1 Hillfort studies and the Wessex Project by Andrew Payne

The Wessex Hillforts Project was initiated in effort and organisation must have been 1996 to answer a need for more wide-rang­ involved in their construction, the reasons ing data on hillfort interiors for the purposes why they were constructed are more difficult of placing their future management on a to comprehend. The term hillfort has been sounder footing and enhancing knowledge applied to many different types of site and of the internal character of the various hill- their varying sizes, morphologies and situa­ fort types represented in Wessex. It was tions strongly suggest a range of different hoped that the combined results of the pro­ motives for their construction, spanning a ject would considerably extend academic considerable date range (Fig 1.1). understanding of the socio-economic role of We usually associate hillforts with the hillforts in southern England during the 1st Iron Age, the period when many new hill- millennium BC, thereby allowing a greater forts were built, but the origins of hillfort level of interpretation to be offered to visi­ building lie at least as far back as the Bronze tors at those sites with public access. Age. During the 800 years before the The primary methodology employed by Roman invasion of Britain (the period that the project was geophysical survey supple­ we conventionally term the Iron Age) the mented by examination of aerial photo­ role of hillforts seems to have changed. New graphic evidence, documentary research and evidence is only gradually being uncovered selective digital modelling of site micro- that helps to extend our understanding and topography. The examination of each hillfort we still have very little information about was to be as comprehensive as possible with­ hillfort interiors in general and the range of out resorting to more costly and unnecessar­ functions they might have fulfilled. ily destructive intrusive techniques. Generally, but not exclusively, set on ele­ vated or other locations conferring natural The context of the study defensive advantages, sites classed as hill- forts in southern Britain can range in size Hillforts have attracted archaeological inter­ from less than one hectare to many tens of est for much of the last century and debate hectares. Their structural complexity varies on their function and significance continues from simple univallate earthworks to vast to be central to the academic study of the multivallate fortresses with labyrinthine later Bronze Age and Iron Age (broadly the entrance passages. Although hillforts are 1st millennium BC). Although some hillforts among the most numerous of all our surviv­ have been damaged by development or ing prehistoric monuments – nearly 1500 levelled through ploughing, those that were listed in the Ordnance Survey’s 1962 remain are some of the most impressive Map of Southern Britain in the Iron Age alone ancient monuments still visible in the coun­ (Fig 1.2) – our knowledge of the majority of tryside today. Such prominent landmarks sites is still quite limited because often their naturally attracted the interest of antiquaries sheer scale is such that there have seldom and pioneers in archaeology from earliest been sufficient resources for extensive exam­ times, an interest that has continued with ination of their interiors. the development of scientific field tech­ Conventionally, hillforts have always niques and modern methods of excavation. been seen as primarily constructed for Writing on social organisation in Iron Age defence, but their disparate sizes, topo­ Wessex, Haselgrove (1994, 1) concluded, graphical settings and architectural forms, ‘there can be little doubting the significance suggest that this need falls far short of pro­ of Iron Age hillforts, given the labour viding a wholly adequate explanation for all invested in their construction, so under­ of them (Harding 1979; Ralston 1996). The standing their role is clearly vital’. While it is vast majority of the sites examined in this clear from the scale of these sites that great project are classic hillforts occupying highly

1 Fig 1.1 1. LARGE HILLTOP ENCLOSURES The major categories of hillfort types represented in Southern Britain illustrating the broad three-phased development of hillfort forms within the region of Wessex (from Cunliffe 1991 and Sharples 1994).

2. EARLY IRON AGE HILLFORTS

A B C G H I

Based on Darvill 1987, Figure 80. N

D J K L

N

M Based on Cunliffe 1991, Figure 14.24

E F

Based on Sharples 1994, Figure 26.2 3. DEVELOPED HILLFORTS

O P Q R

rampart

A Bathampton Down, Avon B Balksbury Camp, Hampshire C Martinsell Hill Camp, Wiltshire G Quarley Hill, Hampshire K Danebury, Hampshire O South Cadbury Castle, Somerset D Norbury Camp, Gloucestershire H Yarnbury Castle, Wiltshire L The Trundle, W Sussex P Maiden Castle, Dorset ALL PLANS TO SCALE I Chalbury, Dorset M Maiden Castle, Dorset Q Hod Hill, Dorset E Walbury Camp, Berkshire 0 500 1000m F Bozedown Camp, Oxfordshire J Figsbury Rings, Wiltshire N Blewburton Hill, Oxfordshire R Hambledon Hill, Dorset HILLFORT STUDIES AND THE WESSEX PROJECT

1.2 - 6 hectares Over 6 hectares

0 200 kms

visible elevated positions dominating their and chronology and, with a few exceptions, Fig 1.2 surroundings (such as ridge ends or escarp­ fieldwork was concentrated on the compara­ Hillfort distribution in south­ ment edges), where the hillfort ramparts tively small-scale excavation of hillfort ern Britain (based on Cun­ enhance an already naturally defensible defences and gate structures. The question liffe 1991 without revision) position. A minority of the sites examined of the function of the hillfort in its social and – not intended to be defini­ possess defences that are of hillfort propor­ economic environment was hardly voiced tive. Non-verified, less visible tions but are situated in locations that confer (Collis 1981, 66). hillfort-type sites probably little or no altitudinal advantage. Clearly Although some hillforts had been dug exist in the survey area; defence was not always the primary consid­ into before 1900 by pioneers of field archae­ evidence for some is discussed eration and it is likely that the wide spec­ ology such as Augustus Lane Fox (better in Chap 2. Classification as trum of sites to which we apply the term known as Pitt Rivers), it was not until the hillforts of newly or recently hillfort performed a range of functions of early years of the 20th century that archaeo­ identified ploughed-out sites which defence was but one. logical interest was sufficiently awakened for depends on how strict our Until the 1960s hillfort studies were major campaigns of excavation to be organ­ definition is. ‘Hillfort’ is often dominated by problems of cultural affinity ised on regional groupings of sites. Between applied loosely to some low-lying sites and sites of less obvious defensive character. THE WESSEX HILLFORTS PROJECT

1907 and the 1940s the combined work of 4th century BC. Originating from Spain and Maud Cunnington in Wiltshire, E Cecil Brittany (Armorica) these invaders initially Curwen in Sussex and Christopher Hawkes thrust into the western parts of Britain, in Hampshire was instrumental in trans­ spreading into Dorset and the Cotswolds, forming knowledge of the many examples of where they built hillforts characterised by hillforts in these areas. The lack of a profes­ massive multivallate defences. This second sional infrastructure and resources for fund­ wave was assigned to the Iron Age B period. ing and employing archaeological staff at Finally, some time around 75 BC, Belgic this point in time did not allow for long term invaders entered the Thames Valley and or extensive programmes of archaeological Kent, spreading into Essex, while a little investigation. They nevertheless provided a later, as a result of Caesar’s military con­ useful sample of evidence from a large num­ quests in Gaul, refugees from northern ber of sites. France landed on the shores of the Solent The first serious attempt to bring and moved into central southern Britain. together the evidence amassed through these These invaders were defined as the Iron Age excavations in a nationwide synthesis was a C peoples. During this period in the south­ paper entitled simply ‘Hill-Forts’ published east of England, hillforts declined and dis­ by C F C Hawkes in the journal Antiquity in appeared to be replaced by large fortified 1931. The paper reflected the historical par­ towns, usually in more low-lying situations adigm then current among prehistorians, commanding river crossings, as for example which sought to explain changes in the at Orams Arbour in Winchester (Whinney archaeological record and defensive architec­ 1994). In territory that fringed the areas of ture at hillfort sites during the Iron Age as a Iron Age C penetration, such as Dorset, the product of successive waves of population continuation of old style hillforts marked movements (or invasions) from continental native resistance to the Belgic influence. Europe (Hawkes 1931; Wheeler 1943). Under the historical paradigm the most Invasionist theories of this nature are no important question was ‘when?’ and longer widely accepted as the explanation involved the dating of hillfort horizons as for cultural change in the British Iron Age, indicators of political change. The excava­ but at the time they seemed to provide a tion methods of the pre-Second World War plausible model against which to interpret era were almost entirely orientated to this the archaeological evidence. The view that problem with great emphasis on the trench­ there had been large-scale invasions in the ing of ramparts and the clearance of prehistoric period had analogies with the entrances, but little work on the interiors historical period with its invasions of Nor­ (Collis 1981, 66). Excavations of this nature mans, Vikings, Saxons and Romans; and provide information concerning the Caesar, writing of Britain in the 1st century chronology and structural history of individ­ BC, talked of incursions of Belgae from ual sites and are a necessary prelude towards northern France and the Low Countries understanding a site, but were rarely taken into the south-east of the country. It was forward to include investigation of the inte­ against this background that Christopher rior on a scale sufficient to enable the recon­ Hawkes in 1931 proposed a three-phase struction of buildings, structures and chronological system – the ABC of the features in the hillfort, let alone the spatial British Iron Age – to explain the various organisation of the interior. The first serious stages of hillfort development in southern attempt to open up large areas of a hillfort England. This system was to form the basic interior was Sir Mortimer Wheeler’s excava­ chronological framework for hillfort studies tion at Maiden Castle in the late 1930s for the next 30 years or more. (Wheeler 1943). The view propounded by the ABC sys­ Hawkes’s ABC scheme, further elabo­ tem envisaged a movement of Celtic peoples rated by Gordon Childe and others (Childe from central and northern Europe spreading 1935, 1946; Piggott C M 1950; Piggott S into the south-east of Britain in the 6th cen­ 1966) to embrace Iron Age defensive struc­ tury BC and fusing with the native populace tures in the whole of Britain, found general to form the Iron Age A culture. This period acceptance and influenced most hillfort was associated with an initial phase of wide­ research published before the mid-1970s. spread hillfort building activity in central- However, with the increased use of radio­ southern and south-east England. The next metric dating and a changing theoretical stage of the scheme involved the arrival of a stance from the late 1960s onwards this par­ second wave of invaders arriving early in the adigm of invasion and response fell out of

4 HILLFORT STUDIES AND THE WESSEX PROJECT favour. Regional developments are now gen­ The post-war period saw the emergence erally agreed to have been more influential of open area excavation and a growing inter­ on the growth of hillforts, including the est in both the form of occupation within demonstration of prestige or status on the hillforts and in the economic and social part of the hillfort builders – or more partic­ stimuli that led to their development. In the ularly the decision makers who controlled 1960s and 70s, the realisation that the social their activities – as well as the wish to give and economic functions of hillforts could physical definition to the limits of jurisdic­ only be addressed through an understand­ tions (social, ritual, economic or political) ing of their internal layout led to the large (Ralston 1996). scale excavation of a number of hillfort inte­ Since the collapse of the historical para­ riors including South Cadbury in Somerset digm, a new chronological framework has (Alcock 1968a, 1968b, 1969, 1970, 1980; only slowly begun to be developed. Unlike Barrett et al 2000); Crickley Hill in Glouces­ Hawkes’s system and those tied to it, there is tershire (Dixon 1976, 1994); Croft Ambrey, now no single chronological scheme that can Credenhill and Midsummer Hill in the be applied to hillfort development over the Welsh Marches (Stanford 1967, 1974; Stan­ whole of Britain and currently we only have ford 1971; Stanford 1981); and Balksbury, a detailed comprehension of the chronology Winklebury and Danebury in Hampshire of hillfort development in certain regions of (Wainwright and Davies 1995; Smith 1977, Britain where sufficient research has been 1979; Cunliffe 1984a, 1995, Cunliffe and carried out. Prior to the use of radiometric Poole 1991). Despite the increased atten­ dating, earlier pre-war dating saw hillforts as tion given to hillfort interiors since the a relatively late development after 600 BC; 1960s, only a very small proportion nation­ most were not built until after 300–250 BC ally have yet been investigated on anything and multivallate forts not until after 50 BC. approaching a reasonable scale. The prob­ These dates are now known to be wrong, lem has been accentuated by the general with radiocarbon evidence linked to changes lack of success of aerial photography at in pottery form and decoration. It is evident revealing features inside hillforts, even when that some hillforts were occupied as early as they are regularly ploughed and cultivated, the Late Bronze Age and many more date often in contrast to their surrounding land­ from as early as the 7th or 6th centuries BC. scapes. This continuing lack of extensive At the same time it has become clear that data is reflected in the most recent compre­ many, if not most, hillforts in southern Eng­ hensive survey of hillfort studies (Cunliffe land were abandoned round about 100 BC 1991) where much of the discussion of the (Atrebatic area) or shortly thereafter (Durot­ available evidence continues to revolve rigian area). This dramatic shift in possible around the morphology of hillfort defences. time-span has superseded the chronologies Within the small sample of hillforts that in many older excavation reports, adding have been examined on a sufficient scale for considerable confusion to an already com­ the nature and density of their internal fea­ plex picture. tures to be adequately characterised, there is From the 1960s onwards, following the considerable variation in the complexity of abandonment of the Hawkes ABC system of internal characteristics and intensity of culture change, an increasing concern with occupation. Some sites reveal evidence of the definition of hillforts led to the appear­ free-standing buildings within their enclosed ance of a number of proposals for their clas­ areas while others contain few traces of sification. These rested mainly on the occupation. The latter group are believed to structure and placement of the ramparts, have served a variety of purposes including a siting (for example cliff-edge forts) and the range of agricultural uses (such as coralling size of area enclosed (see, for example, Avery of livestock), settings for ritual or display 1976). Closer consideration of such evi­ and as temporary refuges (Ralston 1996). dence suggests, however, that any typology Some of the earliest known Wessex hillfort based on shape and situation will be an sites such as Balksbury in Hampshire oblique record of the local topography and (Wainwright and Davies 1995; Cunliffe may carry little archaeological significance. 2000) contained very few internal features Much of the discussion on hillforts still (Fig 1.3). This suggests that they performed focuses on the form of construction of the a very different function from the later hill- hillfort ramparts and less on internal charac­ forts, such as Danebury (Fig 1.4) and ter, which is generally more elusive without Maiden Castle, that developed in the early resort to excavation. Iron Age but continued in use into the

5 THE WESSEX HILLFORTS PROJECT

Fig 1.3 Plan of all excavated features inside Balksbury Camp, Andover, Hampshire (from Wainwright and Davies 1995).

