastic happened to the original pavement, resulting in much of it being grubbed up and some of the tesserae reused, either as aggregate or perhaps elsewhere in the villa. It seems likely that the opus signinum binding

The Phase 5 pit 614 dug into the north-east corner of the room cut through both floors and was filled with wall and floor debris from building phases 3 and 4, proving that

Fig. 10.7. Selection of tesserae from mosaic 615 in fill 903 of pit 614

174

Internal decoration

Fig. 10.8. Tesserae from mosaic 615 in fill 903 of pit 614

Fig. 10.9. Joined tesserae in fill 903 of pit 614

175

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) was below the level of the other rooms of the north wing by about 0.3-0.4m, which ensured that the remains of the original surface of the corridor was not contaminated by later destructive ploughing. Some wall plaster remained in situ on the north wall of the corridor, but the tesserae were all loose above the chalk floor and included large and small red terracotta and brown, white and greyish stone.

the mosaic gradually failed; it was found to be in a poor state on excavation. In addition to the visible strata below the later mosaic, gaps in the coarse borders of Mosaic 5 show, in places on the west and north sides, that they covered over both the smaller red border tessellation of the earlier mosaic together with the same elegant quarter round moulding found in the adjacent rooms (Beeson 2015-16, 8, fig. 12). This suggests that only part of the earlier mosaic was destroyed completely, so presumably only the area of the floor that was problematic had to be removed. Where evidence exists, it seems that new floors were generally laid directly above existing pavements in the manner of linoleum in recent centuries. It is extremely rare to discover evidence of grubbing out of an earlier floor, although it was suspected in Room N7 at Fishbourne (the Cupid on a Dolphin mosaic) when it was lifted, as parts were missing and pick marks existed on its mortar base. It would, of course, have been foolish to lay a new mosaic over any part of an old one that had ‘blown’ or come loose from its base. Any area that was unstable would therefore have been removed.

The East Range There were four phases of building encountered in the East Range during its transition from a rectangular hall to an aisled one. No mosaics were found in situ, but it is quite possible that it had areas of tessellation during the aisled hall periods, as much opus signinum and very many tesserae were recorded during excavation. These were mostly large terracotta, limestone or chalk, but small and medium sized tesserae in all the colours found in the North Range were also present, suggesting at least one decorative panel. There was limited evidence for room partitions at the north end of the building; the collapsed wall and roof which were left in situ may cover such divisions. Heavy ploughing had removed any traces further south.

The fact that, at both periods of redecoration, Room N10 had the most impressive and expensive mosaic in the complex, suggests its importance in the social life of the villa either as an official dining or reception room.

The West Range It seems certain that no mosaics were laid in the West Range, which functioned as workshops, but a considerable number of tesserae were discovered there. This raises the possibility that they were being produced from ceramic tiles, either for use at the villa or for sale elsewhere. The deposits were dominated by large and medium sized terracotta tesserae, but a good number of smaller chalk and some other examples were also present. Tesserae made of stone from Dorset are found throughout the south of England, and their production may well have been a cottage industry that might have supplemented an estate’s income, especially in the winter months.

Room N11: Mosaic context 99 The front corridor at Druce was added to the original strip building at the same time as the expansion northward of the North Wing. It is remarkable for its narrowness, being more of a passage at 1.2m wide than the usual size of around 2.7m generally encountered in Romano-British villa architecture. It was provided with a red tessellated floor in opus signinum over a chalk foundation, that featured coarse tessellation (with tesserae around 30mm square from context 812) and at least one area of mosaic with finer tesserae around 10-20mm in size. A detached fragment from this (840) bears the imprint of lost tesserae as well as those that remain in situ (Fig. 5.27). It features two tesserae from a very slightly curving line that is attached to an area of similarly sized red tesserae. The latter are set on an angle to the line, perhaps suggesting that they formed part of a circular pattern. Depending on the original colours of the lost tesserae, whose impressions survive on the edges of the fragment, either the red tesserae carried on to a decreasing angle or a brown curving line formed the apex and edge of the piece and joined with the straight one. This curving line might then have formed a simple hub with the straight line and others, resembling a spoked wheel. Whatever the nature of the device, it seems likely that this finer mosaic formed a threshold panel opposite to the villa’s porch, although a narrow band of decoration, such as a simple fillet down the centre of the floor, may also have existed. The provision of mosaics and painted walls, and the problem of frost damage, rules out an open arcaded corridor, there was perhaps a series of windows here, either glazed or closed by shutters. The north-south slope of the land dictated that the corridor

Glass tessera A single dark emerald opaque green glass tessera (Fig. 13.11), measuring 7mm x 5mm x 6mm, was recovered from section 929 of Ditch 313 (Trench 35) on the eastern side of the site, and south of the East Range. The tessera was in the top fill (930) of the ditch. Associated pottery dated c.150-300 CE. The layer below this produced pottery (all BB1) dated c. 160-270 CE. Thus these fills apparently date to the site’s Middle Roman Period c. 120-250 CE. The tessera was not originally recognised as glass, and its true nature only became clear during post excavation study. There was no mortar adhering to any surface. Glass tesserae are rare in Britain’s villas, mostly occurring in urban contexts or used as highlights in figural mosaics. When found loose they are generally seen as evidence of wall or vault mosaics, or from the decoration of interior lararia or nymphaea. The early Fishbourne-inspired villa at Southwick produced examples, as did the 4th century complex at Keynsham, Somerset. They were also used 176

Internal decoration in portable emblemata, highly detailed pictorial mosaics made in workshops and laid in terracotta or stone framelike trays. These luxury items, executed with tiny tesserae in the technique known as opus vermiculatum, might be either mortared into a mosaic as its central emblema, or set into a wall like a painting. As they were in solid frames they could be removed at a later date and sold.

signinum and numerous tesserae suggest that they too probably had tessellated floors. Druce Villa is an important, but decidedly enigmatic site that raises new questions. What was going on in Room N3 resulting in over 5,000 small, scattered terracotta tesserae? Was there a first floor mosaic or a tessellated dado in an upper room? The presence of a glass tessera, and an element that may hint at an early opus sectile panel or pavement, raise more questions concerning its decoration that further excavation may one day solve.

The dating of the context of the tessera makes it unlikely to have been part of any of the known mosaics unless it was part of the decoration of the apsidal room. It could have decorated an internal lararium or a water feature in the villa. Alternatively, it may have been used as inlay and ornamented a piece of furniture. It is further discussed by Allen in Chapter 13.

Druce Farm was a wealthy, possibly originally official, and complicated site and an amazingly long-lived one and is a very welcome addition to the corpus of both villa and mosaic studies.

Discussion

Painted wall plaster Andrew Morgan

The villa at Druce Farm has produced a fascinating collection of new mosaics of at least two periods with evidence for earlier lost floors. The predominately red floors of Rooms N6, N7 and the borders of the lower Mosaic 6 in Room N10 seem most likely to date from the time of the aggrandisement of the original villa during the late 2nd-3rd century CE. Although not fail-safe for dating, the presence of red terracotta borders with slightly smaller coarse tesserae than those used on later mosaics might endorse this. That these pavements date from this period would seem a sensible conclusion when considering that the laying of mosaics was a major undertaking, and part of a combined system of refurbishment in league with builders, plasterers and painters. Floors had to be in situ before the plastering or painting of walls and ceilings was undertaken, and then were finished off by washing and polishing after the mess of those other processes had been removed. It was a sequence that would have taken weeks, if not months, to achieve and to dry out. Tesserae used in the lower mosaic (615) in Room N10 seem similar to those found from context 1059 underlying the later mosaic in Room N2, and perhaps suggest that they were coeval. Mosaics 3 and 4 in Rooms N6, N7, and presumably those of N8 and N11, were quite simple, but well laid. Mosaic 4 was particularly attractive and such red and white pavements remained popular in the area for at least another century, often occurring in Dorchester and at other Dorset villas. Mosaic 6 appears to have been the most ambitious of the new floors and may not have looked so very different tonally from the floor that replaced it in the mid-4th century.

