Just north of Bulford, inside the MOD’s Salisbury Plain estate, lies a site called Dunch Hill. This spot is close to the military drivers’ training area, and in 1995 excavations there by Wessex Archaeology had revealed the presence of roundhouses beneath a track that was being laid. In 2020, as part of an assessment of the effects of ground pressure from military and farm vehicles on archaeological deposits, we were fortunate enough to excavate the footprint of another roundhouse adjacent to where the previous investigation had taken place. Working between COVID-19 lockdowns – outdoors and socially distanced – our team was led by Wessex Archaeology’s Phil Andrews, as the 1995 excavation had been, this time joined by Operation Nightingale.
Operation Nightingale began in 2011 with the mission of improving the lives of Wounded, Injured, and Sick (WIS) military personnel and veterans by involving them in archaeological fieldwork. Issue 338 of Current Archaeology has an overview of some of the initiative’s work, but specific projects have also had dedicated articles – within Salisbury Plain these include the Bronze Age barrow and Anglo-Saxon cemetery at Barrow Clump (CA 306), and a camp associated with ‘Easy Company’, the American D-Day paratroopers better known as the ‘Band of Brothers’ (CA 354). Further afield, veteran volunteers have been involved in the investigation of a high-status Roman building at Caerwent (CA 282); the graves of convicts or Napoleonic Era prisoners of war on ‘Rat Island’ in Portsmouth Harbour (CA 339); an 18th-century camp that once housed 8,000 German mercenaries outside Winchester; and a network of First World War practice trenches at Barry Buddon, near Dundee (CA 346). The resulting archaeological discoveries have been undeniably illuminating, but the results for the veterans, which are measured, have been especially encouraging – even for those who are simply looking for some respite from the daily pressures of life, there are positive outcomes. The global trials caused by the COVID-19 pandemic have only served to heighten the challenges for these participants, with a particular emphasis on mental health and wellbeing. And this is where a recent programme based around Late Bronze Age roundhouses has proved invaluable.
While the focus of our fieldwork was excavation, we also wanted to set Dunch Hill in its wider landscape context, undertaking many hectares of geophysical survey across the location of Early Bronze Age barrows, so-called ‘Celtic Fields’ (actually Middle Bronze Age field systems), and Late Bronze Age linear ditches – documenting a real who’s-who of the Bronze Age. As for the excavated evidence, we found quite a few sherds of Late Bronze Age pottery, a Late Bronze Age disc-headed pin, and – most excitingly – lots and lots of post-holes. There were so many, in fact, that it proved quite a challenge to decipher the footprint of any individual house structure, and it was quite a ‘Eureka!’ moment when Phil Andrews succeeded in identifying one, forming a rough circle c.6.5m in diameter. The next step was to determine how the long-decayed roundhouse that these post-holes had once supported may have looked.
Researching roundhouses
One of the wonderful things about archaeology is that there really is a job for everyone, whether that task is physical or intellectual. Even during the period of restrictions on movements in 2020 we were still able to continue our work remotely, drawing together a team of volunteers to research the nature of Bronze Age structures. At Dunch Hill, our starting point was a series of post-holes, so we had a fairly clear canvas to work on (albeit with dimensions, orientation, and ancillary structures, such as fences and ‘four-post’ structures, set in stone – chalk, in this case). Thanks to the opportunities afforded by the internet, and digital resources such as local Historic Environment Records and the Archaeology Data Service, volunteers could all familiarise themselves with their local Bronze Age deposits and see what might be applicable to Salisbury Plain. Previous work by Rachel Pope, who has considerable experience in researching roundhouses, proved really important to us, too, during the design process.
The excavated evidence indicated that our roundhouse had no stone walls, and the local geology was chalky, so these parameters were key in devising our design. No elements were allowed to breach what we had found at Dunch Hill (although we did ultimately make one exception in terms of its orientation, as we will explain later). Structural evidence was slim, but luckily the 1995 excavation had identified a number of more straightforward Later Bronze Age roundhouses in the immediate vicinity of our example, which we could use to help inform our reconstruction.
Nonetheless, we needed all the help we could get, so we spoke to Trevor Creighton and Claire Walton from Butser Ancient Farm, an experimental archaeology centre in Hampshire, and they then brought in Rachel Pope to advise on the build. Fortunately, Rachel agreed that the irregularly arrayed post-holes were solid evidence for a roundhouse – and said that she was keen to see the Dunch Hill structure reconstructed precisely because of the lack of evidence for walling.
