Anglo-Saxon Cross

Monument Class Descriptions

(a key aspect of the Monuments Protection Programme)

The Example of 'Large Multivallate Hill Forts'

Compiled January 1989

  1. Definition
  2. Date
  3. General Description
  4. Distribution and Regional Variation
  5. Rarity
  6. Survival and Potential
  7. Associations
  8. Characterisation Criteria
  9. Bibliography
  10. Acknowledgements

This description has been copied from the Monuments Class Descriptions pages on the English Heritage Website. Further descriptions for another 233 classes of monument can be found there in addition.


Large multivallate hillforts are fortified enclosures, located on hills, and are defined by 2 or more lines of concentric earthworks, which demarcate an internal area of between 5ha and 85ha. On the majority of sites these earthworks will be closely spaced, at intervals of up to 15m. Examples have been recorded where the spacing between the earthworks is wider, either for all or part of the circuit.

Each earthwork usually comprises a rampart accompanied by a ditch, although some are formed by ramparts alone. These multiple boundaries will describe a complete circuit, except on sites situated adjacent to a particularly steep slope or precipice. In examples of this nature, the largest and greatest number of earthworks will occupy the gentlest slope. They tend to continue in reduced or univallate form along the steepest part of the hill, and on some sites are entirely absent in this area.

Most large multivallate hillforts have two entrances, although examples with one and more have been noted. These may be marked by simple gaps; inturned ramparts; offset ramparts; oblique approaches; guardrooms; hornworks; and outworks.

The majority of excavated and surveyed sites have produced evidence for intensive occupation. The main building form comprises oval or circular houses, constructed in a variety of materials, which in many cases appear to have been rebuilt on a number of occasions. These show a considerable variation in diameter and a tendency to cluster in specific areas. Some seem to be aligned along streets. They may be associated with annexes, or located within palisaded enclosures. Four- and six-post structures, interpreted as raised granaries, are also fairly common. A small number of sites include buildings which appear to have functioned as temples. Other features comprise platforms; paved areas; pits; gullies; fencelines of stakes or posts; hearths; and ovens. There is evidence for bronze and iron working, and potting on a number of sites. At the same time the range of finds may include artefacts from well outside the immediate locality.

It might be possible to confuse large multivallate hillforts with univallate forms if the outer earthworks have been levelled. The problem can usually be resolved by careful survey, which should, at least, result in the identification of ditches. A second source of confusion may lie in the distinction between large multivallate hillforts, where earthworks are separated by distances exceeding 15m, and multiple enclosure forts. This is likely to occur mainly in south-west England, where the distributions overlap. Distinctions between the two classes are relatively numerous and include differences in size, topographic location and the scale of the earthworks. Unlike large multivallate hillforts, most multiple enclosure forts are smaller than 5ha and tend to be located on hillslopes rather than hilltops. The scale of the earthworks, surrounding the inner enclosure on large multivallate hillforts, will be much greater than on multiple enclosure forts. In addition, widely spaced earthworks on large multivallate hillforts usually occur in conjunction with closely spaced ramparts.

Large multivallate hillforts are usually seen as centres of permanent occupation, and it has been suggested that some represent proto-urban settlements. Many interpretations stress the defensive nature of the earthworks and represent their construction as a response to increasing warfare. In recent years, this emphasis has changed, and large multivallate hillforts are seen as one manifestation of the power struggles between competing elites. Their ability to mobilise the labour, necessary for monumental works on such an enormous scale, is seen as the ultimate expression of this competition.


The chronology for large multivallate hillforts is based upon a relatively large number of examples, which have produced diagnostic artefacts. It finds further support on a few sites, where samples have been submitted for archaeomagnetic dating or radiocarbon assay. The resulting dates indicate that the origins of large multivallate hillforts lie in the 6th century BC, and that they were constructed and used until the mid-first century AD. Recorded exceptions include The Breidden, with a univallate phase dated to the late Bronze Age, and Salmonsbury, which was occupied until the fourth century AD.

On a number of large multivallate hillforts, early dates seem to refer to initial phases, when the defences were univallate in form. The evidence suggests that occupation was continuous throughout the history of these sites, although exceptions may include Yarnbury and High Rocks, where there is a period of abandonment between the univallate and bivallate phases. At Maiden Castle and Danebury univallate hillforts were constructed during the 6th and 5th centuries BC. Similar dates have been suggested for univallate stages at other sites, such as The Wrekin, Ham Hill, Hod Hill and Yarnbury. Other hillforts, also with univallate phases may have begun slightly later, during the 4th or 3rd centuries BC. Examples include Bredon Hill, Ravensburgh Castle and High Rocks.

Dates for the development of these sites in multivallate form are less secure. Many published sources suggest that multivallation should be placed either in the 1st century BC or in the 1st century AD. The excavated evidence does not entirely support this, although multiple ramparts were erected relatively late at some sites. The multivallate phase at Maiden Castle, for example, is thought to date to the early 1st century BC. In contrast to this, the outer earthwork at Danebury seems to be of 4th century date, while additional ramparts at The Breiddin may have been constructed at about the same time.

The dates for sites where multivallate earthworks were part of the original design, again range fairly widely between the early and late Iron Age. The bivallate phase of Hambledon Hill, for example, has been placed in the 6th or 5th century BC, while Salmonsbury seems to have been contructed during the 1st century BC.

A number of large multivallate hillforts, such as The Breidden and High Rocks, were abandoned during the 2nd or early 1st century BC. Others were occupied until the mid-1st century AD, when enlargements and extentions to entrances and ramparts were apparently interrupted by the Conquest of AD43. Unfinished earthworks, dating to this period, have been recorded, for example, at Hod Hill and Maiden Castle.

