Stonehenge and the archaeology of the prehistoric roof

Postholes: cult or craft?

Most of the prehistoric archaeological sites of Britain, even Stonehenge, are covered with postholes, and it is my central contention that these posts were primarily the foundations of timber buildings. Further, since building is a rational process, postholes can be understood in terms of how posts were joined together to create roofed space.

Unfortunately, many academics have convinced themselves that some postholes are the product of a mysterious cult, whose rituals involved placing posts in the ground.

Despite considerable and ongoing research in this cult of postholes and their cosmology, the reasons for this strange behaviour are still not entirely clear. However, unlike the anthropomorphic polytheist religions evident elsewhere in Europe, adherents of these rituals, with remarkable prescience, seemed mainly to be acting out key themes from modern anthropology.

So it boils down to whether postholes are a product of rational process or ceremonial ritual -- or me, and possibly you, against mainstream academic archaeology. It is no real contest, because I, or we, can ‘prove’ a building by using geometry to understand how it worked.

Please remember, we are the good guys; this strange cult has stolen the Bronze Age peoples’ built environment, so they have nowhere to live, and are forced outdoors to do bizarre rituals in all weathers. But we will be giving back our ancestors their buildings, and perhaps some respect.

The craft

When, seven thousand years ago, the first LBK farmers moved into Northern Europe, they brought with them the woodland practices [silviculture] to grow the timber from which their material culture was largely fabricated. They also brought the specialist skills needed create the built environment on which their culture and economy was dependent.

Turning the trunks of trees into the mechanical structure of a roofed space is a very specialised skill. Builders /architects / masons create the largest and most culturally significant artefacts: they create space and security.

Architecture can provide both utilitarian and abstract expressions of power and belief. If you want to understand a society, a good starting point is their built environment. Ancient buildings and ruins are a main focus of our engagement with the past, although mainly with those cultures with the foresight to build in stone.

The beautiful church at Borgund, Norway, is built entirely from timber, and has no metal fixings, and it is about eight hundred years old. [above] [1] It is quite a small building, but has a complex multilevel roof with some splendid decorative flourishes. It is an example of local craftsmanship and materials, and part of an accomplished ancient tradition of building in timber.

It is one of a group of surviving wooden churches, with walls built from staves, [interlocking vertical timbers]. It is also part of a wider Christian tradition of church building, with its own range and ancestry. In this broader scale it is not a significant building; it is after-all, only a church, not a cathedral.

Borgund Church interior [2]

It has a roof supported on timber posts, but these do not appear to dominate space. Notice how the 45° bracing of the structure is blended with other horizontal and vertical timbers. These details embody a deeper truth: if it was not built with such massive posts, trusses, braces, and ties, it would not have survived its first Norwegian winter.

It was built between AD 1180 and 1250, and survived in part because its posts were set into sill beams, so they lasted longer than those of earlier generations of timber buildings with earthfast posts.

Questing the invisible

If Borgund Church had been built with earthfast posts, once they had decayed and the ground had been ploughed, it would have left a plan something like that to the left.

From this plan, it would be possible to make some assumptions about the shape of the roof and the overall form of the building, but not about the quality of the craftsmanship.

From its context, it would be clear it was Christian, but these postholes will tell you nothing about the important beliefs of the congregation: they will shed no light on transubstantiation or original sin. There may be evidence of repairs and restorations, but not necessarily of the Reformation. If this building had been built a few centuries earlier, it might have been a ‘pagan’ building plan.

However, a modern [post-processual] archaeologist would not necessarily even see this as a building; they might attempt explain this plan in terms of the beliefs and cosmology of the people who dug the postholes. You are expected to believe prehistorians can see into minds of people in the past, and can use this insight to explain why they created postholes.

I think this is a problem, but it’s not mine; I simply have to prove it is a building.

So how do we even know it’s a building just from its foundations?

Well, it’s fairly simple, but with few complicated bits., and I have broken it down (somewhat artificially) to:

Proportions – Is it laid out like a building?

Geometry – Would the timbers linking the posts be parallel?

Assembly – Do the components fit together?

For the purposes of TSA, a building is a roofed structure: A roof relies on geometry to distribute and constrain the static forces it generates, as well as the dynamic loads caused by the climate and its occupants. For this reason, roofs are usually symmetrical.

What we are looking for is symmetry, and starting with the principal vertical components, the posts, we try to build an accurate ‘on paper’ model of the roof and its supporting structure in plan and section. The foundations of the supporting structure will reflect the geometry of the roof, but how directly depends on how precisely the components are put together, or assembled, at the top of the posts.

