ROMAN VILLAS IN BRITAIN

By Guy de la Bédoyère

Back to the Home Page

Contents

What is a villa?

When were they built?

How were they planned?

What were they built of?

Manufacturing tile and brick

Decorations

Who lived in them?

What did Roman Villas look like?

What happened to villas after the end of the Roman period?

Books to read

Visiting Roman villas in Britain

What is a villa?

The classical Italian Roman villa was a working rural business, with farmland and vineyards. It was also a luxury retreat. Italians had a real passion for rural idylls – escaping Rome with all its traffic, smell and noise was popular. The Roman writer Seneca fled to his ‘place at Nomentum … as soon as I left that crushing air in Rome … I noticed an improvement in my condition. You can imagine just how invigorated I felt when I reached my vineyards’ (Letters 104.1).

So what is a Roman villa in Britain? Archaeologists in Britain call almost any Roman house in the countryside a villa. The range is colossal. Some like Woodchester (Glos) compare with eighteenth-century stately homes. They sported lavish mosaic floors, wall-paintings, marble statuary, columns and balustrades. But few Romano-British villas were palatial. The majority were considerably smaller and included houses like Sparsholt (Hants), farmhouses with outbuildings where the owners could only afford one mosaic.

It is worth stressing that the villa house was only one part of the villa. A villa estate did not just include the main building, and attendant structures like baths, barns, garden shrines and mausoleums. Villa estates could encompass villages of workers several miles away. Unfortunately, without any specific estate records to work from we can only guess at the resources any one villa controlled.

When were they built?

Soon after the Roman invasion began in AD 43, 1960 years ago, there was a radical change in how houses were built in southern Britain though some influence had already spread from the continent. The invasion speeded up the process. Traditional timber and thatched roundhouses – efficient, durable, and comfortable – fell gradually out of fashion, but remained in use in remote areas. Within forty years most new houses were rectangular, had several rooms, often tiled roofs, and were usually built with at least foundations and wall footings made of stone.

Left: plan of the early house at Quinton (Northants) (after Taylor). North at left. The stone rectangular house (about 7.5 metres wide) was built in the late first century over an early-first-century round house. The new house fell out of use c. 170 and thanks to this early termination the evidence for its development was preserved. Since the new house lies over the old one there must have been a hiatus in occupation but this might only have been for a year or two while the new building was erected.

The biggest effects were in the Roman towns. By the late first century places like London and St Albans (Verulamium) were filled with timber strip houses opening onto the street. By the second century stone townhouses were being erected in all the province’s colonies like Colchester, and regional capitals like Cirencester.

The countryside was different. Fishbourne villa (West Sussex) was a freak (left). Built on a palatial scale in the mid to late first century, it either belonged to a tribal client king or a Roman governor and may even have had its origins before the conquest. It’s the exception that proves the rule.

Rural change generally was slow and while a few other significant stone houses like Eccles (Kent) started to appear in the first century, it took longer for the Romano-British to invest big money in country houses. It was the late third century before the golden age of Romano-British villas began.

This was also when towns stagnated, and decayed. While public buildings like basilicas, forums and theatres were going out of use or being adapted for different functions, some people started investing vast sums in villas. The likelihood is that the aristocracy, which had once supported urban development, decided to spend their time and money on rural estates.

How were they planned?

Because land availability wasn’t usually a problem and it was difficult to light interiors, Roman villas tended to be rows of rooms, or wings. The simplest were one row, usually with a corridor. Then a pair of small wings might be added – we call this the ‘winged corridor’ villa (very common). This isn’t a particularly clever observation – you can see the same phenomenon in English palaces like Hampton Court or eighteenth-century stately homes.

Left: suggested reconstruction of the façade of the winged-corridor house at Newport (Isle of Wight). The grey rectangle at far left represents a water tank that might have sat on the stone plinth. I have based this on a Roman funerary model of a winged-corridor house from Fontoy-Moderwiese (illustrated by D Perrin, The Roman House in Britain, Routledge 2002, p. 116). Compare with the Sparsholt reconstruction below, based on a similar plan.

To make a house bigger the wings were extended around a courtyard. Then another courtyard and more wings might be added. There was one more type: the aisled villa, built on a plan similar to a church with a nave and aisles. It was a Roman design used for many sorts of buildings.