6 HILLFORT STUDIES AND THE WESSEX PROJECT

Middle Iron Age by which time they were earthwork surveys (Corney 1994) have intensively occupied and strongly defended tended to form part of more geographically fortress town-like settlements with struc­ restricted archaeological surveys of particu­ tures laid out on a rudimentary street-plan lar landscapes rich in cultural remains (see (Sharples 1991; Cunliffe 1984a, 1995, Cun­ for example McOmish et al 2002; Riley and liffe and Poole 1991). Wilson-North 2001), thematic studies of Often over-shadowed by excavation, regional or national distributions of specific non-invasive archaeological techniques, led monument types (see for example Oswald et by analytical earthwork survey continue to al 2001), or casework and project led sur­ make an important contribution to broad­ veys of individual sites such as Maiden Cas­ ening understanding of hillforts through tle, South Cadbury Castle and Cissbury detailed mapping and investigation of their (Balaam et al 1991; Riley and Dunn 2000, surface remains. Deserving of mention in Donachie and Field 1994). The historical this respect are the numerous hachured sur­ contribution of earthwork survey to the veys of hillforts undertaken by the Royal study of hillforts is discussed in greater Commission on the Historical Monuments depth in Chapter 3. More recently, geophys­ of England in the counties of Dorset, Wilt­ ical survey has played an increasingly signifi­ shire and Hampshire and the work of the cant role in revealing patterns of occupation former Archaeological Division of the Ord­ inside hillforts that complements the evi­ nance Survey (working between the 1920s dence obtainable from the study of the sur­ and 1970s) on whose surveys the majority of viving earthwork evidence. Traditionally the plans in this volume are based. The used as an aid to the planning and targeting RCHME surveys were initially undertaken of excavations, as at South Cadbury in the for county inventories in the case of Dorset 1960s, geophysical survey is increasingly (RCHM, 1952, 1970a, 1970b, 1970c). Fol­ employed in its own right or alongside lowing the abandonment of this county-by­ earthwork survey as a powerful non-invasive county approach, more recent analytical tool in hillfort archaeology.

Fig 1.4 Plan of all excavated features inside Danebury hillfort, Hampshire (from Cunliffe 1995).

7 THE WESSEX HILLFORTS PROJECT

A number of criticisms of traditional synthetic studies have resulted in the publi­ approaches to Iron Age archaeology began cation of a number of volumes (eg Cham­ to emerge from the late 1980s. The gener­ pion and Collis 1996; Gwilt and Haselgrove alised, pan-European view of the ‘Celts’ was 1997; Hill and Cumberpatch 1995) though replaced by an emphasis on the distinctive no thoroughly worked-though new Iron Age nature of relatively small regions. This view ‘story’ has yet emerged. relied directly on archaeological evidence In 2001, Understanding the British Iron and took a more critical approach to the lit­ Age – An Agenda for Action (Haselgrove erary sources that had formed the main et al 2001) was published. This detailed plank of the traditional view. At the same research agenda based on five themes: time, the idea of hillforts as ‘central places’ chronological issues, settlements, land­ and elite residences came under increased scapes and people, material culture, region­ scrutiny and was found wanting, since even ality and processes of change proved extensively excavated settlements yielded relevant to hillfort studies in several ways. remarkably little evidence of social differen­ Despite completion before the publication tiation. The very existence of elites in the of the agenda, the Wessex hillforts survey Middle Iron Age was questioned (Hill 1995) and geophysical survey of Iron Age settle­ although the reduction in the number of ments in general had already begun to occupied hillforts after 300 BC does address in part some of the recommended nonetheless suggest some concentration of avenues for future research, including: power at this time (Haselgrove 1999). The view of the period as one dominated by • revealing spatial organisation of settlements endemic warfare is also being overturned. and divisions of settlement space The construction of fortified enclosures • exploring the landscape for evidence of activ­ appears to have been connected as much ity outside visible settlement boundaries with status as defence (Haselgrove 1999, • carrying out surveys of poorly understood Ralston 1996) and increasing emphasis is sites of the earlier Iron Age being placed on the non-defensive aspects of • analysing landscapes around important loci the role of hillforts, concentrating on issues of activity such as the environs of hillforts such as the symbolic use of enclosed space (eg Bowden and McOmish 1987; Hingley In areas with established frameworks, such 1990) and the cosmological significance of as Wessex, new fieldwork should focus on east and west-facing entrances (Hill 1996). clearly defined research themes, as well as There are numerous examples in southern exploiting any significant new opportunities England of the placement of hillfort that may arise. Although the Wessex Hill- defences well down-slope, thus rendering forts Survey was opportunistic in nature it is the interiors visible from the adjacent low­ hoped that it might stimulate other similar land. This may indicate a largely non-mili­ projects elsewhere in Britain where the tary purpose and suggests that display of methodology is effectively applicable and power was more important. information is lacking. A major survey of Northumberland hillforts on the flanks of That power was based on more than simply the Cheviot Hills was started in 2000. The the control of armed force seems clear for three-year project, involving detailed analyt­ many Celtic-speaking societies. The wish to ical earthwork survey of twelve hillforts, is demonstrate status, the need to monitor being carried out by the Archaeological access to markets, to industries, to food, or Investigation team from theYork Office of to luxuries, or the desire to control participa­ English Heritage in partnership with the tion in ritual activities, are amongst many Northumberland National Park Authority factors which may equally have contributed (Ainsworth et al 2001; Frodsham 2004). to the decision to erect hill-fort type earth­ Detailed mapping of the surface evidence is works, as well as influencing the form they more appropriate at these sites than geo­ took (Ralston 1996). physical survey because much of the archae­ ological evidence is spectacularly well It is increasingly appreciated that much of preserved and observable above ground. the Iron Age material recovered during Geophysical techniques are also less effec­ excavation provides only a selective and dis­ tive here due to underlying igneous geology, torted picture of everyday life owing to the thin soil cover and bare rock exposures. ritual nature of many deposits placed in set­ One of the few parts of the country that tlement contexts. These new theoretical and can confidently claim to possess a well

8 HILLFORT STUDIES AND THE WESSEX PROJECT understood hillfort chronology is the Daneb­ pottery styles comparative to those present ury area, following four decades of intensive at Danebury. Here the various phases of the research by Cunliffe (Cunliffe 2000). The hillfort, spanning the Late Bronze Age to excavation campaign at Danebury was the the early Roman period, are defined by most sustained investigation of any hillfort in characteristic changes in pottery form and Western Europe, taking place over some 20 style (ceramic phases 1–7) that have been years and resulting in the excavation of some tied to a sequence of radiometric dates. It is 57 per cent of the interior of the site (Fig therefore possible to arrive at a broad date 1.4). The research on Danebury has con­ range for a given hillfort based on the range tributed to the formulation of a broad model of pottery styles present on the site. In some of hillfort development with, it has been cases gaps in the ceramic sequence suggest assumed, at least regional applicability periods of abandonment followed by reoc­ (Cunliffe 1991, 344–64). In simple terms cupation – commonly linked to refurbish­ this represents a three stage chronological ment of defences or redefinition of enclosing progression from slight univallate forms to ditches – in a later period. A long uninter­ those of increasing elaboration and size. rupted sequence of changes in ceramic style Large multivallate hillforts, discussed under indicates continuity and longevity of occu­ the heading ‘developed hillforts’, represent pation comparable to Danebury. By contrast the final stage of this model (see Fig 1.1). a limited range of pottery generally indicates Hillforts of developed type, where excava­ a single, probably short-lived phase, of tion has demonstrated long sequences of activity uncomplicated by any later phases. occupation and a high density of internal How broadly applicable this model is cannot activity similar in character to Danebury, are be known without more survey both in the known in Dorset and Somerset at Maiden wider Danebury region and farther afield Castle and South Cadbury Castle. Others into neighbouring regions that also posses a that have not been extensively excavated can high density of hillfort sites but have differ­ be recognised from the form of the defensive ent defining characteristics, such as soils earthworks (and in some cases the density of and geology (for example the Jurassic Ridge internal features surviving as earthworks) and west of Cranborne Chase). Comparison elsewhere in Wessex (for example at Yarn- of the evidence with neighbouring regions bury Castle, Wilts; Fig 1.5). and even other areas of chalkland landscape The dating of the construction and occu­ in central southern Britain is problematic pation histories of the other hillforts in the because no other area has been studied with Danebury area is based on the presence of the same intensity as the Danebury area

Fig 1.5 Aerial photograph of Yarnbury Castle, Wiltshire displaying several of the characteristics of a ‘developed’ hillfort including multiple banks and ditches and a single entrance with elaborate outworks (NRMC; NMR 15406/15, SU 0340/149).

9 THE WESSEX HILLFORTS PROJECT

sification previously developed in the mod­ els. While not invalidated, the present mod­ els require further elaboration to incorporate the additional variation in hill- fort sites now shown to exist. The Wessex Hillforts Survey Project was initiated pre­ cisely in order to contribute towards the additional data needed to place the evidence from Danebury and its environs in an even wider regional context. In order to provide sufficient back­ ground data on the regional setting of the Wessex Hillforts Survey it is necessary at this point to describe in some detail the results of the Danebury Environs Project where it relates to hillforts, as well as the results of a recent study of hillfort distribu­ tion in the neighbouring region of Sussex (Hamilton and Manley 1997).

Hillfort development in the Danebury Environs

The Late Bronze Age to Earliest Iron Age The earliest forms of hillfort recognised in the region are hill-top or plateau enclosures at the site of Balksbury and the outer pre­ hillfort enclosure on Danebury Hill (Fig 1.7 Fig 1.6 (28 seasons of intensive research excavation). and see Fig 1.3). Although there is some dis­ Hillfort sites and other Iron Partly as a result of environmental factors, parity in the structural form of these two Age enclosed settlements Iron Age sites in northern England generally sites, both seem to have been established in investigated by The produce far less ceramic material with little parallel with systems of linear earthworks Danebury Environs Project variation in form over time, rendering the that indicate a growing emphasis on bound­ in Hampshire from construction of detailed chronologies in aries, enclosure and barriers at the end of 1989–96 (from Cunliffe these areas far more difficult in comparison the Bronze Age, thereby transforming the 2000). to Wessex (Haselgrove 1999, 114). previously open landscape of the Early The earlier model for Wessex, based on Bronze Age. excavations at Balksbury (Wainwright and Both enclosures were protected by simple Davies 1995) and Danebury (Cunliffe earthworks and show only minor traces of 1984a, Cunliffe and Poole 1991) in Hamp­ internal activity in the form of post-settings. shire, has been considerably refined by the At Balksbury, a bank and ditch defined a work of the Danebury Environs Project (Fig roughly triangular enclosure of some 18 1.6) and the resulting publication (Cunliffe hectares in extent (Wainwright and Davies 2000) has provided the greatest insight yet 1995). Three distinct phases of construction into the history of hillfort development and have been identified, beginning with a slight occupation within a region of central-south­ ditch with a low un-revetted bank on one ern England. The extended research on side, the ditch being twice recut. At the one neighbouring hillfort sites, other enclosed entrance, located at the south-eastern cor­ settlements and linear boundaries in the ner, three phases can also be seen in the tim­ Danebury Environs has highlighted the ber revetment of the entrance passage. complexity in the archaeological record and Although a considerable area of the inside the danger of over simple generalisations of the enclosure was thoroughly excavated about hillfort origins, development and (see Fig 1.3), a number of four- or five- function. Although the three-phase model of post buildings of the kind conventionally hillfort development is still broadly applica­ regarded as ‘granaries’ (or platforms for stor­ ble and has by no means been discredited by ing hay or other fodder) and possibly three this new work, it is now evident that the circular post-built houses found in the south­ archaeological reality defies the simple clas­ ern part of the site were the only evidence of

10 HILLFORT STUDIES AND THE WESSEX PROJECT activity in the Late Bronze Age phase of the to function as a communal focus after site. A well defined pottery assemblage of c 9–800 BC, although it was later used as the Late Bronze Age date was also recovered. site of an un-enclosed farmstead from The defensive enclosure at Balksbury the Middle Iron Age through into the appears to have been abandoned and ceased Roman period. This later nucleus of activity

Fig 1.7 The main phases in the development of Danebury hillfort (from Cunliffe 1995).

