Excavation of an Enigmatic Multi-Period Site on the Isle of Portland, Dorset 9781407306162, 9781407321646

This report involves a multi-period site in a corner of the large sports field of the Royal Manor Arts College in Weston

201 15 37MB

English Pages [155] Year 2009

Report DMCA / Copyright

DOWNLOAD PDF FILE

Table of contents :
Front Cover
Title Page
Copyright
TABLE OF CONTENTS
CONTENTS LIST OF CD
ACKNOWLEDGEMENTS
ABSTRACT
PART 1 INTRODUCTION
PART 2 SITE EXCAVATION REPORT
PART 3 BIOARCHAEOLOGY AND SUBSISTENCE
PART 4 BUILDING MATERIAL, ARTEFACTS AND OTHER FINDS
PART 5 ASSESSMENT OF THE SITE AND FINDS
BIBLIOGRAPHIES
FIGURES
Recommend Papers

Excavation of an Enigmatic Multi-Period Site on the Isle of Portland, Dorset
 9781407306162, 9781407321646

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

l na tio ne di nli ad l o ith ria W ate m

BAR  499  2009   PALMER  

Excavation of an Enigmatic Multi-Period Site on the Isle of Portland, Dorset

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND

Susann Palmer with Denene Reilly

with contributions by Joanna Bird, Malcolm Lyne, Christopher Sparey-Green, Mark Maltby, Michael Allen, Debra Costen, Jane Yeo, David Ashford and David Dungworth

BAR British Series 499 9 781407 306162

B A R

2009

Excavation of an Enigmatic Multi-Period Site on the Isle of Portland, Dorset Susann Palmer with Denene Reilly

with contributions by Joanna Bird, Malcolm Lyne, Christopher Sparey-Green, Mark Maltby, Michael Allen, Debra Costen, Jane Yeo, David Ashford and David Dungworth

BAR British Series 499 2009

ISBN 9781407306162 paperback ISBN 9781407321646 e-format DOI https://doi.org/10.30861/9781407306162 A catalogue record for this book is available from the British Library

BAR

PUBLISHING

i

ii

CONTENTS ACKNOWLEDGEMENTS ........................................................................................................................................ vii ABSTRACT .................................................................................................................................................................ix PART 1 INTRODUCTION 1.1 Introduction. ........................................................................................................................................................... 1 1.2 Environment .......................................................................................................................................................... 3 Land shells by Michael Allen ....................................................................................................................................... 6 PART 2 SITE EXCAVATION REPORT 2.1 Excavation strategy.................................................................................................................................................. 8 2.2 General stratigraphy............................................................................................................................................... 10 2.3 Structures, Features and burials. With Denene Reilly ........................................................................................... 11 PART 3 BIOARCHAEOLOGY AND SUBSISTENCE 3.1 Marine molluscs .................................................................................................................................................... 24 3.2 Bones: mammals; birds and fish by Mark Maltby ................................................................................................ 27 3.3 Grain by Debra Costen .......................................................................................................................................... 44 PART 4 BUILDING MATERIAL, ARTEFACTS AND OTHER FINDS 4.1 Building material -Stone ....................................................................................................................................... 45 4.2 Building material Ceramic: Ridge tiles by Jane Yeo ............................................................................................. 48 4.3 Samian Ware by Joanna Bird and Malcolm Lyne ................................................................................................. 51 4.4 Pottery – coarse ware and fine ware (other than Samian) by Christopher Sparey- Green .................................... 54 4.5 Lithic artefacts - prehistoric and later .................................................................................................................... 65 4.6 Metal objects ......................................................................................................................................................... 69 4.7 Coins by David Ashford ........................................................................................................................................ 70 4.8 Jewellery ............................................................................................................................................................... 71 4.9 Glass ...................................................................................................................................................................... 74 4.10 Kimmeridge shale ................................................................................................................................................ 74 4.11 Fuel Ash Slag by David Dungworth .................................................................................................................... 75 4.12 Charcoal ............................................................................................................................................................... 75

iii

PART 5 ASSESSMENT Conclusions ................................................................................................................................................................ 76 BIBLIOGRAPHY ....................................................................................................................................................... 90

iv

CONTENTS LIST OF CD APPENDICES to sections of the text (compiled by members of A.P.A) 1. Context Record 2. Small Finds Details 3. Lithic Small Finds 4. Jewellery 5. Metals 6. Glass Special Finds 7. Miscellaneous Stone and Clay Special Finds 8. Building Material 9. Medieval ceramic tiles 10. Red Mudstone 11. Shale 12. Fuel Ash Slag, Magnetic and Vitrified Material 13. Charcoals 14. Molluscs 15. Pebbles 16. Soil samples TABLES Bones: Section 3.2; Mark Maltby 1. Contexts 2. Animal bones recorded by group 3. Preservation Indicators 4. Mammal bones recorded by group 5. Totals and percentages of major domestic mammals by group 6. Percentages of cattle, sheep, goat and pig from R.B. sites in Dorset 7. Sheep/goat element counts 8. Sheep/goat body areas represented 9. Elements displaying cut, chop and saw marks 10. Cattle, sheep, goat and pig mandibular tooth ageing data 11. Epiphyseal fusion data 12. Common measurements of sheep and goat from Weston Road and Dorchester 13. Cattle element counts 14. Pig element counts 15. Element counts of other mammals 16. Bird bones recorded by group 17. Bird element counts 18. Fish bones recorded by group 19. Fish element counts Grain: Section 3.3; Debra Costen Table 1 Database Samian Pottery: Section 4.3 by Joanna Bird and Malcolm Lyne Appendix 1: Fabric Coding (M.L.) Table 1: Incorporated in main text Table 2: Catalogue of all Samian Pottery - Coarse Ware and Fine Ware (other than Samian) Section 4.4 by Christopher Sparey-Green Appendix 1: Context dating from pottery Appendix 2: Dorset Black Burnished Category 1 v

Tables: 1 Total number of sherds (incorporated within text) 2 Roman pottery, all Roman phases 3 BB1 Fabric Typology Early Contexts 4 BB1 Fabric Typology Late Roman Contexts 5 Dorset BB1 Forms Early Roman Contexts 6 Dorset BB1 Forms Late Roman Contexts 7 Mortaria (incorporated within text) 8 Amphora and Flagons

Please note that the CD referred to above has now been replaced with a download available at www.barpublishing.com/additional-downloads.html

vi

ACKNOWLEDGEMENTS Thanks must first go to the Local Heritage Initiative, supported by the Heritage Lottery Fund, Nationwide Building Society and the Countryside Agency for the grant of £25,000 which enabled the project to take place. The whole project was later transferred to the supervision of the Heritage Lottery Fund. The project became possible as a result of the continuing support and interest of the Principal, Mr Paul Green, and staff of the Royal Manor Arts College and the owners of the field Dorset County Council: Messrs Andrew Price, then Head of Planning, Steve Wallis of Archaeology Services, John Lowe, Malcolm Turnbull and other officers of the Environment Services, the Education Services, and Fred Wellstead of the Property Management Division; thanks also to the Council’s Cabinet and Councillors for their decision to relocate the proposed all-weather sports field to another part of the site so that excavations could take place of the area which had been uncovered by JCB during preliminary work. Much gratitude is also due to the following for support which helped to secure the project and who also showed interest throughout the project: Mr Simon Williams of Weymouth and Portland Borough Council with its Councillors and staff, Trudie Coles, then an employee at the Council; Portland Town Council; the Council for British Archaeology; the Editor and photographers of the Dorset Echo with its local reporter, Hilda Swinney for constant interest and publicity; Julian Richards for Radio publicity.; Edward Griffiths of the Dorset magazine for publicity and a set of photographs. A substantial number of professional archaeologists and specialists gave their unstinted support and their knowledge helped in a great measure to add to the interest of the work and this report; Citizen’s Advice Office for advice on legal queries which arose. Many thanks to the Trustees of the Isle of Portland Heritage Trust for renting us a comfortable workroom and storage space throughout the period of the project. A number of ‘outside’ specialists in different fields contributed immensely to the quality of the research: A.C. Archaeology Peter Cox (reading the report) and staff, particularly Joe Whelan, John Valentin and Moira Laidlaw; Bournemouth University: Dr John Gale (senior lecturer), Drs. Mark Maltby and Sheila Hamilton-Dyer (bones); Drs Joanna Bird and Malcolm Lyne (Samian pottery) and Christopher Sparey-Green (Romano-British and medieval pottery); Dr. Michael Allen (land snails), David Ashford (coins); David Dungworth (fuel ash slags), Dr Debra Costen (grain), Carol O’Hara (site plans); Janet Bartlett (illustrations of metal objects). Staff of the Dorset County Museum and staff of the Dorset History Centre, Dorchester and Weymouth libraries generously gave their time to assist A.P.A. members with queries and research. The project was undertaken by members of the local community and members of the Association for Portland Archaeology and its director Susann Palmer, a retired museum curator and archaeologist. She was ably and constantly assisted by Denene (Dee) Reilly who acted throughout as Site Supervisor, main work manager and assistant, also in the post-excavation period and helping in the preparation for publication. The majority of the excellent photographs were taken by her. Mike Osmond fulfilled the difficult role of holding the purse strings and assisting in the reports which had to be made to the fund providers. Janet Harris acted as finds assistant while the excavations were in progress. Colin Cade, Stan Wedlake and Dudley Cooper helped regularly with the excavation work and processing of finds after the excavations were completed. Stuart Morris was responsible for the wonderful aerial photographs of the site, other photos of finds, providing answers to questions about history and helping with computer work on the report. Jane Yeo researched the medieval glazed ceramic ridge tiles and wrote a short contribution about them for this report. Dr David Hargrave examined some of the skeletal remains in situ. Edith Cory-King helped with archival and bibliographical research and read and corrected the draft report. Rod Wild helped with the excavation work and completion work on the project. Martin Blundell (A.P.A. Secretary) and Viv Kimber (committee member) helped with the excavations, ‘office’ work and organising events. The following volunteers all worked on the excavation project or afterwards in different capacities, for shorter or longer periods. Without their hard work, the project could not have progressed. Many thanks to each and everyone: Roger Aidulorth; Meredith Andworth; Deborah Bull; Glyn Bartholomew; Brenda Beaucanon; Brenda Beaumont; Sylvia Bradley; Sophie Braund; Tracy Braund; Beryl Bray; Victoria Bristow; Rick Caistor; Catherine Clark; Carol Cooper; Victoria Cole; Greta Coltart; Jane Dale; Ann Davis, John Davis, Mary Denley, Stella De Villiers, Nick Dowling, Hazel Dunning, Kevin Eames, Robert Edwards, Susan (Suan) Emptage (reconstruction drawings), Nigel Ereerd, Liz Fawkes, Tim Fawkes, Robert Fitch, Michael Flame, Jane Flame, Ruth Flame, Peter Glover, Christopher Harpley, Leslie Hitchings, Will Hodder, Meryl Hodder, Natalie Howes, Gill Jennings, Mode Kourla, Theo Kourla, Victoria Keating, Michelle Fenner-Keating, Laura Kimber, Jo Kimber, Tracy Lambrecht, Rhiannon Mann, Pat March, Val McGowan, Andy McLaughlin, Stuart McLeod, Emily McLeod, Peter Montague, Anders Nomin, Jesse Palmer, Nicholas Pilszack, Genevieve Poitier, Oliver Reichmann, Jennifer Rifer, Natasha Roberts, Jake Rowland, Liam vii

Rowland, Michelle Rowland, Sarah Rumble, Michelle Rumble, Barbara Sergison, Julie Shauhnessy, Lucy Shauhnessy, Glenda Sheppard, Julie Shields, Carol Simpson, Samantha Spracklen, Linda Stewart, Jan Stakes, Philip Stakes, Ian Stott, James Talbot, Jean Tatlor, J.M. Traves, Ross Watkins. Lastly, many thanks must go to the Portland Museum Trust and the Isle of Portland Heritage Trust for offering to make it possible to keep the large assemblage of material excavated from the site on the Island for the benefit of the local community and all future interested people who wish to study our rich heritage.

viii

ABSTRACT This report involves a multi-period site in a corner of the large sports field of the Royal Manor Arts College in Weston Road, Portland. Excavation took place following the proposed development of an all-weather sports field, which was shown to contain many structures and other remains during preliminary assessment work by commercial archaeologists. A Local Heritage Initiative Grant (later Heritage Lottery).was obtained by the Association for Portland Archaeology to investigate the site at a more academic level than was possible under the requirements of PPG16. The on-site work took place over a period of about 15 months. A large number of features and a very large quantity of finds were revealed. Specialists in different fields have contributed to the study of the main categories of finds and numerous photographs and drawings give a clear indication of the interest of the site and its assemblages. The major problem of dating structures on the site was the fact that the residents of all periods represented, all used the same materials - limestone and lime plaster - in much the same manner. They almost certainly also ‘recycled’ materials of previous periods and possibly even utilized foundations and walls of earlier periods on which to build new structures on. The following periods were recognised with a good degree of certainty: Mesolithic There is a small scatter of chert artefacts with a Mesolithic aspect all over the site, but no evidence of a definable site of this period. The presence of cores with the débitage does, however, suggest that some knapping may have been done in the area. Late Neolithic/Bronze Age There is a substantial presence of lithic material of this period all over the site. This consisted, rather unusually, mainly of débitage, utilised flakes and a quantity of well-made scrapers. A very large quantity of beach pebbles, some utilised as hammer stones, chopping tools or polissoirs, have been found. There were no structures of this period. Iron Age A small quantity of typical pottery sherds of the Late Iron Age have been found scattered on the site and it is possible that most of the crouched/flexed burials on the site belong to this period. It is possible, but not certain, that some of the later limestone structures incorporated utilised building material or even features of this period. Roman or Romano-British period The majority of the many finds and structures on the site belong to this period which appears to cover several phases, starting with the first century AD up to approximately the fourth century. The vast quantity of pottery consists mainly of black-burnished ware, some New Forest and other British made ware. The character of the finer ware, such as Samian, is predominantly continental and includes some exceptionally rare examples, including black-slipped Samian from central or eastern Gaul. A number of clay and stone spindle whorls have been found. Metal finds include one or two rings, bracelets, pins and fibulae of the first/second centuries, including a rare one of a tiny duck inlaid with enamel, many hob nails, building nails and numerous pieces of other metal objects. Only four coins were found, all badly eroded. Pieces of worked limestone are present in all parts of the site. Saxon period The evidence for this period is meagre. No clearly classifiable remains have been identified, but a quantity of pottery sherds may possibly be of this period. Medieval Period In the south-eastern part of the site, the remains of a large stone rectangular building were very prominent. A quantity of finds associated with the top levels are all of a medieval nature, including green-glazed and brown –glazed ceramic ware and coloured ridge tiles of thirteenth/fourteenth centuries date, all of good quality. A few trial trenches indicated that this overlies complex earlier structures. THE MAIN STRUCTURES ON THE SITE The structures on the site were built during several phases in the main periods but possibly not with a great lapse of time between them. There are walls overlapping earlier walls in a few places. Many finds were associated with all the structures.

ix

Round houses of the Roman /Romano-British period One of the most conspicuous features on the site is the foundations of a stone circular building with an internal diameter of about 6 metres and walls averaging nearly one metre wide. It could, therefore, have been a fairly substantial building. It has a square entrance area made of limestone slabs. The back (western) part of the wall is attached to a straight wall 0.7m wide, possibly forming part of an enclosure wall. There is evidence of fires having been lit in the building and a clay object was found, resembling a very tiny oven, perhaps for burning incense. Half of a second round building of the same size has been found at the southern edge of the excavation area, the other half still being under the grass verge round the site. In places the wall of this building is approximately a metre wide and it too was, therefore, a building of some substance. Fire-places or ovens Between the two round buildings is an area with two adjoining highly compacted clay and red burnt features, possibly the basal parts of one or more ovens or where a brazier had stood for a long period. Animal and bird bones and a quantity of burnt grain have been found next to the structures, including very rare bones of the extinct Great Auk and unusually large quantities of sheep. Near the burnt area is a small square limestone slab (approx 0.3m sq) which may have had a small column/statue on top of it. A small oval structure approximately 3m wide Adjacent the southern side of the furnaces is a small oval or pear-shaped feature. The floor of this structure was covered by several large flat limestone slabs. These slabs were above a stone lined gully. Its function was not apparent, but it can be surmised that it had been a place for drying meat, fish or even skin. Ditch A ditch approximately 2.50m wide and deep runs through at least part of the site from east to west. It has not been possible to explore the whole length of it. It contained quantities of romano-british pottery sherds and a few medieval, suggesting that the ditch may have been backfilled in the latter period. Feature with gullies and broken quern stones Immediately adjacent to the western side of the possible ‘enclosure’ wall of the round building S60, is another walled structure (301) which contained two short lines of gullies. Half a red sandstone quern-stone was wedged upright between the stones of each of the two gullies. Although enigmatic, this feature has the appearance of a meaningful structure and may have had a symbolic or ritual function. Miscellaneous features Many other structures and features of a Roman nature were partly revealed but could not be explored due to limited time. The Medieval Building The medieval building is also likely to have been erected over different phases. In several places it is clearly visible that the building overlies substantial structures of the Roman/Romano-British period. This building was not excavated as it would have required a great deal of time, but was only tested by a number of sondages. This structure is clearly dated by the fact that all finds found on the upper exposed level, are medieval in character. All along the northern wall glazed green/yellow/brown ceramic ridge tiles have been found - a certain sign of wealth and importance. Burials A number of burials of adults as well as babies have been found in close proximity to the above structures. Some of these burials are of an unusual nature and raise queries as to the nature of the site. A large number of burials were also found and excavated by AC Archaeology in the ground of the new sports pitch, west of the site forming the subject of this report.

x

Conclusions The excavated area forms only a very small part of a much larger site and all conclusions can only be tentative. All the specialists who contributed to the project are of the opinion that the site is a ‘most unusual one’. The main characteristics of the roman/romano-british remains are those of trade, ritual and wealth and it is suggested that during the main occupation period the excavated area was possibly the ritual or ceremonial area associated with a settlement primarily involved with trade in England and the Continent. It can also be suggested that during the medieval period the structure on the site was either of a religious nature or the abode of a wealthy influential person. Even though the details about the site are fragmentary, there are adequate indications that aspects of it are very unusual and it is a site which has the potential of adding greatly to our understanding of the archaeology of southern England and Dorset in general.

xi

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET 1.1

Countryside Agency. Excavation work on the site eventually started in July 2005. It was to be a community-based project under the leadership of qualified and experienced members of A.P.A. and supported by a number of specialists in particular subjects. The project was later transferred to the supervision of the Heritage Lottery Fund alone.

INTRODUCTION

The archaeological site involve an area covering approximately 44m by 24m, which is a small part of a very large sports field of the Royal Manor Arts College in Weston Road at SY68457170 on the Isle of Portland, Dorset. It is 300m south of St Georges Church and is at an elevation of 89.31m OD on a level terrace within a gently southward sloping terrain. The area had in the past been used for grazing cattle and in part for allotments, and had a wall round it until the 1970’s. The name ‘Court Barton’ has been associated with the field in the past and the Old English name Beretun indicates a corn farm or outlying grange (Mills, 1986/1996). It is thought possible that the name ‘Court’ may, in this instance refer to the fact that Portland was a Royal Manor and controlled by a royal court.

The object of the project was to excavate the site according to good academic standards so that it could be determined, in reasonable detail, what the significance of the site could be to the archaeological background of Portland and the wider region of Dorset and Wessex. Background to the interest of the site Since at least the 16th century it has been recorded that Portland had produced a vast quantity of finds of the Iron Age and Roman periods (Palmer, 1998 and 2005). This included jewellery, pottery, limestone coffins and such exotic items as imported continental iron ingots. Many of these finds were concentrated round the Verne area in the north of the Island where several older authors have maintained that there were clear features of a hill fort visible and which the building of the present prison (originally a military fort) has almost completely obliterated. Many burials have been found of Iron Age and Roman type in several places on the Island and there were many enigmatic structures. There have also been cursory reports of finds all over the Island of possible Roman buildings with tesserae or mosaics in Southwell and Fortuneswell (Palmer, op cit).

On October 11 2004 work started in the north-east corner of the playing field (opposite the entrance to the cemetery) to construct an all-weather playing field for the College. Archaeologists from AC Archaeology had been commissioned to do a site watch for a few days during the removal of the topsoil. It was thought unlikely that the area would produce much archaeological material, although the finding of a well, full of Roman pottery, had been reported in the vicinity ‘in the 1860’s (Palmer 2000, 2005). This suggests the possible presence of a substantial settlement site somewhere nearby. Aerial photographs of 1949 and 1954, kindly supplied by R. Perry and Stuart Morris, also indicated quite clearly that there were some or other remains in the field being the subject of this Report, as well as in the area to the south and also to the west (excavated by AC archaeology in advance of the construction of the playing field).

Despite all this, Portland has not featured in any major studies of the Romans in Dorset or Wessex and there have not been any systematic local excavations of this period, apart from a number of brief ‘rescue’ investigations when features showed up during development. This meant that it has up to now been impossible to get any coherent picture of what really happened on the Isle in proto-historic periods.

Immediately on removing the topsoil, it became apparent that there were extensive archaeological remains present with many walls, a round building, a large rectangular building and much pottery of the Late Celtic/Roman period (Romano-British). There was also a substantial scatter of Portland chert Stone Age artefacts, mollusc shells and Chesil Beach pebbles over the whole area.

In the 1860’s a local shepherd David White had discovered a well ‘full of Roman pottery a little south of St. Georges church’, i.e. at a place somewhere very near the present site, still unlocated. Excavation by some local antiquarians showed it to be ‘about 16 ft deep with a diameter of 2ft 6in (approximately 5m x 0.75m) and nearly full with Roman pottery’ (National Monuments Records 451697). This represents a large quantity of pottery and very obviously indicated that a settlement site must be somewhere in the vicinity. It also raised queries as to who threw the pottery into the well, and why and when.

At the instigation of the Association for Portland Archaeology (APA), and with the strong support of the Weymouth and Portland Borough Council, Town Council and the Council for British Archaeology and many members of the public, a campaign was started on the same day to delay or cancel the development in order to enable an intensive research excavation to take place. In December of the same year the Cabinet of Dorset County Council generously agreed with the Head of the Environment Department and the College that the sports field should be relocated further to the west on the field so that the local archaeological group could carry out excavation work in order to make an assessment of the significance of the discovery.

Geophysical examination of most of the playing fields has revealed some possible structures south of the present site, but has, generally, not been very helpful. On Portland this method of investigation is, unfortunately, known not to produce very good results because of the presence of the natural limestone beds, often only covered by very shallow brown-earth and loam deposits.

APA successfully applied for a grant from the Local Heritage Initiative Fund, supported by the Heritage Lottery Fund, Nationwide Building Society and the 1

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET NOTES:

Results are frequently also confused by the presence of natural fissures or irregularities in the rock seams.

1. Where no separate authorship is indicated the analysis of the finds and materials was undertaken by members of A.P.A. and the main author of this report (SP). 2. The numbers of tables and appendices for the special sections by named authors have been retained as given by these specialists in their contributions. Tables not included in the main text of their reports are on the CD enclosed with the report.

There are indications of the presence of possible archaeological remains on aerial photographs taken in 1949 and 1954, but the details are not clear as the photographs were not taken in conditions suitable for detailed archaeological research. However, recent aerial photos, taken of the current excavations by a member of APA from a helicopter in 2006, shows reasonably clear indications of structures all over the surrounding fields. Field-walking during the recent dry weather has also revealed possible areas with archaeological features. It can, therefore, be surmised that only a very small area (the north-east corner) of a large settlement site was available in 2005-2006 for investigation. A.P.A. was generously granted a licence to excavate the site for a year (extended to 15 months). There were also time limitations in connection with the terms of the grant and the work on the site depended on voluntary help from members of A.P.A. and members of the community. It was decided to complete removal of the topsoil by hand where the JCB had left off and to expose as far as possible the extent of features on the site. It was realised that it would be impossible and unnecessarily destructive, to attempt excavating, the whole available area fully and that it would be more productive to concentrate primarily on assessing the main visible features. As it became apparent that the large rectangular building was probably medieval and overlying extensive Roman remains, it was realised that it could only be satisfactorily researched during a long extended excavation and, apart from a number of small sondages, would best be left for future generations of archaeologists to explore more fully. While the group of volunteers worked on that part of the field which had been destined for the all-weather sports pitch, members of AC Archaeology started work on an area to which the proposed sports pitch had been relocated to the west of the site forming the subject of this Report. There they found a number of burials and other features (report pending). These archaeologists continued being very supportive of our work and we have permission to consider some of their findings too as is necessary to understand in the perspective of what was found on the main part of the site. It is to be hoped that at a future date it would become possible to explore the rest of the large sports field to the south and west of the present excavated area. All the comments and conclusions resulting from the present excavations, must be regarded as tentative as A.P.A. only had available for research the small north-western corner of a possible very large archaeological site with a great deal more evidence than it was possible to glean during a very limited time.

2

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET (Thomas in Palmer, 1999, pg110-111; see also Allen, this Report).

1.2 ENVIRONMENT Introduction

However, those samples were taken from the area of a medieval lynchet with one sample from the pre-lynchet horizon and another from the actual lynchet deposits, with one from the modern soil for comparison. There are marked differences between the contents of the three horizons. The pre-lynchet soil can be taken to correspond roughly to the natural environment at the time of the Weston Road site’s main occupation phase and that appears to have been supported, most probably, by short turfed grazing grassland.

The excavation of the Culverwell Mesolithic site at the Bill area of Portland provided an excellent opportunity to study the early Holocene environment of the Island (Palmer, 1999 and Thomas, 1999). The Weston Road site has provided the opportunity, albeit limited, of gaining some information about the local environment and possible land usage and human exploitation of the landscape during the first centuries AD. During the Mesolithic occupation of Portland the bioarchaeological remains from the early site indicated that the Island was almost certainly far less devoid of vegetation and natural running water than today. Marine shellfish dominated the economy and the collection of edible plants and the occasional small wild animal and bird probably constituted the limits of the terrain exploitation. The majority of shellfish were gastropods from rocky inter-tidal zones although diverse species from sandy and muddy shores were also abundant, particularly in some of the earliest levels. Mussels (Mytilus edulis), although a rocky shore species, was conspicuously absent, although found on other Mesolithic sites such as Westward Ho! Only one artificially pierced specimen of scallop (Pecten maximus (Linnaeus) was found in a feature regarded as of a ritualistic nature.

Historical and circumstantial evidence indicates that Portland may in the past have had a much better natural water supply than today. The village of Fortuneswell used to have two wells, Southwell had at least one well and a stream which fell over the cliffs in the form of a waterfall, south of the present houses, and Weston had a pond just a short distance south of the archaeological site forming the subject of this report. It is known that a well, full of. Roman pottery was found in the late nineteenth century somewhere in the vicinity of the site, not yet rediscovered (Palmer, 2005). Several stone-lined drains were excavated on the site and a whole series of such drains were found in the adjacent burial ground excavated by AC Archaeology (see Part 1). This indicates clearly that there was a supply of running water on the site during occupation. This may, in fact have been at least one of the reasons why this area was popular for occupation during several periods of the past.

Fresh water was obtained from the nearby Culver Well stream and the indications were that a palaeo-gully running through the site, probably did not contain running water but was muddy at times. The land and freshwater molluscs indicated a shade-loving and shadecompatible environment with the lesser occurrence of a wetland species (Anisus leucostoma) indicating areas with higher water content.

The local Landscape during the Site Occupation Periods The Weston Road site is within a short distance from the cliffs, towards the northern end of the Isle of Portland, at an elevation of approximately 100m O.D. It must be noted that Portland is really a peninsula, attached to the mainland at Abbotsbury by the Chesil Beach, but to the ancient inhabitants it would have appeared as an Island, as it is cut off from the mainland by the water of the Fleet, running through a narrow gap (now at Ferrybridge). The area of the Mere would here have provided excellent facilities for landing boats. Anybody standing on the site anciently would have had the advantage of clear views to the sea to the west over Chesil Beach towards West Bay and to the whole of the south towards the open sea. Other research on the Isle has shown that there would have been at least one strong stream, running south, near the site in the Reap Lane area (Bellamy, personal communication).