Continuous occupation into the Romano-British period is rare, and Salmonsbury is perhaps the only example where settlement continued relatively unchanged until the 4th century AD. Several examples, including Ham Hill, Hod Hill and South Cadbury, were used as short-term military encampments immediately after the Conquest. All three sites were abandoned before AD 60.

A number of large multivallate hillforts were reoccupied during the later Romano-British period. South Cadbury and Maiden Castle, for example, were used as sites for 3rd or 4th century temples, while a villa of a similar date was constructed within Ham Hill.

There is considerable variation in the duration of occupation at individual sites. The figures may be slightly over-estimated, since they are based on diagnostic artefacts in use over fairly long periods of time and on radiocarbon dates with relatively large standard deviations. The shortest period of occupation seems to have been around 200 years, and is represented by sites such as Bury Wood. The majority of examples, including The Wrekin and Bredon Hill, were used for between 300 and 400 years. The duration of a small number of sites, including Hod Hill and Danebury, at 500 years is slightly longer, while Maiden Castle was in use for 600 years.

To a certain extent, this variation seems to reflect regional differences. Large multivallate hillforts, occupied for periods of between 500 and 600 years, are located in Wessex and Somerset, although these areas also include examples used for much shorter lengths of time. In addition, the dates for the origins and abandonment of large multivallate hillforts, are also regionally determined. Sites in Kent and Surrey were not constructed until the end of the 4th or beginning of the 3rd century BC. Examples in Salop appear to have been abandoned by the 2nd or 1st centuries BC. In other areas, where the dates are more varied, phases of construction and remodelling seem to cluster in four periods: the 5th/6th centuries BC; the mid-4th century BC; the late 2nd/1st centuries BC; and the mid-1st century AD. Early Roman military occupation is isolated to sites in Dorset and Somerset, while continuous settlement from the late Iron Age into the Roman period is only noted at Salmonsbury, in Gloucestershire.

General Description

A large number of publications, concerning hillforts as a general category, have been produced since the beginning of this century. The vast majority of early accounts distinguish multivallate from univallate hillforts, but provide very little in the way of detailed classification or discussion. In the 1930's, Hawkes and Piggott suggested that multivallation was introduced to this country from Europe during the late Iron Age. This interpretation found general acceptance, and influenced most accounts published before the mid-1970's. Around this time, the concept of waves of Iron Age invaders was gradually discredited and abandoned, although the late chronology for multivallate hillforts remained much in evidence. The greater emphasis on hillfort classification, which emerged during the 1960's, led to the publication of a number of general accounts, in which multivallate hillforts were divided into groups according to size and the number of defences. Large multivallate hillforts figured in each of these schemes, although they were identified by a variety of terms. None of these classifications gained general acceptance. In 1974 Cunliffe suggested a developmental model for hillforts, which he discussed further in a number of papers published into the 1980's. In general terms, this represents a chronological progression from slight univallate forms to those of increasing elaboration and size. Large multivallate hillforts, discussed under the heading 'developed hillforts', represent the final stage of this model. Alternatives have yet to be put forward, but the excavated evidence suggests that Cunliffe's model may not be applicable outside Wessex. Even within this area, there seems to be considerable variation in the developmental history and chronology of individual sites.

Accounts of excavations within large multivallate hillforts, published before the 1940's, give details of narrow trenches through the earthworks and entrances. Some of these were extended into the area immediately behind the ramparts. Excavations of this nature provide information concerning the chronology and structural history of individual sites. Many produce evidence for fairly dense occupation of the interior, but the trenches were usually too narrow to allow for the reconstruction of buildings. The first open-area excavation, which investigated the occupation of a large multivallate hillfort and related it to the development of the defences, took place at Maiden Castle and was published in 1943. Further investigations have taken place at the site in recent years. Other key sites include Hod Hill, published in 1968; Danebury, with the results of the first 10 years published in 1984; and Hambledon Hill, which is not published in complete form. The information from Maiden Castle, Hod Hill and Hambledon Hill include the results of detailed surface surveys, which at the latter two sites allowed for the partial reconstruction of the layout of different features in the interiors. At Maiden Castle and Danebury surveys of the landscape surrounding the hillforts, allowing for a reconstruction of the contemporary settlement pattern, await publication.

The most prominent features of large multivallate hillforts are the closely-set multiple earthworks. The majority are bivallate, although there are also relatively numerous examples with three or more ramparts. The overall width of bivallate earthworks can vary between 15m and 50m, while measurements for multivallate examples range between 30m and 140m. Width is a fairly useful index of the scale of the ramparts and ditches. In general, the wider earthworks are also the largest. The only exception occurs on sites where the ramparts are spaced at intervals in excess of 15m.

On sites where the topography is fairly undifferentiated, the earthworks describe a complete circuit. Differences are encountered on large multivallate hillforts where the contours are more varied. When one or more sides of the hill are particularly steep, the earthworks tend to be absent altogether, or reduced in size and/or number. In fact, some large multivallate hillforts have a single rampart running along a sheer drop, which joins an arc of multiple earthworks situated across the easiest approach to the site. It has been suggested, although not substantiated by excavation, that a palisade may have marked the edge of a precipice, on sites where the ramparts are absent in this area.

The configuration of each line of earthworks, and the distance between them, often shows considerable variation within the same site. At Hambledon Hill, for example, the second rampart exactly reflects the course of the inner bank, but the third shows a number of deviations. As has been suggested, this may reflect a later construction date for the outer earthworks. It could equally reflect alterations made to take advantage of the local topography. Where this is relatively undifferentiated, as at Maiden Castle, the distances between earthworks and the course which they follow is very similar. In contrast, at sites such as South Cadbury, the natural slope of the hill is used to its maximum advantage, with ramparts occuring at uneven distances where they best serve to enhance the contours. In fact, the wider spacing between ramparts on a number of similar sites can be explained in this manner.

