Archaeological evaluation of the extramural monumental complex ('the Southern Canabae') at Caerleon, 2011

CARDIFF STUDIES IN ARCHAEOLOGY ARCHAEOLOGICAL EVALUATION OF THE EXTRAMURAL MONUMENTAL COMPLEX (‘THE SOUTHERN CANABAE’) AT CAERLEON, 2011 An Interim Report By P. Guest, M. Luke & C. Pudney CARDIFF STUDIES IN ARCHAEOLOGY SPECIALIST REPORT NUMBER 33 33 ARCHAEOLOGICAL EVALUATION OF THE EXTRAMURAL MONUMENTAL COMPLEX (‘THE SOUTHERN CANABAE’) AT CAERLEON, 2011 Interim Report by P. Guest, M. Luke & C. Pudney with contributions by P.Webster, M.Lewis & A.Powell CARDIFF STUDIES IN ARCHAEOLOGY SPECIALIST REPORT NUMBER 33 ARCHAEOLOGICAL EVALUATION OF THE EXTRAMURAL MONUMENTAL COMPLEX (‘THE SOUTHERN CANABAE’) AT CAERLEON, 2011 Cardiff Studies in Archaeology Specialist Report 33 © The authors 2012 P. Guest, M.Luke and C.Pudney, ISBN 978-0-9568398-2-4 Published by the Department of Archaeology & Conservation School of History Archaeology and Religion Cardiff University, Humanities Building, Colum Drive, Cardiff, CF10 3EU Tel: +44 (0)29 208 74470 Fax: +44 (0)29 208 74929 Email: adminshare@cardiff.ac.uk All rights reserved. No part of this publication may be reproduced in any form or by any means without permission of the authors. Designed by Cardiff Archaeological Illustration and Design Software: Adobe Creative Suite 4 Design Premium Contents Introduction 1 Background 3 Project Aims & Objectives 9 Methodology 11 Results of the excavations 13 Trench 1 15 Trench 2 25 Trench 3 29 Trench 4 33 Trench 5 39 Trench 6 51 Trench 7 57 Trench 8 61 Trench 9 67 Artefact & Environmental Assemblages 73 Pottery - preliminary observations 79 Discussion and interpretation 87 Bibliography 95 Appendix 1 - Trench Matrices 97 Appendix 2 - Community Engagement 107 Appendix 3 - Project Team 113 Introduction Excavations were undertaken across the area of the newly discovered complex of monumental buildings to the southwest of the legionary fortress of Isca at Caerleon between the 4th of August and the 1st of September 2011. This work was intended to evaluate the nature of the archaeological remains in this part of Caerleon and to provide important new information on the history and role of Isca in the Roman period, which, as one of only three permanent legionary fortresses in Britain, is a site of signiicant international importance. Greater knowledge of the extramural complex of monumental buildings will lead to a better appreciation of Caerleon’s part in the conquest, paciication and acculturation of Britain, and improve understanding of the River Usk’s role in connecting the fortress with the network of auxiliary forts in southern Wales, as well as other parts of Roman Britain and the Empire. The project was directed by Dr Peter Guest of Cardiff University and Mike Luke of Albion Archaeology, and the core project team consisted of 13 staff and 23 student archaeologists from Cardiff University. From the outset the Caerleon excavations have linked internationally signiicant research with undergraduate training and a broad mission to engage with the public. The engagement strategy this year was to continue to raise the public’s awareness of, and participation in, archaeological ieldwork and the remains of Roman Britain. The excavation also provided an excellent opportunity to involve 23 undergraduate students and numerous young volunteers in knowledge transfer and community engagement activities that will provide them with signiicant employability skills. The ields containing the Southern Canabae complex are privately owned and we are grateful to Mr Michael Haines of Broadway Farm for permission to carry out the excavations. The area is also a Scheduled Ancient Fig. 1. Aerial photograph of the 2011 Caerleon Southern Canabae trenches, the River Usk and the legionary amphitheatre 1 Introduction Monument and Scheduled Monument Consent was granted by Cadw. Funding for the excavations was provided by Cardiff University, the Roman Research Trust, the Haverield Bequest (University of Oxford), Time Team, Newport City Council, and the Caerleon Tourist Forum. Cadw provided funding for the initial post-excavation archiving work and preparation of this interim report and we are grateful to Jonathan Berry and his colleagues at Cadw for their continued support. Finally, we would like to thank the staff at the National Roman Legion Museum (National Museum Wales) for their encouragement and assistance. while Adrienne Powell completed the assessment of the animal bone assemblage. We are pleased to include summaries of their reports in this interim, which begins the process of integrating the stratigraphic narrative and the related inds evidence. A summary of the community engagement activities devised for the 2011 season is also provided here (a full report is available from the authors). Professor Bill Manning kindly commented on a draft and we are very grateful to him for helping to remove the inconsistencies and glitches in the text. We would also like to thank Tim Young for providing images of the geophysical survey results as well as his interpretations. Ian Dennis of Cardiff University prepared this report for publication with his usual skill and patience. This report summarises the results of the evaluation and includes the stratigraphic sequences recorded in each of the nine trenches, an overview of the inds assemblages, and a discussion of the excavation’s signiicance for understanding the legionary fortress at Caerleon. Mark Lewis and Peter Webster undertook the preliminary analysis of the pottery from the excavations, Fig. 2. Almost the complete Caerleon Southern Canabae 2011 excavation team, 31st August. 2 Background The site of Roman Isca, which lies beneath the town of Caerleon near Newport in South Wales, is one of the best known legionary fortresses from the Empire - a result of the intensive accumulation of knowledge obtained over a century-and-a-half of antiquarian and archaeological exploration at Caerleon which enables the fortress’ layout and history to be described in some detail. The signiicance of the site has been understood for centuries and Isca’s ruins were conspicuous enough in the medieval landscape of South Wales to merit comment in the work of chroniclers such as Gerald of Wales and Geoffrey of Monmouth in the twelfth century. The irst antiquarian work was undertaken by John Edward Lee in the 1840s, though the framework of understanding of the fortress really only really began to be established during the course of several important excavations carried out by the National Museum of Wales from the 1920s to the 1980s (Boon 1972; Boon 1987; Jones 2001; Knight 2001: 48). These revealed many of the buildings inside and around the fortress, including the amphitheatre, various barrack blocks (of which the most important are those in Prysg Field), the headquarters building, a possible hospital, the fortress CAERLEON CARDIFF CAERLEON 0 100km EXCAVATIONS 2011 Fig. 3. Location of Caerleon, South Wales (maps reproduced by permission of Ordnance Survey on behalf of HMSO. © Crown Copyright. All rights reserved.) 3 Background modern city of Newport. Isca was founded in c. A.D. 74 or 75 probably by Legio Secunda Augusta during, or perhaps in advance of, the inal campaigns of the governor of Britannia Julius Frontinus against the Silures and other native tribes of this western part of Britain. The Usk allowed the legion to be supplied by sea and also to provision the auxiliary units based in the upstream forts at Abergavenny and Brecon, while the road that crossed the Usk at Caerleon led to the major Roman settlements at Gloucester and Wroxeter in the east (both of which had been earlier legionary bases), and westwards towards the fort at Cardiff and beyond to Carmarthen. baths, a tribune’s house, various supposed workshops, and a quay on the right bank of the Usk. From the 1980s numerous watching-briefs, evaluations and large-scale excavations have been undertaken by the GlamorganGwent Archaeological Trust and other commercial units at Caerleon, of which the most signiicant were those on the Roman Gates site (Evans and Metcalf 1992) and the extramural settlement to the east of the fortress on Mill Street (Evans 2000). The insightful accounts published by George Boon summarised the state of knowledge in the 1970s and 1980s (Boon 1972; Boon 1987), while up-to-date summaries and discussions of speciic themes relating to Caerleon have appeared more recently (Brewer 2002; Manning 2004; Evans 2010). Initially many of the buildings within the fortress would have been built in timber, though the techniques employed by excavators in Caerleon during much of the twentieth century means that any evidence for timber structures would not have been recognised in their narrow trenches. The amphitheatre was irst constructed in about 90 by which time the decision had probably been taken to make Isca the Second Augustan Legion’s permanent base in Britain, and most buildings appear to have been Isca lies on a spur of gently rising ground between the meandering River Usk and its loodplain to the east and south, the Afon Lwyd stream to the north, and Lodge Hill to the northwest (Fig. 3). Oriented almost exactly southeast to north-west, the fortress covers 20.3 hectares (some 50 acres) on the right bank of the Usk at the river’s lowest bridging point before it enters the Severn Estuary at the Fortress North Compass North Fig. 4. Plan of the legionary fortress of Isca at Caerleon as known in 2002 4 Background rebuilt in stone between the late irst and early second centuries. Soon afterwards the legion was redeployed to northern Britain where it assisted in the construction of irst Hadrian’s Wall and, subsequently, the Antonine Wall. For the succeeding decades of the second century it is likely that the legion would have left only a small force at Caerleon while the majority of its men were occupied with the various demolitions and renovations of the Walls that took place up to c. 160 before it was inally decided that Hadrian’s Wall was to be the northern frontier of the Roman Empire. Epigraphic evidence points to several rebuilding events in Caerleon during the third century unsurprising given the age of the original buildings and the extended period of semi-abandonment during the second century. century, notably from the southern and eastern parts. The identity of these late-Roman inhabitants has been a matter of considerable debate with some suggesting a reduced military garrison, while others have imagined civilians living in abandoned barracks and other military buildings until the end of the Roman period when, like most places in Britain, the archaeological evidence for occupation disappears (Gardner 2007). In 2003 Cadw grant-aided the Glamorgan-Gwent Archaeological Trust to produce a research framework for Caerleon (Evans 2004). This document set out how much was known about fortress, but also highlighted the areas where knowledge was lacking and suggested measures that could be taken to provide valuable information to ill in those gaps. The research framework emphasised the plan of the canabae and the development of the River Usk as research priorities, and also highlighted the potential of modern excavations to produce important information relating to the society and culture of the inhabitants of Caerleon and its environs (particularly inds and environmental evidence). This initiative set in motion a It is thought that the military occupation at Isca ended c. 300 when the legion was redeployed, possibly to a new fort at Cardiff before appearing as the garrison of the Saxon Shore fort at Richborough in the early ifth-century Notitia Dignitatum. The fortress, however, has produced evidence for occupation during much of the fourth Fortress North Compass North Fig. 5. Updated plan of Isca after geophysical surveys from 2006 to 2008 (newly-discovered buildings shown in red) 5 Background number of research projects at Caerleon, including the programme of geophysical surveys of all the remaining areas of open ground within the fortress undertaken by GeoArch and Cardiff University between 2006 and 2008 (Fig. 5. See also Guest and Young 2007; Guest and Young 2010), and the major excavations by Cardiff University and UCL of a legionary store-building in Priory Field in 2008 and 2010 (Guest and Gardner 2008; Gardner and Guest 2010). area, as well as a more extensive series of anomalies on the eastern side of the courtyard aligned on a different orientation to the surrounding building or the fortress. The magnetic gradiometer recorded the strongest readings from this building along the incomplete southern range closest to the right bank of the Usk. The eastern range of the building appears to be wider than the western side and seems to have been subdivided longitudinally into four or perhaps ive parallel, though narrow, rows of rooms. The northern range furthest from the river is apparently as wide as the eastern side, but the internal area here seems to be divided into large rectangular spaces with their narrow sides facing the courtyard and ambulatory (although there are also indications of smaller rooms too). In 2008 a gradiometer survey was undertaken of open ields around the amphitheatre to the southwest of the fortress. These continued until 2011 and ultimately revealed a previously unknown complex of large publicstyle buildings extending over an area of about 5 hectares outside the fortress (Figs 6 and 7). Previous work in the 1950s hinted at the presence of large Roman buildings in this part of Caerleon, including a bath-house and a building with a monumental entranceway, yet the scale and nature of the newly-discovered suburb was nevertheless surprising. Elsewhere on this side of the fortress, civilian occupation seems to be conined to fairly rural-looking structures along either side of the main road leading out of Isca’s west gate (Chapman 2011, 324-26; Young 2012). Zone 2) The large courtyard building lies on the loodplain of the Usk and the land to the north rises up to the low promontory on which the amphitheatre and the fortress are sited. The second area of buildings located on the gradiometer survey lies on the upper part of this slope and the top of the higher ground, apparently connected to the back of the main courtyard building. The higher geophysical readings in this area suggest a complicated picture of walls criss-crossing this part of the complex. There are also several very high anomalies in this ield that might be related to recent agricultural activity such as the digging of pits and post-holes, or possibly the excavation of the amphitheatre in the 1920s (particularly the dumping of spoil). At this stage it is dificult to discern the layout of discrete buildings or internal walls, although these certainly followed the same alignment as the rest of the complex. The monumental complex that comprises the Southern Canabae includes one of the largest buildings known from Roman Britain fronting onto the Usk (the courtyard alone enclosed an area into which Caerleon’s amphitheatre would have itted with room to spare). Other buildings extending northwards from the river included further courtyards and possible basilica-like buildings. The whole complex appears to have been built as a single entity and was orientated approximately west-northwest to east-southeast and, therefore, on a different alignment to the fortress itself. This part of Caerleon has long been under rough pasture and the evidence for relatively recent activity is limited to a few discrete areas of metal dumping, lines of old fences and relict ield boundaries. The detailed examination and interpretation of these geophysical surveys is available elsewhere (Chapman 2011, 324-26; Young 2012), but in light of the 2011 excavations the Southern Canabae complex is described here as consisting of at least three major buildings or groups of buildings: Zone 3) Immediately north of this indistinct middleground lies a northern zone of two courtyards and associated buildings aligned on the same orientation. The irst courtyard measures 45-50m square and is apparently enclosed on three sides by a narrow corridor, possibly an ambulatory, while the eastern side is bounded by what seems to be a major aisled building. This building, or perhaps another adjoining building, continues eastwards towards the amphitheatre and the bath-building revealed by the Wheelers in the 1920s (Bath A) , though today it is obscured by farm-buildings (Evans 2000, 492-5). On the opposite western side of the courtyard, and apparently built up against the western ambulatory, is a row of six or seven large rectangular rooms whose narrow ends face towards the courtyard and the Usk loodplain. Zone 1) The southernmost building closest to the Usk is also the largest. Measuring 140m from east to west and at least 120m from north to south, it consists of a large square courtyard covering an area of approximately 1 hectare, possibly provided with a covered portico or ambulatory, and surrounded by ranges on all four sides. The courtyard itself contains two features that are likely to be buildings or structures of some kind. Both show as positive white anomalies (in contrast to the ranges whose walls produce negative readings), and include a small square structure on the main north-south axis of the open Although the second courtyard is imaged less clearly on the magnetometer results, it is possible to make out the main features of the structures that occupied this most northerly part of the complex. It seems to consist of an irregular square or rectangular open space bounded on its southern side by the courtyard just described, its western and northern sides by walls, or possibly roads, and on its eastern side by another long north-south building that 6 Background Fortress North Compass North Fig. 6. Combined results of the gradiometer surveys to the south of Caerleon 2008-11 (© GeoArch) Fortress North Zone 3 Compass North Zone 2 Zone 1 Fig. 7. Interpretation of the geophysical results 2008-11, showing the Southern Canabae complex (© GeoArch) 7 Background appears to be subdivided into internal rooms. There are clearly buildings continuing from this point eastwards towards the amphitheatre though they also lie beneath modern farm-buildings. During excavation in advance of the construction of these agricultural buildings, however, Nash-Williams revealed a large courtyard structure with hypocausted rooms and a monumental porticoed entranceway (Building D) that is almost certainly a continuation of this northernmost series of buildings identiied during the geophysical surveys (Evans 2000, 492-6). Large extramural courtyard buildings similar to the Zone 1 example at Caerleon described above have been identiied at the legionary fortresses at Carnuntum, Nijmegen, Mirebeau and Vindonissa on the continent (though only at Nijmegen in the Netherlands has one been excavated), yet Isca seems to be unusual in not otherwise having developed a signiicant civilian settlement outside its walls (Goguey and Reddé 1995; Hartmann 1986; Willems and van Enckevort 2009). The immediate questions posed by the results of the geophysical surveys concerned the dating and function of the complex of monumental buildings. Was it associated with the military occupation of the site, or was the intention for Caerleon to become a centre of civilian administration for western Britain like York in the North? The alignment of the complex suggests either that it is earlier than the fortress, or contemporary but somehow separate (possibly not ‘military’). Two small trial trenches excavated in 2010 over the southern range of the very large courtyard building closest to the Usk (Zone 1), revealed a wall constructed from deliberately broken and relaid tegulae (roof tiles), which was thought to be part of Roman Caerleon’s main quay on the Usk. Material dumped outwards from this wall was interpreted as the remains of landing stages or jetties projecting out into the river to allow larger ships to dock at Caerleon. The 2010 trial trenches suggested that the Roman remains survive very well in this part of Caerleon’s outskirts, and there was little evidence for extensive medieval and modern disturbance (Gardner and Guest 2010). Fig. 8.Test pits 1 and 2 excavated in 2010 close to the right bank of the river Usk 8 Project Aims & Objectives Geophysical surveys carried out between 2008 and 2010 identiied an extensive complex of public-style monumental buildings between the amphitheatre and the Usk. This part of the fortress canabae includes several large courtyard structures, one of which measures approximately 140m by 120m. The precise layout, date and function of these buildings are not known at the present time, but their ground plans suggest it is possible that they could have been either associated with the legionary command or part of a formal civil settlement, and it seems likely that the complex included parts of a port on the River Usk. • • • • • Layout of the buildings identiied by the gradiometer surveys Date of the buildings’ abandonment and, potentially, some indication of when they were constructed Function of these buildings and, therefore, the purpose of the monumental complex Remains of the suggested quayside wall and landing stages, including their construction and histories Extent of erosion to the remains caused by the River Usk, and if the river continues to erode this important archaeological resource. Another objective of the 2011 excavation was to increase the public’s awareness of, and participation in, archaeological ieldwork and the rich historic environment of Wales. The 2011 project aimed to contribute to many of the ambitions set out in the Welsh Historic Environment Strategy published in 2009 by offering the opportunity to participate in archaeological discovery and increasing awareness of Britain’s Roman past. This promoted heritage tourism in South Wales and beneitted the local economy. The discovery of these buildings is a remarkable addition to the fortress at Caerleon, but many questions remained about their dating and functions. The 2011 evaluation trenches were located to investigate the full extent of the monumental complex that comprises the Southern Canabae, speciically aiming to improve our knowledge and understanding of the following research topics: 9 Methodology The 2011 evaluation consisted of nine trenches exposing a total area of 410 m2, or less than 1% of the extent of the Southern Canabae (Fig. 9). The trenches were located so as to ground truth the geophysical results and to provide as much information as possible on the plans, functions and dating of the buildings comprising the monumental complex: • • • • • Trenches 1, 2, 3 and 4 investigated different parts of the very large courtyard building closest to the Usk (zone 1) Trench 5 was positioned to the rear of the centre of the northern range of the same courtyard building and extended upslope into the indistinct area of buildings in the centre of the complex (zone 2) Trench 6 examined the eastern part of zone 2 lying adjacent to the amphitheatre Trenches 7 and 8 were located at the eastern and western sides of the southern courtyard and adjoining aisled building in zone 3 Trench 9 was positioned over the probable aisled building to the east of the most northerly courtyard also in zone 3. Seven of the trenches measured 20m by 2m, Trench 3 measured 10m by 5m, and Trench 1 also 10m by 5m but with a 15m by 2m extension to the north. All were excavated to the irst signiicant archaeological deposits, except for Trench 1 where Scheduled Monument Consent allowed excavation to the full depth of the surviving stratigraphy. • A dedicated project page for the excavations in Caerleon on the Council for British Archaeology’s Community Archaeology Forum website where the dig blog was updated daily as the season progressed. People could also follow the latest discoveries on Facebook and Twitter. • Volunteers could work on the excavation where they gained experience of excavation, archaeological recording and inds work (cleaning and recording). • All visitors to the site were given a tour of the excavation where they found out more about the background to the project, the excavation techniques being used, and the archaeological remains in the trenches. Objects from the excavation were on display in the Finds Tent where visitors could see a range of artefacts uncovered during the season. • A 3-day Open Weekend was held during the Summer Bank Holiday when visitors could visit the site throughout the day and take part in various handson activities. The engagement element of the project involved Cardiff University’s dedicated Community Archaeologist and Community Engagement Team. Channel 4’s Time Team ilmed the excavation over three days for a programme dedicated to Caerleon that was broadcast in March 2012. Trench 3 was excavated by Phil Harding with Cardiff University students and volunteers, and the programme promoted the archaeology of Roman Caerleon to a mainstream audience across the UK. The topsoil and any overburden were removed by machine after which the trenches were entirely hand-excavated (all machined spoil was metal-detected). On site recording was completed according to Cardiff University standards and bulk and small inds were processed and recorded on site. Samples were taken from deposits where it was considered that surviving environmental remains will make a contribution to the understanding of a building’s functions or dating. The trenches were backilled by machine at the end of the excavation. Building on the success of the Priory Field excavations in 2008 and 2010, the Southern Canabae project encouraged the public to explore Roman antiquity by becoming involved with the archaeological ieldwork at Caerleon. The following resources and opportunities were available during the 2011 season: 11 Methodology Fortress North Compass North Fig. 9. Caerleon Southern Canabae 2011, trench location plan 12 Excavation Results The following sections describe the archaeological remains encountered in the nine evaluation trenches excavated in 2011. The results are compared to the gradiometer survey results and the trench narratives attempt to arrange the stratigraphy into a sequence of phased episodes of construction, use, disuse and dereliction. The trenches are described in some detail with summary plans and photographs, while the accompanying stratigraphic matrices can be found in Appendix 1. Walls and negative features such as pits, foundation trenches and post-holes, are indicated with square brackets [123], while layers and ills are shown within round brackets (123). Common abbreviations include ORS (Old Red Sandstone) and CBM (ceramic building material). The preliminary analysis of the inds and environmental evidence is included after the site narratives. The metal and other registered small inds are due to be cleaned and conserved in the near future, but it is possible to discuss the pattern of small inds between the nine trenches. The bulk inds, including brick and tile and the animal bone assemblage, are also described, while the extensive work on the pottery assemblage has important implications for the dating of the Southern Canabae and the identiication of the activities that took place there. Trench 1 Trench 1 was located to investigate the geophysical anomalies that run parallel to the River Usk. The area closest to the river produced higher gradiometer readings than other parts of the Southern Canabae and it is thought that the buildings exposed in the trench included part of Caerleon’s port facilities as well as the southeastern range of the large courtyard building. The trench was positioned at the eastern end of the building’s south-north axis, looking towards the building in the courtyard (partly excavated in Trench 3) and in line with the middle of the building’s northern range where Trench 5 was located (Fig. 9). The provisional stratigraphic matrix for Trench 1 can be found in Appendix 1. 11). The main part of the trench was 5m wide, while the 15m long extension was 2m wide. The latest Roman deposits in Trench 1 lay some 0.30.4m below the modern ground surface and were sealed by layers of alluvial silts deposited during more recent looding events. Other evidence for post-Roman activity was restricted to a ditch and revetted bank at the west end of the trench. The River Usk is likely to be actively eroding the eastern side of the buildings identiied in Trench 1, although the modern lood defences may well be reducing the rate of this destruction. After the 2010 season the tegula wall was interpreted as part of a quayside structure on the edge of Usk’s riverbank, though the 2011 excavation showed that this was almost certainly not the case and that the river during the Roman period must have lowed some distance to the east of its current course. Instead it is believed that the tegula wall separated a row of buildings of some kind on its riverward side from a roadway running parallel Two 5m by 2m test pits excavated in this area in 2010 revealed the upper courses of a wall constructed of broken roof tiles that appears to have produced the positive magnetic linear anomaly that extended for at least 80m alongside the modern riverbank. Trench 1 included the northernmost of these test pits (TP1) and extended for a distance of 25m from southeast to northwest (Fig 10 and es tr or th or sN F 1.6 1.5 1.4 1.3 1.2 1.1 0 10m Fig. 10. Trench 1 showing main walls superimposed over gradiometer results. Numbers refer to 'Areas' as described in the trench discussion. 