Unfortunately no demarcation between the prehistoric and later Roman and Medieval deposits could be made on the Weston Road site as a result of human exploitation of the same terrain by different occupants (see Part 2). No contexts were encountered which could be assigned to any specific one period only. This makes a comparative study of the environment of the different periods of site usage virtually impossible and any assessment is tentative. Area Assessment Molluscan and other environmental evidence along the Dorchester by-pass indicates that the predominant scene during the Roman and Medieval periods was that of open country with grasslands and interspersed with a localised riverine pattern (Allen in Smith, Healy, et al 1997). This may well have also been the situation over much of Portland during the main periods of site occupation as the analysis of samples of land molluscs from a post-midden (i.e. post Mesolithic) profile at Culverwell also indicated an environment of open country and dry grassland

Mesolithic period No environmental or other evidence was found during the excavation of the Weston Road site which can be definitely related to the early prehistoric periods and any comments about that is extrapolated from data from the Culverwell site. That site was occupied during the Late

3

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET Boreal-Early Atlantic (Holocene) climatic period, about 8,300 years ago (Palmer, 1999).

The Iron Age and Romano-British/Roman Period Although never tested by excavation, there appears to be adequate evidence in the form of banks and ditches to show that the Verne was occupied by a hill fort, destroyed by the building of the Verne fort (now a prison) in the late nineteenth century, quarrying and other developments (Palmer, 1998; 2005). A great volume of finds has been found to indicate that the Verne and the whole area round it, was fairly densely occupied in prehistoric and probably also later periods.

The actual waterfront at the Bill would have been between 100 – 200m further out. The climate would have been about two degrees warmer than today. The main exploitation for food was around the rocky pools and around the cliff tops with the occasional foraging further inland and to the Fleet, then a marshy freshwater area. Edible wild plants grew amply in the fields along the cliffs. The rising sea only broke through into the Fleet at about 5,300 BC when it would have become necessary to use boats.

As there is a paucity of military equipment amongst finds from the Verne, and Portland in general, it could be suggested that although the Verne had defensive/protective features, it may have been a place more in the nature of an oppidum or protected settlement site such as was common on the Continent during this period.

Although connected at Abbotsbury to the mainland by the Chesil Beach or Bank, Portland would, to all intents and purposes, have appeared as an island to the early settlers. The landscape would have been dominated by the great heights of the area now known as the Verne, approximately 135m above sea-level. It would have been very impressive from the sea and from the land to the north (now Dorchester area) and from the coast to the east and west. Some stone extraction for the purpose of hut building would already have started from exposures round the cliff edges.

The stone huts of habitation areas, both of earlier prehistoric periods and the native Iron Age people, would have been scattered all over the areas round the Weston Road site and evidence for this has been found nearby in quarries. Added to this, would have been the burial sites of the Romano-British people during the main occupation period of the Weston Road site, as revealed during the excavations and development work there.

During the subsequent post-Atlantic climatic phases, the climate developed to become much as during the other phases when finds indicated activity on the Weston Road site.

All round the coast there were massive exposures of limestone, with overlays of rubble, suitable for easy usage as building material. Some of these exposures would also have been exploited by limited quarrying to build, at least, some of the homes and public buildings on the Island, as well as for stone to make coffins for the richer members of the population.

Later Prehistoric Periods (Neolithic/Bronze Age) No definite habitation sites of the later prehistoric periods on Portland have been excavated, with the possible exception of some of the features revealed during assessment work done by Terrain Archaeology in the Reap Lane area of Southwell (Bellamy, op cit). An increasing impact of man-made features would by now have been visible in the environment and would have affected the general ecology.

There were areas of woodland, particularly in the shady coves round the coast, where there were direct access points to the rocky coast, with large bird populations on the cliffs. Several streams flowed from the heights through the green grazing land and ended in waterfalls in the south.

Many archaeological finds made over the past centuries (Palmer, 1999) have shown that dotted all round the Island would have been clearly visible the remains of a sizeable number of burial mounds, round as well as some large long barrows (e.g. Kingbarrow, a short distance east of the Weston Road site), several henge monuments (e.g. nearby two in the Grove, see Palmer, 2005).

Much of the open grassland would have been used for subsistence purposes, particularly for grazing sheep and some of the field systems (which was intensively used even by medieval times all over the island) would, possibly already have been in existence to cultivate grain.

At least some of the enigmatic so-called beehive structures (possibly also tombs, some only a few hundred meters from the Weston Road site) may already have made their first appearance during these periods. Some of them contained small quantities of grain, which could have been either the residues of storage or even the remains of offerings. These structures almost certainly continued in use into later prehistoric times (as indicated by finds in them) and may have been visible still during the first centuries AD.

Landing places at Church Ope Cove on the eastern side of the island and the Mere in the north would have been areas of constant activity by boats coming from the continent with goods and pack animals carrying people and goods to Durnovaria and possibly beyond, The presence on the Weston Road site, and on Portland in general, of continental goods (e.g. Samian pottery) may well suggest trade and a great deal of diverse activities. The local native population became romanized in culture and technology. The large quantity of imported and good quality roman finds on the site, suggests that it is likely 4

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET that a strong element of actual Romans or Romanized Europeans may have been living on Portland Even so, the general character of the Island probably remained an insular one, which appears to have continued into postRoman times.

Soil Samples Sixty soil samples were taken from different places within 29 different contexts. Samples were assessed for the nature of the deposits, their contents and magnetic susceptibility, all of which details are noted in the Appendix.

The Weston Road site was on a flat area within a gentle southward sloping terrain, with an adequate on-site water supply, possibly some shady hedges and small gardens or short grass areas and surrounded by some cultivated fields and open grassland on which animals (particularly sheep) grazed. The cliffs would have had a large bird population.

All of the samples indicated that a very high content of human occupation debris was present in a matrix of clayey loamy colluvial soil of varying degrees of lumpiness and stickiness. The colours varied from dark brown to very nearly black in a few samples. Some of the black samples which were tested reacted positively to Hydrogen peroxide and the colour was, therefore, at least partly due to the presence of fine charcoal dust although it is likely that rotted organic material (not tested) may also have been present in some cases. None of the samples gave any clear indication of the activities which could have taken place in the sample area.

The Saxon/Medieval Periods Very little tangible evidence for the Anglo-Saxon and very early Medieval periods have ever been found on Portland, possibly because it has all been destroyed by the expansion of population and many activities by later people.

Seven samples from (199) came from the area of F192, the remains of a presumed furnace/oven in the area of S200. With the exception of one sample, the matrix of the deposits from this area examined was generally black finely granulated earth with not a very high degree of stickiness/cohesion and this was regarded as indicative of cooking or baking activities resulting in fairly ‘dry’ deposits. One sample was very lumpy with limestone fragments, red burnt clay and visible charcoal particles and tested positive for magnetic susceptibility. The duck brooch (SF 149) was found in the area of this sample. The three samples of dark brown clay from the ditch fill (101) were relatively free of visible occupation evidence, apart from some charcoal which could have washed in.

The Weston Road site has produced possibly slight evidence of the Anglo-Saxon period on Portland (see Sparey Green, on pottery, this report Part 4). The geomorphological and environmental scene would probably have been much the same as described above for the Roman period, with a likely expansion of quarrying, agricultural and fishing activities as well as even animal husbandry. Although some of the structures of the Prehistoric and Roman periods would still have been standing, probably in a ruined condition, – the majority may have been plundered for their stone and other material. There would have been a fairly rapid increase of activity and structures relating to the expansion of the Christian religion onto the Island. The Island by now had attracted the attention of nobility and became a Royal Manor. The Lord of the Manor would have demonstrated his authority by establishing monastic institutions on the Island. Although Portland features in the Anglo-Saxon Charters and in Domesday Book (see bibliography this report and in Palmer, 1999), it is difficult to make an estimate of population density as no settlement site of this period has yet been excavated.

A number of samples when tested with a magnet responded positively although there was no evidence of any metallic items or fragments. This response was almost certainly due to very fine residues of the Old Red Burnt Mudstone containing iron sulphide from the Kimmeridge clay in some of the adjacent structures (building material – Part 4.1). Environmental evidence from landsnails Some insight into the nature of the Weston Road site scene during the period of occupation may be gained from the analysis of samples of land molluscs from sample contexts. As with the marine mollusca, not all finds of land snails were kept because of the very fragmented condition of these assemblages. A list of all samples collected is included with the marine molluscs in an Appendix.

5

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET LAND (AND MARINE) SHELLS All assemblages are dominated by two of the three largest taxa; Cornu aspersum and Cepaea spp. These are catholic species enjoying a wide range of micro-habitats from grassland to open woodland, but include hedges and are synanthropic enjoying environments such as gardens and walls (Evans 1972; Kerney & Cameron 1979; Kerney 1999). Only one individual of four different species was recovered from the interstices of the shells. These included Pupilla muscorum, Discus rotundatus, Oxychilus cellarius and Trochulus hispida. Although individually these represent a range of micro-habitats from open country to shade-loving (cf. Evans 1972) they are all species typical of garden and synanthropic environments, and have been recovered, for instance, in urban Romano-British environments in Dorchester (Allen 1993).

by Michael J. Allen Large caches of marine shells were excavated and handcollected. Shells considered to be land snails were removed from these assemblages and four of these caches were provided by the excavator for comment. The handpicked assemblage is strongly biased towards the larger specimens in the British fauna (typically Cornu aspersum, Cepaea nemoralis, Cepaea hortensis and Arianta arbustorum), to the detriment of the other over 100 species in the British fauna. The assemblage recorded is, therefore, not representative of the former palaeofauna, further more the bias towards the larger species (Cornu aspersum and Cepaea spp.) are typically catholic or intermediate species (Evans 1972) that have few habitat preferences.

Caches of land snails; natural or anthropogenic? Methods The fact that the incidence of land snails amongst the marine shells was higher than elsewhere on the site begs a number of questions. The marine shells represent discarded waste from human collection. Do the caches of land snails represent food waste that was collected and discarded, or are they a natural phenomenon? The larger species are all certainly edible, especially Cornu aspersum (common garden snail or Roman snail) which are still eaten in France (petit gris). Examination of all apertures showed no damage potentially caused by removal of the snail for consumption, nor were any burnt.

In an attempt to recover other portions of the molluscan microfauna, each assemblage of hand picked shells were submersed in water with hydrogen peroxide to disaggregate and dislodge soil retained in the interstices of the shells and recover other examples of the palaeofauna – see Allen 1995 (Round-the-Down, Sussex) and Allen 1997 (Lewes Priory, Sussex). The shells were decanted on to a 0.5mm aperture mesh (sensu Evans 1972) and the larger species removed and identified, and the fine residue sorted under a x10 to x 30 stereobinocular microscope. All shells were quantified (count of apices or aperture) and identified following nomenclature outlined by Anderson (2005). The results are presented in table 1.

The larger species Cornu aspersum and Cepaea spp. hibernate in winter amongst loose soil, rubble and debris (Allen 1995, 16) and a proportion of the colony do not survive; this could account for these concentrations of shells, as indeed it did in loose ditches fills in Sussex (Allen 1995). The combination of this fact and the increased preservational qualities of the microenvironment created by the dumps of marine shell may also account for these concentrations.

The land snail assemblages The shells were collected from four caches described and phased as follows: ??IA/RB

Context 198

??IA/RB Romano-British

Context 209 Context 202

Romano-British

Context 189

fill within half of round building S197 oval stone structure above floor round building S60 deposit above limestone near furnace

Conclusions We can tentatively suggest that the very small and unrepresentative assemblage of the smaller shell species suggest a synanthropic garden and occupied environment with open ground, longer grasses (shade) and piles of loose detritus (rubble, stones, waste). The presence of the larger species reinforces this interpretation and although they may be the discarded food waste, they are more likely to represent a proportion of the natural population hibernating within piles of loose marine shells. Likelihood of preservation is increased here due to higher levels of calcium carbonate (marine shells), and the physical nature of the context.

Preservation of these shells has been enhanced by the local microhabitats created by the caches of calcium carbonate-rich marine shells (periwinkle and limpet; table 1). Assemblages of hand-picked shells are not representative of the palaeo-fauna and cannot provide detailed information about the local palaeo-environments (Evans 1972). Washing the large shells to recover smaller species from their interstices (cf. Allen 1995; 1997) produced few shells. Nevertheless, some useful comment can be made about these assemblages.

6

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET

Phase Feature Context LAND MOLLUSCA Pupilla muscorum (Linnaeus) Discus rotundatus (Müller) Oxychilus cellarius (Müller) Trochulus hispida (Linnaeus) Arianta arbustorum (Linnaeus) Helicigona lapicida (Linnaeus) Cepaea nemoralis (Linnaeus) Cepaea hortenis (Müller) Cepaea spp. Cornu aspersum (Müller) TOTAL

198

202

189

209

1 7 26 6 40

1 1 12 2 1 17

1 4 16 5 26

1 2 1 4 13 1 22

MARINE Littorina littorea - Periwinkle Patella vulgata - Limpet TOTAL

5 + 5

2 2

19 + 19

3 + 3

?IA/RB

Romano-British

Table 1. Land and marine Mollusca from sub-samples of hand-picked assemblages

7

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET All areas were excavated by hand to the archaeological horizon and planned at a scale of 1:20. Sample excavations were undertaken with resources aimed at those features which seemed most likely to contribute to the understanding of the site. All structures, features, fills and finds were recorded and photographed. Section drawings were drawn at a scale of 1:10.

2.1 EXCAVATION STRATEGY AND METHODOLOGY The topsoil of an area approximately 120 metres square was removed under archaeological supervision by members of AC Archaeology during a preliminary site watch. The evaluation defined the area as requiring additional archaeological investigation. A.C.A.’s reference for the site was AC 886. After that all archaeological features were excavated by hand, evaluated, drawn and recorded by the Association for Portland Archaeology with help from members of the local community (see Introduction). The site was given a new reference number RMF (Royal Manor Field) to denote its connection with a field of the Royal Manor College of Art.

A total of 299 context numbers were allocated. Numbers 10 – 92 (number 38 cancelled) were used by AC Archaeology. Numbers 100 –317 were used by the Royal Manor Field project. Five structures which appeared to represent buildings or parts of buildings were recorded; they were located within a compact area and did not appear to have been contemporary. Twenty features were identified, some containing structural elements, of which three were linear features and one which was a ditch. A number of miscellaneous features were also recorded and are described in the report.

The recording of the excavation was done using the single context system and each structure, feature, cut, layer and fill was assigned an individual number and, where appropriate, all numbers that had been assigned by AC Archaeology were used by the “Royal Manor Field” project. In some cases two context numbers have been given for the same structure or feature (e.g. S42/S299) where this would facilitate recording and analysis. Where this has occurred both numbers will be given.

Nine burials were uncovered: and partially excavated but not lifted: two multiple crouched pit burials, one crouched cist burial and six infants. Above one of the pit burials four disarticulated long bones were also discovered which may have been another disturbed burial. Although permission was obtained to excavate and lift the burials, it was decided not to excavate them so fully that they could be lifted because of the restraints of time and the lack of space to store skeletal remains in accordance with Home Office requirements. Burials were examined, measured, recorded and photographed after removal of top soil and as much soil as possible round them to reveal details, without disturbing the remains. Two days before the site had to be vacated by the excavators according to their contract with the site owners, a large probable cist burial, surrounded by substantial limestone slabs, was revealed near the northwest corner of the rectangular structure S40/299 but lack of resources prevented excavation to confirm its nature. A further eight unexcavated features of soil discolouration were recorded as being possible graves. All burials were backfilled and left in situ where found.

Within the Report text the context numbers will have the prefix S for structure, F for feature and double brackets only e.g. (189) will indicate cut, layer and fill. Single context recording meant that all significant artefacts and features were recorded individually at different stages of the excavation. There has been no attempt made to provide a detailed description of every context excavated in the body of this report, though this information is available in the project archives. The entire area of the site was set out with a 10 metre square grid. It was not intended to excavate fully the whole area available for excavation as it would not have been possible to do this to an acceptable standard within the time limitations placed on the excavators. It was decided, firstly, to clear as far as possible all remaining deposits which could be seen to overlie structures/features in order to obtain a reasonably coherent plan of the site topography and lay-out. As S60 was one of the most prominent structures revealed by the JCB and its outline was the most complete, it was decided to excavate that structure more fully by the quadrant method, leaving one quadrant intact as a guide, should there ever be another opportunity in the future to re-excavate the site. It was, likewise, decided not to attempt an intensive excavation of the large rectangular building S42/299, despite its obvious prominence and complexity, as it would have required a great deal of work to do it justice and too much interference in the present limited time, could have hampered further work at some future date.

Daily entries were made in a ‘Site Book’ and all new contexts were recorded on Context Sheets as soon as revealed. Special Finds (SF) were recorded when found and all other records kept regularly. Post-excavation strategy Some washing, marking, and recording of all categories of finds took place on site each day during the excavation work and this continued after the period of on-site work in the workroom. The Site Book and all documents were kept up-to-date. Finds were stored according to categories and contexts in plastic bags within cardboard or plastic containers suitable for museum storage. All finds were individually examined and recorded by the project director with the help of the volunteers and special assemblages were dealt with by specialists. 8

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET Geophysical Survey A geophysical survey was carried out for AC Archaeology on behalf of Dorset County Council, by GSB Prospectors Ltd of Bradford (Royal Manor School Report 2004/81). Archaeologists of The Royal Manor Field Project have kindly been given access to the report. The objective of the survey was ‘to locate and identify the nature and extent of archaeological remains that may be present’ within areas of the playing field which were being assessed for the relocation of the all-weather playing field after the JCB had removed the topsoil and revealed numerous structures in the research area of the present Report. As some excavation work had already commenced in the excavation area by the time of the geophysical survey, it would have been counterproductive to have included that in the geophysical survey and that survey, therefore concentrated on surrounding areas of the large field. In the interest of future research of the field, the summary of the results of the geophysical survey pointed out that: ‘extensive areas of magnetic interference had been identified and a scatter of small ferrous objects has been recorded. Despite evidence of past ground clearance and landscaping, a number of anomalies were identified that have cautiously been interpreted as representing building remains. If correct, they appear to extend southward over a well-defined terraced area.’ As the geophysical survey is not strictly relevant to the present report, it will not be included in this volume, but a copy will be deposited with the rest of the archival material.

9

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET It is known that the whole area of what is now the playing fields of the Royal Manor Arts College were used since later medieval times as grazing ground for animals and other agricultural activities, causing further landscape changes which would have facilitated the burial of the site’s archaeological remains so that no record whatsoever of them remained. Some areas of the ground were also used as allotments before and after World War Two.

2.2 SITE GEOMORPHOLOGY AND STRATIGRAPHY The excavated site is on a level terrace within a terrain of which the general topography slopes gently north-south so that sections through the over-burden above the archaeological remains to the north are considerably deeper than those in the south. This raises the distinct possibility that the surface gradient of the area had been increased by a combination of human activity and natural forces of soil movement after the terminal period of site occupation in antiquity.

It is suspected that the latest episode during the twentieth century which increased the depth of soil covering the northern part of the excavated site, was caused by landscaping of the whole area to allow for the lay-out of the cemetery, the playing field and the wide grass verge adjacent Weston Road. The overburden directly over the excavated area was probably also truncated by landscaping for the playing fields. A south facing sample section in the north-east corner of the site, immediately south of the track and adjacent to the site entrance, was recorded by AC Archaeology during preliminary soil removal by the JCB and they have given permission for their data to be incorporated into the following observations.

As the whole Island of Portland is overlooked by the hill of the Verne (more than 130m O.D.) there must have been a constant movement of soil to lower ground and ultimately into the sea during rainy episodes in the past. Some of this soil accumulated in natural fissures in the limestone beds and in a number of palaeogullies formed in late glacial or early Holocene times, leaving a scene of very undulating aspect.

Context 10

Depth 0.16m

11

0.120.19m

12

0.25m

13

Not known

Description Dark greyish-brown silty loam with thin band of limestone inclusions at base Grey-brown silty loam with band of limestone inclusions; and patches of yellowish brown clay. Dark grey-brown silty clay, fairly clear of inclusions in upper level, evidence of features in lower part White or light yellowish-brown clay with some inclusions at top

Interpretation Turf and topsoil brownearth -recent disturbance Subsoil. Possible early cultivated soil recently disturbed. Buried soil horizon covering incorporating archaeological remains

and

Natural deposit pre-occupation

On the site the natural yellowish clay level was recorded at a depth of approximately 0.70m below horizon (12) but below 0.45m towards the centre of the site. The overburden in this south facing section above the archaeological horizon (12) increased significantly to the west where it reached a depth of two metres or more in places and where there was additional evidence for more recent activities. Unfortunately the overburden above the medieval structure S42/S299 was removed by the JCB during the evaluation work without any recording of the stratigraphical details. The foundation walls would have cut through all or nearly all of deposits (12) and (13) in that part of the site. A number of sondages through and adjacent the walls of that large structure give a limited indication of stratigraphical details (see Part 2.3).

10

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET Underneath the layers removed by the JCB was a 0.20m0.30m deposit of dark brown silty loam with charcoal flecks and a high concentration of limestone inclusions in the east and north quadrants. In the western quadrant there was a concentration of limestone slabs, within a matrix of dark brown loam containing charcoal flecks and burnt clay. These random limestone slabs sloped west to east and had the appearance of tumble from a disturbed feature.

2. 3 SITE REPORT: STRUCTURES, FEATURES AND BURIALS with Denene Reilly Structure 60 Circular Building and Possible Related Features (Figures 2, 3 and 20 – 23) The circumference wall of structure 60 was revealed as soon as the JCB had removed the topsoil from the site and some of the apparently related walls also became visible. As it was apparently a prominent feature on the site, it was decided to excavate this structure as fully as possible. It was divided into quadrants and the north, east and west quadrants were excavated.

Following the removal of this layer in the east, west and north quadrants an occupation layer was exposed. This consisted of a dark reddish/brown silty clay loam with a few small limestone inclusions. Charcoal, burnt clay and burnt limestone was present throughout the fill. An area of highly burnt red clay (F166) in the centre of the occupation level could represent the remnants of a central hearth.

Structure 60 was a well-made dry limestone walled circular building with an entrance on part of the eastern circumference represented by a 1.05m gap. It lay towards the north-western limit of the excavation area immediately adjacent and to the north of ditch 22. Bonded by overlapping stones to the external western circumference was a contemporary north/south aligned wall (F33/F260). The interior surface had a floor of compacted clay, which had been laid directly onto the original ground surface.

A compact cream coloured clay/lime plaster floor (F128, F312 & F201) had been laid on the original ground surface. This level had been subject to burning and had areas of red/orange clay, burnt limestone and large quantities of charcoals imbedded in its surface together with pottery, bone, metal and other items., There was evidence that at least part of the compacted clay floor may have been covered with limestone slabs or other flooring material (243). There was an area approximately 2m x 2m of flat limestone slabs which averaged 0.30m x0.25m x0.05m on the southwest side of the northern quadrant which extended into the western quadrant. They were lying on top of the occupation level in the northern quadrant, with some of the slabs sloping at a downward angle west to east.

The circular wall had an average external diameter of 7.75m with an internal diameter of 6.0m and survived to a height of 2 courses of flat limestone slabs on the eastern circumference and 5 courses of slabs on the southern circumference. The wall had an average width of 0.85m and was built of horizontally laid limestone slabs, typically 0.30m x 0.25m x 0.05cm in size which was laid directly onto the original ground surface. The wall achieved its maximum width on either side of the doorway. Part of the wall on the southeast approximately 2m from the entrance had collapsed inward revealing a fill of cream coloured clay (F311). It is possible this could have been a grave (see below). There was no evidence for an eaves drip gully or drain.

Two features which were set into the clay floor surface were uncovered in the east quadrant. The largest of these features (F146) was approximately 2m west of the entrance. It consisted of 50-100mm roughly square pieces of grey limestone covered by pieces of burnt orange clay. The feature was pear shaped, 1m. in length and was 0.70m at its widest point narrowing to 0.30m at its northern limit.

Entrance Area A laid surface of natural and shaped reddish-yellow limestone slabs occupied the internal entrance, where some of the slabs showed signs of scorching. The entrance was not fully excavated but some evidence of stone built walls was revealed outside the doorway at both sides suggesting the possible existence of a porch or passageway. A large limestone slab (F75) measuring 1.02m x 0.89cm x 0.06m lay directly within the passageway.

In the northeast area of the northern quadrant was a smaller feature (F244) consisting of pieces of grey coloured limestone set into the floor of the building with a small orange coloured dome shaped unfired clay structure on top. Although in a very crumbling condition, the feature resembled a small collapsed oven in shape. It measured approximately 0.30m x 0.20m in area with part of a curved opening 0.15m in length on its eastern side. A small section was cut through the feature to try and determine its function. The fill (245) was a reddish brown clay loam with charcoal flecks throughout. Burnt clay and limestone pieces with pottery shards and burnt bone were present in the area around it.

Interior The interior of the circular building was excavated in 3m x 3m quadrants with the eastern quadrant being excavated down to the original natural ground surface. The northern and western quadrants were excavated to the floor surface of the building. The floor surface of the western quadrant had been disturbed by limestone tumble from a possible later feature (F261) (see below). The southern quadrant was unexcavated.

A band of lumpy clay/plaster with linear impression marks, approximately 10mm wide on one side and smooth with a black deposit on the other, was located on the floor surface around the interior face of the wall in 11

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET Large quantities of coarse ware pottery were present in the fill to the north of structure 60 together with many finds of metal, chert, shell, bones, shale and beach pebbles.

each excavated quadrant. A proportion of the clay/plaster (F211 & F213) in the west quadrant showed evidence of burning. In the northern quadrant on the southern side were two limestone slabs on the clay floor which had been burnt, Also in the eastern quadrant south of the entrance were two highly burnt features. These were 0.50m apart (F175 & F180) and they also showed in the north facing section of the unexcavated southern quadrant. The fill of these excavated features contained burnt clay, pottery, bone and charcoal.

Three features which may possibly be grave cuts were found within the enclosed area immediately north -of S60 (for details see Burials below). Feature 248 was aligned north/south 1.50m in length and 1m in width. Feature 237 was also aligned north/south 1.50m in length and 1m in width. Approximately 1m to the west of feature 237 was another possible grave cut (F311) which was partly under the collapsed southwest section of the wall of structure 60. Adjacent to this on the east side was a limestone slab (F247), vertically inserted into the soil. This slab, 0.25m in length and approximately 0.10m in width, had weathered and split into several pieces lengthways. This feature could have been a grave marker.

Wall of possible enclosed area round Structure 60 Abutted to the southern external circumference of the circular building was part of a curvilinear wall comprising limestone slabs, which had been cut by the later west/east ditch (F22). It is possible that this wall represents the remains of the northern circumference wall of a further possible round structure (F200) (see below).

STRUCTURE 197 CIRCULAR BUILDING (NO 2) Two metres to the east of structure 60 were the remains of a dry stone wall (F39/124). This north/south wall was 0.70m wide, 8m in length with two courses of limestone remaining. Its southern limit (F124) had been cut by the later east/west ditch (F22). It had also partially collapsed on its western edge above feature 237. The northern limit (F39) had been lost or possibly disturbed by a burial (F306). A break in this wall was visible and was aligned with the east entrance of structure 60.

This structure is a very well made round dry stone walled structure similar in size to S60 and situated at the southern edge of the excavation. Only half of the structure was visible and partly excavated, the remainder lying under the southern grass verge and not available for research. Underneath the layers removed by the JCB, between the east west wall of F17 and the southern baulk was a layer (129) of limestone rubble within a matrix of dark brown loam. The fill was 0.20m in depth and contained many finds of pottery, metal, shale, bone and mollusc shells. The wall of structure 197 was immediately below this fill.

To the west, a north/south dry limestone wall (F33/F260) was carefully bonded by limestone slabs to the external western circumference of the round building and was 10m. in length, 0.70m in width, with a surviving height of 0.22m (4 courses). A section approximately 5m. in length had been badly disturbed by another north-south wall (F261) immediately on its western side (see F301). The remaining southern part of the wall (F33) was approximately 2m in length and had been cut by a later east/west aligned ditch (F22). This wall extended northwards to the limit of the excavated area. A 3m length of wall (F260) at its northern limit had partially collapsed on its western side. It sloped downwards east to west and continued under the northern baulk of the excavation.

The circular building was constructed from horizontally laid limestone slabs averaging 25cm x 20cm x 5cm in size with up to five courses remaining and built directly onto the ground surface. The width of the wall varied with the eastern section being 1.25m wide and narrowing to 0.90m at the north. The limestone had a worn and weathered appearance, suggesting that it may have been exposed to the elements over a long period. It is also possible that the limestone slabs were reused from an earlier building. A section, approximately 0.50m in length, of the circular wall on the western side had been cut through by part of a later wall (F219),of unknown nature aligned east to west (see site plans). F219 apparently forms part of another structure of which the remainder is under the grass verge. Also abutting the western exterior of the circular wall of 197 were the remains of a north south wall (F317), 1.20m in length and 0.80m in width which had partially collapsed on its western side. The southern visible part of this wall 317 had also been cut by the later east west wall 219 but it originally may have continued further south into the area presently covered by the grass verge. It was not evident what relationship, if any, there was between wall 317 and the circular building S197 but it appears that as both were cut by wall 219, at least two phases of building were involved in this area.

