Iron Age interior design: Mapping the inside of Dorset’s hillfort enclosures

14 mins read
Overlooking Hambledon Hill, one of Dorset’s most famous hillforts. Recent geophysical survey has shed new light on the archaeology of its interior. (IMAGE: Jo and Sue Crane)
Across Dorset, impressive earthworks mark the location of Iron Age hillforts. Until recently, though, little archaeological attention had been paid to what lay within these mighty ramparts. Now, thanks to modern geophysics, the picture is beginning to come into focus, as Dave Stewart and Miles Russell explain.

Hillforts are among the most dramatic and visually striking archaeological monuments in the British Isles. Appearing from the late Bronze Age and reaching their peak in the middle Iron Age, Dorset is particularly rich in such enclosures, with 35 – including the internationally famous sites of Badbury Rings, Hambledon Hill, Hod Hill, and Maiden Castle – having been recorded to-date. In recent years, detailed geophysical surveys have been undertaken, with some success, at a number of British hillforts, but the Dorset sites have unfortunately been rather neglected. This is something we have been keen to correct.

Since 2009, Bournemouth University has been investigating Iron Age and Romano-British settlement patterns across south-western Britain, and initial excavation at a number of sites around Winterborne Kingston, near Bere Regis (see CA 281 and 313), was complemented by an intensive programme of geophysical survey. Creating detailed records of the subsurface features of hillforts was, we felt, particularly important in order to help improve understanding of the form, phasing, structure, and possible function of these impressive prehistoric enclosures.

What, then, is a hillfort? The term is, as many have noted, one of convenience: a useful, if not always entirely topographically accurate, archaeological descriptor, not all of such monuments being found on the highest points in the landscape. Neither, it must be acknowledged, is an association with fortification always helpful, for many enclosures may originally have been designed for purposes other than simply defence and military engagement.

Iron Age hillforts are scattered across Dorset. Many have yet to see archaeological exploration.

Most recognisable forms of hillfort appear from around 600 BC with the univallate (single rampart) enclosures of the early Iron Age. Averaging 5ha in extent, the shape of these earthwork circuits generally conforms to the contours of specific points in the landscape, and their entrances are usually quite simple, with either one or two opposing breaks in the banks. By the middle Iron Age, many of these primary enclosures had been abandoned, although a few were extended and developed into more complex multivallate sites like Maiden Castle. Unfortunately, until the late 1970s, the archaeological exploration of hillfort interiors was relatively rare, with excavators tending to focus instead on sequences of rampart construction. Since the 1980s, though, detailed investigation, geophysical survey, and aerial photography have transformed the field, suggesting that, while certain enclosures contained areas of settlement and storage, others possessed very little evidence of intensive activity at all.

The appearance of these more defined and complex forms of hillfort enclosure may reflect broader developmental changes in early and middle Iron Age society, with communities feeling the need to increase control both of the land and the storage of resources. These sites could have had diverse functions: some hillforts may have acted as welldefended zones of semi-permanent elite settlement, others as places of storage and redistribution, while still more may have performed a more overtly religious and ceremonial function. By the 2nd century BC, though, only a limited number of British hillforts remained in use. Those that survived to this point developed rampart sequences and increasingly elaborate forms of entranceway – but in most areas of southern Britain, throughout the very late 2nd and early 1st century BC, hillforts seem to have been in a marked decline, with only a few apparently continuing in use at this stage.

Maiden Castle was excavated by the Wheelers in 1934-1937. Mortimer Wheeler interpreted their findings as evidence of a bloody Roman attack on the hillfort. (PHOTO: Jo and Sue Crane)

It is the final phase of hillfort activity that has dominated discussion of the Dorset sites, with much of the debate concerning the apparent ‘death’ of enclosure systems rather than their life and period of use. By and large, this end had traditionally been imagined to have been violent (see box opposite): a war-perspective influencing the first detailed investigation of hillforts in the county, most notably in the work of Mortimer and Tessa Verney Wheeler at Maiden Castle (1934-1937) and Ian Richmond at Hod Hill (1951-1958). As one of the largest and most impressive of all hillforts in Britain, Maiden Castle certainly appealed to Mortimer Wheeler’s sense of drama, offering him the opportunity to link archaeological data with a specific historical event, namely the Roman invasion of AD 43. His final summing up of excavation results there was a masterclass in archaeo-historical interpretation, evocatively describing the relentless march of a Roman legion towards the hillfort’s ramparts and the bloody battle that followed. Similar perspectives permeate Ian Richmond’s Hod Hill report, where discussion centres on the forcible removal of the indigenous population and the establishment of a Roman fort at the site.