There was usually a clear distinction between ‘public’ and ‘private’ rooms. Villa-owners of modest means might put a mosaic in the main reception room to create as grand an impression as possible. The wealthiest could afford to have whole wings given over to rooms for entertaining and dealing with clients.

But every villa in Roman Britain was unique. Most conformed to a basic type but with huge differences in detail. Bignor (W. Sussex) and Lullingstone (Kent) are two of the most famous villas but they are totally unalike. Bignor in its fourth-century form was ranged round a large courtyard, and was made by joining up the original second-century house to various outbuildings including a baths and barn (left). Lullingstone is a single block crammed onto a river terrace. It changed in size and shape but not very much.

Almost every villa had extensions and alterations. An archaeologist has to work backwards from the remains of a house in its final form. A common pattern found is a late Iron Age roundhouse knocked down and replaced with a modest rectangular house. That new house might be demolished and replaced with a bigger one, or extended again and again. Roman villa owners could continually ‘upgrade’ if they had the resources, or downgrade when they ran short or changed their minds. At Lullingstone for example, the baths were demolished while the rest of the house carried on for decades.

Villas didn’t have to be a single building. At Sparsholt there were three completely separate buildings around a courtyard. The main house was probably for the family, one was for the baths and the other might have housed slaves and farm animals. Bignor was similar until the owner joined them all together in one large courtyard villa. A villa complex could include barns, latrines, shrines and even family tombs.

What were they built of?

Roman villas were mostly built from local materials. That meant timber and local stone. Tile and brick was made from nearby clay in kilns set up for the job, or the local stone was used for slates. Mosaicists worked on site, or brought in prefabricated panels from their workshops.

TIMBER

Timber was vital for Roman house building. The simplest had posts driven into the ground. But timber frames were also used: horizontal sleeper beams and then other posts and beams to make up a frame. Not unlike medieval houses, these timber frames could be built on stone footings and stone foundations, or straight into the grounds. Walls were made of wattle-and-daub, timber cladding, rubble, mud brick or cob.

Timber-frames continued to play a large (but unknown) role in buildings that survive in the record as stone footings. But almost any house that started life as a wattle or timber-framed building, was eventually (and sometimes quite soon) replaced in stone, though not always.

Timber was essential for roofing (see Manufacturing tile and brick below), but since no Romano-British villa (or any other building’s) roof timbers survive we have a problem of interpretation. When timber does survive, the evidence suggest the Romans were competent and skilful carpenters. A large section of an early second century timber warehouse was recovered from waterlogged levels at Southwark in London in the late 1980s. The summary report includes the following observations:

‘The sill-beams were joined by means of edge-halved scarf dovetails over which a central beam was clipped. As well as having dovetail and lap-housings the sill-beams also had mortices cut into them … the wood held together extremely well with complete pieces up to 5m. During the lifting various tool-marks came to light as well as further complicated jointings.’ J. Dillon, ‘A Roman Timber Building from Southwark’ in Britannia, vol. xx (1989), 229-30

At Carlisle a number of timber structures have been identified in waterlogged Roman levels. This is a description of a ‘very large rectangular timber building’ erected some time in the first 40 years of the second century:

‘The walls were most carefully constructed, with timber posts morticed and tenoned into sill beams joined by dovetail scarf joints.’ M. McCarthy, Roman Carlisle (2002, 75-6)

If a Southwark warehouse could be built by competent carpenters, it is fairly plain that skilled craftsmen were working in Britain early in the province’s history, probably originating in the Roman army. Evidence from literary sources shows that they also knew about using wooden pins or trenails for securing joints (also known as dowelling). Cato, in his de Re Rustica, talks about clavi corneis, ‘pins of cornel-wood’ (xviii.9) being used to secure dovetail joints as does Pliny the Elder (xvi.206). Cornel-wood was, and is, known for its exceptional hardness. A possible example of a trenail made of ash from the late first or early second century AD has been found at Frocester villa (Price 2000, vol. I, 149, no. 29). Oak trenails were found securing joints in both the County Hall and Blackfriars Roman ships found in London. But it is impossible to know how much use they made of such joints because we don’t have the physical evidence.