Wall 33/260 on the western circumference of the structure and wall F39/124 to the east of S60 may represent the remains of a walled enclosure around the circular structure. Their southern and northern walls are unknown and may have been removed at some stage during the past. Interior of the enclosed area The deposit in the enclosed area was a dark/brown clay/loam with a large amount of limestone inclusions. This area was not fully excavated, but the limestone rubble had the appearance of having collapsed from the walls and structures.

12

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET that the internal diameter may have been approximately 5m and the wall approximately 1m thick. A gap on the eastern side of 1.50m could represent an entrance. This structure was built onto the natural ground surface.

The exterior edge of the circular wall on the north was immediately adjacent to part of another straight wall eastwest-aligned (F17). This short wall abutted both the northern circumference of structure 197 the south and another structure S59 (see below) to the west. As only part of it remained, it was not possible to determine whether this wall was contemporary with or earlier than the adjacent structures and the relationship between the three was unclear.

There were also the remains of an east/west-aligned wall (S207) outside the entrance on its northern side. Approximately 2.50m to the east, was part of a short north-south wall which may be the southerly continuation of wall S39/124 which is also east of S60. No made-up floor surfaces were recognised in the interior of the structure but a number of distinct deposits and features were uncovered in the interior (102, 207, 208,256 and 206 (see below).

Interior The topmost deposit (198) in the interior of the building was a silty brown loam with limestone inclusions which was 0.10m in depth and many finds of various kinds in the fill. Below this was a layer (216) of fine friable loam with burnt clay, charcoal and burnt shale present throughout the fill. An occupation/floor layer (218) of hard packed clay and lime, similar to that of structure 60, was exposed with areas of very highly burnt red clay with charcoal and burnt shale embedded in the surface. This burning was running diagonally northwest to southeast and continued under the grass verge. To the centre of the floor were limestone blocks (217) set into the clay. This feature was not excavated but had the appearance of the possible remains of a hearth. There was also a quantity of clay with impression marks similar to that found in structure 60 round the interior of the wall.

Wall 257 (part of the wall of structure 200). The dry stone curving wall was constructed of limestone slabs averaging 0.20 x 0.20 x 0.05 in size with up to five courses remaining. On the southern side the remains of a later east west dry stone wall (S17) 5m in length was visible sitting above the surviving wall of structure 200 with approximately 0.10m of the internal curving edge of the wall (S257) showing below. At the east end of the wall on the south side of the entrance gap was a limestone feature (208) roughly triangular in shape. Constructed from shaped flat limestone slabs with oblong blocks of limestone on its internal edge and running under wall S17. On further excavation, it became clear that the inside curving edge of feature 208 was part of wall S257. On removing some of the stone from feature 208, an infant burial (210) was discovered. Another two further infant burials (241, 253) were also discovered clearing back layer (205) against the internal edge of feature 208 and wall S257. A further fragmented skull of an infant (271) was uncovered against wall 257 directly south of F192. (These burials are further discussed below).

Structure 200 – Enclosure round highly fired/oven area: (Figures 5 and 24-31). Underneath the layers removed by the JCB between ditch 22 on the north and the northern circumference of S197 (the half round building) on the south, were 3 distinct deposits. The upper deposit was a layer of dark brown loam with limestone inclusions. Cutting through part of this deposit in this area, part of a west east linear feature (27) was visible (see Miscellaneous Features S27). Underlying this layer was a spread of limestone rubble in a dark brown silty loam with charcoal flecks throughout. The limestone rubble diminished in depth from 0.15m to 0.05m east to west respectively and was a probable demolition layer. A quantity of mollusc shells and pottery sherds, including Samian ware, were found throughout the fill. Below the rubble layer was another deposit of a fine dark brown loam (189) with charcoal, burnt clay and pottery sherds throughout the fill. The layer varied in depth from 0.05m in the east to 0.15m in the west.

Part of the wall on the south west 1.50m in length was sitting on the north east curving wall of oval structure (S59). The southern part of the wall had been cut by or abutted to another dry stone circular structure (197) (see below). It is unclear what the relationship of these two structures is to structure 200, as they were not fully excavated. The remains of the western circumference had been badly disturbed and partly cut by the ditch 22 on the north but the internal edge and some of the external wall was still intact. A number of limestone slabs were aligned with a wall S33 which may have been bonded to structure 60 but damaged by the 2m wide ditch 22. This suggests that wall S33 may have continued south to structure 200.

While removing the layer of loam, the remains of a curving wall on the western side of the structure became visible and were given the context number 257. After further excavation, it was realised that this wall was part of the very fragmented remains of a possible circular or oval dry limestone walled structure S200 to the south of structure 60 A small section of it was cut into by the southern circumference of S60. It was also significantly damaged by the cutting of an east-west-aligned ditch F22 on its northern circumference and disturbances to sections of the west and north east circumferences. Despite the uncertainty caused by the damage, it could be estimated

Feature 207 On the north side of the entrance gap to the structure were a number of limestone slabs in a broken row surviving to the height of a single course. They sloped downward south to north and had been badly disturbed and cut by the later ditch (22) into which the rest of this feature may have fallen. It was made from limestone slabs 0.20 x 0.20 x 0.05 with an approximate width of 0.70m

13

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET Structure 301 and Associated Features: (Figures 4 and 32-33).

Feature 192 Here a defined area of hard baked clay (F192) orange/red in colour, approximately 2.50m x1m in area, was also uncovered. There was slight evidence of a structural element with a curved and shaped edge at the southern end, suggesting the remains of an oven or contained fire utensil. A quantity of Samian ware was present in the fill (205) round and to the north east side of this feature.

After removal of the topsoil by JCB, it became apparent that there was another north-south wall immediately touching the western circumference of S60, the round building. (See site plans). This other shorter wall S261 had the appearance of being attached to part of the northsouth wall S260 which was bonded to the western part of the circumference of S60 (see S60), but the relationship was not clear and the two walls may merely have abutted each other. A short west-east stretch of wall S295, 2.10m in length and made from flat limestone slabs, formed a right angle with the northern end of S261. These two walls had the appearance of possibly forming the corner of a room or building adjacent the western exterior of structure 60 and its ‘enclosing’ walls.

A one metre square sondage was excavated through feature 192 and ditch 22 at its northern end. The burnt clay was 40mm to 50mm thick and was lying over a fill of limestone rubble (258) in a fine dark brown reddish loam. The later ditch 22 had cut through both of these fills. Fragments of a skull were uncovered against the interior of wall 257. This probable infant burial (287) was recorded but not excavated (see also ‘Burials’ below).

In the corner formed by walls S261 and S295 were three linear features running north to south. They consisted of two sets of double rows of vertical limestone slabs, approximately 0.25m wide, both approximately 0.8.m long and 1m apart from each other. They resembled gullies or water channels. They were set into the original ground surface. Three slabs of limestone were placed over the southern end of the eastern ‘gully’. Midway between the two double north-south rows was a single north –south row of limestone slabs, slightly shorter than the two outer rows, but also set upright into the soil. A few small broken bits of limestone lying by this third row, could suggest that it too may also have been a double row of slabs originally, but this is not sure.

Feature 256 Feature 256 was one metre from the southern internal edge of wall S257 consisting of two rows of upright limestone slabs, each approximately 0.60m in length with a gap of 0.80m between them and running east to west. They were sitting in a layer of fine dark reddish brown loam (205) with approximately 0.20m visible above the surface of the loam. The eastern set of stones had a limestone base on its northern side, triangular in shape and made from small square like pieces of limestone. These stones had been blackened by heat, suggesting that they were once on a single slab that had broken into smaller pieces. On the west side of the eastern upright stones, a small round hollow in the clay floor measured approximately 0.12m in diameter and 0.10m in depth.

The northern ends of the two double rows were connected by another gully-like feature running east to west, also 0.25m wide and also formed by two sets of upright limestone slabs. It was capped with small flat limestone slabs. This gully was approximately 2.50m in length. Together the three double rows (gullies?) therefore appeared to form the three sides of a rectangular feature with the open end to the south.

A 0.50m square sondage (255) was cut through layer 205 directly north of the eastern upright stones. The deposit below was the natural loamy colluvial clay of a very light cream colour found over most of the site below the occupation layers. There were occasional flecks of charcoal, derived from the upper deposits. There were no features visible in this deposit. The western set of stones was not excavated.

Both of the north-south gullies had incorporated into their structure, wedged between upright stone slabs, half of a rotary quern made from red sandstone and also set upright on their edges into the ground surface. The piece of broken quern incorporated into the western uprights had an oblong slot cut into it which was for the handle of the rotary quern.

Feature 206 Approximately one metre west of the end of feature 207 was a flat limestone block 0.50m square by 0.25m in depth with rounded shaped edges. The limestone feature (206) had been subjected to heat, was grey/pink in colour and had cracked into several pieces. The cutting of the later ditch 22 had slightly damaged its northern edge. A quantity of charred grain was associated with this feature as well as pieces of fine glassware.

The fill (302) around and between the stones was a dark brown clay/loam with limestone inclusions and with some pieces of wall plaster of fired/burnt clay with linear impression marks on them.

In the fills (189 and 199) around this feature several small finds were recorded: a duck brooch, glass beads, and more fine pieces of glass as well as pottery which included Samian ware, metal, beach pebbles, burnt shale. Other areas of burnt clay were visible throughout the fill.

This feature was not fully excavated and as the quern stones were in a very fragile state, they were left in situ.

14

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET Oval Structure S59: (Figure 5 and 34-25).

Ditch F22: (Figures 1 and 38).

A dry stone walled structure, approximately oval in shape, lay immediately southwest of structure 200 and northwest of structure 197. The wall was constructed from flat limestone slabs with an internal width of 3m east to west and 2m north to south. The wall varied in width and depth. A section of the curving wall approximately 3m in length on its western side was 0.80m in width at its northern end, narrowing to 0.70m at the southern end with two courses of stone remaining. It was constructed from limestone slabs with an average size of 0.30m x 0.40m x 0.05m.

Ditch 22 was located approximately 4.50m to the north of the southern grass verge, aligned west-east and was visible when the mechanical digger stripped the topsoil. This feature was more than 22m in length and 2.5m in width at the surface and continued beyond the western extent of the current excavation area. The eastern extent was unknown and had been truncated by wall S34 of a later possibly medieval feature (S42/299). The ditch was investigated in two sections. Members of AC Archaeology cut a short section through the ditch to a depth of 0.50m but it was not fully excavated. It showed this feature to be an average of 2m wide with the top 15cm a fill of dark grey brown silty clay (23) which was virtually free from inclusions. A layer of rubble (78) with a clay deposit (79) below was visible on the north-side of the ditch in a fill (24) of grey brown silty clay with very few limestone inclusions.

It curved eastward where it butted against a rectangular worked block of limestone with another large irregular block of worked limestone next to it. The south and east section of the wall had 6 courses remaining and were constructed from smaller limestone slabs than the west, averaging 0.25m x 0.20m x0.05m. The external southeast edge of the wall was cut by structure 197. The remains of another wall (S317) running north to south touched the southern exterior of structure 59 but it was not clear whether it was contemporary with it or earlier. (Wall 317 also appeared to be under the wall S219 discussed in the context of S197).

The section was extended and re-cut during a later phase of the original excavation. This north-south sondage (226) was 2.50m wide by 1.20m deep and contained five fills which was composed predominantly of dark greyish brown silty clay. The ditch had cut through earlier features 60 and 200. On the north side it sloped at roughly 45° to a depth of 0.90m then became vertical for a further 0.30m onto a flat base approximately 0.50m wide and cut into the natural clay. The cut on the south side had been stepped into the natural clay at roughly the same angle as the north.

The northern section of the wall S59 had less of a curve than the rest of the structure with part of the southwest wall of structure 200 sitting above it. A gap of 1m between the northern section of the wall and the internal edge of the western wall of structure 59 was visible and may have been an entrance. The remains of a north south wall which appeared connected to the exterior of the northern part of the structure, just east of the entrance, was aligned with wall S33 and may be a continuation of that wall which was bonded to structure 60.

The top fill (203) below the topsoil was a dark brown sticky loam with very few limestone inclusions. There was a spread of charcoal throughout this fill with powdery degraded limestone on the north side. Oyster shells, bone and a sherd of medieval pottery were present in the top fill.

Interior The interior floor surface of the structure had been laid with very large flat limestone slabs of which some appeared to be possibly conjoining but others had gaps of approximately a centimetre between them (F77). These slabs may possibly have been one slab originally which had been broken in antiquity. The floor was half sectioned for excavation and a number of the slabs were removed. There was a layer of dark brown loam (285) beneath the limestone slabs which was approximately 0.04cm in depth with evidence of burnt clay and charcoal on the southern side against the southern interior wall by the block of limestone. A small quantity of bones and pottery was recovered. This fill was removed to reveal limestone slabs (303) of various sizes which continued partly under the walls of structure 59 on the eastern side. Underneath the slabs was a fill of dark brown very sticky loam which was removed to a depth of 0.40m. A double row of limestone slabs running east to west and planted vertically into the loam, formed a gully like feature (F305) which continued under the eastern wall and western section of the slabs.

Below the top fill was a fill (215) of dark grey brown clay loam with very few limestone inclusions but with charcoal flecks throughout the southern side. The next fill down (222). had many angular pieces of limestone ranging from 5cm to 15cm in a matrice of dark grey brown loam This probably represents the deliberate back fill of the ditch with building debris and silting up of the ditch during and after this activity. The primary fill (225) at the base of the ditch was a sticky dark grey brown fine silty loam. The north and south sides of the ditch had a primary fill (224) approximately 4cm in depth of very fine dark grey brown loam with pieces of limestone present. Structure S122: (Figure 1 and 7). Structure S122 was revealed below a layer of dark brown loam (121) which was very extensive between wall S36 to the north and the later large rectangular structure 299/42 on the south. This fill included fuel ash slag, 15

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET tabular pieces were visible below the course of the wall. These burnt pieces were next to a spread of fuel ash slag, associated to the south with Feature 122. As the limestone slabs of the wall were not burnt at this point, the indication could be that the wall was later than the episode which caused the burning of the slag.

charcoal and many finds. Structure S122 was situated in the south eastern part of this space, 2m from the excavated site’s eastern grass verge. An area approximately 2.5m x 3m. of large limestone slabs, possibly forming a floor or some other surface, was uncovered. The limestone surface at its north east end was curved with evidence of burning.

MISCELLANEOUS FEATURES: (see Figure 1, Site Plan)

The bottom parts of four limestone pillars (S150 in the north west: S151 in the north east, S152 in the south east and S153 in the south west were set into the soil between the limestone slabs., The pillars were 1m apart at right angles to one another in the form of an open square. They were oval in shape with an average size of 25cm. x 30cm. The north east pillar had an even flat top surface with visible tool marks but the other three were uneven suggesting they their tops had been broken off.

Most of the following features were not excavated, except where indicated limited excavation took place but not enough to confirm their nature. Feature 290: (Figure 69). Six metres due north from Structure 42 the remains of two lines of limestone slabs were visible within layer (121c) set at right angles to each other. The east west line was 2.0m in length and constructed from flat limestone slabs with an average size of 0.25m x 0.05m x 0.30m and set vertically into the ground. At the west end of this line, the partial remains of a line of stones running north to south for 2.50m were visible and at the eastern end a few stones were lying. These stones suggest that they may have been a structure which enclosed feature 283 & 287 (see below). Three infant burials were uncovered next to the east-west line of stones of structure 290 on the south side but it was not clear whether there was an intentional or planned connection between these features (see Burials).

A west facing test trench (276) 1m in length was cut against pillar 151 and through the limestone floor running south. The cut by the pillar on the south revealed that it was 1m in height, had a diameter of 0.50m with sides slightly tapered to an uneven base at a depth of 0.52m. It was approximately 0.20m in diameter at its base. The hole in which it was, had been packed with large pieces of limestone in a fill of dark brown clay loam. Below the surface of limestone slabs was a fill (277) of sticky dark brown loamy clay with charcoal and burnt clay throughout. The limestone in the rectangular space between the pillars was highly burnt. Charcoal and 11gm of carbonised grain was collected in the fill (262) between the pillars (see Part 3.2 Grain).

Feature 283 – possibly associated with F290 (above) Inside the area which appeared to be possibly enclosed by F290, in the southeast corner against the line of stone in layer (121c), was an area approximately 0.50m in diameter of burnt fragmented clay and shale as well as sheep bones. They were lying over slabs of limestone which had been subjected to heat.

Structure 36 – Wall: (Figure 8 and 36). The base of a dry stone wall was located to the north of the complex of buildings on the excavated site and was aligned approximately east to west. It continued under the grass verge round the excavated site in the east and faded out and was lost to the west. The wall was 12m in length with an average width of 0.80m and had a number of curves in it. A maximum of two courses of flat slabs of limestone of the original wall remained.

Feature 287 The possible cut of a grave became visible inside feature 290 when layer (121c) was removed and below layer (291c). It was 1-50m in length and 1.0m wide and filled with a pale yellow brown clay. It was running from the northwest corner diagonally to the south east. The feature was not excavated.

A 1.50m sondage was put through the wall near the point where it disappeared under the grass verge. The cut extended 0.50m north and south of it. Immediately below the stones of the wall was a fill of clay loam with small limestone inclusions (269).

Feature 286 Grid Ref: 112.5E/108N One and three quarter meters south of feature 290 and 1.0m east of wall 39 an area of flat limestone slabs (286) was uncovered below layer 121c. It was 2.0m in length east to west and 2.25m north to south. The north side was edged with triangular shaped pieces of limestone which were in a straight line east to west. A large slab to the south on the east side, approximately 0.75 x0.75m had been shaped with a half round cut into its south west edge. It had sunk and was lying at an angle from south to north. Immediately to the south of this stone in a fill (293) of sticky dark brown humic loam a quantity of animal bones were found along with pottery sherds, pebbles and marine shell. A concentration of limpet

Adjacent the northern edge of the wall, the fill (270) was of limestone pieces in a medium brown clay loam which was probable tumble from the wall. Adjacent the southern edge of the wall, the fill (271) was of a dark brown loam with burnt limestone inclusions and some fuel ash slag and charcoal throughout. At the eastern end of the wall, by the grass verge, an area of 1.50m length of heavily burnt pinkish-grey limestone 16

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET clearing the topsoil. No stone lining was visible. The centre fill (47) of the feature was a rubble layer of small limestone pieces in a brown silty loam this was above a fill (48) of pale yellowish brown silty clay. This feature was not excavated.

shells, 0.50m in diameter and 0.20m in depth were also found in fill 293 1.0m to the south of the animal bones. None of these features were fully excavated. Feature 307: (Figures 66 and 67). Grid Ref: 118E/108N Immediately south of feature 286 was feature 307 which had the appearance of being a possible gully. Running north-east to south-west a section of two rows of vertical laid limestone slabs were uncovered below layer 121c. They were running under a layer of limestone slabs, some of which capped the vertical slabs at the south west end. The north eastern end appeared to terminate by a limestone slab laying flat across the width of the feature at its base, possibly being an end stone to the feature. The rows were 1.75m in length and approximately 0.28m apart from each other. They were set at a depth of 0.56m into a layer (291c)) of dark brown loam with limestone inclusions on to a base of natural yellow clay. The rows were constructed from flat limestone slabs of varying sizes and in some places, one on top of the other and overlapping each other. There was only one fill (308) between the rows of stones. It was a dark brown clay loam with limestone inclusions the same as fill 291c. A few pottery sherds, bone and shells were present in the fill. It was not fully excavated. Southwest of F307 at a slightly elevated position, a shell midden (316) was located.

Context 31 Grid Ref: 108.5E/117.5N This was a roughly circular feature of limestone uneven blocks with a diameter of 0.67m. AC archaeology recorded it as a possible post pit or post pad. As this feature was directly north of pit burial 117 it is also possible that it could be a grave marker. Context 32 Grid Ref: 109E/119N To the north of feature 31 at a distance of 1.25m was another roughly circular feature similar in construction to 31. This was half sectioned (105) west to east and found to be a probable post pit 32). It had a diameter of 0.63m and a depth of 0.25m with slightly sloping sides cut onto a flat base of natural pale yellow clay. The pit’s sides were packed with uneven small blocks of limestones with a horizontal flat limestone at the base of the pit in a fill (106) of dark brown crumbly clay loam. One piece of pottery was found in the fill. THE FOLLOWING FEATURES WERE VISIBLE WITHIN THE WESTERN UNEXCAVATED PART OF THE SITE

Feature 292 Grid Ref:120E/110N Running north west to south east and possibly cut by feature 287 to the north west and feature 307 to the south east was the remains of a dry stone wall. It had a rubble fill of small limestone pieces with larger limestone flat slabs on the outer edges. A 2.0m section directly north west of feature 307 and a 1.0m section directly north east of infant burials 282 & 284 was visible.

Linear feature 58 Grid Ref: 98E / 95N In the south west area of the excavated site, against the southern grass verge, below the Roman soil level, was a linear spread of limestone rubble (58) in a layer of dark brown silty loam (57) It was 10.0m in length, approximately 1.50m in width and was running approximately east to west and continued under the western grass verge.

FEATURES NORTH OF WALL 36 Context 28, 29, 30, 31, 32 Very little excavation took place north of wall 36 and the 5 features present had been recorded by AC archaeology after removal of the topsoil by JCB. Two of the features 30 & 32 were excavated further by A.P.A.

A few of the stones had been worked, suggesting the possible remains of a Roman wall within the rubble. The average size of the worked stones was 0.46 x 0.35 x 0.04m. It is unclear whether the eastern end of the possible wall (58) had been cut by S219 or was part of it. It is also possible that this feature was part of a very substantial wall or feature, more than 1m wide, which was revealed by AC Archaeology in a metre square trial trench just outside the south-west corner of the fenced area.

Context 28 Grid Ref 104E/107.5N A pale brown coloured rectangular feature aligned N-S 2.80m in length by 1.70m in width was uncovered when mechanically clearing topsoil. The machine had clipped the top of the feature revealing a stone lining (90) visible at its southern edge. In the centre of the feature was a rubble layer (44) of small limestone pieces in a brown silty loam. This was above a fill (45) of pale yellowish brown clay. This feature was not excavated.

A piece of quern stone was found in the rubble spread which was found to conjoin with another piece of the same quern, previously found on a spoil heap resulting from the initial clearance work on the site.

Context 29 Grid Ref: 118.5E/118.5N Another pale brown coloured rectangular feature 2.50m east to south east of feature 28 aligned N-S 2.90 in length by 1.20m in width was uncovered when mechanically

Linear Feature 26 Grid Ref: 96E/97.5N A linear band of pale yellowish brown silty clay with a quantity of tiny limestone inclusions was visible when the top soil was removed by machine. It was also running 17

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET soil which had few limestone inclusions, below which was limestone rubble.

west to east, 12.50m in length with a maximum width of 1.93m and was 2.50m to the north of feature 58. The western end continued under the turf line, the eastern end widened, curved to the south and was lost. It is possibly a robbed out wall or drainage feature.

Western wall S 34 The western end wall (34), running north south, was 6.8 m in length and 1.2m. in width and ten courses survived above the foundations to a height of 0.50m. It was constructed of dry stone walling which had been carefully and solidly laid from flat limestone slabs approximately 0.25 x 0.15 x0.05m. in size. The southern section of the wall had sunk and had partially collapsed, possibly because the ground was unstable because of the earlier ditch (22) which ran under the southern end of the wall.

Linear Feature 27. Grid Ref: 100E/101N This was another linear band of pale yellowish silty clay with tiny limestone inclusions 3.50m to the north of feature 26. It was also running west to east, 16.50m in length with a maximum width of 0.98m. A section (119) 0.50 meters in length was excavated through the feature. It was approximately 0.15m deep with sloping sides onto an uneven base. Two limestone slabs were placed onto the base of the cut with a quantity of smaller limestone slabs irregularly placed above. The pale yellowish silty clay fill (120) above the stones contained some pottery and bone.

Outside the northwest corner of the building was a large hand cut block of limestone, which was slightly damaged and may have been derived from an earlier building elsewhere and reused as a corner stone of the building S42/299. The interior surface of wall (34) of structure 42 was constructed from large uneven blocks of limestone four layers deep above which were five or six courses of dry stone walling. There was no visible bonding between the internal edge of the north wall (49) and wall 34. (But see below sondage 169).

Feature 81 A 2.0 by 1.0m sondage was put through layer 82 in the north-west corner of the excavation site. It revealed a spread of limestone slabs, some of them worked (81), with a maximum size of 32 x 21 x 4cm and a minimum size of 10 x 9 x 3cm. No edge or alignment of the feature was visible. STRUCTURE 42/299 (Figures 6 and 39-47).

MEDIEVAL

A sondage 1.20m wide was excavated through wall 34 in the northwest corner of the interior of the building. The uppermost layer (263), which was 0.47m in depth, was of redeposited demolition material of common limestone rubble fragments in pale yellowy brown clayey silt. Below this was a 0.14m in depth layer (264) of stony clay. Fourteen pieces of limestone roof tiles were found in (263) and 12 in (264), some with holes and some with mortar adhering to them (see Part Building material). Under these tiles was a very thin band of redeposited clay no more than 0.02m in depth. This was a sterile layer and probably the bedding layer for the robbed out floor onto which the roof tiles had collapsed. Below this was a sterile natural layer with many tiny pieces of water rolled gravels on top of clay which was excavated to a depth of 0.16m.

BUILDING:

A large rectangular structure in the southeast section of the excavation site was uncovered when the mechanical digger removed the topsoil. Only the foundation walls of this structure had survived. The external measurements of the whole building were approximately 6.80m wide and 18m in length west to east and continuing under the eastern grass verge. Only a limited amount of excavation was done on this structure as it was realised that its complexity would require a great deal more work than what could have been achieved in the limited time available for work on the site. It was originally given the context number S42 by AC Archaeology (ACA) for the entire structure. It was later decided to give the eastern part of the structure a different context number S299 as the floor level was at a slightly higher elevation than the western part and also differed from it in other aspects. The walls of the two parts were, however, all on the same alignment and their bases (where tested) were on the same level and were visibly parts of the same building.

The northern wall S49 of structure S42 A small section of wall S49 was removed and a sondage 0.50m square (169) was excavated through it to the deposits below, 1m east of the northwest cornerstone. The fill was of limestone and gravel in a loose dark brown loam and rubble. It was clear that this section of the wall had been robbed of much of its stone and bonding material. Only two layers of large blocks of limestone were still in situ. The stone visible in the bottom of the section was a large shaped stone.

With only a few exceptions, the finds of a medieval character came from contexts associated with this building and, more specifically, most of the ceramic ridge tiles came from fill associated with the northern wall(s) (49) and (281). Western part of the building S42

Southern wall S184 A sondage (182) cut through wall S160 at its eastern end and by the junction with wall S184, showed that the foundations of the southern wall at this point were constructed from limestone rubble.

The interior had a layer of gritty clay mixed with limestone demolition rubble. The southern half of the interior had sunk and was filled with a dark brown humic

Near the junction with wall S160 (the southern wall of S299) there was a 1m gap in the wall, probably an entrance. Walls S92 and S89 were each side of the 18

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET A 1m x 1m section (139) of wall S281 was removed to reveal the base of the curving wall and a small area of an occupation layer (300) inside the feature. The occupation layer was of hard packed clay with evidence of burning. Fired clay and charcoal flecks were present throughout the fill. The skeletal remains of a small dog were found approximately 0.10m below the gritty layer of the interior of structure 299 and immediately above the remains of feature 274.

entrance and formed a square feature, possibly an entrance lobby, to the south. Feature 92 on the east side of the entrance gap was 2m in length and 0.84 to 1m in width. It was constructed from shaped limestone blocks above of which three courses of dry stone walling survived. A metal fitting was embedded in its northwest corner stone but was shown to be almost certainly a very recent addition (perhaps for tethering animals?). Feature 89 on the west side of the entrance gap was 2m in length north to south and 0.5m wide. It was constructed from limestone blocks and slabs with 4 courses remaining, which were laid onto a fill of limestone rubble (182).

Southern wall 160 Set into the southern wall 2.75m from the eastern grass verge, were two steps (181). The top step was constructed from 2 uneven and chipped oblong shaped blocks of limestone, which had been well weathered. Each stone was approximately 0.75m x 0.35m x 0.23m and laid horizontal end-to-end with a filling of limestone pieces packed against the southern side. The bottom step inside S299 was also constructed from two weathered limestone blocks, approximate the same size as those of the top step. The edges of the stones were not as rounded as could be expected from regular use over a long period of time. There were no indications whether the steps led into/ out of the building or connected with another room or structure.