11 THE WESSEX HILLFORTS PROJECT

within the abandoned former defences was box-timber form of construction. The first concentrated in a comparatively restricted hillfort ramparts, given their style of con­ area of the old enclosure. struction, are probably broadly contempo­ At Danebury (Fig 1.7), 16.2 hectares of rary with the timber revetted hillfort the hilltop were enclosed by a slight ditch, ramparts at Bury Hill I and Winklebury. possibly with two entrance gaps, almost At a slightly later date (probably during entirely recut on a more substantial scale in the early 5th century BC) several more hill- the Middle Iron Age (the Outer Enclosure). forts were built in the Danebury area at Figs- The north-eastern side of the enclosure bury, Quarley Hill and Woolbury (Fig 1.6). ditch joins with a linear earthwork (the These sites are all remarkably comparable in Danebury Linear), possibly a later addition. size, structure and date: contour works Internal features of the enclosure in this enclosing similar areas with dump con­ period consisted of some large pits, which structed ramparts (but no evidence for tim­ may have held timber uprights (possibly ber framed or revetted construction) with with some ritual function), and a group of two opposed entrances. There is no evidence four-post structures. (Although common in of extensive debris-generating activities at the later hillfort, these examples were shown Quarley, Figsbury, Woolbury and Bury Hill I to predate the first phase of hillfort in this period, suggesting very low levels of defences.) A small assemblage of Late internal occupation activity. This interpreta­ Bronze Age pottery was also recovered tion is backed up by the results of magne­ from contexts predating the construction tometer surveys at Bury Hill and Woolbury of the later hillfort. (this volume) which suggest an almost total Other possible examples of the type of absence of internal structures. site represented by the Late Bronze Age While it may have had exactly the same enclosures at Balksbury and Danebury have range of functions as the other early hillforts been tentatively identified at Beacon Hill, at the beginning of the 5th century BC, Harting (West Sussex); Martinsell Hill, Danebury differed from them in that the Wiltshire and Walbury Hill, Berkshire on enclosure was used extensively for the con­ the basis of the form of the enclosing struction of storage pits (which were con­ earthworks and the size of the enclosures. centrated in the centre around a focus of The latter two sites were included in the rectangular structures that may have been programme of geophysical exploration shrines) and for the building of circular carried out for the Wessex Hillforts Survey houses occupying a peripheral zone in the and the results are presented in Chapter 2 of lee of the rampart. Four-post storage build­ this volume. ings and a dendritic pattern of roads com­ pleted the plan. Once established, Early Iron Age occupation seems to have been continuous, Of the two sites enclosed in the Late Bronze extending throughout the 5th and 4th cen­ Age, only Danebury remained a significant turies. The implication of this is that, in location and was redefined by a stronger addition to its social and religious functions, rampart and ditch, possibly towards the Danebury served as a focus for a population end of the 7th century BC (Fig 1.7). Bury who occupied the site either permanently or Hill (fort number 1 or Bury Hill I) – a hill- for a significant period during each year. It fort 10 hectares in extent defined by a chalk is interesting to note that this change to resi­ rampart fronted by a timber palisade – dent occupation seems to have taken place probably replaced Balksbury as the main at about the time that the forts of Quarley, communal enclosure in the Danebury Figsbury and Woolbury were constructed – region in the late 7th–6th-centuries BC. events that may be related. By the end of the A similar enclosure dating to the same 5th century BC, Danebury was a defended period is known at Winklebury, to the settlement of considerable extent with an north-east near Basingstoke (Smith 1977). exceptional storage capacity and a cluster of Both sites are apparently largely devoid centrally placed communal structures, while of evidence of internal activity (based on the countryside around was quite densely limited areas of excavation and magnetome­ scattered with farmsteads. Towards the ter survey). The first phase of hillfort periphery of what could be regarded as the defences at Danebury (enclosing a smaller core territory of Danebury, hilltop fortifica­ area of 5.3 hectares within the earlier tions of comparable size were being outer enclosure) was also established at erected at Figsbury, Quarley and Woolbury. some time during the 6th century BC using a The lack of occupation within these sites

12 HILLFORT STUDIES AND THE WESSEX PROJECT suggests that they may have been created sample of Bury Hill II that although the new as strategic points to command the per­ defences had enclosed a settlement, the ceived boundaries of a territory centred duration of the associated occupation was upon Danebury. relatively short (limited to the period defined by ceramic phase 7 at Danebury). Developments from the end of the In chronological terms this could well have 4th century bc (300–100 bc – The Middle been restricted to the early part of the 1st Iron Age) century BC, placing Bury Hill II in the Late On the basis of the distribution of pottery Iron Age. styles in the region, it seems likely that the In summary, the evidence from the hill- political geography of Wessex changed in forts in the region supports the view that the early 3rd century BC. It was at this time, during the 3rd and 2nd centuries only after a diminished level of use, that Daneb­ Danebury remained in use and with a ury underwent a major phase of reconstruc­ greatly enhanced level of activity, until the tion and took on many of the defining construction of a new hillfort at Bury Hill characteristics of a developed type of hillfort late in the occupation history of Danebury. (see Fig 1.7). The south-west gate was Occupation within the newly constructed blocked and the rampart was augmented hillfort ran parallel with the last stages of with material from internal quarries imme­ occupation at Danebury. diately inside the rampart. Finally a corridor approach and projecting hornworks were The Late (immediately pre-Roman) Iron added to the single remaining entrance. For Age (100 bc–ad 50) the next 200 years or so the interior was The hillforts at Danebury and Bury Hill II heavily utilised. A massive storage capacity (both in active occupation at the turn of the in the form of rectangular post structures century (100 BC)), were abandoned by the and below-ground silos was maintained; end of the first half of the 1st century BC. close packed circular houses in the lee of the The end of the occupation at Danebury may rampart were rebuilt every 20–30 years and be linked to the firing of the gate structure; a religious focus continued to develop once this occurred only a very low level of towards the centre of the fort. The intensity occupation persisted into the period follow­ of activity measured in terms of material ing 50 BC. By the very end of the Iron Age, discarded was greatly increased from earlier some of the site was being put to agrarian periods. While the contrast to the earlier use (comparable with Cissbury in West Sus­ period is dramatic it is one of intensity sex). Once the hillforts were finally aban­ rather than range. The layout and the struc­ doned other enclosed settlement sites in the tures were not significantly different, but the region re-emerged, such as Suddern Farm quantity and variety of material deposited in and Houghton Down (Cunliffe and Poole the later period is strongly suggestive of a 2000c, 2000e), which continued in occupa­ greatly increased level of activity (or differ­ tion into the Roman period (see Fig 1.6). A ent attitudes to the disposal of material) and number of the earlier disused hillfort sites, also a greater range of functions (including a such as Woolbury, were also reoccupied by centre of craft production and a place where farming communities (often defined by exchange systems were articulated). small paddocks and enclosures) at this time, There is no evidence that the neighbour­ again continuing into the Roman period. ing hillforts in the area (Figsbury, Quarley and Woolbury), established in the Early Iron The overall pattern Age, were still in use after the end of the 4th In the Danebury area, the desire for hilltop century BC. All retained their simple enclosure began with the construction of entrances of undeveloped form. The situa­ Balksbury and the Outer Enclosure at tion at Bury Hill was quite different. Here Danebury and continued throughout the 1st the early, long abandoned hillfort was refor­ millennium BC, culminating in a spate of tified, though the area enclosed was hillfort building in the 5th and 4th centuries reduced. The new defences (Bury Hill, fort BC. Thereafter the dominance of Danebury number 2 or Bury Hill II) differed from the suggests that some unified authority had traditional form of Middle Iron Age emerged only to be challenged some two defences in that they were composed of two centuries later by a polity setting up fortifi­ massive concentric ramparts with a single cations at Bury Hill. After a period of transi­ ditch in between and are therefore multival­ tion in the early 1st century BC, the late in form. It is clear from the excavated emergence of new ditched enclosures – no

13 THE WESSEX HILLFORTS PROJECT

longer on dominant hilltops – points to a region takes place at Bury Hill II with the new socio-political grouping, but one that construction of multivallate fortifications still adhered to the massive enclosing ditch on the site of the earlier abandoned hillfort. as a symbol of authority. This development possibly represents the Cunliffe identifies Sidbury and Yarnbury emergence of a rival polity challenging the in Wiltshire (18km and 28km from territorial control of Danebury. Danebury respectively) as possible candi­ 5. Abandonment of the remaining two hillforts dates for developed hillforts functioning in the region at Bury Hill II and Danebury to in a similar way to Danebury during the be replaced by other forms of settlement 3rd and 2nd centuries and controlling including banjo enclosures, Suddern Farm- neighbouring territories. No dating evidence type enclosures bounded by impressive ditches has been obtained from Sidbury, but the and clustered enclosure settlements. Areas form of the earthworks suggests it is within some earlier hillforts continue to be of the developed variety. Other excavated occupied by small farming communities from hillforts farther afield in Wessex that the Late Iron Age into the Roman period. conform with the developed model (defined by such characteristics as elaborate defen­ Taken together, the evidence from the three sive earthworks and entrance approaches hillforts and others in the Danebury region and occupied intensively over long has enabled the construction of a coherent periods of the Iron Age) are Maiden picture, showing for the first time something Castle in Dorset and South Cadbury Castle of the complexity of the situation at this in Somerset. level in the settlement hierarchy. It is now The growth of Danebury, after its clear – from the Danebury region at least – major phase of re-defence in c 270 BC, when that many hillforts should be seen as succes­ the hillfort became a major focus of intense sors of earlier hillforts. The settlement pat­ activity, was directly related to the abandon­ tern is constantly shifting from one location ment of all other sites within a radius of to the next and the distribution pattern of up to 10km (based on the absence of hillforts that we see in the landscape today is ceramic phase 7 pottery from settlements therefore the culmination of a series of in the environs of the hillfort). A similar developments over a considerable period of situation has been noted around the hillfort time and does not represent a group of sites of Maiden Castle at this time (Sharples all in contemporary use. The result is con­ 1991, 260). siderable complexity in the surviving archaeological record – borne out by the Table 1 Summary of the sequence of work in the Danebury Environs. hillfort development in the Danebury Environs from 800 BC–AD 50 The pattern in neighbouring regions 1. Large Late Bronze Age hill-top/plateau enclosures (Danebury and Balksbury). The hillforts of Sussex 2. Simple univallate hillforts initially with timber framed or revetted ramparts Hamilton and Manley (1997) have recently succeeded by later univallate hillforts defined attempted analysis on a regional scale of the by dump ramparts frequently built at focal pattern of hillfort distribution in the two points on the system of earlier linear counties of Sussex (Fig 1.8). Three main boundaries. With the exception of Danebury groupings have emerged, reflecting three none of these new forts show evidence of main phases of hillfort development in suc­ significant internal occupation and the cessive periods. A striking aspect of the re­ upkeep of the defences is generally short- analysis of the dating of later prehistoric lived. One interpretation of these sites is that enclosures is that the greatest proportion of they are peripheral markers of a territory the sites belongs to the Late Bronze Age. A centred upon Danebury, explaining the low particular emphasis of the paper in Sussex level of use in comparison to Danebury. Archaeological Collections is to consider how a 3. The defences at Danebury are continuously greater appreciation of the topographical augmented and the site develops into a position of the sites might enlighten our major centre of population with evidence of interpretation of them. intensive occupation from the 5th century The Sussex hillfort sites are classified until the late 2nd/early 1st century bc. simply into three divisions by period (based 4. The latest hillfort development in the on available dating evidence, which is often

14 HILLFORT STUDIES AND THE WESSEX PROJECT

1. 2.

3. 4. ? N ?