15 Excavation results, Trench 1 118 107 149 Areas 1.1 and 1.2 114 148 114 116 149 115 115 105 124 125 Areas 1.3 128 132 132 Roadway 1.4 133 1 Trench N 133 168 0 Fig. 11. Trench 1 from east to west as excavated to these buildings. Two northeast-to-southwest aligned walls on the other side of the roadway appear to belong to the main courtyard building, probably marking the position of its narrow front wing and the boundary with the courtyard itself (see Fig. 12). The spaces separated by these various walls in Trench 1 are number 1.1 to 1.6 in the following discussion. 2m Building 1.5 188 189 177 126 CONSTRUCTION AND USE The earliest deposits exposed in the trench consisted of alluvial clays (194) and (198) in the northern part of the trench. These sloped gradually down towards the river, though the shallowness of the gradient indicates that the river bank in the Roman period must have been some distance further east of this point. 171 176 120 127 122 The tegula wall [115] passed through the full 5m width of the trench (Fig. 13). The wall was built over cobbled footings (190), offset on the wall’s western side by 0.4m, which seem to have been constructed on top of the underlying surface rather than within a foundation trench (Fig. 14). A line of large stones (175) above the remains of a wooden plank/beam (193) on the eastern side of the wall possibly represent revetting against which the footings had been piled in order to stop these slipping down the natural slope (Fig. 15). 170 Courtyard 1.6 Fig. 12. Final plan of Trench 1 showing walls and main features 16 Excavation results, Trench 1 Fig. 13. East end of Trench 1 showing tegula wall [115] and associated structures in Areas 1.1, 1.2 and 1.3 Fig. 14. West face of tegula wall [115] and underlying cobbled foundations (190) 17 Excavation results, Trench 1 Fig 15. East face of tegula wall [115], showing underlying burnt wooden beam (193) and large stones (175) against the wall’s face Eleven courses of tegulae survived forming a wall that was 0.6m wide. The tiles had been deliberately broken lengthways and each half set slightly apart to create a central space that was illed with smaller pieces of broken brick and tile. The langes of the broken tegulae formed the wall’s faces and the entire structure was bonded with solidly packed clay. The langes of the uppermost tegula course had been carefully chipped away to leave a lat surface, on top of which sat a twelfth course of faced stones at the northern end of the trench and broken bricks at the southern end. It is possible, therefore, that this wall was not built much higher than it survives today and that it acted as a sleeper wall for a timber or colonnaded superstructure. (157) could be the remains of a temporary partition or an attempt to drain the area during construction. After the tegula wall had been built, the footings and the lowest courses of tegulae were sealed by extensive deposits of clay intermixed with other material that raised and levelled the ground to either side of the wall. Surface (149) within the southern Area 1.1 consisted of an intermittent layer of large ORS slabs with smaller angular stones, as well as pieces of brick and tile in between and patches of pebbled metalling (Fig. 17). This surface continued beyond the southern edge of the trench and was not excavated. Areas 1.1 and 1.2 The footings of an east-west wall were identiied in the southeastern corner of the trench (excavated in the 2010 test pit). The wall had been robbed but the 0.6m wide foundations survived, comprising closely packed angular stones [148]. The sequence of clay make-up deposits to the north and south of the foundations were different and the wall must have been upstanding when the ground was raised. The surfaces either side of the wall were also different indicating that it separated different buildings, or separate rooms within a single building (Fig. 17). East of the tegulae wall the lowest clay dump (179) contained a dark deposit (192) including large lumps of slag and charcoal (Fig. 16). Further clay and silty levelling material (173/161/155/160/156/154), lay below a north-south arrangement of stones, brick and tiles (153) that probably formed a temporary division or revetment to control the dumping. This was sealed by further clay deposits (146/145). The sequence of material on the western side of the tegulae wall (186/180/184/185/183/157/151) indicates that the wall was already upstanding when the levelling dumps were deposited against its faces. A shallow gully [159] within In Area 1.2 north of wall [148], the levelling deposits were overlain by a series of surfaces, the earliest of which consisted of patches of angular ORS slabs and crushed CBM fragments (130) and metalling (129). A shallow semi-circular structure (116) butting against the east face of tegula wall [115] was constructed on top of these primary surfaces (Fig. 18). This unusual feature consisted of a crude surface of roughly hewn ORS blocks with intermittent patches of crushed tile, while a row of smaller stones and broken brick on its eastern side seemed 18 Excavation results, Trench 1 Fig. 16. North section of Trench 1 showing natural clays and levelling deposits on the riverward side of tegula wall [115] Fig. 17. East-facing view of foundations [148] and surface 149 in Area 1.1 (on left of photograph) Area 1.3 West of the tegula wall in Trench 1 was a narrow strip of ground that appears either to have been part of a lean-to building against the wall, or a pavement of some kind between the buildings described above and the roadway to the west (Fig. 19). This area included the remains of a 2.3m long wall [128] projecting perpendicularly from the west face of the tegula wall. Only the upper surviving course was exposed and, therefore, the level from which it was constructed is not known. It was built using roughly-hewn ORS blocks and was clay bonded, and it seems most likely that it was part of an open-ended to create a narrow gap or groove on its front face. The purpose of this structure is unclear, but the surface around it within Area 1.2 was relaid and repatched several times with mortar (135) and brick and tile (130/138), possibly incorporating a crude drainage channel (141). Eventually, most of these deposits were overlain by a inal more substantial stone and CBM surface (123/136), associated with a mixed charcoal deposit containing iron and copper objects as well as metalworking slags (118) that indicate some of the activities that took place in this building at the end of its life (though it could also have been dumped here after the building was abandoned). 19 Excavation results, Trench 1 Fig.18. Semi-circular feature (116) in Area 1.2 (bottom) and surface 149 in Area 1.1 (top) building of some kind constructed against the tegula wall (against which it butted), whose entrance faced directly onto the road. brick, tile and two pieces of a rotary quern set in a clay bedding (132) (Fig. 20). The roadway petered out on its western side where it met wall [168]. The surfaces either side of wall [128] consisted of sandy silts with occasional pieces of brick, tile and stone. These were darker and more mixed (125) north of the wall, and lighter to the south where a patch of crushed tile (124) against the wall’s south face suggests that this is most likely to have been the interior of a building with which it was associated. The interior and exterior surfaces were both at a higher level than the road surface to the west and appear to have partly encroached over it. Building 1.5 The extension of Trench 1 extended across the full width of a narrow corridor-like building identiied on the gradiometer survey results. This structure was aligned on the same orientation as the riverside building, the tegula wall and the roadway. It is likely that this was the courtyard building’s front wing facing the river and the port. Building 1.5 was deined by north-south walls [168] and [120] on its eastern and western sides respectively, creating a space between them of slightly more than 6m. Wall [168] was 0.55m wide and consisted of crudely faced clay-bonded ORS blocks with a rubble core (Fig. 21). At just under 1m wide, wall [120] was broader than [168] and the short length of this wall exposed in the trench showed that it had been built using two different techniques (Figs. 22 and 23). The northern part consisted of faces of stones of various sizes and a rubble core bonded with earth, while the southern part was constructed with triangular bricks and a core of CBM and rubble that were bonded together with orange clay. That the two parts are clearly bonded together demonstrates that the different constructions were part of the original build, and it is likely that this difference suggests that one part of the wall was the end of an entranceway. The Road 1.4 The roadway ran between the possible lean-to structure or pavement against the riverside building and the long corridor-like building to the west (Fig. 19). It was between 3.5m and 3.75m wide and was partly excavated in the northern part of the trench. The lowest layer associated with the road was a possible metalled surface (162) above the uppermost levelling deposit, which in turn was overlain by a darker deposit with fewer stones (124) that continued into the area of the later lean-to building. The latest substantial surface of the road probably consisted of lagstones, which only survived in one area. The lagstones were heavily worn and fragmented and had been extensively patched and relaid with clay, stone, brick and tile (133). The eastern edge of the roadway was raised with a roughly laid surface of large lat stones, 20 Excavation results, Trench 1 Fig. 19. Area 1.3, including wall [128], and the patched roadway 1.4 west of tegula wall Fig. 20. Quern stones reused in patching of roadway 1.4 (132) 21 Excavation results, Trench 1 Fig. 21. North-south wall [168] separating roadway 1.4 (on left) from building 1.5 (on right) Fig. 22. North-south wall [120] separating building 1.5 (on left) from courtyard 1.6 (on right). The two east-west walls in the trench edge ([171] to left and [127] on right) possibly mark the southern side of an entranceway into the courtyard. 22 Excavation results, Trench 1 Fig. 23. West face of wall [120] and the latest cobbled surface in courtyard 1.6. Note the two different methods of construction and wall [127] projecting into the courtyard area geophysical survey results indicate that the middle the courtyard building’s front wing lay just to the north of Trench 1, in which case the stone-built part of wall [120] exposed in the trench was most probably the southern end of the main entranceway from the corridor building into the courtyard itself. wall had been robbed and that (167) was part of the ill of the robber trench that was not recognised during the excavation. Courtyard 1.6 The geophysical results suggest the western end of Trench 1 extended into courtyard of the very large building on the southern side of the Southern Canabae. A later ield boundary truncated or covered a large part of the end of the extension, but it was possible nevertheless to reveal the inal surfaces of the open area next to the building’s east wing (see Figs. 22 and 23). An east-west oriented wall [171] on the very southern edge of Trench 1 butted against the inside faces of walls [168] and [120]. Only its northern face was visible in the trench and it is not known how wide this wall would have been, although enough was exposed to show that it had been crudely faced with earth-bonded ORS blocks of varying sizes (Fig. 22). The western end of this wall butted up to the brick-built part of wall [120] and it is possible, therefore, that it marked the south side of the putative entranceway. A short brick wall [127] projected into the open area from external wall the wing’s [120]. This continued beyond the southern edge of the trench and how wide the wall was is therefore unknown, but, like the southern part of [120] (against which it butted) it was built of claybonded courses of triangular bricks and a CBM core. The wall extended for some 1.4m into the courtyard and its western end was squared. Wall [127] lies on the same alignment as wall [171] inside the corridor building 1.5 and it is probably a continuation of the southern side of the entranceway through the building’s front wing from the riverside roadway. It seems likely that this short wall would have served as a buttress, plinth or base of some kind to monumentalise the southern side of this entrance when viewed from the courtyard. The latest surface of the courtyard butted against wall [127] and consisted of large river cobbles and some stones bedding onto clay (182/191). A possible early stone surface in the corridor / entranceway (176/177) was found between walls [120] and [171], although the surviving stones blocks were irregularly shaped and could also be part of a construction deposit. This area was subsequently illed by various layers of rubble (174), silts (172/121) and then more rubble (126/187/188/189). The uppermost of these rubble deposits lay at the same height as the roadway further east and they probably represent the latest loor surfaces in the corridor / entranceway. The eastern wall [168] appeared to have be sealed by a layer of silt (167), although it seems likely that the 23 Excavation results, Trench 1 Fig. 24. West end of the Trench 1 extension showing the oblique medieval or post-medieval ield boundary and Roman structures beyond DISUSE, DECAY AND ROBBING In the corridor building 1.5, possible evidence for demolition/collapse deposits consisted of rubble and CBM (119/126) which were in turn overlain by a very compact clay (111). These were similar to layers (107) and (108) in the eastern half of the trench. In general, however, the deposits that seem to be related to the decay and abandonment of the buildings in Trench 1 are much shallower than comparable layers in the other trenches. This might be explained if these buildings had been abandoned and dismantled earlier than elsewhere, or if this part of the Southern Canabae closest to the River Usk had been subjected to extensive looding which led to these buildings’ abandonment and also carried away much of the material that remained after their demolition. The effects of the looding experienced in this part of Caerleon are shown in the thickness of alluvial clay (106) that overlies all of Trench 1 (Fig. 16). ORS blocks and pieces of Roman brick and tile. The bank and ditch were part of an unusual semi-circular ield boundary and it is possible that they could have served as a lood defence. FIELD BOUNDARY A substantial Medieval or post-Medieval boundary was located at the very western end of the trench extension. It consisted of a ditch [102], which was still visible on the surface, and an adjacent wall [170] and bank (178) to the east (Fig. 24). The wall had been built using reused 24 Trench 2 Trench 2 was intended to examine how the central open area of the Southern Canabae’s main courtyard building was surfaced, as well as to investigate the geophysical anomalies in the courtyard’s northwestern corner which it was thought could be of post-Roman origin (Figs 25 and 26). A thin (c. 0.05m thick) charcoal-rich layer (223) overlay the natural geology in the western sondage, though nothing similar was found at the east end of the trench. This was sealed by a series of thick clay deposits that extended across the entire trench, including (214/221/222) at the eastern end and (226-230) in the western sondage. It is possible that the charcoal layer is derived from the clearance of vegetation in advance of construction work in the early Roman period, the irst stage of which is represented by the succeeding clay levelling layers. Analysis of the few sherds of pottery from these layers will conirm if these indicate Roman activity or pre-date the arrival of the legion in Caerleon. It was found that the courtyard surface consisted of open ground upon which cobbles and stones had been thrown down as hard standing. These had been cut by a post-hole and a shallow pit. The geophysical anomalies seem to correlate with a substantial rubble structure that appeared to have been deliberately laid. The only evidence for post-Roman activity in Trench 2 was a post-Medieval ield boundary. The provisional stratigraphic matrix for Trench 2 can be found in Appendix 1. STONE STRUCTURE An area of large angular blocks of stones was revealed in the central part of the trench that appears to have formed part of a structure of some kind (Figs 28 and 29). These stones extended between 3.25m and 3.8m from east to west, and the lowest layers of blocks (207) were probably laid directly on top of the underlying clay levelling deposits. The western edge of the feature was straight and here the stones gave the impression of having been laid with their lat sides facing upwards to form a face of at least two courses. The largest stones were found on this western side, while smaller blocks and large cobbles were more prevalent to the east where the feature’s edge was diagonal to the trench. The upper layers of stone (208) were all large and there did not appear to be any internal CLEARANCE AND LEVELLING After hand excavation had conirmed no sensitive or signiicant archaeological remains were present, and with permission from Cadw, a sondage was dug by mini-digger at the eastern end of the trench. The natural geology, consisting of a light grey-blue clay containing river-borne cobbles (231), was located at a depth of 1.00-1.15m (Fig. 27). A hand-excavated sondage at the western end of the trench discovered similar grey-blue clay at a similar depth, demonstrating that the underlying geology in this part of the loodplain is relatively level. r Fo 0 sN s tre th or 10m Fig. 25.Trench 2 superimposed over gradiometer results and showing extent of stone feature (207) 25 Excavation results, Trench 2 Fig. 26.Trench 2 from east as excavated Fig. 27. Machined-excavated sondage at east end of Trench 2, showing clay and river cobbles (231) 26 Excavation results, Trench 2 Fig. 28. Stone feature (207) from east (upper stone layer (208) visible in the section) Fig. 29. Stone feature (207) from north 27 Excavation results, Trench 2 The surfaces, pit and posthole were sealed by deposits comprising thin pink clay (209) and then a compact yellow-grey silt with pebble inclusions (205) and (206). It is uncertain if these represent later ground levels or post-Roman alluvial deposits. arrangement to the structure at this level (a dark soft silt (215) was tentatively identiied separating the lower stones from those above). It is possible that this structure might contain the remains of a collapsed north-south earth-bonded wall, though its existence and dating need to be conirmed. POST-MEDIEVAL FIELD BOUNDARIES The two shallow north-south ditches [203] and [212] that passed through the centre of the trench are the remnants of relict ield boundaries. These cut layer (204) that, in turn, sealed the top of the rubble structure (207). The course of ditch [212] was observed as a slight undulation in the ield to the south of Trench 2 and fragments of clay pipe were recovered from its ill. COURTYARD SURFACES AND ASSOCIATED ACTIVITY Thin layers of silty clay with occasional small stones (210/213/216) overlay the clay levelling deposits and appeared to butt against the stone structure (207). These layers were very irm in places and appear to have been deliberately compacted. Although the presence of iron pans could indicate the effects of natural processes too, the presence of charcoal, fragments of CBM, pottery and slag suggests these layers that are probably the remnants of the earliest Roman-period ground levels in the courtyard (Fig. 30). The discovery of a posthole [219] and a shallow circular pit [225] cut into these layers supports the interpretation of these as open ground. The post had been packed with pieces of ORS and clay (218/217), while the ill of the pit included what appeared to be metalworking debris (224). Fig. 30. Trench 2 from west, showing latest cobbled ground level cut by post-hole [219] and pit [225] in foreground 28 Trench 3 NATURAL AND SUBSOIL After hand excavation had conirmed no sensitive or signiicant archaeological remains were present, and with permission from Cadw, a sondage was dug by mini-digger at the northern end of the trench. The natural geology was located at approximately 1.4m below the modern ground surface and consisted of horizontal pinkish clay overlain by river cobbles and gravel (309). Trench 3 was located within the internal open area of the main courtyard building of the Southern Canabae to investigate an apparently square positive anomaly identiied during the gradiometer surveys. The structure lies on the courtyard’s main north-south axis towards its northern side and the trench was positioned to expose this possible building’s northeastern corner as well as a substantial area of the courtyard itself (Figs 31 and 32). Trench 3 was supervised by Time Team and excavated in three days towards the end of the 2011 season. The natural was sealed by some 0.75m of clay subsoils, the uppermost of which might have been brought in to level the area prior to construction of the courtyard building (308/313/323). The geophysical anomaly turned out to be a rectangular structure consisting of low clay-bonded brick walls. These were not rendered or plastered and no evidence for a loor was found inside the structure. The courtyard appears to have been open ground, while several large lat stones set into the yard surface were identiied as possible post-pads for other more supericial buildings. The provisional stratigraphic matrix for Trench 3 is located in Appendix 1. rtr s es Fo CONSTRUCTION AND USE Structure 3.1 Two walls were located in the southern part of the trench (Fig. 33). The east-west wall (303/304) was 0.8m wide and was bonded to the end of a north-south wall (302) that was 0.6m wide. Wall (303/304) was the outer wall of th N or 3.2 3.1 0 3.1 5m Fig. 31. Trench 3 showing main walls superimposed over gradiometer results. Numbers refer to ‘Areas’ as described in the trench discussion. 29 Excavation results, Trench 3 308 Trench N 309 315 308 319 317 318 316 310 307 307 320 304 303 321 323 322 311 313 314 302 306 305 0 2m Fig. 32 Final plan of Trench 3 showing walls and main features the structure (this continued beyond the trench to the east and west), while (302) appears to have been an internal wall dividing the inside of the structure into two parts (this continued beyond the southern edge of the trench). either wall were observed and it is possible either that they had been built directly onto the existing ground surface, or that the cobbles (321/322) observed in a small sondage against walls (304/302) were footings contained in a trench whose edges were not recognised during the excavation. Both walls were clay-bonded and consisted of faces of triangular bricks on either side of a core of large pieces of broken brick. The walls were bonded together and had therefore clearly been constructed as part of the same build. Up to six courses survived of the outer wall and ive of the narrower internal wall. No foundations for Trench 3 showed that the anomaly on the magnetometer results was a rectangular brick-built structure some 9m wide and aligned southeast to northwest. It is more dificult to be certain about its length but it must have 30 Excavation results, Trench 3 Fig. 33. Structure 3.1 from south Fig. 34. Rubble spread (310) and possible post-pads (315-319) in Area 3.2 31 Excavation results, Trench 3 extended for at least 13m in total and was divided into two parts, of which the northwestern was possibly the largest. Excavation of the interior of the structure produced no evidence for an internal loor of any kind and the deposits encountered here (311/312) were very similar to those found outside the building in the north of the trench. The deposits associated with the structure produced very few inds of any kind and the absence of bricks and tiles from the vicinity of the structure suggests either than any superstructure had been extensively robbed, or that the walls supported a timber building. It is also possible, however, that these walls had never supported a building at all in which case the structure could have functioned as the base of a platform of some kind. Area 3.2 Courtyard surface and possible post-pads There was no indication that the courtyard had been paved in the area of the trench and it appears that the subsoils served as the contemporary ground levels. Several individual and groups of large lat stones (315/316/317/318/319) were found embedded in the top of the clay subsoil (Fig. 34). The stones were irregularly 32 shaped and individually measured c. 0.3m by 0.2m (315/316), while some seem to have been placed together to form roughly semi-circular groups c. 0.6m by 0.6m in extent. These were found close to the brick structure and they are tentatively interpreted as post-pads (although was no obvious spatial patterning and similar lat stones (314) were found within the brick structure too). A larger area of ORS blocks (310) also lay above the clay subsoil on the north side of the brick structure. These possibly butted against the face of the brick wall and appear to have been thrown down rather than laid. ABANDONMENT Deposits of silty clay extended across the entire trench illing the areas inside structure 3.1 as well as outside (305/306/307). It is likely that these were laid down here during looding events after structure 3.1 and probably the entire courtyard building had been abandoned, though how long afterwards is not certain. There was little evidence for Medieval or modern activity in the trench. Trench 4 Trench 4 was positioned to investigate the southern end of the range of rooms on the northwestern side of the very large courtyard building in the east of the Southern Canabae (Fig. 34). The gradiometer results suggest that this northwestern range possibly was divided into large rectangular spaces with their narrow sides facing the courtyard, while a long narrow area adjoining the courtyard could have been a corridor or ambulatory. Trench 4 cut diagonally across the Roman buildings in this area and, according to the geophysical results, it included parts of perhaps two of the large spaces within the range, as well as parts of the possible ambulatory and another room or building at its southern end. evaluation trench, several possible walls and at least two phases of construction were identiied. The main walls had been partly robbed, although one stretch of a wall in the south of the trench was still standing several courses high (Fig. 35). The Roman archaeology in Trench 4 was undisturbed by later activity and survived only a few centimetres below the modern ground level. The provisional stratigraphic matrix for Trench 4 is located in Appendix 1. EARLIEST SURFACE AND STONE STRUCTURE The earliest deposit within the trench was a surface (431) located approximately 1.2m below the modern ground level (6.75m over datum). Stones and cobbles had been laid lat within a clay bedding layer that extended over the northern part of the trench (Fig. 36). The southern edge of this surface appeared to begin approximately 1.4m from the adjacent rubble structure (430) and was aligned parallel to it on a northeast-southwest orientation. The The walls and surfaces exposed within Trench 4 were aligned approximately north-south or east-west and several separate buildings or rooms were revealed. Although the archaeology in this part of the Southern Canabae was dificult to understand in a 2m wide th ss e rtr N 416 Fo 425 4.4 r No 431 419 4.3 421 419 419 430 420 429 4.2 430 418 434 435 0 4.1 10m Fig. 34. Trench 4 showing wall and surfaced areas superimposed over the gradiometer results. Numbers refer to ‘Areas’ as described in the trench discussion 33 Excavation results, Trench 4 Trench N surface continued beyond the trench to the east and west, while it also extended beneath a later stone surface at the northern end of the trench. 