The intervals between earthworks may also vary at different points in their circuit. This usually occurs in the area of the entrance or entrances, where at least two lines of earthwork diverge to define a fairly wide space. Examples include The Wrekin and Old Oswestry.

Some sites, such as Salmonsbury, are associated with small annexes. The earthworks, which enclose additional areas of this nature, are invariably much smaller than those surrounding the main hillfort.

In addition to multiple earthworks, large multivallate hillforts are distinguished by the size of their internal area. This appears to fall into one of three clearly defined groups. The first and largest of these comprises sites of between 5 and 10ha. The second includes examples measuring between 12 and 29ha, while the earthworks surrounding the third and smallest group of multivallate hillforts enclose between 40 and 85ha.

The shape of large multivallate hillforts is particularly varied, and seems to depend mainly on local topography. The earthworks are usually slightly irregular and, on the majority of sites, are curvilinear. They may be roughly circular; elongated; elongated and sinuous; oval; trapezoidal; or polygonal in outline. A small number of sites are more angular. Examples include Bury Wood, which is triangular; Salmonsbury, which is roughly square; and Hod Hill, which is rectangular.

Each line of earthworks surrounding large multivallate hillforts usually comprises a rampart and a ditch, although some are formed by ramparts alone. Internal quarry ditches or scoops, and counterscarp banks are also common components in this area of the site.

The structure and scale of the ramparts can vary considerably, not only between different large multivallate hillforts, but also within the same site. At the Wrekin and Old Oswestry, for example, the earthworks include scarps cut into the steepest part of the slope, stone-revetted banks, and ramparts of dump construction; while on a number of sites with glacis ramparts, such as Maiden Castle, the entrance passage is stone revetted. If there is a difference in scale, the outer ramparts tend to be smaller than the inner. At Salmonsbury, for example, the inner rampart is 18m wide and the outer is 12m wide; while at Caesar's Camp, the inner rampart is 17m wide and the outer is 7m wide.

In southern and eastern England, glacis ramparts are the most common form. These are constructed from dumps of earth and rubble, placed in such a way that the outer face of the bank forms a continuous slope with the inner face of the ditch, at an angle of between 30 and 45 degrees. Examples are found at Danebury, Hod Hill and Maiden Castle, where they range between 16 and 25m in width. At Maiden Castle, the Period 3 rampart was reinforced by internal walls of chalk and limestone. On a number of sites glacis ramparts replace stone, timber, or turf revetted banks, which mark earlier univallate hillforts.

Dump ramparts also occur in similar areas and sometimes on the same site. Again these are constructed from dumps of earth and rubble, but are separated from the ditch by a narrow berm. Danebury, where the inner earthwork consists of a glacis rampart, and the middle and outer banks are of dump construction, provides an example.

Ramparts, with a stone revetment on one or both sides, are common in areas where the bedrock is close to the surface. Examples include The Breidden, High Rocks and Salmonsbury. The revetment is always dry-built and usually consists of fairly large stone blocks. These are mostly derived from the immediate area of the site, either from rock-cut ditches or outcrops, although at South Cadbury the Lias slabs must have come from some distance away. On some sites, such as South Cadbury, the revetment is formed by a combination of drystone walling and vertical timbers.

In south-western and western parts of the country, some of the ramparts surrounding large multivallate hillforts are formed by scarps cut into the hillside. These increase the angle of slope, which is usually marked at the bottom by a ledge, terrace, or bank. This form of rampart is found at The Wrekin and Ham Hill. It is sometimes referred to as 'downward construction', and accounts for the presence of banks not accompanied by ditches, which represent accumulations of material thrown downwards from further uphill.

These different rampart forms have a number of features in common. The majority of excavated sections produce evidence for several phases of rampart enlargement. Some sites, such as High Rocks, had palisades along the rampart crests; while others, such as Hod Hill had stone parapets and rampart walks. Palisades may also occur in conjunction with other forms of rampart. At Hod Hill, for example, a palisade was located beyond the inner ditch, and was later reinforced by a wide low bank. Small banks, interpreted as markers for the position of the main earthwork, have been noted on a few excavated sites, such as Bredon Hill.

Ditches are usually located along the outer edge of the ramparts, although on a number of sites they appear to have been used as quarries for banks located further downhill. Alternatively, one ditch may have provided material for two banks. This form of double bank occurs at both Caesar's Camp and Danebury. The majority of ditches on large multivallate hillforts are V-shaped, although shallow U-shaped examples are also recorded. Except on sites with glacis ramparts, they are separated from the bank by a berm. Recorded widths vary between 1.5 and 15.5m, while depths range from 1.2 to 7m. As with the ramparts, the outer ditches on a number of sites tend to be smaller than the inner ditches.

Internal quarry ditches and scoops are found on many large multivallate hillforts, where they provide a convenient source of material for large inner banks. Counterscarp banks, which represent accumulations of material derived from phases of ditch cleaning, are also common features.

Large multivallate hillforts usually have two entrances, which can be single or dual portal in form. There is considerable variation in ground plan between the entrances on different sites, and indeed, between different entrances on the same site. They may appear as simple gaps, or can comprise long passages; bastions; barbicans; guardrooms; overlapping ramparts; hornworks; and outworks. Other associated features include roadways; gates; and bridges. There is usually evidence for multiple alterations within the entrances, which generally become increasingly elaborate with time. In addition, many of the gates and revetments appear to have been destroyed by fire and later replaced.

Simple entrances are marked by a straight approach through each line of earthworks, and are formed by corresponding gaps, of up to 10m in width, in the ramparts and ditches. The width of the ramparts on either side the entrance may be slightly increased, creating a bulbous effect.