0 102 0 Km

KEY TO SITES REFERRED TO IN THE TEXT

1. Torberry Alluvium Greensand Ridge Promontory 2. Harting Beacon Chalk Contour 3. The Trundle High Weald Other (? uncertain dating) 4. Cissbury Coastal Plain Low Weald

limited), and hillforts of several different Late Bronze Age and Early Iron Age Fig 1.8 forms, size and type are present in each of periods, including small forts 1–2 hectares Distribution of hillforts in the periods. Under this scheme there is no in area (for example Chanctonbury, Sussex related to geological distinction made between large hilltop Hollingbury, Thundersbarrow and Wolston­ zones (based on Hamilton enclosure type sites and smaller univallate bury) plus some large forts comparable and Manley 1997). forms of hillfort in the Late Bronze Age to hilltop enclosures in Wessex (Harting to Early Iron Age. Distinct geographical Beacon and Bell Tout). There is a tendency patternings of hillfort distribution can appar­ for these sites to occupy peripheral down- ently be observed in each of the three land locations (possibly to observe outwards periods and, like Wessex in the middle the landscape and people in the surrounding period (corresponding to the Middle Iron area). The enclosures in this period, Age), hillforts seem to be fewer in number being sited on the boundaries between but exhibit intensification of internal activity. different geological and environmental Discussion on the function of the zones, are also suitably placed to access a sites revolves around their topographical varied range of natural resources both position and the tendency for them to favour downland and river valley. The enclosing particular topographical positions at differ­ earthworks consist of a mixture of timber- ent periods. This leads the authors to suggest revetted and dump-style rampart construc­ that they may have functioned differently tion similar to the techniques employed in each of the three phases identified. in the Danebury Environs and on the They believe it is inappropriate to explain Ridgeway Hillforts (Chapter 2, this sites in terms of continuums of develop­ volume). Evidence of domestic use of the ment, such as increasing socio-economic sites is generally lacking. Few if any of the centralisation and developing hierarchies sites are known to contain internal features, (models that have been applied in the past such as pits, and associated artefact finds to Danebury), and that the successive are normally few in number. Despite a phases of hillfort construction are linked reasonably large area excavation of the inte­ more to position in the landscape, reflecting rior at Chanctonbury Ring, very few fea­ aspects of symbolism and territoriality. tures were uncovered, suggesting that the By far the largest number of sites site was not primarily used for occupation belong in the first phase, spanning the (Bedwin 1980). Harting Beacon is known to

15 THE WESSEX HILLFORTS PROJECT

contain four- and six-post structures similar successive periods are seen as being to hilltop enclosures in Wessex (Bedwin reflected in a shift in their topographical 1978, 1979). Highdown Hill and Holling­ position and location in relation to valued bury do show signs of occupation – resources, such as land suitable for a mixed including the presence of round houses, range of agriculture and industrial raw metalwork hoards, fine-ware pottery and materials in the case of the Late Iron other occupation debris. Age pattern. The number of hillfort type enclosures in Sussex is dramatically reduced in the The Jurassic Ridge Middle Iron Age. Only four sites are pre­ The pattern of development in Wessex sent (the Caburn, Cissbury, the Trundle outlined by Cunliffe (Cunliffe 1991) would and Torberry) spaced at even intervals and appear to hold true for the hillforts located centrally within each major block of the Jurassic Ridge bordering Wessex of downland defined by the north–south to the north and north-west including the rivers of the Sussex Downs. A greater Cotswolds and parts of Gloucestershire, intensity of activity took place within these Oxfordshire, Northamptonshire and sites compared to the Late Bronze Age/ Worcestershire. Early Iron Age enclosures, as evidenced by Large enclosed sites that appear to large numbers of internal pits. As is also share similar characteristics with the early generally the case in Wessex, most of the hilltop enclosure class of site in Wessex have Sussex Middle Iron Age forts were pre­ been recognised at sites such as Norbury ceded by Late Bronze Age or Early Iron Camp and Nottingham Hill, Gloucester­ Age activity. In some cases the defences of shire. As in Wessex, early hillforts seem to the enclosures were subsequently substan­ be prolific while far fewer developed tially remodelled in the Middle Iron Age, as hillforts of the Middle Iron Age have been at Torberry (Cunliffe 1976). This reconfig­ identified. The excavated site of Crickley uration has traditionally been seen as relat­ Hill (Dixon 1976, 1994) is the best known ing to the emergence of central places (the example in the region of an Early Iron Age former Danebury model) which replaced hillfort, with a construction date for the first socio-economic functions previously dis­ phase of defences (a massive timber-laced persed across several enclosures, but rampart with an external stone facing) in Hamilton and Manley argue for a function the 7th or 6th century BC. The main as territorial landmarks or ‘landmark features within the fort at this time were enclosures’ situated in prominent central rectangular post-built structures (either downland positions to be seen from a dwellings or rows of storage buildings). In distance all around. The substantial the late 6th or early 5th century BC the ramparts that define this group of sites defences were reconstructed and the earlier emphasise them from afar (a trend contin­ rectangular buildings replaced by circular ued into east Hampshire at the hillforts of timber buildings. Old Winchester Hill and St Catherine’s Conderton (or Dane’s) Camp, in Hill). Hamilton and Manley suggest that Worcestershire (Thomas 2005) and Huns- the sites in this period may not have been bury in Northamptonshire (Fell 1937) primarily defensive nor settlements in the share certain features in common with conventional sense. The pits that have been those Wessex hillforts that originated in the found inside the sites need not necessarily Early Iron Age period but continued to be imply a settlement function. Instead the occupied on a more intensive scale during sites could have acted as foci for selective, the Middle Iron Age (the so called devel­ patterned deposition. The point is also oped form of hillfort). The small 1.5 made that the elaborate entrances at some hectare hillfort at Conderton Camp on Bre­ of the sites may be as much to do with the don Hill, Worcestershire displays a relative ‘theatre of presentation and approach’ as paucity of internal activity in the period fol­ protection from attack. lowing its initial construction in the earlier In the Late Iron Age (the final phase) part of the Middle Iron Age (c 300 BC). In in Sussex, enclosure activity shifts away the succeeding period the defences were from the chalk downland and concentrates remodelled and strengthened, the enclosed in the Weald, suggesting involvement area was retracted and one of the two with iron working and the importance of the opposed entrances was blocked (a develop­ natural iron resources of the area. The ment paralleled at Danebury). The second differing functions of the Sussex sites in period of the hillfort is associated with

16 HILLFORT STUDIES AND THE WESSEX PROJECT dense internal occupation activity sug­ Prospecting techniques in gested by a row of tightly packed circular houses (possibly with several successive hillfort archaeology phases of construction) in the eastern half of the fort and an area given over to a very Hillforts in the landscape dense grouping of as many as 100 storage pits in the western half. This interpretation It is now appreciated that hillforts are only a of the site is based on limited excavation single element in a complex and changing carried out at the end of the 1950s and pattern of landuse in the 1st millennium BC more recent geophysical survey (see that encompassed many other forms and Thomas forthcoming). Artefacts recovered types of settlement both enclosed and un­ from the interior, such as iron currency enclosed. An understanding of hillforts can­ bars, are also indicative of the developed not truly be achieved without some status of the site. appreciation of the wider systems in opera­ The multivallate hillfort at Hunsbury tion, necessitating research into the interac­ near Northampton (Fell 1937), possessing tion and chronological relationship of a evidence of intensive occupation in the particular hillfort with contemporary non­ Middle Iron Age and a range of finds hillfort sites (including field systems, suggesting craft and exchange functions, is boundaries and trackways) as well as neigh­ another possible contender for developed bouring hillforts and other enclosed settle­ hillfort status in the region. ments in the surrounding landscape.

Fig 1.9 The hillfort of Danebury in its landscape context based largely on aerial photographic evidence (from Cunliffe 1986).

17 THE WESSEX HILLFORTS PROJECT

To understand the role of a hillfort in archaeological projects to employ geophysi­ society it is necessary to understand how it cal methods on an ambitious scale not only relates to the surrounding settlement pat­ as a predictive method to assist targeting tern. Intensified activity within a hillfort at a of excavation but also to provide a wider given point in time may be reflected in the context within which to interpret the simultaneous abandonment and depopula­ excavations (Musson 1968; Tite 1972). tion of settlements in the surrounding land­ Similar, equally successful, exercises linked scape. This might be interpreted as the to sample excavation were carried out dur­ consequence of a time of crisis or the hillfort ing this period at Conderton Camp, taking on the role of a semi-urban central Worcestershire and Rainsborough Camp, place (Danebury and Maiden Castle). Northamptonshire (Aitken and Tite 1962; The Maiden Castle Project in Dorset Tite 1972). These projects were a successful (Sharples 1991) and the Danebury Environs early demonstration of the effectiveness of Project in Hampshire (Cunliffe 2000) are magnetometry for exploring hillfort interi­ notable examples of projects that have in ors and characterising the relative density recent decades attempted to achieve this of occupation features they contained. greater understanding using the systems What was lacking was the ability to collect approach. The theme of studying the sufficiently high resolution data-sets, due to hillfort in its chronological and landscape con­ the slow mode of operation of the instru­ text has been continued in recent years by the ments, and the means to manipulate South Cadbury Environs Project in Somerset the data subsequently to produce easily centred on the hillfort of South Cadbury Cas­ interpretable visual representations. The tle (Coles et al 1999; Leach and Tabor 1997; approach first pioneered in the experiments Tabor and Johnson 2000) and at Castle Hill, of the 1960s at sites such as South Cadbury Wittenham Clumps, Oxfordshire (Oxford was not repeated until the early 1980s at Archaeology 2003, Payne 2002b, 2002c). Aer­ Maiden Castle in Dorset, by which time ial photography was used to great effect in the geophysical techniques in archaeology 1980s to provide detailed evidence of archaeo­ were coming of age with the arrival of logical sites in the environs of Danebury hill- routine digital data recording and comput­ fort (Fig 1.9; Palmer 1984), but in more recent erised plotting of the data. The complete years the use of ground-based archaeological magnetometer survey of Maiden Castle, prospecting has proved to be as important in undertaken by the Ancient Monuments studies of this nature, particularly in areas such Laboratory (AML) between 1984 and as the South Cadbury Environs where the 1985 (Balaam et al 1991, Payne 1996) was a value of aerial photography is restricted due to striking reaffirmation of the benefits of link­ both predominantly pastoral land use and a ing large scale overall geophysical coverage limited archive of available aerial photographic with smaller targeted research excavation material. Recently, magnetometer survey has of hillfort interiors (Fig 1.10). Digital cap­ begun to provide a rich archaeological context ture of the data from Maiden Castle on for the hillfort of South Cadbury of similar portable field computers heralded the rou­ quality to the results achieved from aerial tine use of this method with resulting reconnaissance in the mainly arable landscape improvements in data presentation. The (favourable to the formation of crop and soil computer-plotted halftone or greyscale marks over archaeological sites) around plots that became the norm in archaeologi­ Danebury in Hampshire (Fig 1.9). A pro­ cal geophysics from the late 1980s onwards, gramme of aerial reconnaissance undertaken coupled with the development of increas­ by the National Mapping Programme (Bewley ingly powerful information technology, 2001) has recently begun to provide evidence allowed the results of geophysical surveys of the contemporary landscape setting of the to be seen in much greater clarity than ‘Ridgeway Hillforts’ (Segsbury, Uffington Cas­ ever before and enabled the recognition of tle and Alfred’s Castle) on the North Berkshire even the weakest anomalies from features (or Lambourn) Downs, although most of this such as ring-gullies. As the number of geo­ data has yet to be published. physical surveys of hillfort sites increased during the 1990s it gradually became appar­ The role of geophysical survey ent that, largely due to the technological improvements of the preceding decade, The original excavations at South Cadbury archaeological geophysics had the power in the 1960s (Alcock 1968a, 1968b, to contribute much to our understanding 1969, 1970, 1971) were some of the first of hillfort interiors.

18 HILLFORT STUDIES AND THE WESSEX PROJECT

Fig 1.10 The magnetometer survey of Maiden Castle in Dorset carried out by EH prior to excavation in 1985 (from EH, Ancient Monuments Laboratory).

19 THE WESSEX HILLFORTS PROJECT

Fig 1.11 Geophysical survey in the 1990s as of informed conservation management. Oblique aerial photograph of an aid to site management Although an increasing number of sites are Old Winchester Hill, Hamp­ now sympathetically managed in favour of shire. The site is crossed by Overall responsibility for the conservation of preserving any buried archaeological fea­ several long-distance footpaths hillfort sites – the majority of which have tures present inside them, a considerable and is managed primarily as statutory protection as scheduled ancient number still face pressure from gradual a nature reserve by English monuments – is the duty of English Her­ degradation by agricultural activities such as Nature. Footpath and track itage. A problem to date has been the lack of ploughing, grazing and arboriculture, as erosion converges on the trian­ extensive data on hillfort interiors, which well as burrowing and visitor erosion. The gulation pillar and along the has deprived English Heritage of even the need to improve our understanding of the ramparts (NMRC; NMR most basic information on the archaeologi­ internal layout of hillforts, both for practical 15393/23, SU 6420/53). cal content of many hillforts – a prerequisite reasons of site management and in order to

20 HILLFORT STUDIES AND THE WESSEX PROJECT

Crown copyright, RCHME N

657

656

655 -3.20 -0.23 2.73 5.70 nT Scale Linear greyscale - data enhanced by 1m Gaussian low-pass filter 0 30 90m 640 641 642 643 continue to improve our academic compre­ Department of English Heritage, primarily Fig 1.12 hension of the role and functions of this to provide information to support casework Magnetometer survey class of monument without resorting to aimed at stabilising the management of the carried out in 1995 of a costly and undesirable ground disturbance, sites in order to better secure their preserva­ sample of the interior of were the two main underpinning reasons for tion for the future. The surveys were able to Old Winchester in relation the development of the programme of significantly enhance the data available on to the RCHME hachured largely geophysical survey-based research each of the hillfort interiors and were a suc­ earthwork survey (from described in this volume. cessful demonstration of the power and EH, Ancient Monuments The understanding of hillfort develop­ affordability of fluxgate magnetometry to Laboratory and ment in central-southern England applicable transform knowledge of archaeological sites RCHME). to the Danebury region is based on limited that may be clearly-visible, well-defined earth­ information derived from relatively small scale works but are otherwise poorly understood, sample excavations (Cunliffe 2000), but it particularly in terms of their internal archaeo­ was believed that it could be markedly logical contents and arrangements. The sur­ enhanced, refined and extended by access to veys – all carried out in a relatively short space the level of information that geophysical sur­ of time – made a significant contribution to vey was potentially capable of providing. Dur­ furthering understanding of the sites. ing the early 1990s, geophysical surveys had The magnetometer survey at Letcombe been undertaken by the Ancient Monuments Castle, linked to a Countryside Stewardship Laboratory of English Heritage on several agreement that converted the site from hillforts in central-southern England, includ­ arable to stable grassland, was particularly ing Buckland Rings and Old Winchester Hill useful, allowing the characterisation of a in Hampshire (Figs 1.11, 1.12), Caesar’s hillfort site for which negligible archaeo­ Camp in Berkshire and Letcombe Castle logical information had previously been (Segsbury Camp), Oxfordshire. These sur­ available (Figs 1.13, 1.14). The availability veys were commissioned by the Conservation of such data has clear benefits for the