As crafts and trades were handed down and disseminated these carpentry skills would have become more and more widely available. Nevertheless, there was a vast range of Roman villas in Britain running from simple corridor houses right up to the palatial courtyard complexes like Woodchester so of course skills would have varied according to time, place and available resources, as they do today. But the idea that Roman carpenters were universally jerry-building slobs, which has been put forward, is untenable both because the evidence is very limited and some of what there is, suggests they knew what they were doing (a useful link with technical descriptions of joints is at Roman woodworking).

Many early villas began life as timber-frames built on masonry foundations. Boxmoor (Herts) villa had loose stone footings with a timber frame on top, and wattle-and-daub walls. The owner added some mortar floors and wall-plaster. By the early second century the house had been burnt down, probably deliberately so that a new house could go up. The new cob house was built on the same alignment, exceptionally sealing the limited evidence for the timber house. Cob was more resistant to fire, but a later owner rebuilt again in stone.

Left: the south end of the winged-corridor villa at Plaxtol (Kent) as it might have appeared in the second century. The form of the windows, height, use of timber frame and so on, are all hypothetical.

Unfortunately, because villas survive normally only as foundations and the lowest courses of stone, it is very hard to know how much any one building made use of timber. Timber staircases for example simply do not survive so we do not usually even know if there was an upper storey. Measuring wall thicknesses or depths of foundations doesn’t provide an answer because buildings were not erected according to precise calculations. Moreover, a building might have been built with a single storey, but was later given an upstairs.

STONE

Stone and tile are usually the only structural materials that survive on a villa site, apart from traces of window glass and iron grilles. Timbers rarely survive because they rot and archaeologists can only detect wooden houses when they burned down, leaving carbonized remains, and that happens most often in towns.

Stone survives in foundations and lower wall courses. Occasionally imported exotic stone was used. Fishbourne is the most obvious early example, but it was expensive and normally restricted to decorative features. Very few villas produce traces of imported stone.

Stone walls depend on stone foundations unless a level bedrock platform is available. The Romans dug a foundation trench and filled it with rubble, stone and concrete, depending on what was available. If necessary, timber piles were driven into damp ground.

Most Roman villas in Britain are in the south and east, but much of this area is short on stone suitable for carving and into blocks. Instead, in Kent and the south-east the only abundant stone is ragstone flint.

Every flint nodule is uniquely randomly shaped. Building a wall means gathering cartloads of them, laying each one carefully into place and packing it with lime mortar. Lime mortar is made by burning chalk in a lime kiln and then adding water and an aggregate like sand. But the mortar needs to cure. Freezing destroys the curing process, so lime mortar building was only carried on between April and September.

Strength came from weight and packing rather than the mortar. Constant maintenance was need to stop water getting in and eroding the wall. Neat corners and edges are almost impossible with irregular flints. Roman flint walls were normally held in a frame made up of tile and brick, or with pieces of dressed stone imported from elsewhere. This part of Lullingstone shows a brick arch (above) and on the right brick quoins forming a door jamb. The doorway was later blocked up with more flint nodules. Throughout the building tiles were used as levelling courses. These helped provide regular flat platforms onto which a new band of flint wall could be built.

Limestone from the Cotswolds and Lincolnshire is easy to quarry and work. Both areas are excellent agricultural areas and were dotted with villas built from the limestone. Inherently more stable than a flint wall, limestone walls derive their strength from more regular courses of stone than is possible with flint. Regularity of courses improves the fairer spread of structural load. This wall (left) at Chedworth shows neat courses of stone, with a vertical channel for flue tiles to carry fumes from the hypocaust heating system below the mosaic floor.

The ease of carving some limestone meant brick and tile quoins were less necessary, and the stone could even be used to make roof slates (generally diagonal, and pierced with a nail hole just as today). Pieces of fully dressed stone were used for corners, window and doorframes, lintels and decorative features, like this example found at Chedworth. Owners of flint villas had to import stone for columns and balustrades, or used wood that has now rotted away.

Sometimes villa owners made use of handy pieces of stone even if that meant appropriating public property. At the Clanville (Hants) villa, a milestone seems to have been removed from the road three miles (5 km) away and brought to the villa where it was tooled ready for building work. Since the stone names the emperor Carinus (282-3) that helps date the work, since it must have occurred after 283.