An LBK longhouse from Elsoo, Netherlands [3]

Although the posthole foundations of a prehistoric timber building may appear irregular, or asymmetric, this is to facilitate a symmetrical rafter pair, the main timbers forming the roof. The rafter pair is not directly supported by the posts, but on horizontal timbers, and so the postholes actually mark where the rafter is not.

The theoretical model of Elsloo 32 [above], has offset ties, allowing a symmetrical rafter pair. It is also laid out so that the ties and 45° bracing would trap the internal posts, increasing the rigidity of the structure. Part of the building has a stave wall, indicated by the trench without postholes.

It is essential to understand the nature of the structure's assembly, or how, and in what order, the pieces of wood fit together to create the timber structure of the roof.

Profane Geometry

A longhouse roof has a nominal axis running down the central ridge, with rafter pairs at right angles to it. There is a limited range of variations where rafters may run parallel to the ridge, creating a valley were the two roof surfaces meet.

Basic roof forms

Many prehistoric roofs were either circular or annular in form, so that rafter pairs and supporting beams radiate on many different axes. While small round buildings may seem simple, their more complex geometry creates problems when you scale them up. This challenges preconceptions that ancient buildings should be ‘simple’, and these large circular roofs would be considerably more complex to construct than rectilinear ones.

Their builders had to contend not just with more complex geometry, but with roof surfaces that widened towards the base. In the case of annular roofs, the inside slope narrows towards the base, while outer slope gets wider. This results in a much greater mass on the outside of the roof, although roofing the central area with some form of cone might serve to counter this imbalance.

However, conical roofs become problematic when we scale them up; their surface area increases disproportionately with their diameter, so that a 20m cone can weigh four times as much as a 10m one. So there are theoretical limits to the how big a conical roof it is safe to built, but precisely where that limit is, is another matter. Iron Age roundhouses don’t come bigger than 17-18m, and the average is around 10m.

The bigger the building becomes, the greater the forces it generates, and the stronger and more robust the structure itself has to become. Once you have big buildings, especially circular ones, it is unavoidable that engineering solutions become more complex. So now for the complicated bit.

Interlace Theory

Circular and annular structures are not simple to conceive or build. For one thing, the wood is straight, so the building is always some form of polygon, not a true circle. I have indicated the complexities involved in building roofs of this size, but to model how it was actually achieved on a technical level, I had to devise interlace theory.

In an annular structure the roof is formed by rafter pairs, arranged radially, supported on horizntal timbers at right angles, which in turn are supported by supported by posts. Typically, each rafter of the pair will be supported at the ridge, at its base, and at some point roughly midway between the two. So each rafter pair must have at least five parallel horizontal elements supporting it.

In a structure based on a series of concentric rings of posts, the multiple radial angles of rafters and their parallel supporting structures can be creating by joining the posts in a ring together in different ratios, so that posts are linked 1:1, and also 1:2, 1:3 and even 1:4.

So in this case, each post in the ring could have eight elements attached to it. Each of these eight elements must joint the post at a different height, and not cut the path of neighbouring elements. As a result, each polygon occupies its own level, or horizontal space, between the posts.

As the model of the Woodhenge ridge piece (Ring C ) illustrates, it supports a series of interlocking polygons which form a conical stack of horizontal elements. The longer 1:4 ratio elements (C4) are lower in the stack and interlock with the next post ring [D].

This process is repeated on each ring, creating a multilevel series of interlocked polygons, further strengthened by ties at the arcade and roof plate level.

You can’t get the wood . . .

It is important to understand that trees are grown in managed conditions to optimise the production of the timber required. This does not preclude harvesting ‘wild wood’, but managed woodland has great logistical benefits, providing what is required, where it is needed, and on a predicable basis

In a previous article I explained that a nominal young oak could yield a 50’ tapering stem, which is neither too thick at one end, nor too thin at the other, to be jointed. In addition to their use as rafters, these trees provide the ties that run across the base of the roof, and thus also limit the scale of roofs to a nominal 50’. [below]

This is why the inner ring and outer ring of a timber circle are this distance apart, and why the posts of the inner circles are smaller that those of the outer.

The ring midway between the two has the largest and deepest postholes, marking the taller posts supporting central ridge of the roof. These posts may well be in excess of 50’ high, since it is only at the top, where they join the ridge pieces, that their diameter is important, and thus, taller, larger trees may be available.