Introduction In total 3,556 pieces of painted wall plaster were recovered from the villa. Most of the assemblage was retrieved from the North Range, with a much smaller quantity coming from the Period 4.2 East Range Aisled Hall. No painted plaster was recovered from the West Range. The material was very fragmented, and much of it was abraded. Only two examples were found in situ. After recovery, the painted plaster was cleaned and consolidated using a PVA Universal Adhesive and Bonding Agent diluted 1:3 with water. The assemblage The largest assemblage derived from Room N1, with 1,453 pieces from several layers of building material over the mosaic pavement. Material from layers was also recovered from Rooms N2 (221 pieces), N4 (371 pieces) and N11 (75 pieces). Small amounts were recovered from layers in Rooms N3 (13 pieces), N6 (11 pieces), N7 (22 pieces) and N9 (2 pieces). There was a limited amount recovered from contexts in Room N10 (16 pieces). Additional material (176 pieces) was retrieved from the infill of a large pit (614) which had been dug through the mosaic floors in that room, probably in the final/post Roman period, and is considered to have derived from this room. No wall plaster was recovered from Room N8, and that from Room N5 could not be securely identified as originating there. There were three additional deposits of discarded building material containing painted wall plaster associated with the North Range. The first related to the Period 4.2 redevelopment of Room N3 and consisted of material used to fill what had been a semi-sunken room. The other two were midden deposits created in the post-Roman period, one located east of Room N4 and another south of Room N3. A limited amount of wall plaster (86 pieces) was recovered from collapse contexts in the Aisled Hall, with additional material (309 pieces) recovered from infills of three pits within the building. All collapse layers of the

The new burst of activity that saw the re-structuring of Rooms N1 and N2 and their mosaics in the mid-4th century also saw a replacement mosaic for Room N10. The ‘retro’ designs of Rooms N1 and N10, probably by mosaicists based in Lindinis, and the unusual mosaic of Room N2 must date from this era around the 350s CE. Modern ploughing on the site has erased much of the evidence for mosaics in several of the rooms of the North Wing and that of the Aisled Hall, but the presence of opus 177

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) complex arrangements with up to six elements (L6). The individual elements are described by depth using the convention devised by Davey and Ling (1982, 81), i.e.: Line 20mm. 3. Imitation Marble: four styles were identified showing a simple representation of marble created by applying splashes and dots of coloured paint on a background of grey (IMG), light grey (IML), pink (IMP) and white (IMW). 4. Decorative Shapes: a small number of pieces had evidence of distinctive decorative designs and are further divided into four types: a. circles (DC), b. vegetal forms of leaves, stems or petals (DV), c. geometric designs of precise painted shapes (DG), d. irregular forms (DI). These are a small number of polychrome shapes suggestive of naturalistic designs, some possibly used in paintings. Note: that the material is limited and none of the examples could be combined to determine meaningful forms.

villa buildings contained thick deposits of fine sandy material comprising decomposed wall plaster and mortar. Small quantities of painted plaster were also recovered from the fills of ditches and middens. Most of the painted plaster can be attributed to the final phase of construction that occurred during the Late Roman Period 4.2 (c. 300-350/70 CE). A small amount of material can be attributed to the Middle Roman Period 3.2 (c. 200250 CE) and a lesser amount was possibly from the Early Roman Period 2.2 (c. 70 -120 CE). Methodology Each piece of painted wall-plaster was recorded by context. A style catalogue was created, based on background colour and the presence of any design features. In total, 91 decorative styles were identified, which were then grouped into four main categories: 1. Monochrome Colours: areas of painted plaster of a single colour (M1): yellow, red, white and pink being the most common. Fragments showing two adjacent areas of colour were recorded as (M2). 2. Linear Designs: a large number of linear designs were classified by the use of different colours and width of the elements. They were sub-classified by the number of elements, from a simple single line (L1) to more

Although developed independently, the basis of this approach is similar to that used by Wessex Archaeology when examining the material from Charles Street, Dorchester (Adam 1990, 56), where 31 decorative types were identified (6 monochrome, 23 linear, 1 geometric and 1 naturalistic). A series of photographs (Figs 10.10-15) illustrate examples of decorative styles in use at Druce.

1. Style-6: Grey imitation marble (IMG). Room N12, Phase2

2. Style-7: Light grey imitation marble (IML). Room N1, Phase 4

3. Style-4: Pink imitation marble (IMP). Aisled Hall, Phase 4

4. Style-61: White imitation marble (IMW). Aisled Hall, Phase 4

Fig. 10.10. Examples of imitation marble designs (1-4)

178

Internal decoration

1. Style-13: Blue Composite Linear Design (L6), example of imitation architectural moulding. Room N1, Phase 4

2. Style-12: Red Composite Linear Design (L6), example of imitation architectural moulding. Room N1, Phase 4

3. Style-11: White stripe and dark blue band forming division between central zone and dado (L2). Room N1, Phase 4

4. Style-8: White line embellishing join of areas of yellow and red (L1). Room N1, Phase 4

5. Style-24: Green stripe combed, between areas of red and white (L1). Room N4, Phase 4

Fig. 10.11. Examples of linear designs (1-5)

Room N1

Table 10.1 lists the styles associated with Room N1; the full style catalogue and photographic record are held in archive. The different decorative styles from each room were assessed using the standard three decorative zones found in villas throughout North West Europe: the lower zone or dado, the main or central zone and the frieze or upper zone (Davey and Ling 1982, 31). An attempt was made to identify ceiling decoration by examining the nature of the plaster, the colour scheme, and the decorative styles. Each room is considered in turn, with special emphasis on Room N1 which features the major assemblage, followed by the other rooms in the west wing, then the remaining rooms and associated features of the North Range, then the Aisled Hall and, finally the other sources.

This room was extended and refurbished during the Late Roman Period 4.2 (c. 300-350/70 CE). It had three load bearing walls constructed in flint, with an internal timberframed wall to the south. It is likely that the room was open into the roof space and did not have a ceiling (see Chapter 5). In total 1,453 fragments of painted plaster were recovered where they had fallen as the building collapsed. The area of recovered plaster was approximately 2.43m2 which represents only c. 4.6% of the wall area, based on an estimated wall height of 2.4m and potential door and window openings. Most of the plaster was retrieved from the north, east and west sides of the room. Examples of 179

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

2. Style-14: Arrangement of red circles on yellow (DC). Room N11, Phase 4

1. Style-14: Alignment of red circles decreasing in size on yellow (DC). Room N1, Phase 4

4. Style-47: Single row of grey circles linked by grey line on red (DC). Room N3 infill, Phase 3

3. Style-60: Two rows of red and pink circles on grey, adjacent to combed red (DC). Room N3 infill, Phase 3

5. Style-93: Solid black geometric design (DG). Midden south of Room N3

6. Style-62: Black and grey geometric design (DG). Ditch 313

Fig. 10.12. Examples of circles and geometric designs (1-6)

180

Internal decoration

1. Style-40: Top of floral motif in cream on yellow, with red and white linear borders (DV). Room N4, Phase 4

1. Style-40: Part of floral motif with stems or leaves in cream on yellow (DV). Room N4, Phase 4

1. Style-41: Part of white floral motif on grey (DV). Room N4, Phase 4

1. Style-42: Red stems (possibly reeds) on pink (DV). Room N10, Phase 4

1. Style-57: White and red leaf shapes between red and yellow join (DV), with evidence of over-plastering. Room N10, Phase 3

Fig. 10.13. Examples of designs using vegetal motifs (1-5)

27 decorative styles were recorded, although 13 were represented by less than ten pieces including four that were represented by single fragments.

was 73mm with indentations showing it had been applied to a flint wall. Other pieces featured indentations of reed bundles, indicating attachment to a timber framed wall.