416 425 To the south of the surface was an area of packed rubble and medium-to-large cobbles (430) that extended for a distance of 7m across the southern half of the trench. On its northern edge the angular stones appear to have been deliberately laid lat to form at least ive rough courses, while the southern part of the structure was more mixed with cobbles, some of which were very large, and there was no hard edge on this side (Fig. 37). The rubble was between 0.7m and 0.8m thick and the stone blocks and cobbles were densely packed throughout, except in the centre where a narrow stone-free channel might have formed a drain (429). The northern edge of faced blocks suggests that the area exposed in the trench was part of a wide linear platform of some kind on a northeast-tosouthwest alignment. Area 4.4 431 419 Area 4.3 421 419 Structure (430) was not excavated and, although its relationship with surface (431) is not certain, it is possible that they were in use at the same time. The packed rubble and cobble platform could be the explanation for the darker linear anomaly on the gradiometer results that it was believed might be the courtyard’s ambulatory. 419 LEVELLING AND SECOND PHASE CONSTRUCTION A short length of a wall [418], revealed in the southern end of Trench 4, was aligned east-west. This appears to correspond with a negative linear anomaly on the geophysical survey results and it separated a cobbled surfaced area to the north from a different room, or more likely building, to the south. The wall had been partially robbed, though six courses survived close to the western trench edge (Fig. 38). 430 Area 4.2 420 429 The wall’s construction trench [434] was 0.76m wide and contained footings of packed angular stones in a loose brown-yellow sand (435). The lowest course of faced stone was as wide as the footings, but the upper courses were only 0.66m wide. White mortar bonded the wall’s stone faces with its core of rubble and CBM pieces, and a string course of bricks and lat stones sat between the third and ifth courses of squared stones. The lowest stone courses were abutted by the levelling deposits to either side, indicating that it was built from a lower level before the spaces to the north and south were subsequently raised and surfaced. 430 418 Area 4.1 - southern room or building The gradiometer results suggest that the space to the south of wall [418] was situated towards the centre of an east-west row of rooms or adjoining buildings in this part of the courtyard building. Although only a very small area of the interior of this room was included in the southwestern corner of the trench, it was possible to 433 434 Area 4.1 435 0 2m Fig. 35. Final plan of Trench 4 showing wall and main features 34 Excavation results, Trench 4 Fig. 36. Earliest stone and cobbled surface (431) in the northern part of Trench 4. Stone and white mortar layer (416) is visible in the top right-hand corner Fig. 37. Crudely faced northern edge of rubble structure (430) 35 Excavation results, Trench 4 Fig. 38. Southern edge of Trench 4 showing stoney surface (415) and white mortar surface (433) in the foreground, with wall [418] and footings (435) beyond, and the sequence of deposits in Area 4.1 in the corner Area 4.3 - northern (open?) area. The early stone and cobble surface (431) was sealed at the northern end of the trench by a thin dark accumulation or occupation layer (427), while immediately to the south of this some 0.6m of compact grey clay (432) had been dumped against the north face of rubble structure (430). These layers were overlain by redeposited natural (424/426), mixed together with quantities of domestic debris that appear to have been dumped deliberately to raise the height of the area in the north of the trench to the same level as the top of the rubble platform (430) to the south. There was no indication of any cobbling equivalent to (415/420) here and it is possible that the dumps of redeposited natural would have served as surfaces in this part of the central area (in which case the ground here was lower than the cobbled surface to the south). observe the sequence of deposits here in the side of the trench that had robbed the wall (Fig. 38). The lowest deposit (439), possibly natural or redeposited natural from the digging of the wall trenches, lay beneath a layer of light brown orange silty sand with frequent small stones and occasional CBM inclusions (438), that is likely to have been levelling for a loor. No actual loor surface was visible, however, and overlying (438) was a charcoal rich layer (437) that possibly originates from the use of the room or its abandonment. Area 4.2 - central cobbled area To the north of wall [418], the earlier rubble (430) was sealed by a series of layers (412/422/428), deeper in the south than the north, that seem to have served to level the original uneven platform before a new cambered surface was laid. This consisted of small pebbles (420) and small angular stones (415) bedded in silty-clay (Fig. 39). The northern side of this surface followed precisely the same northeast-southwest alignment as the underlying rubble (430), while in the south the surface butted against the third course of wall [418] suggesting that the cambered surface illed the same long narrow space as rubble (430). In the southeastern corner of the trench white mortar (433) seems to have been poured on top of the levelling layers, forming a separate but adjoining surface to (415/420). The sharp northeast-southwest boundary between surfaces (433) and (415/420) indicates the presence of a partition of some kind that is likely to have butted against the north face of wall [418]. Area 4.4 The northern side of this apparently unsurfaced space was demarcated by an area of packed ORS blocks mixed with large quantities of white mortar aligned northeastsouthwest (416). Only a small part of this was exposed in the corner of Trench 4 and, although in plan this stoney deposit gave the impression of being the ill of a robber trench, after the area to the south in the trench was excavated it became clear that this feature was only one or two stones deep (Fig. 36). It is likely, therefore, that (416) is part of another rubble surface extending northwards, or perhaps the remains of a collapsed wall. Nevertheless, the clear differentiation between this deposit and the stone–free area to the south indicates that these were 36 Excavation results, Trench 4 Fig. 39. Trench 4 looking north, showing stone surface (415) and white mortar surface (433) in the foreground, pebbled surface (420) in the middle ground, and possible rubble surface (421) beyond rubble across the trench, but it was not possible determine if this was the result of deliberate selection to create building platforms or other factors. distinct spaces, separated by a partition of of some kind. It is possible that a shallow trench [410] adjacent to (416) marks the location of such a wall, though no trace of footings was found in this putative robber trench. Trench [408] to rob wall [418], like with cut [410] in the northern part of Trench 4, was dug from the level of these rubble deposits. This indicates that the robbing activity occurred after the buildings in this area had fallen down, though the fact that part of wall [418] survived suggests that this was not undertaken systematically. ACCUMULATION AND POSSIBLE RESURFACING The northern edge of cobbled surface in Area 4.2 (420) was overlain by a series of mixed silty sand layers (414/419/425), possibly natural accumulations of soil, which also sealed the levelling layers and possible clay surfaces in Area 4.3 to the north (424/426). Two adjacent areas of rubble above (419) on the eastern side of the trench could be the remains of later surfaces, or platforms of some kind, at the same height as the cobbled area (420) to the south (Fig. 39). Layer (421) consisted of large lat blocks of ORS that appeared to form a very uneven surface, while an area of smaller stones in a sandy clay matrix (417) extended southwards from (421) where it almost joined cobbled surface (420). DECAY, COLLAPSE AND ROBBING Extensive deposits of rubble and CBM fragments covered the latest Roman levels in Trench 4. These included (403/405/406/407/436), presumably derived from the abandonment and collapse of this part of the courtyard building. There was some variation in the density of 37 Trench 5 Trench 5 was located at the centre of the rear range of rooms on the northern side of the very large courtyard building alongside the River Usk (zone 1), and continued up the hillside into the middle part of the Southern Canabae complex (zone 2). The trench cut diagonally across the Roman buildings and was positioned at the northern end of the courtyard building’s main north-south axis, overlooking the building in Trench 3 and also in line with the middle of the building’s southern range closest to the Usk where Trench 1 was located. The gradiometer results suggested that this spot could mark the position of a narrow entranceway leading northwards from the large courtyard building into the buildings of zone 2 (Fig. 40). slope allowing the construction of the various corridors and rooms of the Roman buildings along the edge of the higher ground (Figs 41 and 42). The long walls exposed within Trench 5 were aligned approximately east to west and several separate rooms or spaces were revealed, which for the purposes of this discussion are numbered 5.1 to 5.6 (from south to north). The archaeology in this part of the Southern Canabae was very complicated and the northern part of the trench produced possible evidence for several phases of rebuilding and major structural alterations. The main walls had been robbed, but several others in the northern part of the trench were still standing several courses high above the internal loors with which they were associated. Other than one pit that cut through a wall, the Roman As the ground in this area rose gently upwards from east to west, a series of shallow terraces had been cut into the th ss e rtr Fo 5.6 r No 5.5 5.4 5.3 5.2 5.1 0 10m Fig. 40. Trench 5 showing main walls superimposed over gradiometer results. Numbers refer to ‘Areas’ as described in the trench discussion 39 Excavation results, Trench 5 518 Area 5.6 563 561 562 561 511 532 Trench N 545 Area 5.5 560 558 557 546 556 555 533 533 505 553 554 533 550 552 550 551 Room 5.4 550 530 Room 5.3 549 543 548 516 Fig. 41. Trench 5 from south as excavated 515 archaeology in Trench 5 was undisturbed by later activity and survived only a few centimetres below the modern ground level. This remarkable level of preservation is partly explained by the unrobbed walls at the top of the slope creating spaces that became illed with hillwash after the abandonment of the Roman buildings. The provisional stratigraphic matrix for Trench 5 can be found in Appendix 1. 509 516 537 524 Area 5.2 523 CONSTRUCTION AND USE The main wall that separated the large courtyard building (zone 1) from the buildings in the centre of the Southern Canabae complex (zone 2) would have passed through the centre of Trench 5. The wall and its foundations, however, had been completely robbed and all that remained was the east-west robber cut [516] and its ills. The robbing trench was some 0.95m wide and the bottom lay between 0.45m and 0.55m below the level of the earliest Roman loors to either side, suggesting that the original wall foundations had been relatively wide but shallow. It is unusual for wall foundations to have been so extensively robbed and it could be that, unlike all other wall foundations observed in the 2011 trenches, this main wall was provided with masonry rather than cobbled foundations that were worth the effort of removing. The lowest deposits contained within the cut were both very clean, including what appears to be redeposited natural 564 509 529 517 528 Area 5.1 0 2m Fig. 42. Final plan of Trench 5 40 Excavation results, Trench 5 clay (536/567), and it is possible that these could be the base for foundations at the bottom of a construction trench rather than inilling of the robber cut. base for a timber wall. It seems more likely that [523] supported a colonnade and that the narrow c. 1.6m-wide space between this wall and the robbed main rear wall was a corridor or passageway at the back of courtyard building’s rear range (Area 5.2). The surviving surface of this passageway consisted of patches of worn and degraded compact white mortar or plaster (544) that could have been the bedding for lagstones of some other type of durable looring. Area 5.1 and Passageway 5.2 The part of Trench 5 to the south of the robbed main wall most likely lay at the very back of the zone 1 large courtyard building’s northern rear range and consists of two spaces separated by a low brick north-south wall surviving at least ive course high [523]. The wall’s two faces were built using triangular bricks (the apex of each brick pointed inwards) and the core was composed of smaller pieces of broken brick, all of which were bonded together with clay (Fig. 43). It is certain that this wall did not rise any higher because the sloping capstones of a drain (564) along its southern side rested on the top edge of the wall’s uppermost course of bricks. The other side of the drain was formed by a row of irregular blocks of ORS placed on their ends between 0.5m and 0.8m from the wall. The bottom of the drain was unlined and the capping was formed by a row of complete bricks and irregular slabs of ORS laid at an angle (possibly in groups of three), with their bottom edges resting on the top of the lining stones and their upper edges on the wall itself. The ill of the drain was a loose dark silt that contained no inds (566). On the south side of the brick wall [523] was Area 5.1, which extended beyond the trench edge to the east, south and west. The geophysical results suggest that this lay within the range on the northern side of the main courtyard in zone 1, but it is not clear if it was roofed or an open space. No traces of loors were found here and it appears that this area was bare earth perhaps with cobbles thrown down to form crude surfaces. The earliest of these was a compacted clay layer (541) and a large lat stone set within it (547) might have served as a post-pad for a structure of some kind. The clay, however, was very clean and lies below the capped drain against wall [523], suggesting that if it was a surface then the wall and drain (as well as other features described below) must post-date its use. A more convincing surface in this area was a layer of compacted silty-clay mixed with small fragments of CBM and cobbles (517) above a thick levelling deposit It is uncertain if the brick courses continued further down and it is possible that they formed, however, a Fig. 43. Wall [523] and adjacent drain with sloping capstones in situ. Looking northwards from Areas 5.1 to 5.2 41 Excavation results, Trench 5 Fig. 44. Drain [529] in Area 5.1, from west Fig. 45. Flagstone feature (528) in Area 5.1, from east. Drain [529] cuts across the trench from the near edge 42 Excavation results, Trench 5 (540). This was some 0.3m higher than (541) and gave the appearance of a rough yard surface that had seen considerable use. Set within bedding layer (540) and sealed by the surface (517) was a second stone-lined drain (529), this time not on the same alignment as the buildings. The edges of the drain were formed by two courses of unmortared irregular ORS, blocks and one large lat ORS capstone was found in situ at its eastern end (Fig. 44). The drain did not extend across the full width of the trench, but instead terminated some 1.10m in from the eastern trench edge (the western end of the drain ends obliquely with its southern side extending only 0.9m into the trench). Like the previous drain, no stones lined the bottom but, unlike the irst drain, the ill (542) produced quantities of pottery, animal bone and small inds. possible that it was intended to hold water or some other liquid. At irst glance the absence of a rim on its northern side is problematic, but there is some indication that this end originally would have butted against a wall in which case a rim on this side might not have been necessary. The geophysical survey suggested that an north-south wall ran through this part of Area 5.1 and, even though the excavations did not identify the remains of a wall, it is possible that it had been completely robbed and was simply missed in the narrow conines of the evaluation trench. Circumstantial evidence for the existence of a 0.40.5m wide wall against which feature (528) could have butted is provided by the drain (529), whose western end terminates obliquely about half way into the trench and, therefore, on the other side of a possible robbed out wall to the lagstone trough or tank. Also set within bedding layer (540) in Area 5.1 was part of an unusual structure composed of horizontal rows of large rectangular ORS slabs, with a raised bevelled rim on its southern side formed by upright narrower ORS slabs (528). This was aligned parallel to the east-west brick wall [523] between this area and passageway 5.2 and, as it extended beyond the western edge of the trench (and there was no sign of a second rim on its northern side), it seems safe to assume that a third row of lagstones and rim would have taken it close to, if not butting against, the brick wall or its adjoining drain (Fig. 45). The function of this structure is uncertain, though the lagstones from which its base was made had been carefully laid and it is Rooms 5.3 and 5.4 The rooms exposed in the northern part of Trench 5 are different to the northern range of the main courtyard building. The main wall robbed out by trench [516] separated the lower areas to the south from a series of raised rooms to the north (the loors were some 0.4m higher than the passageway 5.2), of which parts of two were excavated. The sequence of walls is extremely complex here and it is possible that there had been several episodes of structural alterations in this part of the complex, though it was not always possible to fully untangle this complexity. Fig. 46. Room 5.3 from southeast, showing latest loor (548), internal dividing wall [530], and beam slot (543) 43 Excavation results, Trench 5 Fig. 47. Room 5.4 from east, showing latest loor (551) Fig. 48. Brick end of wall [555/556] from room 5.3. Butted against by wall [505] on left and [533] in the foreground 44 Excavation results, Trench 5 The two rooms upslope of the main wall were separated by an internal north-south partition wall [530] and a slot (543). Wall [530] was some 0.5m wide and had been built with roughly faced ORS blocks set in a friable silty-sand mortar. The face of this wall in room 5.3 was unrendered and the southern end of the wall appears to sit directly on top of an earlier surface without any foundations (explaining the uneven nature of the wall’s courses). The slot (543) was some 0.2m wide and extended for a distance of 1.1m from the southern end of wall [530] to the robber trench for the main wall (which formed the eastern side of the room). Originally the slot would have been illed by a wooden beam at the base of a doorframe between rooms 5.3 and 5.4. Concrete was poured into the rooms thereby preserving the position of the original beam, which was either removed or had simply rotted away. The loor in room 5.3 consisted of a yellow-brown concrete mixed with small pebbles but no crushed tile (548), which butted against the lowest course of wall [530] (Fig. 46). This loor was not excavated, but the sequence of deposits beneath it was observed in the side of robber trench [516], which showed that it had been laid on top of a bedding layer consisting of stones, broken brick and tile, as well as a thin deposit of burnt material in the area of the beam slot (upon which the beam itself had originally sat). There was no evidence for an earlier loor in this room, suggesting that the dividing wall [530] was an original feature. that these and the piece of wood came from the collapsed lath and plaster ceiling in this room. Room 5.4, some 2m wide and 4m deep, was adjacent to room 5.3 and was formed by walls [530] to the south, [533] to the north, [505] to the west, and robber trench [516] to the east (Fig. 47). Wall [533] was more substantial than the others, measuring 0.9m wide and built with mortared courses of large neatly faced blocks either side of a rubble core (it’s northern face had been partially robbed). Painted plaster was found adhering to the surviving face of this wall in room 5.4, decorated with horizontal and vertical red lines above the loor (552). Wall [505] butted up to the continuation of [533] and was clearly a later addition. Although only a short length of this wall was exposed in the trench, it survived to a height of eight or perhaps nine courses above the loor. Some 0.4m wide at the top, it seems to have been built in two episodes: the lowest six or seven courses use small neatly cut and faced stones, while the uppermost two courses do not appear to have been bonded with mortar and were made with larger more roughly-hewn stones and a brick (it is not certain whether the upper part of this wall is Roman or post-Roman reuse of a truncated wall). The face of this wall in room 5.4 had been rendered with course plaster, as was the north face of the southern wall [530], though in neither case was there any indication that they had been decorated. Room 5.4 was separated from area 5.5 by wall [533] and its continuation [555/6], and the architecture of these adjoining spaces appears to have been entirely different. During the excavation this main dividing wall was thought to have consisted of three separated phases of construction, though it more likely that the length of walling contained in the trench included a blocked doorway 1.05m wide (the northern part of [533]), between jambs built of alternating courses of stone and brick, of which only the northern side [555] was completely exposed (Figs 48 and 49). The latest loor in room 5.4 consisted of a thin layer of crushed tile set in a degraded mortar matrix (551). Originally this might have looked like opus signinum but it had become worn in several places, particularly near to the walls and close to the doorway into room 5.3, revealing either areas of mixed burnt material (550) or large pieces of brick upon which the loor had been laid (and similar to the sequence observed beneath the concrete loor in room 5.3). Area 5.5 The area of Trench 5 to the north and east of room 5.4 was packed with the remains of two parallel east-west walls and, although the excavation continued to a depth of 1.2m, no substantial loors were discovered here. The western wall of the narrow building comprising areas 5.3-5.5 [562] lay at the northern end of the trench. Measuring 1.2m it was wider than the opposite wall of this building [516] to the south and had been robbed to the top of its foundations of mortared cobbles (563). It is possible to conirm, therefore, that this building adjacent to the main courtyard building to the east was just under 7m wide, and that there must have been another space equivalent to 5.5m some 2.5m wide behind room 5.4 (and also probably 5.3). The corner of area 5.5 between east-west wall [562] and north-south wall [533/555] contained two further east-west wall-like structures, [560] and [557]. Wall [560] was not fully exposed, but it is likely that it had been built up against the inside of main wall [562] and, therefore, must have been about 0.7m wide. It seems to have consisted of two abutting courses of bricks and stones without a core, and its southern face comprised at least three courses of lat bricks, then a course of neatly cut ORS blocks, followed by two courses of rectangular tufa blocks all bonded with a very solid pinkish-red sandy mortar. The two visible stones of the ORS course were laid as stretchers, while the tufa blocks of the surviving upper courses appear to have been laid as headers (Figs 49 and 50). A thin piece of wood with the remains of three iron nails through it was discovered at the base of the wall in a pile of collapsed plaster (SF 5072). Some of the loose pieces of plaster bore grooves on their inner faces and it is likely 45 Excavation results, Trench 5 Fig. 49. Area 5.5 from east, showing walls (from left to right) [533], [505], [555/556], [557] and [560]. Note the possible illed doorway in [533] (below left-hand ranging rod) which in turn was overlain by several collapsed courses of degraded tufa blocks set in a similar yellow mortar, some of which had been cut into voussoirs and were mixed with ORS voussoirs (513) (Fig. 52). Two adjoining complete box lue tiles were also incorporated into this collapsed structure. The coursing of the stones within (513) was very clear and the collapsed roof retained part of its original curvature when excavated (Fig. 53). The box lues had been laid end-on-end and were built within the tufa coursing (they showed no signs of sooting). It is thought that rows of box tile were used to form ribs supporting the vaulted ceiling that was constructed mainly of tufa, but with ORS courses perhaps to give extra strength. Lying only 0.4m south of this wall was another parallel east-west wall [557], although this was faced only on its southern side. Wall [557] butted against wall [556] and consisted of nine irregular courses of narrow ORS blocks bonded with a loose sandy mortar that sat upon a wide rubble foundation observed beneath its south face (see Figs 49 and 50). The space between the face of wall [560] and the core of [557] was illed with a loose red sand and mortar deposit (558) very similar to the bonding of [560]. The space on the other side of wall [557], in the corner between it and wall [533/555/556], was packed with rubble and broken brick and tile within a loose mortary matrix (546) and (554). The walls in this small part of area 5.5 and the material iling the spaces between them were overlain by several deposits that must be the collapsed remains of the superstructure of wall [560] (Figs. 50 and 51). The irst of these overlay [558] and consisted of a tangled mass of wedge-shaped tufa blocks within a matrix of only slightly less solid pinkish-red sandy mortar (545), and bonded on its northern side to three courses of triangular bricks (532). Wall [557] possibly formed an additional support or bracing wall for the vaulted ceiling. If this interpretation of these walls in this area is correct, the loor this ceiling covered is probably the clay surface beneath the rubble deposits (546) and (554) that illed the space between walls [557] and [533]. If so, the plain loor suggests that, despite the elaborate roof arrangement, this space was not intended to be as impressive as the neighbouring rooms 5.4 and 5.3 (whose loors were considerably higher too). These deposits appear to be the remains of part of the lower courses of a barrel vault, of which [558] was the northern supporting wall from which it sprung. How far this roof or ceiling extended is unknown as the opposite supporting wall to [560] lay beyond the trench, but it is clear from its collapsed remains that this had been a substantial structure. Against and on top of (545) lay a row of shaped ORS voussoirs in a yellow mortar (559), Room 5.6 The northwestern corner of Trench 5 exposed a small part of a room in the building beyond wall [562], which had been covered with a tessellated pavement (518). Unfortunately disturbed by the robbing of the wall, the tesserae were all black or white and presumably formed the border of a large mosaic (Fig. 54). 