In some examples, the length of the passage is extended by parallel ramparts, which project into the interior of the hillfort for distances of up to 40m. These may be revetted by turf, timber, stone, or a combination of any of these materials. At Bredon Hill, the flanking walls on either side of the entrance end in semi-circular bastions. Fairly well constructed drystone walls replace timber and turf revetments on a number of sites. This form of refurbishment is often connected with the construction of additional ramparts, or with the enlargement of existing earthworks.

As an alternative, the length of the approach may be increased by the addition of hornworks, which swing outwards from the main hillfort ditch. At Danebury, the inner hornworks form a passage 6m wide and 46m long. On the same site, the outer hornworks are used to form a forecourt or barbican on either side of the entranceway. Similar features have been noted on a number of other sites. Indeed, the same effect is created on large multivallate hillforts where the main ramparts diverge from one another adjacent to the entrance.

In south-western and western areas, guardrooms, located towards the inner end of entrance passages, are relatively common. They may be C-shaped or rectilinear, and sometimes have internal divisions and hearths. The walls are constructed of timber, stone, or a combination of the two materials.

On a number of large multivallate hillforts oblique approaches to these forms of entrance are created, either by overlapping outer ramparts, or by the construction of hornworks and outworks. The evidence suggests that they represent late additions. Certainly, Hod Hill appears to have been abandoned before the completion of the hornwork. Such earthworks usually comprise ramparts of dump construction and accompanying ditches. Counterscarp banks are also recorded. The hornworks at Hod Hill and Hambledon Hill come from one side only and cross the straight approach to the main entrance, greatly increasing the length of the passages.

Outworks are usually placed centrally across the main access route, creating a staggered and dual approach to the entrance. They may be located in line with or beyond the outer rampart, as at Yarnbury, or within the forecourt defined by multiple hornworks, as at Maiden Castle.

The entrances into the majority of large multivallate hillforts are approached by pronounced hollow ways or terraces. As they pass through the ramparts these appear to have been surfaced by compressed layers of small stones, cobbles, or large slabs and heavy blocks. At Danebury, a palisade ran along either edge of the road as it passed over the ditch terminals.

Gates were often located towards the inner ends of passages, and are frequently marked by large postholes on either side of the entrance. Central postholes, marking double gates, are also recorded. Paired posts, noted on either side of the entranceway on a number of sites, are usually thought to represent the remains of a bridge across the break in the rampart. Some large multivallate hillforts were entered through two sets of gates. At Danebury, for example, a second gate was located adjacent to the outer ditch.

The interiors of large multivallate hillforts have produced evidence for dense occupation. Excavated features include: circular and oval buildings, marked by platforms, posts, stakes, gullies, earth and stone walls and floors; annexes; palisaded enclosures; hearths; ovens; four- and six-post structures; temples; streets; pits; hollows; and a variety of scattered postholes, stakeholes and gullies with no coherant plan. With the exception of four- and six-post structures, rectangular buildings are rare, although examples have been noted at Danebury and Maiden Castle. On the second site this building was internally partitioned.

On a small number of sites, reconstructions of the overall plan of these features have been made possible by open-area excavation or detailed survey. At Danebury, roundhouses tend to concentrate around the periphery of the site, in the lee of the rampart, although they also occur in the central and northern part of the site. Similar arrangements have been noted at Maiden Castle, South Cadbury and Hambledon Hill, where some of the houses are arranged in a linear fashion, possibly along a street. At Hod Hill, roundhouses are more uniformly spaced across the unploughed area of the interior. Roads have been noted within a number of sites, including Danebury, where a number of tracks branch off the main street. Four- and six-post structures at Danebury are mainly in the southern part of the site and are aligned on metalled streets, while storage pits occupy the central and northern zones. A similar clustering of storage pits has been noted in the centre of Yarnbury.

The numbers of circular and oval houses within large multivallate hillforts are high. Two hundred and seven hut platforms were surveyed in the interior of Hambledon Hill. In addition, numbers of similar although rather more ephemeral features were noted. Excavated structures usually produce evidence for superimposition and multiple periods of rebuilding. Within the same site hut platforms may occupy the slopes, while in more level areas roundhouses will have been placed directly on the ground surface. Some, although, not all of these buildings have porches; while others, such as those at Hod Hill have conjoining annexes and/or are placed within palisaded enclosures. Some have a ring of posts, marking central roof supports and others do not. There is also quite a wide variation in the diameter of different houses within the same sites (5-15m). These differences may reflect status distinctions between the occupants of the houses.

The materials used to construct these buildings are varied. In areas where stone was accessible it was used for the outer walls. Elsewhere roundhouses were built with vertical timbers; planks; single or double rings of stakes, occasionally set in gullies; and walls made of clay or a substance very similar to cob. Some of these forms were accompanied by circular drainage gullies. A number of excavated structures have produced quantities of daub imprinted with wattle marks. The floors comprised beaten earth, stone paving, or cobbling. Stone thresholds were uncovered during excavations at Maiden Castle. Hearths, either stone-built or marked by fire-reddened areas of subsoil, are often found within these buildings.

Quite apart from hearths located within roundhouses, a number are found outside these structures. In such situations, they may be surrounded by semi-circular wind-breaks, marked by stakeholes. At Bredon Hill, one of the hearths appears to have been used for bronze-working.

Many large multivallate hillforts include four- and six post structures, measuring between 1.5 and 4m on the long axis. These are usually interpreted as raised granaries. At Danebury, a single example with 5 posts, and several with 9 were also noted.

Temples have only been noted on a small number of sites including Danebury, South Cadbury and Maiden Castle. At Danebury and South Cadbury these were rectangular and constructed in timber; at Maiden Castle the temple was circular and constructed in stone. At both South Cadbury and Maiden Castle these buildings were approached by roadways. The temple at South Cadbury was associated with numerous animal burials and shallow pits, containing iron objects.