Left: the Clanville milestone (RIB 98) made of sandstone, taken from a road and reused at the villa. It reads M AVR KARINO N CAES, ‘For Marcus Aurelius Karinus, most noble Caesar’ (AD 282-3)

Manufacturing tile and brick

So long as suitable clay, sand and water was available, tiles and bricks were manufactured on-site or nearby. They were laid out to dry before firing and it is very common to find footprints and pawprints of dogs and even children who wandered around across the tiles.

Roman clay roof tiles were generally of two types: the tegula and the imbrex. Tegulae were large rectangular tiles with raised flanges on each side (very approximately around 300x360mm, weight 4.9kg). They were laid in rows down the roof, with the flanges on each side running beside the flanges of the adjacent rows. Some were nailed onto the roof timbers.

The imbrices were curved rectangular tiles (around 15x360mm). A row of imbrices covered the tegulae flanges. The result resembled roofs in Italy and Spain today and I have used one to show how a Roman roof might have been constructed.

In areas where easily-worked limestone or sandstone was available (like the Cotswolds and Lincolnshire), hexagonal stone roof tiles were often used. Characteristically these have straight sides, pointed tails and heads, and chamfered edges, and a nail hole at the top (see left). They vary enormously in size from as little as 260x200mm to 460-290mm). They were laid out in diagonal rows, starting from the bottom and create an impression of diamonds laid in horizontal rows. One of the most useful discussions of their use is in Eddie Price’s publication of the excavations at Frocester Court (Stonehouse, 2000), see Volume 1, p. 131-8.

Of course thatch, and wooden shingles, were also probably used but evidence never survives for obvious reasons. A lack, or shortage, of tile though does not prove a villa had a thatched or shingle roof: tiles were routinely robbed from the ruins of Roman villas and other buildings, even during Roman times. Later, Saxon church-builders, for example, used them and the practice lasted well into the Middle Ages.

But no-one knows exactly how the timbers of a Roman villa roof were arranged because none has survived. They had to be steep enough to make sure rain ran off but not so steep the huge heavy tiles would fall off. Some collapsed villa walls give an idea of the angle of the roof pitch, but this clearly varied a little (see What did Roman Villas look like?) below.

Roof tiles were also used for other jobs. One at Lullingstone covers the drain on the mosaic (left). At the Rockbourne (Hants) villa imbrex tiles supported the floor in a hypocaust central heating system. Bricks served as quoins and leveling courses in flint villas, as the frames for windows and doors and as floor tiles.

A Roman tiler in Kent called Cabriabanus used a roller-die to impress his name on the flu-tiles he made for central-heating systems. It identified his work, and it also made a relief surface that helped the tile adhere to mortar. Normally tillers just used a toothed tile comb to create grooves in diagonal stripes across the tile to do the job.

Cabriabanus’s work has been found at several places in Kent so he must have been a jobbing tiler, taking commissions as and when they arose. But the villa at Ashtead (Surrey) seems to have been a tiling business. Tiles manufactured on a big kiln there, and distinguished by their decorative animal scenes, have turned up at various other Roman villas in the region.

Left: one of the distinctive flue tiles manufactured in the kilns at the Ashtead (Surrey) villa. The design was invisible in use, and served only as a keying for mortar but it served as a useful trademark and helps archaeologists identify their use on other sites.

Decorations

Roman villas were all about status. The richest decorated their best rooms with mosaics and painted wall-plaster, exotic stonework, furniture and statuary. The subject matter showed off the owner’s taste and knowledge of classical art and literature. Further down the social class, the less well-off villa owners commissioned just one or two mosaics from less accomplished craftsmen.

The super-rich installed grandiose bath-suites or dining rooms. Lufton (Somerset) villa had an ostentatious octagonal baths built on (pictured left). Littlecote (Wilts) had a dining room or cult chamber built as an apsidal hall with a mosaic depicting Orpheus (pictured below left). The less well-off might have a little bath suite tacked on, or just made out of a converted room within the house.

Central heating (hypocaust) systems, based on circulating hot air under floors and through walls, weren’t universal. They were installed in bath suites, and sometimes reception rooms. But most rooms had hard floors, and were heated by braziers.