The centre is usually marked by small diameter posts, indicating the narrow end of the ties; so, following the same structural logic, we can suppose that the central roof structure will be built from small diameter timber, but in long lengths.

When I started to model interlace structures like Woodhenge, there seemed to be a requirement for 60’- 65 lengths of wood. This seriously challenged my assumption about 50 ‘ ties and rafters.

The technical solution is to use wood split from larger trunks, in particular timber's ‘boxed heart’, where the outer parts of a large trunk are spit off to leave a baulk of heartwood.

What length of timbers was available in this period depends on what types of trunk could be harvested. The issue is what height the tree reaches before it starts to branch and develop a crown, which in turn depends on the nature of the woodland in which it is growing.

Trees such as oak, growing in ancient forests and wildwood, with many generations of light competition, may be 40m high, and certainly 30m is not considered tall for this type of tree. So, in essence, we are looking for straight lengths of timber, split from a trunk that is 30–50 % of the total height of the tree; perhaps not common, but certainly possible.

The timbers used in a stave wall were usually manufactured by splitting, as were boards, and the larger rectangular baulks used for sill beams.

Postholes are usually round because nothing is gained by reducing the timber, but above ground a post can be squared. Rectangular [post]holes can reflect the use of squared timber, and are not necessarily for stone. I have modelled those at Woodhenge as marking the position of the stairs, one context where very wide baulks might be used as a foundation.

It is self evident that a timber building's first weakness is its earthfast posts, so it is not uncommon to find postholes that have been recut, which may indicate rebuilding and refurbishing.

Understanding the plans of ancient roofed spaces

What are we looking for in our plan of a building?

1] Proportion: in the structure as a whole, and symmetry in the disposition of post rings.

2] Geometry: a series of parallel components at different heights between the posts, which will allow the path of a symmetrical rafter pair to be plotted.

3] Assembly: between the inner and outer posts, ties and braces should pass close enough to internal posts to be supported at least at one point.

Examples of these basic aspects will be illustrated using archaeological plans from sites that have been mentioned in previous articles. This article is not about reconstruction, but about demonstrating spatial relationships; the evidence is in the form of plan and section drawings, and these convey all the critical information.

The Sanctuary

The Sanctuary is a prehistoric site on Overton Hill near Marlborough, Wiltshire. The site was first dug by Maud and Ben Cunnington in 1930. One hundred sixty-two postholes were excavated, including some with doubles, and some with postpipes.

Just so as we understand what is at stake here, this is what the most recent investigation concluded about this structure:

"This was not a building like a cathedral, architectural styles succeeding one another over centuries. It was not a monument at all - it was a process.

The important thing was the ceremony, the activity. Going out into the wildwood to find the trees, felling them, dragging them to site, carving them, digging the pits, hauling the things into place. And this process was so important that to allow it to continue, the posts had to be removed and the pits filled. Then the cycle could begin again."

Mike Pitts [4]

The plan of this structure is much discussed, largely because of the complexity of the postholes in rings C & D. For the same reason, I am reluctant to advance too prescriptive a model of the building and its geometry, as a small change in the position of a post, or the presence of more than one post sharing a hole, has serious implications for accurate modelling. This is an important consideration; a model can only be as accurate as the archaeological plan.

All the rings are sufficiently concentric and spatially related, with similar geometry, to suggest that it had one principle structural phase, perhaps with alterations and repairs.

At about 20m, this site is just small enough to be modeled as a cone, although it is laid out like an annular structure, with its largest postholes in its middle ring, C. The model shows ties running between the outer ring and the small posts in ring F, and others radiating from the post in the centre.

So, while it shares many features with other buildings of its period, it probably had a very different form of roof, which I suspect was quite tall. The building is clearly not domestic, in that, like Stonehenge, it contains a stone circle, which fits neatly between the posts of the second circle.

Mount Pleasant

Mount Pleasant, Dorsett, like many of these buildings, is set within a large enclosure, and immediately surrounded by its own ditch.[5]

This is a remarkably regular ground plan, with a strong geometry defining its principal axes. The overall proportions and geometry of the post rings work perfectly well, and it is modeled as an annular roof, probably with projecting radial ridges or clerestories on its main axes. The double alignments on these axes would allow one rafter pair above another, and suggests the presence of some form of clerestory windows.

The ties that are illustrated weave trough the inner posts so that most could be supported by more than one post. This pattern of ties also suggests that the roof was multilevel, and hints at the overall complexity of the whole structure, where ties at multiple angles would radiate like the spokes on a bicycle wheel.