Within the assemblage, 119 pieces of plaster were extremely abraded and remained unclassified (style-0). All these pieces were backed by sandy plaster; the thickest example

The dominant colours from the assemblage were yellow and red. The monochrome yellow (Style-1) comprised 489 pieces which measured 1.23m2 in area and represented 51% 181

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

1. Style-35: Irregular shapes of different colours (DI), possibly from a small painting. Room N4, Phase 4

2. Style-30: Irregular shapes in turquoise overpainted on red (DI). Midden south of Room N3

3. Style-48: Irregular decorative shapes in pink, yellow and orange (DI), possibly from a small painting. Room N4, Phase4

4. Style-36: Azure blue, possibly from a small painting. Room N4, Phase 4

Fig. 10.14. Examples of designs using irregular shapes (1-4)

of the recovered material. The paint had been absorbed into the top layer of plaster (intonaco) having been applied when the plaster was damp (the fresco technique). Evidence suggested that yellow was the background colour for much of the room, with other colours and designs painted over this after it had dried. There were variations in the shades of yellow, but these were attributed to different environmental conditions experienced before and after collapse and subsequent burial. Several examples displayed evidence of decoration processes, with 36 pieces showing brush strokes captured in the top layer of plaster, and three pieces with incised lines made by a sharp implement, although none related to any surviving decorative scheme. The pieces were backed by fine-grained sandy plaster, apart from three with a high lime content.

the top layer of plaster in the fresco style, but some areas showed that the red paint had been applied on a surface previously painted yellow, and partial flaking revealed the background colour. There was no evidence for a decorative scheme painted over the red. There were 18 pieces featuring a brighter red colour, eight of which displayed paint applied directly on the coarse middle layer of plaster. This was interpreted as evidence of a functional repair to the original decoration, comparable to repairs undertaken on the mosaic floor in this room as described by Beeson above. Brush strokes were recorded on 29 pieces and two featured an incised line. The majority of pieces were backed by a sandy plaster, but 32 had plaster with a higher lime content, of which nine were very high. Some of these had reed indentations (each with an approximate diameter of 6mm). There were 74 pieces featuring areas of yellow contiguous with areas of red (Style-5). Three pieces exhibited signs of repair, including one piece with a crudely applied top layer of plaster with a depth of 5mm. Most fragments were backed by sandy plaster, apart from six lime-rich pieces.

The second-most abundant decorative style was a monochrome red (Style-2) with 257 pieces forming an area of 0.75m2, representing 31% of the recovered surface area. This largely featured paint that had been absorbed into 182

Internal decoration

1. Style-77: Green band on pale green and white join, with evidence of over-plastering. Room N10

2. Style-77: Green with evidence of over-plastering and top coat of yellow. Room N10

3. Style-69: Vibrant yellow adjacent to dark red, with evidence of overplastering. Room N10

4. Style-59: White line and curved bands in shades of red, purple and dark grey with blue spots. Room N3 infill

5. Style-2: Red with evidence of over plastering. Room N10

Fig. 10.15. Examples of Phase 3 decorative designs (1-5)

183

STYLE TYPE No

Name

Total Pieces

Description

Context numbers with number of pieces 150/197 138 150 151 151 156 182 197 236 858

184

A

A

Architectural Pieces

21

 Red

21

 

 

 

 

0

UN

Unclassified

119

Devoid of sufficient colour or pattern

98

 

4

12

4

1

M1

Yellow

489

Monochrome yellow

479

 

3

 

2

2

M1

Red

257

Monochrome red

246

3

4

 

3

M1

White

36

Monochrome white

30

 

 

4

IMP

Pink imitation marble

15

Pink imitation marble effect with random spots and splashes in black, white and red

14

 

5

M2

Yellow/ Red

74

Contiguous areas of yellow (St-1) and red (St-2) joined along straight edge

57

6

IMG

Grey imitation marble

81

Medium grey imitation marble effect with random spots and splashes in red, blue, white, yellow and black

7

IML

Light grey imitation marble

5

8

L1

White stripe

9

L1

10

 

 

 

 

 

 

 

 

1

1

 

 

 

4

1

1

2

 

 

 

 

 

 

 

6

 

 

 

 

 

1

 

 

1

 

16

 

 

 

 

 

74

1

 

6

 

 

 

 

 

Light grey imitation marble effect with random spots and splashes in blue, white and black

5

 

 

 

 

 

 

 

 

93

White line 7-8mm dividing yellow (St-1) and red (St-2)

91

 

 

2

 

 

 

 

 

Red stripe

22

Red stripe 11-12mm on yellow (St-1) and white (St-3)

22

 

 

 

 

 

 

 

 

 

L1

Red line

63

Red line 3-6mm on yellow (St-1) and White (St-3)

63

 

 

 

 

 

 

 

 

 

11

L2

White stripe and dark blue band

19

Twin linear elements of white and dark blue dividing red (St-2) from grey stylised marble (St-6): 1. White stripe 10-14mm: 2. Dark blue band 20-22mm

19

 

 

 

 

 

 

 

 

 

12

L6

Red Composite Linear Design

41

Six elements on yellow background: 1. white line 7mm: 2. pink stripe 12mm: 3. red stripe 12mm: 4. dark red stripe 12mm: 5. yellow stripe 12mm: 6. dark red line 6mm

41

 

 

 

 

 

 

 

 

 

13

L6

Blue Composite Linear Design

61

Six elements on yellow background: 1. white 7mm: 2. pale blue 12mm: 3. med blue 12mm: 4. dark blue 10-12mm: 5. yellow 10-15mm: 6. dark blue 3-7mm

61

14

DC

Red circles

13

Aligned round and oval red circles on yellow (St-1). The circles range in diameter from 4-20mm

12

 

 

 

 

 

1

 

 

15

L1

Red band on yellow

2

Red band 30mm thick painted on yellow background

1

 

 

 

1

 

 

 

 

 

17

L2

Two white lines on yellow and blue join

6

Two white lines overlapping yellow and blue background: 1. white 4mm: 2. blue 4-6mm: 3. white 5mm

6

 

 

 

 

 

 

 

 

 

18

L2

Double stripes

9

Two parallel lines painted onto white background: 1. dark grey indeterminant thickness: 2. light grey 10-12mm

9

 

 

 

 

 

 

 

 

 

19

L1

Blue stripe

10

Light blue line 6mm painted on yellow.

10

 

 

 

 

 

 

 

 

 

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Table 10.1. Phase 4 decorative style catalogue for Room N1

Table 10.1. Continued STYLE TYPE No

Name

Total Pieces

Description

Context numbers with number of pieces 150/197 138 150 151 151 156 182 197 236 858

21

L1

Yellow line

2

Yellow line 2-5mm thick with right angled T-junction on red background

2

 

 

 

 

 

 

 

 

 

22

M1

Green

2

Green (some variations)

1

 

 

 

1

 

 

 

 

 

23

M2

Green and pale green

1

Green contiguous with pale green

 

 

 

1

 

 

 

 

 

24

L2

Green band on white 6mm below join of dark red and white

4

Dark red contiguous with white, and green band 30mm positioned on white 6mm below join

4

 

 

 

 

 

 

 

 

28

M2

Dark red/ brown

1

Dark red contiguous with area of brown

 

 

 

 

1

 

 

 

 

 

61

IMW

Imitation white marble

6

Irregular dark grey spots ranging in diameter from 5-13mm on white background. Possible suggestion of a pattern

 

 

 

 

 

 

 

 

6

 

97

DI

Yellow, red, pink decorative

1

Possibly similar to St-48

 

1

 

 

 

 

 

 

 

 

 

 

TOTALS

1366

6

11

34

13

2

2

2

12

5

1453  

185 Internal decoration

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) The absence of colour blending confirms that different colours were applied with care. One fragment was found in situ near the bottom of wall 120. It was noted during excavation that this grey area continued seamlessly into an area of lighter grey (Style-7) represented by only 5 pieces (Fig. 10.10/2), with smaller and less frequent coloured shapes. The examples of imitation marble using a pink background in Fig.10.10/3 were from a total of just 15 pieces (Style-4), featuring smaller dots and splashes from 2mm to 10mm in diameter, in white, black and red. The pieces were backed by a standard sandy plaster, apart from a single piece which was applied to a lime-rich plaster. Six pieces had a backing plaster containing ceramic fragments, and three pieces exhibited a middle layer of red coloured plaster. It is possible that the pink imitation marble formed a small but conspicuous contrasting panel within the dado. This pink simulated marble effect is found elsewhere in the villa in Room N2 and in the Aisled Hall. It is a common decorative style and is also recorded at Charles Street, Dorchester (Adam 1990, 56) and at Wortley, South Gloucester (Wilson et al. 2014, 133). There were six pieces with a white background featuring irregular black spots ranging in diameter from 5mm to 13mm (Fig. 10.10/4), which were identified as examples of white imitation marble (Style-61).