Storage pits are most common in large multivallate hillforts in southern areas of the country. Large numbers are often found within the same site. At Danebury, for example, 1,122 pits were revealed in an area representing 23 percent of the interior. The profile of storage pits is varied and includes beehive, conical, sub-rectangular, or cylindrical shapes. The diameters range between 0.5 and 2.2m and the depths between 0.3 and 3m. Many appear to have been recut, with a dividing wall separating adjacent pits. At Maiden Castle, a layer of clay at the top of these features was thought to represent the remains of seals. Once they ceased to function for storage, many of the pits were backfilled with a variety of rubbish. A small number were used as burial places. At Danebury and South Cadbury, 'special deposits' of animal bones, particularly the skulls of cattle and horses, were recovered from some of the pits.

Shallow hollows, usually with irregular outlines and with diameters of between 8 and 15m, are also recorded on a number of large multivallate hillforts. Their purpose is rather obscure.

Various schemes have been put forward, which divide large multivallate hillforts into groups according to the size of the internal area and the number of ramparts. While the first seems sensible, the second is open to criticism. In the first place counting ramparts, allowing for sites where some of the earthworks are incomplete, results in the creation of groups which are only relevant to single sites. In the second, it fails to recognise the fact that some bivallate earthworks required an equal, and sometimes greater amount of labour, than sites with a larger number of fairly slight ramparts. It seems more useful to adopt a scheme which allows for an estimation of the scale of the earthworks, and which makes allowances for the wider spacing encountered on some sites. With these considerations in mind three main types can be recognised (see Figure 1). These are:

  1. Large multivallate hillforts with an internal area of between 5 and 10ha.
  2. Large multivallate hillforts with an internal area of between 12 and 29ha.
  3. Large multivallate hillforts with an internal area of between 40 and 85ha.

These can be divided into 5 sub-types (see Figure 1):

  1. Sites with closely spaced earthworks measuring between 15 and 30m in width (eg. High Rocks and Ravensburgh)
  2. Sites with closely spaced earthworks measuring between 35 and 70m in width (eg. Hod Hill and Salmonsbury)
  3. Sites with closely spaced earthworks measuring between 75 and 140m in width (eg. Maiden Castle and Yarnbury).
  4. Sites with widely spaced earthworks measuring between 90 and 110m in width (eg. The Breiddin and Bredon Hill).
  5. Sites with widely spaced earthworks, defining an additional area in excess of 5ha (eg. The Wrekin and Danebury).

This results in 12 possible permutations, since Type 3 only occurs as sub-types A and B.

Three large multivallate hillforts, including South Cadbury, Hod Hill and Ham Hill were used for a short time during the early Roman period as military encampments. In fact, Hod Hill has a 4 acre Romano British fort and associated buildings in the NW corner of the its interior. After a period of abandonment, some sites were re-used during the late Roman period. Ham Hill became the site of a villa and of quarries for building stone, while Maiden Castle and South Cadbury were occupied by 3rd and 4th century temples. It is worth emphasising the point that these hillforts were also the sites of similar Iron Age buildings.

South Cadbury was reoccupied during the 5th and 6th centuries, while the rampart was refortified with the addition of a timber fighting platform. By the 11th century this same site had become a burh, defended by a masonry wall and bank, with monumental gateways.

During the Iron Age large multivallate hillforts formed centres of permanent settlement. The size of the areas enclosed and the scale of the earthworks suggests that they must have occupied a fairly elevated position, in relation to other contemporary classes of hillfort and settlement. Their occupants were obviously able to mobilise the large labour force, necessary for construction on such an enormous scale. This may have had as much to do with displays of power as with defence. Evidence of violent action has been uncovered on a few sites. Numbers of bodies, apparently of young adult males, have been discovered in the entrances of Maiden Castle, South Cadbury and Bredon Hill. Burning in entrances usually provokes discussions of skirmishing, although accidental fires could equally account for such evidence. The caches of slingstones, recovered on many large multivallate hillforts, are also thought to be connected with defensive action. Some large multivallate hillforts, still in use at the time of the Conquest, may have been attacked by Roman troops. At Hod Hill, for example, 11 iron ballista bolts were found in the south-east angle of the hillfort.

In general, the jewellery, weaponry and tools recovered from large multivallate hillforts, seem indicative of the high status of their occupants. They were certainly able to draw on fairly wide ranging contacts. The finds often include items from well outside the immediate locality. Examples include the briquetage at Danebury; and the Gallo-Belgic coins, from Kent and Europe, at Hod Hill. Iron currency bars are particularly common, and their presence suggests a control over circulation. Indeed, bronze and iron working seems to have been taking place within some sites. Other manufacturing activities included spinning; weaving; potting; the working of bone and antler tools and ornaments; and the processing of agricultural produce.

Distribution and Regional Variation

Large multivallate hillforts are recorded in: the Weald, in East Sussex and West Kent; the chalk downs, in Kent, Surrey, Hampshire, Berkshire, Wiltshire and Dorset; the Mendips, in north Somerset; the Cotswolds, in Avon, Gloucestershire, and Warwickshire; the Malverns and the Welsh Marches in Hereford and Worcester; the Welsh Marches, in Salop; the Chilterns, in Buckinghamshire; the Northamptonshire Uplands; the East Anglian Heights, in Hertfortshire and Cambridgeshire; west Essex; and central Nottinghamshire. The main concentration of large multivallate hillforts occurs in Dorset, Wiltshire and west Hampshire, with a second, although smaller group, in Salop. The distribution of the remaining sites is fairly scattered, and at its northernmost extent is represented by a single outlier in central Nottinghamshire.