Villa owners never left their gold and silver plate behind if they could help it, so it almost never turns up in villas. Very occasional finds are made within a villa building, like the huge hoard of silver coins found in the villa at Shapwick (Somerset). Concealed beneath a floor around the year 224 or a little later, the 9238 coins were never recovered for unknown reasons. The villa was demolished shortly afterwards so there was little chance of it being rediscovered in antiquity. The fabulous plate from Mildenhall, or the gold and silver found at Hoxne, must also once have belonged to villa owners. These fabulous finds have transformed our idea of what the wealthy in fourth-century Britain owned.

Mildenhall Great Dish (detail) Brantingham: mosaic nymph

What did Roman Villas in Britain look like?

The straight answer is that we don’t know because not a single one survives intact or even partly intact. Appearance of course is partly dictated by the ground plan, but that’s only a component of the story. As explained above, It’s almost impossible to say whether a villa had an upper storey because even thick walls can just be an indication of over-building.

Left: drawing of part of a wall-painting found at Trier that seems to depict part of a villa in a stylized form

The picture of Littlecote in the preceding section is completely hypothetical but some of the ideas that were used come from mosaics in Tunisia that show villas and their outbuildings, and a wall-painting from Trier. Unfortunately there is nothing equivalent from Britain, so the comparisons could be quite misleading.

In antiquity building was not based on tight financial calculations, or precise knowledge of the properties of materials; instead builders were inclined to make mistakes by overlooking filled-in ditches, but often structures were more substantial than they might have needed to be. In very rare instances collapsed walls give a good idea of a villa’s possible appearance. The picture on the left shows the villa at Redlands Farm, Stanwick (Northants). The façade of a wing was found where it fell, preserving the use of stone and even an idea of the pitch of the roof.

Another, very different, building was found at Meonstoke (Hants), pictured here on the left. Here part of the gable end of an aisled villa was found showing that brick and tile had been used to create a decorative blocked window effect. The section is preserved in the British Museum. Enough information survived from the site to indicate the pitch of the roof, showing that here at least this might have been as much as 50 degrees which is difficult to reconcile with the weight of the tiles. At Stanwick it was far less, perhaps as little as 20 degrees though this seems far too gentle – water would have been liable to seep back under the tiles. Most buildings were probably around 30-40 degrees.

When it comes to modern reconstructions of Roman villas all sorts of problems come into play, not least of which is planning reconstructions. The drawing below left shows a restoration of the gable end of the aisled house at Meonstoke. I used this and what I knew about Roman techniques and other buildings to produce the drawing below of what the villa at Sparsholt (Hants) might have looked like (below). It has an upper storey but I have no idea if there was one. The Discovery replication of this building (see Rebuilding the Past) resulted in a very different structure because planning restrictions outlawed the clerestory, while other factors meant the wings (shown here as the left and right gable ends), and various other decorative and functional components, were omitted or set aside. I included decorative features from Meonstoke because this villa is from the same region. But you can compare this with the Newport drawing at the top of the page, where the plan is similar. It just shows how much uncertainty there is about this.

Left: Meonstoke façade restored. Above: reconstruction of what Sparsholt might have looked like in its final form

What about verandahs and corridors? Were they open to the elements. Depending on who you listen to, you’d be forgiven for thinking some people knew for certain. I’ve put an outline of the problem at Verandah.

Who lived in them?

Villas weren’t evenly distributed in Britain. Most were in the south and east. There were almost none in Wales, or beyond Exeter, and even in southern England distribution was patchy. The Weald of Kent and Salisbury Plain had virtually none. The Weald was too wooded, but Salisbury Plain might have been an imperial estate where natives had to carry on living in roundhouses. In the north villas only turn up in river valleys and where farmland was good, like Lincolnshire, the Vale of York and East Yorkshire.

So, let’s be clear: villas represent a tiny proportion of the population in terms of accommodation. We know of around 1100 ‘villas’ but these range from farms to palatial establishments. Suppose on average an extended family of 15 live there and they have three times as many household slaves – 60 x 1100 = 66,000 people but that is probably an exaggeration. Seeing as Roman Britain probably had 3-5 million people (based on medieval comparisons) that means just 1.3-2.2 percent of the population lived in villas. Even taking towns and the army into account by far and away the majority lived on the land in nameless and unnoticed farmsteads, roundhouses and villages though many of these people might have worked the land on villa estates. We pay villas a lot of attention because of their visibility (to us).

