So second time around, and now in colour, I give you Systematic Irregularity, a phenomenon which can be observed in many aspects of the prehistoric Celtic world, which is a problem, as I can’t show you everything that isn't square; but you can trust me – I’m an archaeologist.
We are all familiar with the restless flowing curves characteristic of Celtic decoration, particularly the Iron Age La Tène Style, exemplified by pieces like the Battersea Shield [left], not a place you would really expect an unnatural shape like a square; however, it goes much deeper than this.
I first noticed there was something was odd about Celtic enclosures when I was writing up the excavation of the late Iron Age site at Orsett ‘Cock’. It had two phases of overlapping enclosures, surrounded by as double ditch system enclosing both. The enclosures were roughly the same slightly irregular shape, with two parallel sides and two right angles: technically trapezoidal, almost square, but with one corner misplaced.
This concerned me. Why did both enclosures have a similarly irregular shape? Clearly, ditches may have to respect topography, existing structures, and even trees, and you would naturally expect hillforts to be hill shaped, but Orsett was a gravel terrace.
As I looked at other sites, it became clear that trapezoidal (2 parallel sides), usually with 2 right angles, enclosures were quite common, and that some even had 2 sets of parallel sides and no right angles, like a pushed over rectangle or square, a parallelogram or a rhombus respectively (more rich pickings for the pub quiz). Everywhere I looked, nothing was quite square, regardless of scale, not even quite small enclosures such as that surrounding an important building within the Hillfort at Hod Hill.
For an archaeologist concerned with understanding the built environment, this had several important implications, not least for finding and identifying buildings. Faced with the large number of postholes sites often produce, the first thing that most archaeologists will look for is circular structures and small square ones. As I have discussed previously, the most important factor associating postholes of the same building is depth, as the Little Woodbury examples perfectly demonstrate, but this is not widely appreciated, and structures are often identified by overall shape alone.
Some excavators have even used computers to find ‘square’ structures and associated postholes. This was tried at the important Iron Age fort at Hengistbury Head, with the computer looking for squares with an accuracy of 150mm, and while posthole size was also considered, the postholes' depth, as such, was not. Systematic irregularity suggests the computer would better have looked for structures with an ‘inaccuracy’ of about 300-400mm. In the end, the actual results were generally discounted, confidence in the computer being somewhat undermined by its habit of using the same posthole in more than one structure.
So our search for buildings and structures in posthole palimpsests is slightly more complicated than anticipated, but builders are unlikely to have compromised their building because of this, so any irregularity would probably not affect the roof. Noticing which post is misplaced may actually give us a vital clue about the superstructure it supports.
- How far does the phenomenon extend geographically?
- In what other aspects of Celtic culture does it manifest?
- What does it tell us about Celtic culture, belief, and religion?
- When does it appear and disappear?
- What happens when the Celts interact with cultures like the Romans, where squares are not similarly regarded?
A plan from the Iron age cemetery at Fin d'Ecury, N France 
The symbolism of a square is based on the balance and regulation of four elements, or two opposing pairs, such as the points on the compass, or the four ‘elements’: Fire, Water, Earth, Air. A square is more often female in association, because of its relationship with the earth, as is the case in Hindu and Chinese symbolism. 
This provided a good example where religious belief impacts the way craft is practiced, and it is important to note that this is no way detrimental to the quality of the finished article. The imperfections being subtle, they have no significant effect on form or function.
I am unsure as to when systematic irregularity first appears. In the Neolithic, sacred spaces are assumed to have been defined by circular structures such as stone circles and henges. People are thought to have lived in rectilinear houses, and were buried in communal long barrows. By the Bronze Age we see roundhouses, and round barrows, and sacred space presumably became square at, or after, this point.
I have found systematic irregularity in buildings from the late Bronze Age, but the prehistoric built environment is very poorly understood, so it is not the easiest place to look. Most classic trapezoidal enclosures date from the late Iron Age, along with the square temples.
Religion can be a difficult business. Gods, rulers, and people can be fickle, so quite what became of systematic irregularity when parts of the Celtic world came under Roman control is difficult to judge. In this context it is impossible to avoid mentioning the druids, who, according to the Romans, were responsible for such things when they got there. Julius Caesar tells that the druids were staunchly against writing and literacy. As presumed keepers of the oral history and laws of the people, they had considerable vested interest in illiteracy. It may be that systematic irregularity was another practice that served to help differentiate the Celtic culture from the destructive forces of the Classical world.
It is clearly difficult to demonstrate that a phenomenon like systematic irregularity is real, let alone prove the theory that it was rooted in the belief that sacred space was defined by squares, but it works for me as an explanation. The apparent alternative, that Celtic surveyors and architects, our ancestors, were serially incompetent, is not something I am prepared to believe.
References and further reading: G. A. Carter, 1998: Excavations at the Orsett ‘Cock’ enclosure, Essex, 1976. East Anglian Archaeology Report No 86.
 Illustration cobbled together from: A. Chadwick, 1999: “Digging Ditches, but Missing Riches? Ways into the Iron Age and Romano-British Cropmark Landscapes of the North Midlands,” in Northern Exposure: Interpretative Devolution and the Iron Ages in Britain, B. Bevan, ed. Fig 10.3, p153 [Scrooby Top]. B. Cunliffe, 1978: Iron Age Communities in Britain: An Account of England, Scotland and Wales from the Seventh Century BC Until the Roman Conquest. 2nd edition. Routlage & Kegan Paul. Figs: [11.6] Aldwinkle Northhamptonshire, [11.5] Casterley Camp, Wiltshire, [11.14] Portsdown Hill, Hampshire, [2.4] South Lodge, Dorsett
 Ibid [13.26] Hod Hill, Dorsett
 Ibid [15.7] Danebury
 G. Bersu, 1940: Excavations at Little Woodbury, Wiltshire. Part 1, the settlement revealed by excavation. Proceedings of the Prehistoric Society, 6, 30 -111
 Taken from: Downs, Jane, 1997: The Shrine at Cadbury Castle: Belief enshrined. In Adam Gwilt and Colin Haselgrove, eds: Reconstructing Iron Age Societies. Oxbow Monograph 71, Oxford, 145–152
 B. Cunliffe, 1987: Hengistbury Head Dorset. Volume 1: The prehistoric and roman settlement, 3500BC–AD500. Oxford University Committee for Archaeology. Monograph No. 13
 W. F. Grimes and J. Close-Brooks, 1993: Caesar’s Camp, Heathrow, Middlesex. Proc Prehist Soc 59, 299-317
 A. C. King & G. Soffe, 1994: The Iron Age and Roman temple on Hayling Island, in A. P. Fitzpatrick and E. L. Morris, eds.: The Iron Age in Wessex: recent work, Salisbury: Trust for Wessex Archaeology, 114-16
 S. Piggott, 1965: Ancient Europe, Edinburgh University Press: Fig131, p 233
 See: J. Collis, 1977: Pre-Roman burial rites in north-western Europe. In R. Reece, ed., 1977: Burial in the Roman World. CBA Research Report 22, London
 I. M. Stead, 1971: ‘Yorkshire before the Romans: some recent discoveries’, in R. M. Butler, ed.: Soldier and Civilian in Roman Yorkshire, 21–43, Leicester
 J.E Cirlot, 1971: A dictionary of symbols 2nd edition, Routlage & Kegan Paul
 From Figure 5: http://www.geometricdesign.co.uk/perfect.htm Taken from: Martin Lings, 1987: Splendours of Qur'an Calligraphy and Illumination ISBN: 0500976481 Interlink Pub Group Inc.