Excavation of the area between wall S184 and the grass verge, revealed an earlier short length of wall (S50) 2m in length and 0.60m in width, butted up against S92 and external to structure 42. This later feature had postholes which were lined with wood cut into the east and west corners on its northern side. S299 The Eastern part of the building The south (160) and north (281) walls of the structure were 4.50m apart and 7.50m in length. Part of wall S281 (numbered S49 by A.C.A) had become damaged through post-depositional disturbance or may have been partly robbed. The top surface of the interior of the building, below the soil removed by the JCB, was a hard packed surface of gritty yellowish brown clay with a scatter of limestone rubble.

A small section of gully (309) made from upright limestone slabs was revealed running parallel against the bottom of the lower step inside the building.

A layer of uneven limestone slabs (159) approximately 4m x 2m was uncovered at the eastern end, which continued under the eastern grass verge.

Feature 178 A sondage was excavated down in the small space between the southern wall(s) S160/184 and the grass verge on the south. This revealed that the foundation trench for wall S160 had been cut into a dark brown silty loam (171) and was 1m wide 0.30m deep with vertical sides and a flat base. It had been filled with hard packed clay, small gravels and limestone inclusions. Only three courses of limestone walling were uncovered.

Northern wall 281 A sondage cut through the northern wall showed that its foundations had also cut into an earlier feature (274) at its eastern end (see below) At its western end was a section filled with a later deposit of dark brown loam virtually free of inclusions. This was directly opposite feature 92 in the south wall.

The sondage also revealed that the foundation wall(s) at this point had cut through earlier features (178). To the south of the wall were the earlier remains of a curving well-made dry stone wall. It curved from northeast to the west, was 1.5m in length 0.70m in width and had been cut by the foundation wall S181 as well as by features 50 and 92, near the junction with wall S184 (see above). This feature could not be further excavated without disturbing the walls overlying it.

Water channel/Gully (41,158) Immediately to the north of the building 1m from the foundation wall (281) was a stone lined gully (41), running west to east where it disappeared under the site’s eastern grass verge. A 2.40m sondage (158) was excavated at its eastern end. The feature was constructed from two rows of vertical limestone slabs 0.30m apart which varied in thickness and length. The gully was 0.20m in depth and filled with a dark brown sticky loam with limestone inclusions. It had cut through an earlier curving structure (274) at its eastern end.

BURIALS AND POSSIBLE GRAVE CUTS: (Figures 9 and 48-69). All the following burials were examined as fully as possible to confirm their nature without disturbing them or lifting them. They were all left in situ and reburied (see Site strategy). The estimated ages of infants were based entirely upon limb bone length. Possible grave cuts or possible graves not confirmed by adequate excavation are also recorded below in this report.

Feature 274 Part of this earlier curving dry stone wall structure was uncovered at the eastern end on the north side of the building. The wall was curving from under the eastern grass verge southward and survived to five courses high. It had been cut by the gully and northern wall S261. 19

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET with the head to the NE. The body had been interred in the foundation of Feature 208 in a recess within an oval shape of limestone pieces. A small limestone block had been placed to the north near its head. The burial was in the same fill as skeletons 4 and 5. Pot sherds were found in the fill but these may have been intrusive from the surrounding soil. This infant was smaller than 4 and 5. The length of the tibia was approximately 30mm in length putting the age at approximately 24 weeks.

Note: The burials are not discussed in numerical order as discovered and recorded but are given below in categories. Infant burials in Structure 200 Skeleton 4 Context 241 Grid ref:108.5E/98.5N Neonate The infant was orientated SW-NE with head to the SW and had been placed parallel to the internal edge of the southeast feature 208 of structure 200. It was interred in a shallow grave below fill 189 and in fill 205b. A flat beach pebble was found in fill 205b above the burial. The skeleton was crushed and the skull fragmented. The body was in the supine position with its arms across the chest. The legs were slightly flexed and both knees faced out in a bow legged position. The fill (205b) surrounding the skeleton was the same as the rest of the soil in the area and although the burial was not fully excavated, it could be seen that a semi circle of small limestone pieces may have surrounded the burial on the west side. A rimmed pottery sherd was found above the skull to the west but this may have been intrusive from the surrounding soil. An approximate age for the infant was estimated from the length of the long bones. The tibia was 75mm in length and the Radius 55mm putting the age at approximately 43 weeks. This was the only burial found on site that was in the supine position and was part of the same complex of burials which included infant burials 5, 6 and 7.

Skeleton 7 Context 273. Grid Ref. 105E/98N Infant Part of a fragmented skull of an infant was uncovered immediately south of feature 192 when clearing back in fill 205 to the west against the internal edge of wall 257 of Structure 200. No further excavation took place and the fragment was left in situ and reburied. Infant burial group in area of fill 121B and C: Skeleton 2 Context 220 Grid ref: 115E/110N Neonate The infant was orientated N-S with head to the N and had been placed at the south west end of a 2.0m length of upright limestone slabs which was part of feature 290 and immediately south of another possible grave (F 287) (see below). The skeleton was found below the roman soil level in a layer (121c) of dark brown loam with limestone inclusions. No grave cut was recognised. The fill (121c), which was around the body, was the same as the rest of the soil in the area which contained many small beach pebbles, pottery sherds and animal bones. Although the skeleton was partially disarticulated, crushed and fragmented, the orientation could still be determined. The body was in a crouched position lying on its right side with hands near the head and facing west. One single limpet shell was just above its head directly to the north. To the east 10cm from the body was a small block of grey limestone which could possibly be a grave marker. An approximate age for the infant was estimated from the length of the long bones. The right femur was 85mm in length and the tibia 70mm placing the age at approximately 43 weeks.

Skeleton 5 Context 253 Grid Ref: 108E/98N Perinatal The infant burial was directly to the north east of skeleton 4. It was orientated SW-NE with head to the SW and had been placed parallel to the internal edge of feature 208. It was interred in a shallow grave and buried in the same fill as skeletons 4 & 6. Although the skeleton was partially disarticulated, crushed and the skull fragmented, the orientation could still be determined. The body was in a crouched position lying on its left side facing west with hands near the head. A small curved and shaped piece of limestone was placed to fit around skull to the SW and limestone pieces which may have formed a semi circle around the burial were also visible to the west of the skeleton. Pottery sherds were found in the fill but these may have been intrusive from the surrounding soil. This infant was larger than 4 and 6. The length of its tibia was approximately 77mm in length putting the age at approximately around birth.

Skeleton 9 Context 284 Grid Ref: 117E/109N Neonate The infant burial was directly to the east of skeleton 2 and was orientated N-S with head to the north. It had been placed at the east end of the 2.0m length of upright limestone slabs which was part of feature 290 and in the same layer as skeletons 2 & 8. The skeleton was also crushed and the skull fragmented but it was still possible to determine its orientation. The body was in a crouched position lying on its right side, facing west with its hand near the head. The skeleton was too crushed to measure

Skeleton 6 Context 210. Grid Ref. 108.5E/98N Late foetus to neonate The infant burial was directly to the south east of burials 4 and 5. The skeleton was also disarticulated and the skull crushed but the orientation could still be determined. It was in the crouch position lying on its right side facing west with hands near its head and was orientated NE-SW 20

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET any of the long bones but its size was approximately the same as skeleton 2.

Skeleton 1A The skeleton was orientated approximately N-S with the head to the N facing W. It was articulated and in the crouched position. It was immediately to the east of skeleton 1 and was placed with its back against the NE side of the pit. The head appeared to be pushed against the side of the pit and was slightly higher than the rest of the body. It was crushed and tilted forward to the right with part of the jawbone visible, which was broken into several pieces. The body was lying on its right side with the right hand near the head. The upper torso was twisted slightly to the left, nearly into the supine position. The left arm was across the chest at a right angle. The radius and ulna of the left arm had been damaged and broken at the wrist and the bones of the hand were not visible. The lower limbs and pelvis were crushed and broken and part of the lower femur and upper tibia and fibula were broken and fragmented at the knee joint. Resting on these lower leg bones was part of the pelvis of skeleton 1. The damage to the arm and leg bones suggests the possibility that skeleton1 may be later than 1A although no re-cut of the pit was recognised. Above the left shoulder by the head in fill 232 was an oblong V shaped piece of chert. The age of the skeleton at death is not known and the long bones were too damaged to measure but the general size was that of an adult. The burial was not fully excavated, left in-situ and reburied.

Skeleton 8 Context 282 Grid Ref: 117E/109.5N Infant Directly south west of skeleton 9 part of a badly disturbed burial of an infant was uncovered in the same layer as skeletons 2 & 9. Parts of the leg bones, pelvis and ribs were visible and part of a very fragmented skull to the north of these. No further excavation took place and the skeleton was left in-situ and reburied. Multiple and Cist burials Skeletons 1, 1A Context 117 Grid Ref. 109E/106N Adults To the north approximately 6.0m from Structure 60 two adult crouched articulated skeletons were found buried in a shallow pit. It was unclear whether the burial had cut through part of a possible wall (F299), forming part of the ‘enclosure’ wall of the round building S60. or whether the wall was in fact a feature of the burial. The pit (cut 231) was oval in shape approximately 1.0m in diameter and had been cut into the natural clay. The fill (232) of the pit, which was not fully excavated, was a dark brown loamy clay with small limestone inclusions and was the same as fill (221) which was above it. Part of a human long bone was found in fill (232) above Skeleton 1. It was bleached white and very pitted and had the appearance of a bone left exposed to the elements. Pottery sherds were also found in the fill but these may have been intrusive from the surrounding soil.

Skeleton 3 Feature 30 Cist burial Grid Ref: 115E/123N Below the soil removed by the mechanical digger in an area to the north east of the site a layer (80) of brown silty clay with very few limestone inclusions was uncovered. The outline of a rectangular feature (30) was the clearest evidence of activity visible in this fill. The cut (103) of the feature into the natural clay was 1.40m x 0.86m and aligned NW - SE and the fill (104) was of pale yellow hard packed clay. The feature was half sectioned and the north section was removed first.

Skeleton 1 The skeleton was orientated NW-SE with the head to the NW facing SW It was articulated and very tightly flexed, suggesting it may have been bound. It was lying on its right side with hands near the head and knees close to the chest and the upper back was against the north side of the pit. The skull had been slightly crushed, the jawbone broken and out of line. The ribs were close together and pushed downward nearly parallel to the backbone. The visible left arm and leg bones each had several breaks in them. The left fibula and tibia had also been broken and disturbed where they join to the foot. Both sides of the pelvic bone were visible, part of which was lying across the bones of the left fibula and tibia of skeleton 1A (see below). Part of an animal bone was found lying N-S across the abdomen. It was 10cm in length, the north end cut at an angle, the south end broken. The age of the skeleton at death is not known but the approximate length of the left tibia 300mm and the left fibula 400mm suggests it was an adult. The skeleton was not fully excavated, left in- situ and reburied.

On clearing back against the section, approximately 30cm down in the centre of the feature, the cut (107) of a small square pit with a width of 0.30m was noticed. The fill (108) of this small pit was of pale yellow hard packed clay mixed with light brown loam. In the centre of the fill sherds of a broken black burnished bead rimmed pot (SF 30) were found resting on a flat limestone slab with another flat limestone slab at a 30 degree to the northwest of it. All the sherds were facing with the outside of the pot and base uppermost, i.e. the pot had been placed upside down on the limestone slab. The pot has been classified as being of the Maiden Castle War Cemetery type (F8 Fabric 5) of the mid or third quarter of the first century AD (Sparey-Green Part 3). It had been disturbed at some stage and several sherds were missing. On removing the rest of the fill (104) from feature 30 the capstone of a limestone-lined cist was uncovered at a

21

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET the knee bone was missing from the femur. The tibia long bone had been placed the wrong way round with the knee joint also to the southwest so that the skeleton’s two legs would have faced in the opposite directions. The patella was lying against the lower part of the femur. It was decided to leave the legs in-situ and further excavations took place directly to the east and west of them.

depth of about 0.5m. The capstone had broken in several places and collapsed. The broken capstone was removed at the northwest, revealing a crouched inhumation, of which, only the skull, scapula and one arm were exposed. The skeleton appeared to be lying on its left side, facing north with its head slumped towards the southwest. To the north of the head were two black burnished ware pottery vessels. A small cooking pot (SF 173) which had fallen on its side and broken into several sherds and a bowl (SF174) (dog dish) which remained whole and contained several small animal bones which were also under the dish. An iron nail was also above the bones.

A double adult crouched burial was partially uncovered 0.2m below Sk10 and fill 289. Skeleton 10a was to the east and Skeleton 10b was just visible below the fill and skeleton 10a and directly to the west of skeleton 10. Skeleton 10a was orientated N-S with the head to the north and facing west. It was articulated and in the crouched position and lying on its right side. Its left arm was at a right angle to the body. One finger bone was visible near the head. As skeleton 10 could not be further excavated, it was unclear whether Skeleton 10a had been placed against the side of the pit.

The grave was not fully excavated and the skeleton and grave goods were left in-situ and reburied. Multiple crouched pit burial Skeleton 10 &10a and 10b Context 306. Grid Ref: 113E/113N Adults

Skeleton 10b was orientated N-S with the head to the north, facing west and was lying on its right side. It was directly to the west of skeleton 10. The back of the skull, backbone and part of the pelvis first became visible. Part of the leg bones, which were tightly flexed, and part of a hand near the head became visible on the west side below fill 289 and partly under Skeleton 10a. Skeleton 10b was slightly smaller than skeleton 10a and the tightly flexed legs suggest this body may have been bound.

A pit burial was uncovered at the northern end of wall S39, which has the appearance of forming an enclosure round the circular building S60 (see above Structural Features). It was at a point where that wall had either collapsed or was cut into. The edge definition of the cut (288) was poor and was approximately 1.50m in diameter. Part of a curving line of limestone slabs three deep was uncovered on the west and south eastern sides suggesting that the top of the exterior to the pit had been marked with flat limestone slabs. The pit, where excavated, had sloping sides cut into a light brown clay loam.

FEATURE 314 Possible Cist Grave: (Figures 9 and 68). Grid Ref: 119E/ 109.5N Directly to the north at the north east end of feature 307 below layer 291c a large possible stone lined cist was uncovered during the last day of the excavation work. It was 1.50m in length northwest to southeast and 0.80m wide and had been cut into the natural yellow clay to an unknown depth. It was lined with large flat vertical limestone slabs and had been capped by three large limestone slabs. The capping stones had collapsed into the fill (315) of redeposited clay. The pottery sherds found in the fill may have been intrusive from the layer above. Two further horizontal large limestone slabs were visible above and lying to the north east of feature 314. A triangular block of burnt mudstone was lying outside the cist near the north east corner (see Building Materials Old Red Burnt Mudstones). The nature of this feature and the surrounding area was not confirmed by further excavation.

In the very top of a fill (289) of dark brown loam with limestone inclusions, below the edge of the grave cut, was a sub-circular feature of same-sized limpet shells. They were in the north-west of the pit 15-20cm below the roman soil level and had been placed against the side of the pit. To the north-east 1.50m from the shells and at a slightly higher level in the fill, was a squarish feature of same sized small round beach pebbles set into the ground. Skeleton 10. To the south of the shells and lower down in the pit, the long bones of the left and right leg of an adult were uncovered. Only the femur and tibia of the left leg were present and were positioned in the south section of the pit. The femur which had broken in two was orientated east to west with the head of the femur to the east. Just above the knee joint was a 4cm in length deep cut in the bone. No healing to the bone was evident, suggesting the person had died shortly after receiving this injury. The tibia was damaged and broken in several places with part of the bottom of the leg missing.

FEATURE 193, Oval with limestone slab – possible grave This feature was 2m to the west of burial feature (30). It was oval in shape and roughly 1m in diameter with limestone pieces around its edge. A flat limestone slab 0.30m x.25m lay within the feature, slightly more to the south east side. A possible stake hole was to the east of the limestone slab.

The right leg bones, directly to the north of the left leg bones consisted of a femur, tibia and patella. They were orientated north-east to south-west, 10cm from each other and not in a flexed position. The femur long bone had been placed with the femur head to the southwest. Part of 22

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET

The cut (194) or edge of feature 193 had uneven sloping sides which were poorly defined. An uneven base of natural clay was reached at a depth varying between 0.15m to 0.35m and appeared as though it had been artificially compacted, perhaps trampled. The fill (195) consisted of a pale brown clayey loam with a few small limestone inclusions and no finds were present to indicate whether or not it was a grave, perhaps robbed out.

23

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET It was not regarded as feasible to attempt AMS dating or isotope analysis of the shells as there was not an adequately firm stratigraphic relationship in the distribution of organic and cultural material on the site.

3 BIOARCHAEOLOGY AND SUBSISTENCE The evidence suggests that the site occupants of all periods lived very well off local produce: molluscs, fish and meat of sheep, cattle, possibly horse and also sea birds. Without a substantial series of radiocarbon dates from all the different species, it is not possible to reach absolutely certain conclusions about which species were used most in which period(s). As the site’s deposits, from early prehistoric times onwards, had been disturbed by subsequent site occupants, scientific dating may not be as conclusive or informative as could be wished for. The composition and nature of the assemblage of meat (animal, fish and bird) bones has been judged by Maltby (see below) to be more likely to be from the period of the Roman or Romano-British occupation. This is also the major period evidenced by the artefactual and cultural remains.

Any connections between distribution ‘patterns’ of molluscs and possible functions of the structures can only be tenuous, particularly as it was not possible to fully excavate any of the features or areas (apart from possibly S60) to an adequate degree. A small arrangement of equal-sized winkles was revealed at the eastern, head end, of Burial 10 with a pebble arrangement at the western end above the head(s. This unusual arrangement probably had a strong symbolic or ritual meaning, particularly as the burial itself was also very unusual. This is further discussed in Part 2.3 and 5. List of edible invertebrate species recovered from the Weston Road site

3.1 MOLLUSCS Patella sp probably most vulgata Monodonta lineata (da Costa) Littorina littorea (Linnaeus) Littorina obtusata (Linnaeus) Buccinum undatum Linnaeus

Some of the mollusc shells found on the site, present a problem as many of the species - particularly the limpets and winkles – were so commonly found in early prehistoric times. Limpets and winkles were extremely prolific on the Mesolithic site of Culverwell and also Site 1 on Portland (Palmer, 1999). However, as no structural remains, or other evidence of settlement, of the Mesolithic or later Stone Ages were found, it may be reasonable to tentatively treat the subsistence remains as though they were mostly from one broad period, namely the Romano-British. Only a small quantity of molluscs, amounting to 233gms, were found in contexts associated with the rectangular building S299/42 and will be considered below.

Mytilus edulus Linnaeus Ostrea edulis Linnaeus Cardium edule Linnaeus

A total of approximately 21,757kgms of shells (including approx 5,057 land snails) were collected for classification (see Appendix). As there is no certain evidence that any of the land snails were purposefully harvested by the site occupants, they have been treated as environmental data and discussed in Part 1 Environment.

Cardium aculeatum Linnaeus Sphaerium corneum

These figures for edible mollusca, purposefully brought to the site, cannot be regarded as a precise reflection of the volume of mollusc meat consumed on the site as extremely fragmentary remains of shells were present, as also found in many of the soil samples taken, It would have been too time consuming to collect them all, and possibly also counterproductive, as classification of very tiny remains would not always have been possible. The following discussions can, therefore, be regarded as based on a sampling of all identifiable mollusca found in all excavated contexts.

Cirripedia sp.

Thick topshell Edible periwinkle Flat periwinkle Whelk Common mussel Common European oyster Common edible cockle Large offshore cockle Perrey carrow (Pea mussel) Barnacle

Only 1 specimen

Slightly salty, mouth of stream Only 1 specimen

From the above list it is evident that the site occupants at Weston Road harvested food from off-shore, on- shore as well as from streams and muddy and sandy areas. This reflects the fact that access to different parts of the coast round Portland was still relatively easy, the area of the Mere at the northern end of the Island, was rich in food supply and streams flowed into the Fleet and harbour (see Part 1.2 Environment). There are three species of Patella but they can only be identified from the living mollusc; the most common is vulgata.

In recording the weight of shell specimens collected, it is also relevant to note that the individual species vary greatly in size and weight and, also, some are more likely to fragment than others. Very few specimens were complete, apart perhaps from the more robust oysters. 24

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET This does not appear to have been the case at the Weston Road site. A total of 3.702kg of oyster shells were found as opposed to 9.324kg of limpet shells, i.e. nearly three times more limpets in weight. Here it should also be considered that limpet shells are lighter and far less sturdy than oyster shells and hardly any limpet shells were found that were complete. This makes the difference between the two assemblages really significant.

Distribution of mollusc shells There was a very wide random distribution of all the species of shells but in the following table an attempt will be made to ascertain where the largest quantities (totals of different species) were collected. Context 109 110 121 129 189 198

Quantity Details of Context (see Site plans)

1256gm 702gm 1793gm 2409gm 1628gm 2392gm

Most of the oysters 1,467kgms were in Context (23) and (203), both contexts being top fills of the ditch (22), and which can be regarded as ‘rubbish’ thrown away or washed into the ditch (see Part 2.30. The other molluscs in (203) were: land snails 34gms, limpet 22gms, winkle 19gms and mussel 1gm.

Inside linear structure west of S60 North of S60, inside ‘enclosure’ Outside northern wall of S42/S299 Above S198, fill, half of round bldg. Area of S200 & square base F206 Interior fill S198, half of round bldg

The above figures for (129) and (198) as fills of S197, may possibly be significant and could suggest that S197, the half of a round building (see Part 2.3), may, amongst other possible functions, have been used for the storing or processing of molluscs. The collected total of shells from the two contexts amounts to about 5,134gm, nearly 25% of the total collected. Only half of this building was available for research, the other half being under a grass verge, and the available half was not fully excavated. Further work could, therefore, either have changed the picture or else confirmed it.

Oysters were also found in the following contexts which can be tentatively related to structures on the site: (66) fill of the oval structure 59; (129) fill of S197 the half of a round building; (138) and (199) fill of S200 the area of ovens/furnaces and (279) fill against the southern wall of the rectangular building S299. These contexts are all reasonably clearly associated with structures which otherwise contained mainly roman material, except for (279) which is just outside a wall of a structure which is regarded as probably medieval (see Part 2.3and Part 5). As the latter structure is thought to overlie roman remains, it can, therefore, possibly be assumed that most (if not all) the oysters were consumed by the roman/romanized population of the site. However, it can also be considered that the largest quantity of oysters in the ditch were found by the builders of the rectangular building and were swept up and dumped in the ditch, which would lead to the further assumption that the ditch was still open at that later time. (See further discussion Part 5).

Context (110) is a large area which contained no features but which appeared to be within a possible enclosure surrounding the round building S60. Context (109) is within part of a rectangular or square structure attached to a north-south wall which is also attached to S60 on the west side. The connection of finds from these two contexts and the round building is possible but not clear and is further discussed in Part 5. The material from Context 189 is the fill over a square limestone base S206 within the area of S200, which contained evidence for a hearth/furnace, and could indicate that some or other activity took place involving an object, whatever it was, placed on the square base. This will also be reviewed in relationship to other finds and further discussed in Part 5.

The further comment can be made that, although the Roman/Romano-British site inhabitants did eat oysters, it was not their favourite mollusc dish. This could perhaps be explained by the fact that the oyster beds were not as conveniently harvested as the limpet sources. For as long as is known, the nearest source of oysters to Portland has been beds in Weymouth harbour, which would have entailed taking them by boat to a landing place somewhere near the northern end of the Island. There is a possibility that the only beds available in Roman times may also have been there in the harbour and this would have meant that harvesting oysters would have been a bit more laborious than collecting limpets from the numerous rocky shelves round Portland, near the place where they were processed. (See also Part 1. Environment).

The large quantity of mollusc shells from the large area of Context 121, between the ‘wriggly wall’ S37 and the northern wall of the rectangular building S42 and S299 (regarded as probably medieval), could perhaps suggest that this was an area into which refuse was deposited. Ratio of shell species to each other In attempting a discussion of the ratio of the different species of molluscs to each other, it should be borne in mind that although most shells in excavated areas were collected, the overall picture may perhaps have been different if more areas could have been researched.

Only 326gms of scallops were found. They are bottom dwellers off-shore and would have required diving unless occasionally washed up on shore or caught up in a fishing net. The collected specimens came from the following contexts: 20, 128, 129, 144, 162, 170, 171, 188, 189, 202, 209, 223 and 235 with only one or two specimens per context in most cases. Larger quantities came from a few contexts as follows:

Cool (2006) has suggested that oysters predominated in the Roman or romanized British diet over other molluscs.

25

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET Context 129 Context 188 Context 189 Context 202

199gms 25gms 15gms 25gms

The sample from (129) came from fill associated with Structure 199, the half of a round building where oysters were also found and a large quantity of other mollusc shells (see above). The specimens of (202) came from fill within S60, the complete round building west quadrant while those from (189) are from fill within the area of F192 which contained remnants of one or more furnaces/ovens and an enigmatic square carved slab of limestone F206. Mussels also occur in only a few contexts: (198) 698gm; (199) – 18gm; (203) – 1 gm; (304) – 1gm. Context 198 is also a fill within the half of the round building, already noted above, which tends to confirm the suggestion that molluscs were stored or processed in that building. Mussels would have been easy to harvest from areas where streams entered the sea, e.g. along the Fleet and Weymouth bay and possibly also from (now extinct) mouths of streams on Portland. From the above figures it would appear that not all species represented on the site, were harvested regularly and there appears to have been some issue of preferences or, maybe, discrimination. Land snails A quantity (approximately 5057gm) of shells from land snails were collected from the site. As there was no evidence that these were involved in the subsistence pattern of the site occupants, they are discussed in Part1.2 for their environmental implications.

26

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET

(Numbering of tables, as given by Maltby, has been retained. Tables not included in the text can be found on the CD)

structures and layers but other groupings are more arbitrary, simply amalgamating contexts of similar types. Finds associated with the upper levels of structure S299 are the only one which produced significant numbers of medieval artefacts. This building may have been built upon an earlier Romano-British structure and most of the bones appear to be from lower levels of this building and therefore of pre-medieval date.

Methods of Animal Bone Analysis

Size of Faunal Assemblage

All bones and teeth recovered from the excavations were recorded individually onto a relational database (Microsoft Access), which forms part of the site archive. In the main table, where appropriate, the following information was recorded for each specimen: species; anatomical element; part of bone present; approximate percentage of bone present; gnawing damage; erosion; weathering; burning; fusion data; other comments. Separate tables linked to the main table by an individual identification number were created for metrical, butchery and tooth ageing data. Where necessary, identifications were confirmed by reference to the comparative skeleton collection housed in the School of Conservation Sciences, Bournemouth University. The fish and the wild bird bones were identified by Sheila Hamilton-Dyer. Tooth eruption and wear descriptions for cattle, sheep/goat and pig follow the method of Grant (1982). Most measurements for mammals and birds are those recommended by von den Driesch (1976) and most of the fish bone measurements follow Morales and Rosenlund (1979).

A substantial sample of 13,649 animal bone fragments was recovered from the excavations (Table 2). Considering the small area excavated, this represents a dense accumulation of material. It is also testament to the high standard of recovery despite the lack of a sieving programme. Samples of over 1,000 fragments were obtained from structures S60, S197 and S200, rubble layer 121, other rubble layers, soil layers 100-110, and other soil layers.

3.2 ANIMAL BONES FROM WESTON ROAD, PORTLAND by Mark Maltby

Bone Preservation General comments on bone preservation were made for each of the 131 contexts. Only two assemblages were categorised as poorly preserved with many fragments displaying significant surface erosion. Five (4%) were designated as quite poorly preserved with slight erosion on many of the bones. Thirty-five (27%) of the assemblages are moderately preserved with good surface preservation but with high percentages of fragmented and gnawed bones. Most contexts (66%) have quite well preserved assemblages with good surface preservation and less fragmentation. Only two samples, however, have excellent preservation. A total of 9,368 (69%) of the bone fragments are from very well or quite well preserved assemblages. Only 54 (0.4%) of the fragments are from poorly or quite poorly preserved context assemblages.

Sub-Division of the Sample Animal bones were recorded from a total of 131 contexts. As will be discussed below, the faunal sample is fairly homogeneous across the site, displaying relatively little variation in species and element representation. Although this is a multi-period site, it is believed that the vast majority of the bones are of Romano-British origin. For the purpose of this analysis the sample has been divided into the following 14 groups:

Surface erosion was recorded on only 306 (5%) of the identified elements. This figure varies between 0% and 10% in the 14 groups (Table 3). The highest incidence of eroded fragments is from soil layers 100-110 but there is relatively little variation in surface preservation between different context types. The relatively shallow deposition of the assemblages sometimes in layers that probably accumulated quite slowly accounts largely for the presence of eroded elements.