21 THE WESSEX HILLFORTS PROJECT

management of the site: for example, the Because of the degree of overlap, it is neces­ information provided by the survey is of sary at this point to provide a brief review of practical use for determining if a zone that geophysical survey of hillfort sites in south­ is suffering from erosion due to burrowing ern England that led up to the development or heavy footpath wear also contains of the Wessex Hillforts Survey programme. vulnerable archaeological features. Mitiga­ These surveys, carried out between 1989 and tion measures can then be taken to decrease 1995, were a major influence on the design the threat of erosion in the vulnerable area of the subsequent project carried out (for example by re-routing foot-paths). between 1996–8. Other ground disturbance such as the erection of fences and sign-posts can be The hillforts of the Lambourn and Marl­ avoided in areas where the survey has indicated the presence of archaeological borough Downs (or the Ridgeway group) features. In addition to the surveys carried The survey at Segsbury Camp (or Let­ out for management purposes, the ability combe Castle) carried out from 1993–5, of geophysical methods to help address provided the clearest illustration of the con­ substantial archaeological questions related siderable academic potential of geophysical to hillforts was also emphatically demon­ methods in hillfort research (Fig 1.14). Let­ strated by a succession of surveys in combe is one of the grouping often referred support of the Danebury Environs and to as the Ridgeway Hillforts which, with the Uffington White Horse Hill Projects exception of Uffington Castle, had been between 1989 and 1991 (Cunliffe 2000, subject to very limited investigation before Miles et al 2003). 1993 (some excavation by the Hillforts of

Fig 1.13 Oblique aerial view of Letcombe Castle or Segsbury Camp, Oxfordshire (Copyright reserved Cambridge University Collection of Air Photographs, BIT 36, 1972).

Fig 1.14 (opposite) The original trial magnetometer transect across Letcombe Castle/Segsbury Camp undertaken in 1993 (from Ordnance Survey and EH, Ancient Monuments Laboratory).

22 HILLFORT STUDIES AND THE WESSEX PROJECT

847 N

846 ring gullies ­ possible hut sites

845

844 Dense concentration of occupation features (pits, hearths, ovens etc)

843

384 385 386 387 Scale SU 38 84 0 50 150m

23 THE WESSEX HILLFORTS PROJECT

Fig 1.15 the Ridgeway Project has since taken place). various scheduled monuments on White Oblique aerial view of Nevertheless, they excited considerable Horse Hill, by means of limited excavation, Uffington Castle, Oxford­ speculation about their function within the to help inform their future management shire (NMRC; NMR Iron Age settlement pattern of the region. and public presentation by English Heritage 15073/17, SU 2986/30). Nothing was known of the interior layout of and the National Trust who share joint Segsbury prior to the initial magnetometer responsibility for conserving the sites. As survey transect in 1993 (Payne 1993b). the archaeological excavations carried out Now, with total survey coverage and evi­ by the Oxford Archaeological Unit had to dence for at least 20 circular structural fea­ be small in scale to disturb as little of the tures within the hillfort combined with large monuments as possible, the wider use of agglomerations of pits (Chapter 2, this vol­ geophysical survey was an important ume), we can confidently attribute Segsbury additional component of the project. A very to the class of Danebury-style developed similar approach was adopted by the hillforts with probable functions as a centre Danebury Environs Project for the internal of population and an enhanced storage investigation of the hillforts of Woolbury capacity (Payne 1996). This emphasises that and Bury Hill during 1989–90 (see below). our perception of what constitutes a devel­ The Uffington Castle survey (Fig 1.16) oped hillfort should be as much about the was carried out to provide information on evidence inside the defences as features such the archaeological content of the hillfort as multivallate ramparts and elaborate interior to augment a limited archaeological entrances traditionally associated with such investigation through the surrounding sites but less recognisable at Segsbury. perimeter earthworks. The purpose of the Fig 1.16 (opposite) In 1989 a magnetometer survey was excavation was to recover information on Plan of the hillfort of carried out by the AML, inside the neigh­ the origins and development of the hillfort Uffington Castle with the bouring hillfort of Uffington Castle, with minimal disturbance to the site; the interpretation of the Oxfordshire (Figs 1.15, 1.16) in support of excavated section therefore exploited an magnetometer data (from the White Horse Hill Project (Miles et al existing breach through the ramparts. EH, Ancient Monuments 2003). The overall objective of the project Exploration of the hillfort interior was Laboratory). was to enhance understanding of the limited to non-intrusive investigation by

24 HILLFORT STUDIES AND THE WESSEX PROJECT

White Horse N 866

Prehistoric and Roman burial mound

Bronze Age round-barrow 865

864

863

AREAS OF ANOMALOUS MAGNETIC ACTIVITY 862 1. ?ferrous material

2. ?geological LINEAR/CURVI-LINEAR ANOMALIESN

linear anomalies alongside ramparts

former cultivation patterns

tentative anomalies A Y E W 861 R I D G Scale DISCRETE ANOMALIES T H E 0 30 90m occupation features

298 299 300 301 25 THE WESSEX HILLFORTS PROJECT

magnetometer survey (see Payne 2003a). The conclusion drawn from the geophys­ Despite the availability of numerous aerial ical results from Uffington (based on the photographs of the site, few of them had density and range of features mapped within revealed any detail of archaeological features the hillfort) was that it had only been occu­ within the hillfort except for traces of pied for a relatively short period of time dur­ Medieval or Post Medieval strip cultivation. ing the earlier Iron Age. Excavation has now Magnetometer survey, therefore, had an demonstrated further activity on the site important role in mapping the density and during the Roman period that resulted in layout of any buried archaeological features the incorporation of material of Roman date present underneath the relatively blank in the partially filled up earlier Iron Age pits. physical topography of the hillfort interior. In this respect the site parallels other hill- The pattern of discrete magnetic anomalies forts in the region, such as Woolbury, which mapped by the survey (Fig 1.16) suggests after a period of disuse when the defences that the site contains a moderate density were no longer maintained (often lasting of pits dispersed fairly evenly across the many centuries) were reoccupied by farming interior, with some loose clusters of pits and communities from the Late Iron Age into closely paired pits in places but otherwise the Roman period. Liddington Castle, sited few indications of any other forms of in a similar position to Uffington above the occupation (such as ring gullies). northern scarp of the Marlborough–Lam­ Subsequently during 1994–5 some small- bourn Downs, probably also had a similar scale excavation took place inside Uffington history of occupation, as suggested by finds Castle as part of the Hillforts of the Ridge- of early Iron Age and Roman material way research project undertaken by the (Bowden 2000; Hirst and Rahtz 1996). Oxford University Department of Continu­ ing Education (Miles et al 2003). The areas Hampshire hillforts of the hillfort interior that were opened up Although no large scale geophysical survey were carefully positioned to investigate areas took place at Danebury itself, during the containing geophysical anomalies mapped early 1990s the Ancient Monuments Labo­ by the earlier 1989 survey. Of the sample of ratory (AML) took part in the subsequent magnetic anomalies investigated by excava­ research on the Danebury Environs (see Fig tion, ten were shown to represent pits with 1.6), providing a series of fluxgate magne­ fills containing Iron Age and Romano- tometer surveys on several of the neighbour­ British material and another one was found ing hillfort sites to Danebury (Cunliffe to be an oven of Romano-British date. The 2000). The aim of the Danebury Environs availability of the geophysical data was cru­ Project was to arrive at a broader under­ cial for enabling the precise targeting of standing of the interaction of the hillfort small excavation areas (strictly limited in with its contemporary environment by extent by the terms of the Scheduled Monu­ studying the development of settlement and ment Consent to excavate) onto features of contemporary systems of land allotment in interest, thus avoiding unnecessary ground its locality from the end of the Bronze Age disturbance and wasted effort on opening up to the beginning of the Roman period. The unrewarding trenches. The relative paucity eventual objective was to understand the of features inside Uffington Castle (see role of the hillfort in the context of the below) compared to other hillforts with long changing social and economic systems in sequences of habitation (such as Danebury the wider Danebury area during the 1st mil­ and Maiden Castle) presented the very real lennium BC. danger of opening up blank areas and miss­ As a first step in the study it was clearly ing the archaeological features that were crucial to examine the several other hillforts being sought to provide material evidence in the immediately surrounding area to for the occupation history of the site. The assess their development relative to Daneb­ magnetometer survey and subsequent exca­ ury (addressing questions such as: when vation at Uffington demonstrated that large they were established, how long they were and medium sized pits were easily detectable occupied for, how many phases of occupa­ with a traverse separation of 1.0m and a tion were represented and when did they go reading interval of 0.25m along traverses out of use?). Magnetometer surveys played (1.0 × 0.25), but smaller post-hole type fea­ an integral part in this process. tures generally failed to register appreciable The nearest hillfort to Danebury, located anomalies, even when the traverse interval 4 miles (6.4km) to the south-east, is at was reduced to 0.5m (Payne 1996). Woolbury near Stockbridge, Hampshire

26 HILLFORT STUDIES AND THE WESSEX PROJECT

(see Fig 1.6 and Fig 2.30). Woolbury tion in each fort by surveying sufficiently appeared, on the basis of the surviving earth­ large areas to show contrasting or recurring work remains, to represent an example of a patterns of activity. The results suggested simple, Early Iron Age hillfort, constructed that the early fort was largely devoid of sig­ at about the same time as the first phase of nificant features, in sharp contrast with the hillfort defences at Danebury (in the 5th later fort, which appeared to contain a mod­ century BC). The straightforward construc­ erately high density of pits of various sizes tion of the ramparts suggested however, that, scattered evenly across the area surveyed. unlike Danebury, it was potentially unen­ Following the survey, excavation in the two cumbered by Middle Iron Age occupation. forts showed that Bury Hill I was probably The results of the fluxgate magnetometer never used intensively, whereas there was survey carried out by the AML between plentiful evidence of high status activity (of 1989 and 1990 clearly indicated a low level the Late–Middle Iron Age) within the of magnetic activity inside the hillfort, sug­ defences of Bury Hill II (Cunliffe and Poole gesting that settlement activity within Wool- 2000b), fully confirming the initial expecta­ bury was of a much lower intensity than at tions based on the magnetic data. Danebury. This interpretation was subse­ Magnetometer survey of a sample of quently confirmed by excavation, which the interior of Old Winchester Hill hillfort revealed that, unlike Danebury, Woolbury (see Figs 1.11, 1.12) carried out by the did not develop as a major focal point of AML in 1995 – again for the purposes of habitation (Cunliffe and Poole 2000a). The improving management and presentation magnetometer survey also confirmed the of the site (in a publicly accessible nature location of the missing eastern section of the reserve) to visitors – produced very similar hillfort ditch, which later excavation showed results to those obtained from the hillfort had been gradually infilled and levelled by of Woolbury. On the evidence of the cultivation during the late Iron Age and magnetic data, Old Winchester appears to Roman period, when a small farming com­ contain only thin scatters of pits inter­ munity was established in the abandoned spersed with empty areas, although features hillfort. This farmstead, which consisted of a associated with a linear group of round series of enclosures defined by narrow barrows occupying a central position within ditches, was detected by the magnetometer the later fort were also detected. as a group of linear anomalies in the eastern part of the survey area. Off-chalk sites In 1990, the second year of the Daneb­ The results of magnetometer surveys at Buck- ury Environs Project, at Bury Hill (4 miles land Rings (Hampshire) and Caesar’s Camp (6.4km) north of Danebury on the outskirts (Berkshire) in 1993 and 1995 (Payne 1993a; of Andover), it was again critical to define Linford 1995) were less informative than the status and development of the hillfort in those obtained from hillforts on chalk geology relation to the neighbouring forts in the area or chalk plateau drift, possibly reflecting less at Balksbury, Danebury and Woolbury (see than optimal geology for magnetic prospec­ Fig 1.6). Bury Hill (see Fig 2.13) had been tion. Buckland Rings (NGR SZ 31 96) lies off interpreted as the remains of two hillforts the chalk on a spur of Pleistocene plateau and (Hawkes 1940) – a smaller, strongly forti­ river terrace gravels deposited over Tertiary fied bivallate enclosure (Bury Hill II) super­ sands of the Bagshot Beds on the south-east imposed on a larger, more lightly defended edge of the New Forest near the coastal town fort with a single rampart (Bury Hill I). The of Lymington. Caesar’s Camp in Windsor earthworks of Bury Hill I are now under­ Forest (NGR SU 864 657) is situated on sim­ stood (Cunliffe and Poole 2000b) to repre­ ilar geology consisting of plateau gravel over sent the remains of an Early Iron Age sands of the Barton Beds. hillfort which, after a long period of disuse, The results of the magnetometer survey was succeeded by the fortification of Bury at Buckland Rings were poor by comparison Hill II. In 1990 the AML carried out flux- with some of the forts surveyed in the years gate magnetometer surveys in each of the previously on the Hampshire chalkland to forts, covering 47 % of the area enclosed by the north. With the exception of sections of the inner fort (Bury Hill II) and a more lim­ the defences, the position of the entrance- ited area of the remaining part of the earlier way plus evidence for a former archaeologi­ outer enclosure (Bury Hill I). It was hoped cal intervention detected along the eastern that magnetometer survey would be able to degraded side of the fort (Hawkes 1936), demonstrate the relative intensity of occupa­ anomalies that could relate to archaeological