There is nothing inherently wrong with the approach taken by Richmond and Wheeler, but by focusing so resolutely upon the ‘violent demise’ of the Hod and Maiden Castle enclosures, both men unfortunately shifted attention away from the hillforts themselves. In the years that followed, it is fair to say, we have sometimes lost sight of the context that hillforts operated within and the societies that brought them into being.

In this photograph of Hod Hill, which faces east, the prehistoric enclosure and the Roman fort that was later built within the ramparts lie in the north-west (bottom left) corner. (IMAGE: Jo and Sue Crane)

With the development of nondestructive forms of geophysical survey in the 1980s and 1990s, though, both the perception of hillforts and the way in which they could be examined and researched have been significantly altered. Large-scale survey programmes such as the Wessex Hillforts Project (initiated by the Ancient Monuments Laboratory in 1996) generated more wide-ranging data on these sites, demonstrating the possibilities for resolving their internal structure, layout, and organisation. It was clear that the potential for adding the Dorset enclosures to this database, for the purposes of contrast and comparison, was considerable.

In embarking on our project, there was much to consider. The majority of hillforts in Dorset are located on chalk, although a few occur on the limestone of Purbeck and in the clay valleys to the extreme west of the county. Of the 35 sites identified, more than a third have been subjected to some form of archaeological investigation – although only Chalbury, Hod Hill, Maiden Castle, and Pilsden Pen could claim to have been sampled to any significant degree. The problem is not that hillforts are particularly difficult sites to investigate archaeologically, but the sheer acreage of land enclosed within their rampart circuits means that only a small percentage of any interior has, to-date, been sampled. Geophysical survey, therefore, represents by far the best way of trying to understand the internal form of these sites.


Mortimer Wheeler conjures an extraordinarily vivid account of a Roman assault on Maiden Castle, but more recent reassessment of excavation data from the site has called this narrative into question. What Wheeler interpreted as evidence for burnt-out houses is in fact traces of iron-working, while human remains dubbed a war cemetery represent nothing of the sort, bodies having been laid to rest over a very long period of time. The famous ‘ballista bolt’, found in the spine of one skeleton, is actually a spearhead (possibly of British manufacture), while most of the other injuries recorded in the cemetery may relate to execution rather than combat. At Hod Hill, the ballista bolts found during Ian Richmond’s excavation could easily (and more plausibly) have come from the Roman fort, where catapult foundation platforms were discovered; it is quite probable that they could have been discharged during target practice rather than in some speculative attack. Crucially, archaeological evidence also suggests that both Maiden Castle and Hod Hill had been largely abandoned by 100 BC, a century and a half before the Romans arrived.

A version of this text was originally printed in CA 335 as a response to a reader’s letter.


Having carried out more than one survey in sub-zero temperatures and gale-force winds, it seems amazing that anyone would voluntarily live on these exposed hilltop sites, yet people were clearly doing just this, and in some numbers. Geophysics has produced compelling evidence of roundhouses invisible to the naked eye – so many of them that it is probable that the actual level of habitation on all such sites has been severely understated. Across the survey, we could see a wide range of discrete internal forms, ranging from limited scatters of pits and features to pits clustered into distinct zones, pits associated with roundhouses, and intensive distribution of roundhouses and other built forms divided and arranged by roads and trackways.

Hod Hill has been completely surveyed under the current geophysics project: the resulting magnetometry plot is shown here.

Three distinct forms of prehistoric roundhouses emerged during our project. The first, defined as a ‘Pimpernestyle’ roundhouse (named after a Dorset site dug in the early 1960s), has a ring ditch measuring between 10m and 15m in diameter, with an entrance generally facing south-east. We have seen examples of such structures at Abbotsbury, Hod Hill, Pilsdon, Maiden Castle, and across the flatter areas of Hambledon Hill. Other house platforms at Hod Hill have no apparent ring ditch: they appear similar to dwellings recorded from the settlement outside Poundbury hillfort, where ring-slot walls measure about 7m diameter.

These ditches were probably intended to help keep the hut floor dry and to prevent the daub wall from getting excessively damp and crumbling, but in steeper terrain they do not seem to have been necessary. In environments like these, house platforms seem to have generally been terraced into the side of the slope. At Abbotsbury hillfort, for example, these surfaces appear to have had a ditch on the uphill side only, with the natural incline presumably carrying any water away downhill. Similar terraces are also seen in some numbers in the south-west quadrant of Hod Hill, a part of the hillfort that has undergone severe plough damage, though no ring ditches can be seen there. Their presence at sites like Abbotsbury could signify that such features did originally exist here too, but have since been erased through modern farming. On Hambledon Hill, the slope is steeper still, and most of the site’s dwellings are terraced into the hillside, but no ditches are evident – yet ploughing is unlikely to have been instrumental in their removal, as the platforms appear generally well preserved. Perhaps here the angle of the terrain by itself was sufficient to prevent the dwellings from becoming waterlogged.