Structure S59 – Oval drystone structure Structure S60 – Round drystone structure Structure S197 – Round drystone structure Structure S200 – Round/oval drystone structure (kiln/furnace) Structure S299 – Rectangular building (medieval) Ditch F22 – East-West ditch Other Linear features Grave fills Other walls of structures Rubble layer context 121 – below S299 Other rubble/demolition layers Soil/loam layers 12, 100-110 Other soil/loam layers Other contexts – miscellaneous

Gnawing damage, mainly by dogs, is slightly more prevalent than surface erosion, affecting 528 (8%) of the identified elements (Table 3). This figure ranges widely between 2% and 17% in the 14 groups, although the extreme figures derive from small samples. Gnawing damage particularly lessens the survival rate of limb bone extremities and the bodies of vertebrae. In many cases these have been completely destroyed, which has had detrimental effects upon the number of elements that have provided epiphyseal fusion and metrical data. Gnawing damage was observed frequently on the major domestic mammal species. Relatively more (13%) cattle elements are gnawed than those of pig (11%) and

The bone-bearing contexts involved are listed in Table 1. The groups are partially based on clearly defined 27

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET sheep/goat percentages within a narrow range of 65%80%.

sheep/goat (7%). This implies that more cattle bones survived scavenging than those of the smaller species, which have more fragile bones susceptible to total destruction. Gnawing damage was also observed on two bird, four fish and two deer bones.

Table 5 compares the NISP counts of sheep/goat, pig and cattle only. The overall percentage of sheep/goat rises to 73%, ranging between 40% and 81% in the 14 subgroups. Extensive comparisons between the relative percentages of sheep/goat, cattle and pig have been made by various authors (e.g. King 1999; Maltby in press). Results from a variety of Romano-British sites from Dorset, Devon, Somerset and Hampshire are presented in Table 6. These show that the Weston Road assemblage contains one of the highest percentages of sheep/goat. There are several, not necessarily conflicting, explanations that need to be considered to account for such high percentages of sheep/goat elements.

Only 184 (1%) fragments are burnt, of which 135 are from unidentified elements. High percentages were obtained from S59 (5%) and miscellaneous wall contexts (8%). The possible kiln/furnace S200 assemblage includes 3% burnt fragments. Sheep/goat provides 41 of the 49 burnt fragments amongst the identified material. Small numbers of cattle (3), pig (2), bird (2) and fish (1) elements were also affected. The slightly higher percentage (0.9%) of burnt fragments in the sheep/goat assemblage may indicate that their bones were more likely to come into contact with fire, either directly during cooking or after food consumption. However, it should be noted that large mammal fragments (41) are quite well represented amongst the burnt fragments unidentified to species.

1) Recovery rates were better at Weston Road than on other sites. There is no doubt there was a high standard of retrieval, which could account for higher percentages of sheep/goat than on some sites where their smaller bones are more likely to have been overlooked than those of cattle. However, if one counts only the elements of sheep/goat and pig – species of similar size – the percentage of sheep/goat (87%) is still among the highest encountered in the samples compared, indicating that other factors are involved.

Modern breaks were visible on many bones accounting for significant amounts of fragmentation beyond breaks made in antiquity. Overall, 6% of the identified elements were recorded as broken. The highest percentages are from assemblages from grave fills (11%) and walls (11%).

2) The high percentage of sheep/goat is indicative that the settlement was rural in nature. King (1999) has argued that “unromanised” indigenous rural settlements in Roman Britain tend to have higher percentages of sheep/goat than more “romanised” urban, military and villa sites, perhaps indicating a continuation in dietary preferences of the native communities. The results presented in Table 6 partially support this argument. Nearly all the urban samples have much smaller percentages of sheep/goat than rural sites. However, it is suspected that the site at Weston Road is part of a much larger settlement complex than a rural farmstead and the density of bones could also suggest this. The artefact assemblage is also richer than would normally be expected on such a site.

A total of 6,919 specimens are unidentified (Table 3). The percentage (51%) of unidentified fragments is indicative of the fragmented nature of the assemblage. The highest incidence of unidentified fragments (66%) was found in the assemblage from grave fills. The small sample from other contexts produced the lowest percentage (38%) of unidentified fragments. Further indicators of preservation will be discussed when reviewing element representation and fragmentation patterns of the major species. Sheep/Goat Abundance Although most of the sheep/goat elements could not be further identified, it was possible to distinguish between sheep and goat from some of the better preserved skull, mandibles and limb bones using criteria recommended by Boessneck (1969) and Payne (1985). At least 454 elements belonged to sheep compared with only 19 identified as goat. The dominance of sheep within sheep/goat samples (96%) is typical of Romano-British sites (King 1991; Maltby in press). It is marginally lower than at Greyhound Yard, Dorchester (97.4%) (Maltby 1993).

3) The high percentage of sheep/goat reflects the local, insular pastoral economy that was heavily reliant on sheep production. The historical importance of sheep on the Isle of Portland is well known (Jones 2005) and the generally thin soil conditions on much of the island are much better suited for sheep grazing than for other animals, particularly cattle. People resident on the island in the Roman period, no matter how influenced they were by the Romans, are likely to have relied largely on local food resources.

Sheep/goat elements (NISP = 4,384) dominate the identified portion of the assemblage, providing 71% of the mammal fragments, their contribution varying between 40% and 81% in the 14 sub-groups (Table 4). However, the lowest and highest percentages are from small samples and the lowest percentage is from S299, which is biased by the inclusion of a partial dog skeleton in the counts. The nine largest assemblages have

4) Sheep were very important in the pastoral economy in Dorset during the Romano-British period. Similar very high percentages of sheep/goat in relation to pig have been found at sites such at Rope Lake Hole (Coy 1987a), Whitcombe villa (Buckland-Wright (1990) and the small sample from the Hod Hill fort (Fraser 1968). However, other sites do not have particularly high values compared 28

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET tend to favour elements such as the skull that easily fragment but can still often be identified. MNE counts generally have the effect of increasing the percentage of upper limb bones and this sample is no exception. There is, however, no clear indication that there is a heavy bias towards joints of lamb and mutton being imported. Many of these bones belong to animals that were brought to the vicinity of the site for slaughter and processing. However, cranial and foot elements do tend to be less well represented in the MNE counts than the more robust upper limb bones at least, which could imply that some joints of meat were introduced.

with sites in Hampshire, for example, particularly if cattle are excluded from the calculations. 5) Some Romano-British temples have a high percentage of sheep/goat bones associated with them (King 2005). Perhaps, the most relevant comparison here is with the assemblage from Hayling Island. However, species representation there is dominated by both sheep and pig with very few bones of cattle (Table 6). Sheep/Goat Elements Represented Table 7 provides a summary of the different types of element identified for sheep/goat. Three methods of calculation were carried out. The first involves a simple count of specimens (fragments) identified to each type of element (NISP). The second method records the zones of the element represented, enabling the determination of the minimum number of each element present (MNE). The third method involves the calculation of whole bone equivalents (WBE). This takes into account how much of the complete element has survived. Values of 1.0, 0.75, 0.50, 0.25 and 0.10 were assigned to each specimen based on the approximate proportion of the element present. Table 7 provides the summated totals of these counts for the major bones.

There are only modest amounts of variability in sheep/goat element counts in the 14 sub-groups. For example, the percentage of loose teeth only ranges between 22% and 36% in the seven largest assemblages. Gross comparisons of body parts represented (Table 8) show some variations between groups but no clear patterns. Foot bones tend to be slightly less common in structures than in other groups and S197, S200 and S299 all had higher than average percentages of cranial elements. However, the overall impression of the assemblage is one of homogeneity. Sheep/Goat Fragmentation and Butchery Evidence Given the fragmentary nature of the sheep/goat assemblage, it is of no surprise that there are very few groups of associated bones. There are only three examples recorded. The first consists of pairs of complete tibiae and metatarsals probably from the same young lamb in S200 (context 199). S60 produced the distal epiphysis of the left tibia, astragalus, calcaneus, centroquartal and metatarsal of an immature sheep (context 202) and the distal epiphysis of a right femur, tibia, astragalus, calcaneus, centroquartal and metatarsal of another immature sheep (context 245). Although neither group has any evidence of cut marks on any of the bones, it does not rule out the possibility that these were deposited after the carcasses had been at least partially processed.

Interpretation of sheep/goat element representation must take into account problems of differential survival and retrieval. Small bones such as the phalanges, carpals and tarsals are nearly always poorly represented because many have been overlooked during excavation. The high incidence of gnawing will also have had a greater detrimental effect on more fragile elements such as vertebrae and limb bone epiphyses. The high percentage of loose teeth (27%) confirms how fragmentary the sheep/goat assemblage is and its moderate survival. High percentages of loose teeth indicate many jaws have been destroyed leaving only the denser teeth. Apart from teeth the best surviving elements are the mandible (8%), tibia (7%), radius (6%) and metatarsal (5%). Again, this largely reflects the fact that these are relatively robust elements, whose shafts can survive well compared with more fragile elements such as the scapula and femur. MNE counts reveal that at least 146 tibiae are represented by distal shafts. Assuming that there are equal numbers from both sides of the body, this indicates that at least 73 sheep/goat are represented in the assemblage. Other elements with high MNE counts are the humerus (112) and pelvis (104). Mandibles (89) and metatarsals (70) are much less prominent when MNE counts are employed.

A summary of observations of butchery marks for all the identified species is provided in Table 9. Only 104 elements had cut or chop marks observed on them, representing 3% of the sheep/goat sample excluding loose teeth. High fragmentation and the destruction of many articular surfaces by canid scavenging partially account for this low figure. These areas are where butchery marks most often occur but any such marks will have been often obliterated. In addition, the majority of butchery particularly on the limbs was carried out using knives. Experienced butchers using knives often do not leave many cuts on the bones.

If NISP and MNE counts are modified to express them as percentages of the best represented element (excluding loose teeth) (Table 7: columns 18 and 20), it is possible to see that MNE calculations significantly increase the percentages of humerus, ulna, pelvis, tibia, tarsals and neck vertebrae. Conversely, relative percentages of mandible, skull, metatarsal and phalanges are lower than indicated by NISP counts. The reasons for these discrepancies lie in a combination of differential fragmentation and other taphonomic factors. NISP counts

Butchery marks were most commonly observed on tarsal elements. Twelve astragali, five calcanei and five centroquartals were recorded as butchered. Apart from one astragalus, all the marks were made with knifes. Most of the cuts on the astragali run transversely across the anterior surface. These would have facilitated skinning 29

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET enough for this assessment to be made. The numbers assigned to each stage are listed in Table 10. It was possible to separate the jaws of immature sheep and goats on the basis of their morphology (Payne 1985) but it was not possible to separate these species in the older mandibles. However, it is likely that most of these belonged to sheep.

and enabled the tibia and the rest of the upper hind limb to be separated from the feet. Similar transverse marks were observed on various aspects of the calcanei and centroquartals and reflect the same activity. Very few marks were observed on metapodials and phalanges but transverse cuts on two first phalanges are skinning marks. Butchery marks were also quite commonly noted on sheep/goat scapulae (9) and pelves (11), reflecting processing associated with segmentation of the carcass and filleting. Two scapulae bear superficial chop marks above the neck associated with separation from the humerus. The remaining marks are incisions found on various parts of the neck and blade. Most of the butchery marks on the pelvis are located on the ilium and all but two consist of knife cuts. The location and nature of most of the incisions suggest they were made during segmentation. Several upper limb bones also bear butchery damage (Table 9). Apart from a humerus and a tibia, which both had been chopped through the shaft during segmentation, all the marks consist of fine incisions, usually located near the ends of the bones.

Five mandibles belonged to neonatal mortalities (Stages 1). Four of these were identified as goat compared with only one sheep. Given that goats are relatively rare in this assemblage; this may be of some significance. Goats may have been more commonly milked than sheep and this could be reflected in the slaughter of newborn kids. Goats may also have been kept within the settlement making it more likely that the bodies of any neonatal mortalities (either natural deaths or deliberately killed) would be present. All five of these mandibles are associated with structures (S197, S200, and S299). Their slaughter and deposition may also have ritual connotations, although they are found amongst large amounts of other material. A substantial number (28%) of mandibles are at Stage 3, representing animals killed in their first year, mainly between 6-12 months of age. Of those that could be distinguished, 12 belonged to sheep compared with only one of goat. This may indicate a peak of slaughter of surplus lambs not required for breeding, milking or wool. They had not have reached full size, however, and may have been slaughtered to control the size of flocks utilising the available pasture on the island. Ten (19%) of the mandibles are at Stage 4 and would have belonged to sub-adult animals killed in their second year. Seven of these belonged to sheep whereas no goats were identified.

Butchery marks were rarely observed on cranial elements. However, the horncore of a male sheep has been chopped from the skull near its base presumably to release the horn sheath for working. Three parietal fragments also bear chop marks and in one case the skull had been cleaved open to remove the brain. Chop marks were also observed on the medial aspects of the diastemae (fronts) of two mandibles from S197 (context 198). These were made when the jaws were split. The only other mandible with butchery evidence bears knife cuts on the ramus just below the condyle. These were made when it was detached from the skull.

Twenty-two (42%) of the sheep/goat mandibles from the site had all three molars in wear (Stages 5-7) and belonged to animals over two years of age. Eight of these (15%), however, did not possess any heavily worn teeth (Stage 5) and are mostly from animals aged between two and four years old. Animals culled between these ages may have provided some fleeces and offspring prior to slaughter but were killed before old age. Fourteen mandibles of sheep/goat over four years of years of age (Stages 6-7) provided 27% of the sample. These would have been animals kept for breeding, wool production and milk products prior to slaughter. Only one specimen, however, was assigned to Stage 7, indicating that the majority of animals were culled prior to very old age.

Butchery marks were observed on 30 vertebrae (Table 9). Evidence for the use of a cleaver is more prevalent on these. All but one of the butchered lumbar and thoracic vertebrae and four of the cervical vertebrae (excluding the atlas and axis) bear chop marks. All are related to the division of the trunk into two sides. In two cases the vertebrae have been chopped through close to the midline but the majority bear marks that are situated off-centre. There is no evidence for transverse chops on any of the lumbar or thoracic vertebrae, although they were found on one cervical vertebra. In contrast, all bar one of the butchery marks on the atlas consist of knife cuts mostly running transversely. These were made during separation from the skull. One atlas from S60 (context 245) has been sawn in half. This is very unusual butchery for a Roman site and it is possible that this bone is a later (possibly post-medieval) intrusion. Butchery marks were observed on or adjacent to six rib heads. In all but one case these were made with knives.

It was possible to record large amounts of epiphyseal fusion data (Table 11). Fusion data tend to be less reliable than tooth ageing data in samples that have been subjected to significant amounts of scavenging by dogs, as the unfused bones of younger animals are more prone to destruction. For example, only 16% of the youngestfusing epiphyses in the sample are unfused. These represent animals that died under a year old and therefore represent a lower estimate of first-year deaths than the mandible data. A substantial number (352) of porous sheep/goat elements were recorded. These belonged to lambs and kids certainly under a year old and many

Sheep/|Goat Ageing Evidence Recording of the eruption and wear of mandibular cheek teeth followed the method of Grant (1982). Mandibles were then assigned to one of seven stages of development. Fifty-three mandibles were complete 30

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET measurements taken and compares some of them with the results from Greyhound Yard, Dorchester (Maltby 1993; in press). Generally, sheep in Roman Dorset appear to have been quite small even by Romano-British standards. The range and variety in size of the sheep from Weston Road is broadly similar to those from Dorchester. All the mean sizes that have been compared, however, show that the average size of the sheep in the Dorchester assemblage is slightly greater than in the sample from Portland. Differences in breadth measurements may simply reflect that there are more bones of larger males and/or slightly older animals in the Dorchester sample rather than any overall differences in the stature of the animals. Males are likely to be less well represented amongst the distal tibiae measured at Weston Road, as it is likely that a higher proportion of males were culled before the epiphysis fused than at Dorchester. All the other measurements compared are subject to a small degree of post-fusion growth and the higher percentage of immature sheep represented at Weston Road could be reflected in these results.

probably were substantially younger. This represents 11% of the sheep/goat elements (excluding teeth) and supports the evidence from the mandibles that the sample includes significant numbers of neonatal and juvenile sheep and goats. In addition a number of metapodials had shafts that were unfused or just fused, which is indicative of late foetal, or more likely in this case, neonatal mortalities. Of the later-fusing epiphyses, results from the second phalanges should be disregarded as they have probably been heavily biased by less efficient retrieval of the smaller unfused specimens. Results from the first phalanges should also be treated with caution for the same reason, although substantial numbers of unfused specimens were recovered, as should be expected. Results from the distal tibia and metapodials are fairly consistent (Table 11). Overall 61% of the surviving specimens have unfused epiphyses. These epiphyses generally fuse between 15 and 28 months, indicating that well over half of the sheep and goats represented were probably under two years of age, an observation that supports the tooth eruption data.

It is interesting to note, however, that the average withers height estimated from the maximum lengths of complete limb bones, is also slightly lower in the Portland sample (56.3cm) than in the Dorchester assemblage (58.1cm). Unfortunately, the sample is quite small and tests showed that the variation is not statistically significant.

Late-fusing epiphyses, particularly the proximal humerus, are the most susceptible to destruction and the sample sizes are consequently smaller and less reliable. Overall, 64% of the epiphyses are unfused representing animals that died under 3-4 years of age. The results suggest that there was not a significant slaughter of animals aged between two and four years old, which slightly conflicts with the mandibular ageing data. However, both methods indicate that approximately only around a third of the animals represented were older than this.

Another factor that needs to be considered is the possibility of chronological variation. There is some evidence to suggest that the average size of sheep became greater in the later Roman period at Dorchester (Maltby 1993). Unfortunately, the nature of the Weston Road sample precludes the investigation of possible diachronic changes.

Interpretation of these results needs to take into account the possibility that some stock may have been imported or exported. High percentages of first year animals, particularly those equivalent to Stage 3 of the tooth ageing sequence, have commonly been found in British Iron Age samples (Maltby 1981; Hambleton 1999; Albarella 2007), including some from sites in Dorset such as Maiden Castle (Armour-Chelu (1991) and Gussage All Saints (Harcourt 1979). On these sites, sheep/goat in the 1-2 year bracket are often less well represented than those aged between 6-12 months. Although there is a lot of variation in Romano-British assemblages, there is often a greater proportion of animals aged between 12 and 36 months (Maltby 1981; in press; King 1991), indicating an increase in the importance of meat production. This is not reflected in this assemblage and contrasts with the Roman assemblages from Dorchester, where there were very few specimens at Stage 3 in the tooth eruption sequence (Maltby 1993; 1994). A greater demand for meat in the town may account for this contrast.

Sheep Horncore Data The relative abundance of horned and hornless sheep has been shown to vary between different Romano-British sites, even between those located in the same region. Hornless sheep are largely absent in Iron Age samples but are found in variable numbers on later sites, including Dorchester (Maltby 1993; 1994; in press). At Greyhound Yard, hornless specimens form no more than 5% of the sheep skulls in the earliest Roman samples, a figure that rose to 37% in some later Roman deposits before decreasing to 11% in the latest Roman phases (Maltby 1993). In addition to the loose horncores (Table 2), horned sheep are represented by 27 skull fragments at Weston Road, including some from neonatal lambs. Only three (10%) skulls of hornless specimens were found (two from soil layers 100-110; one from S197). Again interpretation is difficult because of the problems of chronology but the percentage of hornless sheep falls below the maximum levels found in Dorchester, perhaps implying that the sheep at Weston Road came from less diverse sources.

Sheep Metrical Data Measurements were taken on 333 sheep/goat elements. Details of all measurements are stored in the archive. Table 12 summarises the data from the more common

31

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET between the WBEs and NISP counts for these elements (Table 13). The majority of these elements consist of 10% or less of the complete bone. Such fragmentation to an extent indicates that the bones were commonly broken during processing, particularly to release marrow. However, subsequent taphonomic processes such as weathering and trampling probably account for significant amounts of further fragmentation. The metapodials are less fragmented than the upper limb bones indicating that they were not as intensively broken. Although only six (18%) metacarpals and three (11%) metatarsals survived as 75% of more of the complete bone, such relatively complete bones are still more common than in the generally better preserved urban and suburban assemblages from towns such as Caerwent and Winchester (Maltby 2007). This implies that marrow processing may in fact not have been as intensive at this site.

Cattle Abundance Cattle elements (960) are fairly poorly represented, providing only 16% of the identified mammal assemblage (Table 4). Apart from the small sample from linear features, in which cattle provide 47% of the identified elements of cattle, sheep/goat and pig (Table 5), none of the sub-groups produced over 29% cattle. The percentage of cattle elements is significantly smaller than in most Roman assemblages from southern England (Table 6). Only the small sample from Hod Hill and the assemblage from Hayling Island have figures below those found on this site. Possible reasons for the high percentages of sheep/goat and the consequent low percentages of cattle and pig have been noted above. Such a low percentage of cattle is generally not typical of a Romano-British urban assemblage. It would be interesting to obtain samples from elsewhere on this settlement to see whether such percentages are reflected more broadly. However, the evidence from these excavations suggests that the inhabitants of this part of the settlement did not rely as heavily on beef as contemporary communities in the region.

Butchery marks were observed on 37 cattle elements, representing 5% of the total sample excluding teeth (Table 9). This is a low percentage compared with most cattle assemblages. For example, 29% of the cattle elements from Greyhound Yard, Dorchester were recorded as butchered (Maltby 1993). This reflects the very fragmentary nature of the sample from Weston Road. It also reflects the absence of some characteristic processing marks found in abundance on urban sites. Of the butchered specimens, knife cuts were observed on 17 and chop marks on 19. A mandible from context (209) has been sawn through the diastema. The use of cleavers became more prevalent during the Romano-British period but they were used with greater consistency by professional butchers resident on urban sites in particular (Maltby 2007). The relatively low number of bones with cleaver marks is more typical of rural assemblages. The use of knives to disarticulate cattle carcasses was prevalent on Iron Age sites and their use continued into the Roman period at some rural settlements. This is the case at Weston Road. Cuts on the medial aspects of a proximal radius and femur, on the anterior aspects of two astragali and on the lateral aspect of a mandibular ramus are typical of such butchery practices.

Cattle Elements Represented Table 13 summarises the cattle element data. The best represented elements are loose teeth (249), which provide 26% of the cattle assemblage. This indicates the fragmentary nature and moderate preservation of the cattle bones. Excluding these, mandibles (79) skull fragments (51) and first phalanges (48) provide the highest NISP counts. Phalanges and other small foot bones such as the carpals and sesamoids are generally much better represented than in the sheep/goat sample, reflecting differential recovery rates and indicating that sheep/goat are probably under-represented in comparison with cattle in the assemblage. The larger numbers of such bones largely accounts for the fact that 34% of the cattle assemblage (excluding teeth) consists of foot elements. Cranial elements are also better represented (25%) than in the sheep/goat assemblage. This could imply that there is a greater bias towards elements of low meat utility in the cattle assemblage. However, this is more likely to be due to better retrieval of foot elements and the greater fragmentation of skull and mandible fragments. The highest MNE counts were obtained from the metacarpal (15), mandible (15) and humerus (15), elements of both low and high utility (Table 13). There is therefore no convincing evidence that the assemblage is biased by the disposal of large amounts of primary processing waste. On the other hand, there is also little evidence that joints of meat of cattle butchered elsewhere were commonly imported. The cattle assemblage is too small to carry out detailed comparisons of body parts found in different sub-groups.

Knife cuts were also found on two occipital condyles made during separation from the spine. Seven first phalanges bear knife cuts on the peripheral and/or anterior surfaces of the shaft. These were made during skinning. Knife cuts on the frontal bone on the top of the skull are also likely to have been made during skinning. All the butchery observed on seven cattle vertebrae are marks made with cleavers. Most, as in the case of sheep/goat, are related to the division of the trunk into two sides. An atlas and axis have been chopped through close to the midline but in three other cases such marks are situated off-centre. Two thoracic vertebrae have been chopped through transversely during segmentation of the spine. Although chop marks were observed on several limb bones, there is only one dubious case (on an ulna) that has evidence for broad blade marks than run longitudinally along the bone. Such skims are

Cattle Fragmentation and Butchery Evidence No associated bone groups of cattle were recorded. Fragmentation levels of the upper limb bones and mandibles are very high, as indicated by comparisons 32

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET 1) Many of the herds kept on the island may have been sent to Dorchester for slaughter to satisfy demands for prime beef (young adult) cattle. 2) Cattle were not commonly kept on the island and the remains include large numbers of imported animals. However, if this was the case, they were brought to the island alive, as there is no clear evidence for the import of joints. Nor is there evidence for specialist butchery. 3) The exploitation of cattle on Portland was restricted by the lack of good grazing land. To preserve pasture, many cattle were slaughtered at a young age. 4) The slaughter of young cattle and the deposition of their bones reflect special circumstances related to the nature of this part of the settlement. Romano-British temples often have produced unusually large numbers of bones of immature animals, particularly of sheep/goat and pig, perhaps indicating the selection of young animals for sacrifice. However, few of these sites have large numbers of cattle bones and there is no evidence from any of them that high percentages of young cattle were selected for slaughter (King 2005).

characteristic of filleting marks made by specialist butchers who practised in towns such as Dorchester (Maltby 2007). There is also no evidence for longitudinal splitting of upper limb bones. This practice has also been encountered on a number of urban and villa sites. This suggests that bulk processing of marrow was not taking place on this site. There is therefore no evidence for the presence of specialist butchers at this settlement. Generally, the nature of cattle carcass processing is more similar to that encountered on rural settlements such as Owslebury in Hampshire than that found on urban sites (Maltby 2007; in press). Cattle Ageing and Metrical Data The fact that only four mandibles survived with any cheek teeth still embedded in the jaw is testament to the heavily fragmented nature of the cattle assemblage (Table 10). Larger samples of epiphyseal fusion data were obtained (Table 11). There is quite a high percentage (17%) of unfused early-fusing epiphyses, belonging to calves mainly under a year old. There are also quite large numbers of porous bones (84) that comprise 12% of the cattle elements excluding teeth. This demonstrates that bones of a significant number of juvenile animals were deposited. Some of these calves may have been slaughtered as a by-product of dairy production. Veal may also have been popular, although most RomanoBritish samples tend to produce high percentages of adult cattle (Maltby in press). However, a significant number of jaws and bones of young calves were also found at Greyhound Yard, Dorchester (Maltby 1993; 1994). Again some cattle may have slaughtered at a young age to preserve pasture.

Unfortunately, very few cattle bones could be measured to determine whether there was a bias towards males or females amongst the bones from adult animals. A complete metacarpal has a greatest length of 168mm, which is from a small animal with a withers height of 103mm. This lies towards the lower end of the size range of cattle from Dorchester (Maltby 1993). Three horncores are from animals with very small horns. Pig Abundance Pig elements (676) provide 11% of the identified mammal assemblage (Table 4).

Although there is only a small sample of distal tibiae and metapodials, there is a high percentage (69%) of unfused specimens indicating that over two-thirds of these bones belonged to cattle aged under 24-30 months old. Similarly, only 19% of the latest-fusing epiphyses are fused, indicating that only a small number of adult cattle are represented. Such a bias towards immature cattle is very unusual on Roman sites, where most samples usually contain high percentages of adult cattle. Unfortunately analysis of epiphyseal fusion data was not carried out on the Dorchester Greyhound Yard sample. However, the analysis of mandibular tooth ageing data showed that half the cattle mandibles belonged to fully mature cattle probably all over four years old (Maltby 1993) and most urban sites have even higher percentages of adult cattle mandibles (Maltby in press). In assemblages from Roman Exeter 69% of the latest-fusing epiphyses have fused (Maltby 1979) and at Winchester this figure is even higher (75% - Maltby in press). High percentages of mature cattle have also been found on rural settlements such as Owslebury (Maltby 1994).