27 THE WESSEX HILLFORTS PROJECT

Fig 1.17 (opposite) features in the interior were all but absent. undertake a strategic programme of The location of the Wessex Magnetic susceptibility (MS) values from geophysical survey, in order to extend the Hillforts Survey area the topsoil were low, suggesting geological potential shown by the earlier surveys to indicating the sites included conditions unfavourable to the detection of explore the diversity of hillfort settlement in the project and other features such as pit fills. The apparent patterns at a regional level. main hillfort sites in central absence of magnetic anomalies indicative of Southern England. archaeological features inside the hillfort The development of the Wessex could therefore reflect the local geology hillforts survey programme rather than a genuine lack of internal activ­ ity. Despite the uncertainty over the internal In the wake of all the relatively unstructured character of the hillfort, the survey still pro­ activity described above, came the realisation vided valuable information for informing that non-destructive geophysical survey tech­ the future management of the site, in partic­ niques could make a wider contribution to ular by identifying the position and form of broadening knowledge of hillfort origins, the main eastern entrance through the function and development in central-south­ defences into the hillfort along part of the ern England. The result was a proposal for a defensive circuit where the earthworks are more ordered and wide-ranging thematic poorly preserved. survey project on hillforts focusing on the In 1995, a survey of a sample of Caesar’s chalk downland of Wessex (Fig 1.17), where Camp carried out by the AML (Linford a sound database of knowledge of Iron Age 1995), succeeded in detecting an internal archaeology was already in existence, quarry ditch inside the line of the inner ram­ acquired over many years through the part and a thin scatter of possible pits research by Cunliffe on Danebury and its together with an aggregate of pits in the environs and earlier archaeologists such as interior. Magnetic susceptibility (MS) was Hawkes and Cunnington. This programme highest in the vicinity of the ramparts (sug­ of survey was christened the Wessex Hillforts gestive of occupation concentrated in the Project or Wessex Hillforts Survey. Unlike most area close to them) but MS values recorded earlier hillfort related projects in southern over the rest of the site were very low (sug­ England, the study was designed to be more gesting a lack of iron rich minerals in the ambitious in scale, investigating hillforts topsoil developed over the site). Assuming spread across a wide region but at a relatively that the magnetic evidence is a reliable indi­ coarse level of detail, rather than examining cation of the buried features present within groups of sites in a smaller locality in some the fort, the results from Caesar’s Camp considerable detail as had already been done suggest a relatively sparse degree of activity by the Danebury Environs Project. Although within the area sampled and provide little magnetometry is only capable of providing a evidence for sustained occupation or a relatively coarse level of detail of the buried wealth of interior structures. However, as archaeological features present in a given hill- was the case at Buckland Rings, it was fort, compared to what can be achieved by thought that the identification of subtle intrusive means, a large number of sites can magnetic anomalies would be unlikely on a be covered economically and in a short space site with such extremely low topsoil and of time. The project was designed to bridge subsoil MS values. the gap between these two levels of investiga­ Although in the first half of the 1990s tion and extend the study of hillforts into the geophysical survey on hillfort sites in south­ areas immediately beyond the Danebury ern England was targeted on a largely Environs, drawing upon the backdrop of pre­ piecemeal basis according to management vious detailed research to provide a context priorities, magnetometer survey in particu­ within which to interpret the results from the lar proved capable of making a substantial new sites. One of the principals of the project contribution towards the study of hillfort was to include as many different types of hill- sites. In the majority of cases, geophysical fort (in terms of size of area enclosed and the survey provided the means of assessing form of the defences) as possible, in order to the distribution and intensity of settlement obtain a representative sample of the diverse activity within the interior of a particular range of hillfort sites present in the area (see hillfort, thus providing an insight into the Fig 1.1). This was a particularly important length of occupation of the site, how aspect of the project, designed to enable the space was organised and where different possible interrelationship of hillfort form and activities were carried out in the enclosed function to be examined. The fact that the space. There clearly was, then, scope to project was based entirely on non-invasive

28 Ul

Principal sites referred to in the text : Ca Hd Rh Uc 2. Bw 3. Bn Banbury Hill Hb Hambledon Hill 1. 5. Bd Badbury Rings Hd Hardwell Camp 4. Bs Bh Bathampton Down Ht Harting Beacon 6. Gm Bt Battlesbury Hi Highdown Fo Bi Bilbury Ring Hl Holbury Cs Cg 8. Cb Bw Blewburton Hill Ho Hod Hill Wo Bh Ry 7. 13. Bo Bowry Walls Ld Lidbury Br Bratton Camp Mc Maiden Castle 9. 18. Ld Br 10. Wi Bk Buckland Rings Og Ogbury Sd Bl Bulbury Os Old Sarum Cc Bs Bussock Wood Ow Old Winchester Hill Bt Sb Bo Bz Buzbury Qu Quarley Hill 12/16. Td Ch Cd Vs Cg Cadbury Congressbury Pd Poundbury Yn Qu Og 11. Cs Caesar’s Camp, Rh Rams Hill 17. (Easthampstead) Rw Rawlsbury Bi 14. Ws Gv Fg Cc Casterley Camp Ry Rybury Os Ca Castle Hill, Sb Scratchbury Ci Little Wittenham Sd Sidbury Sc 15. 19. Cr Ce Cp Castle Piece Sc South Cadbury Dn Wh Tt Hl Tb Cr Castle Rings, Donhead Sp Spettisbury Hh Ow Cl Chalbury Tr The Trundle Ht Cb Chisbury Td Tidbury Hb Fr Ci Chiselbury Tt Toothill Camp Bn Ho Cp Tr Bz Ce Clearbury Tb Torberry Rw Ch Cley Hill Uc Uffington Castle Hi Bd Cd Codford Circle Ul Uleybury Sp Dd Dudsbury Vs Vespasian’s Camp Dd Wb Wd Bk Dn Dunwood Camp Wb Weatherby Castle Bl Fg Figsbury Ws White Sheet Hill Pd Fo Forest HIll Wh Whitsbury Castle Mc Fr Frankenbury Ditches Gm Grimsbury Castle Wi Winklebury Cl Gv Groveley Castle Wd Woodbury Hh Ham Hill Wo Worlebury Over 6 hectares enclosed Yn Yarnbury 1.2 - 6 hectares enclosed Under 1.2 hectares enclosed PROJECT AREA 0 50 Kms

1 - 19 Sites included in the geophysical survey programme :

1. Perborough Castle, Compton. Berks 7. Martinsell Hill Camp, Pewsey, Wilts 13. Walbury Camp, Inkpen, Berks 2. Segsbury Camp (or Letcombe Castle), 8. Oliver’s Camp, Bromham, Wilts 14. Woolbury, Stockbridge, Hants Letcombe Regis, Oxon 9. Fosbury Camp, Tidcombe, Wilts 15. Castle Ditches, Tisbury, Wilts 3. Alfred’s Castle, Ashbury, Oxon 10. Ladle Hill Camp, Litchfield, Hants 17. Norsebury Ring, Micheldever, Hants 4. Barbury Castle, Ogbourne St Andrew, Wilts 11. Danebury, Nether Wallop, Hants 18. Beacon Hill Camp, Burghclere, Hants 5. Liddington Castle, Liddington, Wilts 12. Bury Hill and environs (16), Upper Clatford, Hants 19. St Catherine’s Hill, Winchester, Hants 6. Oldbury, Cherhill, Wilts 29 THE WESSEX HILLFORTS PROJECT

methods was another element in its favour, defined by any natural physical features and the project represented a rare opportu­ such as geological boundaries or major river nity to demonstrate that geophysical field­ valleys. The eastern boundary follows a work was capable in its own right of making a north–south line across chalk downland contribution to solving substantive archaeo­ approximately parallel with and just to the logical problems without the need for any east of the A34 main trunk road from disturbance to the sites. Winchester to Newbury as far as the Goring Broad issues that it was hoped geophysi­ Gap. This line places the Tertiary deposits cal survey would potentially be able to of London Clay and Bagshot Beds east of resolve included such questions as: Basingstoke and Newbury largely outside the eastern boundary of the project. The • Are all large, slightly defended early Iron south-east corner of the study area Age enclosures actually largely devoid of coincides approximately with the city of settlement activity – as the few excavated Winchester. The western edge of the project examples suggest? area runs in a north–south line through • Which hillforts appear to exhibit comparable the middle of Salisbury Plain, 10km east densities of occupation to developed hillforts of the towns of Shaftsbury and Warminster such as Danebury and Maiden Castle? up to Devizes in the north-west corner of • Where a series of hillforts have been postu­ the study area. In total the study area covers lated as the largely contemporary centres of approximately 6,000 sq km and includes adjacent territorial blocks (such as those on parts of the counties of Berkshire, Hamp­ the Ridgeway and the South Downs), do shire, Oxfordshire and Wiltshire. they exhibit a similar density and character Collis (1994) has recently stressed the of occupation? pre-eminence of Wessex for British Iron Age • Where two or more hillforts are in unusually studies, and it was clearly important that close proximity, do they exhibit similar densi­ a pilot project involving the large scale ties of occupation? or does one appear to be geophysical survey of hillforts should take more intensively occupied? place against as comprehensive a backdrop of • Does occupation commonly occur outside interpretative data as possible. In addition, hillforts? the efficacy of geophysical techniques on chalk substrates has been amply demonstrated In addition, site-specific issues could be (David and Payne 1997, Payne 2000a) and examined, for example: Does the frequently the selection of primarily chalkland sites was referenced ‘unfinished’ hillfort at Ladle Hill a deliberate attempt to maximise the probabil­ actually contain a settlement? ity of achieving successful results. The diversity of hillfort sites in the area The survey area would allow the study of hillfort interiors relative to the area enclosed and the com­ The area chosen for the study was the plexity of the defences, enabling possible eastern half of Wessex, comprising three relationships between site form and internal main blocks of undulating chalk downland layout to be recognised. Although a wide broken by river systems, including the range of hillfort types are represented in the Hampshire Downs, the North Berkshire area (see Fig 1.1), few have yet been exca­ Downs and the eastern part of Salisbury vated on any scale and therefore the internal Plain (see Fig 1.17). The area contains at characteristics of the majority of the sites, least two major groupings of hillforts: those and the variation in these between sites, of the Danebury region studied by the largely remained a mystery. Danebury Environs Project and the ‘Ridge­ The area also possesses the potential for way hillforts’ of the Marlborough and integrating geophysical survey with access Lambourn Downs on the edge of the chalk and management schemes in association escarpment overlooking the Vale of the with a number of countryside and environ­ White Horse to the north. The area is mental agencies such as The National Trust, bounded by the Upper Jurassic geology of English Nature, local authorities and the the Vale of the White Horse and the Countryside Commission who own or are Thames Valley to the north and the Tertiary involved with the management of several deposits of the Hampshire Basin to the hillfort sites in the region, with scope for south. In contrast to the northern and informing the public about the archaeologi­ southern limits of the project area, the cal significance of the sites. Hitherto the eastern and western boundaries are not lack of data has prevented these agencies

30 HILLFORT STUDIES AND THE WESSEX PROJECT from doing this at more than a very basic Surveys could only be carried out with the level. Furthermore, there was a pressing full consent of the landowners and in need to identify sites with high archaeological one case (Tidbury Ring, Hampshire) per­ potential presently in unsympathetic man­ mission was not forthcoming requiring agement in order to help prioritise and the substitution of an alternative site target conservation initiatives aimed at secur­ (Fosbury, Wiltshire). ing the preservation of sites where archaeo­ The resultant list of sites was then logical information was potentially being considered in terms of its methodological gradually degraded over time through lack and academic integrity. In methodological of intervention. terms it was important that the sample contained a balance of sites with surviving The sites included in the earthwork remains in the interior (for project and selection criteria example Beacon Hill) and sites under per­ manent cultivation with largely plough flat­ The sites selected for survey make up a tened interiors (for example Norsebury representative sample of the various hillfort Ring). In addition it was proposed to types identified in the region. For reasons survey an unexcavated area inside Danebury of cost and because of practical considera­ to provide a control method for assessing tions such as tree cover on some sites, it was how representative geophysical data is of the not possible to attempt a systematic and full archaeological content of a hillfort exhaustive study of all the hillforts in the where it is known from excavation. project area. Two hillforts in close proximity In academic terms the sample was to one another just north of Newbury at checked and, where necessary, augmented Bussock Wood and Grimsbury Castle had to ensure that it included the following: to be omitted from the sample because both are currently in wooded areas. Other 1. Examples of recognised hillfort types such hillforts close to expanding towns such as as large hilltop enclosures, eg Walbury, Andover, Basingstoke and Southampton had Martinsell; univallate contour hillforts, not escaped being built over by housing eg Liddington Castle, St Catherine’s Hill; and road development. Because of the multivallate hillforts, eg Barbury Castle, constraints of such land use on the effective Castle Ditches; and small hillforts, application of geophysical methods an initial eg Oliver’s Camp, Alfred’s Castle selection process was operated whereby a 2. Examples from previously suggested ‘group­ short-list of the most suitable sites for ings’ of hillforts, eg the ‘Ridgeway forts’ survey was prepared from English Heritage (Barbury Castle, Liddington Castle, Uffington management sources. The selection of sites Castle and Letcombe Castle (Segsbury Camp)) also reflected management priorities based 3. Examples from the Danebury Environs on perceived threats to the sites such as (Bury Hill and Woolbury) pressures arising from cultivation and other 4. Examples of hillforts in unusually close forms of erosion. The short-list of sites proximity (eg Danebury and Woolbury; included in the survey programme (see Beacon Hill and Ladle Hill) Fig 1.17, Sites 1–19) was arrived at by the 5. Examples of special interest (eg the following means: ‘unfinished’ hillfort at Ladle Hill).