46 Excavation results, Trench 5 Fig. 50. Area 5.5 from west, showing wall [560] on left and wall [557] bottom right, sealed by collapsed deposits (545) and (513) NW SE 9.60 9.60 532 513 539 545 559 546 558 560 557 Top of wall stone, lower courses made of tile 0 1m Fig. 51. Section through Area 5.5 47 Excavation results, Trench 1 Fig. 52. Tufa and Old Red Sandstone voussoirs from Area 5.5 DISUSE, DECAY AND ROBBING Area 5.1 and Passageway 5.2 The deposits overlying the latest surfaces in area 5.1 consisted of thin mixed silty accumulations (514/517). The last loor in passageway 5.2 was sealed by a thicker deposit of light brown silty material (537) and an area of rubble that could possibly have been intentionally laid (524). Although this is far from certain, the fact that these stones extended over the top of brick wall [523] indicates that this wall was no longer in use. Above (537) and (524), a thick layer of mid white-brown sandy silt with degraded mortar, plaster and concrete lumps (509) extended across the passageway area. Containing quantities of pottery, animal bone and oyster shell, this deposit appeared very midden-like and suggests a combination of building decay and rubbish dumping. earliest of these is a mixed deposit of mortar, decayed plaster and broken roof tile in room 5.4 (527), which was overlain by more compact layers with more tile and less mortar (526) that also illed room 5.3. The main episode of roof collapse is represented by (522), which also included three intact, or near intact, box lue tiles, perhaps from the collapsed vaulted ceiling in area 5.5 (that this material also overlay wall [530] indicates that the dividing wall between these rooms was not longer standing above ground at this time). Layer (522) was probably part of the same roof collapse event as (507) in area 5.1 and 5.2, but these had been cut by [516], thereby conirming that the main wall of the building was robbed some time after the roof had collapsed. The other walls that formed rooms 5.3 and 5.4 were not robbed, probably because the rooms became illed with building debris and rubbish that covered their lower courses until they were no longer visible above ground. This suggests that the building’s main load bearing walls must have continued to support the building’s roof for some time after the internal walls had fallen down. Layer (509) in area 5.2 was overlain by large quantities of densely packed broken roof tiles covering much of area 5.1 as well (507). The southern edge of the fragmented tiles was marked by a line of more intact tegulae and broken imbrices aligned southeast to northwest across the trench (508). It is not clear if these were tiles that had simply slipped from a nearby roof or had been deliberately placed here during a later period of occupation (see Figs. 55 and 56). Area 5.5 As described in the previous section, the vaulted ceiling in this part of the building seems to have fallen down into the room below, probably as a single event rather than gradually over time and sealing rubble deposits (539) and (546) dumped on top of the latest loor. The survival of the remains of the tufa and stone ceiling is explained partly by the very solid mortar with which it had been bonded and perhaps the shape of the stones from which it was constructed, many of which had been cut into voussoirs that would have been less useful to those looking to recycle building stone. Rooms 5.3 and 5.4 The last concrete and crushed tile loors in rooms 5.3 and 5.4 were sealed by thin layers of silty material that produced animal bone, pottery and several metal small inds (531 / 535). In turn these were overlain by a series of deposits that illed the rooms, particularly room 5.4 where these reached a depth of 0.5m. These layers consisted of dumps of sandy silt mixed with decayed plaster and mortar, pieces of concrete looring, as well as very large quantities of brick and tile fragments, animal bone, oyster shells, pottery and numerous glass and metal inds (534 / 538). These appear to have originated from the discarding of rubbish in these, presumably abandoned, rooms. Cut [561] marks the robbing of wall [562] between area 5.5 and room 5.6. The lowest ills of this trench were similar to some of the collapse deposits further south, suggesting that the wall could have been robbed relatively early after the abandonment of this part of the Southern Canabae complex. The next layers in the rooms are associated with the decay and collapse (or perhaps demolition) of the building. The 48 Excavation results, Trench 1 Fig. 53. Northern end of Trench 5, showing remains of collapsed vault - brick and tile coursing (545) in the foreground,, with courses of tufa and ORS voussoirs (513) beyond. Note the two intact box-lue tiles The remains of the collapsed ceiling in area 5.5 and the tessellated loor in room 5.6 were sealed by a layer of mixed stone and CBM (506), similar to layers (503 / 504) further south in Trench 5 where they became gradually thicker towards the bottom of the slope. The latest activity in Trench 5 was feature [520] cut from the level of context (503/4), which could be a very late trench or pit to rob wall (556), although only part of it was contained within the trench. Above this and extending across the entire trench was topsoil (501/502). Fig. 54. Half-sectioned robber trench [561] of wall [562]. Note the small piece of tessellated pavement on the right (Area 5.6) 49 Excavation results, Trench 5 Fig. 55. Stacked (laid?) tegulae and imbrices (508) in Area 5.1 Fig. 56. Deposit of broken roof tiles and box lue tiles (507) in Areas 5.1 and 5.2 50 Trench 6 Trench 6 was positioned diagonally across the northeastern part of the middle zone 2 of the Southern Canabae complex, close to the amphitheatre. The gradiometer results suggest that the southern end of the trench lay within a rectangular structure, possibly an open area, and extended into a narrower rectangular building to the northwest (Fig. 57). The uneven topography of this area today was believed to be a result of the dumping of spoil from the Wheeler’s work in the amphitheatre in the 1920s, and aerial photographs of the excavations show narrow-gauge railed trackways leading out from the amphitheatre’s south entrance. It was thought, therefore, that large quantities of redeposited spoil from these relatively recent excavations might cover any remains of Roman buildings in this trench. lagstone looring, all of which is believed to date to the Roman period. The walls divide the area exposed in the trench into three areas (6.1-6.3), which appear to have belonged to two adjoining buildings (Figs 58 and 59). The provisional stratigraphic matrix for Trench 6 can be found in Appendix 1. CONSTRUCTION AND OCCUPATION Area 6.1 The southern east-west wall in Trench 6 [637] was some 0.95m wide and consisted of a rubble core between two courses of facing stone, and it appears to have been clay rather than mortar bonded. Understanding the construction of wall [637] was complicated by the fact that a lead water-pipe had been built across it and that a later phase of robbing had removed the courses overlying the pipe within the trench (Fig. 60). The water-pipe (634) ran from north to south at right angles to wall [637] and had been built into it so that it would have buried beneath the loor surfaces. The main pipe had a diameter of 0.12m and a bulge showed where two lengths had been joined As it turned out, however, the evidence for twentiethcentury spoil dumping in this part of the ield was limited to a layer of stone in the northern part of Trench 6. Below this two east-west walls were revealed, as well as a stone-lined drain, a lead water pipe and an area of th or e rtr Fo N ss 6.3 6.2 6.1 0 10m Fig. 57. Trench 6 showing walls and surfaced areas superimposed over gradiometer results. Numbers refer to ‘Areas’ as described in the trench discussion 51 Fig. 58. Trench 6 as excavated, from south Excavation results, Trench 6 613 Trench N 632 Area 6.3 BAULK Opus Signinum 631 622 624 Area 6.2 629 629 Fig. 60. Water-pipe (634) crossing wall [637] and sealed by cobble core ((628), from northwest 605 together in the centre of the wall. The pipe continued beyond the eastern edge of the trench, but its western end ended just within the trench with a round collar or plate pierced by iron nails (Fig. 61). This must be where the lead-pipe had been attached to a wooden tank (or pipe) from which water was drawn down slope. On the southern side of wall [637] the main pipe was tapped by a narrower branch heading in a more southeasterly direction. Presumably both pipes fed fountains or other water-features somewhere within the large courtyard building. 626 637 634 637 628 628 619 The lead-pipes do not appear to have been laid within a trench, but sat on stone bedding before being covered over by large cobbles [628] and the make-up deposits in area 6.1. The geophysical results suggest that this was the western side of a wide rectangular space that had been made-up and levelled by some 0.2m before a stone loor was laid (604) (Fig. 62). This surface, possibly open to the elements, was exposed in the southern part of the trench and consisted of four rows of large ORS lagstones on the same alignment as wall [637] (though there is some indication that the northern part of the surface had 617 604 0 Area 6.1 2m Fig. 59. Final plan of Trench 6 52 Excavation results, Trench 6 tile (612/625) on top of a clay bedding layer (624/626), both of which lay around and on top of the lead pipe. A stone drain (629) passing down the centre of room 6.2 presumably lay beneath this loor, though subsequent robbing activity within the trench (possibly to take away the capstones) has removed the relationship between these surfaces and the drain itself (Fig. 64). The drain was oriented on the same alignment as the building and its sides were built from courses of lat stone and large pieces of broken brick. A layer of compacted clay formed its bottom surface (636), while a lat stone capping the drain was observed in the east side of the trench. been laid in the opposite direction). The lagstones lie directly below the topsoil and during the excavation it was suggested that they could have been laid during the 1920’s amphitheatre excavations, perhaps the bedding for a trackway to remove spoil. While this is possible, the lagstones exactly follow the alignment of the Roman building and it seems more likely that they represent the latest Roman surface, most likely a courtyard, in this part of the Southern Canabae. Rooms 6.2 and 6.3 From wall [637] northwards Trench 6 seems to have extended into the interior of a single building, divided into two rooms by the second north-south wall [622]. Only one course of this wall was revealed but it did not appear to have been particularly well built. Some 0.55m wide, it was earth bonded and even within the narrow conines of the trench it was clear that it had been constructed slightly off straight (Fig. 63). It is assumed that this was an internal wall dividing a building into two rooms. There was no indication that this wall had been robbed, suggesting either that it supported a timber wall or that it was dismantled during or soon after the Roman period. A similar sequence of bedding (623) and degraded tile surface (631) was uncovered in room 6.3 north of wall [622], though here compact clay (650) had been introduced to patch the worn loor. DECAY AND ROBBING Room 6.3 was illed with 0.08-0.2m thick deposits of burnt material, consisting of numerous dumps of charcoal and possibly metalworking waste (616 and 608), while the last Roman loor in rooms 6.2 was sealed by a variety of deposits including a patch of broken roof tiles (607) and decayed mortar mixed with crushed tile fragments (614). Presumably these indicate a period of abandonment and decay followed by collapse or demolition as evidenced The latest loor in the southern room 6.2 (i.e. between walls [637] and [622]), consisted of a thin layer of crushed Fig. 61. Northwestern end of water-pipe (634) showing collar and iron nail heads 53 Excavation results, Trench 6 Fig. 62. Southern end of Trench 6, showing lagstone surface (604) Fig. 63. Internal wall [622], from west 54 Excavation results, Trench 6 Fig. 64. Drain (629) in Area 6.2, from west by the substantial silty layers containing larger pieces of brick and tile as well as rubble in these rooms (609/610/615). A small possible posthole [613] in the northern end of the trench suggests limited activity in this area after the building had collapsed or been demolished, though it is not possible at this time to offer a precise date for this feature. A wide V-shaped trench was cut seemingly along the line of drain in the middle of room 6.2 [640], possibly to remove the drain’s capstones. The linear cut [619] in the southern part of the trench was for the robbing of the substantial wall [637] between rooms 6.1 and 6.2. It is fortunate that the robbers were not interested in the wall’s cobble core which lay on top of and covered the leadpipe. The robber cuts and lagstone loor were all sealed by layers of rubble that covered the entire trench. A concentration of rubble in the northern part of Trench 6 might be spolia from the amphitheatre excavations in the 1920s, although if this is the correct interpretation of this material it suggests that dumping here was less intensive than previously thought. This lay directly beneath the subsoil and topsoil (602 and 601). 55 Trench 7 The gradiometer survey results indicate that the western side of the irst courtyard in zone 3 of the Southern Canabae comprised a series of large rooms or adjoining buildings that were built against the southwestern wall of the courtyard. These structures lay on the edge of the higher ground overlooking the loodplain of the River Usk as it meandered towards the Severn Estuary, and they would have been the irst buildings at Caerleon visible from boats and ships as they sailed upstream. Trench 7 was positioned across two rooms on the southern side of this range and it also included the area to the south of this courtyard at the western end of the Southern Canabae’s central zone 2 (Figs 65 and 66). lie immediately below the modern ground surface and survive very well indeed. There was very little evidence for post-Roman activity in this part of the ield. The provisional stratigraphic matrix for Trench 7 can be found in Appendix 1. CONSTRUCTION AND OCCUPATION Rooms 7.1 and 7.2 The two walls in Trench 7 had both been robbed, the external wall down to its foundations some 2m below the modern ground surface. This wall’s foundations consisted of large river cobbles (716) tightly packed and bonded with mortar in the 1.30m wide wall trench [719], while the internal wall [722] between rooms 7.1 and 7.2 (some 0.8m wide) had been reduced to a layer of sandy mortar between stone courses below the rooms’ loors but still above its foundations. The interior of the rooms had been raised with large quantities of make-up material and were considerably higher than the ground to the east. The excavations discovered that the end room had been furnished with a hypocausted loor and elaborately decorated walls. No traces of loors or surfaces were found in the area beyond the rooms to the south and it is thought that this could have been open during the Roman period (Fig. 67). The archaeological remains of these structures In room 7.2 a thick make-up deposit (723) of pinkish clean clay raised the loor level by some 1.15m above the original ground surface from which the foundation trench had been cut (724). Overlying this and across the entire room was a levelling layer of stones and brick and 10m 0 rth es tr or F 7.1 o sN 7.2 7.3 Fig. 65. Trench 7 showing walls and surfaced areas superimposed over gradiometer results. Numbers refer to ‘Areas’ as described in the trench discussion 57 Excavation results, Trench 7 Trench N 726 Area 7.3 732 702 734 731 707 707 712 719 720 708 708 716 Fig. 66. Trench 7 fully excavated, from west tile fragments (738), which also acted as bedding for the concrete surface (714) that was part of the only loor that this room had been provided with. Several pilae from a hypocaust were arranged in rows on top of the concrete, which sloped downwards from north to south and was covered by a patchy layer of burnt material, presumably from the iring of the underloor heating system (Fig. 68). The concrete extended to the eastern wall of the room, but stopped some 0.6m short of the western wall and it is possible that this space would have been illed by columns of box lue tiles running up the inside face of the wall. 718 718 717 718 714 Room 7.2 714 At total of 24 pilae and mortar bases for seven more were uncovered in the exposed part of room 7.2 (718), while the material that subsequently illed the room contained numerous bricks and stones from other disturbed pilae. The pilae had been arranged in rows of nine across the room and each surviving pila consisted of a square brick on a small mortar spread, some of which had an irregular lat piece of stone mortared on top. None of the second tier of stones had mortar on their upper surfaces and it is not certain how high the pilae originally would have been. It seems unlikely that the space beneath the raise loor was only some 0.12m deep, but the absence of mortar on the stone is certainly intriguing. The inilling deposits also contained fragments of large square or rectangular ORS slabs that presumably had been laid 725 705 722 705 Room 7.1 709 0 709 2m Fig. 67. Final plan of Trench 7 58 Excavation results, Trench 7 Fig. 68. Hypocaust in room 7.2 showing bases of pilae (718) on concrete bedding (714). Photograph taken from south Fig. 69. Painted plaster overlying the hypocaust pilae in room 7.2 59 Excavation results, Trench 7 Fig. 70. Half-section through room 7.1 from east, showing levelling deposit (730), stone and brick bedding layer (729), followed by an opus signinum surface (709) over the pilae (the largest piece had bevelled edges and mortar adhering on one side). The loose bricks and slabs were found among large quantities of decayed mortar that must have formed the loor’s surface when laid over the slabs, but which had been broken up when the slabs were later removed. These deposits also produced numerous pieces of plaster painted with various colours, including red, yellow and blue, and the heated room’s walls seem to have been decorated with diamond patterns and loral motifs (Fig. 69). though whether this occurred while the buildings were in use is not certain. The interior of room 7.1 also had been raised above the existing ground surface (layers (737) and (730)), though exactly by how much is unknown as it was not possible to examine the deposits to their full depth in this part of the trench. Again a mixed layer of stones and brick and tile fragments (729) lay above these and below the room’s loor surface, which this time consisted of opus signinum (709) (Fig. 70). As previously described, the in situ pilae in room 7.2 were covered by the disturbed remains of the hypocaust loor (717), including mortar, plaster (some painted), broken bricks and pieces of ORS slabs, which presumably had to be broken up in order to remove the underlying slabs. This was overlain by a layer of broken roof tiles and stone that suggest decay or the demolition of the building (710), after which the eastern external wall was robbed (trench [720]). Stones (703) on top of the loor in room 7.1 (709) indicate a period of decay before the building collapsed or was demolished, though the western internal wall appears to have been robbed at a later date. Later activity in 7.3 is evidenced by a north-south linear stone feature (731) that partly overlay some of these rubbish dumps and might be a poorly constructed drystone wall. It was dificult, however, to understand the nature of this arrangement of stones in a narrow evaluation trench and it could be that the feature derives from the decay or demolition of the neighbouring building. Area 7.3 Area 7.3 covered half of the trench from the robbed eastern external wall. The latest surface here was approximately 0.9m below the loor in room 7.2 and consisted of compacted cobbles (736), which suggests that this part of the trench lay outside a building, perhaps a road or alley through the Southern Canabae. Above the cobbles a series of rubble layers interspersed with deposits containing animal bone, pottery and oyster shells appears to indicate the dumping of rubbish against the eastern wall of the courtyard buildings (712/726/732/739/740), In area 7.3 outside the building, the rubbish dumps on top of the last surface and the possible wall were all sealed by a series of silty deposits (primarily 728/726/727), containing pieces of brick, tile and stone, that eventually raised the level of ground here almost to the same height as further west. Thin topsoils covered Trench 7 beneath the turf (701/702). 60 Trench 8 Trench 8 was oriented northwest to southeast across the large apparently aisled building on the eastern side of the irst courtyard in zone 3 (Fig. 71). The trench revealed three internal areas of the building, at least one of which had been furnished with an opus signinum loor, while two underwent later alterations. An extensive external area was exposed at the southern end of the trench. The Roman period deposits were found between 0.15m and 0.45m below the modern ground surface and were very well preserved (Figs 72 and 73). adjacent building. It was assumed before the excavation that these were either internal ‘aisles’ of the building or possibly porticoes around the courtyard to the south, yet no sign of the presumed east-west masonry wall separating the two corridors was identiied during the excavation. This is possibly because a wall lies deeper than Trench 8 was excavated, though it is more likely that the geophysical anomaly was caused by a concentration of charcoal and other burnt material that might represent the upper ill an unexcavated linear feature (perhaps the remains of a timber wall). The building had been raised above the surrounding ground level and the discovery of a possible step on its southern side indicates the building was accessed from the large courtyard in this direction. There was little evidence for signiicant post-Roman activity in Trench 8. The provisional stratigraphic matrix for Trench 8 can be found in Appendix 1. Instead, the entire southern part of Trench 8 consisted of a series of compacted mixed orange-brown silty clay layers, primarily (846) and (856), that appear to be the latest ground level in this part of the complex (see Fig. 72). There is no indication that this space was paved or provided with any form of durable surface, while water from a drain that ran beneath the adjacent building seems to have lowed directly out onto this area where it would have simply soaked into the ground. On the basis of the excavated evidence it is suggested that this part of the complex was an external area and was connected to the courtyard rather than the aisled building, though only further excavation will conirm this. CONSTRUCTION AND OCCUPATION Area 8.1 – possible open area with step and drain The southernmost 7.5m of Trench 8 appears to have been part of a single external space, though at irst glance the archaeology within the trench does not appear to coincide with the gradiometer results that suggested this area included two wide corridor-like structures parallel to the 8.4 8.3ii rth es rtr Fo o sN 8.3i 8.2ii 8.2i 8.1 0 10m Fig. 71. Trench 8 showing walls and surfaced areas superimposed over gradiometer results. Numbers refer to ‘Areas’ as described in the trench discussion 61 Excavation results, Trench 8 Trench N 836 802 Room 8.4 820 821 820 Room 8.3ii 842 840 838 802 853 840 813 Room 8.3i 853 811 812 811 835 805 832 Room 8.2i Room 8.2ii Fig. 72. Trench 8 from south, fully excavated 815 The building’s southern external wall was 0.50m wide and had been surprisingly badly constructed [815] (Fig. 74). Only two courses of this wall were visible in the trench and there was no sign that the masonry had been bonded with mortar (the mortar of the other walls survived reasonably well so it is unlikely that the bonding in [815] had simply eroded away). Butting up against the southern face of this wall was a 1.1m - 1.2m wide stone platform consisting of a single course of large lat stones and also seemingly unmortared [826/829]. This was probably a step allowing access from the courtyard into the interior of the building and it is possible that this feature ran along the full length of the front wall. A curved stone-lined drain [825] had been built into both the wall [815] and platform [826] which channel to water, presumably from the eastern end of the building, into the adjacent open space via a V-shaped outlet made from fragments of broken tegula (possibly part of a single tile). The tegula pieces had been carefully selected and deliberately positioned to reduce splashing, while crude sloping faces of stones and tile were found to either side of the drain opening (Fig. 75). 826 844 825 829 825 844 845 827 854 852 Area 8.1 846 865 859 808 847 851 The poor quality of the construction of these features suggests that wall [815] was not intended to be load- 0 2m Fig. 73 Final plan of Trench 8 62 Excavation results, Trench 8 Fig. 74. Southern narrow external wall of building [815] and broad step (826/829) from external Area 8.1 (left). Drain (825) cuts through both the wall and step. Taken from the east Fig. 75. Outlet of drain (825), consisting of broken tegula. Taken from the south 63 Excavation results, Trench 8 bearing and it is dificult to imagine that this wall and the step formed the building’s main entrance. Later decay or demolition deposits from Areas 8.2 and 8.3 produced a number of segmented bricks, each one quarter of a circle, that indicate the presence of brick-built columns somewhere in this part of the building. While the original location of these columns is not known, the external wall [815] is perhaps the best candidate in Trench 8 for a colonnaded wall. The presence of the rudimentary drain that would have deposited large quantities of water from the step directly into the courtyard suggests that this was most likely a utilitarian area where ornamentation was not a major concern. an existing surface without foundations, again suggesting a later alteration to the layout of the building. Only a short length of this wall was exposed in the trench and it appears to have been originally bonded to, or butted against, a similarly shallow north-south wall that has since been completely removed but whose presence is shown by the robber cut [853] along the eastern trench edge (Fig. 77). This cut was only found in 8.3i indicating that in its later phases this part of the building’s central ‘aisle’ consisted of rooms perhaps accessed from a narrow corridor (8.3ii). The geophysical results have identiied other possible internal divisions in this ‘aisle’ further west. The latest surfaces in both excavated areas of 8.3 consisted of thin deposits of compacted pinkish clay (813 and 842) suggesting that these spaces were provided with rudimentary beaten clay loors. Area 8.3ii was excavated to a level below the latest phase and here a series of lat stones, possibly deliberately laid, might suggest an earlier stone loor (842). Areas 8.2, 8.3 and 8.4 - the aisled building The part of the building contained in the central and northern parts of Trench 8 included three internal spaces separated by east-west walls, all of which had been robbed to below the levels of the latest loors. The southern space 8.2, bounded to the south by wall [815] and to the north by the wall [812], was 2.75m wide, the central space 8.3 between [812] and the next wall [821] was 5.5m wide, and only 1.2m of the most northerly space 8.4 was exposed in the trench (the geophysical results suggest that this part of the building was probably as wide as 8.3). Walls [812] and [821] were more solidly built than [815] with neatly dressed facing stones of ORS bonded with durable mortars (the cores of both walls contained fragments of CBM). It is noteworthy that the middle wall [812] was wider than the walls to either side (0.7m compared to 0.55-0.60 m), and it is possible that the weight of the roof was supported on this wall or that this part of the building was taller. DECAY AND ROBBING Overlying the uppermost clay surfaces in area 8.1 was a thick layer of clayey silt containing large quantities of broken roof tiles and some rubble (824/827), which presumably derives from the removal of the adjoining building’s roof. Within this deposit was a curvilinear arrangement of faced stones (823) abutted by stacks of possibly deliberately placed broken bricks and roof tiles (828) (Fig 78), perhaps indicating activity of some kind in this area during the period of the building’s abandonment (it is also possible that this feature relates to the drain mentioned previously). Above this were various rubble deposits related to the collapse or demolition of the building (there was more roof tile in this part of the trench that above the building to the north) and topsoil. Area 8.2 was divided into two by a slightly crooked northsouth mortared wall [805] that ran between walls [815] and [812], but was not bonded to them (Fig. 76). The wall was 0.46m wide and consisted of a core of rubble and large pieces of broken roof tile between faces of roughly cut ORS, some of which still had lime mortar adhering to them. The wall was not constructed with foundations, but was simply built on top of an existing surface and it appears that it was was a later addition to the building inserted to create separate rooms 8.2i and 8.2ii (the geophysical results suggest that this southern part of the building could have consisted of a row of rooms). The wall was associated with poorly-preserved opus signinum loors in the two rooms (832/835), while the presence of patches of lime mortar on both faces of [805] suggests that both rooms were rendered (no evidence of painted plaster was recovered). The loors in Areas 8.2 and 8.3 were overlain by a series of deposits containing quantities of degraded plaster, mortar and CBM fragments that must have originated from the building’s abandonment and dereliction. It was from this level that cut [853] to rob the later wall dividing 8.3 was noted, which if correct indicates that some of the building’s internal walls were being dismantled while the superstructure was still standing. Thick layers of dark silty soil with large quantities of building rubble and broken roof tiles (the tile was concentrated in the central and southern parts of the trench) lay on top of these deposits, including (839), (804/806) and (808/822). These rubble deposits were cut by lat-bottomed and vertical-sided trenches for the robbing of the major east-west walls ([820], [811] and [814] ran along walls [821], [812] and ([815] respectively). The fact that wall [805] was not robbed suggests that it had been partly dismantled already (or had collapsed) and was not visible when the robbing took place. This, combined with the absence of large quantities of building stone in the upper parts of the trench, indicates To the north the wider Area 8.3 had also been subdivided into two, though here the additional wall [838] ran parallel to the direction of the building (i.e. east-west) creating smaller spaces some 2.9m and 2.2m wide (8.3i and 8.3ii respectively). This wall was 0.83m wide with a rubble and tile core and also had been built directly onto 64 Excavation results, Trench 8 ig. 76. Wall [805] separating rooms 8.2i (top) from 8.2ii (bottom). Taken from the east Fig. 77. Section of wall [838] from the east. The wall separated rooms 8.3i (left) from 8.3ii (right), and probably butted against a north-south wall that has been robbed (edge of robber trench [353] cuts the clay surface in 8.3i) 65 Excavation results, Trench 1 Fig. 78. Arrangement of stones and CBM in Area 8.1 perhaps that the major robbing event occurred sometime after the building’s demolition. A rubble layer with CBM inclusions (807) post-dated this robbing event in the southern end of the trench, which lay beneath the topsoil (801). 