We wanted to test the efficacy of various building styles and materials so that the project had genuine academic worth, as well as being good for participants’ morale. We have always set out to do right by the archaeology as well as by our Operation Nightingale volunteers. Once we had an idea of how the roundhouse may have looked, this led to discussions with Trevor and Butser Ancient Farm on the potential for trying to build a physical reconstruction of our findings. Interpretations of Iron Age roundhouses or crannog sites are familiar, as too now are ‘Durrington Walls’-type structures from the Neolithic period (such as those outside the Stonehenge Visitor Centre). Butser itself has a magnificent Neolithic longhouse based on the findings at Horton (see CA 375 and 377), as well as a cluster of Iron Age roundhouses, a Roman villa, and two Anglo-Saxon houses. One period, however, was underrepresented at the Farm: the Bronze Age.
We wanted to redress this, and thanks to our partners (South Downs National Park, the Armed Forces Covenant Fund, Breaking Ground Heritage, and Step Together) this wish has now become a reality. In what always seemed to be sun-drenched days spanning spring-autumn 2020, the team stripped turf, dug holes, shaped timbers, painted murals, smelted copper, cast bronze, carved figures, and more: there was a job for everyone. During the project the volunteers forged their own bonds and friendships, as well as making enduring archaeological connections – some are now guides at Butser, or have gone to university to study archaeology. So far, this has been the Operation Nightingale project of greatest longevity and diversity of activities – in perhaps the most trying of times, too. Feedback from participants has been outstanding, and our task now will be to build on this (wattle-and-daub or clunch!) framework to maintain the benefits. The project will need maintenance, which is in itself a real positive, providing important academic data on these structures and the longevity of their experimental walls. So, what have we built?
Reconstructing Dunch Hill
Before construction began, we spent a lot of time talking about central post-holes. This is an uncommon feature of roundhouses generally – only 3% have one – but a clear central post had been used to support the roof of two of the structures excavated at Dunch Hill before, and could also be identified in the archaeology of our building. An obvious question arose: if central posts are so useful in roof construction, why are they so uncommon? The simplest answer would be that this scarcity is a result of the usually central position of the hearth, for which the safest location was under the apex. Dunch Hill lacked any evidence for a hearth which, together with its orientation, might suggest it was a seasonal building rather than a permanent dwelling. Such structures are well-attested archaeologically. Interestingly, when Butser’s founder Peter Reynolds built the centre’s first reconstruction, an Iron Age roundhouse based on remains excavated at Maiden Castle in Dorset, he believed that the use of a central post had actually accelerated structural collapse by introducing further stresses.
As mentioned above, at Dunch Hill the archaeological evidence comprised a ground-plan of substantial post-holes, set in a slightly ‘flattened’ circle c.6.5m in diameter. The sizes and locations of the posts are defined by the archaeology, suggesting a framework of upright posts, topped with a horizontal ring-beam, or ‘wall-plate’. Such a construction is an excellent way to build a strong support for the roof, and an interpretation familiar to us from the numerous Iron Age round-houses built at Butser, including Peter Reynolds’ Pimperne House, constructed in 1976 and following another Dorset discovery. In the absence of evidence for walls, we at least knew they had to be set within this frame. But what form should the walls take?
Rachel said that a major feature left to experiment with in prehistoric roundhouse reconstruction was organic walling, which accounts for one in four prehistoric houses. Although we’ve said that there was no evidence for walls at Dunch Hill, it is more accurate to say that the absence is the evidence. Whatever they were made from, the house walls were more ephemeral than the posts; and because the post and wall-plate frame – a strong timber cylinder – carries the entire load of the roof, we had considerable freedom to explore walling options. The only prerequisite was the expectation of minimal archaeological evidence surviving three millennia hence. When it came to wall height, this was determined by practicality: we built our post and ring-beam frame c.1.5m high, interrupted by a doorway facing south-west – this was our only compromise on the excavated evidence, as the original northward orientation unfortunately did not work with the location we had been given at Butser – and the entrance posts held a 1.6m-high lintel.
The reconstruction’s walls are made principally of earth. Topsoil, as well as underlying chalk, would have been a readily available and easily won resource for local Bronze Age builders, but because the wall material is extracted from the ground on which the house is built, the archaeology left behind on the structure’s collapse/erosion might be relatively difficult to distinguish from natural materials. By building walls of this nature, then, we can test our own ideas on ways they can be built, their practicality, and their durability. In time, we can also assess what archaeological evidence they leave behind after demolition, however subtle. With this in mind, we built four different wall designs in discrete segments in the spaces between the posts, so that we can observe how they endure and change over time.
