Huge middens in southern England hold a special secret
How did societies change when iron first came to Britain? Kate Waddington, Niall Sharples, Richard Madgwick, Tom Higham and Alex Bayliss revisited an extraordinary mound at East Chisenbury in Wiltshire, hoping to create the first requirement of answering that question – a scheme to say what happened when. They succeeded
For archaeologists the key difference between the Bronze Age (2200–800bc) and the Iron Age (800bc– ad50) has traditionally been about the type of metals in use. The introduction of iron on was a significant technological shift. In Britain, making bronze involved finding ding scarce resources, especially copper and nd tin, and turning them into distinctive e artefacts using a melting and casting technology that evolved over one and d a half millennia. Iron, in contrast, smelted in high temperature furnaces s and shaped by repeated hot hammering ing (smithing), was locally abundant. At least initially in Britain, it was used to o produce relatively simple cutting-edge ge tools and ornaments.
What actually happened as one era a changed to the next is an interesting problem. It was initially thought that t people adopted iron technology from m the continent because they wanted better tools that were easier to make. . However, it has recently been observed ved that rare iron objects had been around nd for centuries before the major transition, and that in the Early Iron Age there were very few objects of either kind, especially compared to the he immediately preceding Late Bronze
Age. It seems there were more reasons for the adoption of iron than the simple appeal of a new technology.
There is evidence for several important changes between 800bc and 600 or 550bc, a period known as the Late Bronze Age/Early Iron Age ( lba/eia) transition or the Earliest Iron Age, and characterised by Llyn Fawr type bronze metalwork and pottery known as Post-Deverel Rimbury decorated wares. It seems that the transition was actually driven by social transformations:
• Most obviously there was conspicuous burial of large hoards of bronze metalwork. These are often mad made up of standardised objects such as facet faceted axe blades (303 of which were foun found together at Langton Matravers, Dor Dorset in 2007). Faceted axe blades are not functional, f due to the alloy which has t too much tin; another type, the Bret Breton axe, has too much lead.
•S • Settlements were enclosed by signi significant boundaries, such as those at Pimperne Pi in Dorset and Old Down Farm in Wiltshire, and the earliest hillfo hillforts appear, such as Balksbury Cam Camp in Hampshire.
•S • Substantial timber roundhouses, such as that at Pimperne, were not only arch architecturally imposing structures but cons consumed a large amount of prime timb timber. Limited numbers of people seem to have lived in them, suggesting the development d of family hierarchies.
•M • Midden sites appear – extensive deposits depo of thick, dark earths rich in dung and ash, and large assemblages of animal an bone, pottery and a diverse range rang of other artefacts.
Un Unfortunately, chronological resolution reso of this important time is very poor. poor The effectiveness of radiocarbon dating is hampered by something known as the Hallstatt plateau. Samples that date to between around 800 and 400bc give very similar radiocarbon ages, which means that when they are calibrated the possible range of dates covers the whole of this time period.
The only way to resolve this problem is to use independent dating methods, such as relative sequences provided by stratigraphy in the ground. Multiple radiocarbon dates can be combined with such archaeological information by using Bayesian statistics. This
approach seemed particularly appropriate to analysing the great lba/eia middens. Butchered animal bone and organic residues on potsherds could provide radiocarbon dates from primary rubbish – items directly related to the deposition of the material in their associated layers of midden.
Monumental middens
In the 1920s Maud Cunnington became the first to identify a midden – a huge waste deposit – recovering a vast collection of occupation material at All Cannings Cross on the north side of the Vale of Pewsey, Wiltshire. The finds were so numerous and distinctive that it became the type site for Early Iron Age pottery, and token sherds were distributed to archaeology departments and museums throughout the country.
A full understanding of the significance of All Cannings Cross came only in the last 30 years, as a result of excavations at a number of sites, including Llanmaes in the Vale of Glamorgan, Whitchurch in Warwickshire, Potterne and East Chisenbury in Wiltshire, Worth Matravers in Dorset and Runnymede Bridge in Surrey. These sites – there are now around 30 – belong predominantly to the first half of the first millennium bc. The soil matrices are largely composed of dung, stabling waste and hearth ash, suggesting one of the principal features was the presence of large numbers of animals. The rich bone and pottery assemblages, on the other hand, point to human activity such as feasting, and human skull fragments – modified at Potterne and All Cannings Cross – suggest ritual activity.