The majority of large multivallate hillforts are located on escarpments or spurs, above one or more major river valleys. Both Hod Hill and Hambledon Hill, for example, overlook the Stour and Iwerne, which provide access routes inland, from the coast to the chalk downs. Bredon Hill is situated above the Avon, Yarnbury above the Wylye, High Rocks above a valley leading to the Medway, and so on. The only site which is actually placed on low ground adjacent to a river is Salmonsbury, which lies in an angle between the rivers Windrush and Dikler, in Bourton Vale.

A smaller number of large multivallate hillforts are located at a greater distance from river valleys. Danebury, for example, is placed on the highest point of the chalk plain, 5km west of the valley and floodplain of the River Test. In situations of this nature, sites often occupy hills which rise above the surrounding countryside, even if the difference in height is not particularly great.

In Dorset, Wiltshire, west Hampshire, and Salop large multivallate hillforts are spaced at intervals of 15 to 20km or more. Exceptions are found in a cluster, which extends from north-east Dorset into south-west Wiltshire. Here, the distance between sites falls below 5km. Examples include Hod Hill and Hambledon Hill, which are around 3km apart; and Bilbury Rings, Hanging Langford and Yarnbury, which are each separated by about 2km.

Elsewhere in the country the distribution is more scattered and examples are spaced at intervals which exceed 20km. Exceptions are recorded, and include Midsummer Hill and Herefordshire Beacon which are separated by less than 5km.

All large multivallate hillforts with widely spaced ramparts, belonging to sub-type D are located towards the western part of the distribution. Those with closely spaced earthworks, belonging to sub-type C, tend to concentrate in Dorset and Wiltshire. The very large sites belonging to Type 3 are rare. Single examples occur in Northants, Berkshire and Somerset. Most areas have a combination of sites belonging to both Types 1 and 2. Exceptions include regions where the smaller sites, belonging to Type 1, predominate, as in Wiltshire; or are the only type found, as in Hereford and Worcester, Avon, Hampshire, Buckinghamshire, Hertfordshire, Cambridgeshire and Surrey.

Large multivallate hillforts which were used into the first century AD are located mainly in Gloucestershire, Avon, Somerset and Wessex. Evidence for early Roman military occupation is found on sites in Dorset and Somerset.

Some of the components described in Section 3 tend to occur more frequently in certain regions. Glacis ramparts are most common in southern and eastern parts of the country, while scarping is a feature of sites in south-western and western areas. Similarly, outworks and hornworks are found on large multivallate hillforts in southern England, while guardrooms occur towards the western edge of the distribution.


Around 50 large multivallate hillforts are recorded in England. An estimated 75% of these sites fall within the size range described by Type 1, with about 20% belonging to Type 2. Only three large multivallate hillforts of Type 3 are known. A large proportion of hillforts (40%) belonging to this class are located in Dorset, Wiltshire and west Hampshire.

Survival and Potential

The extent to which the appearance of large multivallate hillforts closely resembles their original form, depends mainly on later landuse and on the scale of the ramparts. A number of large multivallate hillforts have been used as quarries for building stone, others are occupied by forestry plantations. Examples on military land may be damaged by trench digging or by the movement of heavy vehicles. Different areas of the site may support different forms of vegetation. Parts of the ramparts and interiors of many sites, for example, are covered in woodland or scrub.

On sites surrounded by massive Iron Age earthworks, the crests of banks may still be 10 to 15m above the top silts of the ditches. In contrast, smaller ramparts now appear as low features of between 0.5 and 2.5m in height. These are relatively easily levelled by agricultural activity, and on many sites appear in part as crop or soil marks, or are entirely absent in certain areas. The outer earthworks are particularly vulnerable to ploughing and have been removed, either partially or completely, on a number of large multivallate hillforts. This means that it is very difficult to identify examples which were originally incomplete.

The structure of the ramparts will also affect survival. Glacis and dump ramparts tend to retain their original profile, although a degree of slumping is likely. Timber revetments will obviously have rotted, while the upper courses of stonework will have tumbled. Collapse is seen most frequently above exposed and steep scarps.

Certain components are more vulnerable than others and are, therefore, found on fewer sites. Features associated with the crest of the rampart, such as palisades, parapets, or walkways, are unusual, simply because this area rarely survives. Counterscarp banks, being fairly insubstantial features, are rare on sites in areas where intensive agriculture is practices. Palisades adjacent to the ditches will also be removed by this kind of activity. In contrast, the ditches usually remain relatively unaltered.

The earthworks surrounding many large multivallate hillforts are broken by gaps which were not part of the original design. These are usually fairly easy to identify, since the causeways across the ditch tend to slump. Yarnbury is an example, where at least one break in the ramparts is associated with the annual fair held on the site until 1916.

The ramparts on either side of the entrance will usually have slumped, and where timber or stone revetted, may have collapsed, filling the original roadway with debris. In rare cases it may be possible to identify component features. At The Wrekin, for example, the guardrooms were identified before excavation. Many of the entrances into large multivallate hillforts have been used repeatedly since the Iron Age as points of access. This will have led to the erosion of the central part of the passage and to the removal of any features in this area.

Features within large multivallate hillforts are far more vulnerable to erosion, and in many cases the interiors will survive in a very different condition from the ramparts and ditches. The most common forms of earthwork in this area include platforms; internal quarry scoops or ditches; and the backfilled ditches of earlier univallate hillforts. These are only visible within the minority of sites. Later activity, and particularly ploughing has removed all traces of such features within the majority of large multivallate hillforts. In some cases the plough has cut into the tail of the inner rampart.

