05 October, 2009

34. The Curious Case of the Lengyel Longhouses

There are some things that are just wrong, and Lengyel longhouses transcend odd, through strange, into weird; buildings should not taper. Well, not at 7°, they shouldn’t. It's just not a right and proper thing for them to do. At least two sides should be parallel, otherwise your roof will be in some way irregular. It is all very worrying for a structural archaeologist. So why would Neolithic builders, like those who created this example from Biskupin, Poland [left], [1] make a building with such an unusual shape?

In the previous article we looked a LBK longhouse from Elsloo, Netherlands, using a theoretical model to investigate how the builder might have dealt with the inconvenience of using tapering timber in creating a near parallel-sided building. However, slightly later in the Neolithic, archaeological sites with determinedly tapering buildings, known as ‘Lengyel Culture’, became quite widespread in parts of Central Europe.

In this article we will look briefly at this strange phenomenon. Not that I would claim to have resolved the puzzle, but that is the point: it is important to consider and discuss what you don’t really understand equally as thoroughly as the things you do.

Lengyel Culture
This culture, named after Lengyel in Hungary, is seen as the successor or later contemporary to the Linear Pottery Culture [LBK] in many parts of central Europe. These were farming people, growing grain and keeping livestock, mainly cattle, some pigs, and a few goats and sheep. There is still plenty of fish and game in the diet.

The buildings we shall look at come from Poland, on the northern extent of this cultural spread, which is in part recognised by its trapezoidal longhouses. The evidence for these structures usually comprises a continuous bedding trench presumed to have held a stave wall similar to those associated with LBK longhouses. [2]

A simplified plan of The Lengyel Settlement of At Brześć Kujawski [3]

At Brześć Kujawski, a settlement dating from 4500-3900 BC [4], the remains of up to 50 such buildings have been found, many superimposed in a complex archaeological puzzle. Part of the solution is shown on the left. They are associated with the characteristic borrow pits used to produce the daub for the walls. Buildings range in size, and some are associated with workshop features.

The site was home to a stable and long-lived farming community, one of many in the area, which would have had similar architectural needs to the LBK farming settlements. The emphasis on the outer stave wall rather than using posts is a departure from LBK longhouse tradition. LBK longhouses often have stave walls at one end, and even continuous timber walls, but have considerably more internal postholes than are evident on Lengyel sites.

And they don’t usually taper quite as much.

Biskupin Longhouse Theoretical Model
Simple models of a tapering roof based on Biskupin Longhouse

At the outset, there appear to be three main possibilities to resolving the shape of the roof on a tapering structure:

Option A: The pitch of the roof is constant, the wall is level, and so the ridge slopes down at the narrower back.

Option B: The ridge and the wall are level, and so the pitch of the roof increases as the roof narrows towards the back.

Option C: The ridge is level, and the roof pitch is even, but the height of the wall increases towards the narrower back.

Option A has been suggested as a solution, partly supported by ethnographic parallels from the Pacific.[5] However, in theoretical structural archaeology, what people do in Papua New Guinea, an area with a markedly diverse architectural culture, much of it based on the use of bamboo, is not considered cogent to the North European Plain in the fifth millennium BC. Which is not reason to dismiss this form of roof.

The main issue with A is things' not being level, since not just the ridge, but also the other ‘horizontal’ roof timbers would have to slope to stay parallel to it. In a previous article [6], it was shown how heavy snow can double the static load of a roof; throw in dynamic wind loading, and the disadvantages of having a tilted structure probably outweigh any perceived advantage like lower wind resistance.

Option B would also give the builder a headache, in that each pair of rafters would slope at a different rate, giving each ‘frame’ in the roof a different geometry. At least everything in the option is level, which has much to commend it as a structural principle.

My preferred option would be C, or rather should be, as it cleverly incorporated the idea of tapering wall plates, [D, above]. This shows a series of tapering wall plates set on level stave wall sections, creating a continuous sloping wall plate rising to the back.

But there is a problem: it doesn't really work. It is the geometry of the model that does not work, and making a building work goes to the heart of our methodology.

Above / below left: How an oak tree grows [7]

The main problem is the amount of taper. The model tapers at about 12:1 cm/m, or 6:1 cm/m on each side, which is far more than I would expect to see in timber.