While midden sites are found across southern Britain, there are clusters: these include a concentration of sites in the Vale of Pewsey in Wiltshire and in Purbeck in Dorset, and a string of sites close to or in the River Thames. There is no obvious reason for this pattern. Some people have argued that the Thames and Pewsey sites indicate a concern with routeways, but that doesn't seem to explain the Purbeck sites, as the coastline here is characterised by high cliffs and exposed bays. It seems more likely that the determining factor was access to particular localised raw materials – shale and salt, for example, were of particular importance to the Purbeck sites.
This leads to an important point: while the sites share many similar characteristics, they are certainly not identical. Shale working was significant on Purbeck; prestigious metalwork and smithing characterised Llanmaes, Runnymede Bridge, Potterne and Whitchurch. Large numbers of cauldrons and axes were deposited at Llanmaes; a gold bracelet at Potterne; and a large collection of miniature bronze axes at Whitchurch.
Likewise, there are significant
variations in animal bones; sheep dominate most sites, but substantial quantities of pig support the idea that feasting was important. At Llanmaes an unusually specialised deposit focused on the right fore-quarter of pigs, clearly a cut of meat that was favoured for special events in south Wales. There is considerable variation in the depth of the deposits, from 50cm at Runnymede Bridge to 3m at East Chisenbury.
These sites had complex histories. At Potterne, for example, ceramic styles change and iron objects appear towards the top of the midden – the most recent part. This suggested to us that it might be possible to refine chronologies with Bayesian methodology (feature Jul/Aug 2011/119). There was a problem, however.
At most excavated middens, when you get to the bottom of the organic soil and expose the natural subsoil, you find the bases of numerous features – postholes, pits and gullies. Some of these had clearly been dug down through the midden from above, but their dark fills are completely indistinguishable from the midden soils. This was very clear at Whitchurch where in a trench measuring 5m by 2m we found eight pits cut into the subsoil, not one of which we had spotted as we dug through the midden. This makes it very difficult to be certain of the stratigraphic integrity of any material. An object might just as easily have been buried in a feature cut through pre-existing midden deposits, as in the midden itself: would radiocarbon analysis date the growth of the midden, or a pit dug down at a later date?
East Chisenbury
Fortunately, there was one midden where this problem could be overcome. East Chisenbury is one of the most
impressive of the midden sites. It survives as a very visible mound on the north of Salisbury Plain, just south of the escarpment overlooking the Vale of Pewsey and on the east side of the River Avon. It was rediscovered during the Royal Commission survey of the plain in the 1990s, and David Field, David
McOmish and Graham Brown excavated two small test trenches. Defence Infrastructure Organisation conducted further investigations in 2016−17, supported by Operation Nightingale/Breaking Ground Heritage and Wessex Archaeology. These included a geophysical survey that defined the extent of a substantial and broadly contemporary enclosure ditch around the midden. Postholes attested to the presence of numerous roundhouses beside the ditch beyond the midden’s edge.
The 1992 excavations recovered some 10,000 potsherds and almost 1,500 animal bones, as well as a range of small finds that included a Sompting axe, a Late Bronze Age-looking axehead dating from the lba/eia transition period. Most importantly for our purposes the excavations revealed the midden was, unusually, made up of a variety of clearly distinguishable layers, interspersed with chalk dumps and platforms or hard standings within the otherwise unstable heap. Similar deposits have been identified in other sites, such as All Cannings Cross and Runnymede Bridge, but they are much less frequent there. Interestingly, two test pits excavated in the East Chisenbury midden in 2016–17 revealed a very different sequence, and produced far fewer finds than the previous excavations; it is not a homogenous mound.
This visible layering meant that at East Chisenbury the large numbers of well-preserved animal bones and large residue-encrusted re potsherds could be securely se associated with contemporary activity: ac we could the trust the stratigraphic str order. It could provide the th key constraints for an effective Bayesian Ba model.
We selected 28 samples of short-lived material, m animal bone and carbonised residues, re from the midden layers in the two tw 1992 trenches. The results have been be exciting. They confirmed initial suspicions su that accumulation had begun on a cultivation soil, which was rich in midden m material. This horizon started around ar 1020–855bc (95% probability), probably pr between 975–890bc (68%); and an ended 795–700bc (95%) and probably pr 790–745bc (68%) – within the th Late Bronze Age.