Their remains were found quite consistently in the various sub-groups, providing between 8% and 17% of the cattle, sheep/goat and pig fragments (Table 5). Comparisons with contemporary assemblages from southern England show that the abundance of pig elements from Weston Road is fairly low (Table 6). Most urban samples have higher percentages of pigs perhaps reflecting variations in dietary preference (King 1984) and the availability of pigs (Maltby 1994). However, low percentages of pigs have been found in some suburban as well as rural assemblages. For example, pig percentages from Colliton Park and Alington Avenue, Dorchester are significantly lower than those from Greyhound Yard. Pig Elements Represented Loose teeth (148) provide 22% of the pig assemblage (Table 14), indicating only moderate preservation of their bones. Excluding teeth, cranial elements are more abundant than in the sheep/goat and cattle assemblages (33%). This is typical of many faunal assemblages as these elements survive better in pigs. Foot bones are slightly less abundant (19%) than in the sheep/goat assemblage, even though pigs have more bones in their

Therefore, to find such high percentages of immature cattle is very unusual. Several explanations can again be put forward:

33

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET There is, however, evidence for a significant cull of pigs aged between about 6-12 months old (Stage 3). Eight (40%) of the mandibles and nine (41%) of the maxillae belonged to pigs of this age. Assuming that the majority of pigs were born in spring, this would indicate that many were killed in the autumn and winter. The high kill-off of first year animals is supported by the epiphyseal fusion data, which shows that 35% of the early-fusing epiphyses are unfused (Table 10). Sub-adult animals (Stages 4-5) are represented by nine (45%) of the mandibles (Table 9) indicating that a substantial proportion of second year mortalities are represented. This is also supported by the later-fusing epiphyses, which include 78% unfused specimens (Table 10).

feet. This is also evident in the MNE counts with no foot element having a MNE higher than seven compared with 21 for all the cranial elements. Some of the upper limb bones such as the humerus (MNE=16) are also better represented. This may imply that some joints of pork or bacon were brought in after their trotters were removed. However, several other upper limb bones are no better represented than some of the tarsals and metapodials and there is no conclusive evidence that this practice was particularly common. Pig Fragmentation and Butchery Evidence No associated bone groups were identified. The degrees of fragmentation of the upper limb bones are similar to those of sheep/goat. Butchery marks were observed on 32 (6%) pig elements (Table 9). Seventeen of these consisted of knife cuts and 14 of chop marks. A pelvis from S60 (context 223) has been sawn through the acetabulum. The use of saws is unusual in Romano-British butchery. Therefore it is possible that this bone represents a later intrusion. Knife cuts on the anterior surfaces of two pig astragali are similar to those commonly found on sheep/goat and were inflicted when the feet were separated from the upper hind limbs. Knife cuts were also found on a centroquartal. Similar knife cuts were observed on the lower shafts of four tibiae. Knife cuts associated with dismemberment were also observed on a scapula, pelvis and two lumbar vertebrae. However, two other tibiae have evidence of chop marks on the shaft indicating alternative methods of dismemberment were used. Similarly the distal ends of a femur and a humerus have also been chopped through. A mandible has evidence for being split through the symphysis. Two atlases bear superficial transverse chops made when the skull was removed. Another atlas and three other vertebrae and a rib have chop marks associated with the division of the trunk into sides. One skull shows evidence of being chopped open to remove the brain. Three other skull fragments have knife cuts. Overall butchery methods are similar to those observed on the other main species.

Only three (15%) of the mandibles (Table 9) and three (14%) of the maxillae belonged to adult pigs over two years of age (Stages 6-7) and only one (3%) of the latestfusing epiphyses have fused (Table 10), indicating that nearly all the pigs were under 36-42 months old. Pigs can tolerate high levels of immature slaughter but the percentage of immature pigs represented here is unusually high. At Greyhound Yard, Dorchester, 25% of the pig mandibles belonged to adults (Maltby 1993; in press). Adult animals are even more common in most urban samples (Maltby in press). Slightly higher percentages of adult pigs were also found at Hayling Island, although first and second year mortalities are in the majority (King 2005). Unfortunately only three of the mandibles discussed above have surviving canines to determine whether they belonged to males or females. The oldest mandible belonged to a sow, which presumably was kept for breeding. Two younger specimens belong to males. Five other female and four male mandibles were also recorded. Twenty out of 21 loose lower canines belonged to males. However, these have a much better chance of recovery than the smaller canines of females. Metrical data of pig bones are restricted because of the high levels of immature slaughter. The lateral lengths of seven astragali range between 36.6mm and 44.2mm (mean = 40.4mm). These represent pigs of similar sizes to those found on other sites in southern England (Maltby in press). Several pig bones belonged to large animals, although unfortunately none could be measured. These probably belonged to large domestic animals, although the presence of wild boar cannot be totally discounted.

Pig Ageing, Sexing and Metrical Data Tooth eruption data are available from 20 mandibles (Table 10) and 22 maxillae. In contrast to the sheep/goat and cattle samples, there are no neonatal animals represented by these jaws (Stages 1-2). Only 17 (3%) of the pig elements were recorded as porous. As discussed above, the presence of neonatal goats and calves may be by-products of dairying, which would not been an issue in pig husbandry. The near absence of neonatal pigs contrasts with Greyhound Yard, Dorchester, where 25% of the pig mandibles were at Stages 1-2 (Maltby 1993; 1994; in press). It has been suggested that some pigs were kept in Romano-British towns to meet increased demands for pork (Maltby 1994). In contrast there is little evidence to suggest that pigs were kept in the vicinity of Weston Road.

Other Mammals Horse Although it is possible that equid remains could include mule as well as horse, there is no clear evidence for the presence of the former at this site and it is assumed that the majority or all of the remains belonged to horse. Only 37 elements were identified, representing less than 1% of the mammal assemblage (Table 4). These include a large portion of skull, mandible, humerus, thoracic vertebra, rib and two first phalanges from S197 (contexts 198 and 34

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET There are several other possible associated bone groups of dog. S197 produced 22 elements from adjacent contexts (129) and (198). Although at least two dogs are represented by left pelves, most of the bones could have belonged to one small adult dog. Bones from most part of the body are represented and the remains could represent a badly disturbed complete skeleton. A caudal vertebra from this group has evidence of a healed fracture. S200 produced a pair of femora, tibia, fibula, metatarsal and a rib from context (189) that probably belonged to the same sub-adult dog. The shaft of the fibula has fused pathologically with the tibia. The assemblage from S60 includes three metapodials from context (227) probably from another immature dog. The evidence indicates that dog carcasses were treated differently from the mammals discussed previously in that, even when disarticulated, they survive more frequently in groups, indicating that they were probably originally deposited whole. With the exception of the partial skeleton from S299, various taphonomic processes and disturbance have resulted in the disarticulation and diminution of their skeletons.

216). All epiphyses are unfused and it likely that all seven bones belonged to the same foal. The low proportion of horse is more typical of urban assemblages. Comparing counts of horse and cattle only (Table 4), horse provides only 4% of the sample. This, however, is higher than at Greyhound Yard, near the centre of Dorchester (2%) but very similar to percentages obtained from the Colliton Park site, situated near the edge of the town (Maltby 1994; in press; Hamilton-Dyer 1993a). However, percentages of horse are often more common on Romano-British rural settlements, including Alington Avenue on the outskirts of Dorchester (18%). Thirteen (35%) of the other horse elements are loose teeth, indicating poor preservation of their bones. Postcranial elements are mainly from the lower limbs, particularly small elements such as carpals and tarsals (Table 15). The presence of teeth and bones of immature equids is unusually high in this assemblage. In addition to the partial skeleton from S197, four of the loose teeth are deciduous and a proximal radius and distal metatarsal are unfused. All these indicate the presence of immature horses. Sub-adult or adult animals are represented by the other teeth, two fused distal metapodials and a fused distal radius. In most Romano-British assemblages very few immature horses or mules are represented (Maltby in press). At Greyhound Yard, Dorchester, a skeleton of a foal was deposited at the bottom of a shaft, which subsequently received large numbers of depositions of other animals, complete pottery vessels and other small finds (Maltby 1993; Woodward and Woodward 2004) along with other material more typical of other deposits. This may have been a votive offering. However, hardly any remains of immature horses were recorded in the rest of the assemblage (Maltby 1993). Horses and mules would have been valued mainly as pack and riding animals and could expect to have relatively long lives.

Four upper limb bones from the skeleton from S299 produced shoulder height estimates (following Harcourt 1974) between 37.1cm and 37.9cm. This animal was about the size of a fox terrier. Several other dogs represented are from smaller animals, although unfortunately few could be measured. A thoracic vertebra from S60 and mandible from ditch F22 are from miniature dogs. A wide range in sizes of dogs has been encountered on Romano-British sites (Harcourt 1974) including in the large assemblage obtained from Dorchester (Maltby 1993). No bones of neonatal puppies were found but older puppies are represented as well as adult dogs. Cat

No butchery marks were observed on any of the equid bones. Although butchery has been recorded on horse bones from Roman sites, they are uncommon and so their absence here is not unusual. Only two horse bones were measured. A radius and third metacarpal have distal breadths of 75.7mm and 47.3mm respectively. These are comparable in size to specimens found in Dorchester (Maltby 1993).

Eighteen bones of cat were found in four of the subgroups (Table 4). Thirteen of these are from the western part of layer (110). These consist of bones of the left hind limb of an adult (ABG1). Several knife cuts were observed on the distal part of the tibia shaft. These probably represent incisions made preparatory to skinning, after which the leg was discarded. Although most of the remains are foot bones, the presence of the complete tibia and part of the femur indicates that the limb was not disarticulated and the foot bones did not remain attached to the skin. It is possible that the skinned carcass was deposited complete and subsequently disturbed.

Dog Dog remains were found in nine of the sub-groups and in total 102 elements were recorded, representing 2% of the identified mammal assemblage. Most of their elements were found in deposits associated with structures (Table 4). These include a partial skeleton of an adult dog (ABG6) found in S299 (context 140). The group consists of 42 bones that include the skull and jaws, all of the upper limb bones apart from the tibiae, several vertebrae and ribs, and three metacarpals. There is no evidence of butchery on these bones, which are in excellent condition. Some of the bones may not have been recovered. Its location at the base of a wall could indicate it was a foundation deposit.

The greatest length of the tibia in this group measured 113.4mm, comparable in size with some of the smaller domestic cats recovered from Greyhound Yard, Dorchester (Maltby 1993). A second metacarpal from S200 is from a much larger cat, having a length of 32.6mm. This is either from a large domestic specimen or possibly a wild cat.

35

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET when the partial skeleton from S60 is excluded. This is higher than usually found on rural settlements but still well below the average recorded from other types of Roman settlement (Maltby 1997). At Greyhound Yard, Dorchester, the domestic fowl: pig percentage averaged 25% (Maltby in press). However, comparable calculations at Alington Avenue produced a total of only 3% domestic fowl.

Wild Mammals Only three red deer elements were identified. One of these consists of a fragment of eroded antler from S197. A calcaneus and a fragment of tibia may represent remains of joints of venison brought to the site. The only element identified as roe deer is a humerus from a rubble layer (context 209). Three bones of hare were also recorded. An atlas of a cetacean was found in context (291). This is from a dolphin or porpoise that was possibly stranded on the Isle of Portland. The most likely species concerned are common dolphin, harbour porpoise or bottle-nosed dolphin.

Domestic fowl elements represented (Table 17) reveal the presence of all parts of the body except the head. No identified bird skull fragments were recorded but they are unlikely to have survived intact in these deposits. Only one porous bone of an immature bird was recorded indicating that most of the identified bones are from adults. This again contrasts with Greyhound Yard, Dorchester where 12% of the domestic fowl bones were porous (Maltby 1993). Both hens and cocks are represented at Weston Road. Three unspurred tarsometatarsi belonged to the former and four spurred specimens to the latter. Five fragmented leg bones allowed investigation of the shaft cavity. Two specimens contain medullary bone, showing they belonged to laying hens. Three other bones have no evidence of such deposits. Measurements were taken on 11 domestic fowl bones. All fall within the size range of those from the much larger sample from Greyhound Yard, Dorchester (Maltby 1993). No butchery marks were observed on any of the domestic fowl bones.

Twelve small mammal bones were recorded, of which two were further identified (Table 4). A maxilla of a field vole was found in rubble layer (209) and a tibia of a black rat was recovered from soil layer (162). Although black rats have been occasionally recorded in Roman Britain, particularly on urban sites including Dorchester (Maltby 1993), there is a strong possibility that this bone could be intrusive. Bird Bird bones are fairly poorly represented in the assemblage. Only 192 were recorded representing 1.4% of the total assemblage (Table 2). This figure never rose above 3% in any sub-group. Fifty-six (29%) of the bird bones could not be further identified (Table 16).

Ducks and Geese Domestic Fowl Only 17 bones from at least three species were identified. These include seven goose bones, which provide 5% of the bird elements (Table 16). Unfortunately it has not been possible to identify these more specifically because of the close morphological similarities between bones of different species and the fragmentary nature of the material. Some of the bones, such as a tarsometatarsus from ditch F22, are large enough to be from domestic geese, although Albarella (2005) has questioned whether ducks and geese were domesticated in Roman Britain. However, wild grey lag geese (Anser anser), usually accepted as the species which was domesticated, are rarely observed in and around Portland (PBO 2002). Indeed the dark-bellied Brent goose (Branta bernicla) is the only species nowadays to be more than a rare winter or migrant visitor. Quite large numbers can be found near Ferrybridge in the winter months. It is possible that some of these bones could be from that species. Goose bones form only 1% of the bird bones identified at Greyhound Yard, Dorchester (Maltby 1993).

Domestic fowl (chicken) bones (62) are comfortably the most commonly represented, providing 46% of the avian assemblage. Twenty-two of these bones belonged to a skeleton (ABG2) of an adult bird found amongst rubble near a large stone on the south side of the east entrance of S60 (context 186). Most bones of the wings were recovered along with six vertebrae and the pelvis. Although present on a few British sites from the early Iron Age, they are found, at best, in very small numbers on Iron Age sites in central southern England (Maltby 1981). They are more common in Romano-British samples, although there is a substantial amount of variation in their abundance in assemblages from different types of settlement. They tend to be found most commonly on large urban sites (Maltby 1997). Comparisons with sheep/goat elements show that domestic fowl contributed only 1.4% of the remains of these species (including the partial skeleton). This is more typical of rural assemblages and is much lower than at Greyhound Yard, Dorchester where the mean domestic fowl: sheep/goat percentage was 18% (Maltby 1993; 1997; in press). In contrast, the percentage of domestic fowl is below 1% in occupation deposits at Alington Avenue, a rural settlement just outside Dorchester, (Maltby 1997), although skeletons of domestic fowl were recovered from some of the graves (Maltby 2002). Domestic fowl provides 8% of the total pig and domestic fowl elements at Weston Road. This figure falls to 6%

Ten bones of duck were identified representing 6% of the bird assemblage (Table 16). The bones are all smaller than those of mallard (Anas platyrhychos) but larger than those of teal (Anas crecca). It is very difficult to distinguish between medium-sized ducks from their fragmentary skeletal remains. Three carpometacarpi have greatest lengths of 50.2mm, 50.4mm and 53.6mm. This size variation could indicate the presence of more than one species. A number of species of sea ducks are found 36

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET However, in the cases of Halangy Down and Portland, it is more likely that the great auk bones are from birds that were nesting on the islands alongside other colonies of seabirds. They would be very difficult to catch at sea. This indicates that the breeding range of the species extended to the southern English coast in the Roman period. The Weston Road find indicates that at least one bird was butchered. Auks and other seabirds could also have been exploited for their feathers, oil and eggs.

around Portland (PBO 2002). Most of the duck bones, however, are closely comparable with those of widgeon (Anas penelope). Large numbers of widgeon nowadays winter on the Fleet lagoon on the mainland side of Chesil Beach. Five of the duck bones belonged to the left wing of one adult bird (ABG8) found in S60 (context 223). The furcula has been chopped through near its midpoint showing that the carcass has been split. Overall, duck remains are less common at Weston Road than they are in Dorchester, where a wider range of ducks, including mallard/domestic duck, has been found (Maltby 1993).

A left humerus of a gannet (Morus bassana) was identified from a possible midden deposit (context 293). This species does not now breed on the island but is frequently seen offshore (PBO 2002). Although the bird involved may have been caught offshore on the fishing grounds, it is more likely to have been either a weak or dead bird that was washed ashore or a bird that was captured on land on or near its nesting site. Gannets nest on cliff tops and it is plausible that colonies did reside on Portland in Roman times. A number of archaeological sites in Scotland have produced both great auk and gannet remains (Serjeantson 2001). The humerus bears a superficial chop mark near its distal end, inflicted when the lower part of the wing was removed.

Seabirds Bones from at least seven species were recovered (Table 16). Most of these belonged to the auk family. Six bones of razorbill (Alca torda) were identified. A seventh bone could have belonged either to a razorbill or a guillemot (Uria aalge). Breeding colonies of both species are now found around Portland Bill (PBO 2002) and their distribution could have been more widespread in the past. All the bones are from adult birds. Eleven bones were identified by Sheila Hamilton-Dyer as great auk (Pinguinus impennis). The bones were found in three contexts in the vicinity of structure S200. Seven of these consist of pairs of coracoids, scapulae and humeri and part of the furcula of an adult bird (ABG7) in a demolition layer (context 189). None of the bones have survived complete and ancient breaks were observed on the humeri, only the proximal parts of which were found. The left humerus bears knife cuts just above the break on the medial aspect of the shaft. The cuts are deep and would have facilitated the removal of the lower parts of the wing. A shallower oblique knife cut was probably a trial cut. A vertebra in the same context could have belonged to the same bird. Similarly, a complete right femur and a small portion of sternum from context (199), and the posterior part of a left mandible in context (205) could have belonged to the same great auk.

Great black-backed gull (Larus marinus) is represented by seven bones, including a pair of mandibles. Nowadays birds of this species on Portland include breeding residents supplemented by winter visitors and migrating birds. Six bones from smaller gulls were recorded, four of them from S200. These belong either to herring gull (Larus argentatus) or lesser black-backed gull (Larus fuscus), both of which can be found commonly on and around Portland throughout the year (PBO 2002). A humerus of a fulmar (Fulmarus glacialis) was also found in S200. Nowadays this species breeds in small numbers on Portland (PBO 2002). Six wing bones of great northern diver (Gavia immer) were identified. These include the slightly charred remains of a left humerus and ulna, probably from the same bird in S200. Great northern divers currently winter in small numbers around Portland harbour (PBO 2002).

The discovery of great auk on the Isle of Portland is a significant find. This large (60cm long) flightless bird became extinct in the 19th century largely due to human over-exploitation. Its distribution ranged across the Northern Atlantic (Serjeantson 2001). Although there are both pictorial and archaeological records of great auks from the Isle of Man as late as the 17th century (Fisher 1997), nearly all of the archaeological finds from Britain are from Scotland, particularly from sites in the Orkney Islands and the Outer Hebrides. Even in those areas, finds of great auk become very scarce in the historic period (Serjeantson 2001). The only other example recorded from the British Isles is from 3rd century AD deposits from Halangy Down, St Mary’s, one of the Scilly Isles (Locker 1996; Serjeantson 2001). Bones of great auks have also been found on four Roman sites in the coastal zone of the Netherlands (Groot 2005). Here, given the lack of suitable nesting sites (low rocky shores), it was suggested that the remains were from weakened birds that were washed ashore in the winter.

Seabird species therefore provide 29% of the identified bird bones. This is a very high figure for a RomanoBritish settlement and shows how local coastal resources supplemented the food supply. None of the species discussed in this section were present in the sample of over 2,500 identified bird bones from Dorchester Greyhound Yard, although there was one bone of a shag (Phalocrocorax aristotelis) (Maltby 1993). The rarity of great auk from Roman sites has already been noted. In addition, in his survey of bird bones from Roman Britain, Parker (1988) listed only one site which has records of the presence of great northern diver. This was the Saxon shore fort at Portchester Castle and the remains there consisted of a partial skeleton (Eastham 1975). This was found in a well along with complete pottery vessels and several complete animal skulls of several species. Fulford (2001) has suggested these may have been ritual depositions. Guillemot has been identified in Romano37

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET (Pica pica) a good match for the other. These species are all likely to have been resident on the island.

British levels at Ower on the edge of Poole Harbour (Coy 1987a) but not on other sites listed by Parker (1988). A single bone of an auk (cf. guillemot) has been found in recent excavations at Caerwent (Maltby in press). None of the sites surveyed by Parker (1988) produced identifications of razorbill, gannet or fulmar. The only record of herring gull came from the fort at Caerleon (O’Connor 1986) and the only find of greater backblacked gull was noted from early excavations in Exeter (Bell 1915) but not from the much more substantial (and reliable) later excavations of the Roman fortress and town (Maltby 1979). A recent survey of bones from RomanoBritish major towns has failed to add to the records listed above (Maltby in press).

Three bones of the thrush family (Turdus sp.) were identified. Two are a good match for song thrush (Turdus philomelos) and the other is very similar to blackbird (Turdus merula). Although these are the most likely species to be represented, the close similarities between the skeletons of different Turdidae means that other species cannot be totally ruled out Fish Despite the lack of sieving, surprisingly large numbers of fish bones were recovered. From all deposits, 763 fish bones were recorded, providing 6% of the total assemblage (Table 2). By comparison, fish bones formed only 2% of the faunal assemblage from Greyhound Yard, Dorchester in hand-collected samples (Hamilton-Dyer 1993b). The higher percentage of fish at Weston Road cannot be the result of better preservation as bone survival at Greyhound Yard was generally much better. It may reflect more effective retrieval of bones and/or the fact that fish were consumed and their remains deposited on the site more frequently. It must be emphasised, however, that the percentage of fish bones is almost certainly an underestimation. Similarly, it is likely that the sample is biased significantly towards the retrieval of bones of larger fish. Four hundred of the fish bones could be further identified (Table 18).

Other Bird Species The only bone of a wader identified at Weston Road is an ulna of a woodcock (Scolopax rusticola). This species is now a relatively uncommon winter visitor to the island (PBO 2002). However, it appears to have been one of the most common game birds exploited in Roman Britain. Parker (1988) listed 28 sites where it has been recorded and the number of sites has been expanded substantially in recent years. At Greyhound Yard, Dorchester, woodcock was the fifth most common bird species identified, providing 4% of the identified bird bones (Maltby 1993; in press). Its presence has also been recorded in other excavations in Dorchester including those at Colliton Park (Hamilton-Dyer 1993a; Maltby in press).

Sea Bream (Sparidae) Seven bones identified as those of the pigeon family were recorded. Although it is possible that these are from domestic birds, most of the bones are from small and slender birds comparable with those of the wild rock dove (Columba livia), which could have been nesting on the same cliffs as the auks and other seabirds. Bones of doves and pigeons have been found on many Romano-British sites (Parker 1988) including towns such as Dorchester, where they formed 3% of the bird assemblage from Greyhound Yard (Maltby in press).

Bones from at least four species of sea bream were recovered from nearly all the sub-groups. A total of 107 bones were recorded, representing 27% of the identified fish (Table 18). Most of these are vertebrae, which were not identified to species. However, it was possible to identify a number of other bones (Table 19). Eight jaw elements were identified as Couch’s sea bream (Sparus pagrus). Eight other elements were identified as red sea bream (Pagellus bogaraveo). Three jaws belonged to gilthead (Sparus aurata) and a parasphenoid of a black sea bream (Spondyliosoma cantharus) was also identified. General estimates of the sizes of the sea bream ranged from medium to very large, with most towards the larger end of the range. It should be remembered, however, that the sample will be biased towards the recovery of bones of larger fish.

Two bones of raven (Corvus corax) were found in S197, one of which was from an immature bird. There was one resident breeding pair of raven on the island in 2002 (PBO 2002) but numbers of this species are likely to have been much greater in the Roman period. Parker (1988) recorded them on 39 sites, which made them the third most frequently occurring species behind domestic fowl and mallard-sized ducks. They would certainly have been attracted as scavengers to settlement sites, although their ubiquity on Romano-British sites has also led to suggestions that they were sometimes deposited as votive offerings (Fulford 2001; Woodward and Woodward 2004; Maltby in press).

Locker (2007) noted that sea bream have been found more commonly on sites in SW England than elsewhere in Roman Britain. They were listed on seven of the 13 sites she surveyed from this area. Significantly in nearby Dorchester, they are well represented in the Greyhound Yard assemblage and are also present in the Colliton Park sample (Hamilton-Dyer 1993b; 1993a). They are warm water fishes and their greatest concentrations around Britain nowadays are around the SW coast. Most of the red sea bream bones recovered belonged to fish of between 40cm and 50cm long, which are towards the top end of the size range for this species. Red sea bream can

Five bones of smaller corvids were identified (Table 16). An ulna is from a bird the size of a rook (Corvus frugilegus) or crow (Corvus corone). The other four bones are all from smaller birds, with jackdaw (Corvus monedula) being the closest match for three and magpie 38

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET which migrated to these waters and were caught there in the summer.

be found inshore but larger fish are often found in deeper offshore waters. The black sea bream parasphenoid is from a very large specimen towards the top of the species’ size range of 50-60cm. This species can be found around wrecks and rocky outcrops (Wheeler 1978). Black sea bream is currently the most common of the Sparidae in southern British waters.

Cod Family (Gadidae) Bones of gadids (154) were the most commonly identified, providing 39% of the fish elements (Table 18). Many of the vertebrae and some other elements could not be identified beyond family level (Table 19). Of those that could be further identified, most (67) belonged to cod (Gadus morhua). They ranged in size between medium and large specimens. Most of the bones fall into the latter category and belonged to fish over a metre long. These cod are therefore substantially larger than those that are currently caught around the inshore waters of Portland. Adult fish of this size are often found in deep waters and can migrate large distances. The Atlantic and North Sea populations were increasingly exploited from the medieval period onwards. However, some groups of cod are known to be more stationary and remain relatively close inshore and it most likely that these fish are from such a group. It implies that the waters around Portland were substantially better stocked with larger cod than they are now. Cod bones have been identified on 21 of the sites listed by Locker (2007), with the largest numbers coming from sites in London. They have not, however, been previously recorded on other sites in the south and SW of England

The gilthead bones consist of a pair of maxillae from a large fish and a very large premaxilla (greatest height = 40.5mm), which must have been from a fish towards the top of its 70cm long size range. This species is found infrequently in British waters, although it has been increasing in numbers recently around the Channel Islands and off the Cornish coast. It schools in quite shallow waters preferring sandy substrates (Wheeler 1978). It has been identified only on three of the 109 Romano-British sites surveyed by Locker (2007). One of these was from a fish imported to the villa at Castle Copse in Wiltshire (Jones 1997). Another was from Halangy Down (Locker 1996) and two were found at Ower, Dorset (Coy 1987b). The Couch’s sea bream bones belonged to fish of various sizes but most are from large specimens. One of the largest is represented by an articular and dentary probably from the same individual found in S200. The articular has an anterior width of 9.0mm; the dentary has an anterior height of 15.0mm and a tooth pad width of 8.7mm (Morales and Rosenlund 1979). An even larger dentary, also from S200, has a tooth pad width of 10.5mm. Four premaxillae from S200 have tooth pad widths of 5.9mm, 11.6mm, 11.8mm and 12.6mm. Couch’s sea bream rarely exceed 50cm in length, although fishes as large as 75cm have been recorded (Wheeler 1978). Unfortunately reference material for specimens as large as those found at Weston Road was not available but the largest bones could be from fish from near the top end of the range. They have been until very recently been infrequently caught in the English Channel but large specimens have been found recently near the Channel Isles and off the Cornish coast.

There are substantially more precaudal than caudal vertebrae in the gadid assemblage (Table 19), although their skeletons contain twice of many of the latter. This is almost certainly a consequence of retrieval rates. Precaudal vertebrae are larger than caudal vertebrae and are more likely to have been recovered. Nevertheless, the presence of substantial numbers of precaudal vertebrae and head bones does suggest that whole fish tended to be brought to the site, which again supports the argument that these were local catches. Pollack (Pollachius pollachius) is the other gadid species identified in some numbers with 17 elements identified. The specimens at Weston Road range widely between those from small immature fish (likely to be underrepresented) to those from very large adults. At least five specimens fall into the latter category, and were from fish probably over a metre long. Although pollack can grow up to 130cm, fish of about 50cm are much more common (Wheeler 1978). Such large fish are found most commonly above rocky reefs and wrecks including those around the Isle of Portland. Although they are currently found quite commonly around the southern and western coasts of Britain, the only other specific identifications of pollack from Roman Britain are from two sites in the Scillies (Locker 2007).