Taper in trees is a complex business. It is dependent on growing conditions and species, but I think a 2--3° rate of taper would be as much as I would like to see, The figures for trees like English oak, illustrated above, if modelled as cones, give taper rates in the region of 1--2 cm/m.

It would be strange if timber with excessive taper were all that was available to Neolithic builders, given that silviculture is an essential ingredient to successful buildings. It might be argued for an individual building, where timber was sourced from natural 'wild wood', that irregular timber, probably split from larger trees, had to be used, but in a well-established community the necessary timber would be expected to be grown locally.

Building is also about geometry. That is its secret, the syntax of architectural form. It is how the forces are tamed and brought in equilibrium, and the eye is pleased. This is very much a problem of geometry.
Looking at the geometry of our model [left], what interests us is the value ‘h’, which is the theoretical difference, or rise, in wall height between the front and back. The value of the reduction of width ‘w’ on each side is a constant at about 1.7m.

In a roof with a pitch of 45°, h will equal w. In other words, the wall would have to rise 1.7m to make the rising wall model work. Over the length of the building [L], that corresponds to a rise of 6 cm/m. If we slacken the roof pitch to 37°, this reduces the rise to 4.4 cm/m, and h to 1.25m (which we can calculate from the tangent rule). This is closer to taper in timber, but not close enough. If we increase the roof pitch to 53°, a more realistic angle for thatch, then h will increase to 2.26m, making the problem worse.

So even slackening the roof to a point where only turf might the viable surface, the wall's rising due to taper in the wall plate still does not appear to work.

There is something else about the geometry of our model, something it shares with other Lengyel longhouses: the narrow end is roughly half the width of the wide end, or (Wf =Wb/2). In terms of w, then, the wide end is 4 x w, and the narrow 2 x w.
This is significant.

In terms of the geometry of a traditional longhouse, an arcade half the width of the building supports the roof, i.e., 2 x w wide, its position corresponding to the position of the back wall. An interesting example of a tapering building comes from Grossgartach [B8], Julich-Welldorf [left].[8] It is constructed with posts, and the position of the arcade and ridge piece can be seen clearly running down the centre of the building to the narrow end, where they intersect the tapering wall line.

If these stave wall structures are built with reversed assembly ties supporting the wall plate [right], as suggested for Elsloo 32 [9], then this is the obvious place to support an upper floor in the roof space.
A typical longhouse is about 24’ wide, and by simple geometry, the centre of the roof will be 12’ from the edge, and its [arcade] supports 6’. In a 45° pitched roof the headroom at the arcade plate would be about 6’, so would roughly mark the edge of usable headroom, as well as being a convenient position to define the upper floor.

So this width of 2 x w, or Wb, also marks the limit of usable roof space, offering a different partial solution to our original problem.

If the roof ridge and other timbers remain horizontal, and the wall is reasonably level, at some point near the narrow end it becomes possible to create a clerestory, to allow illumination and ventilation of the upper floor. What happens in the middle of the building would still be problematic; in our posthole example, only half the building tapers distinctly.

So there it is, the curious case of tapering Lengyel longhouses. Hardly a definitive solution to the mystery, but that is not the point; it would have been wrong to ignore these apparent exceptions to the norms of good structural design. These basic considerations, rooted in the geometry of the models inferred from a structure's foundations, are fundamental to this approach to archaeological interpretation. To see these principles apparently defied is worthy of investigation.

While a better solution may present itself in the future, I like my timbers level. Having a pitched roof is problem enough without tilting it in another plane, so I would be prepared to sacrifice my even-pitched roof first. So throw in some form of clerestory -- and that is a close as I can get, without a lot more study.

Local Issues
We have already strayed a long way from our “ . . . especially the archaeology of the lost timber built environment of prehistoric Southern England” mission statement, but given that only a handful of early buildings are known from here, it is hardly surprising. It is one of the reasons theoretical structural archaeology came into being, and explains why I have been looking at continental Neolithic longhouses. In a future post, I will look at what all this tells us about how to look for, and find, Neolithic buildings in native posthole palimpsests.