The full magnetometry plot for Maiden Castle, revealing new clues to the features that lie within its ramparts.

Other clues to how these settlements operated have also come to light during our survey: the layout of the recorded house structures shows a significant degree of interior planning, and where gaps between groups of structures and pits can be identified, these may be interpreted as some form of roadway or thoroughfare. Some of these incorporated earlier features: at Maiden Castle, a Neolithic bank barrow splits the hillfort lengthwise, apparently forming the spine for a herringbone pattern of cleared Iron Age pathways. The long barrow within Hambledon hillfort may similarly have formed part of a track through the narrow central part of the enclosure, while a broad path takes a direct route between the two southern gates. At Spettisbury, two possible tracks branch from the entrance, terminating against the southern rampart, perhaps betraying remodelling and an earlier second entrance that had since become defunct. Meanwhile, a single trackway through Buzbury hillfort indicates that a gap in the southern rampart, lying east of the current entrance, was this site’s original gateway.

At Hod Hill, magnetometry revealed over 200 circular anomalies representing structures, as well as an extensive road system. This detail of the survey plot shows trackways extending from the Steepleton Gate at the north-east (top right) and some of the densely packed roundhouses, hearths, pits, and fence lines.

Within Hod Hill, the most striking aspect of the geophysical plot is the number of circular anomalies representing structures – over 200 in total – which are so densely packed that their footprints often overlap, suggesting a long period of occupation. Each of these appears to contain one or more strong pit-like anomaly – comparison of magnetometry and resistivity plots here indicate a mixture of hearths and pits, some probably combining both. A line of similar pit-like features without ring ditches appear near the eastern rampart, and a scattering of the same have been detected in the south-west.

To the north-east of the enclosure, we can also deduce further evidence of planning: the tree-like branches of a road system, extending from the Steepleton Gate, is hinted at from the lack of cut features rather than as a positive feature in its own right. Through such negative features, we can see other Iron Age amenities in great detail: some sections of both main and minor trackways are bounded by ditches, apparently created by the gullies surrounding structures being linked to form a continuous drainage channel. A higher-resolution plot of one such section, near a junction, showed it to be also bordered by rows of fourpost structures, generally about 2.5m square with four or, in two instances, five post-holes, probably representing storage buildings such as granaries.

This detail of the Maiden Castle survey shows the Neolithic bank barrow that splits the fort interior lengthwise, and the two enclosure systems.

Various other branches of the trackways terminate at areas of magnetic scatter below the ramparts or appear to divide the internal area into blocks of features. Within each of these blocks, a closer look at the magnetometry reveals at least one structure surrounded by its own enclosure, possibly relating to areas of administration and control. Finally, in the south-east corner there are two areas abutting the quarry pits that appear to be screened with palisades. We can see traces of possible smallscale internal enclosures at Poundbury and Banbury hillforts too, although here the survey responses are generally weak. Maiden Castle’s two enclosures are much more clearly apparent: the northernmost seems to have started life as a single oval measuring 30m long, before being later doubled in length, while that established to the south is some 60m across and could have formed an individual enclosed farmstead in its own right.

How should we interpret these findings? The potential number of dwellings inside Dorset’s developed hillforts seems excessive if the enclosures originally acted only as the type of temporary refuge with which they have been traditionally associated. On the contrary, overlapping buildings indicate occupation over a long period, which also makes it unlikely that the structures simply housed the huge workforce required to raise the ramparts in the first place. The traces of industry and the artefacts found in the various excavations at such sites suggest that, by the middle Iron Age, developed hillforts had become permanently occupied centres, possibly akin to market-towns, with resident storage facilities and craftspeople.

Magnetometry clearly shows Hod Hill’s Roman fort in the north-west corner of the hillfort, overlying the faint traces of earlier circular structures.

Taken together, our geophysical surveys, combined with a reanalysis of earlier archaeological evidence, strongly suggests that all Dorset’s enclosures were entering a period of decline long before the start of the 1st century AD, with none being actively refortified at the time that the Roman legions arrived in AD 43. Indigenous reoccupation of the hillforts also seems to have been very limited during the Romano-British period. Roman-period traces at Hod Hill, for example, are limited to a fort, well defined in the geophysical survey, which was established in the enclosure’s north-western corner as an ‘observe and control’ establishment overlooking native settlements in the river valley of the Stour. The double ditch and rampart surrounding the fort can be clearly seen in the survey plots, as can the two titulus ditches that served to bolster protection of the entranceways.