The other example of a monochrome colour was white (Style-3), with a limited assemblage of 36 pieces. The colour was variable, which again was interpreted as discolouration after application. All pieces were backed by sandy plaster. The areas of yellow and red would have dominated the main or central zone of the decorative scheme in Room N1. It is likely that this zone comprised two-dimensional, upright rectangular panels, typically three or five per wall. From the relative quantities of the prime colours it would seem the panels were probably painted yellow surrounded by a broad framework of red. The schematic Figure 10.16 illustrates a representation of the possible decorative scheme. The use of two-dimensional panels is the most common decorative scheme found in British villas and was used throughout the Roman period (Davey and Ling, 1982, 31). There were four distinctive decorative styles featuring splashes of colour on a plain background, which are simple forms of imitation marble (Fig. 10.10/1-4) typical of the later Roman period (Davey and Ling, 1982, 31). This style often associated with the dado zone is found in many villas throughout Britain, including Tarrant Hinton, Dorset (Lucas, 1993c, 56). The largest assemblage of 81 pieces (Style-6) measured 0.31m2 representing 13% of the surface area (Fig. 10.10/1). The grey background was covered in random spots and irregular shapes in white, red, black and yellow that had been applied by stippling and splashing, evidenced by small splash marks created on impact. These shapes ranged in diameter from 2mm, with occasional larger shapes with lengths in excess of 30mm.

A total of 12 linear designs were recorded (Fig 10.11). All were painted with precision, being straight and having a consistent width. A simple white stripe of 7-8mm width (Style-8) was used to highlight lengths where areas of yellow were contiguous with areas of red (Fig. 10.11/4). This design featured on 96 pieces, forming a length of approximately 2.50m. The white line was always applied

Fig. 10.16. Schematic decorative scheme for Room N1, Phase 4

186

Internal decoration on the yellow (secco), and several pieces show that the white paint had not formed a strong bond and had flaked off, sometimes represented only by a slight raised ridge on either side. A single piece exhibited a second white line at right angles creating a T-shaped junction. All the pieces were backed by sandy plaster, and a few had a grey coloured middle layer, possibly contaminated by charcoal during preparation. Two further linear designs were used to create a simple framework within the yellow panels. The first, recorded on 63 pieces, used a thin red line (Style-10) with a width of 4mm, including two fragments featuring examples of right-angled corners. The second, with 22 pieces, used a red stripe (Style-9) with a width of 11-12mm. Three of these pieces had right-angled corners and a further two pieces had a right-angled T-junction where it was abutted by a red line. All these pieces were backed by a sandy plaster and three pieces had indentations suggestive of a timber framed wall.

assemblage of painted wall plaster from Tarrant Hinton (Lucas 1993c, 56), although only the dark red and red/pink elements were recorded. Monochrome white (Style-3) was found on 36 pieces of painted plaster. Slight variations in colour are explained by different environmental conditions experienced since application of the paint, and possibly from the conservation process. Some examples that appeared to be slightly cream were painted on a yellow background and may represent a later repair or redecoration. All pieces had a sandy mortar backing. It is possible that white was used for the upper layer of the frieze surmounting the complex linear designs, as used in the Catterick mansio (Davey and Ling 1982, 34). An additional six pieces with a white background featured a random array of small dark-coloured spots 3-10mm in diameter (Style-61). There was evidence of splash marks showing that they had been applied by sprinkling in the manner used for the imitation marble effect, and this is interpreted as an example of white imitation marble.

The dado zone was surmounted by a strong linear design with two elements: a white stripe of 10-14mm on top of a dark blue band 20-22mm (Style-11). This created a distinctive division between the dado and the red framework of the central zone. There is evidence for a gap between the red and the dark blue into which the white line was painted (Fig. 10.11/3). All pieces had a sandy backing plaster and one had impressions on the bottom and side suggesting a corner location. One piece with a slip of coarse plaster on top of the painted surface is taken as evidence of shoddy repair.

There was some evidence for other decorative embellishments, the most distinctive being an alignment of seven small red circles which decreased in diameter from 20mm to 4mm (Style-14, Fig.10.12/1). These were used to decorate a yellow panel and were possibly displayed as a T-shaped formation of spots and set diagonally (Davey and Ling 1982, 33). There were 22 pieces identified as architectural plaster, shaped to accommodate openings in the walls, possibly for windows, doors or even a recess; all were painted red, supporting the use of red as an outer border. Ten pieces had chamfered edges, five set at approximately 30⁰ and another five set at approximately 45⁰. A further 12 pieces had very slight undulating surfaces, the purpose of which is not known and may just be a consequence of an uneven base surface.

There were two examples of more complex linear designs, each made up of six elements with an overall width of c. 61mm. Although based on different colour schemes, both designs conformed to the same decorative conventions. The Blue Composite Linear Design (Style-13) comprises 61 pieces forming a sequence of six parallel elements: white stripe 7mm, pale blue stripe 12mm, medium blue stripe 12mm, dark blue stripe 12mm, yellow stripe 10-15 mm and a dark blue line 3-7mm. The overall width is 5665mm (Fig. 10.11/1). There is evidence for partial incised lines, probably used as a guide to align the design. All the pieces were backed by a sandy plaster. It was possible to assemble the pieces to create a pattern with a length of 1.40m; this included one corner piece set at a right angle. The Red Complex Linear Design (Style-12) is represented by 41 pieces and follows the same conventions: a white stripe 7mm, pink stripe 12mm, red stripe 12mm, dark red stripe 12mm, yellow stripe 12mm and a dark red stripe 6mm (Fig. 10.11/2). It has an overall width of 61mm. The complete sequence is painted on a background of yellow. All the pieces were backed by sandy plaster. The top and bottom were not obvious, but it is possible that the darkest colour would have been at the top, becoming lighter towards the bottom. Such complex linear designs with multiple elements were used to create the illusion of an architectural form such as plaster mouldings (Davey and Ling, 1982, 32). It is likely that both these design styles were used to define the frieze zone. The use of a comparable linear design to Style-12 was present in the

There was no convincing evidence for ceiling plaster. Some fragments had a plaster backing with a high-lime content and showing indentations derived from reeds, sometimes in tied bundles, that is often regarded as evidence of ceiling plaster (Timby and Morgan 2014, 93), but in this instance this material is believed to have been associated with the timber-framed south wall 235. In conclusion, the decorative scheme of Room N1 was functional but, nevertheless, colourful and impressive; a schematic reconstruction is shown in Fig. 10.16). Arranged in the standard three zones, it is likely that the central zone comprised a number of two-dimensional upright rectangular panels painted in yellow, usually three or five along each wall, set within a broad framework of red. Some, if not all, of the panels were enhanced by a simple inner framework using a thin red line or stripe, some were further embellished with small decorative features such as an alignment of red circles, possibly running diagonally from a corner. In places, possibly above the panels, the join between the yellow and red was highlighted by the 187

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) use of a distinctive white stripe. At the top of the central zone the frieze was likely defined by a complex linear design with six elements using four shades of blue or red, creating the effect of architectural moulding. The upper part of the frieze was probably painted in white. The dado was created using a stylised marble effect; the main part was a background of grey featuring strong random spots and irregular shapes in red, dark blue, yellow and white, the grey becoming lighter and the shapes smaller and less colourful towards the floor. It was surmounted by a linear design formed by a distinctive white band over a dark blue band separating the grey marble effect of the dado from the red interval of the central zone. There is no convincing evidence for ceiling plaster.

of white and dark blue linear design (Style-11) which probably surmounted the dado. A number of other design features found in Room N1 were also represented, such as complex linear designs giving the impression of architectural moulding defining the bottom of the frieze (Style-13).

Room N12

There were additional contexts probably associated with earlier development phases, but the samples recovered were insufficient to give any indication of a decorative scheme. Two pieces were associated with Phase 2 but were devoid of any colour or decoration. Phase 3 was represented by a single green painted fragment. Of interest were several pieces in red and white (Style-51) with a gloss-like finish that suggests an oil-based paint.