However, we do have some interesting indirect evidence in the form of Long Barrows, like West Kennet, Wiltshire [left], which are trapezoidal in shape. In Neolithic Britain, these were tapering elongated mounds formed from materials dug from flanking ditches, with a communal tomb at the wider end. This end often featured megalithic stones, and a forecourt with traces of 'activities'. There is seems no reason to avoid the obvious interpretation that the tomb represents a house for the dead.[10]

In a strange episode of serendipity, Martha Murphy, unaware of this article, who was editing my article “Why almost nothing in the Celtic world was square”, wrote to remind me about the importance of cattle in the Neolithic, and mentioned that she had read that the tapering structures might in some way be symbolic of cattle.[11]

For the structural archaeologist this presents problems, since it implies a level of architecture above and beyond the call of duty. As a rule, the stance so far in TSA is that symbolism is secondary to structural form. Subsequently, those aspects of a form that acquire particular meaning may be emphasised to enhance a building's symbolic value, but not to the extent where it undermines the structural stability of the original form.

One reason this is difficult, and to be avoided, is that we can read a standing building with its surface decoration intact with some degree of confidence, but this level of preservation is exceedingly rare in prehistory. We must never forget that we are dealing with foundations, which is why it is ‘theoretical’. While we can define some rules for modeling roofed space from its foundations, we must not overreach ourselves. To go too far beyond this, to consider the symbolism and perception of, the spaces we can define, requires a different knowledge base and approach.

However, utility is another matter, and in the next article we are returning to Poland to look at what may be the earliest Neolithic cart shed ever discovered.
Sources and further reading:
[1] Bogucki, P. 1988. Forest Farmers and Stockherders. Early Agriculture and its Consequences in North-Central Europe. Cambridge: Cambridge University Press. p70, fig. 4.9 [Taken from: Grygiel, R. 1986. The household cluster as a fundamental social unit of the Lengyel Culture in the Polish Lowlands. Prace i Materialy Muzeum Archeologicznego i Etnograficznego w Lodzi 31: 43-334.]
[2] Bogucki, P., The Neolithic Mosaic on the North European Plain. School of Engineering and Applied Science: Princeton University. http://www.princeton.edu/~bogucki/mosaic.html [Accessed 01/10/2009]
Also: Czerniak, L., Raczkowski, W, & Sosnowski,W., 2003. New prospects for the study of Early Neolithic longhouses in the Polish Lowlands. Antiquity Vol 77 No 297 September 2003.

[3] Bogucki, P. 1988. Forest Farmers and Stockherders. Early Agriculture and its Consequences in North-Central Europe. Cambridge: Cambridge University Press. p67, fig4.8 [Taken from Grygiel, R. & P. Bogucki. 1997. Early Farmers in North-Central Europe: 1989-1994 Excavations at Oslonki, Poland. Journal of Field Archaeology 24: 161-178]
[4] Bogucki, P., 1991, Changing Neolithic Landscapes at Brzesc Kujawski, Poland.
Paper presented at the annual meeting of the Society for American Archaeology, New Orleans, Louisiana, April 1991. http://www.princeton.edu/~bogucki/landcape.html [Accessed 01/10/2009]
http://antiquity.ac.uk/projGall/czerniak/czerniak.html [Accessed 01/10/2009]

[5] Marshall, A., 1981. Environmental Adaptation and Structural Design in Axially-Pitched Longhouses from Neolithic Europe. World Archaeology, Vol. 13, No. 1, Miscellany (Jun., 1981), 101-121 . Stable URL: http://www.jstor.org/stable/124216
[6] http://structuralarchaeology.blogspot.com/2009/02/22-iron-age-graphs-important-discovery.html
[7] James. 1989. Forester's Companion. Cambridge University Press. p 355
[8] Whittle, A. W. R., 1988. Problems in Neolithic archaeology. Cambridge. Fig. 3.7, p46.
[9] http://structuralarchaeology.blogspot.com/2009/08/33-elsloo-32-neolithic-longhouse-made.html
[10] For example see: Hodder, I., 1984. Burials, houses, women and men in the European Neolithic. In: Miller, D., Tilly, C., eds. Ideology, Power and Prehistory. Cambridge.
[11] Quite possibly: Darvill, T., 2004. Long Barrows of the Cotswolds and Surrounding Areas, Tempus 2004