The square setting of large posts in the centre of the structure suggests that, in this case, a simple pyramid roof could cover the central area, on the inside of the annular roof.

Navan Fort

This 40m in diameter circular timber structure excavated at Navan Fort in County Armagh has been dated by dendrochronology to 95BC. It was later destroyed by fire, and covered by a mound.[7]

Navan is a protohistoric site, in that it can be identified as Emain Macha, ancient capital of the kings of Ulster, known from Iron Age aural tradition like the Ulster Cycle, which was first written down and preserved in the Early Christian period. From his great halls at Emain Macha, King Conchobor ruled his kingdom, with his warrior troop, the Red Branch Knights, and it was here that the young hero Cú Chulainn came to be trained. One tradition, recorded in the Annals of the Four Masters, was that Emain Macha burned down after the defeat of the king Fergus Foga of Ulster at the battle at Achadh Leithdheirg in AD 331. [7]

The model illustrates an annular roof, and a slightly higher radial roof running E-W from a central post. I interpret the long slot, sloping down towards this posthole, as the foundation of a set of stairs rising to the centre of the building.

The wall is interpreted as being composed of squared timber buttresses, linked by a double skin stave wall. The space between these was probably partly filled with earth or rubble. The remarkably massive construction of its outer wall may indicate a military function. It also distinguishes Navan from earlier buildings, although the symmetrical layout of its post rings is remarkably similar to Mount Pleasant, which is probably 2,000 years earlier. This only serves to emphasise that we a dealing with technology for supporting roofs.

Stonehenge

It is with some reluctance that I turn to Stonehenge. It is a unique and exceptional monument, and it carries a considerable weight of romantic and scholastic preconceptions. It has strong visual culture reflecting its current state, and to suggest it was once a wooden building seems counter intuitive.

Of more concern is the plan, which is not entirely complete or accurate. The numerous postholes are seldom shown on plans of Stonehenge, prompting very little comment. My interpretation is based on the position and disposition of the features known as Y, Z, Q, and R holes, and it should be stressed that there is absolutely no real evidence that these ever contained stones.

In essence, the building has an outer roof plate [Y holes], a ridge piece, [Z holes], and the inner roof plate is represented by a double ring of postholes [Q & R holes]. There is no arcade of posts; in its stead there is a circular post and lintel stone wall, on the inside of the ridge.

This could support ties running across the base of the roof, which in turn, by means of queen posts, could support the arcade plate.

Ties can still be plotted passing next to the ridge posts, ensuring a viable span.

The double posts [Q&R] are assumed to be at different heights, facilitating a multilevel roof, particularly near the entrance where four lines of posts are present. One or both of these post rings probably contributed to the roofing of the centre of the building.

While this could have been an open court, the five sets of trilithons could have equally well served to support ties cutting across part [chord] of the circle defined by the Q & R holes. These ties could form the basis of a roof, but this is unlikely to be a simple cone because of the size of the central area.

Thus Stonehenge presents certain technical problems, principal of which is the sheer scale of the area to be roofed in the middle. The general proportions would allow the builder to use another annular roof, and then fill its centre with a cone.

However, since TSA is not about speculation, I am happy to say I don’t know how, or even if this was done.

There is no reason why this building should not work, even without an arcade, but it would have been surprising, audacious, and unique. It is usually significant, but perhaps not unexpected, to find structures that push the technical limits in terms of scale. In architecture this is a perfectly common aspiration of both patron and craftsman. However, combining wood and stone in this way, each with its own foundation, is perhaps a little unwise, as there is a significant difference in the way these two materials react to changes in their environment.

In terms of the stonework itself, the singular point of interest is why this type of engineering, unremarkable in the Mediterranean, should turn up in southern England, and be integrated into a timber structure. Since the these annular buildings are an existing local tradition, it is best explained by ‘imported craftsmen’, perhaps even slaves. This is a perfectly well-documented practice elsewhere in this period, and may representing a single unique episode or event of gift exchange or commissioning.

Once you accept that Stonehenge was a building, in which the sarsen ring and trilithons are just important components, then it is clear that the Bluestones were the content. Then Stonehenge may have been a temple that was built to house and control access to the Bluestones.

Stone circles, with widely-spaced open rings of stones, were part of an older tradition, and, like long barrows, in some senses communal. Here the mystery is enclosed, sealed off, and perhaps even imprisoned, behind a wall of stone and wooden posts.

