Chedworth in Gloucestershire is one of the largest and finest examples of a Roman villa in Britain, and recent research suggests it may also have been exceptional in its longevity. Radiocarbon dating has now indicated that a wealthy family continued to live there, in some style, well into the 5th century – the first time that this has been convincingly demonstrated for any of the villas of Britannia.
Lying at the head of an east-facing coombe with fine views across the Colne Valley, Chedworth was begun c.AD 120, and by AD 370 the villa had developed to become a distinctive place of luxury with three ranges of rooms constructed along the south, west, and north sides of two courtyards. Important visitors would have approached from the east, moving through a gateway into the lower courtyard, and from here steps would lead you to the upper courtyard where the best apartments of the family were situated on the sunnier and more prominently positioned west and north sides.
A spring issued from the hill slope between the west and north ranges, and two sets of baths – including cold plunge baths, and steam-heat and dry-heat baths – were created on either side of this water source. These facilities, alongside rooms for dining, meeting, relaxing, and sleeping, were decorated with painted wall plaster and fine mosaics: richly decorated spaces that demonstrate the exceptional wealth of the people who once lived here. The status of its inhabitants is confirmed by two recent exotic finds: a fragment of marble quarried from an island near Athens, and a piece of glass probably made in the Crimea on the shores of the Black Sea. Each was an artefact type that had not been found in Britain previously, suggesting that they were very exclusive imports.
But while Chedworth was undeniably grand, it was only one of many villas in the countryside surrounding the local provincial capital of Corinium (Cirencester), albeit a particularly elaborate example. This density of villas demonstrates both the economic significance of the area and, by the late 4th century, the quality of life that the locality offered. One is bound to wonder, then, how this community declined. What became of this network of leading landowners occupying their Romanised grand houses within their farming estates?
Up to now, the generally accepted narrative has been one of rapid decline. By the early 5th century, Britain was no longer part of the Roman Empire. In AD 407, the provincial government had backed a man given the title of Constantine III, who had taken troops to Gaul in a bid to become emperor. Initially he had some success, but Britain rejected him in AD 409 and he was defeated in AD 411. The legitimate emperor, Honorius, had problems of his own and subsequently made no attempt to reoccupy Britannia. The result was a messy end to over 350 years of membership of the Roman Empire.
Emperor Honorius… made no attempt to reoccupy Britannia. The result was a messy end to over 350 years of membership of the Roman Empire.
According to this traditional picture, with no Roman army occupying Britain, coinage to pay the soldiers ceased to be imported from across the Channel, and nothing seems to have been produced locally to replace it. Other artefacts also melt away from the archaeological record, with dateable pottery in particular becoming markedly less abundant. The picture is one of sudden economic collapse, with towns depopulated and villas abandoned. A once-intricate network of specialist products, markets, and crafts declined, and people turned towards subsistence farming to survive. Increasingly, however, research and interpretation of archaeological records suggests that this story is too simplistic. There is evidence for the continuation of a Romanised lifestyle of some sophistication, which can be discerned though a scattering of distinctive finds.
The problem with Chedworth is that it was uncovered quickly following its discovery in 1864, but the details of the excavation were never published. By 1868, a museum had been built beside a lodge in the centre of the villa, and the best finds placed there – but as no stratigraphic record, drawings, or photographs survive from this dig, the context of these artefacts has been lost. Moreover, evidence of the last occupation of Chedworth was swept away in the excitement of uncovering its decorated floor surfaces. Subsequent excavations have also remained largely unpublished, but this is being addressed – Professors Simon Esmonde Cleary and Peter Salway, with Jason Wood and Emma Durham, have created a monograph of all pre-2010 archaeological work, drawing together information from previous excavations and applying fresh analysis to the site’s surviving structures and artefacts. Curator Julie Reynolds guided the project and enabled the monograph to be sent to the publishers – it will be available in 2022.
Significantly, research for this volume has highlighted a number of unusual finds from the villa, demonstrating that wine and olive oil were still reaching Chedworth from the East Mediterranean in the 5th and 6th centuries. These late finds come from areas of the site unexcavated in the 1860s, away from the walls of the villa; they were discovered during excavations by Sheffield University in 2001, and include part of a wine amphora from Palestine which was found in the upper courtyard, and another from Asia Minor (Turkey), which was recovered from garden soil. Meanwhile, at the edge of the dining room (Room 32) at the east end of the North Range, three pieces from a corrugated amphora that once contained olive oil brought from Cyprus or Cilicia were found.