Within the fort, we can also see the outlines and internal divisions of rectangular barrack buildings – signals are particularly strong in the south of the camp – and these buildings overlie the faint traces of circular structures that can also be seen at the northern edge of the site. One very clear circle, located at the heart of the camp, is thought to represent the footprint of a Bronze Age barrow – an interpretation under which it has been granted scheduled status – although the possibility of it being a roundhouse dating from this later phase of the hillfort’s life cannot be ignored.

We can also see Roman-era activity at other hillforts. Maiden Castle contains a Romano-British temple, and another such structure lies just outside the ramparts of Badbury. At Poundbury, a disturbed square structure appearing in the geophysical plot may also be a temple – or, alternatively, the site of a 17th-century plague house that is known to have existed on the site. As this latter interpretation suggests, use of the hillforts did not cease with the end of the Roman occupation. In fact, one enclosure, that of Hambledon Hill, even saw military action in the 17th century. During the English Civil War, a group of peasants known as the Dorset Clubmen rose up against the hardship being caused by the armies of both sides foraging on their land. In August 1645, a force of some 2,000 Clubmen made a stand on Hambledon Hill, but were routed by no more than 50 of Cromwell’s dragoons.

An aerial photo of Badbury Rings hillfort, which was used as a vantage point during the Second World War. It still bears traces of military installations from this period. (IMAGE: Jo and Sue Crane)

This defeat vividly demonstrated that the defensive capabilities of hillforts were no match for ‘modern’ weapons – though the prominent positions of Lambert’s Castle and Nettlecombe Tout hillforts were later put to good use with the building of semaphore signalling towers to counter the threat of Napoleon, providing an early warning system of possible invasion. World War II also saw several hillforts exploited as vantage points, with traces of military installations remaining at Abbotsbury, Badbury, and Poundbury.

Beyond the martial sphere, rectangular features of uncertain date at Spettisbury are thought to possibly represent temporary buildings erected to accommodate workers engaged in the construction of the railway there in the late 19th century. Meanwhile, the creation of pillow mounds at Pilsdon, Coney’s Castle, and Eggardon reflects the intensive farming of rabbits in the area (see CA 261 for more on the archaeology of this industry), while the ‘burrowing’ of antiquarians can be detected at Maiden Castle and Pilsdon Pen. Designation of Dorset hillforts as Sites of Special Scientific Interest and nature reserves is just the latest stage in a continuum of activity that has seen many of these enclosures change shape and serve multiple purposes over their lifetimes.


Thanks to geophysical survey, what originally went on inside the great hillforts of Dorset, both in prehistory and in the centuries after, is becoming less of a mystery. It is also apparent, when combining the new fieldwork with finds from earlier excavations, that, other than low-level secondary reuse activities such as burial and metalworking, there is no proven link between the hillforts of Dorset and the Durotriges tribe who inhabited the area in the later Iron Age. Nor were these sites battle-ready fortifications at the time the Roman legions arrived, having been in serious decline since the beginning of the 1st century BC. This is significant, for it means that the Dorset enclosures were not after all a ‘breed apart’ from their neighbours, the hillforts to the east and north, as they were in fact created, used, modified, and eventually abandoned in broadly the same way. Certainly, there is nothing in the archaeological record to suggest that the hillforts of Dorset were ‘flourishing’ at the end of prehistory – such activity there was marginal at best.

Hambledon Hill, the subject of our lead image, is shown in an entirely new light in this magnetometry plot. Geophysical survey has proven to be a remarkably effective tool for mapping the interiors of prehistoric landmarks like this.

This raises the question of where, if not in the hillforts, were the major centres of Durotrigian settlement in the mid-1st century AD? And were these in anyway similar to the late Iron Age oppida identified in tribal lands to the east? In both cases, the answer remains to be seen. Only further excavation and more extensive fieldwork will more fully clarify issues of status, use, function, ownership, resettlement, and subsequent modification of hillforts in Dorset and the wider South-west. Until such a time, it is hoped that the new survey work will not only help to clarify the internal form of these sites, but will also start the process of putting to rest one of the more potent myths of British archaeology. It is time to consciously uncouple the Durotriges from the enclosures of Dorset, and to assess the hillforts in their own right – without need to mention the Roman army.


Dave Stewart and Miles Russell, Hillforts and the Durotriges – a geophysical survey of Iron Dorset, Archaeopress, £30, ISBN 978-1784917159.

This article was featured in CA 336.

Leave a Reply

Your email address will not be published.