One section consists of stacked turves, another of loose topsoil, and both form sloping banks that critically work to buttress, as well as insulate, the structure. These were retained by light wattle hurdles – horizontal hazel rods woven through hazel uprights, or ‘sails’ – anchored in shallow holes in the ground and holes bored into the wall-plate. The keen-eyed observer might reasonably argue that, by anchoring the hurdles in the ground, we have contradicted the archaeology. Our justification is that the holes are so insubstantial – about 40-60mm round and deep – that they would leave no evidence in truncated archaeology (as argued by Graeme Guilbert). We acknowledge this as an unsupported rationalisation – an alternative strategy could have been to anchor them instead into the vertical posts.
The third section is a ‘gabion’ wall, with two parallel hurdles set in a post gap: one panel beneath the inner edge of the wall-plate and one beneath the outer edge. This offset provides a c.25cm space into which we have packed wool as an insulating material – a type found in the Tyne–Forth region during the Bronze Age. Each of these walls is lined on the inside with daub, which also seals the exterior of the gabion. Our fourth type of wall is cob walling, made from crushed and lump chalk mixed with straw, hair, wool, and water. They have no damp course, such as a base of flints, as Dunch Hill revealed no such evidence. As a result, they are vulnerable to erosion by water, but are arguably the most interesting in terms of interpreting the original archaeology.
Raising the roof
So, to the roof. The existence of other impressively circular and symmetrical roundhouses at Dunch Hill suggested to us that the irregularity of our excavated ground-plan was not for lack of skill, but suggested an unusual roof design. Supported by the legacy of nearly 50 years of roundhouse construction at Butser Ancient Farm, we feel able now to move beyond the strict adherence to structural engineering principles – always creating a neat circle/cone – and instead are happy to let the archaeology lead. If irregular roofs are so challenging, we thought, why do many prehistoric roundhouses have an irregular ground-plan? It’s not difficult to draw a good circle with a peg and a bit of rope.
In our reconstruction, the well-founded central post, 5.5m high, had been fitted with two 20cm-long pegs in a V-shape near the top, simulating a forked trunk, to act as a prop for the four principal rafters. Set at a 45º angle, these are the only rafters to meet at the apex, so avoiding messy overcrowding. They act as ‘principals’ by supporting the other 22 rafters, which were fastened to them just below the apex. The rafters were spaced roughly equidistant from each other, and were fastened to the wall-plate using rope and pegs. Multiple rings of hazel rods, lashed together at the ends to form a continuous length, encircle the roof, providing both a system of attachment for the thatch, and stabilisation for the structure. We knew the central post would aid construction, but were surprised by its effectiveness: the post and four principal rafters were erected in 2.5 hours without mechanical aids. However, because our central post was not truly central, neither was the roof apex. Combined with the irregularity of the frame, this created a roof that we found very challenging to build and thatch.
Our Dunch Hill building, then, is something of a tasting menu of ideas and questions. We are not suggesting that the original was of the ‘57 Varieties’ type. Rather, our aim is to maximise its experimental potential by testing a range of suggestions about how long-vanished structures might have been built. As evocative of the past as they are, experimental buildings are not representations of the past as it was. They are testbeds for possible solutions to problems posed by incomplete archaeology: structures created by the joining of dots, to be evaluated for performance, longevity, and residue. In this, we stay true to the thinking and practice of Peter Reynolds, as we look forward to celebrating 50 years of Butser Ancient Farm in 2022 (watch this space for a future feature marking Butser’s anniversary).
Further reading
P Andrews (2006) ‘A Middle to Late Bronze Age Settlement at Dunch Hill, Tidworth’, Wiltshire Archaeological and Natural History Magazine 99: 51-78.
R E Pope (2008) ‘Roundhouses: 3,000 years of prehistoric design’, Current Archaeology 222: 14-21.
R E Pope (2015) ‘Bronze Age architectural traditions: dates and landscapes’, in F Hunter and I B M Ralston (eds) Scotland in Later Prehistoric Europe (Edinburgh: Society of Antiquaries), pp.159-184.
P J Reynolds (1979) Iron Age Farm, London: British Museum.
For more information on Butser Ancient Farm, see www.butserancientfarm.co.uk.
All images: Harvey Mills Photography.