Another fragment of Asia Minor amphora, mixed with sherds of late Roman shelly ware, lay among the debris on the north side of the North Range. This discovery provides useful dating evidence, as the kilns making late Roman shelly ware only begin production after AD 360, and when this British pottery is found it is a rare indicator of 5th-century occupation. Nor was this its only occurrence at Chedworth; in 2000, a fragmented but near-complete jar made from this pottery was found placed in a pit cut through the floor of Room 27 in the North Range.
National Trust archaeologists have been busy at Chedworth in recent years. Between 2010 and 2012, excavations within the West Range uncovered rooms and mosaics so that they could be seen under a new protective cover building (CA 305). Once this was completed, between 2013 and 2018 research was then carried out in the North Range to help us understand the dating sequence of its rooms, and to establish the extent of the mosaics that still lay beneath the grass. It was hoped that this information would enable funding for another cover building, so that these mosaics could also be seen by visitors. Specialist reports from these excavations are now being brought together so that the project can be written up and published – and the results of this analysis have already allowed new information about sub-Roman Chedworth to be linked to the earlier discoveries.
A key development came in 2017, when we excavated a series of trenches within the North Range where information on surviving floor surfaces was insufficient. This included opening trenches in the corners of Rooms 27-30, while the entire surviving remains of a mosaic were uncovered in Room 28. Two trenches were excavated in the north-east and south-east corners of Room 27, beside the wall dividing it from Room 28. Once the turf was cut and the topsoil cleaned down to the surviving Roman remains, we could see that the floor surface had been almost entirely worn away, apart from a thin strip surviving against the north wall. There, a remnant of a crushed brick and mortar opus signinum floor remained, supported by a hardcore of limestone gravel mixed with mortar. Removing this exposed the Cotswold limestone bedrock, which had been cut to create a foundation trench for the dividing wall; the wall had been built in the trench, and then the gap between the wall and the trench edge was filled with a dark soil.
The finds recovered from this foundation trench filling were crucial in dating the wall. They consisted of two small fragments of animal bone and a black piece of pottery, while the soil also contained fragments of twig charcoal which were collected for radiocarbon dating. It was clear that the dividing wall had been inserted between the north and south walls of the North Range, as it was not bonded to them and its style of building was of roughly dressed rubble – a stark contrast to the neatly coursed limestone blocks that the earlier walls were made from. More charcoal was found in the foundation trench for the south wall; this was collected too, and there were further clues to come from Room 28. There, the mosaic proved to be fragmentary, with only about 30% surviving. The central design had been completely worn away and there were traces of burning caused by later use of the space. Someone had created two hearths in the middle of the room, one using recycled quern fragments as its floor, while the other used a limestone kerb surrounding three hypocaust box-flue tiles, laid side by side. The tile hearth was also radiocarbon dated, with samples taken from the burnt material within the cavities of the box-flue tiles.
As we waited for results of the dating analysis, we thought that the south wall would probably date to the later 2nd century, the wall between Rooms 27 and 28 would be later 4th century, and that the hearth built within the worn central area of 28 would be sub-Roman in date, somewhere in the 5th-7th century. We were partly right – but some of the dates proved to be rather more surprising.
The results are in
We were right about the south wall: our radiocarbon date of AD 75-219 (at 95% probability) was backed up by diagnostic pottery found in the soil filling the foundation trench, as well as within the layers it was cut through. Incidentally, we thought that for the 2nd-4th centuries, the broad date range of radiocarbon results usually made pottery evidence more precise than radiocarbon dating, which was a disincentive for trying the technique. Despite this, we took the samples from the foundation trench fillings and this decision was greatly rewarded. The result from the charcoal sample, collected from the dividing wall foundation trench filling, came back as AD 424-544 at 95% probability.
This was very unexpected because, if this wall were built after AD 424, then the mosaic within Room 28 had to be later, as its intricate pattern fitted exactly within the area of the room, defined on its west side by this newly dated 5th- to 6th-century wall. The accepted narrative for sub-Roman Gloucestershire does not include the polite refurbishment of buildings and the laying of new mosaic floors. In any case, what about the hearths and evidence of a later workshop that had damaged the middle of Room 28’s mosaic? Once again, our date estimate was wrong. The two dates from the hearth, constructed from reused box-flue tiles, placed it not in the sub-Roman period, but much later, and revealed that it was medieval, built at some time between the 12th and 15th centuries.
The 5th- to 6th-century charcoal date from the foundation trench needed back-up, and one of the bone fragments was sent for analysis. Results were delayed by the first COVID-19 lockdown, and when the date finally arrived it was less precise – but it did confirm that the wall was built late in the life of the villa, with a date ranging from AD 337 to 537 at 95% probability. As another piece of the puzzle, the fragment of black pottery from the foundation trench filling was also examined and identified as late Roman shelly ware, the pottery type which has not been found in deposits pre-dating AD 360.