There is then a gap of 30 to 145 years before be the major deposition of midden begins,be around 795–565bc (95%). The midden stops accumulating between 500–355bc (95%), probably between
455–385bc, which indicates it had a life span of between 170 and 320 years. In rough terms, the midden was growing at some time between 700 and 400bc.
Thus middens were used for a long period, and people were active at these special places over several generations. There was sufficient time for changes in artefact use and ceramic style, such as were observed at Potterne. Most importantly, dates show that middens, Post-Deverel Rimbury decorated wares and the Late Bronze Age/Early Iron Age transition period, continued much later in the first millennium bc than people have previously thought. Most estimates suggest the transition period ended roughly 600 to 550bc, but this is clearly over a century too early; it ended in the mid-to-late fifth century bc.
Changing values
Important confirmation for this date range comes from work at the Iron Age hillfort of Danebury in Hampshire. Bayesian analysis of the chronology suggests the small hillfort enclosed by a monumental bank and ditch was constructed around 450bc. Scratch Cordoned Bowls – a very distinctive and diagnostic Early Iron Age fineware – are found in large quantities at Danebury but are completely absent at East Chisenbury, implying that the sites do not overlap in time. Nevertheless, some hillforts produce decorated PostDeverel Rimbury pottery, similar to that found at East Chisenbury, suggesting activity contemporary with the middens. At All Cannings Cross, in contrast, Scratch Cordoned Bowls suggest that activity continued there into the Early Iron Age.
The only other midden to have an extensive series of radiocarbon dates is Runnymede Bridge, which again shows
a sequence. It begins with the construction of a waterfront palisade. Midden accumulation starts in 870– 805bc (95%) or 850–820bc (68%) and ends around 795–745bc (95%) or 790–765bc (68%). Clearly, this predates the midden at East Chisenbury, which is to be expected as the large ceramic assemblage shows very little sign of the elaborate decoration found at Chisenbury.
Particularly satisfying is that the project has provided a useful chronological model for the period of the lba/eia transition, despite the presence of the Hallstatt plateau. The results have a significant impact on the accepted settlement and monument sequence in Wessex, as midden sites had not previously been seen to extend into the fifth century bc, a century more traditionally associated with the development of hillforts. As such, published Iron Age periodisation models may require some adjustment, even if this can currently be applied only to Wessex. More work is needed to refine the chronology of other midden sites and contemporary phenomena, such as the early hillforts and the monumental roundhouses. Only when the chronology of these settlement forms has been refined will the complex nature of the transition be understood.
One theory for these middens is that they represent periodic gatherings of the dispersed communities of southern England, coming together with herds of cattle and sheep to undertake exchanges of animals and scarce goods, and arrange marriages and other alliances. The consumption of vast amounts of food and the destruction of enormous quantities of pottery, suggest these meetings were significant and prolonged feasting events that were associated with a range of other activities, including ritual and competitive destruction of material wealth.
That this occurred in the lba/eia transition is likely to reflect the destruction of a value system based on the circulation of bronze metalwork. This system had maintained relationships between the dispersed households and communities in southern Britain, and its demise required a new strategy for mediating relationships. Seasonal gatherings seem to be one of the social strategies used to lin link different communities, but large ho houses, continued bronze exchange and hi hillforts were alternatives strategies in th the transition period. In southern Br Britain these alternatives were short liv lived. By the Iron Age, hillfort co construction became the dominant m mechanism for social control.
British Br Archaeology has reported on excavations ex at several midden sites: All Cannings Ca Cross (News Jan 2004/74), Westbury W (News Jul/Aug 2005/83), Whitchurch W (feature Jul/Aug 2007/95), Chisenbury Ch (feature Jan/Feb 2012/122) and an Llanmaes (feature Jul/Aug 2016/149). Kate Ka Waddington is lecturer in archaeology at Bangor University; Niall Sharples is professorp of archaeology at Cardiff Un University; Richard Madgwick is lecturer in archaeological science at Cardiff Un University; Tom Higham is director of the Radiocarbon Accelerator Unit at the University of Oxford; Alex Bayliss is head of scientific dating at Historic England. En See “Histories of deposition: creating cre chronologies for the Late Bronze Age–Early Ag Iron Age transition in southern Britain,” Br Archaeological Journal (2018)