66 Trench 9 Trench 9 was the most northerly of the trenches excavated in 2011 and was positioned across the full width of the narrow northwest-southeast aligned building that lay between the western courtyard of the Southern Canabae (zone 3), and Building D excavated by NashWilliams in the 1950s. The trench exposed two raised rooms or corridors, provided with opus signinum loors and walls decorated with painted plaster, between two probable external areas (Figs 79 and 81). The deposits associated with the building’s use and demolition lay no more than 0.3m below the modern ground surface and were remarkably well preserved (Fig. 80). Ruts caused by modern agricultural vehicles passing over this part of the ield were shown to be actively cutting into the Roman archaeology in the southern part of the trench. The rooms or corridors of the building were considerably higher than the surrounding ground level and had been clearly raised with substantial quantities of levelling material. The absence of steps or other means of entering the building from the north and south indicates that access into and through the building itself was gained from its short eastern and western sides. The excavated evidence combined with the gradiometer results suggests that the narrow building could have been the rear wing of the monumental Building D whose entrance opened onto the amphitheatre. Trench 9 produced very little Medieval or modern material, suggesting that the Roman building was the irst and last phase of signiicant occupation in this part of Caerleon. The provisional stratigraphic matrix for Trench 9 is located in Appendix 1. CONSTRUCTION AND OCCUPATION Rooms 9.2 and 9.3 - the ‘corridor’ building. The east-west walls of the 12m wide building had been robbed either to their lowest courses or their foundations. Although the southern external wall [931] had been completely robbed to its cobbled footings, these showed that it was considerably wider than the northern external rth s tre 9.4 r Fo o sN 9.3 9.2 9.1 0 10m Fig. 79. Trench 9 showing walls and surfaced areas superimposed over gradiometer results. Numbers refer to ‘Areas’ as described in the trench discussion 67 Excavation results, Trench 9 Trench N 939 934 952 945 965 946 958 932 956 Area 9.4 952 957 945 962 960 902 932 962 963 947 949 917 919 924 940 Room 9.3 Fig. 80. Trench 9 fully excavated, from south wall [917] (0.98m compared to 0.60 m), and it is possible that this was the main (rear) wall of the Building D complex. A third wall [918], 0.75m wide, divided the building longitudinally into two long internal spaces (at some 5m the southern, or rear, internal area 9.2 was wider than the northern internal area 9.3 which measured 3.7m wide). The lower courses of walls [917] and [918] had not been robbed and consisted of rubble cores between faces of roughly cut blocks of ORS bonded with orange sandy mortars. 918 922 914 Room 9.2 Several unexcavated mortary deposits visible in the sides of the later robber trenches possibly relate to the construction phases of these walls. The robber trenches also showed that a series of substantial deposits had been dumped between the walls, particularly (936) in 9.2 and (940) in 9.3, to raise the interior of the building before the irst the loors were laid. These deposits were up to 0.4m deep in 9.2, but decreased in depth from south to north within the trench as the builders worked with the natural slope of the ground on which the building was constructed. 941 943 951 950 923 931 916 Area 9.1 916 The make-up in room 9.2 was overlain by a layer of medium-sized angular blocks of ORS (943), over which the concrete for the irst opus signinum loor 0 2m Fig. 81. Final plan of Trench 9 68 Excavation results, Trench 9 Fig. 82. Room 9.2 from south, showing earliest opus signinum loor (941) beneath second opus signinum loor (907/914 (941) was poured and levelled (Fig. 82). This loor had been intensively used and the surface shown signs of considerable wear-and-tear. A possible accumulation layer consisting of ine brown silts (932) lay on top of this loor, which in turn was sealed by a second and inal opus signinum surface (907/914). The same construction sequence was observed in area 9.3 where a layer of ORS (930/925) levelled the make-up deposits (940/926) and formed the base for an opus signinum loor (910). Unlike in 9.2, however, the loor in this internal area was not replaced and the original surface was used throughout the duration of the building’s life (Fig. 83). from wall [917] it must predate the decay or demolition of the building. Furthermore, the pit had been illed with dumps of sandstone and tufa before being deliberately levelled with deposits of sand and decayed mortar (929), as if the pit had been dug and illed while 9.3 was still in use within the standing building. Area 9.1 - southern ‘yard’ The southern end of the Trench 9 included a small part of what is likely to have been an open area behind the rear wall of the corridor building just described (the geophysical results identiied a probable courtyard, some 25-30m wide and potentially up to 40m long, immediately to the southwest of the building). The evidence provided by the building’s walls and the dimensions of its two internal spaces suggests that the narrower northern area 9.3 possibly served as a corridor, while the wider southern area 9.2 might have have been subdivided by lateral internal walls (outside the 2m wide evaluation trench) to form a range of rooms with doorways from the corridor. Further excavation would conirm the precise layout of this building, but the geophysical results suggest that such an arrangement of corridor leading to rooms is certainly possible. Although the trench [923] that robbed wall [931] extended to a depth of more than a metre below the loors inside the building, no surfaces of this external area were visible in the southern face of the trench. Instead the sequence of deposits in area 9.1 consisted of a mixed layer of sandy silt (916) with large quantities of pottery, animal bone and occasional glass fragments, below a similar deposit containing numerous small inds as well as rubble and CBM fragments (908). These probably represent episodes of rubbish dumping followed by the accumulation of debris from the building’s decay and demolition. Therefore, the ground level of the courtyard must lie beneath these, although how much lower is uncertain. This demonstrates the extent of the terracing in this part of the Southern Canabae that was required to In area 9.3 most of the opus signinum loor (910) and its underlying rubble bedding (930/925) had been cut away by a large shallow pit [924] whose western side lay just within the trench and which extended for an unknown distance beyond the trench to the east. The purpose of this pit is unclear, but as it was sealed by fallen plaster 69 Excavation results, Trench 9 Fig. 83. Room 9.3 from east, showing half-sectioned pit [924], and opus signinum loor (910) on right create a platform for the corridor building, and the height difference and the absence of steps must mean that the building was not accessed from the southern courtyard. [960] may have run perpendicularly eastwards from wall [932], though only a short length of it was included in the trench and the evidence for this wall is less convincing. Area 9.4 - northern area The area to the northeast of the corridor building in the northern part of Trench 9 was also considerably lower than the corridor building. The trench was excavated to a depth of 1.2m below the modern ground level and as no substantial surfaces had been observed at this point it is uncertain, therefore, if this part of the trench was inside another building or had been an external area. A series of brown and orange sandy layers – including (945/946/947/949/95/965) - butted up to the lower courses on both sides of walls [932] and [958], and these must have been deposited while the structures formed by these walls were standing. These layers gradually became shallower to the north as the pre-existing ground surface sloped upwards, though whether they represent construction activity or surfaces is unclear. Three connected walls were discovered in area 9.4 that were part of a structure of some kind (Fig. 84). These were narrower and more poorly constructed than the building’s walls, consisting of ORS blocks of various sizes bonded with a yellow sandy mortar, but without a consistent rubble core. A north-south wall [932] extended from the north wall [917] of the corridor building and continued northwards beyond the end of the trench. This wall varied in width from 0.5m to 0.6m and was not quite straight along its observed length. Unusually, it was bonded into the lower courses of [917] but not the wall’s two uppermost surviving courses, suggesting either that it had supported a timber superstructure or, perhaps more likely, that it was a sleeper wall for a colonnade or portico. A second wall [958] at right angles to wall [932] extended for an uncertain distance eastwards, and the fact that these walls were bonded together shows that they had been constructed at the same time. A possible third wall The north-south wall [932] continued beyond the northern end of Trench 9 towards the monumental Building D excavated by Nash-Williams. Lying immediately west of the amphitheatre, this building was oriented on the same alignment as the Southern Canabae and its northwestern side included a row of rooms that apparently head directly towards Trench 9. It is possible, therefore, that walls [932] and [958] are a continuation of this range of rooms and that Building D extended as far as the corridor building described previously. If this is indeed the case, the position of wall [958] suggests that the rooms, were they of equal size, would have been approximately 4.3m wide. DECAY AND ROBBING Rooms 9.2 and 9.3 Dumps of painted wall plaster (909) in the northern part of room 9.3 lay on top of the opus signinum loor 70 Excavation results, Trench 9 Fig. 84. Area 9.4 from north, showing walls [932] and [958] (910) as well as the in-illed pit [924] (Fig. 85). Red and white were the predominant colours on the plaster and, although fragmentary, it appears that the decorative scheme comprised red lines forming squares or rectangles on a white background. It is most likely that this had decorated the internal face of wall [917], but it is not certain whether the plaster fell while the building was still standing (presumably no longer in use and partially derelict), or during the robbing of the building. Similarly fragmentary deposits of painted wall plaster (913) were found overlying the loor (914) in room 9.2 close to wall [918], indicating that this space had been decorated too. foundations in trenches [919] and [922] respectively. The ills of these robber trenches contained large quantities of building debris as well as other material including pottery, and the building was overlain by shallow silt deposits including concentrations of medium-sized ORS rubble mixed with quantities of rooing material and decayed mortar. Area 9.4 In the northern part of the trench dark silt deposits (929/934) overlay the sandy layers around the walls, and they also appear to overly the uppermost surviving courses of north-south wall [932] (possibly supporting the observation that this had been a low sleeper wall that was no longer in use when these silts were deposited). Layers of mixed ORS and tufa rubble (some blocks with mortar still adhering to them) in a dark sandy matrix overlay the dumps of collapsed wall plaster in both rooms 9.2 and 9.3 (904 and 906 respectively). These were then cut by robber trenches that removed the lower courses of the building’s two external walls [917] and [931] as well as its internal wall [918]. All of the lower courses of the southern external wall [931] had been removed and the robber cut for this wall [923] bottomed on its cobbled foundations at just over 1m below the latest Roman loor surfaces. Walls [917] and [918] had not been completely robbed and the lowest courses survived above their Although the evidence is not entirely clear, it appears that a shallow cut [952] along the wall [932] closest to the corridor building indicates that even this low wall had experienced some robbing, though this activity must have occurred before deposits (935/920) and (902) were laid down as these seal the uppermost ill of the robber trench. No evidence for robbing was noted for the short section of wall [958] within the trench. Layers (935/920) produced large quantities of animal bone, pottery, shell 71 Excavation results, Trench 9 and charcoal, suggesting that the area immediately north of the corridor building was used for the disposal of rubbish after the structures formed by walls [932/958] had gone out of use or been robbed away. Trench [919], associated with the robbing of the corridor building’s northern wall [917], was cut from the level of layer (902) indicating that the structures in the northern part of Trench 9, whatever form these took, went out of use and were robbed earlier than the building itself. The uppermost layers in the northern part of the trench consisted of thick deposits of ORS rubble with some fragments of CBM lying directly below the topsoil. The stones here were generally much larger than those found overlying the corridor building where more roof material was recovered. Fig. 85. Area 9.3 from west, showing collapsed wall plaster (909) above the opus signinum loor (910). The foundations of wall [917] are visible on the left of the photo 72 Artefact & Environmental Assemblages The 2011 excavations produced a typically large and varied range of material culture and environmental evidence. At the time of writing the work of conserving, identifying and describing this material is on-going; Adrienne Powell has completed the and assessment of the animal bone, Drs Peter Webster and Mark Lewis have inished the preliminary identiication of the pottery assemblage, while the metal and other small inds are scheduled to undergo conservation in the laboratories at Cardiff University from the Autumn of 2012. The pottery assemblage is presented in the next section, while the following discussion summarises the extent and nature of the various categories of non-ceramic inds, including an initial examination of the patterns of deposition between the nine trenches. Trenches 1 and 5 produced the largest collections of small inds, though probably for different reasons. In Trench 1 the majority of artefacts came from deposits used to raise and level the buildings on the east side of the tegula wall (rooms 1.1 and 1.2), while in Trench 5 the rubbish deposits overlying the latest Roman loors in rooms 5.3 and 5.4 produced the greatest quantity of small inds from this building. Although the processes that led to the deposition of large numbers of objects in Trenches 1 and 5 were different, it is likely in both cases that these artefacts were not directly associated with the functions of these buildings. Iron, lead and glass inds were particularly common from Trench 1, whereas glass and copper allow objects were found more frequently in Trench 5. Elsewhere, a surprisingly large assemblage of glass was recovered from Trench 2 (though not all of it was Roman), Trenches 3 and 4 produced very few inds, while the objects from Trench 9 included relatively large quantities of iron as well as coins. SMALL FINDS / REGISTERED ARTEFACTS A total of 475 objects classiied as small inds were recovered, including a wide range of artefacts of copper alloy, iron, lead, glass, worked bone, ceramic, and jet or shale. All stratiied ind spots were recorded in three dimensions and the distribution of small inds between the nine trenches is shown on Table 1 and Fig. 86. Coins Copper alloy* Iron Lead Glass ** Anteixa Ceramic *** Bone Jet / shale Stone Intaglio Total Total 20 94 83 92 134 4 10 10 2 25 1 475 Trench 1 4 24 32 40 34 T2 1 1 1 5 14 Although the small inds are yet to be cleaned and conserved, it is apparent that the assemblage includes far fewer military inds that was the case in Priory Field. Instead, the majority of registered artefacts from T3 T4 2 7 4 7 10 2 1 1 4 9 10 10 11 1 32 94 1 6 145 24 1 9 Table 1. Distribution of small inds / registered artefacts (by count) * non-numismatic objects ** beads and vessel glass *** decorated samian and grafiti 73 T6 23 14 11 30 1 3 5 1 6 5 1 1 T5 T7 2 7 4 4 19 3 T8 2 12 3 11 7 T9 9 9 14 3 5 1 1 1 3 1 6 3 1 42 48 39 42 Artefact & Environmental Assemblages 50 coins copper alloy (non-numismatic) 40 iron lead glass (beads and vessels) Quantity 30 20 10 0 Trench 1 Trench 2 Trench 3 Trench 4 Trench 5 Trench 6 Trench 7 Trench 8 Trench 9 Fig 86 . Distribution of small inds from the Southern Canabae the Southern Canabae seem to be personal, domestic and craft items. This perhaps relects the activities that took place in the complex, or possibly the domestic and industrial nature of the rubbish dumped into the abandoned buildings in Trench 5 and possibly elsewhere. buildings produced greater quantities of such material, although even here there were signiicant differences between the material recovered. Trenches 1, 4 and 5, for instance, produced large assemblages of animal bone and pottery, while Trenches 6, 8 and 9 did not. BULK FINDS The material classiied as bulk inds, and therefore not registered in the same way as small inds, includes pottery, animal bone, iron nails, non-vessel glass, metallurgical and glass slags, tesserae and clay-pipe. The distribution of these inds is shown on Table 2 and Fig 87, where it is apparent that different quantities of material were recovered from the nine trenches. In part this relects the depth to which the trenches were excavated, but these patterns must also relect the intensity and type of occupation in different parts of the Southern Canabae too. For example, Trenches 2 and 3 in the centre of the main courtyard building produced very little material at all, suggesting that this large open area was not used in a way that resulted in the loss of iron nails, broken ceramic vessels or animal bone. The trenches located within other It likely that this is an effect of the various processes that formed the stratigraphic deposits within which this material was contained, such as the dumping of rubbish in rooms and spaces in Trenches 4 and 5 when these buildings were no longer in use. It is important to bear in mind, therefore, that the material recovered from the trenches could have originated elsewhere and does not necessarily refer to the original functions of the excavated buildings themselves. In Trench 1, however, much of the ceramic assemblage was recovered from levelling deposits within the buildings closest to the River Usk and this will be important when understanding the chronology of these buildings’ construction and use. It is noteworthy that these deposits in Trench 1 produced similarly large quantities of metallurgical slags but relatively little animal bone, indicating perhaps that the material used to Total Trench 1 T2 Animal bone 4674 168 3 Ceramics Iron nails 6882 785 1582 169 363 17 Glass* 1806 15472 259 Slag ** Tesserae** Clay-pipe T4 T5 T6 T7 T8 T9 751 1810 227 1182 157 376 116 6 1472 82 2063 170 309 32 436 112 239 73 302 124 581 8 235 329 38 143 66 147 6978 616 687 25 1670 851 3195 1480 1030 8 57 741 1099 140 1 11 7 8 2 1 17 47 3105 94 T3 Table 2. Distribution of bulk inds * window and modern glass, by weight (g) ** by weight (g) 74 Artefact & Environmental Assemblages 2500 2000 animal bone ceramics iron nails Quantitiy 1500 1000 500 0 Trench 1 Trench 2 Trench 3 Trench 4 Trench 5 Trench 6 Trench 7 Trench 8 Trench 9 8000 6000 glass (window and modern)* weight (g) slag (metal and glass)* tesserae* 4000 2000 0 Trench 1 Trench 2 Trench 3 Trench 4 Trench 5 Trench 6 Trench 7 Trench 8 Trench 9 Fig. 87. Distribution of bulk inds from the Southern Canabae raise the interior of these buildings may well have been brought here from an area were metalworking took place. from Trenches 5 and 8, while almost non-existent from Trenches 2 and 3. In the latter cases the absence of roof tiles is presumably because the structures located in the open area of the main courtyard building either did not have tiled roofs or were not roofed at all. It is interesting to note, however, that smaller quantities of roof tiles were recovered from Trenches 1, 4, 6, 7 and 9, all of which exposed parts of Roman buildings that must have been roofed with terracotta tiles like those in Trenches 5 and 8. The differential recovery of tegulae and imbrices is likely, therefore, to be in part the result of some roofs being salvaged (Trenches 1, 4, 6, 7 and 9), while others appear to have remained in place until they collapsed and became part of the archaeology of the buildings they had once covered (Trenches 5 and 8). The discovery of tesserae from mosaic loors was limited to Trenches 4, 5 and 7. Those from 5 and 6 were small and predominantly black or white, which in the case of Trench 5 must have come from the disturbed tessellated loor partly revealed in the northern end of the trench. The two rooms exposed in Trench 7 were not tessellated (the hypocaust loor in room 7.1 was opus signinum), and it is likely that a mosaic existed in one of the adjacent rooms. In contrast, the tesserae from Trench 4 were larger, crudely fashioned, and all white or light grey. Again there was no trace of a mosaic in the trench, but the concentration of tesserae towards the northern end suggests a relatively unsophisticated tessellated loor close by. CERAMIC BUILDING MATERIAL Over three tonnes of brick and tile was recovered during the evaluation excavations. The distribution of this material by type between the nine trenches is shown on Table 3 and Fig. 88. Assuming that bricks and tiles will not have moved very far from where they had been used in walls and roofs, these reveal details of the buildings’ original architecture as well as indicating the different histories of buildings once they were abandoned. Tegulae and imbrices from roofs were most common 75 Bricks were recovered from seven of the nine trenches, but are most common from Trench 7 where many of the examples found there must have been used in the pilae to support the heated loor. Trench 7 also produced several pieces of box lues that presumably were also used in the hypocaust, though box lue tiles were most common from Trench 5 where they appear to have been used to reinforce the barrel vault at the north end of the trench (the single complete box lue from Trench 5 was 0.39m long, 0.19m wide, 0.13m deep and with sides 0.02m thick). Artefact & Environmental Assemblages Tegula Imbrex Brick / pila Box lue Segmented circular tiles Unidentiied Total Total Trench 1 T2 T3 T4 T5 T6 T7 T8 T9 1090 396 99 26 11 3.5 1 0.5 49.5 15.5 40 0.5 399.5 218 32 20.5 125.5 22.5 2 69 13 96 2.5 295 86 22 41 11.5 24 0.5 3.5 25.5 4 109.5 215.5 501 1174.5 141 569.5 42.5 123 217 23.5 33.5 1285.5 3045.5 146 271 11.5 27 33.5 34 163.5 313.5 137 317.5 Table 3. Distribution of ceramic building material by weight (kg) 500 tegula imbrex brick (incl. pila) 400 box flue segmented circular tiles Weight (kg) 300 200 100 0 Trench 1 Trench 2 Trench 3 Trench 4 Trench 5 Trench 6 Trench 7 Trench 8 Trench 9 Fig 88. Distribution of ceramic building materials from the Southern Canabae Several trenches also produced examples of segmented circular tiles, though they were most frequent in Trench 8. Each tile formed one quarter of a circle and they would have been laid in courses to build columns or engaged (half) columns. Trench 8 showed that at least two sizes of brick-built column were used in the Southern Canabae: the smallest had a radius of 0.175m and the bricks were 0.05m thick, while the largest had a radius of 0.24m and the bricks were 0.07m thick. Such segmented bricks are not a common feature of Romano-British architecture, military or civilian, and only a few are known from within the legionary fortress at Caerleon. ANIMAL BONE1 All bone was scanned and counted and the presence of ageing, sexing, and butchery data was noted, as were measurable specimens. Most of the bone was unwashed at the time of evaluation and hence some surface features such as shallow gnawmarks and ine cutmarks may 1 The following is an abridged version of Adrienne Powell’s assessment of the animal bone assemblage (analysis was undertaken in the Osteoarchaeology Laboratory at Cardiff University). have been obscured. Vertebrae and ribs were counted as unidentiiable. Phasing and context information were not available at the time of evaluation hence all bone is treated as single phase. There is a total of 4674 fragments in the assemblage (Table 4), excluding bone from the topsoil, of which 2159 fragments are identiiable. Most of the material comes from Trench 5 (39%), Trench 7 (25%) and Trench 4 (15%); these trenches also contained greater proportions of identiiable bone than the other trenches. The overall level of identiication, at 46%, is quite high and suggests some retrieval bias. The main taxonomic groups present are shown in Table 5. The major domestic species, cattle, sheep/goat and pig, comprise 49% of the identiiable assemblage. Of these, the most frequent species overall is pig (61%), although the frequency varies between trenches. No systematic recording of anatomical elements was made during evaluation, although foot bones, particularly metapodials, are notably common. Cattle is the next most abundant taxon overall (24%) and is particularly frequent in Trenches 8 and 9, even outnumbering pig in the latter. Sheep/goat makes a relatively minor contribution overall 76 Artefact & Environmental Assemblages Trench Identiied Bone Unidentiied Bone Total % ID 1 2 4 5 6 7 8 9 Total 25 143 3 391 901 148 584 107 238 2515 168 3 751 1810 227 1182 157 376 4674 15 0 48 50 35 51 32 37 46 360 909 79 598 50 138 2159 Table 4. Summary of the animal bone from the Caerleon Southern Canabae evaluations (15%) and includes two goat specimens. A group of 46 sheep/goat bones from context (534) appear to be mostly from the same, juvenile, individual. Small numbers of equid, cat and dog bones are also present. The dog bones from Trench 5 include a partial skeleton of a dwarf dog comparable with a modern corgi reference skeleton, as well as an isolated tibia from a dwarf animal. Remains from a small, non-dwarf, canid either fox (Vulpes vulpes) or small dog are also present. Butchery evidence is infrequent but may have been obscured by the unwashed state of most of the material. There a few examples of worked bone in the form of sawn antler segments, a deer metacarpal with the distal end sawn through and a sawn segment of large mammal bone. Measurable elements are frequent as are ageable bones and jaws: the assemblage includes neonatal bones from cattle, pig and dog. Sexable fowl tarsometatarsi are frequent, sexable pig canines are present but less frequent. Some evidence of pathological conditions has survived. There is both charred and calcined bone present, and much bone has a dark colour, although it is unclear whether this is due to charring of staining. Carnivore and rodent gnawing damage was present. Wild mammals present include red/fallow deer (Cervus/ Dama), roe deer (Capreolus capreolus), fox, hare (Lepus sp.), hedgehog (Erinaceus europaeus), water vole (Arvicola terrestris) and unidentiied small mammal. Bird bones are unusually numerous, as a group outnumbering the pig remains; they are most abundant in Trenches 5 (43% of total identiied) and 6 (46%). Bones of domestic fowl are the most prevalent, however, other species present include goose, duck, pigeon (Columba sp.), raven (Corvus corax) and medium and small passerines. A small amount of ish bone is present, concentrated in Trenches 5 and 6. Other species present include frog (Rana sp) and toad (Bufo sp), most of which occurred in a large (ca. 250 bones) concentration in context (706). Trench Cattle 1 15 Sheep/ goat 3 4 49 29 159 2 5 52 76 300 1 6 1 2 31 1 2 Pig Equid Comparison of this Southern Canabae evaluation animal bone assemblage with previously published groups from Caerleon (Hamilton-Dyer 1993, O’Connor 1986) highlights several interesting features. Firstly, the predominance of pig bones in contrast to the assemblages from the Scamnum Tribunorum and the Legionary Fortress Baths where cattle predominate (save for the smaller samples from drains in the latter where sheep/ goat bones are more numerous). Cattle bones were also the most frequent in the unpublished material from the Dog 7 Other mammal Bird 27 93 1 57 394 6 5 36 3 50 93 1 4 9 9 152 32 657 Fish Other Total 25 0 2 7 65 30 97 8 17 3 17 9 Total 48 247 10 153 32 643 2 8 20 5 3 28 Table 5. Animal taxa present in the Caerleon Southern Canabae assemblage 77 360 3 909 79 255 598 50 1 12 1 259 138 2159 Artefact & Environmental Assemblages recent excavations in Golledges Field. Perhaps related to the species proile is the low incidence of observed butchery in the evaluated bone: in both published assemblages cattle butchery is indicated primarily by chopmarks, whereas knife cuts are more prevalent in the sheep/goat and pig remains and these are more likely than chopmarks to have been obscured by the unwashed state of the current assemblage. Secondly, the frequency of wild mammal bones (7%) is greater than in the published material. Thirdly, the frequency of bird bone is remarkably high. Fourthly, the presence of dwarf dog bones: bones from small dogs were also noted in the Legionary Baths assemblage and at Caerwent (Noddle 1983), however, beyond describing them as lap-dogs little information about their morphology was provided and it would be interesting to compare these bones to see if different types can be identiied. 