Even so, we don’t have the name of a single Roman villa owner in Britain, or any records of how any one estate functioned, where its money came from, how many slaves it had and so on. This picture shows a wall painting from the villa at Brantingham (E Yorks). Perhaps this was the lady of the house. It is impossible to say. At Thruxton (Hants) a mosaic seems to name a man called Quintus Natalius Natalinus, perhaps the owner. The meaning of the latter part ET BODENI, ‘and of Bodenus’, is unknown.

But we know the Romans ruled by delegation. They used local chiefs and ruling families to rule for them, by installing them in positions of responsibility. The descendants of tribal leaders that had fought against Rome probably sat on city councils and built those early stone townhouses. It was probably these families that later owned many of the villas. Others might have been owned by retired soldiers, or even immigrants from Gaul.

But like so many wealthy people at different times, getting out of town became fashionable in the late third century. They retreated to the privacy of rural estates and spent their money there, where they could parcel up local economies and politics in peace. They’d also started to model themselves on the Romans of old. Their mosaics had classical myths depicted on them, and some of the most pretentious had Latin verse laid out on the floors.

This (left) is the main mosaic at Lullingstone. It depicts Europa being carried off by Jupiter in disguise as a bull. This mythical event is mentioned in the first book of Virgil’s Aeneid. The floor has a metrical couplet on it:

INVIDA SI TAVRI VIDISSET IVNO NATATVS

IVSTIVS AEOLIAS ISSET ADVS QVE DOMOS

It means ‘If jealous Juno had seen the swimming of the bull, she would with more justice have gone all the way to the halls of Aeolus’.

The allusions of all this are too dense to go into here, but the style of the composition recalls the work of Ovid in the way each line ends. So, at fourth-century Lullingstone, the owner was showing off his knowledge of the work of Virgil and Ovid, poets who had been dead by then for more than 350 years. It was rather like one of us imitating Shakespeare to show off today and it says a lot about how the educated villa owners of Roman Britain wanted to see themselves and pose to their visitors.

What happened to the villas after the end of the Roman period?

Forget the idea that the Romans ‘left’. What happened was that the Roman government stopped administering Britain. That meant it stopped paying the soldiers, and raising taxes. Town government also faded out. That cut right to the quick of the whole system. Money stopped circulating, and while that didn’t stop farming or trade, it did stop the need to sell produce to raise cash to pay taxes.

Since villa economies depended on urban markets, communications and other support industries, they couldn’t carry on. Some, like Frocester (Glos) lasted well into the fifth century. But there was neither the money nor the services to maintain the houses. One by one they fell into ruin. The reason was not that they had been built by incompetents, but simply that any building requires maintenance to stay in one piece. The same applies today, and has throughout history. A few villas did collapse prematurely, like Meonstoke, but these are the exception.

Once the roof is damaged by the weather, or a fire starts, the timbers rot or collapse. Without the compression force of the roof, the walls start to lose stability and disintegrate. Wall-plaster fell off in lumps. Rabbits undermined mosaic floors, collapsing into central heating voids below. Abandoned buildings attract stone and tile robbers, and the voids they left further compromised the structures. The huge villa at Dinnington (Somerset), pictured left, eventually left no trace on the surface. Modern ploughing eventually started to cut its way through the mosaics that were left and finally revealed the lost house.

To be honest, it is still a mystery why people let go a more comfortable way of life. But they did. For the wealthiest families, who lived in places like Chedworth and Bignor, the answer might lie in England’s super rich of the Edwardian era. Generations of those landowners had grown utterly dependent on armies of servants to do everything for them. When Roman Britain’s ‘system’ collapsed, perhaps the great villa owners were soon confronted with houses and estates they had absolutely no idea how to maintain and run without slaves and estate workers.

As centuries passed the houses disappeared from view and were entirely forgotten about until chance discoveries exposed them once more. But only with the classical revival of the eighteenth century did those discoveries lead to excavation, which in those days could lead to veneration or destruction or both. Little has changed!

Plaxtol (Kent). The winged-corridor villa, featured in a reconstruction drawing above,

under excavation c. 1986 in a view from the north. See how close the remains are to the surface,

yet the villa had remained undiscovered until a modern drain was laid.

Villa ruins (or indeed any Roman ruin) were the equivalent of a free supply depot. Medieval church builders helped themselves to stone and tile. Carved columns and other decorative features were taken away too. By the time archaeologists started to take an interest there was scarcely a villa visible above ground.