1. Surface conditions were required to be After this procedure was carried out, the suitable for survey with minimal surface total internal area of all the sites selected obstruction from vegetation or modern was calculated and an attempt made to ferrous contamination. match the amount of survey coverage 2. The underlying geology should be favourable required to the budget available. The short­ for magnetometer survey and reasonably list was finally adjusted to include the widest consistent across the total sample (chalk, possible range of hillfort types including greensand or clay-with-flints). some of the larger examples, such as 3. Where sites were under grassland, priority Walbury Hill Camp in Berkshire, within the was to be given to sites in public manage­ budgetary constraints. This allowed a total ment (such as Barbury Castle) or with of 18 sites to be included in the project with extensive public access. an additional external survey area at Bury 4. Sites with existing adequate geophysical Hill in Hampshire. survey coverage (such as Uffington Castle) Of the 18 hillfort sites selected for were excluded. study by the project, excavation had only

31 THE WESSEX HILLFORTS PROJECT

previously been carried out inside five opposed to more traditional and costly intru­ (excluding Danebury): Bury Hill and sive techniques) for the investigation of hill- Woolbury for the Danebury Environs fort interiors. Using a planned sampling Project (Cunliffe 2000); earlier work by C F strategy (involving a selection of representa­ C Hawkes at Bury Hill and St Catherine’s tive hillfort types), the project attempted to Hill (Hawkes 1940 and 1976); Liddington rectify not only the historic excavation bias Castle in 1976 (Hirst and Rahtz 1996) and towards hillfort defences, but also combined an excavation at Oliver’s Camp near Devizes investigations into the nature of early and by M E Cunnington, published in 1908. developed hillforts, spatial differentiation of All of these documented interventions were function, regionally and at an intra-site level. small-scale and based on a single season Also included in the research design (Trow of excavation. et al 1996) was the exploratory assessment of Seven out of the 18 sites selected for sur­ a number of methods including magnetic vey under the project possessed scope for susceptibility and digital terrain modelling, improved interpretation in their manage­ for rapid characterisation of hillfort interiors ment as public open spaces. and settlement intensity. This approach rep­ Wherever practical, 100% samples of the resented a measured response to archaeolog­ interior of each hillfort were surveyed. In ical problems that might otherwise have some cases this was not possible due to par­ demanded a massive investment in tradi­ tial tree cover or other unsuitable terrain tional excavation, but without being directly such as quarried areas. threatened by development the sites included in the project were unlikely to see The aims and objectives such an investment in the foreseeable future. of the project The project was designed to solve substan­ tive archaeological problems explicitly using In his 1976 introduction to Hillforts: Later geophysical data and data from other non­ Prehistoric Earthworks in Britain and Ireland, invasive sources. Avery writes: The over-arching aims of the project were to provide data for improved manage­ We need the exploration of the interiors of ment and interpretation as well as widening both major and minor forts, and also the academic comprehension of the diverse hill- exploration of nearby settlement sites, on fort types in Wessex, particularly in terms of a scale large enough to throw light on the their relative socio-economic function and population, social structure and economy varying occupation histories as reflected in of these sites. Just as no two sites reflect their internal layout. identical approaches to tactical defence, The specific objectives of the project as so all sites will vary in social structure set out in the 1996 Project Design (Trow and economy. The task of the next 40 et al 1996) were designed to address the years must be to create sound data, and following research questions and academic a sound chronology, as the basis for an issues relating to hillfort sites in southern understanding of these aspects. England:

The Wessex Hillforts Project was initiated i) To support English Heritage casework in an attempt to contribute to this long relating to the conservation and management process of broadening understanding. To of hillforts in the South East and South date our knowledge of hillforts in general West Regions by providing high quality, has been reliant on a limited number of wide-ranging and detailed data on the intensively studied sites such as Danebury, internal archaeological content of hillforts while the bulk of sites remained poorly under­ to assist the putting in place of appropriate stood. The Wessex Hillforts Project was management measures at each of the sites designed to help right this imbalance, there­ starting from an informed basis. This aim fore allowing a more synthetic approach to stemmed from the premise that it is difficult hillfort study. to effectively protect a site if you are largely In a recent collection of papers entitled ignorant of the range of archaeological Science in Archaeology: an agenda for the future features that are preserved within it. (Bayley 1998; Gaffney et al 1998) the Wes­ ii) To obtain information on the internal sex Hillforts Project is described as an exam­ arrangements of hillforts that might other­ ple of a site-based project that employed wise be gradually lost over time as a result geophysics as the prime methodology (as of agricultural erosion. Obtaining such

32 HILLFORT STUDIES AND THE WESSEX PROJECT

information by excavation would be The methods employed by prohibitive in terms of cost due to the quantity and scale of the sites in unsympa­ the project thetic land use. iii) To contribute to improved on-site interpreta­ Survey techniques tion for visitors to the monuments, to promote increased public understanding, Fluxgate magnetometer or gradiometer awareness and enjoyment of the archaeologi­ survey (Fig 1.18) cal heritage. iv) To broaden academic understanding of the Magnetometer survey is the preferred geo­ diverse hillfort types in Wessex, particularly physical method for the initial location or in terms of their socio-economic function general planning of archaeological sites as reflected in their internal layout. On (English Heritage 1995) and for this reason completion of the data collection it was was the principal geophysical survey tech­ hoped that it would be possible for the first nique adopted for the project. Rapid ground time to understand: coverage (at a rate of around 1.5 hectares a • The nature of the internal arrangement day) and the ability, under suitable condi­ of early hill-top enclosures tions, to detect a wide range of buried • The range of internal patterns exhibited archaeological features are the principal by early hillforts advantages of the technique. • The consistency of dense internal activity Magnetic surveying is a passive geophysi­ within the category of developed hillforts cal technique involving the measurement • The functions of small hillforts and their of minute variations in the magnitude or gra­ difference from, or similarity to, enclosed dient of the Earth’s magnetic field at close settlements (numerous examples of which intervals (1.0m or less) across the ground have been surveyed in Hampshire and surface (English Heritage 1995; Clark 1996). Wiltshire and a smaller number in adjacent Modern magnetometers are capable of Gloucestershire and Oxfordshire; Source: detecting magnetic variations or anomalies English Heritage Geophysical Survey over 50,000 times weaker than the natural Database) ambient field strength. Magnetic anomalies • Recurring patterns of spatial organisation. occur in association with archaeological v) To assist the design and development of features due to magnetic susceptibility differ­ appropriate methodologies for the non- ences between their composition and the sur­ intrusive archaeological assessment of major rounding deposits that occur when iron-rich earthwork monuments under different landuse regimes and pressures. vi) To demonstrate the potential of thematic programmes of non-destructive survey in the development of regional research frameworks.

Given the historically proven effectiveness of aerial archaeology on the Wessex chalkland, it was decided that understanding of the individual sites largely based on the geo­ physical data could be markedly enhanced by a study of the existing aerial photo­ graphic (AP) record, held in the National Monuments Record (NMR) at Swindon, from the locality of each hillfort site. The decision was taken to examine the AP evi­ dence within a 2km radius of each site and assess its archaeological significance and Fig 1.18 possible relation to the actual hillforts cen­ Magnetometer survey in tred on. This data is presented in Chapter 2 progress using a Geoscan together with discussion of the topographi­ FM36 Fluxgate cal siting of each hillfort, the interrelation­ Gradiometer (from ships between sites and the ground plan and Archaeometry Branch, EH, surface morphology of each site. Centre for Archaeology).

33 THE WESSEX HILLFORTS PROJECT

minerals in the soil form more strongly ferro­ showing as bunching or striping of alternate magnetic materials such as magnetite or lines of data (by equalising the mean of magheamite. This magnetic enhancement is each line of readings). In some instances the usually related to burning, although more data were also smoothed slightly, to improve subtle inorganic and bacterially controlled the definition of archaeological anomalies mechanisms may also play a part under suit­ greater than a metre in width, by the use of a able soil conditions. Such conditions occur Gaussian low-pass filter with a radius of naturally in most topsoils, providing a source 1.0m (Scollar et al 1990). of magnetically enhanced material that The range of archaeological features becomes incorporated in archaeological fea­ generally detectable by magnetometry at tures and so produces almost indelible mag­ hillfort sites of Bronze Age and Iron Age netic signatures, even where features have date on chalkland geology includes: infilled been all but erased by intensive agriculture. ditches defining internal enclosures or Magnetometers also respond to the strongly divisions and other earth-filled features magnetic signals produced by heavily fired including silo and rubbish pits, irregular structures that have become permanently quarries or scoops, and shallow ‘working magnetised as a result of intense heating. hollows’. Annular gullies defining the This permanent thermo-remanent magnet­ former positions of round houses of Iron ism is found in domestic and industrial fea­ Age date were detected at Segsbury Camp tures containing fired clay such as hearths, and subsequently confirmed by excavation. kilns, furnaces and ovens, and in some cases Numerous other examples exist at hillfort burnt stone structures (Aitken 1974, 141–7). sites both in Wessex and farther afield, Magnetometry, coupled with aerial pho­ including South Cadbury Castle, Somerset tography, has been recognised for many and Conderton Camp, Worcestershire. years on the Wessex chalkland as a powerful Ovens, furnaces and hearths, both of indus­ method for planning prehistoric settlements trial and domestic type, would also be and landscapes. The series of surveys car­ expected to register appreciable magnetic ried out for the Danebury Environs Project anomalies. One noteworthy example of a from 1989–96 (Payne 2000a) demonstrated large oven of key-hole shaped plan, detected that the technique is particularly effective on by magnetometry and subsequently con­ the chalk and chalk plateau drift of this firmed by excavation, occurred at Uffington region, where anomalies, caused by higher Castle (Payne 2003a). magnetic susceptibility of the soil concen­ It would be misleading to suggest that trated in buried archaeological features (pri­ magnetometry can provide a complete picture marily the infilling of features cut into the of all the activity and occupation within a hill- chalk such as ditches and pits), stand out fort. Some important categories of features clearly against the relatively much lower can be missed. This applies in particular to magnetic background from the surrounding some smaller, shallow and less substantial fea­ natural substrates. tures such as gullies and post-holes (especially All of the magnetometer surveys carried where truncated by ploughing), and also some out for the Wessex Hillforts Project pits and graves, which may only offer a poor employed Geoscan FM36 Fluxgate Gra­ magnetic contrast between their fill and the diometer type instruments with built-in surrounding natural chalk (for example a pit data-logging facilities enabling digital data filled with chalk rubble). capture of about 16,000 readings in a two Comparison of the geophysical data from hour survey session. The instruments are the excavated samples of Uffington Castle sensitive to changes in magnetic flux density and the inner camp at Bury Hill provides a of a tenth of a nanotesla (nT). In all cases clear example of these limitations (Payne the data were collected on a 30m grid, at 2000a, 2000c, 2003a). Generally only the 0.25m intervals, along traverses spaced 1.0m larger pit-type features (and in the case of apart. This represents a compromise, by Uffington, the oven) were represented in the which larger area coverage was achieved at magnetic data, while the majority of the the expense of possibly missing smaller smaller features recorded during excavation archaeological features that might have been were not visible. The application of more detected by narrower instrument traverses sensitive caesium magnetometers in recent (halving the separation between traverses years is now improving the detection rate of from 1.0m to 0.5m, for example). Data pro­ narrow circular gullies and slots and post­ cessing involved the initial elimination of the hole structures within Iron Age settlement effects of thermally induced instrument drift, complexes (Payne 2004).