78 Pottery1 INTRODUCTION All pottery (approximately 93.25kg) from the 2011 excavations has been examined and an archive list produced dividing the material by context and source and, where possible, giving ‘spot dates’ to individual vessels (quantiication has been by estimated vessel numbers and by weight). The comments below are based on this archive list and a brief summary by context has also been produced (the quantiied archive and synopsis are available from the authors on request). century loruit. This is apparent if we show in histogram form all vessels to which more than a generalised date can be given (Fig. 89). There are, however, differences between the trenches and we have, therefore, included a separate histogram for each trench below. When considering these it is important to remember that the yield per trench will vary according to its size and the depth of stratigraphy investigated. The total number of vessels speciically dated from each trench is as follows: Trench 1: 175; Trench 2: 51; Trench 3: 10; Trench 4: 156; Trench 5: 178; Trench 6: 40; Trench 7: 59; Trench 8: 28; Trench 9: 41. It will be clear, therefore, that Trenches 1, 4 and 5 are likely to provide the best indications of the pattern on the site as a whole and that Trench 3 can do little more than provide a glimpse of what may be there. Before making speciic comments it must also be pointed out that the nature of a preliminary excavation imposes limits on the amount of information which can be deduced from the pottery recovered. The various trenches were not all of equal size making direct comparison between them dificult. Not all were excavated down to 70 60 Vessel loss per decade 50 40 30 20 10 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 0 Fig. 89. Roman pottery from the 2011 Southern Canabae excavations Taken as a whole, the pottery includes some pieces from the very beginning of the Roman occupation of the fortress including examples both of the samian form 29 and the ‘Hofheim’ lagon type, both of which could have appeared with the irst construction of the fortress and are unlikely to have been current much after the early 90s AD. However, all trenches show a preponderance the natural subsoil, a factor which must have bearing on the chronological comments below. Nevertheless, a fairly consistent pattern does seem to emerge. CHRONOLOGY In general all trenches show a spread of pottery from the later irst century to the early/mid fourth and with a second POTTERY1 1 The following is an abridged version of the authors’ pottery assessment. These are preliminary comments based on a review of the entire pottery assemblage and the next step will be to consider the material in relation to the stratiied grouping of contexts. 79 The Pottery of late irst and second century vessels and, thereafter, a diminution. Most areas have some vessels which belong to the later third to fourth centuries but the number of fourth century pieces is not such as to suggest occupation beyond the early to mid century and this is backed up by the absence of pieces which certainly belong to the second half of the century. mid fourth century pieces (only Trench 7 shows a similar pattern). Trench 6 shows a clear early to mid 2nd century peak and a marked decline in activity after c. A.D. 200. Trench 7 shows a similar pattern to Trench 5 but with a more even distribution of types between the early second and the early/mid fourth centuries (Fig. 93a). Trench 8 igures are again based on a small sample, but show a late irst to mid second century peak tailing off across the remainder of the second century and with little thereafter (Fig. 93b). Trench 9 has comparatively more later pieces than most areas. The main loruit is again in the second century and, as with all areas, there is a third century dip. The increase in late third and fourth century pieces is, however, noticeable here (Fig. 93c). Looked at on a trench by trench basis we can see some variation across the extramural complex. Trench 1 has a preponderance of types from the period between the very late irst century and the very end of the second (Fig. 90a). Thereafter, there appears to be some activity on or near the area up to the early/mid 4th century. Trench 2 igures are based on a much smaller sample, but there is an even greater concentration of late irst to mid 2nd vessels and very little from any later period (Fig. 90b). Trench 3 igures are based on a very small sample indeed but seem to mirror those from the nearby Trench 2 (Fig. 91a). The almost total absence of third century pottery from either trench is notable. Trench 4 igures are based on a more extensive collection but are almost entirely concentrated on the period from the late irst to the very end of the second centuries (Fig. 91b). We may note that the trenches are numbered across the site from east to west and it looks as if the main courtyard building and port (i.e. Trenches 1 to 4) saw little activity after the very end of the second century. Many sherds appear to have been moved around in the ground and only a few contexts appear to have produced near complete vessels. However, Trench 5 contexts (534-5) appear to be an exception and contain large fragments of vessels freshly broken when they entered the ground. Trench 5 yielded the largest collection of pottery and the most later (third and fourth century) material (Fig. 92a). There is again a late irst to late second century peak but there are also signiicant amounts of third and early to The shallow depth of topsoil produced, as one might expect, a number of contexts which look as if they may be disturbed. However, this is mitigated by the small quantity of Medieval and later pottery (far less than in 20 5 A B Vessel loss per decade 4 12 8 3 2 1 0 0 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 4 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 Vessel loss per decade 16 Fig. 90. Roman pottery from Trench 1 and Trench 2 20 2.5 A B 16 Vessel loss per decade 1.5 1 12 8 4 0 0 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 0.5 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 Vessel loss per decade 2 Fig. 91. Roman pottery from Trench 3 and Trench 4 80 The Pottery 6 15 A Vessel loss per decade Vessel loss per decade B 5 12 9 6 3 4 3 2 1 0 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 0 Fig. 92. Roman pottery from Trench 5 and Trench 6 5 4 A 4 B Vessel loss per decade Vessel loss per decade 3 3 2 2 1 1 0 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 0 4 C Vessel loss per decade 3 2 1 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300 310 320 330 340 350 360 370 380 0 Fig. 93. Roman pottery from Trench 7, Trench 8 and Trench 9 the light railway system leading in the general direction of the 2011 site and some spillage of amphitheatre spoil over the site cannot be entirely ruled out. There are a number of layers with suspiciously low numbers of rims, for instance, making the presence of pottery ‘weeded’ from the amphitheatre assemblage at least a possibility. It will be worthwhile to examine the contexts of these layers in due course. the Priory Field excavations of 2008 an 2010). One may suggest that medieval disturbance was probably slight, perhaps because the area was rough grazing and not used for arable farming. There is no evidence of late medieval and early post-medieval activity but some for the period between the mid 17th and the 19th centuries, although it may be noted that this seems to concentrate in the area of the more northerly Trenches 8 and 9. The spreading of material from middens over this part of the site seems possible. SOURCES With a few exceptions, the sources of Roman pottery from the 2011 excavations are as one would expect for a Caerleon site. There are some signiicant differences with the Priory Field material and some interesting presences and absences, so that a brief resumé of the main sources of pottery is worthwhile. Prior to the 2011 season it was thought likely that the area being excavated had acquired its uneven appearance due to the spreading of spoil from the nearby amphitheatre excavations. This was clearly not the case but it remains true that photographs of the Wheeler excavations show 81 The Pottery Finewares As one would expect from a site with a second century loruit, samian forms a major component of the inewares. There are some South Gaulish pieces including a couple of form 29s, but only one Les Martres piece (Table 6). The remainder are mainly Central Gaulish but there are a few pieces of East Gaulish ware also. the numbers of decorated vessels seem low and disproportionately weighted in favour of Trench 1, as the breakdown on Table 7 makes clear. The greater incidence of the early South Gaulish form 29 over the somewhat later South Gaulish form 37 in Trench 1 is unusual on any Caerleon site with occupation through into the second century and may indicate some early building in the vicinity. A representation of samian as a percentage of all pottery again makes clear that Trench 1 is exceptional, while the percentage of samian from the more westerly trenches seems low (Table 8). Samian forms are, of course, more easily identiiable from very small fragments than most other pottery shapes and the igure of 189 vessels may be disproportionately high (approximately 25% of all vessels to which a irm date could be given). Within the samian assemblage SOURCE SG LMdV CG EG TOTALS FORM 18 18R 18 or 18/31 18/31 27 29 36 37 67 78 C.11 cup bowl 18/31 33 bowl 18/31 18/31 or 31 27 31 31R 33 36 37 38 45 46 72 C11 C15 C21 Cup Bowl Beaker Flagon 31 31R 45 C11 Lud.Sb Bowl T1 2 T2 4 1 3 5 5 1 4 1 2 2 T3 T4 T5 T6 1 T7 T8 T9 1 1 1 1 3 1 1 2 1 3 3 1 3 2 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1 6 3 3 1 1 1 6 3 3 5 1 2 3 1 12 1 1 1 1 1 1 1 6 1 1 1 2 4 6 2 3 1 1 2 1 1 1 1 2 1 1 3 1 1 2 3 1 2 1 1 1 1 1 1 2 68 17 3 30 31 Table 6 Samian ware from the 2011 excavations 82 11 16 6 1 7 TOTALS 7 1 2 9 11 8 6 11 1 2 2 2 2 1 1 1 4 2 1 20 6 14 3 27 7 1 2 4 1 1 1 1 17 1 1 1 1 1 1 3 1 189 The Pottery SOURCE SG SG SG SG CG CG TOTALS FORM 29 37 67 78 37 72 T1 5 4 1 12 1 23 T2 T3 3 1 1 T4 1 2 T5 T6 T7 T8 1 1 T9 1 5 2 3 5 4 2 1 1 4 2 3 2 1 4 2 8 TOTALS 8 11 1 2 27 4 53 Table 7 Decorated samian from the 2011 excavations source is unknown. Examples in a thin red fabric come from contexts (208), (311), and (426). The impression gained from sorting the pottery is that samian ware, and particularly decorated ware, is underrepresented and one might postulate both that the nearby quays were not those over which samian was imported and that higher status residential structures were at some distance from the areas excavated. There is a certain preponderance of later forms also, with more Central Gaulish pieces (114) than South Gaulish (64). Probably also dating from the later irst or early second century is the roughcast ware from the Argonne or nearby areas of north Gaul. This fabric is found in Britain between the Flavian and mid- Antonine periods but, in the Caerleon area, there is a notable absence of Gaulish roughcast ware from levels later than the early second century as, by then, the local Caerleon Ware kilns were in production. The only known Caerleon kilns are on the raised ground at Abernant above Bulmore, but others are likely to have existed. Among their products were roughcast beakers which, although technically less competent than the north Gaulish examples, were produced in large numbers and were, no doubt, a good deal cheaper and thus drove the competition from the market place. The range of pottery produced by these local kilns was probably extensive. Certainly the Caerleon Ware kilns supplied our site with red slipped vessels imitating samian and metal shapes and red slipped mortaria. Other inewares tend to relect the late irst to late second century activity on the site as the Table 9 shows. There are a number of pieces of local green glazed pottery which was probably manufactured at or near Caerleon in the later irst to early second century and might, on the basis of the known inds of wasters, have been made in the western civil settlement. Although there are no pieces comparable to the green glazed wasters from Nash-Williams’ extramural building VII (Boon 1966, Pl.III.5) there are a few pieces of grey ware with vitreous accretions (eg. from 502) which may repay further examination. Substantial fragments of a langed bowl in this fabric came from Trench 1 context (157), which is closely paralleled by other vessels from the western Canabae (Greep in Zienkiewicz 1986b, Fig.38, 3.2-3.3). all pottery samian samian % In Table 10 we have recorded those Caerleon Ware vessels for which a form can be determined. The 97 Caerleon T1 T2 T3 T4 T5 T6 T7 T8 T9 TOTALS 175 68 38.9 51 17 33.3 10 3 30 156 30 19.2 178 31 17.4 40 11 27.5 59 16 27.1 28 6 21.4 41 7 17.1 738 189 25.61 Table 8 Samian as a proportion of the pottery from the 2011 excavations Also probably of local manufacture (and of late irst to early second century date) is mica dusted ware. Although not plentiful (and liable to have lost its surface in the prevalent soil conditions) a small number of examples were noted. Of probably similar date are a couple of vessels in thin off-white eggshell ware from Trench 6 (605). This fabric appears in small quantities throughout the fortress and civil settlement and has obvious afinities with the eggshell ware found at Holt. A common (and at present unknown) source for both is possible. The 2011 site also yielded a number of small thin-walled cups and small bowls which are in a variety of fabrics and almost equal to the white egg-shell ware in thickness. Again, the Ware vessels amount to some 13% of all recovered vessels, a remarkable total given the comparatively short lifespan of the Caerleon Ware kilns (c. A.D. 110-160/70). What is also clear from the table is that Caerleon mortaria are surprisingly scarce on the site. This seems to relect a general dearth of mortaria (with the possible exception of the later white Oxford vessels) and may relect a nondomestic character for the buildings sampled. The second half of the second century saw the appearance of a number of colour coated wares. The 2011 site produced examples of colour coated beakers imported from Lezoux and from the Mosel (Moselkeramik). These were current from the mid second to the mid third century. In addition 83 The Pottery FABRIC Red Eggshell White Eggshell Terra Nigra Pompeiian Red Mica dusted London type Green Glaze N.Gaul Roughcast Lezoux colour coated Mosel colour coated Nene Valley colour coated Oxford colour coated ?New Forest TOTAL T1 2 T2 1 T3 1 T4 1 T5 T6 T7 T8 T9 1 1 2 1 1 2 2 1 1 4 3 1 2 1 1 2 3 17 2 1 9 1 2 7 7 7 1 1 25 1 1 1 3 1 1 4 5 3 3 2 7 TOTALS 5 1 4 1 5 1 7 8 9 13 13 5 1 73 Table 9 Non-samian inewares from the 2011 excavations produced by the Caerleon Ware potters, but need not, of course, be restricted to the narrow c. 110-160 date range of the red slipped ware. there were colour coated beakers from the Nene Valley which are probably later second to third century in date. Later British colour coated wares are, however, scarce. There are no examples of the Nene Valley colour coated versions of kitchen ware bowls and dishes popular in the fourth century (although one possible ‘Castor Box’ may be noted) and very few Oxford colour coated beakers or bowls. A single beaker fragment possibly from the New Forest was present, but, unlike the Oxford ware, this is a ware on the very edge of its range in South Wales and one would not expect larger numbers. A few pieces of ineware of unknown source were also noted. These will require further work and the possibility of vessels brought in as make-weights in cargo delivered at the quays or in the luggage of troops or merchants cannot be excluded. All except mortaria Mortaria T1 10 4 T2 5 0 T3 0 0 A number of contexts in Trench 5 (and one in Trench 1) included substantial fragments of more than one casserole and lid of a type identiied with an African style of cuisine by Vivien Swan. She has identiied a number of vessels of this type on legionary sites and on the Antonine Wall and postulated the presence of African troops recruited into the legions, especially in the Severan period, as a reason for the sudden appearance of these vessels. While this remains a possibility, it must be pointed out that the spread of a particular style of cooking requiring specialised equipment does not necessarily relect the T4 34 4 T5 15 0 T6 7 0 T7 5 2 T8 2 0 T9 9 0 TOTALS 87 10 Table 10 Caerleon ware forms from the 2011 excavations Other Kitchen and Table Wares A series of local sources are apparent. The local potteries which produced the green glazed and mica dusted inewares probably also produced more mundane ware. There are a few examples of white slipped ware (including mortaria but mainly lagons) which are probably local to Caerleon and late irst to early second century in date. The few examples of everted rim jars and langed and carinated bowls of the Flavian to Trajanic period undoubtedly include examples from local greyware and oxidised ware sources. All Caerleon sites yield both Caerleon Ware and oxidised wares which on occasion are simply Caerleon Ware without the slip but which may be coarser and unslipped versions of Caerleon Ware forms or other forms in these coarser fabrics. All are likely to be ethnicity of the cook using those utensils (as many modern users of a wok will attest). One may also point to the use of these distinctive casseroles in other areas of the western Mediterranean seaboard. The case for Africans at Caerleon is, therefore, unproven, but the appearance of a speciic style of cooking using specially (and locally) made casseroles is clear on a number of Caerleon sites both inside and outside the fortress. Other local wares include the common South Wales Reduced (or Grey) Ware found throughout the region. It is less common at Caerleon than elsewhere, presumably because of the presence of local potteries producing oxidised wares, but, nevertheless, in the 2011 trenches, as elsewhere in and around the fortress, we ind examples 84 The Pottery particularly of jars, wide mouthed jars and bowls in this regional fabric. and perhaps of Oxfordshire colour coated ware may also be signiicant. This is not a ‘grand house’ collection. The small number of mortaria present must also have a signiicance and perhaps points to a non-domestic emphasis in the area. We may note a crucible fragment from context (423) and a few waste tiles but this does not seem to be an industrial area either. But the variety, particularly of earlier inewares does not really support a service or storage function. At present it is dificult, on the evidence of the pottery when considered in isolation to arrive at any clear indication of the usage of the area. In part this may be due to the small scale of the excavations when compared with the large area examined. It may, however, be that when the material is linked to the stratiied sequence, clearer trends will be apparent. One suspects, however, that it is the nature of such exploratory excavations to raise more questions than they answer and that more extensive work on speciic buildings will hold the key to unravelling the function of the site (or sites). Among the coarse wares which have been imported onto the site from elsewhere in Britain, the most common is Black-Burnished Ware. This is present in South Wales from the conquest period onwards but becomes the pre-eminent cooking ware from the early to midsecond century to the late fourth century or beyond. Unsurprisingly, therefore, it is present in most contexts on the 2011 site with the possible exception of a few which may be irst century in date. We may also note a single example of Severn Valley Ware (a jar which might presumably have arrived as a container) and several white mortaria from the Oxfordshire kilns. In general mortaria are scarce on the 2011 site, although a stamped vessel in South West White Slipped Ware may be noted. We have already noted a number of pieces of ineware from unidentiied sources which may have appeared with other cargoes or in luggage. To these we should add four pieces from contexts (535), (902), (920) and (939) of Black-Burnished Ware category 2 (BB2), a fabric which rarely appears in the west of Britain. It was made in the South-East and exported up the eastern seaboard. Our pieces could, therefore, have come from anywhere within the normal exporting range of the ware but they are perhaps best seen as having come in the luggage of legionaries returning from construction duties on the Antonine Wall where BB2 is common. Despite the obvious range of the assemblage there are a number of ceramics notable for their scarcity. There are, for instance very few amphorae sherds (just a handful of Dressel 20 oil amphora sherds, a ish sauce amphora and some South Gaulish wine amphora fragments) suggesting that these bulk commodities did not enter the civil settlement through this part of the riverside – a point probably supported by the scarcity of amphorae on the nearby Priory Field site. The scarcity of mortaria has already been noted and may also relect the usage of the area sampled. The presence of a few items of over-ired tile, perhaps wasters (from 425 and 515) is intriguing. SOCIAL & ECONOMIC IMPLICATIONS Comments on the implications of the assemblage must be tempered by the partial nature of the collection – an exploratory sample from a large area, clearly covering several buildings. Even with this in mind, the pottery is noticeably more varied that that from the recent Priory Field excavations and draws on most, though not all, of the sources of pottery available to Roman Caerleon. Both presences and absences may be signiicant. The dearth of amphorae has been noted and this was clearly not an area over which amphorae were moved. Neither is it likely that the buildings on the site used many amphora-born commodities – or if they did then their service and storage areas were not sampled. The dearth of decorated samian 85 Discussion The excavations at Caerleon in 2011 revealed some remarkable archaeological remains that will make a signiicant contribution to our understanding of the legionary fortress of Isca and, in turn, the early history of western Roman Britain. The conquest and paciication of the native tribes of the west, particularly the Silures, play a major role in the sequence of events described by Roman historians and it is possible that the complex of buildings known here as the Southern Canabae would have been familiar to the Roman soldiers and their commanders who took part in the campaigns against these most obdurate of opponents of Roman imperial ambitions in Britain. Medieval and modern occupation seems to be limited to ield boundaries and the occasional pits and post-holes, most of which are likely to be agricultural in nature. Other major Roman centres in Britain invariably were occupied in one form or another in later centuries and in many instances, particularly the two other permanent legionary fortresses at York and Chester, these sites became important urban centres from the Medieval period onwards. In these places the actions of their post-Roman inhabitants caused considerable damage to the underlying Roman archaeology or led to the Roman remains being covered by complex archaeological remains that need to be carefully excavated before the Roman levels are reached. The deinitive discussion of the excavations and their signiicance will be possible only after the inds have been conserved and the respective specialists’ reports have been written. This will allow the integration of the excavated stratigraphy and the inds information, but the purpose of this interim report is to present an outline of the archaeological sequences identiied in the nine evaluation trenches and to explore, at least in part, whether or not we might be in a position to present provisional answers to the research questions that provided the original motivation for the 2011 excavations. In the area of the Southern Canabae, however, this did not happen. For whatever reasons the post-Roman successor to Isca grew up within the walls of the fortress and never expanded into the extramural complex that otherwise might have obscured or disturbed the Roman archaeology we now know is so well preserved in these ields. We are also fortunate that this land has been under pasture for many years and that any ploughing has been sporadic and relatively shallow. No plough marks were noted in the evaluation trenches, although, because the archaeology lies so close to the surface, the movement of tractors and other vehicles across the ields, particularly in wet weather, is causing some damage to the uppermost remains in the western part of the complex. The project’s aims and objectives were grouped into a series of related themes and topics, including: • Layout of the buildings identiied by the gradiometer surveys • Date of the buildings’ abandonment and, potentially, some indication of when they were constructed • Function of these buildings and, therefore, the purpose of the monumental complex • Remains of the suggested quayside wall and landing stages, including their construction and histories • Extent of erosion to the remains caused by the River Usk, and if the river continues to erode this important archaeological resource. Outreach was an important element of the 2011 season and encouraging public participation and community engagement were written into the original project design. A separate summary of the successful engagement work is provided at the end of this report, but interestingly the questions asked by members of the public were often the same as those we set ourselves: What have you found? How old is it? What were the buildings for? The archaeological remains in the nine trenches have provided us with the means to answer these questions and, as is always the case, ask others that could form the basis for future work on the site. Before exploring these in more detail however, it is worth summarising the nature and extent of the surviving archaeology that lies beneath the ields to the south of the amphitheatre at Caerleon. In all trenches it was found that the remains of the Roman buildings survive very well and the uppermost deposits associated with the