The accepted narrative for sub-Roman Gloucestershire does not include the polite refurbishment of buildings and the laying of new mosaic floors.
We had also found late Roman shelly ware filling a drainage gully in Room 29a in 2017, as well as in soil packed either side of a limestone drain built along the outside face of the North Range. Other late finds included fragments of Saxon pottery of a type that was is in use up to the 8th century, which was scattered in various late debris deposits around the villa – another indicator of the late occupation of the site.
Evidence for 5th-century refurbishment has not been found across the whole site, however. Chedworth’s South Range is located downslope and consequently has a less sunny aspect than its West and North counterparts. It is thought to be the place where the working or service rooms were located, and where perhaps the servants had their accommodation. This area remained unexcavated in 1864, but a small trench was dug here beside the lower courtyard in 1997 and 2000, uncovering evidence for a grain-dryer cut into the South Range portico. The archaeology suggested a building in decline, and burnt grain found in this feature yielded a radiocarbon date of AD 385-539 at 95% probability.
Further evidence for Chedworth in decline came in 2012, when the West Range corridor was uncovered. We found that concentrated pedestrian access across the central door threshold and along the corridor northwards (heading towards the baths) had eroded away much of the mosaic floor – the surface had been patched up with clay mixed with large sherds of pink grog-tempered pottery. The eroded area and the pottery suggested a late Roman or sub-Roman date, but the building’s occupants had commissioned no new mosaic for this part of the villa. In Room 6 (also in the West Range), though, a baked clay structure was found to have been built into the worn central part of the mosaic. This probable hearth remains undated but, like the hearths in Room 28, it suggests a lower-status, utilitarian use for this room.
Chedworth in context
What can we make of this evidence for both refurbishment and decline at sub-Roman Chedworth? Of the three styles of mosaics identified by Stephen Cosh at the site, the latest and least sophisticated mosaics are confined to the east part of the North Range. These are the Room 28 mosaic, the gallery mosaic, and a small surviving fragment further east in Room 31a. This part of the villa had been extended in the 4th century – perhaps the owners, under reduced circumstances, concentrated their polite accommodation within this part of the complex, while the West and South Ranges were ‘closed down’ or converted to workshops or for storage.
What of the pottery evidence? The fragments of 5th- to 6th-century East Mediterranean amphorae from the site were traded through the ports of western Britain. Tintagel in Cornwall, for example, is well known for its high number of such imports, and in Gloucestershire, Chedworth has the greatest range of late amphorae so far identified. It is not unique among West Country villa sites, though. Jane Timby, who has written the reports on Chedworth’s pottery, has compared finds from seven sites, with sherds of East Mediterranean amphorae identified at the local villas of Great Witcombe, Kingswood, and Frocester. These sites also yielded late Roman shelly ware, as did other villas in the area, such as Kingscote, Barnsley Park, and Wortley. Similarly, there is evidence for 5th- to 6th-century occupation from Cirencester and Gloucester, and smaller sites like Bourton on the Water and Wycomb.
Perhaps then, in the light of this new evidence, we can reconsider the relevance of the scant documentary evidence provided by the accounts of St Gildas and St Patrick, which hint at the survival of a high level of Romanised learning and sophistication into the 6th century. The British kingdoms of the West Country seem to have continued until AD 577, if the entry in the late 9th-century Anglo-Saxon Chronicle can be relied on. For that year, it records the Battle of Dyrham, where three kings were slain and the cities of Gloucester, Cirencester, and Bath were captured. That they were mentioned this late in the 6th century suggests that they were still significant places at that time.
However, as ever in archaeology, more research is required and further sealed samples for dating need to be obtained to improve our understanding of Chedworth and sub-Roman Gloucestershire. At Chedworth, we hope to take additional samples from around the site to build a stronger case for its late occupation, but, until then, the surprising radiocarbon dates from Room 27 currently give the best evidence for a 5th-century mosaic in Britain.
Simon Esmonde Cleary, Chedworth: Life in a Roman Villa, History Press in association with the National Trust, £17.99, ISBN 978-0752486437
Daily blogs from the 2017 and 2018 Chedworth excavations are available at https://archaeologynationaltrustsw.wordpress.com, and you can read about the Room 28 mosaic at: https://archaeologynationaltrustsw.wordpress.com/2017/08/18/day-5-chedworths-room-27