A number of fragments displayed decorative forms. These included one piece with a distinctive curved cream shape on a yellow background (Style- 40), which may have been the petal of an individual flower head painted on a yellow panel. There were also several pieces with a pink background and red swirls with a form that could not be interpreted (Style-42), which were unique to this room.

A small sondage in the south-east corner of N1 provided the opportunity to investigate the infill of material found beneath the mosaic floor (Fig. 5.14/1). It is believed this material came from the apsidal Room N12 that was constructed and demolished during the Middle Roman period 3.1 Phase 2 building works (c. 120-200 CE). The assemblage comprised 26 pieces of painted wall plaster, with 10 pieces of red (Style-2), five pieces of grey imitation marble (Style-7), two pieces of pink imitation marble (Style-4), two pieces of yellow (Style-1), four pieces of white and dark blue bands (Style-11) and one piece of a white stripe between areas of red and yellow (Style-8). Whilst this was a small sample, it is significant that all the design styles also featured in the final decorative scheme used in Room N1 suggesting that this scheme was influenced by, or even replicated, that used in a much earlier phase of development. This echoes the conclusion by Cosh and Neal when visiting in 2013, that the design used for the mosaic floor in Room N1 was typical of a much earlier period.

The final decorative scheme for Room N2 was substantially the same as for Room N1, which is consistent with its use as an ante-room. There were slight variations, including the width of the white band used in Style-11 and the width of the white line used in Style-8, together with a small number of decorative embellishments not found in Room N1. Room N4 The function of this small outer room, attached to the east side of Rooms N2 and N5, is problematical as only the northern extent survived. A total of 371 pieces of plaster were recovered from collapse layers within the room, of which 39 were devoid of colour and unclassified. Thirty five different styles were identified, 20 of which had five or less pieces.

Room N2 This small room probably functioned as an ante-room to Room N1 as it appears to offer the only means of access. Its final configuration was laid out during the Phase 4 building works (c. 300-350/370 CE). The room was badly damaged by ploughing. A total of 221 pieces of painted plaster were recovered including 43 unclassified fragments. Three contexts containing very small assemblages were potentially associated with earlier building phases.

This room featured six main colours. The dominant colours were red (Style-2), with 32 pieces, and yellow (Style-1), with 24 pieces, which were likely to be from panels in the central zone. There were 25 pieces with shades of grey (Style-67); three pieces featured a convex curve probably associated with a door or window opening, suggesting grey was used as a border, 33 pieces displayed varying shades of green (Style-22), and 25 pieces were of pink (Style-44). There were 11 linear designs of which six were represented by a single piece. The most common (30 pieces) featured a white line of 5-8mm between areas of dark red and grey/blue (Style-46). Four pieces of dark red contiguous with an area of white onto which was added a green band that had been combed (Style-24) (Fig.10.11/5) were also recorded.

Painted wall plaster featuring 20 decorative styles relate to this room, of which 14 styles had less than ten pieces, including five represented by single pieces. Ten of the decorative styles were also found in Room N1. The dominant colours are similar to those in Room N1: monochrome yellow with 32 pieces (Style-1), 54 pieces of red (Style-2), 11 pieces showing a white line between areas of yellow and red (Style-8) and 13 pieces of white (Style-3). The dado is represented by 24 pieces of grey imitation marble effect (Style-6) and six pieces of the pink imitation marble (Style-4). There were three pieces

Several fragments had remnants of two floral motifs. The first comprised three pieces with part of a flower head painted in cream on yellow (Fig. 10.13/2), two of which 188

Internal decoration are adjacent to red and white lines (Style-40) which may have formed an associated framework (Fig. 10.13/1). The second is a simple flower head painted in white on grey (Style-41) (Fig.10.13/3). Such motifs were used to enhance panels, such as those found at Wortley villa (Wilson et al. 2014, 135).

with small ceramic fragments that potentially represent degraded opus signinum. The painted material could have derived from an upper room or from a lararium. Room N6 Based on the dating of the plain red mosaic pavement, this room was decorated in Phase 3, Middle Roman period (c. 200-250 CE). There were few fragments of wall plaster in the collapse layers, with 11 pieces of pale brown (Style-32) and a single piece of red (Style-2). While the evidence is limited, the material is consistent with the simple style of the mosaic floor and may reflect the function of the room.

Significantly, this room contained a number of pieces exhibiting irregular polychrome shapes and impressions quite unlike material found elsewhere. Two assemblages have been identified based on the colours. The first uses a palette of red, pink, yellow, orange, light green and brown (Style-35), which have been used as blocks of colour with linear brush strokes suggesting foliage (Fig. 10.14/1). This colour combination was recorded in building 182, Room 17 at Colliton Park, Dorchester, and was considered to be potentially from a freestyle landscape painting (Timby and Morgan 2014, 94). The second assemblage has colours in pink, yellow and orange, again suggestive of foliage, adjacent to a panel of green (style-48) (Fig. 10.14/3). There was a single piece of monochrome azure blue (Style-36) (Fig. 10.14/4), the only example on site, which could have represented the sky. The evidence is limited, but it is possible that the above fragments represent one or more small wall paintings.

Room N7 This room, located opposite the probable main entrance to the North Range, was also decorated during the Phase 3 Middle Roman period. A small amount of painted plaster was recovered from the collapse layer, comprising 12 pieces of dark red contiguous with white (Style-51), five pieces with a brown background (Style-32) and five pieces of pale brown contiguous with white (Style-31). Although the evidence is modest, a simple decorative scheme in red, white and brown is likely. There was a single piece of concave shaped architectural plaster suggesting it derived from a door or window.

Room N5 All the material associated with Room N5 was recovered from clearance layers and its origin is uncertain. The room was extensively plough-damaged and the plaster may have derived from Rooms N2 and N4. Of interest were four pieces in a pale yellow featuring a distinctive top layer (intonaco) which was particularly hard. The underlying plaster was very different, in that it was hard and clay-like (style-B). It may have been a plaque. A similar fragment was also recovered from Ditch 309.

Room N9 Only two pieces were recovered from this probable service room. One showed an area of dark red contiguous with an area of white (Style-51) and the other had the same colours but featured a 30mm green band positioned on white 6mm below the join (style-24). It is likely that the walls of this room were whitewashed, and it is possible that the recovered wall-plaster could have derived from a household shrine; such features were often associated with service rooms (Perring 2002, 197).

Room N3 The stratification of Room N3 was complex (see Chapter 5) with deep collapse layers, especially at the north end. There were many layers of roof tiles and flint from the walls, embedded in a large amount of a sandy material interpreted as decomposed plaster. It would seem that these layers had been disturbed soon after the collapse to remove limestone roof tiles, and only 13 pieces of painted wall plaster were recovered. Of these, nine pieces were unclassified, two pieces were monochrome red (Style-2), one piece was pink imitation marble (Style-4), one small piece was turquoise overpainted on to a red background (Style-30) (Fig. 10.14/2) and one piece was pink with a red decoration (Style-42). The evidence suggests that the walls were plastered, but the absence of significant painted material implies that they were not decorated. It is possible they had been whitewashed, if only for utilitarian purposes. This is consistent with the latest floor of rammed chalk, with no evidence for a paved floor other than a loose collection of c. 5,000 small red ceramic tesserae, which represents less than 1m2, and some isolated areas

Room N10 Centrally positioned in the North Range, this was a probable reception room. It underwent at least two major alterations during Phases 3 and 4, when a second mosaic floor was laid directly over the first. There is evidence for two phases of wall decoration, with a new layer of painted plaster applied over an earlier design. The room was severely plough damaged and only 16 pieces of painted plaster were recovered from the north end of the room. However, pit 614 in the north-east corner was dug through the floor after the villa had fallen into disrepair (see Chapter 5). The pit was half sectioned (Figs 5.43/2 and 5.46) revealing much building material, including 176 pieces of painted plaster believed to have originated from the room. From the evidence of over-plastering seen on a number of pieces (Figs 10.15/2 and 4), 45 pieces from this assemblage have been attributed to building phase 3, the remainder being associated with phase 4. 189