Although Locker (2007) does not specifically list Couch’s sea bream amongst the species found on Romano-British sites, bones of this species were identified at Silchester and Greyhound Yard, Dorchester (Hamilton-Dyer 2000). However, Weston Road has produced the largest sample of bones of this species. There is no evidence from the elements represented that these were preserved fish that were caught in more southerly waters and imported. Bones from all parts of the fish are represented. It is possible that a greater range of sea bream including large adult fish was available for exploitation in the seas off Portland. This infers that northward migrations of some species of sea bream must have been more extensive in the Roman period. This may be due to warmer sea temperatures. It is unlikely, however, that sea bream were spawning in the English Channel, which would suggest that these were ones

Four other gadid species are each represented by a single bone (Tables 18-19). A vomer of a large saithe (Pollachius virens) was found in a soil layer. The size, habitat and range of this species are similar to those of pollack and fish of this size are more likely to be encountered above rocky reefs and wrecks. The only 39

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET found in inshore waters including those around the Dorset coast. Locker (2007) listed six Romano-British sites where scad have been identified. However, these total only 20 bones. This therefore is the largest assemblage of scad as yet found in Roman Britain. With the exception of one bone from York, all the sites that have produced scad are located in the south of England. Ten bones were identified in the sample from Greyhound Yard, Dorchester (Hamilton-Dyer 1993b).

other specific identification of saithe from a RomanoBritish site is from Elm’s Farm, Essex (Locker 2007). An articular of a large hake (Merluccius merluccius) was found in S197. This species can reach 180cm, although it rarely grows larger than a metre. It has a wide distribution but is most common off the south and west coasts where it is usually caught in deep waters today, although this is a consequence of overfishing. It is more likely to be found in shallower waters in summer and rises near the surface at night (Wheeler 1978). Although found quite commonly on medieval sites in SW England (Locker 2001), only three Romano-British sites have produced hake bones (Locker 2007). These include five bones from Exeter (Wilkinson 1979). A post-temporal of a large ling (Molva molva) was found in rubble layer (121). Again this species generally inhabits deep waters but also occur near rocky grounds and wrecks (Wheeler 1978). It has only been recorded on at least three Romano-British sites, one of which is from St Martins in the Scillies (Locker 2007). It has also been recently recorded at Bucknowle in Dorset (Hamilton-Dyer pers comm.).

Wrasses (Labridae)

Most of the gadid bones not identified to species are probably mainly from medium to large pollack and cod. A few of the smaller specimens, however, could have belonged to whiting (Merlangius merlangus), a species commonly found around the island.

Twenty-five bones of wrasses provide 6% of the identified fish. Vertebrae were not identified to species but all the jaw and pharynx bones belonged to ballan wrasse (Labrus bergylta). It is likely that all the vertebrae belonged to the species. The distinctive and robust inferior pharyngeal was the most common element found (9). Most of the wrasse bones found belonged to fish of 30-40cm length, although three bones of smaller fish are also represented. Such fish are abundant around Portland. Their preferred habitat is in areas containing large amounts of algae (Wheeler 1978). Wrasse bones have been found on seven other Romano-British sites in SW England (Locker 2007) and over a hundred were found at Greyhound Yard, Dorchester, including two partial skeletons in a shaft (Hamilton-Dyer 1993b).

Bass

Other Fish Species

Fifteen bones of bass (Dicentrarchus labrax) provide 4% of the identified fish bones. This species is commonly fished both commercially and by other anglers around Portland, particularly in the summer. The largest fish grow to about a metre but usually are no larger than 60cm long (Wheeler 1978). The bass represented from Weston Road range between medium and large in size, although smaller fish are probably under-represented. Bass bones were present on 11 of the Romano-British sites surveyed by Locker (2007), mainly from the south of England. The largest assemblage of bass has been found at Greyhound Yard, Dorchester (Hamilton-Dyer 1993b), although most of these were from a minimum of six fish deposited together in a shaft. They were not present in the Colliton Park assemblage (Hamilton-Dyer 1993a) but small numbers have been found in Exeter (Wilkinson 1979) and at the Uley shrine in Gloucestershire (Locker 2007).

Sixteen bones were identified as conger (Conger conger). The bones are from fish ranging between 50cm and 200cm long. They are commonly found near rocks, reefs and wrecks (Wheeler 1978). Conger bones were only identified on five sites in the 109 surveyed by Locker (2007), three of them in the SW region, in Exeter (Wilkinson 1979), Greyhound Yard, Dorchester (Hamilton-Dyer 1993b) and Halangy Down (Locker 2007). Although a specimen from East Gate, Gloucester can be added to this list (Maltby in press), conger is still a fairly rare species in Romano-British assemblages. Butchered conger bones have also been identified in a 3rd century AD assemblage at Bucknowle, Dorset in a possible foundation deposit (Hamilton-Dyer pers comm.). A single bone of common eel (Anguilla anguilla) was identified. This has been one of the most common species encountered in Romano-British assemblages, being recorded on 72 sites in Locker’s (2007) survey. They have been found in large numbers on several sites where sieving programmes have been undertaken. Significantly perhaps, their bones were rarely found in the handcollected sample from Greyhound Yard, Dorchester but were recovered in soil samples and are abundant in the sieved samples from Colliton Park (Hamilton-Dyer 1993a; 1993b). A single vertebra of common eel was retrieved from Weston Road. It is from a fish of about 60cm in length. Fish of this size could have been caught in inshore waters but they can also be caught in rivers.

Scad Scad (Trachurus trachurus) provides a significant proportion of the fish bones. They were found in ten of the sub-groups and a total of 76 bones represent 19% of the identified fish elements (Table 18). Nearly all of the elements are vertebrae (Table 19). This is because the caudal vertebrae in particular survive particularly well. This survival is enhanced by the fact that most of the vertebrae have evidence of hyperostosis, a condition that develops in many older fish and is not uncommon in the vertebrae of adult scad. The thickening of the bone makes them more dense and robust. A knife cut was observed on one of the caudal vertebrae. Large schools of scad can be

Two bones of John Dory (Zeus faber) were found in S60. A dentary belonged to a medium-sized fish of about 40

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET Comparisons have been throughout this report with the faunal assemblages from Dorchester and it is perhaps appropriate to conclude by highlighting the similarities and differences between the assemblages from these sites.

40cm. However, a precaudal vertebra is from a large fish of about 60cm in length. John Dory are usually found near the surface of inshore waters (Wheeler 1978). The only other record of its presence on a Romano-British site is from St Martins in the Scillies (Locker 2007). They currently spawn in the English Channel but not further North. The fish involved could be summer migrants.

Although we must accept that bones of smaller species are under-represented, it is clear that the majority of the meat consumed in the areas excavated of both settlements was derived from domestic mammals. At Weston Road, 93% of the bone fragments belonged to mammals (Table 2) and 98% of the identified mammal fragments belonged to sheep/goat, cattle and pig (Table 4). At Greyhound Yard, Dorchester, 95% of the identified mammal fragments belonged to these species, the slightly lower figure being largely attributable to a larger number of dog skeletons (Maltby 1993). The equivalent figure at Colliton Park is 97% (Hamilton-Dyer 1993a) and an identical percentage was obtained from an assemblage from Charles Street in the town (Maltby in press).

Two opercular of mullet were recorded (Table 18). It was not possible to identify these two species. Mullet bones have been found fairly commonly on Romano-British sites (Locker 2007). Over 50 were recorded at Greyhound Yard, Dorchester, although most of them belonged to a single skeleton of a golden grey mullet (Liza aurata) recovered from a shaft (Hamilton-Dyer 1993b). These species are also more likely to have been present and caught in the summer months. Surprisingly, only one bone of a flatfish was identified. This was an anal pterygiophore probably of a plaice (Pleuronectes platessa). Plaice and other flatfish can be found around Portland. Flatfish bones have been recovered in fairly small numbers on sites in Dorchester (Hamilton-Dyer 1993a; 1993b).

However, there are significant differences in the relative abundance of the three major species. There is evidence from the NISP counts that suggest strongly that relatively more sheep were being deposited at Weston Road than in Dorchester (Table 6). If one adapts the MNE counts to produce estimates of minimum numbers of individuals, there are a minimum of 73 sheep/goat represented compared with 11 pig and only eight cattle. Even allowing for the fact that cattle produce about eight times the amount of meat than sheep of similar ages, this still suggests that more lamb and mutton was being processed than veal and beef. This is very unusual for a site of this date. The contrast with the evidence from Dorchester is clear. In only one of the assemblages from sites in the town and in the neighbouring rural settlement at Alington Avenue do sheep/goat elements form over half of the total of sheep/goat, cattle and pig (Table 6).

A vertebra of a small shark was found in a soil layer. It was not identifiable to species. The most likely species involved would be the dogfish (Scyliorhinus canicula), which is commonly caught today in the shallow waters around Portland. There is no evidence of bones associated with the production and consumption of fish sauces, which have been found in a few Romano-British sites, particularly in towns (Locker 2007). A large deposit of bones of very small herring (Clupea harengus) and sprat (Sprattus sprattus) has recently been discovered at the Dorchester Hospital site and is thought to represent the waste from the production of allec, a fish sauce (containing bones) produced from the fermentation of small fish (HamiltonDyer pers. comm.).

A higher percentage of the sheep bones deposited at Weston Road belonged to lambs, particularly those slaughtered in the autumn and winter. Most of these lambs were probably males. Whilst residents of Dorchester also ate lambs of 6-12 months of age, they obtained more meat from sub-adult sheep killed between 18-24 months than appears to have been the case on Portland. Mutton from older sheep was consumed in quantity at both settlements. However, it appears that the bones of relatively more sheep over two years of age (55%) were deposited at Greyhound Yard, Dorchester than at Weston Road (42%).

Conclusions Despite its moderate preservation and its very fragmentary nature, the excavations have produced an important bone assemblage of distinctive character, which has a number of very unusual features. Any conclusions must be tempered with the observation that only a very small portion of what was probably a much larger settlement has been investigated. Large amounts of intra-settlement variability is a feature of Romano-British faunal assemblages (Maltby 2007; in press) and there is no guarantee that the assemblage from Weston Road will necessarily be replicated in other areas, particularly as it appears the site was used on occasions during the Roman period for the burial of children and perhaps may have been regarded as a special place associated with burial and other rituals.

There is evidence that a wider range of sheep was available to the inhabitants of Dorchester. They obtained meat from both horned and hornless types of sheep, and possibly hybrids of the two types (Maltby 1993). From Weston Road there is much less evidence for hornless sheep, although they are represented in small numbers. Most of the sheep, however, were small, slender and horned, of a type and stature found widely in Iron Age Britain. It is likely that most of the sheep represented are from flocks raised on the island. These were adapted to the relatively poor grazing available. 41

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET least some of these were ritual depositions (Woodward and Woodward; Maltby in press). There are several other associated dog bones, all of which could have belonged to heavily disturbed carcasses. There are no deposits of neonatal puppies in evidence at Weston Road, whereas they were found in several features in Dorchester. The abundant evidence for gnawing on bones is another indication for the presence of dogs, which clearly had access to the discarded remains of animals after processing. The presence of bones of miniature dogs represents evidence that new breeds of dog found commonly on Romano-British sites including Dorchester were also found on the island.

As noted above, the current evidence suggests that beef formed a much more substantial part of the meat diet at least in the central areas of the town for those in Dorchester than at Weston Road. Of course, it is feasible that filleted beef could have been imported to Weston Road. This would leave little zooarchaeological trace. In the town, the acquisition and distribution of beef was largely in the control of specialist butchers. There is no evidence for the presence of such butchers on Portland. Although cleavers were being used for some of the butchery, there is no evidence for the very consistent butchery methods practised in Dorchester. Both settlements have evidence for the presence of calves and the consumption of veal. However, there are relatively more calves represented at Weston Road than in Dorchester. This may imply that milk production was of greater importance on the island. However, if the animals were mainly drawn from stock kept on the island, it could also reflect the need to keep the numbers of cattle within its carrying capacity. The cattle found generally had very small horns and this may indicate that most were from a single, local type. In Dorchester most beef came from adult but not elderly animals. Metrical analysis has indicated that most of these were cows (Maltby 1993; 1994). The majority of cattle represented at Weston Road had not reached adulthood. Possible explanations for this contrast have been discussed above.

Domestic cats were also probably introduced to the Isle of Portland in the Romano-British period, although an earlier introduction cannot be ruled out. The excavations produced evidence for skinning of at least one individual. Although several cat skeletons have also been found in Dorchester, none of them have as yet produced similar processing evidence. There is very limited evidence for the exploitation of wild mammals at Weston Road, although red deer, roe deer and hare bones were identified. There is also evidence that (probably stranded) cetaceans were occasionally exploited. It is unlikely that there would have been large herds of deer on the island and it possible that these are from imported joints of meat and antler. A slightly richer assemblage of wild mammal remains was obtained from Greyhound Yard, Dorchester, although they still formed only a very small proportion of the overall assemblage.

More pork and bacon appears to have been consumed in Dorchester, at least in the central areas of the town. There are remains of many more neonatal pigs in Dorchester, which could reflect a preference for the consumption of suckling pig but could also indicate the keeping of pigs within the town. Most of the pigs represented at Weston Road were killed between six months and two years of age. There are relatively fewer elements from older pigs than there are in Dorchester.

Domestic fowl appear to have been more commonly exploited in Dorchester than at Weston Road, although they were probably also kept on the island. Eggs as well as meat were exploited at both settlements. However, the rest of the bird assemblage from the two settlements is significantly different. At Greyhound Yard, Dorchester, the rank order of the most abundant bird species is domestic fowl; ducks; ravens and other corvids; woodcock; pigeon family and geese. The only seabird represented is shag. By contrast, the species order at Weston Road is domestic fowl; seabirds; duck; goose; corvids; pigeon family (Table 16). The greater emphasis on seabirds again reflects local adaptations on the island. There is evidence that several species were exploited including winter visitors such as the great northern diver and breeding species such as the razorbill and gulls. The presence of butchered bones of gannet and the extinct great auk indicates that they were eaten. Their discovery can best be explained by the presence of breeding birds on the island, which probably supported a much greater abundance of nesting colonies at that time.

Goats are represented on both sites in small numbers and the presence of neonatal mortalities indicates that they were probably kept at both settlements and that their milk may have been an important factor in their exploitation. Overall, there is evidence for the deposition of higher percentages of immature domestic mammals at Weston Road than in Dorchester. Whether this reflects local husbandry practices adapting to the particular conditions of the island or is a phenomenon associated with the special nature of the site remains to be tested by further excavation. Horse bones form only a small percentage of the mammal remains at Weston Road. There is no evidence that they were eaten. Again, they include an unusually high proportion of bones from immature animals, including a partial skeleton of a foal.

The seabirds may have been part of a generally richer marine environment. Another unusual feature of the Weston Road assemblage is the relatively high numbers of fish bones. Fishing both from shore and offshore from boats is indicated. Even though the recovery of their bones was favoured by retrieval methods, the sizes of the largest bream and particularly cod are well beyond the

A partial skeleton of a dog associated with the lower (Romano-British) walls of S299 could be a foundation burial. Dorchester produced a large number of dog skeletons, mainly from pits and it has been argued that at 42

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET size of the catches usually taken in the area in modern times. Several species represented are now rare or absent from the seas around Portland. This, however, does not necessarily mean they were imported from elsewhere. Indeed, the elements represented would suggest the contrary with complete fish being processed at the site. A combination of slightly warmer sea temperatures and fish stocks not adversely affected by overexploitation would account for this surprisingly rich marine fish assemblage. Some of the fish caught off Portland were taken to Dorchester, as several of the same species were found there, albeit in smaller numbers. There is as yet no evidence for substantial exploitation of common eel or the production of fish sauces on the island, in contrast to Dorchester. Higher marine isotope signatures in burials in lead coffins in the Poundbury cemetery in Dorchester have led to suggestions that fish consumption may have been related to status (Richards and Hedges 1998). This suggestion could also be relevant to the settlement at Weston Road, which has some rich finds. However, the relatively rich fish assemblage could again mainly reflect the peculiar conditions pertaining to meat acquisition on the island. Increased demands for fish particularly from Dorchester would have encouraged the islanders to exploit the waters around Portland. In addition to trading some of their catch to the town, it would be surprising that they did not supplement the food supply of those living on the island, particularly if they included those who were accustomed to or had recently acquired Roman culinary tastes. The dense accumulation of faunal remains, the dominance of one species (sheep), the presence of high percentages of lambs, calves and piglets, the presence of relatively high percentages of fish and seabirds, including rare species are all unusual aspects of this faunal assemblage. It remains to be determined whether this reflects general adaptations to local conditions on and around the island or is related to the specific sites, which might have possessed special social or religious status. Acknowledgements Thanks are due to Sheila Hamilton-Dyer for her identifications of many of the bird and all the fish bones and for access to unpublished information.

43

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET Grain Identification

3.3 GRAIN (Tables are on CD)

by Debra Costen Carbonised grain have been found only in five contexts, three of these from contexts associated with structure 200, an area including evidence for persistent burning or possibly a brazier or oven (see Part 2.3). The grain (10gm) from C189 came from a sealed deposit above Feature 206 which is a square block of limestone, which had possibly formed the base for a plinth, column or even statue to stand on (see Part 2.3). This feature is in the interior of S200 which also contained a feature with signs of severe burning (F192) or ovens, A quantity of grain was also found actually at the bottom of (189). A gram of grain was found in the layer below F192, an area of hard baked clay and orange-red in colour, indicating a probable place for the actual burning of this grain. The sample from (205) came from a deposit of dark reddish brown loam associated with feature 256 consisting of two short rows of upright limestone slabs, also within the burnt area of S200.

Samples for identification had been sub-sampled by the excavators prior to submission, therefore a statistically representative count is not possible and the identifications can only represent the site tentatively as an entirety The numbers of grains in the table represent the number of whole grain in the sample plus the number of equivalent whole grain which was estimated from the fragments of whole taxa. It should be noted that there was no evidence of chaff, nodes, awn fragments or culm bases in the samples submitted. Grains were examined under a Leica stereo binocular microscope at ×10-×40 and identified by morphological criteria and by comparison with a modern reference collection. The plant remain assemblage is quite typical of assemblages found on other Roman military and nonmilitary sites across Britain (Huntley and Stalibrass. 1995) The use of vetch grown in rotation with barley is not new. Traditionally, farmers grew it as feed for livestock and draft animals. It could be grazed green in late winter and early spring, when feed was short; it could be harvested early to make hay; or it could be left to mature, to provide grain and straw for winter feeding. References to rotation with legumes date from Roman times from continental Europe including the British Isles (Bakels. 1999)

The grain samples from these three contexts are, therefore, samples which can, with a reasonable degree of certainty, be regarded as being of the Roman/RomanoBritish period of the site. The other two samples came from contexts where the dating cannot be certain as contamination between the Roman/Romano-British and Medieval periods is a possibility. The small quantity of grain from (298) came from a sondage excavated into the southern side of the large rectangular building S299, where earlier structures, probably Roman, became visible. The largest quantity of grain (11 gm) came from (262), the area just outside the northern wall of the rectangular building where there was a paved area with the bases of four columns.

The practice of planting of legumes with barley or oats continued into the medieval period; legumes were of great importance for maintenance of soil and nitrogen levels and are also good forage (Bolton. 1986). Many estates and priories, especially in the South of England used the east Kent rotation system- wheat/barley plus oats/legumes or fallow (Mate, 1985). Later experiments on intercropping have verified that this practice has many advantages for increasing forage and crop yields (Papastylianou. 2004; Tuna & Orak. 2007)

Barley was one of the principal grains grown during the Roman period and could have been used for beer making or bread or griddle cakes (Cool, 2006). There is some doubt as to whether vetch and oat were grown or just became included into the assemblage as weeds.

44

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET possibly near the cliffs in the north-eastern areas of the Island.

4 ARTEFACTS AND OTHER FINDS 4.1 BUILDING AND STRUCTURAL MATERIAL STONE

Context 36 41 64

Introduction The foundations of all the buildings revealed by excavation on the Weston Road site are of local limestone. Unfortunately, none of the super-structures of these buildings have survived and no clear indications were found of the material they were originally made of. It is likely to have been a mixture of stone and plaster.

SF No if Applicable

Some 58A

121 130

The material will be listed and discussed in relationship to the contexts within which they were found, but only the small percentage of material which have been found in reasonably well sealed layers or contexts, associated with the buildings, can be regarded as possibly indicative of the nature of the buildings. All other material classed as ‘building material’ made of clay (baked or processed) and stone which have been found distributed all over the site can, unfortunately, only be regarded as ‘demolition material’ and with no certain reference to any particular building or structure. A full list of all possible architectural or structural material, other than that discussed below, is given as an appendix. (See CD).

263 264

Some in SF459

Description 2 pieces 1 piece 4 tiles, one with hole 1 complete with 2 holes 13 pieces, all plus holes, some white plaster 14 pieces, 7 plus holes, some plaster 12 pieces, all plus holes, some two holes, some plus plaster

Comments Culvert by S42 Rubble layer next to culvert by S42 Area north of S42/299 Fill of stone gully by S299 Layer within S42 Under C263

A number of pieces of varying sizes of limestone tiles were also found in the spoil dump where they had been placed during the initial clearance of the site by JCB. The Context list of the above limestone tiles appears to indicate that they were found concentrated almost all within or near the large rectangular building (S42/S299 (See Section 2.3). Although the tiles were found in socalled demolition layers (i.e. layers containing material derived from buildings/structures as a result of disturbance) the concentration suggests that this building was probably originally roofed with limestone slabs. This is a heavy roofing material and the building would have required a very substantial wooden roof frame to uphold the weight of the tiles.

All the walls of structures which can, with a good degree of certainty, be classed as buildings, are very substantial and on average between 0.75 to 1.0 metres wide. This width suggests that the buildings were substantial and perhaps even the equivalent of two storeys high. It is also obvious that only a very small percentage of the original building material of all the structures, has remained on the site and been recovered during the excavations. As explained, it is almost certain that the buildings from this early site were raided by the many generations of subsequent Portland inhabitants so that many (if not most) of the old drystone walls and stone cottages still lived in on the Island, may contain material which originated from the Weston Road site and others like it which have completely disappeared, been destroyed or not yet discovered.

Old Red Burnt Mudstone Tiles The presence of anciently burnt deposits of oil shale, referred to as Old Red Burnt Mudstone, in the Jurassic Kimmeridge clay beds round Portland harbour, near Portland Castle and Ferrybridge have been extensively studied, and compared with deposits in Purbeck, by Ian West of Southampton University (West, April and May 2006). A fairly large quantity of this material – much in very fragmented form – has been found on the Weston Road site. It would be safe to assume that most, if not all, of it was derived from the local outcrops.

It should also be noted that some of the materials or items classed as ‘Building Material’ may in fact be derived from architectural, decorative or utilitarian features or structures which were not part of the main structural fabric of the building(s), e.g. screens, room dividers, built-in wall cupboards, shelves, columns, pedestals to stand something on or even free-standing furniture such as stone seats. The small sizes of most of the remaining fragments of material make it very difficult to determine what the original functions of most of these items were.

This mudstone often occurs in tabular pieces of varying sizes, roughly square or rectangular, which are fairly easy to work and polish and it is possible that different types of items could have been made from this very attractive stone. There are apparently slight geological differences between the materials from the different places but the majority of this material is very fine textured. Some samples are a bit more gritty and others contain tiny fossils of Pectern, Discina, Astarte and Lucina. The oily clay contains iron sulphide which burns very easily and

Roofing Tiles A quantity of roof tiles made of limestone slabs, some with holes through them, were found in a number of contexts on the site. It can be assumed that the material for these tiles was obtained from early local quarries, 45

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET fires have been reported in the shale beds of Purbeck and, possibly anciently, round Portland (West, op cit.)

another stage of occupation this floor was disturbed or the tiles removed).

Very tiny bits of Burnt Red Mudstone occur all over the Weston Road site, often so crushed that it is virtually powdered. It is possibly this material that gave the red colouring to the top of the clayey loam of Context 218 in Structure 197, a round building of which only half was available for part excavation (Section 2.3).

It can be assumed that although the rest of the mudstone - completely crushed or in pieces - was found all over the site, other structures may also have had floors or features with this material used as inlays, even though they were not always found in clear association with identified structures.

Only a few pieces of Red Burnt Mudstone have been found at Weston Road which are recognisable as having been neatly shaped - usually triangular - and highly polished on one face for possible use in the same manner as opus sectile floor inlays (as for instance at Fishbourne Palace). These items often also have rounded and carefully smoothed edges and are just over a centimetre thick. Several have remnants of plaster concretions on them.

It can also be considered that the red tiles may have been used, not (or not only) as floor inlays, but as inlays for decorative purposes on the surfaces of stone furniture or features within the buildings. Ten of the above pieces have a small residue of white lime plaster or mortar on them, a clear indication that this plaster was the material used for securing the tiles to some or other surface(s). The tile from C127 (also associated with S60) is much bigger than the rest, suggesting that this type of tile was, indeed used as inlays for different types of surfaces.

A total of 72 pieces of varying small sizes and quality of Burnt Red Mudstone have been collected. This demonstrates the fact that this material was found all over the different areas of the site. The following list is only of pieces which were definitely worked or polished.

Tesserae A quantity of small square blocks of limestone was recovered from several of the contexts. Although limestone often has a tendency to fracture naturally into a tabular form, these small blocks are fairly regular and it must be considered that they could have been used as tesserae. The following list gives details of 14 possible tesserae which were retained and recorded, but it is likely that more were found but discarded with the numerous limestone pieces which are distributed widely on site.

Old Burnt Red Mudstone Context 39 61 109 110 118 121 127 129 130 171 203 209 213 215 223 227 Unstrat.

S.F. No if applicable 374 373

354

359 358 353

Description and Comments 4 shaped pieces 3 pieces, one 7x7x1.3cm; assoc S60 6 smoothed pieces, one 5x4x1.3cm 1 piece with tiny fossils 1 piece with tiny fossils 21 pieces, one with score marks and white plaster 1 complete triangular tile 8.5x9.5x6.5cm: assoc. S60 2 pieces 1 piece 1 piece approx 3.4x3.4cm 1 piece 13 pieces 1 subtriangular piece 1 piece from ditch C22 Piece from triangle; assoc. S60 Fragment, smoothed, some fossils Complete triangle from spoil heap

Possible Tesserae Conte xt 102 110 a 121 129

SF No if applicable 418 429 426 199 (part)

209 227 263 279

451 350 355 460 (part)

Description 4 square blocks approx 2.5x3.02cm 1 square block 2.5x3.0 x1.5cm Square block 3x3x3x1.5cm 2 tiny squares:,3x1.5cm and 5.5x4.8x1.5cm 1 block 3x3.5x1.5cm 3 blocks 1.0x1.0x1.0cm 1 block 3.0x2.8x1.5cm 1 block 3.0x3.0x2.5cm

Although fourteen pieces are not really adequate to permit of any conclusions, the regular dimensions of the majority of these small blocks of stone, may indicate that they were intentionally shaped for use as tesserae. Unfortunately none of them came from a sealed context where a firm connection with a particular building is indicated.

From the above list, it can be seen that only two complete tiles, both visibly artificially shaped and polished on one surface and smooth round the corners, have been recovered. Unfortunately one was found unstratified on the spoil dump but one was found in C127, an occupation layer within S60, the complete round building of which one quadrant was fully excavated. (Part 2). Further pieces which show signs of workmanship were in the rubble layers above S60 in Contexts 61 (3 pieces) and one in C223. It would, therefore, be reasonable to argue that the floor of Structure 60 at some stage of occupation could have had inlays of Red Burnt Mudstone. (even if at

Miscellaneous carved stone items A very substantial quantity of small pieces of limestone was found all over the site but can be regarded as demolition rubble. At least 50 pieces of limestone have been collected from the site which have clear indications of having been carved. These came from those features 46

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET which are regarded as essentially Roman/Romano-British features. Unfortunately they are all too fragmentary to permit of identification but they do suggest that at least some of the buildings on the site had carved decorative features. A small percentage had red staining on at least one surface, perhaps indicative of the usage of ochre or some or other colouring matter.

47

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET 4.2 STRUCTURAL MATERIAL – CERAMIC

Clay: Fabrics A and B

Introduction

The frequent use of Fabrics A and B on the Weston Road site suggests strongly that the clay was obtained from the locally exposed Kimmeridge clay beds and was locally mixed for use as covering on the surfaces of walls or in the construction of features related to the buildings (see Part 1.2 Environment). It is present as colluvial clay on the site.

From the data in the site description it is evident that stone was the most frequently used building material on the site, probably during all periods. Although the excavated foundations of the buildings of the RomanoBritish period all appeared to be dry-stone walling on the site, not enough evidence was found to indicate what material was used in the superstructures, although it is likely also to have been mostly stone.

It is of a consistency that could be used in the making of simple pottery but, with more water added, becomes a paste suitable for application to rough surfaces and can even be more diluted to a white wash. Many of the lumps of Fabric A, as found, have at least one surface which has been very carefully smoothed and some have even acquired a slight sheen. This effect can be the result of friction as, for instance, where clay on floors had been continuously walked or sat on, but it could also suggest that the clay surface had been deliberately worked to acquire a sheen. Where found baked, it may indicate the proximity of later fires, either intentional or accidental.

Substantial quantities of mortar in the form of clay and lime plaster were found on site in most contexts and many of the stone materials listed above in this Report had concretions of a white mortar on them., It seems that at least the surfaces of walls would have been covered with clay, lime wash or plaster which was smoothed. This plaster would also have been used for securing floor tiles and other decorative architectural features, as for instance indicated by the presence of plaster concretions on some of the opus sectile.