occupation and subsequent abandonment of the Southern Canabae lie very close to the modern ground surface, in places only just below the turf. The explanation for this extraordinary level of survival is the absence of any signiicant postRoman activity in this part of Caerleon – the evidence for LAYOUT AND ARCHITECTURE OF THE SOUTHERN CANABAE The evaluations conirmed that the Southern Canabae complex consists of at least three major elements, each comprising separate buildings distinguishable by their layout and architecture. The trenches also found that the outlines of the buildings as imaged by the geophysical surveys were generally borne out archeologically, so 87 Discussion and Interpretation that magnetic anomalies coincided with walls and other features such as drains. It is now possible to state with certainty that the major buildings identiied during the surveys are all Roman. Whether these were built as a single episode or were added to over time remains to be conirmed, though the architecture of the main elements suggests either that they were built at different times or that they were contemporary but served different functions. Fortress North This unusual form of tile wall seems to have divided a series of buildings that extended towards the river (areas 1.1 and 1.2), from a parallel roadway running alongside their landward sides. The presence of a stone course capping the eleven tegula courses perhaps indicates that this wall was a stylobate intended to support a colonnaded superstructure of some kind, and the interiors of the buildings were levelled before being provided with rudimentary stone and brick loors. The roadway consisted Zone 3 Compass North Zone 2 Zone 1 Fig. 94 Interpretation of the geophysical results showing the Southern Canabae complex and the 3 ‘zones’ of buildings that comprise it. Zone 1 – the quayside The main part of Trench 1 exposed a cross section of the courtyard building’s southern range closest to the River Usk. The 2010 test-pits suggested that this area might have included the quayside wall which was part of Caerleon’s port, though the 2011 excavations showed that this was not the case and the situation is somewhat more complicated than initially thought. It is now clear that the Usk has moved westwards since the Roman period and that any remains of a port and quay that once existed here have been eroded away by the river. Instead, the gently sloping pre-Roman ground surface shows that the ancient riverbank, although close, must have been further east of the Trench and that the wall constructed from deliberately broken tegulae [115] was not the quay. of a narrow pavement (1.3) which, within Trench 1, was divided by a short wall perhaps to create lean-to buildings or covered entrances, as well as the road itself (1.4) which clearly separated the structures along the riverbank from the southern range of the main courtyard building to the west. The road surface had been intensively used over time and was patched using brick and building stone, as well as complete and broken querns. It is not known if the buildings closest to the river were part of a port, but it is possible that these were warehouses or similar that fronted the quay that has since been eroded away. Zone 1 – the courtyard building: the courtyard The very large courtyard building on the southern side of the Southern Canabae was explored in Trenches 1, 2, 3, 4 and part of 5. The 1 ha courtyard was shown to have been left as open ground, perhaps intermittently surfaced 88 Discussion and Interpretation with cobbles and pebbles most of which have worn away. The two possible buildings identiied on the geophysical surveys proved to be walled structures, although if these were indeed both Roman buildings is debateable. The stone wall and associated collapse in Trench 2 could have been part of a building, though it is not certain that it is Roman in date and it is possible that it was part of a substantial Medieval ield boundary. The structure on the courtyard’s east-west axis, part of which was exposed in Trench 3 is certainly Roman, although like the tegula wall in Trench 1 the method of construction is unusual in a Romano-British context. from brick to stone courses in the centre of this wall probably marks the position of the main entranceway into the courtyard from the port-side road. The southern side of the entranceway appears to have been marked by a transverse wall separating it from the corridor, and in the courtyard this was perhaps embellished by a brick-built base for a plinth or pilaster. The courtyard building: the northern range Trenches 4 and 5 were positioned to investigate what was believed to be the wide northern range of the courtyard building that the geophysical results suggested could have been divided into large rectangular spaces whose narrow sides faced the courtyard. It is dificult however, always to reconcile the geophysical anomalies with the excavated archaeology in this part of the building, primarily because the southern end of Trench 4 contained at least two phases of occupation that are conlated on the gradiometer results. The irst of these consisted of a laid, possibly linear, dry-stone platform whose function is unclear (area 4.2). This was overlain by a well-made cambered cobble and stone surface that might have formed a wide corridor or portico on this side of the courtyard. The adjacent space to the north of this cobbled area does not appear to have been surfaced at all, while a layer of stone and mortar at the northern end of Trench 4 suggests another space with a hard-wearing loor surface (areas 4.3 and 4.4). How these areas were separated is not clear as no remains of dividing walls were identiied, and it is not certain if these spaces were part of a single building or entirely separate buildings. The use of ired bricks, in this case triangular, in structure 3.1 is reminiscent of Roman buildings on the continent particularly, though certainly not exclusively, in Italy. Opus testaceum (also known as opus latericium), was a common building technique from the irst century B.C. and this method of construction, in which courses of bricks were used to face a cement core, remained popular throughout the Roman period especially for imperial buildings in the western provinces (the Constantinian Aula Palatina in Trier is one of the best known examples of a brick-built building from the fourth century). The use of bricks to construct buildings however, is very unusual in Roman Britain where mortared stone was the universally-favoured building technique in forts and fortresses, towns, cities and villa buildings (see below). The wall of ired bricks in Trench 3 explains why this structure produced positive readings on the gradiometer results (imaged as white rather than black), although it is not certain that these walls were ever part of a building in the centre of the courtyard. The absence of foundations and the use of clay rather than mortar or cement to bond the bricks would have meant that the walls were not strong enough to support a substantial superstructure and roof. The absence of loose bricks and roof tiles suggests that these walls were never built to full height and covered. Instead of a building, the structure in Trench 3 perhaps should be seen a base for a platform of some kind, or an enclosure demarcating this space in the wider courtyard. The rear of the courtyard building’s northern range was also exposed in the southern end of Trench 5. A low brick-built opus testaceum wall appears to have separated a narrow corridor (area 5.2), probably colonnaded, from an unsurfaced area to the south. The presence of a drain along the outside of the stylobate wall suggests perhaps that the southernmost area in the trench was an external open space bounded on its northern side by a covered corridor. The lagstone structure in the southern part of Trench 5 appears to be a water trough or tank, perhaps indicating that industrial activities took place in this part of the courtyard building’s northern range, or that animals were kept here (area 5.1). The area of the courtyard seems to have been the location for other, more supericial, structures. The identiication of possible post-pads and a post-hole might have been part of timber buildings or enclosures, while the shallow pit in Trench 2 illed with metalworking debris hints at the nature of the activities that might have taken place here. The geophysical survey suggests that a narrow passageway or corridor passed though the centre of the northern range, providing access from the main courtyard to the buildings of zone 2 to the northwest. The northern range on the other side of this corridor is unfortunately obscured by magnetic noise, though it is possible a similar arrangement of open spaces and coarse surfaced areas lay beyond this passageway too. The courtyard building: the southern range The main courtyard building’s southern, probably front, range seems to have consisted of a 6m wide corridor (area 1.5) that gave access from the riverside road (see above) into the courtyard itself. The corridor and courtyard were separated by another low opus testaceum wall, again possibly for a colonnade that could have extended around three or four sides of the central open area. A change Zone 2 – central buildings and spaces The buildings on the higher ground overlooking the main courtyard building and the Usk loodplain were exposed 89 Discussion and Interpretation in Trench 6 and parts of Trenches 5 and 7. Trench 5 showed that the Roman buildings constructed against the rear wall of the courtyard building’s northern range had been terraced into the sloping ground. It seems likely that a series of long structures would have risen up the low promontory, perhaps with a monumental entranceway of some kind in the centre leading from the port and courtyard. The northern part of Trench 5 contained two adjoining rooms provided with concrete loors and painted plaster (rooms 5.3 and 5.4), east of which was a space with a vaulted ceiling that is in the correct location for an entranceway. A doorway between the two rooms suggests that this corridor building could not be accessed from the courtyard building to the south and that the only way to reach this part of the building was from the north, possibly via an integral corridor running along its rear wall. The vaulted ceiling over the possible entrance (area 5.5) was constructed from tufa and masonry voussoirs, with the frame of the vault perhaps provided by ribs of box lue tiles. There is no suggestion that the vault was decorated in any way, and the presence of such an unusual and complicated architectural feature suggests perhaps that this part of the building was intended to be two storeys or more high. Only a small area of the vaulted space was exposed in the trench and further excavation is needed to conirm whether the collapsed ceiling is indeed the remains of a barrel vault rather than, for instance, an archway. central square courtyard with narrow corridors on three sides and what appears to be a major aisled building on the fourth, northern, side. In plan the central part of this building is reminiscent of a forum, but the absence of rooms beyond the corridors means that it is unlikely to have been built as a marketplace. The courtyard is similar to buildings known as quadriportici, deined as nearly square peristyle courtyards surrounded by colonnaded porticoes. Trenches 7 and 8 examined buildings on the southern and northern sides of courtyard 1 respectively. The two rooms exposed in Trench 7 were at the east end of a row of large rooms that ran alongside the courtyard’s southern portico. These had been raised in height by at least 1m over the surrounding ground surfaces. One of the rooms was provided with an opus signinum loor (room 7.1), while the adjoining corner room was built with an underloor heating system and elaborately painted walls (room 7.2). These rooms appear to have been part of a small bath suite and the geophysics suggests that the praefurnium, where air was heated before passing into the hypocaust, was located in a small adjacent room accessed from the portico. On the other side of the southern zone 3 courtyard, Trench 8 found that the two wide corridors running along the courtyard’s north side were not provided with internal loors as anticipated, but instead seem to have consisted of beaten earth surfaces. The remains of a large aisled building were found beyond these open corridors and a step-like structure against the southern external wall shows that this building could be accessed from the direction of the central courtyard. A drain built into the wall terminated with an outlet on the outside face of the step from which water must have poured into the adjacent open corridor. The discovery of part of a tessellated pavement at the northern end of Trench 5 demonstrates that at least one room on the crest of the higher ground was provided with a mosaic loor (area 5.6), though only the edge was exposed and it is not known how well the rest survives or if it was decorated. The northern part of zone 2 was explored in Trench 6, where a well preserved lagstone surface may well indicate a small open courtyard between neighbouring buildings (area 6.1). The geophysical results suggest that a narrow rectangular building with a coarse crushed tile loor lay to the northwest of this possible courtyard, and the discovery of metalworking suggests that industrial activities took place here. The discovery of a lead waterpipe shows that water from a tank of some kind located in this building was fed downslope to fountains or suchlike further east. The southern end of zone 2 was investigated in the eastern end of Trench 7, which showed that, in this area at least, an external cobbled area lay between the buildings of zones 2 and 3 (area 7.3). This might have been a road or alleyway, though the geophysical results suggest that these surfaces could have been part of a more extensive open space. The aisled building seems to have consisted of a series of parallel rooms and internal corridors covered either with lagstone or opus signinum loors (rooms 8.28.4). Although some walls were rendered there was no evidence that any of the rooms had been more than white-washed. It is possible that this building was part of a complex that included the bath-house excavated in the 1920s next to the amphitheatre (known as Bath A). The northern courtyard complexes: courtyard 2 The northernmost courtyard building was examined in Trench 9. Although the geophysical results are less clear here, it is likely that it composed a second quadriporticus with an adjoining building on its northeastern side. The full width of this building was exposed in the trench, which showed that it was long and narrow and, like the aisled building in Trench 8, had been raised in height above the surrounding ground surfaces. The interior was divided into two parallel spaces, probably a corridor and a row of room with opus signinum loors and painted walls (rooms 9.2 and 9.3). The absence of any evidence for a step to overcome the difference in height between Zone 3 – the northern courtyard complexes: courtyard 1 Trenches 7, 8 and 9 examined the two courtyards and associated buildings in the north of the Southern Canabae. The southernmost of these consisted of a 90 Discussion and Interpretation the courtyard to the south and the interior of the building suggests that this building was accessed from its narrow ends, and possibly was part of the large courtyard structure excavated in the 1950s known as Building D (also known as Building IX), that included a monumental entranceway and hypocausted rooms. tessellated, at least in its latest phase. Most trenches produced evidence for internal decoration in the zones 2 and 3 buildings, ranging from fairly rough whitewashed render to elaborate wall paintings in rooms 7.2 and 9.3. The discovery in Trench 8 of segmented tiles of two different sizes indicates that a colonnade, possibly around the outside of the southernmost courtyard in zone 3 was constructed from brick-built columns. Segmented tiles are known from only a few excavations in Caerleon (Zienkiewicz 1993, 126-7) and a handful of other sites in Roman Britain, notably the temple to Claudius at Colchester where brick built columns may have formed the main portico of the temple when it was rebuilt after the Boudiccan destruction, if not before (Hull 1953; Crummy 1980; Drury 1984, 41). The northern part of Trench 9 may well have exposed part of the central courtyard of Building D, which again was much lower than the internal loors of the corridor building (area 9.4). If this turns out to be the case, perhaps two major bath complexes seem to have been located on the northern side of the Southern Canabae closest to the amphitheatre and the road leading into the fortress’ west gate. Building methods The Southern Canabae can be divided into two parts that seem to have been built using different methods of construction. The remains of the courtyard building in zone 1 are more supericial than the other buildings in the suburb and many of the walls located in Trenches 1, 2, 3, 4 and 5 were built without foundations and bonded with earth or clay rather than mortar. Furthermore, the presence of at least three CBM-built walls in the courtyard building is very unusual in Caerleon. Opus testaceum, the use of bricks or tiles to face a wall’s concrete core, is more common on the continent than Roman Britain where mortared stone was the favoured building technique (Ward-Perkins 1981, 21-121; Adams 1994, 145-51). Many buildings constructed entirely of brick and tile can be found in Rome, Ostia and Pompeii, as well as elsewhere on the Roman continent, although the CBM walls at Caerleon are not strictly opus testaceum as their faces are clay bonded and clay or earth was also used to pack the walls’ cores. This suggests that these were never intended to be load-bearing walls and that they would have supported arcades or colonnades instead of solid superstructures, although why bricks and tiles were used is unclear. The internal rooms and spaces of the zone 1 courtyard building were provided with hard-wearing loor surfaces made of cobbles, pebbles, beaten earth and scatters of crushed tile. These give a functional impression to these areas that is supported by the general absence of wall plaster from the courtyard building where the walls of rooms must have been entirely unembellished. Layout of the Southern Canabae The suburb of monumental buildings between the amphitheatre and the River Usk appears to have consisted of at least 3 separate elements, all of which were built on a different orientation to the fortress. The very large courtyard building in zone 1 lies next to the port facilities on the Usk, only part of which survives, and comprises an extensive central courtyard surrounded by corridors and ranges of rooms. The building measures at least 140m from east to west and 120m from north to south and is one of the largest structures known from Roman Britain. It is not certain if similar buildings existed in the canabae of the other permanent British fortresses at Chester and York (for example Mason 1987, 149-51), though there are several instances of large buildings located outside the walls of legionary fortresses on the continent comparable with the Caerleon example, notably at Carnuntum, Nijmegen, Mirebeau and Vindonissa. It is perhaps interesting that, like at Caerleon, the courtyard buildings at Mirebeau and Vindonissa were located in close proximity to amphitheatres, while at Carnuntum and Nijmegen the courtyard structures are on the opposite sides of the fortresses to their amphitheatres (van Enckevort 2002; Goguey and Reddé 1995; Hartmann 1986; Jobst 1983, 98-100.). The building complexes in zone 3 of the Southern Canabae would appear to consist of at least two perisytle courtyards and associated buildings. In plan these are reminiscent of quadriportici, a form of courtyard building that was common in Roman cities of the late republican and early imperial periods. Quadriportici from Rome, Pompeii and Hadrian’s villa at Tivoli are among the best known examples from Italy (Berry 2007, 143-7; Coarelli 2002, 178-81; McDonald and Pinto 1995, 95-99; WardPerkins 1981). In the urban examples the quadriportici were located next to theatres (Pompey’s theatre in Rome and the Large Theatre in Regio VIII at Pompeii), or amphitheatres (the 141m by 107m palaestra in Regio II at Pompeii), while at Tivoli the courtyard known as the Water Court or Piazza d’Oro served as an ornamental garden next to a grand nymphaeum with six fountains. The buildings in zones and 2 and 3 were generally constructed using mortared masonry walls and they were more solidly built that the zone 1 courtyard building. In the northern zone 3 the buildings appear to have been long narrow corridor-like structures ranged around open peristyle yards, while in the central zone 2 the paved area discovered in the southern end of Trench 6 could have been an open yard between surrounding buildings. The loors in these parts of the Southern Canabae were usually opus signinum, although room 7.2 was provided with a hypocaust and room 5.6 must have been 91 Pompey’s quadriporticus in Rome also performed a decorative function and the courtyard was illed with a formal garden, including statues and fountains. Although the Pompeian quadriporticus in Regio VII is best known as a gladiator school and barracks, the gladiators moved here only after the earthquake of A.D. 62 and it is not certain what function the building originally was intended to fulil. The presence of a long pool in the centre of the Pompeian palaestra reinforces the importance of water in these buildings, something that is worth bearing in mind when considering the drain and outlet in the irst courtyard of zone 3 excavated in Trench 8. DATING OF THE SOUTHERN CANABAE: CONSTRUCTION, ABANDONMENT AND POSTROMAN ACTIVITY The inal statement regarding the dating of the Southern Canabae must wait for the identiication and analysis of the artefacts recovered from the excavated deposits, but it is already possible to put forward some ideas at this stage of the post-excavation work regarding the construction and abandonment of this important new addition to Roman Caerleon. It is clear from the material produced during the excavation that the Southern Canabae was built very early in the Roman period and there is no reason to suggest that the suburb was later than the construction of the fortress in the mid 70s. It is unclear at this stage if the various different buildings were all built at the same time, though the similar proportions of late irst and early second-century pottery identiied from all nine trenches suggests that they could have been broadly contemporary (in which case the use of different building techniques in zone 1 compared to zones 2 and 3 should be related to these buildings’ functions rather than their dating). Clarifying the nature of the relationship between the amphitheatre and the Southern Canabae is essential if we are to understand the early development of the fortress’ civil settlement. The construction of the amphitheatre c. 90 (or possibly its reconstruction c. 140) appears to have led to the partial demolition of at least one bath-house on its western side. It is possible that this building and the adjacent monumental building D were part of the Southern Canabae buildings discovered by the geophysical surveys and evaluated in 2011. If this turns out to be the case, it demonstrates that the Southern Canabae was certainly in existence before the reconstruction of the amphitheatre, and it could pre-date the amphitheatre’s construction in 90. It would seem likely that Isca would have been provided with a port as soon as it was established in 74/75, from which the legion could be supplied from the sea and also send provisions to the auxiliary forts upstream from Caerleon. From the pottery evidence it appears that much of the Southern Canabae was no longer in use from possibly as early as the late second century. While most trenches produced some third and fourth century pottery, this was less prevalent than in most excavated sites at Caerleon and suggests that several of the buildings in the suburb had been abandoned by the beginning of the third century at the latest. It is clear, however, that the buildings must have remained standing for some time and several rooms and spaces were later used to discard rubbish. Trenches 5, 7 and 9 produced signiicant quantities of pottery from the third to mid-fourth centuries and this material from various rubbish deposits dumped into rooms and against walls outside buildings is evidence of continued occupation nearby in the period after the Southern Canabae was no longer in use. The preliminary inspection of the small coin assemblage from the 2011 excavations did not identify any fourth century coins and only a handful of late-third century issues. If this picture is borne out in the inal analysis of all the inds, the Southern Canabae would appear to have been abandoned perhaps 75-100 years before the legion was withdrawn from Caerleon in c. 300, after which some parts were used for the deliberate disposal of rubbish by occupants living and working in the vicinity. Some of the buildings of the Southern Canabae were deliberately demolished and most appear to have been robbed of their building stone before those inside the fortress. Only the buildings in Trenches 5 and 8 seem to have collapsed leaving much of their roofs in situ, whereas Trenches 1, 6, 7 and 8 produced far smaller quantities of tile indicating that the roofs of these buildings had been removed prior to the robbing of their walls. The robber trenches in all of the 2011 trenches did not contain the later Medieval and modern material that dates much of the robbing within the fortress and it would appear, therefore, that the Southern Canabae was reduced before the fortress, almost certainly in large part between the late Roman period and the reappearance of pottery in the twelfth century. Post-Roman occupation in this part of Caerleon is limited. Possible structures of reused Roman roof tiles were identiied in Trenches 5 and 8 (though these could be Roman), while the upper courses of wall [505] in Trench 5 could be evidence of a standing wall reused at some point after the Roman period. Otherwise, the few signs of post-Roman activity in the area of the Southern Canabae consist of medieval ield boundaries in Trenches 1 and 2, all of which were in use until relatively recently. FUNCTION, STATUS AND PURPOSE OF THE SOUTHERN CANABAE The 2011 excavations have conirmed that the buildings of the Southern Canabae date to the early Roman period in Britain, while the layout of the monumental surburb suggests its construction was part of a wider plan to develop the fortress at Caerleon in the years during the conquest and paciication of the Silures. The Discussion and Interpretation The northern part of the Southern Canabae (zone 3) consisted of extensive unpaved open areas with corridors and long narrow buildings around their sides. The two largest courtyards had large buildings on their eastern sides that seem to have been divided into small rooms accessed from long corridors. These were provided with robust opus signinum or lagstone loors and some of these rooms’ walls were painted. A small decorated bathsuite on the south-western side of one of these courtyards had been raised by a considerable height above the surrounding land and it is tempting to think that this was done in order to improve the view towards the River Usk. It is also possible that the structures in zone 3 continued to the northeast and were connected to the buildings excavated close to the amphitheatre (Baths A and H, and Building D or IX). The extent of these buildings to the northwest however, is uncertain and it is also possible that the bath-houses (if these were indeed separate) and monumental building D were not part of the zone 3 complexes, though this will only be resolved by further investigation. For the time being it seems likely that the buildings in the northernmost part of the Southern Canabae were not domestic spaces, but could have been used as ofices, shops or stalls, or that somewhere in this area lay a mansio, perhaps with a central garden, to serve the needs of the many oficial visitors to Isca (Black 1995, 20-1; Evans 2000, 496). scale of the buildings with their opus signinum loors and simply decorated walls are reminiscent of Roman ‘public’ military and urban architecture, and there seems little reason to doubt that this part of Isca’s extramural settlement was a state-sponsored foundation - the canabae legionis. The excavations discovered tantalising evidence for the activities that took place in the Southern Canabae and, although it should be remembered that the nine evaluation