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) The Phase 3 (c. 200-250 CE) over-plastered material can grouped into 6 decorative styles. The most abundant comprises 25 pieces showing areas of white contiguous to an area of pale green, which is enhanced by a darker green band of 45mm (Style-77) (Figs 10.15/1 and 2). The design is exclusive to this room and phase. It should be noted that green was used as the dado zone at Tarrant Hinton in Building 2, Room 2.1 (Graham 2006a, 155). Another interesting example was an area of white contiguous to a small exposure of black (Style-74), but this was only partially exposed under the new plaster. Three pieces of plaster display a particularly vibrant shade of yellow adjacent to a dark red (Style-69) (Fig.10.15/3). A further decorative style not found elsewhere was elegant, elongated leaves in red and white enhancing the join between areas of red and yellow (style-57) (Fig. 10.13/5). On all these examples, the layer of new plaster was very eroded, but appeared to have thicknesses between 8mm10mm. The surface of the old plaster was noticeably hard and smooth suggesting that it had been highly burnished, which is a feature of earlier 1st and 2nd century Roman decoration (Davey and Ling 1982, 30). The assemblage is quite limited, but the dominant colours used in the decorative scheme for Phase 3 were areas of pale green and white, with others of red and vibrant yellow.

were 25 pieces of red painted plaster (Style-2), five pieces of yellow and five pieces of white, with a further six pieces featuring very faded red contiguous to white (Style-51). There were two further examples of a linear design. The first comprised six pieces with adjacent stripes of dark red and pink (15mm and 20mm wide) on a yellow panel (Style-55). This strong design was similar to that used in Room N1 to define the bottom of the frieze (Style-13). There were also seven pieces featuring bands of dark red and white (35mm and 30mm wide) between an area of white and blue (Style-11), which was the same colour scheme used in Room N1, where it was used as the uppermost element of the dado zone. There was a small piece with an arrangement of red circles on yellow (Style-14) with a suggestion of two rows of circles, but its design cannot be confirmed (Figure 10.12/2). A further piece showed two pink lines converging on a very degraded red background (Style-56); its form is elusive. The dominant colours of this limited assemblage were red and white with some yellow, which may have been used for the central zone. It is possible that the dado was represented by a single piece of light grey imitation marble surmounted by white and dark blue bands. The red and pink stripes on yellow may then have defined the frieze of the upper zone. Room N3 Infill

There were a further 152 pieces with no evidence of over plastering which were regarded as material from Phase 4, including 13 unclassified pieces. The dominant colour of pale pink was represented by 49 pieces (Style-44), 39 of which were small fragments with a thin sandy plaster of 8mm thickness; the remainder had a coarse sandy plaster and were up to 32mm thick. A further nine pieces featured irregular shapes in a dark pink suggestive of foliage stems, possibly reeds (Style-42) (Fig.10.13/4), which implies that pink may have been used for the dado as at Halstock Room 3.21 (Lucas 1993c, 99). The central zone is likely to be represented by 17 pieces of monochrome dark red and nine pieces showing red contiguous with pink (Style-73), and a further five pieces with a 9mm cream stripe on the join of the red and pink (Style-70). There were 25 pieces of monochrome white (Style-3) of which six pieces had a high lime content and four exhibited clear impressions of reed bundles. One example showed indentations for 11 reeds, another with seven reeds; the diameter of the reeds was between 5.29mm and 5.45mm. There was also an impression of the ties used to hold the bundles together. This is the most convincing evidence for the presence of ceiling plaster.

The room was completely redesigned and extended in Phase 4 during the Late Roman Period (c. 300-350/370 CE) and involved the use of layers of building material and ash to substantially raise the floor level of the southern half of the room (Fig. 5.31). This material must therefore represent an earlier building phase and is likely to derive from the Phase 3 redevelopment. Over half of this infill was excavated and 64 pieces of painted plaster and five pieces devoid of colour were recovered. There were examples of 16 decorative styles, of which 15 had less than 10 pieces and four were represented by a single piece. A total of 22 pieces featured the standard monochrome colours of yellow, red and white, seven pieces were of green (Style-22) and a further five pieces were red contiguous with green. Notably there were two styles only found in the Phase 3 decorative scheme used in Room N10: four pieces of a distinctive green decoration (Style 77) featuring a green stripe embellishing the join between pale green and white, and a further three pieces featuring an area of red adjacent to a vibrant yellow (Style-74). There were several pieces of painted plaster of note, from four styles unique to this deposit. The first (Style-60) was a single piece that featured two rows of circles (diameter 13mm) painted on grey to embellish the join with an area of red. The circles had been arranged to alternate, although the alignment was uneven. The top row were pink with red centres and the lower row red with dark red centres. The red area has been crudely combed (Fig. 10.12/3). A second style features a fragment showing grey circles (diameter 13mm) attached to a grey line on the edge of an area of dark red (Style-47) (Fig. 10.12/4). A similar design using circles attached to a line was found in rooms 3.2 and 3.3

Room N11 This narrow corridor runs east-west on the south side of the North Range; the west end was badly plough damaged. In total 75 pieces of plaster were recovered from contexts associated with collapse layers; 18 of these were unclassified. Material representing 14 styles was identified, but 12 had six or less pieces including five represented by a single piece. There was some plaster in situ on wall 441 but this was devoid of colour or decorative form. There 190

Internal decoration associated with the bathhouse at Halstock Roman Villa (Lucas, 1993c, 97). A third design (Style-59), represented by a single piece, has a 5mm white line painted on a red background, followed by bands positioned at an angle suggesting the curve of a possible roundel, followed by an area of dark grey with small blue spots (Fig. 10.15/5). Finally, there are five pieces featuring a degraded black band (Style-58). Although the origin of these pieces is not known, they are distinctive and would have featured in a reception room; they appear slightly more sophisticated than those found in the Phase 4 decorative schemes.

were eight pieces of pale brown colour (Style-36) which were recorded in Room N6. East Range /Aisled Hall Four building phases were identified in the East Range, with a substantial Aisled Hall which was developed in the latter two phases. The northern end featured a collapsed wall that was left in situ, so any associated residual wall plaster remained buried. Only 86 pieces were recovered from the collapse layers. A further 309 pieces were retrieved from associated pits, especially pit 209 which contained 297 pieces. The assemblages from these features were consistent with the material recovered from the collapse layers and are considered to represent the final decorative phase.

Midden - South of Room N3 An area immediately south of Room N3 was formed by layers of discarded building material, including large amounts of decomposed sandy plaster and mortar which included 365 pieces of painted plaster and 56 pieces that were devoid of colour. The feature, interpreted as a midden, was associated with the demise of the villa and created in Phase 6 (c. 430-650 CE). The painted pieces were assigned to 41decorative styles of which 17 were represented by less than ten items, including 13 styles represented by a single piece. Plain white (Style-3) was the most common with 75 pieces, with a further 14 pieces featuring a 50mm wide red band on white (Style-31). In addition, there were 59 pieces of red (Style-2). Other monochrome colours were represented by 18 pieces of yellow (Style-1) and 18 pieces of pale brown (Style-32). There were 12 pieces of pink imitation marble (Style-4) and eight pieces of grey imitation marble (Style-6); the latter was otherwise only recovered from Rooms N1 and N2. There were a number of individual decorative pieces, the most striking being two pieces with shapes of dark pink on light pink (Style-50) suggestive of foliage, and similar to those found associated with Rooms N4 and N10. There was also a single piece showing decorative cream vegetal shapes on yellow (Style-45), the same motif was found in Room N4. One piece displayed an irregular shape in turquoise overpainted on red (Style-30), comparable to a decorative design found in Room N4 (Fig. 10.14/2). Part of a distinctive and unique sharp design was formed by conjoining pieces showing two converging black lines on white (Style-93), suggesting the left hand corner of a possible triangle (Fig. 10.12/5). The midden appears to contain material taken from a number of the nearby rooms and may be the product of systematic robbing of building material.

