21. Thrust, trusses, and not going down the aisle

Much of our experience of height and depth, the z axis of the space we inhabit, comes from the built environment, which has long since outstripped things like trees as tall objects in our lives. By contrast, in prehistoric Britain, when buildings were made of wood, it was difficult to create structures taller than the trees they were made of, and builders got on with the practical problem of creating width. This is one of the perennial problems of architecture, and an important thread in the narrative of the built environment. In this article we shall look at how builders solved the technical problems inherent in wide timber buildings, considering medieval buildings at the end of this process, and how this can be related to the archaeology of Neolithic buildings at the beginning.

However, British archaeology won’t be joining us; it left on a vacation to Africa, seeking its inspiration in the warmer climes of the colonies, from where it has never really returned. Thus, the built environment prior to the arrival of the Romans has been imagined as circular mud huts, which, like those of Africans, are of little relevance, since ‘architecture’ clearly arises from the superior civilisation of the colonising power -- an approach to archaeology that owes more to imperialism that empiricism.

Imagining a 3-dimensional past can be difficult from an archaeological space that is invariably shallow. Everything above ground has long since rotted away, and even the foundations have usually been ploughed over. Several thousand years of architectural heritage has been reduced to shallow features in the subsoil. The only possible understanding of this evidence is structural; these are building foundations, and only make sense in terms of the superstructure they supported.

The last article looked at how foundations work -- no maths and no long words. I think I got away with it, so now we are going to look at how buildings work.

Structural basics: A: a rigid triangle. B: a less rigid square, is subject to C: racking, and requires D: bracing to prevent this.

Imagine nailing together pieces of wood of equal length. The simplest and strongest shape you can make is a triangle. In fact, it is perfectly rigid. However, if you make a square, you will require extra nails to make the shape rigid. A square shape relies on the rigidity of its four corners; otherwise it will ‘rack’. ‘Racking’ can best be prevented by either very rigid joints, or by adding a ‘brace’ between two opposite corners, thereby ‘bracing’ the square to create two triangles.

Bracing is probably the least viable structural element from an archaeological point of view, but it is an important part of what makes buildings work. It is also worth making the point that to create a tenon joint between two pieces of timber, perhaps 50% of the wood is removed from each component, reducing the strength of the components accordingly. A joint is also a source of physical weakness, as well being potentially less rigid than a straight timber.

How thrust is created, and how it is constrained using a tie beam

The next very important concept to grasp is ‘thrust’. Imagine a simple roof as a folded sheet of card stood on your desk. When you push down on it, simulating the weight of the roof, the edges of the card roof will push outwards. It is this outward force that is known as ‘thrust’, and it will tend to push the structure supporting the roof outwards. This is an important consideration in roof design.

If you are particularly stupid or unconvinced, or perhaps wished to make a lasting impression on a child, you could repeat the experiment, using the inverted blades of a pair of scissors placed on the palm of your hand to represent the roof. Press down, and, if you examine the resulting wound, you will find that the blade entered your flesh at an angle representing the resolution of the downward and outward force.

As a building becomes wider, the weight of the roof increases, and so does the thrust that has to be constrained. The answer is to turn the shape of the roof into a triangle by adding a horizontal timber between the rafters known as a ‘tie’ beam, or as a ‘collar’ if it is nearer the apex.

The mass of the roof squeezes or compresses the posts and rafters supporting it, but the ‘thrust’ stretches, or causes ‘tension’ in the tie beam. You can snap a straight wooden stick by bending it; this is its principle weakness. It is much harder to break it by stretching it or squeezing it. It is this ‘tensile’ and ‘compression’ strength that makes timber like oak such a good building material.

As a roof becomes larger, both the weight and the thrust increases

Having a tie beam at the base of the roof allows the builder to create wider, heavier roofs. But as the roof becomes wider, the tie beam running across the roof will be too long to support its own weight and additional posts will be required to support it, as well as the extra mass of the larger roof.

A simple building with ties, and a larger aisled structure

One common solution was ‘aisled’ buildings, with extra sets of supporting posts on either side called ‘arcades’; the ‘aisles’ being the spaces between the side of the building and the arcade. Any bride down an aisle is in fact in something of a blind alley.

The Plan of House 32 Elsloo, Netherlands: a Neolithic longhouse [1]

Part of the Neolithic solution to preventing thrust involved supporting the ridge of roof directly with a series of posts under a ridge piece. This was in addition to two lines of arcade posts. The roof was well supported, but the interior of a longhouse was a post-rich environment. A Neolithic supermarket would have had very narrow aisles.

A theoretical Neolithic building section

~.~

“The ridgepole sags to the breaking point.

It furthers one to have somewhere to go. Success “[Ta Kuo / Preponderance of the Great, hexagram 28, I Ching, ~ The Book of Changes, China – 1st millennium bce]

~.~

The positioning of posts, and hence postholes, will intimately reflect the nature of the superstructure it supports. Clearly the posts are the first thing to be put up, but whether you then put the transverse or the longitudinal timbers up next is significant, and is known as the 'order of assembly’.