trenches between them revealed less than 1% of the total area of the monumental building complex, it is possible to consider what took place inside these buildings. Large rectangular or square courtyard buildings outside legionary fortresses in the Roman Empire have been interpreted as fora, macella, mansiones, and a waggon yard (van Enckevort 2002; Goguey and Reddé 1995, 26-9; Jobst 1983, 98-100; von Petrikovits 1981, 170), and from its ground plan alone any of these functions are possible for the building complex in zone 1 of the Southern Canabae. Only one of the known examples, however, has been excavated and at Nijmegen 50,000 posts were found covering the central 1 ha courtyard, which have been interpreted as supports for a raised wooden loor or fences that would have been needed to corral large numbers of animals, especially cattle (van Enckevort 2002, 388-9; Willems and van Enckevort 2009, 61-4). The Caerleon building was not well built and it is unlikely that many of its low walls bonded with earth and clay would have been able to support a substantial superstructure or a second storey. It is not certain if the brick-built walls of the structure towards the rear of the open courtyard supported a covered building or were the base of a platform or enclosure, but its positioning on the courtyard’s main axis was deliberate. The courtyard itself was left open and there is no evidence that it had been paved, as might be expected if the building served as a marketplace or civic area. It is however, one of the few open spaces within the canabae large enough to accommodate the thousands of men in the Second Augustan Legion and it is also possible that the courtyard and surrounding buildings were a gathering place for people, as well as animals and equipment, arriving from the river port before heading off to other parts of the canabae or the fortress. At Caerleon we are now in the fortunate position of being able to draw on the evidence of the excavated inds as well as building plans when thinking about the functions performed in the Southern Canabae. Final analysis will take place in due course, but initial observations of the various categories of artefacts have identiied some interesting patterns that have an important contribution to make to this discussion. In fact, different parts of the site have produced very different inds assemblages. Pottery and other inds are scarce from the area of the courtyard of the zone 1 building, the port area in Trench 1 produced many metal objects as well as metalworking debris, while pottery and animal bones were common from Trenches 5 and 7 yet less frequent in Trenches 8 and 9. Some of this material will be derived from deposits overlying the last Roman loors and, therefore, belong to a late period of activity when parts of the Southern Canabae were used to dispose of rubbish, though other artefacts will have been incorporated in deposits associated with the buildings’ construction and use. The area of the Southern Canabae on the higher ground overlooking zone 1 produced very complex geophysical results and it is dificult to untangle the buildings’ remains in this part of the surburb. The excavations indicate however, that a long narrow building divided into a series of corridors and small rooms sat on the slope above zone 1, perhaps with a monumental entranceway allowing access through it. Other buildings nearby included a room with a tessellated loor and a possible metalworking area from which water was fed through a lead pipe to water features of some kind in the courtyard. Peter Webster and Mark Lewis have noted the absence of amphorae and mortaria from across the Southern Canabae where decorated tableware is also unexpectedly uncommon. Roman legionaries are well known for consuming large quantities of beef and pork, but Adrienne Powell’s animal bone report highlights the unusually numerous remains of pigs and birds from particular trenches (see King 1999 and King 2005). For the time being it is clear that the pottery and animal 93 Discussion and Interpretation bone assemblages from the Southern Canabae are both distinctively different from the material produced elsewhere within the fortress or outside. The age at which the animals were slaughtered will be one of many themes to be pursued when selected material is examined for the inal report, as will the types and sources of the pottery recovered during the excavations. Whatever the functions performed in the various buildings it is evident that, together with the amphitheatre, the Southern Canabae formed the centre of the oficial settlement around the legionary fortress – the best candidate for a canabae legionis in Britain. Isca, however, did not develop into a major city and, unlike other fortresses, the domestic part of the ‘town’ seems to have remained relatively small and unsophisticated. This apparently stunted development suggests that the civil administration of south-east Wales in the Roman period was located elsewhere, possibly in the civitas capital of Venta Silurum at Caerwent some 9 miles away. The Southern Canabae seems to have fallen into disuse and abandonment during the early third century, clearly demonstrating that the legion at Caerleon was not responsible for administering the canton of the Silures by this time, or that the structures of Roman imperial authority in this part of Britannia needed military protection. These and other questions will be the focus of the remaining post-excavation work and we look forward to presenting the inal analysis of this remarkable site in the full excavation report. 94 Bibliography Adams, J.-P. 1994. Roman Building: Materials and Techniques. London: Routledge. Berry, J. 2007. The Complete Pompeii. London: Thames & Hudson. Black, E.W. 1995. Cursus Publicus. The infrastructure of government in Roman Britain. Oxford: BAR Brit. Ser. 241. Boon, G.C. 1972. Isca. The Roman Legionary Fortress at Caerleon, Mon. Cardiff: National Museum of Wales. Boon, G.C. 1987. The Legionary Fortress of Caerleon-Isca. Cardiff: National Museum of Wales. Brewer, R.J. (ed.) 2002. Birthday of the eagle: the Second Augustan Legion and the Roman military machine. Cardiff: National Museums & Galleries of Wales. Burnham, B. and Davies, J. (eds) 2010. Roman Frontiers in Wales and the Marches. RCAHMW, Aberystwyth. Chapman, E, 2011. ‘Roman Britain in 2010: I. Sites Explored’, Britannia 42, 320-28. Coarelli, F. (ed.), 2002. Pompeii. New York: Riverside Book Company. Crummy, P. 1980. ‘The temples of Roman Colchester’, in: W. Rodwell (ed.), Temples, churches and religion: recent research in Roman Britain. Oxford: BAR Brit. Ser. 77, 243-83. Drury, P.J. 1984. The Temple of Claudius at Colchester Reconsidered’, Britannia 15, 7-50. van Enckevort, H. 2002. ‘The eastern canabae legionis of the Legio X Gemina on the Kops Plateau in Nijmegen’, in: P. Freeman, J. Bennett, Z.T. Fiema and B. Hoffmann (eds), Limes XVIII. Proceedings of the XVIIIth International Congress of Roman Frontier Studies held in Amman, Jordan (September 2000). Oxford, BAR Int. Ser. 1084, 387-93. Evans, E. 2000. The Caerleon Canabae: excavations in the civil settlement 1984-90. London: Britannia Monograph 16. Evans, E. 2004. The Roman fortress at Caerleon and its environs: a framework for research. Glamorgan-Gwent Archaeological Trust Report 79. Evans, E. 2010. ‘Caerleon’, in: B.C, Burnham and J.L. Davies (eds) Roman frontiers in Wales and the Marches. Aberystwyth: RCAHMW, 161-172. Evans, D R & Metcalf, V M 1992. The ‘Roman Gates’ site in the fortress of the Second Augustan Legion at Caerleon, Gwent: The excavations of the Roman buildings and evidence for Early Medieval activity, Oxford: Oxbow Monogr. 15 Gardner, A. 2007. An Archaeology of Identity: soldiers and society in late Roman Britain. Walnut Creek: Left Coast Press Gardner, A. and Guest, P., 2007. Caerleon: Excavations at Golledge’s Field and Priory Field, 2007: Interim report. Unpublished report, Cardiff University / University College London. Gardner, A. and Guest, P., 2010. Caerleon: Excavations in Priory Field and the Southern Canabae, 2010: Interim report. Unpublished report, Cardiff University / University College London. Goguey, R. and Reddé, M. 1995 Le Camp Légionnaire de Mirebeau. Mainz: Römisch-Germanischen Zentralmuseum. Guest, P. and Gardner, A., 2008. Caerleon: Excavations in Priory Field, 2008: Interim report. Unpublished report, Cardiff University / University College London. Guest, P and Young, T. 2007. ‘Mapping Isca: geophysical investigation of Priory Field, Caerleon’, Archaeologia Cambrensis 155 (2006), 117-33. Guest, P and Young, T. 2010. ‘Mapping Isca: geophysical investigation of School Field and Priory Field, Caerleon’, Archaeologia Cambrensis 158 (2009), 97-111. Bibliography Hamilton-Dyer, S. 1993. ‘The animal bone’, in: J.D. Zienkiewicz, ‘Excavations in the Scamnum Tribunarum at Caerleon: The Legionary Museum Site 1983-5’, Britannia 24, 132-136. Hartmann, M. 1986. Vindonissa: Oppidum-Legionslager-Castrum. Windisch. Hull, M.R. 1958. Roman Colchester. Oxford. Jobst, W. 1983. Provinzhauptstadt Carnuntum. Vienna: Österreichischer Bundersverlag. Jones, G.V. 2001. ‘John Edward Lee and Antiquarianism in Nineteenth-Century Caerleon’, The Monmouthshire Antiquary XVII, 3-8. King, A. 1995. 'Diet in the Roman world: a regional inter-site comparison of the mammal bones', Journal of Roman Archaeology 12, 168-202 King, A. 2005. ‘Animal Remains from Temples in Roman Britain’, Britannia 36, 329-69. Knight, J. 2001. ‘City of Arthur, City of the Legions: Antiquaries and Writers at Caerleon’, The Monmouthshire Antiquary XVII, 46-54. Manning, W.H. 2004. ‘The Roman conquest and the army’, in: M. Aldhouse-Green and R. Howell (eds), The Gwent County History 1: Gwent in prehistory and early history. Cardiff: University of Wales Press, 178-204. Mason, D. 1987. ‘Chester: the Canabae Legionis’, Britannia 18, 143-68. McDonald, W. and Pinto, J.A. 1996. Hadrian’s Villa and Its Legacy. New Haven: Yale University Press. Noddle, B.A. 1983. ‘The animal bones’, in: P.J. Casey, ‘Caerwent (Venta Silurum): the excavation of the northwest corner tower and an analysis of the structural sequence of the defences’, Archaeologia Cambrensis 132, 63-67. O’Connor, T.P. 1986. ‘The animal bones’, in: J.D. Zienkiewicz, The Legionary Fortress Baths at Caerleon. Volume II: The Finds. Cardiff: National Museum of Wales, 225-248. von Petrikovits, H. 1981. ‘Die Canabae Legionis’, in: 150 Jahre Deutsches Archäologisches Institut 1829-1979. Mainz, 163-75. Reddé, M., Brulet, R., Fellmann, R., Haalebos, J.K., and von Schurbein, S. 2006. L’architecture de la Gaule romaine. Les fortiications militaries. Paris: Maison des Sciences de l'Homme. Young, T. 2012. Geophysical surveys in Caerleon, Newport, 2007-2011. Unpublished GeoArch Report 2012/05. Ward-Perkins, J.B. 1981. Roman Imperial Architecture. 2nd edition. New Haven: Yale University Press. Willems, W. and van Enckevort, H. 2009. Ulpia Noviomagus - Roman Nijmegen: The Batavian capital at the imperial frontier. Rhode Island: JRA Supplement Series 73. Zienkiewicz, J. D. 1986a. The Legionary Fortress Baths at Caerleon. Volume I: The Buildings. Cardiff: National Museum of Wales. Zienkiewicz, J. D. 1986b. The Legionary Fortress Baths at Caerleon. Volume II: The Finds. Cardiff: National Museum of Wales. Zienkiewicz, J. D. 1993. ‘Excavations in the Scamnum Tribunarum at Caerleon: the Legionary Museum Site 1983-5’, Britannia 24, 27-140. 96 Trench Matrices Appendix 1 Trench Matrices Trench 1 Stratigraphic Matrix 7. Topsoil 101 105 106 6. Alluvium and Medieval Boundary 112 103 109 170 110 104 178 5. Disuse and Robbing 113 114 117 142 165 122 164 181 167 131 141 123 116 118 133 125 135 130 145 146 153 191 151 150 152 162 154 159 156 157 169 161 121 176 180 184 185 179 192 189 188 172 174 183 173 187 143 144 158 155 1. Natural 163 149 160 2. Walls 124 134 129 3. Levelling 126 132 136 137 138 139 140 182 166 119 4. Surfaces and Patching 102 111 108 107 177 186 115 148 175 190 147 193 195 198 194 97 128 168 171 120 127 Trench Matrices Trench 2 Stratigraphic Matrix 7. Topsoil 201 6. Post-Medieval Boundaries 202 211 203 212 204 205 5. Second Phase Surfaces/ Alluvium 209 206 207 4. Post-Courtyard Activity 218 224 217 219 3. Courtyard Surfaces 2. Clearance and Levelling 225 214 223 232 231 98 216 229 228 227 226 230 221 1. Natural 208 213 210 222 215 Trench Matrices Trench 3 Stratigraphic Matrix 5. Topsoil 301 4. Alluvium / Abandonment 306 305 311 312 307 3. Structure 303 302 304 2. Rubble Spread and Possible Post-pads 314 1. Natural and Subsoil / Surfaces 320 322 321 323 310 315 316 317 318 319 313 309 99 308 Trench Matrices Trench 4 Stratigraphic Matrix 401 9. Topsoil 8. Robbing 411 409 440 410 408 7. Decay / Collapse 436 6. Surfaces 437 403 417 5. Accumulation / Levelling 413 4. Surfaces 3. Levelling / Construction 2. Earliest Surface & Stone Structure 1. Natural 406 405 433 415 438 434 439 100 414 425 420 423 429 421 419 412 418 435 407 430 416 428 424 426 422 432 427 431 Trench Matrices Trench 5 Stratigraphic Matrix 501 5. Modern Pit and Topsoil 502 521 506 520 504 503 515 4. Robbing and Accumulation 519 512 565 510 567 511 536 561 516 508 507 522 525 514 509 526 527 524 3. Disuse / Collapse 537 542 566 513 559 534 538 531 535 545 539 558 546 554 517 544 550 540 529 2. Levelling & Floors 518 551 543 548 528 564 549 541 547 552 553 563 1. Walls 523 530 101 533 505 555 556 557 560 562 Trench Matrices Trench 6 Stratigraphic Matrix 601 602 5. Topsoil 603 4. Robbing 641 647 640 619 611 613 3. Decay and Demolition 609 607 615 610 608 614 605 616 2. Floors/Surfaces and Drain 650 631 612 623 624 625 604 626 648 617 635 633 636 629 627 628 634 1. Walls 622 637 102 Trench Matrices Trench 7 Stratigraphic Matrix 5. Topsoil 701 704 705 4. Robbing 702 706 708 711 720 3. Decay and Demolition 703 710 728 726 717 721 733 731 713 707 735 712 709 2. Floors/Surfaces and levelling 718 736 714 729 738 730 723 725 737 715 716 1. Walls 722 719 103 732 734 740 739 Trench Matrices Trench 8 Stratigraphic Matrix 801 5. Topsoil 4. Robbing 841 843 820 837 803 807 810 809 811 814 806 804 802 808 822 3. Decay and Demolition 824 827 840 853 858 817 816 818 836 819 839 828 830 831 833 834 823 2. Floors/Surfaces 842 813 832 835 859 846 845 857 854 847 851 855 1. Walls 821 838 812 805 104 815 844 856 825 829 826 Trench Matrices Trench 9 Stratigraphic Matrix 901 9.6 Topsoil 915 903 921 919 9.5 Decay and Robbing 922 923 902 904 920 935 909 961 905 906 913 928 953 962 952 912 929 959 9.4 Second Surface 930 940 924 907 937 941 910 9.3 First Surface 930 964 944 956 955 945 946 943 925 950 940 914 926 936 938 951 9.2 Levelling and Construction 947 963 949 9.1 Walls 960 958 932 105 917 931 918 Appendix 2 Public Participation & Community Engagement An important element of the Southern Canabae project was the aspiration to bring the results of the excavation to as wide an audience as possible, and to allow members of the public the opportunity to participate in the process of archaeological ieldwork and discovery. The outreach strategy for 2011 included providing places for volunteers on site, running twice-daily tours for visitors, holding special Open Days during the Summer Bank Holiday weekend (27-29 August), and using the internet and social media to communicate with people across the world.1 Several TV companies ilmed the excavation in 2011 and the work at Caerleon featured in two of the BBC’s new series - The Story of Wales with Huw Edwards and The Great British Story: A People’s History with Michael Wood. The excavation was also ilmed for the Channel 4’s Time Team, which brought the Southern Canabae project to an estimated 1.5 million viewers. Under the title ‘The Lost City of the Legion’, the 2011 excavations were a hugely successful public engagement event attracting thousands of visitors to the site and generating interest in archaeological research at this internationally important site from across the world. Volunteering at Caerleon A total of 61 volunteers (approximately 20 per day) joined the team to take part in the excavations. This was very successful and feedback was extremely positive. The volunteers were from a diverse range of ages and backgrounds, from school pupils interested in careers in archaeology, local history enthusiasts, experienced amateurs, metal detectorists, local shopkeepers, teachers, and retired nurses. The volunteers included groups from sixth form colleges in Colchester and Cirencester who each camped with the team for a week. The volunteer scheme was heavily over-subscribed and enquiries about volunteering continued until the last day of the dig. Site Visits and Open Days Through tours and a variety of family-centred interactive activities with tangible learning outcomes, the excavators sought to increase public awareness of the archaeological ieldwork that was being undertaken. Visitors during the Open Days were welcomed by students dressed in authentic Roman costumes and were given guided tours of the site, thereby experiencing an archaeological excavation at irst hand. Members of the public were encouraged to examine the many types of inds recovered Volunteers 2011* 61 2010** 41 2008** 36 Total 138 Visitors total Open Days 6,225 3,292 3,904 1,643 3,090 934 13,219 8,969 Dig blog hits Facebook ‘Interactions’ Twitter followers 19,315 41,102 471 24,188 104 13,103 - 37,291 41,102 585 Table 11 Summary of outreach activities at Caerleon 2011 * 4 weeks ** 6 weeks 1 1. Dr Paula Jones co-ordinated the volunteer programme and the schedule of engagement events on site. Cardiff University’s Community Engagement Team helped organise the engagement strategy and also arranged a series of family centred activities for the Open Days. Staff from the National Roman Legion Museum provided signiicant resources and support for the Open Days including Roman costumes, small suits of Roman armour for children and a range of literature for display. 107 Community Engagement Fig. 95. Volunteers young & younger joyfully help uncover the archaeology 108 Community Engagement Fig. 96. The hustle and bustle of the busy open days 109 Community Engagement during the excavation, and they could take part in various archaeological and fun activities devised and arranged by the students and volunteers. These included • make your own Roman pot workshop • Finds handling activity – learning about pottery and animal bone • authentic Roman food - cookery display (with Edith Evans of the Gwent Glamorgan Archaeological Trust) • childrens’ Roman dig • a ‘gallery’ of colouring sheets & colouring competition for younger children. • trying on Roman armour for younger children visit historic sites. In 2011 we welcomed special group visits by the charities Fairbridge and Scope. In terms of attendance the excavation was a huge success with a 6,225 people visiting during the 4 week season in 2011 - an average of 1,556 people per week. Over 3,200 visitors came to the site during the bank holiday weekend - 1,611 visitors attended on the Monday alone. Scope work with adults with a range of learning dificulties and disabilities, providing care, support, and enjoyable activities. A group from Scope visited the excavations on two occasions, helping out in the trenches as well as cleaning inds. All of the individuals involved are now researching the Roman period in their spare time and have become committed voluntary diggers. Almost 1,400 copies of the commemorative Lost City of the Legion booklet were taken away during the excavation season by members of the public as souvenirs of their visits. Special Visits The Southern Canabae project aimed to engage all sections of society with archaeological research, including people from disadvantaged backgrounds, those with disabilities, and ethnic minorities who tend not to Fairbridge work with ‘disengaged’ young people (between 13-25 years old) and help them to gain the motivation, self-conidence and skills they need to change their lives. Two young people accompanied by Fairbridge staff came to site and spent the day working in the trenches alongside the archaeologists and also having an extended tour of the site and lunch with the team. It was clear that the experience was enjoyed by all, but particularly inspiring to one of the participants who would now like to go on to pursue a course in archaeology in the future. Dig Blog and the Internet The internet allowed the Southern Canabae project to build on the successes of previous seasons and to communicate the results of the archaeological excavations to as wide an audience as possible. The project generated interest from across the world and the team used a variety of means to satisfy the demand for updates about how the excavation was getting on and news of the latest inds as they came Fig. 97. Reconstruction of Isca and the Southern canabae (©7reasons) 110 Community Engagement out of the ground. The 2011 internet resources included: • Project pages on Cardiff University’s website • Dig Blog • Animated digital reconstruction of the legionary fortress • Facebook page • Twitter feed • Youtube channel for dig videos The 2011 Dig Blog, hosted on the Council for British Archaeology’s ‘Community Archaeology Forum’, was updated daily with illustrated archaeological and social stories written by students and volunteers as well as staff. In total the blog received over 19,000 hits during the 4 weeks of the excavation (an increase of some 20% over the previous year). The new animated digital reconstruction of the legionary fortress, including a 90 second ly-thru over Roman buildings and roads, was released to coincide with a new webpage describing some of the spectacular discoveries made in the irst 3 weeks of the excavation. This received extensive coverage and the animation was featured in newspapers, television, radio and online in the UK and in many countries all over the world The Facebook page was new in 2011 while the Twitter account built on the success of the irst trial in 2010. Both were very popular and members of the excavation team updated both sites throughout the day with the latest news from the trenches. The Facebook site received over 40,000 interactions during the month-long season, and 471 people followed the dig tweets. Also new this year was the Youtube channel where 3 videos were uploaded. These described the excavation, showed a hypocaust underloor heating system being dug, and an interview with 2 retired volunteers. The success of the engagement strategy at Caerleon in 2011 is evidenced by the constructive feedback from volunteers and the enormous popularity of the project in terms of visits and positive evaluation. The South ern Canabae excavation shows how archaeology has the power to provoke and inspire people from all over the world to get involved with archaeological ieldwork and cutting-edge university research in the UK. 111 Appendix 3 Excavation Team Ninety people in total worked on the excavations at Caerleon in 2011. Many of the project’s core staff have worked together on previous excavations in Caerleon and elsewhere, though this year we were pleased to welcome Paula Jones, Becky Smith, Meg Tudor and Scott Williams as new members of the team. The excavators were either undergraduate archaeologists from Cardiff University or volunteers in various guises. The students and volunteers did not just dig they also gave tours to the hundreds of visitors to the site, organised and set up various activities during the open day weekend, and coped admirably with the presence of numerous tv crews throughout the season. The small army of volunteers were of all ages and backgrounds, and came to Caerleon from all parts of the UK and beyond – some for a day here-and-there, while others camped with us for several weeks. Many volunteers had worked on the Priory Field excavations in 2008 and 2010, and several former students from these seasons returned during their holidays in 2011 (James Goodsell, for instance, is a veteran of Priory Field who was persuaded to help supervise Trench 2 during his vacation in Caerleon). Time Team spent three days with the project at the end of our third week and we would like to thank Phil Harding, Raksha Dave, Matt Williams and Alex Langlands for their hard work in the trenches. The team camped in Priory Field and we are indebted to Cadw for allowing us to use the ield again. Many people provided the excavation with assistance during our time in Caerleon, but we are particularly grateful to Miguel Santiago of the Priory Hotel for the loan of his generator and for letting us use his hotel’s garden to celebrate Elizabeth Guest’s 8th and Rachael Sarson’s 21st birthdays (including a memorable performance from our special friends The Widders). Sustenance this year was provided by Steve Waite who, ably assisted by Archie Gillespie (and Rob Riddett in the ‘breakfast tent’), did a fantastic job of feeding dozens of hungry archaeologists every day, while somehow maintaining a semblance of civilisation in the camp and keeping spirits up even in wet boots. Thank you to everyone who helped dig Caerleon’s Southern Canabae in 2011 and for making our month on site once again such a rewarding and enjoyable experience. CAERLEON SOUTHERN CANABAE 2011 EXCAVATION TEAM Core staff Peter Guest - Director Anna Gow - Site assistant Mike Luke - Director Robert Riddett - Site assistant Ian Dennis - Site supervisor Rebecca Smith - Site assistant Caroline Pudney - Site supervisor Megan Tudor - Site assistant Chris Waite - Finds supervisor Scott Williams - Site assistant Paula Jones – Community Archaeologist Steve Waite - Cook Archie Gillespie - Site assistant 113 Guest, Luke, & Pudney Student archaeologists Kelsie Armstrong Luke Green Kai Lumber Daevid Bedwell Matthew Gwynn Ana Lutescu Emmelia Booth Jade Hanley Christopher Matthews Emily Bowyer Elliot Heade Rachael Sarson Gwion Dafydd Michelle Humphreys Hollymae Steane Price Cassandra Davis Owen Jones Emma Stephens Nicholas Dawson Rebecca Lewis Rebecca Trower Jonathan Durman Daniel Lovelace Pre-university taster students Jacob Ball Lucy Hannam Marie Wall Dominic Chorney Charlotte Haywood Hannah White Jess Evans James Lorimer Alex Fallows Radu Pitis Roman Society Fieldwork Bursary holders Jack Donnelly (Colchester Sixth Form College) Rebecca Jenkinson (Colchester Sixth Form College) Eleanor Merry (Colchester Sixth Form College) Martha Page (Cirencester College) Oliver Swindall (Cirencester College) Volunteers Sue Adams Morgan Jones Linda Stanton Jane Ashwell Freya Knowles Anne Sterry Amanda Chadburn and Torin MacDonald Jonathan Lambert Greg Tasker Ryan Linton Sally-Anne Taylor David Chapman Phillip Mills Dawn Thomas Kelsey Dronield Warren Moore Lisa Venables Lynn Earley Jennifer Nye Neil Whatley Keith Edger Kate Pannel Sarah Jayne Evans Lesley Parratt Catherine Ferguson Kia Perryman James Goodsell Alex Raymond Sam Grainger Elliot Rees Tony Hearn Babs Roberts Verdun Howells Amelia Schafer-Rutherford Angelo Italiano David Standing David James Hywel Stanton Time Team Phil Harding Matt Williams 114 Raksha Dave Alex Langlands 33 CARDIFF STUDIES IN ARCHAEOLOGY The excavations at Caerleon in the summer of 2011 were focused on the complex of monumental buildings outside the fortress of Isca. Discovered during recent geophysical surveys between the amphitheatre and the River Usk, this impressive suburb extended over about 5 hectares and included some very large Roman buildings. The Caerleon Southern Canabae 2011 season of evaluation excavation produced the first glimpses of these ancient structures’ remains, which lie close to the surface of the fields around Broadway Farm and are very well preserved. This report presents the preliminary results of the nine trenches opened across the full extent of the complex, beginning with the stratigraphic narratives and followed by summaries of the finds and environmental evidence. The preliminary study of the pottery assemblage indicates the suburb could have been first constructed at about the same time as the fortress (i.e. A.D. 70s), but that the majority of the buildings would seem to have been abandoned perhaps as soon as the early-third century. They were possibly used for the disposal of rubbish during the late Roman period, including the remains of unusually large quantities of pigs and birds, after which the buildings were stripped of their stone and tile before disappearing for 1,500 years. The final discussion looks to provide possible answers to the project’s original research questions, concluding with some observations regarding the layout and architecture of the Southern Canabae and its relationship with the fortress and the legionary command in this part of Britannia. Peter Guest is Senior Lecturer in Roman Archaeology at Cardiff University, Mike Luke is Senior Project Manager at Albion Archaeology, and Caroline Pudney undertook the initial archiving and post-excavation work at Cardiff University after supervising during the 2011 season. Caroline currently works at Cadw.