The fragments represent 18 different styles, of which 14 were represented by less than ten pieces and four by a single piece. The predominant colour was monochrome white (Style-3) with 223 pieces, which was probably used for panels in a central zone. This colour is quite common in Romano-British decorative schemes, especially towards the end of the Roman period (Davey and Ling 1982, 37). There were 93 pieces (40 being small fragments) featuring another type of imitation pink marble (Style-4): a pink background with splashes of colour with red spots 8mm, white spots 4-8mm and blue 2mm, with a backing of fine sandy plaster 10mm thick; this is likely to have represented the dado zone. A further four pieces had an area of white adjacent to the pink imitation marble that had been overpainted by a 7mm wide dark/black line (Style-65). This probably defined the top of the dado marking the division with the central zone. There were a further 20 pieces with a white background decorated with an irregular arrangement of blue/black spots ranging in diameter from 5mm to 13mm; the spots had been applied by sprinkling and are interpreted as a white imitation marble design (Style-61) (Fig. 10.10/4). Seven pieces show areas of monochrome red contiguous with the monochrome white, possibly forming a red interval between white panels. The paucity of painted plaster recovered from such a large building suggests that it was only partially decorated, and possibly restricted to a reception area. The decorative scheme was relatively plain, with a central zone painted white, possibly set within a narrow framework of red. The dado was pink imitation marble, which could have been enhanced with panels of white simulated marble.

Midden - East of Room N4 Immediately to the east of Room N4 was another midden area of discarded building material created in the postRoman period. This contained 123 pieces of painted plaster, 26 of which were unclassified. The dominant colours were yellow (Style-1) with 13 pieces, and ten pieces each of red (style-2) and green (style-22). There were 18 pieces with a white line between areas of red and blue/grey (style-86), and six examples of decorative styles only represented by a single piece of plaster. The material in this midden was essentially the same as found in Room N4, although there

Middens and Other Pits Building material was recorded in several ditches and middens and comprised 44 pieces of discarded painted wall plaster. The only design type not found elsewhere were two pieces recovered from ditch 313, having a very distinctive design of adjacent lines composed of a 10mm wide band of dark grey and 12mm wide light grey band on a white background (Style-62) forming a corner with an angle of 45o (Fig 10.12/6). 191

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Discussion

pigments have not been analysed, but locally available pigments such as red ochre, yellow ochre, green earth (glauconite), chalk and charcoal would have met most of the requirements.

The Druce painted plaster suffered a great deal of damage over the years, initially during the decay and collapse of the buildings when building material was removed, and recently, more extensive damage caused by ploughing. This affected the preservation of the plaster in every room and contributed to the fragmentary and abraded nature of the recovered material. The largest assemblage was recovered from Room N1, but even this only represented 4.6% of the original coverage, whilst other rooms provided a very limited amount of material. Most of the wall plaster is associated with the Late Roman Period Phase 4.2 building Phase 4 (c. 300-350 CE), when the last major redevelopment of the villa was undertaken. A small amount recovered from under the floor of Room N1 is believed to be from Middle Roman Period 3.1 building Phase 2 (c. 120-300 CE). A limited amount of material recovered from pits within Room N10 has been assigned to Late Roman Period 4.1 building Phase 3 (c. 250-300 CE), as has the material recovered from the infill used in Room N3.

The most common forms of decoration were linear designs; 39 examples were identified, of which 22 consisted of a single element, 12 used two elements, three examples had three, and two comprised six elements. Single lines and stripes, usually in red but sometimes in blue, were used to form a simple inner framework to enhance panels of monochrome colour, especially yellow. Only right-angled junctions were found in relation to these frameworks, with no evidence to support the use of polygonal shapes. Linear designs were used to highlight or embellish the join between two areas of colour, and especially the division between the three decorative zones. The two most complex designs, comprising six elements imitating architectural moulding, were probably used to define the start of a frieze zone. There were four examples of a simple imitation marble with background colours of grey, light grey, pink and white. The pink background was present in several rooms, whilst a grey background was only used in Rooms N1 and N2. A limited number of decorative embellishments were identified and were represented by a handful of pieces. There are two examples of decorative circles. The first is a simple alignment of red spots that reduce in size. In the second, the use of round shapes, in one or more rows, is used to embellish a join between two colours and was associated with Phase 3 building work. There are a small number of fragments which clearly represent four simple vegetal designs that appear to have been used as individual motifs, possibly within a framework. Two represent flower heads, one in cream on yellow and the other white on grey. A third design is an example of long elegant leaf shapes painted in red and yellow, used as decorative forms along the join of areas of red and yellow. The fourth is dark pink shapes on pink, possibly representing reeds and could have been used to decorate a dado.

A very small amount of painted plaster was recovered from the upper layers of building material found in Room N3, suggesting that this room was not painted. It may have been whitewashed, although there was no evidence for this. Similarly, a relatively small amount of material was recovered from the Aisled Hall, and it is likely that only a small area, perhaps a single room, was painted; again the remainder may have been whitewashed reflecting its use. There was limited evidence for ceiling plaster, but some possible material was located in Rooms N4 and N10. Decorative Schemes The decorative schemes used at Druce villa are based on the standard three zones, using two dimensional panels and the typical Roman colour palette. The schemes are colourful, bold and well-executed, but are quite simplistic in comparison to many other Dorset villas. There were a large number of linear designs, including two examples of an imitation architectural feature where a linear design was used to suggest a stucco moulding. There is limited evidence of decorative flourishes such as floral and vegetal motifs, and the mere hint of a small landscape design, both from Room N4. There is no surviving evidence for the inclusion of architectural features, figurative images or large scale paintings. The Phase 4 decorative scheme used in Rooms N1 and N2 closely aligns with the scheme apparently used in Room N12 during building Phase 2; the use of such a traditional decorative scheme suggests very conservative taste.

Two polychrome design groups were identified comprising irregular shapes. The colours mingle and overlap as though painted in a free form manner to create an impression of depth, and some are suggestive of foliage. Although the evidence is limited, is it possible that these few fragments represent features from a small landscape painting. A single fragment of azure blue may have been used to represent sky. Although the assemblage from several of the rooms is quite modest, three assessments of the decorative schemes were undertaken for each room: 1. Colour usage Comparison by percentage of the use of different colours in each room (Fig 10.17) 2. Decorative styles Comparison of the use of different decorative styles in each room (Fig. 10.18) 3. Linear designs

Across the site, 91 design styles were identified and classified into four categories. There were 14 examples of the monochrome colours that are common in many Roman villas, especially yellow, red, white and pink. Red was found in every room bar one and was the colour most frequently used adjacent to other colours. The paint 192

Internal decoration

Fig. 10.17. Comparison of the main colours used. Percentages calculated using the number of fragments recovered from each room

Fig. 10.18. Comparison of the different categories of decorative styles in each room

193

The Rise and Decline of Druce Farm Roman Villa (60-650 CE)

Fig. 10.19. Comparison of linear design types used in each room

Comparison of the use in each room of linear designs varying from single elements (L1) to six elements (L6) (Fig. 10.19).

were several examples where paint had been combed for decorative effect immediately after application and before it had dried. There was limited evidence of burnishing, which related especially to Phase 3 material from Room N10, creating a particularly hard and smooth finish.

This demonstrates that the lobby-type Room N4 had the most complex decorative scheme, with an intricate arrangement of style types and, in particular, examples of decorative embellishments suggesting that the room had a special status. It also contained material that may represent an example of figurative painting. Room N1 used the most colours and linear designs, with examples of imitation marble. Only Rooms N1 and N2 have imitation moulding. Rooms N6 and N7 were modest by comparison, possibly indicating that the wall decoration was contemporary with their Period 3.2 mosaic floors. In general, the rooms of the west wing of the North Range, particularly Room N10, show a difference between the decorative schemes used in Phases 3 and 4. There is a suggestion that the Phase 3 decoration was better applied and used more sophisticated decorative forms. There is a change in palette from the greens of Phase 3 to the pinks of Phase 4, possibly implying a change in the use and status of the room after the development of newly decorated rooms in the west wing. The simple decorative scheme used in the Aisled Hall suggests that it had a utilitarian rather than domestic function.

Redecoration and Repair There is evidence for complete and partial redecoration in several of the North Range rooms. It would appear that complete redecoration undertaken by skilled tradesmen coincided with major building phases. Only in Room N10 is there evidence for redecoration by over-plastering the previous wall plaster; elsewhere previous plaster was removed during the re-development. Any repairs required in the intervening years were achieved on a best-endeavours basis. It was noticeable that these had been done by a less skilled work force, maybe labourers associated with the villa. Evidence of repairs is shown by poor quality plaster using crudely sorted material with large inclusions, including random ceramic pieces. The ability to match paint colours proved difficult and the paint was poorly mixed. There were many examples where the paint was applied directly onto the coarse underplaster with no evidence of the fine grained top intonaco layer. This compares with the poor-quality, yet functional, repairs made to the mosaic floors especially apparent in Room N1.