A: Normal assembly; B: Reversed assembly; C: Reversed assembly with offset posts

‘Reversed Assembly’ is the next important idea to master. If you look at your roof, you may find the ‘tie’ of the wooden roof truss set on your main walls, because there is nowhere else for it to go. This is ‘normal’ assembly. However, in a timber building, without load bearing walls, you can put a tie under the beam supporting the roof structure: this is ‘reversed’ assembly. In this form of construction, the ties sit on top of the posts, with the roof, or wall-plate, on top of it. While this is perhaps a minor technical issue in what is, after all, an entirely theoretical construct, it has big implications for the positioning of the posts, and hence the archaeological footprint of a structure.

The effect of assembly on the archaeological footprint of a theoretical building

In a ‘normal’ assembly structure, the posts support the wall / roof plate and will be in a straight line under that timber. In a ‘reverse’ assembly, there is an option of offsetting the post along the tie it supports, so it is no longer directly under the edge of the roof. This has four effects:

• Moves post away from the outside of the building.
• Simplifies the jointing of the structure.
• Creates a ‘cantilever’ effect, helping to support the tie.
• Creates something else for a theoretical structural archaeologist to worry about.

This last point is not insignificant, since it follows that the posts of a reverse assembly structure will not necessarily form a straight line along the axis of the roof, since they support timbers running across the building. While this could make identifying buildings more difficult, it can also, in theory, tell you about the nature of the superstructure.

Reconstruction of Shang Dynasty timber house, showing the use of reversed assembly (China, later 2nd millennium BCE) [2]

The traditional timber architecture of China has preserved ‘reversed assembly’ because the roof and its ties are still carried on posts, rigid load-bearing walls having proved more liable to fail in earthquakes. Another significant difference from European traditions is that Chinese timber buildings do not rely on triangular bracing to create rigidity, but use instead elaborately jointed corbelled brackets known ‘dougongs.’

Sections of traditional Chinese timber roofs, showing reverse assembly and dougongs

In Europe, rigid load-bearing walls had certainly started to appear by the Iron Age. Transferring some, or all, of the roof load to the walls creates less need for the posts that previously cluttered up the interior of buildings. It is solving the problems of creating a wide and open internal space that lies at the heart of the development of timber architecture, in terms of both the carpentry of joining wood together and how frames could be assembled and supported. We still have thousands of medieval buildings which representing the tail end of this 5,000 year story, so we know how it turned out.

Types of roof: A: Hammer beam, B: Cruck, C: King post truss

Medieval timber roofs are categorised by the type of truss used, usually named after the rich vocabulary of its component parts. Before the advent of dendrochronology, the type of roofing technology was often the principle indicator of date.

There were several different approaches used. One common and successful type was the ‘hammer beam’ roof, an elegant and complex solution for transferring the roof load to the walls.

A ‘cruck’ roof utilised a matched pair of curving timbers, joined at the top to form and support the roof, offering an entirely different solution.

The aptly named ‘King post’ truss came into use in Southern England by C.14th and proved to be ultimate technical solution to the problem of roof thrust, with a single king post which joins the apex of the roof to the centre of the tie. However, although it appears, intuitively, that the king post supports the apex of the roof from the tie beam, if you have been paying attention, you will realise it is the apex that is supporting the tie beam, and that the ‘post’ could be replaced with a metal rod, since it is under tension, not compression.

The king post truss is an elegant and simple solution that evolved from more complex solutions. From thousand of years of experience builders learnt that supporting the apex directly is unnecessary, and they slowly reduced the number of posts in the interior as experience refined what was felt to be safe.

The C13th Barley Barn at Cressing Temple, Essex [3]

If Neolithic longhouses like Elsloo represent the beginning of the story, the beginning of the end is marked by the great medieval barns like the Barley Barn at Cressing Temple in Essex, built from 489 oaks for the Knights Templar at the beginning of the 13th century.

Westminster Hall, London [4]

However, if English timber roof building has a ‘apex’, it is the truly extraordinary Westminster Hall roof, (1394 –1402), built by Henry Yevele, the architect/builder, and Hugh Herland, the carpenter/designer. They replaced the aisled roof structure with a single 68’(20.72m) span hammer beam roof, using 1m thick hammer beams projecting horizontally 20’ to narrow the span and transfer the roof load to the masonry walls.

Comparative building sections A: Elsloo, Netherlands [reconstructed]; B: Cressing Temple Barley Barn; C: Westminster Hall

There has been a tendency among archaeologists and reconstructers to assume prehistoric buildings and building technology to be simple and minimal. However, this is probably a reflection of their own lack of resources in terms of materials and culture. The lesson here is that prehistoric buildings were over-designed, and while width is difficult to achieve, it is still possible. The widest late Neolithic buildings I have found were between 40 & 50 feet (12-15m) wide. It is just that this could not be achieved without cluttering the interior with posts.

However, what ultimately restricts roof width and pitch is the length of timber available to form ties and rafters. Height, in terms of posts, is subject to same limitation. Precisely what this might represents, in terms of English oak, will be discussed a later article.

Sources & Further Reading:
[1] Startin, W. (1978). "Linear Pottery Culture Houses: Reconstruction and Manpower." Proceedings of the Prehistoric Society 44: 143-59
[2] Clark G & Piggott S 1968, Prehistoric Societies, London, Hutchinson, fig. 85, p.273.
[3] http://www.cressingtemple.org.uk/
[4] http://www.parliament.uk/about/history/westminsterhall.cfm