The majority of excavations on large multivallate hillforts have consisted of narrow trenches through the earthworks and within the entrances. On some sites these were extended into the area immediately behind the inner bank. Excavations of this nature usually produce evidence for several phases of rampart construction. These are separated by turflines, or marked by changes in structure, such as the replacement of revetted banks by those of glacis form. Artefacts from the separate phases, may allow for the dating of such developments. If the rampart was originally timber-laced, voids or carbonised wood may be revealed. Slumping of soil and rubble along the outer edges of the bank will protect palisades trenches or postholes, where these occur. In situ drystone revetment is often recovered in the same area.

Additional features, such as internal revetment walls, or structures along the crest of the rampart, may also be uncovered. Buried soils are preserved below most ramparts. These may contain environmental evidence, including pollen and snails, which allow for the reconstruction of the pre-enclosure landscape. Since most sites are located on alkaline, calcareous soils, snails are more likely to be preserved than pollen. Some excavations have also produced artefacts and traces of pre-hillfort occupation from this area. These, together with samples which might be suitable for radiocarbon assay, provide some sites with a terminus ante quem. This same evidence may indicate different periods of construction for the various multiple ramparts surrounding a site.

Many ditches show signs of recutting. They also contain evidence for rampart revetment, most frequently in the form of large blocks of tumbled stone, but occasionally this appears as charred timbers and planks. Dumps of occupation debris have also been recovered from a number of ditches, particularly in the terminals on either side of the entrance. These deposits give an indication of the total date range of the site being investigated. As with similar evidence from the ramparts, this may reveal different dates for different lines of defence.

Most excavations within the entrances of large multivallate hillforts have uncovered postholes associated with gates, many of which were replaced on multiple occasions. These are usually large, up to 1m wide and 1.5m deep, and where protected by the slumping of the bank, are in good condition. Where the surface of the passage is filled with tumbled debris, features in the central area have been uncovered. These include slots and postholes associated with gates and bridges, together with the remains of road metalling. Replacements of the revetment on either side of the entrance passage are usually discernable. Indications of major restructuring, such as the enlargement of ramparts, the extension of passages and the addition of hornworks are commonly uncovered in this area. The burning the entrances, which occured on a number of large multivallate hillforts, may allow for the recovery of material suitable for radiocarbon assay.

Excavations extended from the earthworks into large multivallate hillforts usually reveal a series of stratified deposits against the inner rampart. The few open-area excavations, which have taken place on these sites, have produced similar evidence. Up to 1m of soil may have accumulated behind the rampart. Internal quarry scoops are often used initially as convenient dumping areas and then for occupation. Buildings in such locations show a long history of reconstruction and replacement. Their association with stratified deposits, which generally contain large numbers of artefacts, allows for the detailed phasing of the site. Similar deposits are likely in hollow areas, in large pits and in the ditches of earlier and much smaller univallate hillforts, backfilled when certain sites were remodelled in multivallate form.

Quite apart from open area excavations information concerning potential is also revealed by detailed survey. On a few sites, ring grooves marking roundhouses, and other features such as internal streets, are still visible from the surface. Excavation on these same sites has demonstrated that it is possible to recover upstanding clay or decayed cob walls in these situations. House floors and the lower parts of ovens and hearths may also be revealed. Exceptions to this occur on steep slopes, where the run-off of surface water has resulted in the removal of occupation layers. On most sites the soil overlying these structures is between 0.15 and 0.6m in depth. The chances of recovering positive features is much reduced in the majority of situations, since the penetration of the average plough is around 0.3m. Several seasons of ploughing and resulting erosion, particularly on sloping ground, will cause considerable damage. It is not unusual to find truncated pits and postholes of between 0.15 and 0.3m in depth. Gullies and stakeholes are often removed altogether in these situations. The only remaining features may be the large postholes, marking four- and six-post structures, and the deep storage pits. Undamaged stake-built structures are extremely rare. If they do survive, it is usually in hollow areas subsequently filled with soil and therefore protected from ploughing. Parts of the site protected by scrub or woodland may have escaped these effects. Although tree roots disturb shallow deposits, the effects are usually very localised.

Large numbers of artefacts are usually recovered from large multivallate hillforts. Since most of the sites are located on calcareous soils, metalwork, pottery and bone is generally in good condition, particularly when recovered from the deeper negative features. The range of finds is wide and includes: ceramics; a variety of bone objects, such as gouges, awls, needles, points, toggles, dice and other gaming pieces; beads of amber, coral and glass; bronze jewellery; iron jewellery, weapons and tools; bone, bronze and iron harness fittings and lynch pins; slingstones; Gallo-Belgic coins; artefacts associated with bronze working, such as crucible fragments; evidence for iron-working, such as iron currency bars, iron slag and bloom; artefacts associated with spinning and weaving, such as bone weaving combs, and clay, bone or stone loomweights and spindle whorls; and objects used for grain processing, such as quernstones and rubbers. Animal bones, particularly of sheep and cattle, are generally recovered in large quantities. Some are placed in 'special' deposits in storage pits. Inhumations are occasionally recovered from the same contexts, while scattered human remains are found on most sites.

Environmental evidence in the form of charred grain and seeds, is often recovered from hearths, ovens and storage pits. These same contexts, together with deposits in quarry scoops, may produce stratified material suitable for radiocarbon assay, microfaunal or pollen analysis.

It is worth taking into account the possibility that extra-mural settlement may occur in association with some sites. This has been uncovered, by accident, outside a number of hillforts, although none of them belong to the large multivallate class. It seems most likely where flattened areas are left outside the defences, as at Caesar's Camp in Kent. The wide spaces between the ramparts of hillforts belonging to Sub-Type E may also have been occupied. Again this has yet to be proven by excavation.