Painting techniques

Application of the Plaster

Craftsmen were employed during the major development phases to apply the plaster and the paint; they were competent and worked systematically. The decorative schemes were simple but well-executed, evidenced particularly by linear designs that were straight and of consistent widths. When different colours were involved, care was taken to ensure that the previous application was dry before the next was applied so that the colours did not merge. There were a number of examples of incisions positioned to inform the painting design, although many did not correspond to a specific design feature. There

In common with most buildings constructed throughout Roman Britain, the wall-plaster was usually applied in three layers: bottom; middle (which together form the arriccio) and top (intonaco). The function of the bottom layer was to cover the uneven surfaces of the bare walls, which in this case were constructed mainly of flint nodules, although some internal walls were timber-framed and infilled with laths or bundles of reeds. The fabric of the walls was occasionally shown by indentations on the 194

Internal decoration bottom layer of plaster, which was of variable thickness, the thickest surviving example from a flint wall being 73mm. The second or middle layer of plaster was applied to create a flat, level surface and had a consistent depth, typically between 10-12mm. Both these layers were usually a sandy plaster, typically comprising two/three parts sand mixed with one part lime; the same material was used for mortar. The sand was usually well-sorted and devoid of large inclusions, suggesting careful preparation, although very occasionally the mixture was coarse, with an aggregate of prominent stone or flint inclusions and, occasionally, fragments of ceramic. There was no evidence that this was functional, and it probably reflected the competence of the tradesmen and availability of material. It was preferable that the middle layer was applied before the bottom layer had fully dried to create a strong bond. However, it was noticeable that many pieces of surviving wall plaster had very flat surfaces, indicating that the middle layer had become detached. In a few examples, the middle layer was coloured red or grey, but there was no evidence that this was functional and is interpreted as chance when the plaster was being mixed. A limited number of pieces had a high lime content, which is often regarded as evidence of ceiling plaster, although this relationship was not conclusive at Druce. The top layer (intonaco) comprised a fine grained limerich white plaster, creating a smooth surface suitable to receive the paint. This was usually very shallow with a depth between 0.5mm to 1mm. The paint was best applied when the top layer was damp so that it would be absorbed into the plaster using the fresco technique. There are numerous examples where the brush strokes used for painting have been captured in the damp top layer of plaster. There were a few examples, especially from Room N10 (Phase 3), where the top layer was noticeably hard and smooth, suggesting that it had been heavily burnished. There was limited evidence for a new decorative scheme being applied over an existing decoration, although in Room N10 the Phase 4 scheme was painted after new middle and top layers of plaster had been applied onto the Phase 3 wall plaster.

195

11 Roman and post-Roman pottery Structure of the report

dominated by mid to late mid-Flavian forms. There are only five pre-Flavian sherds representing four vessels: two Dr 15/17 plates, a Ritterling 9 hemispherical bowl and a Ritterling 12 cup. The composition of the mid to late Flavian assemblage is typical for the period, with plain forms Dr 18 (platters) and Dr 27 (cups) constituting 66% of all identifiable forms. The rarity of the offset on the wall on all of the Dr 18 plates/platters underscores the mid-Flavian onwards date. In terms of services, the combination of platter Dr 18 (nine MNV, i.e., minimum number of vessels) and cup Dr 27 (10 MNV) is a typical Flavian profile and the ratio of cup to platter is close to 1:1. Decorated vessels are scarce, with only two Dr 29, four Dr 37 and one Dr 30 by MNV.

This chapter reviews all pottery types recorded at the villa, whilst Dr Lyne evaluates the assemblages in Chapter 12. Reports on the imported, non-local and local wares are presented for the Roman and Romano-British pottery. The material, however, was dominated by the locally-made Black Burnished ware products. Consideration is given to pot sherds which bear graffiti and, finally, the postRoman and medieval material is discussed. Some of the material which is reported on in this chapter is illustrated in Chapter 12, and the relevant figure numbers are cited. All reports were compiled using the annual detailed notes and catalogues prepared by Dr Lyne. A total of 29,206 sherds of pottery weighing 299,843g was retrieved from 826 excavated contexts; this included small amounts of prehistoric pottery (Chapter 2), and postRoman and medieval sherds. A further 44 sherds weighing 64g came from sieved environmental samples. Dr Lyne identified and catalogued the material annually, enabling comprehensive spot-dating and contributing towards the ultimate phasing of the site. A fabric series was compiled by Dr Lyne (Table 11.1) and, where possible, fabrics were correlated with the National Roman Fabric Reference Collection (Tomber and Dore 1998). Detailed Excel catalogues of all fabric types are held in archive, together with Dr Lyne’s identification/spot date catalogue.

Central Gaulish (Lezoux) products dominate the assemblage and are largely of mid to late 2nd century date, with the most common forms being Dr 31, 31R, 18/31 and 38 (23 MNV). Decorated vessels are represented by Dr 37 (12 MNV), mainly rim sherds with ovolo only surviving. In addition, a single Dr 45 is present. East Gaulish wares, comprising six sherds from Rheinzabern and Trier, are represented by just three vessels, a Drag. 31, Drag. 45 and Curle 21. Discussion This is a relatively small samian assemblage which only starts to reach the site in any quantity from the mid-Flavian period. The small number of pre-Flavian forms could have come to Druce at the same time as the Flavian material. The occupants clearly had access to samian in the later 1st century and, in this respect, the assemblage is in stark contrast to the nearby villa at Dewlish where 1st century samian is scarce, represented by only two sherds (Mills 2021, 160). It appears that much of the Central Gaulish, Lezoux material at Druce belongs to the mid to late 2nd century. The very small quantity of East Gaulish products is similar to the limited supply pattern at Dewlish (ibid.).

Imported finewares The Romans introduced new dining habits and, with these, new ranges of foodstuffs and specialized tablewares for eating, drinking and serving. These are considered below by specific ware type. Samian Mark Corney: initial work by Peter Teekamp In total, 363 sherds of samian pottery weighing 2,310g were recovered, with a low average sherd weight of 6.4g. This is a very small assemblage overall, representing just 1.24% of the total ceramic collection by number, and 0.78% by weight. Full details are on Excel spreadsheets in the excavation archive.

Potters’ stamps Five sherds have incomplete stamps and the following two are illustrated.

The assemblage

Fig. 11.1 1. Form uncertain, possibly Dr 18. South Gaulish, La Graufesenque (LGF SA). Part of stamp within an ansate frame; surviving letters [ ]VR[RA]И with the N reversed. Possibly Murranus, recorded at La Graufesenque 65-75 CE; a similar stamp was recorded at Wroxeter (Dickinson 1999, 127). Murranus was

The assemblage is dominated by Central Gaulish products (233 sherds weighing 1,337g) followed by South Gaulish wares (124 sherds weighing 909g), with only six (64g) East Gaulish products. The 1st century vessels are all from the main production centre at La Graufesenque and 197

The Rise and Decline of Druce Farm Roman Villa (60-650 CE) Table 11.1. Druce pottery fabric codes with NRFC equivalents Ware

Druce Fabric NRFRC Fabric

Imported Wares Samian South Gaulish

F1A

LGF SA

Samian Central Gaulish

F1B

LEZ SA 2

Samian East Gaulish Rheinzabern

F1C

RHZ SA

Samian East Gaulish Trier

F1C

TRI SA

Central Gaulish Black-slipped ware

F2

CNG BS

Moselkeramik Black-slipped ware

F3

MOS BS

Central Gaulish Terra Nigra

F11

CNG TN

Argonne Colour-coated ware (Sinzig roughcast)

F14

ARG CC

Lyon Colour-coated ware

F18

LYO CC

Rhineland White ware

F27

RHL WH

Cream Gauloise 4

A1A

GAL AM1

Pink Gauloise 3

A1B

GAL AM1

Baetican (Early)) amphorae 1 DR 20 fabric (