Some of the lumps of white clay have clear impressions on them of having been pressed onto surfaces of wood, reeds or straw. A few lumps had finger marks on them as though squashed into shape by hand. This may indicate the presence of ‘wattle and daub’ type structures on the site, but could also be the result of straw mats being used on the floor or light partitions of reeds–and-plaster in a building.

Lime plaster and clay material, in various quantities, were collected from the site and are listed in an appendix with details of quantity and contexts. As not all of this material could be collected and kept, the quantities indicated are only a sample of what was present. Fabrics

Clay of Fabric B which often contains much fine quartzitic grit or sand, would have been more suitable for bonding material than Fabric A. It must be noted that although this fabric sometimes has a pinkish colour (due to the colour of the quartzitic or sandstone grit), it differs from opus signinum where the pink colour is due to the inclusion of crushed pottery.

A variety of bonding materials or mortars were used for structural purposes on Romano-British sites, e.g. at Danebury and elsewhere (Poole, 1984: Dix, 1979 and 1982). At the Weston Road site the grittiness and colouring of the material varied from context to context but, in general, there appears to have been three basic types of fabric present. Fabric A consisted of lumps of fairly fine white baked clay with little or even no grit in them whereas in Fabric B types the matrix was still white clay but there were various degrees of inclusions of different sands or grits which varied in colour from black, grey, to pink or even light orange. A third fabric C consisted of light grey to dark grey processed coarse lime plaster with bits of crushed limestone and tiny bits of chert.

Lime plaster In differentiating between the clays of Fabrics A and B and lime plaster Fabric C, it must be considered that all three fabrics are really products of the local Jurassic limestone Beds. The first is amply produced by the sedimentation of fine weathered particles of limestone or dust, while lime plaster is produced by burning lumps of limestone in a hearth or simple kiln, producing quick lime (Calcium Hydroxide), before slaking (mixing) it with water. It is used in the same manner as cement for grouting and covering stone walls and in a more liquid form as white wash on walls, sometimes as a base for painting coloured designs or murals on.

Quantities of burnt clay were found round the area of the possible ovens/fire places by F192, S200 and round Contexts 189,196 and 199 (see Part 2.3). There is a possibility that the hearth(s) were sometimes used for the production of small quantities of lime putty although that may not have been its main purpose. It is also likely that the clay was used in the construction of hearths and ovens, accounting for the frequent evidence of secondary burning on the clay.

It is known that there were a number of limekilns on Portland during later Medieval times and one nearcomplete one still stands in Easton Lane. It is quite likely that there were much earlier kilns on Portland and the plaster used on the Weston Road site was almost certainly locally produced, possibly somewhere very near the buildings on which the material was used. Samples of 48

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET lime plaster were collected and kept from the following contexts: Context 129 138 139 162 164 170 171 188 263 264

Medieval Glazed Ceramic Tiles A total of 6,943kg of glazed coloured medieval tiles were found, all of them fragmented (see appendix and also Special Finds375 - 414). Judging from the thickness of the flat pieces (average 1.0 – 1.5cm) they are, almost certainly, remains from roof tiling as this thickness is too little for floor tiles. Only one piece of unglazed brown tile from Context 310 could be a floor tile. Forty-three pieces of tiles are identifiable as parts of ridge tiles. One piece of larger than average ridge tile (SF413) had, in profile, a very large rounded angle and may in fact have been a hip tile for joining two roofs. The rest are tiles which came from below the ridge; of these five pieces have curved edges but the rest are flat, of which 11 pieces have shaped rounded edges.

Weight 45gm 399gm 213gm 343gm 24gm 315gm 102gm 5gm 7661gm 2800gm

From the above table it would appear that the largest quantity of processed lime plaster was collected from the large rectangular building, regarded as medieval in date (S42 and S299). Contexts 263 and 264 were encountered only in a small sondage excavated in the north-west corner of this big building in order to test the stratigraphy and the relationship to possible romano-british structures underlying the later building cannot be overlooked (Part 2.2).

Distribution: With the exception of the two pieces from Context 121, all these pieces were found in contexts closely associated with fills of the large rectangular building(s) listed as Structure S299/S42 (see Part 2.3). The largest quantity came from (130) which is a layer directly over the north wall of the building. The two pieces (100gms) were found in a large area outside the northern wall where a large quantity of Roman remains were also found, and it can be assumed that they arrived there as a result of post-depositional disturbance.

As there is a difference in appearance in this sondage between the lower layers of walling and the upper layers, there is a possibility that buildings of two (or more) different periods were built on the same foundations. If so, the indications could be that the later builders had extended the wall covering of lime plaster downwards, right over the earlier structures

The fabrics of the tiles are all of a brown, grey, white or pink-red clay and the glazes are shades of brown or very dark brown, yellow (sometimes with a very light green tinge) or pale green to very dark green. The few unglazed pieces found may be pieces from parts of the tile where the salt glaze was patchily applied.

Possible tegulae

The following is a report on a representative sample of the ridge tiles which were compared to assemblages in Dorset County Museum. Although no exact matches were found, the nearest matches were from the Cornhill site in Dorchester. No secure dating evidence was found at the Weston Road site, but relative dating by reference to the pottery could suggest dates between the eleventh and twelfth centuries (see Pottery analysis by Sparey-Green, this report).

Nine pieces of curved tile, possibly tegulae, were found unstratified on the spoil heap where they had been dumped by the JCB during initial work on the site. No assumptions can be made as to which structure on the excavated site they were derived from.

49

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET Sample of Ridge tiles from Weston Road Archaeological Site by Jane Yeo SFN

Context

385

130

No of pieces 4

397

168

1

408

AC 20

4

411

AC 64

5

413

Unstrat.

5

Description

Material

3 white colour material with grey core and a pale green glaze. Vertical knife cuts to base of crest. Ridge crest eroded.

Rough texture with abundant subrounded quartz inclusions Sandy matrix with no obvious inclusions. Sandy matrix with sparse shell? inclusions. Fine sandy fabric with moderate flint inclusions.

1 badly eroded terracotta piece with grey core and areas of green glaze. Oblique knife cuts to base of crest Terracotta with grey inner core and spots of brown glaze on upper surface. Three vertical knife cuts on base of ridge. Ridge crest eroded. 2 pieces of terracotta with small patches of pale green and brown glaze. Evidence of vertical knife cuts on base of one ridge. Ridge crest eroded. 2 pieces of grey material with yellow/green glaze. Evidence of three vertical small knife cuts on base of ridge. Knife or wire cut crest 25mm high. Terracotta with grey inner core and yellow/green glaze on upper surface. Oblique knife cuts to base of crest. Ridge crest knife or wire cut 20mm high. 3 pieces made of white coloured material with pale yellow/green glaze. Two have vertical knife marks at the base of the ridge. Ridge crest eroded. 2 pieces terracotta outer surface with grey inner core with dark green glaze. Evidence of knife cuts on base of the ridge. Ridge crests eroded.

415

162

3

3 terracotta ridge pieces with patches of green glaze. Evidence of vertical knife cuts on base of ridges. Ridge crest knife or wire cut 30mm high but top eroded.

White sandy with sub-rounded quartz grains. Sandy matrix with sparse shell? inclusions. White rough texture with abundant rounded quartz grains. Fine sandy fabric with some small inclusions. Fine sandy matrix without obvious inclusions.

Conclusion: Several types of fabric were identified. Rough texture with rounded quartz inclusions. Fine sandy fabric with few inclusions. Sandy fabric with sparse shell inclusions. All the crests on the ridge appear to be either knife or wire cut and are approximately 22mm in height. There are vertical or oblique cuts into the base of the crests, presumed to aid drying or firing. Similar patterns and materials have been excavated on other sites, including Poole and Sherborne, but most were found in post-medieval demolition contexts. (Hare, 1991; Keen and Ellis, 2000; Orton, et al. 1993; Watkin 1994; Woodward et al 1993).

50

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET The bowl fragment percentage is very similar to that in Table 1 but those for dishes and cups are appreciably down and brought about by the presence of a few large rim fragments from a Drag 18/31 dish and a Dr 33 cup distorting the Central Gaulish Samian dish and cup EVE percentages upwards. The beaker percentage remains unusually high.

4.3 THE SAMIAN POTTERY FROM PORTLAND by Joanna Bird and Malcolm Lyne (The numbering of tables as given by the authors has been retained. Tables not included in the text are on the CD; Figures 11,12).

A further unusual feature of the assemblage is that nearly all of the bowl fragments are from decorated Drag 29s and 37s: there is one fragment each from plain Samian forms Dr 38 and Dr 44. This suggests that we are dealing with a high status site: a view further reinforced by the presence of the high percentage of decorated beaker fragments, including those from two black-slipped Martres-de-Veyre Samian Dechelette 67s. Black-slipped Samian is exceedingly rare in Britain and to have two such vessels from one site is most unusual.

1. Introduction (M.L) The site yielded 302 sherds (2049g) of Samian ware from 46 contexts, ranging in date from the mid-1st to the mid3rd century. Of this material, 26 sherds came from the assessment (AC 886) and the rest from the main excavation (RMF 05/06). 2. Methodology (M.L) All of the Samian assemblages were quantified by numbers of sherds and their weights per fabric and Estimated Vessel Equivalents (EVEs) based on rim sherds were recorded (Orton 1975). Fabrics were identified using a x30 magnification pocket microscope with inbuilt artificial illumination source to determine the natures, forms, frequencies and sizes of added inclusions. The decorated Samian was reported on by Joanna Bird and the plain Samian and an overall view of the entire assemblage by Malcolm Lyne.

Set against the apparent high status make-up of the Samian assemblage is the native appearance of the two circular buildings and the fact that five of the vessels had been broken and then riveted together again. These latter comprise a South Gaulish Dr 37 bowl (Context 14), a South Gaulish closed form(Context 110), a Central Gaulish closed form (Context 161) a Central Gaulish Drag 37 bowl (Contexts 190, 192, 199 and 205) and a Central Gaulish Drag 18/31 dish (Context 262). Thirty eight fragments, including 31 from the riveted Central Gaulish Drag 37 bowl referred to above, had been refired after final breakage.

3. The assemblage (M.L) The total assemblage comprises fragments from 27 decorated and at least 28 plain vessels and was quantified by EVEs as follows:

It may be that the decorated vessels, or the beakers at least, were acquired from a high status site nearby: another possibility is that there was a ritual element in activities on the site which required the use of such exotic beakers and bowls.

The assemblage is rather small for such quantification but Table 1 suggests an overwhelming predominance of bowls and dishes, a poor showing by cups and an unusually high percentage of otherwise rare beaker forms. Further quantification by numbers of sherds per fabric and form was then carried out in an attempt to confirm or disprove this impression. Table 1 Fabric

Jars EVE

SG MDV CG EG

Bowls EVE 0.24 0.05 0.71 P 1.00 36.0%

Dishes EVE 0.14 0.08 0.84 0.10 1.16 41.7%

Beakers EVE 0.02 0.07 0.16

Cups EVE

Morts EVE

0.29

0.08

0.25 9.0%

0.29 10.4%

0.08 2.9%

Others EVE

Total EVE 0.40 0.20 2.08 0.10 2.78

% 14.4 7.2 74.8 3.6

Table 2 Fabric SG MDV CG EG ?

Jars

Bowls 24 4 75 19

Dishes 9 5 32 2

Beakers 6 22 5

Cups 1

Morts

4

1

122 40.4%

48 15.9%

33 10.9%

5 1.7%

1 0.3%

Others 16 1 72 1 3 93 30.8%

51

Total 56 32 189 22 3 302

% 18.5 10.6 62.6 7.3 1.0

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET have closely similar scrolls, both with the heavy corded binding, the rosette in the field and the fine wavy line above a band of circles at the base. Stanfield and Simpson 1958, pl 28,337, also has the cross motif in the field but has different leaves; Terrisse 1968, pl 44, 1060, a mould from Les Martres, has tulip-shaped buds in place of the leaves. The small pointed leaf on the Portland beaker has no exact parallel in Rogers 1974. The slip is smooth and evenly coloured, the body buff, both characteristics of Samian that has been fired black deliberately; X-8 has not apparently been previously recorded as a maker of black Samian (cf Stanfield and Simpson 1990, 87-93). c. AD 110-130. 129-198 Drag 37, EG, in the style of Helenius of Rheinzabern. The motifs are all recorded for him although this particular arrangement is not shown in Ricken and Thomas 2005. The two masks are Ricken and Fischer 1963, types M15 and M21, the cupid M134, the bestiaries with coiled rope M199, the bear T59, the eagle T207, the rosette O34, the column O219, the leaf P79 and the medallion K19. Early to mid 3rd century. 144 Dechelette 64, CG. Only the base of a draped female figure survives, but her position, between vertical ornaments or vases, suggests the work of Libertus or Butrio. c.AD 120-140. 189 (1)Small sherd, probably Dechelette 67; CG, black slipped. The moulded decoration may include an animal, but is too fragmentary to identify. A similar date to the X8 beaker is likely. (2)Drag 37, SG. Trident-tongued ovolo above panel design, which includes an animal fighter with a spear (Hermet 1934, pl 23,253) and a fat corded bud. Flavian. Burnt (3)Drag 37, CG. Freestyle design including a deer, probably Oswald 1936-7, type 1822R. Antonine. (4)Drag 37, CG. Freestyle design with running animals, including a deer. Antonine 190-192-199-205 Drag 37, CG; an eccentric bowl that cannot be attributed to a specific potter. The ovolo (Rogers 1974, type B18) was shared by several potters, mainly of the SacerAttianus group at Lezoux, and the borders of narrow beads would fit their work. The rosette C29, however, is only recorded for Quintilianus and Laxtucissa. The bowl is divided into panels; the surviving complete portions have pairs of shallow panels, but the panel to the left of the largest piece and a small separate fragment have no horizontal division. The large animals are a boar (similar to Oswald 1936-7, type 1641), a hound (Oswald 2029), possibly a bear (cf Oswald 1595) and a large hoofed animal, perhaps a horse and rider. The surviving panels at the base have scenes of small figures fighting animals, and may reflect the burlesque turns in the arena that sometimes involved dwarfs (cf Konig and Ewigleben 2000, 63 and fig 139): they include a hare (cf Oswald 2063A) facing a small naked figure holding what may be a net or noose (cf Oswald 667) and a leopardess (Oswald 1518) confronting a small gladiator (Oswald 1027 is only generally similar); part of a second small gladiator is

4. The Samian stamps (M.L) There are three stamped fragments: Context 26. Central Gaulish Drag 31 basal sherd with fragmentary stamp -----M. c.AD.150-200 121. Base of Central Gaulish Dr 33 cup with eight petal rosette stamp. c.AD.120-160 262. Central Gaulish Dr 18/31 basal sherd stamped ---ATIM. Possibly DIVICATIM. c.AD.140-150 5. Decorated Samian. (J.B.) The decorated Samian recovered from the excavations at Portland comprises 27 vessels, ranging in date from the Neronian/early Flavian period up to the middle of the 3rd century. The 1st-century wares are all South Gaulish, and consist of three examples of form Drag.29 and six of Drag.37; there is also a single globular beaker of form Dechelette 67. The proportion of Drag 29 to Drag 37 emphasises the mainly Flavian date of this small group. Only two pots could be attributed to potters, a Drag 29 in the style of Calvus and a Drag 37 with a fishing scene that was probably made by one of the Germanus group. The Central Gaulish material is the most numerous, at thirteen or fourteen vessels. It includes four further beakers, a most unusual number of relatively uncommon forms for such a small assemblage: these consist of three more examples of Dechelette 67 and a conical beaker of form Dechelette 64. Two of the 67s are black-slipped, and one of them is in the style of the anonymous potter X-8 who is not apparently previously recorded as a maker of black ware. The other forms are all Drag 37 bowls, dating from the first to the third quarter of the 2nd century; whilst numbers can be misleading with such a small group, it may be significant that none of the major later Antonine producers are represented. The latest potter identified is Cinnamus; the others are earlier, and include X-13, Criciro and perhaps X-9, as well as several pieces of an unusual bowl dated Cad 125-150. There are also three or four Drag 37 bowls from East Gaul, and those that were certainly identified originated at Rheinzabern. They include several pieces of a bowl in the style of Helenius, dating to the first half of the 3rd century. 6. Catalogue of Decorated Samian. (J.B.) RMF 05/06 109 Drag 37, CG (may be same pot as 209 (4) below). The vine-scroll is Rogers 1974, type MI or M7, with a small bird next to it. Trajanic – Hadrianic. 129 Drag 37, SG. Panel with tendril. Flavian. 129-189-199-205 Dechelette 67, CG, black slipped; in the style of X-8 of Les Martres-de-Veyre. Two other examples of this form 52

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET AC 886 12 Drag 29, SG. Lower zone scroll with palmate leaves. Preor early Flavian 15 Drag 37, EG (Rheinzabern). Ovolo Ricken and Fischer 1963, type E17, shared by several potters, here above a double medallion. Late 2nd – first half 3rd century. 19 Drag 37, EG (Rheinzabern). The ovolo (Ricken and Fischer 1963, type E44) and wavy-line border (O247) were used by Pupus-Iuvenis II, but the figures are unusual. The standing warrior is a variant of M181, lacking the spear in the right hand; the other is not identifiable. First half 3rd century.

impressed on its side below the hoofed animal. There is also a small figure, perhaps an amazon, at the top of the frieze on one of the fragments. Whilst most of the animals are without exact parallels, the general links of the figures are mainly with the Libertus-Butrio and Cinnamus groups. A date of c.AD 125-150 is likely. Burnt. 199 (1) Drag 29 probably, SG. Fine scrollery. Neronian – early Flavian (2) Drag 37, SG. ? Foliage at base. Flavian (3) Drag 37, SG. Part of a scene with a fisherman seated on rocks and holding three lines. Hermet and Oswald illustrate several similar scenes, some of them used on bowls of Germanus, but not this precise figure (Hermet 1934, pl 22, 201-204; Oswald 1936-7, types 951-956; for Germanus, Mees 1995, pls 73, 1-3, and 78, 10). c.AD 7595 (4) Dechelette 67 probably CG. Tip of probable leaf at top of decorated frieze. Hadrianic – Antonine. 205 (1) Drag 37, CG. Warrior on horseback, a slightly smaller version of Oswald 1936-7, type 247; it is on an unpublished bowl from Staines in the style of X-9. Hadrianic. Drilled hole for repair. (2) Drag 37, C/EG. The figure is not certainly identifiable; it may be a bound captive similar to Oswald 1936-7, type 1146. Later 2nd or early 3rd century. (3) Drag 37, SG. Panel design, including a hare on grass. Flavian 209 (1) Drag 37, SG. Panel design, including a saltire with small vine leaves. c.AD 70-85 (2) Drag 37, CG, in the style of Cinnamus of Lezoux. The ovolo is on Stanfield and Simpson 1958, pl 159, 23, the small horse on pl 163, 70. c AD 145-175 (3) Drag 37, CG, probably in the style of Criciro of Lezoux. The notched rosette is on Stanfield and Simpson 1958, pl 117, 10, the border on pl 118, 17; the ovolo and other motifs are too broken to identify. c.AD 135-165 (4) Drag 37, CG, probably by X-13 of Les Martres-deVeyre (may be the same pot as 109 above). The ovolo is on Stanfield and Simpson 1958, pl 47, 558, and the trifid on pl 44, 505; the vine scroll is Rogers 1974, M1 or M7. c.AD 100-125 (5) Dechelette 67, SG. Vertical beadrows alternating with wavy lines (cf Hermet 1934, pl 90, 6-29). Flavian – Trajanic. 271 Drag 37, CG, in the style of X-9 of Les Martres-deVeyre. The rosette, border and groups of corded sticks are on Stanfield and Simpson 1958, pl 29,352; the figure and hoofed animal are not certainly identifiable. c.AD 110130. Drilled hole for repair. 285 Drag 29, SG. Fine festoons in the upper zone; the lower zone has a wreath composed of a leaf used on stamped bowls of Calvus (Dannell et al 2003, Calvus I G7-8, 1040). c.AD 65-85

53

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET here is confined to the raw sherd numbers and weights, but in view of the fragmented and mixed nature of most groups much of the analysis below is based on percentages derived from the sherd count alone without quoting the weights, the figures from the latter being very similar. Even the amphorae were surprisingly fragmented.

4.4 COARSE WARE POTTERY REPORT by Christopher Sparey-Green (The numbering of tables as given by the author has been retained. Tables not included in the text are on CD; Figures 13-15).

No attempt has been made here to analyse the material by Estimated Vessel Equivalents (EVEs); data from this site might repay further analysis and comparison with other material from the island and other local coastal sites.

Introduction The following report covers the prehistoric, Roman and medieval coarse ware pottery from both the initial work on the site by AC Archaeology (AC886) and from the main excavation RMF05 and RMF06. The ceramics from this site extend from the Late pre-Roman Iron Age to Anglo-Saxon and medieval periods with very little of prehistoric, post-medieval or modern date.

The prehistoric material is largely of the middle or late Iron Age and has been identified by reference to local sites of that date with links across the Channel. For the Roman pottery, reference has also been made to the National Roman Fabric Reference Collection (Tomber and Dore 1998) and the recent publications on local pottery. For the Colour Coated fabrics and Grey wares from the New Forest, Oxford and Alice Holt / Farnham reference is made to the standard studies (Fulford 1975; Young 1977 and Lyne and Jefferies 1979).

The Roman samian ware has been the subject of a separate report and figures for this material are not included in the statistics below (Bird and Lyne). Acknowledgements I am grateful to Professor Sir Barry Cunliffe and Malcolm Lyne for advice on the identification of some Late Iron Age vessels. I am also very grateful to Susan Willetts for much help in examining the pottery and processing the data.

The amphorae and mortaria are described here in outline but deserve specialist study, perhaps in conjunction with other groups from Portland. The small amount of medieval and later wares is described on the basis of material found at Sixpenny Handley and Minchington and by reference to the relevant local literature (Sparey-Green 1997).

Method of Study All the pottery has been examined according to context, the majority derived from the initial clearance of topsoil and the occupation layers associated with the medieval building and the Roman structures adjacent. There are few closed groups derived from ditches, pits, graves or deeply stratified sequences. All sherds recovered from the hand excavation had been retained for study. The condition of the pottery was generally good, given the fairly calcareous soil, the surface of the majority of sherds retaining the decorative surfaces or burnished decoration. The average sherd size of 7 – 8gm indicates the high level of fragmentation and the later groups contained much residual material.

A catalogue of the pottery has been compiled for each context which lists and quantifies the fabrics and forms present. From this a series of tables have been created, as follows. Table 1 shows the total amount of pottery, other than samian ware and medieval roof tile from the site. Table 2 summarises all Roman fabric types in context number sequence, without detailed phasing. Tables 3 – 6 analyse the Roman BB1 coarse ware by forms and fabrics from Roman contexts.

The fabric series is that developed locally from the writer’s work on material from Poundbury, Dorchester, Minchington and Sixpenny Handley with the addition of fabrics identified on Portland. The Roman pottery fabric series is largely based on unpublished material from these sites. Individual fabrics were identified by visual examination with the aid of a hand lens and x 40 microscope. Fabric groups within each context have been quantified by sherd count and weight. Statistical analysis

Tables 7 and 8 cover the mortaria, amphorae and flagons. Table 9 summarises all the medieval pottery. Only unusual or un-paralleled pieces are illustrated.

TABLE 1 TOTAL NO. OF SHERDS These figures include the total from all contexts, irrespective of phasing. Date Prehistoric Late Iron Age; Roman Medieval TOTAL

Fabrics F50-59; 62-63 F1-28; 60, 61; 80; 90-93; 100-127 F29-49

No of Sherds 124 14676 478 15278

54

Weight (gm) 904 112388 3252 116544

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET surface (Fig 1). The profile of one jar or bowl can be recovered, the vessel generally similar to a bowl from Rope Lake Hole of Early or Middle Iron Age date (Davies 1987, fig. 79, 2) and jar type JC3.2 of the Middle Iron Age at Danebury. Other rims in Fabric F53 from (205) and (291) derive from other vessels comparable to other Iron Age forms from these sites (Figs 2-4).

MAIN FABRIC CATEGORIES PREHISTORIC FABRICS F50 –-F56; F58, F59; 6263 LATE IRON AGE WARES Including possible imports F6; F57, F60, F60.1, F61

The other fabric-group of interest comprises F57, F60 and F61 which contained much coarse angular white quartz and a variety of metamorphic minerals including feldspar and schist, varying in the mix of minerals and the size of the inclusions. Only one vessel could be tentatively reconstructed as a cordoned jar from contexts (66/161), (110), (164) and (186) (Fig 5). This vessel could be compared to vessels of type JE2 in fabric B3, as found at Hengistbury Head and derived from Armorica, but this would need analysis of the mineral content to confirm its origin (Cunliffe 1997, 4-5, fig 4).

ROMAN FABRICS LOCAL COARSE WARES F1 - 5 Dorset Black burnished Ware & BB1 variants F7 - 8 Large Storage/Food Processing Vessels F10.1 ? Black Burnished Ware variant F9 –10, 11-17, 110 Grey & other coarse wares MORTARIA F18 –19; 27.1 AMPHORAE/ FLAGONS F20 - 24; 80, 91, 11-123, 125

Fabric F60.1, a dark grey fabric with frequent subangular quartz was represented by sherds of one vessel. This was a thin-walled jar from (164) and (222) produced on a slow wheel or turntable and decorated with bands of opposed hatching demarcated by cordons (Fig 6). This is probably a local late pre-Roman Iron Age copy of an import from Brittany in fabric A3a or A3b (Cunliffe and Brown 1987 Ill.153).

BRITISH COLOUR COATED WARES F25 – 28 New Forest Wares F 93; 100 Nene Valley Ware F101 Oxfordshire Ware NON-LOCAL FINE WARES AND COLOUR-COATED WARES F102 –103 ‘Terra Nigra’ and ‘Terra Rubra’ F104 -109 Continental Colour Coated Wares

There are some rare fragments of finer black burnished wares in F1.10, 5.2, 6.0 and 6.1.; Vessels in these fabrics were characterized by a smooth, dark grey burnished surface and brownish core with fine quartz filler. The vessels are of regular form, suggesting the use of a turntable or slow wheel in their manufacture. The most significant sherds were of a cordoned, pedestal vessel from a Late Iron Age or early Roman context (291) (Fig 7). The inscribed cross design on the underside of the omphalos base is a feature paralleled in vessels from Hengistbury (Cunliffe and Brown 1987, Ill. 181, type ba2.0). This also is probably a local copy of wares produced in North-West France in Fabric A1 (op. cit., 213).

MEDIEVAL & POST-MEDIEVAL FABRICS F29 – 49; 200, 210 PREHISTORIC POTTERY Scattered sherds of pottery with flint or other coarse filler were identified in later contexts. Much of this was from handmade vessels but some, from the regularity of the form and rilling on the inside wall, may have been formed on a slow wheel. The majority of the material described here would probably belong to the middle or late Iron Age but derived from later contexts. There is a possibility that some from the less well-defined fabric groups are intrusive sherds of Saxon origin.

A single sherd of a vessel with flaring rim in this fabric is also copying a vessel of a type represented at Le Yaudet in Armorica (Fig 8).

The most frequent group of fabrics were those with forms of crushed quartz, calcined flint or sand (Fabrics F50, F51, F54, F55, F57.1, F58, F59, F62, F63). No vessel forms were recoverable from the small fragments of this type and no particular date can be attributed to them. They presumably represent fabrics derived from river valley deposits in the chalk lands to the north of Portland.

There may also be a small quantity of fine grey-brown grog-tempered ware (F52.3) from (279). ROMAN POTTERY

Another distinctive fabric group, found in small quantities, had varying proportions of limestone, oolites and shell (Fabrics 52, 53 and 56). These fabrics are similar to fabric H at Danebury (Cunliffe 1984, 308) and Fabric A at Rope Lake Hole (Davies 1987, 151). This filler would be consistent with an origin in Purbeck or Portland itself. Sherds from (97), (121), (138) and particularly (172) were in Fabric F53, a brown fabric with frequent grey oolites and traces of red haemetite on the

The collection of almost 15,400 sherds of this period provides an important opportunity for the study of material from Portland although the relatively small average sherd size (8gms) is a reflection of the absence of sealed pit or ditch groups containing well preserved vessels. There were 108 sherds of medieval pottery included in the stratified deposits, representing a low level of intrusive material. 55

EXCAVATION OF AN ENIGMATIC MULTI-PERIOD SITE ON THE ISLE OF PORTLAND, DORSET Fabric 1.12 Black with orange-red surface. Filler: subangular translucent quartz