34 HILLFORT STUDIES AND THE WESSEX PROJECT

General experience of magnetic pros­ response to larger-scale variation in solid or pecting on chalk in southern England using drift geology (for example an area of plateau fluxgate gradiometers has shown that they drift as on the hill occupied by the hillfort of are rarely equal to the task of locating Woolbury) does not normally occur. How­ smaller post-holes (typically 0.3m in diame­ ever, the partially clay-capped hill occupied ter and 0.3m deep), regardless of the sam­ by Woolbury hillfort shows a more confused pling interval being used (see Payne 1996). magnetic background compared to those Therefore, remains of stake-built structures sites where the geology is more uniform (such as some common forms of Iron Age (Payne 2000b). The problem would appear round house) are unlikely to be detectable to be particularly severe in the case of Wal­ except where associated features such as bury Hill on the northern scarp of the hearths, surrounding gullies or deposits of Hampshire Downs and at the highest point burnt daub are present. of the chalk geology in southern England. This was shown to be the case at an Purely natural pockets of clay-with-flints are Early Iron Age enclosed settlement at known to occur within the chalk at the hill- Houghton Down near Danebury, surveyed forts of Segsbury and Uffington Castle and in 1994 in advance of excavation (Payne produce anomalies similar to those associ­ 2000a, 2000d). Here, the round houses ated with man-made features such as pits associated with the earliest Iron Age phase and quarries. There is therefore a potential of the site, discovered in the process of exca­ danger of misinterpreting natural features of vation, were invisible in the magnetometer the geology as archaeological features. Geo­ data. If generally applicable, this situation logical features might be expected to exhibit would unfortunately result in important cat­ more irregular form and more random pat­ egories of activity at Iron Age sites being terning than archaeological features, but under-represented in standard fluxgate experience shows that it is not always possi­ magnetometer surveys – a limitation that ble to differentiate reliably between the two. should always be borne in mind in the inter­ Magnetic susceptibility survey (Fig 1.19) pretation of the data. Larger than average post-holes (such as those constructed to Detailed magnetic susceptibility (MS) sur­ retain the doorframe posts of timber houses veys were carried out at two of the hillforts or the foundation sockets of large four-post with ploughed interiors – Norsebury Ring structures) are comparable to small pits and and Castle Ditches – where the results of the therefore more easily detectable even at magnetometer surveys proved particularly standard 1.0m × 0.25m sample intervals. A interesting. The magnetic susceptibility sur­ few isolated examples of possible four-post veys were designed to provide additional structures detected by magnetometer survey information to support the interpretation of have tentatively been identified at Uffington the magnetometer surveys. Castle and Perborough Castle in Oxford­ Different materials become variably mag­ shire (Chapter 2 this volume) and at Con­ netised in the presence of the Earth’s mag­ derton Camp in Worcestershire (Chapter 3 netic field. The degree to which soils become this volume and Payne 2005. The latter magnetised in the presence of this external site is situated on particularly favourable induced magnetic field is known as the mag­ geology for magnetic prospection (Middle netic susceptibility (MS) and depends on the Jurassic Inferior Oolite) and in these condi­ concentration of naturally occurring iron tions post-hole type structures would be oxides they contain, and the extent to which expected to be easier to resolve than similar these have been modified to more magnetic features on chalk. forms by various mechanisms. These are not In areas of predominantly chalk geology, as yet wholly understood but seem to be features of geomorphological origin may linked with a past human presence on a site sometimes register in a magnetometer sur­ (Tite and Mullins 1971; Clark 1996, chapter vey, particularly in areas where the superfi­ 4). Concentrations of soils that have become cial geology is variable, or has been artificially magnetically enhanced (increasing influenced by periglacial conditions. The their MS) as a product of human occupation influence of scoring and fissuring of the sur­ can be defined by topsoil magnetic suscepti­ face of the chalk has been noted in magne­ bility measurement. A susceptibility survey tometer surveys of several sites in the may, therefore, supplement and confirm the Danebury environs, including Bury Hill and findings of a magnetometer survey by indicat­ New Buildings. The fluxgate gradiometer is ing the areas within a hillfort where features sensitive only to localised soil changes, so a and debris of domestic and possibly industrial

35 THE WESSEX HILLFORTS PROJECT

origin are most concentrated. This is of therefore employed on a 5m grid to give particular interest within hillforts such as detailed coverage of each site. Additional soil Norsebury and Castle Ditches that exhibit samples were collected at 20m intervals to signs of internal divisions or smaller internal enable laboratory readings to be carried out, enclosures. In such cases, MS survey may be as a check on the field measurements and as capable of defining any concentrations of a test of the consistency of the results from activity associated with these discrete areas, the two techniques. Because of the possibil­ therefore helping to shed light on their func­ ity at Castle Ditches of the readings being tion or the nature of the activities carried out affected by stones in the soil samples, a set of in particular zones of the hillfort. laboratory readings was also obtained after Two alternative procedures are com­ sieving the samples through a 2mm mesh. monly used in archaeological magnetic sus­ The results from the MS surveys are ceptibility surveys, the first of which is to presented in the sections on Norsebury collect volumetric susceptibility readings on and Castle Ditches in Chapter 2 (Figs 2.26 in-situ soil using the Bartington MS2 meter and 2.48). and MS2-D field sensor (Fig 1.19). This method allows rapid ground coverage, but Digital terrain modelling for accuracy it requires close contact by Tom Cromwell, Nick Burton and between the ground surface and the detector Andrew Payne coil. It may therefore produce a slightly dif­ ferent response to the alternative method of Background taking readings in the laboratory directly on This element of the project was undertaken soil samples collected from the site. Labora­ by staff of the former Central Archaeology tory samples are air dried, weighed and mea­ Service (CAS) at the request of the Ancient sured using the Bartington MS2-B sensor, Monuments Laboratory. The aim was to and mass specific susceptibility values can provide topographic models onto which then be calculated by standardising the geophysical data could be ‘draped’ for pre­ instrument readings to a 10g sample weight. sentation and interpretation. The even surface of the rolled plough-soil The advantage of digitally modelling inside the two hillforts provided suitable detail of the site microtopography is that ground conditions for the acquisition of the data (providing the resolution is good quality MS data using the field mea­ sufficient) can subsequently be manipulated surement technique (Fig 1.19), allowing and interrogated to extract information on good contact to be made between the field the most subtle of earthwork features sensor loop and the soil. This method was (see, for example, Chapman and Van de Noort 2001; Newman 1997). This approach is not possible with a fixed map- type view of the traditional hachured kind, although hachured plans have clear advan­ tages of their own, such as indication of phasing between earthworks, detailed ground observation during the survey process and a much greater analytical element. When combined with GIS soft­ ware the digital terrain data can be viewed from different directions and overhead angles in order to highlight specific features and areas such as recessed building plat­ forms terraced into the slopes of a hill. Vertical exaggeration of height readings can be applied to enhance the visibility of very Fig 1.19 slight earthwork features and light shading Magnetic Susceptibility can be applied from various angles and survey equipment directions to emphasise subtle surface detail manufactured by by the shadowing effect this generates. Bartington Instruments (from Archaeometry Survey methodology Branch, EH, Centre for The survey data was collected on a grid Archaeology). pattern of points. The data points needed to

36 HILLFORT STUDIES AND THE WESSEX PROJECT

Fig 1.20 GPS surveying equipment used to produce the three dimensional topographical models of selected hillforts. a) Trimble Navigation 4600LS post-processing GPS equipment b) Leica Geosystems System 530 real-time kinematic equipment (from EH, Centre for Archaeology and courtesy of Leica Geosystems Ltd).

be very accurate, with maximum permissi­ would be quite distinct in a model. Beacon ble error margins of only a few centimetres Hill was also of median size with visible fea­ in Easting, Northing and height in order to tures, and its close proximity to Ladle Hill create models that were accurate at the added academic interest. Ladle Hill was scales at which they could be usefully included because it appeared to be an unfin­ viewed. These models would then be the ished fort, and was thus exceptional. In the next best thing to being out on site. To do case of Ladle Hill the partially constructed this, however, meant surveying each hillfort defences and associated dumps of rampart in great detail. The only practical solution material were fully included in the survey. was to use GPS – a surveying version of the Finally, Oldbury was selected as a very large satellite navigation equipment used in avia­ site with abundant visible features. tion and marine applications. Each site was In 1996, four of the sites (Alfred’s Cas­ first divided into convenient sections using a tle, Barbury Castle, Ladle Hill, Oldbury) baseline through the middle of the site, and were surveyed using Trimble Navigation each section then gridded-out using tapes 4600LS post-processing GPS equipment, and ranging poles to ensure that data was with the roving receivers mounted on a two- collected evenly across the whole hillfort. metre pole that the surveyor carried (Figs The GPS equipment was then carried along 1.20(a), 2.22, 2.36 and 2.45). This equip­ the grid lines, taking readings at fixed inter­ ment required the downloading and pro­ vals to produce an even distribution of data. cessing of data at the end of each day in As the technique was being used to order to turn the raw data into a set of 3-D map topographical detail, only sites with coordinates that could be examined and evidence of surface features in the interior modelled in Computer Aided Design were selected, although in retrospect it may (CAD), a process which made it impossible have been equally valuable to test the to see gaps in the data until after the day’s methodology on sites that are more difficult fieldwork was complete. The receivers were for traditional earthwork survey, in particu­ set to take readings at a fixed time interval, lar those with tall vegetation cover. At such and were then carried along the grid lines at sites the technique may have a particularly a set pace to get an even rate of data collec­ useful role for picking up earthworks that tion. Where significant details were encoun­ can’t be seen by eye because they are tered the pace was slowed to capture more obscured by vegetation. points in order to get smoother models. The The final selection of hillforts for topo­ nominal data interval was 2m between graphical recording was Alfred’s Castle, points, with extra data points around any Barbury Castle, Beacon Hill, Ladle Hill and visible breaks in slope such as the edges of Oldbury. Alfred’s Castle was of interest as a sharply defined features, in order to obtain very small site, not set on a hilltop, with very accurate models using Digital Ground pronounced earthwork evidence in the inte­ Modelling III (DGM3) software that CAS rior. Barbury Castle was of median size, but employed at the time. In the event, the post­ exhibited a wealth of visible features that processing nature of the equipment meant

37 THE WESSEX HILLFORTS PROJECT

that point intervals averaged closer to 3m in ated. Once the raw data was checked most cases. Beacon Hill, by comparison, through CAD modelling, the points were was surveyed in 1999 using Leica Geosys­ exported to ARCINFO or GEOSOFT tems System 530 real-time kinematic equip­ OASIS MONTAJ to be modelled and ment (Fig 1.20(b)), which eliminated draped with the geophysics plots. post-processing by giving Ordnance Survey The results of the GPS surveys are pre­ coordinates instantly through the use of on­ sented and discussed in the relevant section board radios and processors. Experience in Chapter 2. with pole-mounted equipment indicated Documentary research and aerial that height data would not be compromised by a backpack-mounted system, so the photographic analysis backpack-mounted antenna was used and The final stage of the project, following the the pole was discarded. Beacon Hill was sur­ completion of the internal mapping of the veyed at an interval of 1m by setting the subsurface and surface evidence for activity receivers to capture data every time they in the hillforts, was devoted to researching moved more than 1m from the previous the immediate landscape setting and the reading. The equipment also kept track of broader regional context of the sites the grid lines to be walked, guiding the sur­ included in the survey. The first step in this veyor along each line without the need for process was to assemble and interrogate tapes or ranging poles. The results (see Figs existing published sources of archaeological 2.11 and 2.12) were faster, and more accu­ information on each of the sites, and any rate than the previous surveys, with little records of artefactual material they may wasted time. It should be noted that Trim­ have produced, in order to attempt to gain ble Navigation also offers a real-time kine­ some insight (however limited) into relative matic system (the 4800 model) with these dates of occupation. same benefits. This phase of analysis also involved the All of the surveys were plotted relative to study of the morphology of the hillfort and the Ordnance Survey grid (OSGB36). For the preparation of a description of the main the early sites, this was accomplished by sur­ visible surface characteristics of each of the veying the sites on an arbitrary grid with sites included in the project (including ram­ pegs to mark the baseline, followed by a part form, entrances and any visible earth­ control survey to tie the pegs into OSGB36 work features in the interior). by surveying them relative to a series of local In addition, the relationship of each site trig pillars. In the case of Beacon Hill, how­ to the broader pattern of hillfort distribution ever, there was a trig pillar within the site so in Wessex was considered together with the survey grid was established on OSGB36 location, aspect, relationships with geology at the start. and soils, known land allotment patterns in From the outset the project was aimed at the immediate vicinity and evidence for modelling the internal ‘living space’ of each extra-mural settlement – enclosed and open. hillfort, corresponding to the area surveyed The latter component was addressed largely by geophysics. For practical reasons the by examination of aerial photographical topographic surveys were carried up to the records. The aerial photographic material top of the ramparts, thus modelling the from a 2km radius around each site was inner slopes of the defences. examined for the presence of other forms of settlement in the vicinity of the hillfort and Data processing evidence for field systems, tracks and linear All of the point data were imported into boundary ditches in an attempt to recognise AutoCAD for editing and modelling, at any possible relationships between these var­ which point they could be separated into ious features that would suggest a develop­ items such as boundaries and paths. The mental sequence for the site in question. files were divided into appropriate layers. The analysis of the surface and docu­ The first four sites were then modelled in mentary evidence relating to each of the DGM3 to create contour maps and gridded sites and their landscape setting is presented triangular mesh surfaces, but these were under the heading ‘morphology and setting’ subsequently remodelled using Key Terra in Chapter 2 followed by discussion of the Firma IV (KTF4) to produce Triangular geophysical evidence from each hillfort. Irregular Networks (TINs) and contour This format was chosen in order to present plots. The fifth site (Beacon Hill) was also all the information on each site together in a modelled in KTF4, and a contour plot cre­ single unified entry.

38