Reason or Ritual?

Considerable intellectual endeavour and research funding has been channelled into exploring the mysterious beliefs, cosmology, and rituals of the posthole-digging cult and their mysterious activities in the landscape. However, it would be a very odd world without a built environment, and that’s probably why British prehistory has become a very strange place.

Meanwhile, back in the Bronze Age, nobody has anywhere to live, even though Southern England is the principle source of European tin, and controlled the southern route to the copper mining areas of the Irish sea. However, once you accept postholes as foundations, and these structures as buildings, we are well on the way to having a built environment, or at least something fit for the purposes of those at the top of society.

All these ground plans pass the three tests to prove they were roofed. They have the correct proportions and symmetry for buildings, which gives rise to viable roof geometry, especially when Interlace Theory is considered. The satisfactory and conclusive evidence is the assembly, the way the ties fit between the posts in buildings like Mount Pleasant.

They are all in their way remarkable buildings, built at the limit of what was possible with the materials and tools available.

Stonehenge is an exception-- it is always an exception. But does it prove the rule? It has a load-bearing circular wall, but no arcade. However, in all other aspects it conforms to the tests of its being roofed. I am happy that Stonehenge was a temple, with an outer roofed area, and if we understand the sarsens as a wall round a more sacred space, perhaps further divided by the trilithons, this would indicate a three-fold division of the space, a common pattern for temples.

As for the stonework, Mediterranean contacts are not an issue in this period. Similar high status objects appear in the elite graves of both Greece and Wessex. None of this is unusual or unexpected, and it is simple to accept that southern England was part of a wider political and economic world.

In practically every class of artefact where we have evidence, there are examples produced to the highest possible technical standards, using the best materials available. Building in English oak, one of finest construction timbers in the world, is a vital part of the cultural package that makes an agricultural economy possible in our climate.

Timber was produced from managed woodland, and this was used in conjunction with other local materials to produce the built environment society required to house and protect its culture and resources. Postholes are the foundations of the load-bearing posts most commonly used to support roofs. They are the product of a rational and comprehensible activity, which has historical continuity, and a well-established craft tradition.

Sources & further reading

[1] http://www.visitnorway.com/en/Product/?pid=47214 [last accessed 21/6/11]

Exterior photo: http://upload.wikimedia.org/wikipedia/commons/1/1b/Borgundstavechurch.JPG [last accessed 21/6/11]

Section & plan: http://www.greatbuildings.com/cgi-bin/gbc-drawing.cgi/Norwegian_Stave_Church.html/Stave_Church_Section_A.html [last accessed 21/6/11] After Gunnar Bugge. Stave Churches in Norway. Dreyers Forlag A/S, 1983. ISBN 82-09-01929-5. NA 5761.B84 1983. Discussion p5-8. elevation drawing, p67. Transverse section drawing, p66. Longitudinal section drawing, p67. Plan drawing, p67.

[2] Interior photo: http://en.wikipedia.org/wiki/File:Stave_church_Borgund_interior.jpg [last accessed 21/6/11]

[3] PJR Modderman (1970) 'Linearbandkeramik aus Elsloo und Stein 2.' Tafelband, Leiden Univ., Faculty of Archaeology.
PJR Modderman (1975) 'Elsloo, a Neolithic farming community in the Netherlands,' in Bruce-Mitford, R L S, Recent archaeological excavations in Europe, Chapter IX.
PJR Modderman (1985), D'ie Bandkeramik im Graetheidegebiet, Niederländisch-Limburg.' Berichte der Römisch- Germanischen Kommission, 66::25-121.
See also: W. Startin (1978), 'Linear Pottery Culture Houses: Reconstruction and manpower.' PPS 44, 1, pp. 43-159.

[4] http://www.cvni.org/treenursery/tree_heights/tree_heights.html [last accessed 21/6/11]

[5] http://www.britarch.ac.uk/ba/ba51/ba51feat.html [last accessed 21/6/11] ISSN 1357-4442 Editor: Simon Denison
Mike Pitts. Return to the Sanctuary. Issue no 51, February 2000 http://en.wikipedia.org/wiki/The_Sanctuary

[5] http://www.pastscape.org/hob.aspx?hob_id=453935&sort=4&search=all&criteria=mount%20pleasant%20henge&rational=q&recordsperpage=10 [last accessed 21/6/11] [6] www.doeni.gov.uk/niea/navanforteng.pdfhttp://en.wikipedia.org/wiki/Navan_Fort