Both stratigraphic and spatial associations are recorded between large multivallate hillforts and other classes of monument. Broadly contemporary classes include: slight univallate hillforts; large univallate hillforts; hilltop enclosures; enclosures of various classes; field systems; cross dykes; and linear earthworks.

As already described, within section 2, a number of large multivallate hillforts replace univallate forms. Examples include Yarnbury, which replaces a slight univallate hillfort; Maiden Castle which replaces a large univallate hillfort; and Ham Hill which may replace a hilltop enclosure. On the majority of sites the earthworks surrounding the earlier hillfort are re-built on a much larger scale. A smaller number of examples are constructed on the site of a univallate hillfort, but with a greatly increased internal area. Here the multivallate ramparts partly follow the course of the univallate, and partly extend well beyond the original hillfort on one side. Earlier univallate earthworks within the newly defined interior are usually levelled and often become the focus for later occupation. On other sites, such as Yarnbury, the smaller univallate hillfort lies entirely within the multivallate earthworks.

Quite apart from stratigraphic associations, numerous spatial relationships between large multivallate hillforts and other classes of hillfort are also recorded. These can be defined where two sites are located at intervals of less than 5km.

Spatial associations with enclosures of various classes are also noted in the records. Examples include the small double banked enclosure on the ridge adjacent to the Breiddin, and the sub-circular enclosure within the interior of Bury Wood, which from the finds seems to be contemporary with the hillfort.

Regular aggregate field systems have been noted adjacent to a number of sites, including the Breiddin and Bury Wood. A similar system is located partly under the rampart of Hambledon Hill.

Cross-dykes have been noted in association with several large multivallate hillforts. At Caesar's Camp, for example, a large ditch crosses the only level approach to the site. Associations with linear earthworks are less common, and include Silbury Hill, which is placed at the confluence of several linear ditches, and Old Oswestry, where Wat's Dyke is aligned on the hillfort.

Large multivallate hillforts are also associated with earlier monuments including: causewayed enclosures, at Hambledon Hill and Maiden Castle; a system of Neolithic outworks, at Hambledon Hill; long barrows, at various sites including Hambledon Hill; a bank barrow, at Maiden Castle; a pit circle, at Maiden Castle; and round barrows of various classes, at a number of sites including Hod Hill.

Associations with post-Iron Age monuments include: a Romano British fort, in the north-west corner of Hod Hill; early Romano British military buildings, at South Cadbury and Ham Hill; 3rd and 4th century AD temples, at Maiden Castle and South Cadbury; a 3rd and 4th century AD villa, at Ham Hill; a late Roman cemetery, at Ham Hill; early Saxon settlement, at South Cadbury; an 11th century burh, at South Cadbury; and Saxon inhumations, at a number of sites including Salmonsbury.

Characterisation Criteria

The four criteria for assessing class importance apply to large multivallate hillforts as follows:

Period (currency): Long-lived. Large multivallate hillforts were constructed and used for a period of about 600 years, between the 6th century BC and the mid-1st century AD. Early dates usually refer to univallate phases, although a few multivallate sites were built during the 6th/5th centuries BC. Salmonsbury is the only example, which seems to have been occupied continuously into the 4th century AD. Individual sites were used for lengths of time varying between 200 and 500/600 years. Large multivallate hillforts occupied for up to 500 or 600 years are confined to Somerset and Wessex.

Rarity: Rare. Around 50 large multivallate hillforts are recorded in England. Forty percent of these sites are located in Dorset, Wiltshire and west Hampshire, with a second concentration in Salop. Large multivallate hillforts are less common elsewhere.

Diversity (types): Very high. Three main types and five sub-types have been distinguished. Since Type 3 is only associated with 2 of the sub-types, the number of possible permutations is 12. Further variation has been noted in shape, and in the presence or absence of specific component features.

Period (representativity): Medium. Large multivallate hillforts represent one of a fairly wide range of monument classes known to characterise the Iron Age period.

Assigning scores to these criteria following the system set out in the Monument Evaluation Manual, large multivallate hillforts yield a Class Importance Value of 55. This lies towards the top end of the range of possible values (max. = 64), reflecting the large number of sites which are likely to be of national importance. Examples representing each of the types defined above, in a variety of regional and topographic situations, should be included in the sample of nationally important sites. Wherever possible, large multivallate hillforts with different component features should be incorporated.


Cunliffe, B. 1984. Danebury: An Iron Age Hillfort in Hampshire (= Council for British Archaeology Research Report 52, Volumes 1 and 2). London. CBA.

Cunliffe, B. 1984. 'Iron Age Wessex: continuity and change'. In B Cunliffe and D Miles (eds), Aspects of the Iron Age in Central Southern Britain (= Oxford Committee for Archaeology Mongraph 2). Oxford. 12-45.

Dunning, G. C. 1976, 'Salmonsbury, Bourton-on-the-Water, Gloucestershire'. In D W Harding (ed), Hillforts: Later Prehistoric Earthworks in Britain and Ireland. London. Academic Press. 75-118.

Mercer, R. 1986. Hambledon Hill - the Iron Age hillfort and excavations 1986. Interim Report.

Richmond, I. A. 1968. Hod Hill Volume 2: Excavations Carried Out Between 1951 and 1958 for the Trustees of the British Museum. London. Trustees of the British Museum.

Sharples, N. 1985. Maiden Castle Project 1985: an interim report. Dorset Natural History and Archaeological Society Proceedings, 107: 111-120.

Sharples, N. 1986, Maiden Castle Project 1986: an interim report. Dorset Natural History and Archaeological Society Proceedings, 108: 53-62

Wheeler, R. E. M. 1943, Maiden Castle, Dorset (=Reports of the Research Committee of the Society of Antiquaries, 12). London.


Description prepared by F Raymond, January 1989. Release 00, January 1989.

Top of Page