Internal Organisation and Deposition at the Iron Age Temple on Hayling Island (Hampshire)

by Anthony King and Grahame Soffe

Summary

This paper discusses the Iron Age phase of the 'Celtic' and Romano-Celtic temple excavated on Hayling Island 1976-81. A sub-division of the Iron Age features into two periods is proposed, the first comprising a square open-air enclosure set within an outer enclosure, dating to the mid 1st century BC. The succeeding period, from the late 1st century BC to the conquest period, perhaps as late as c. AD 70, saw the probable abandonment of the inner enclosure in favour of a sub-circular building of round-house form that was set on the eastern margin of the enclosure around a pit that had apparently formed an important focus from the foundation of the shrine. The artefactual assemblage is important for the coins and metalwork, and displays characteristics that set it apart from assemblages from settlement sites. The material was almost certainly deposited as a result of ritual activity in the temple enclosure. The zonation of the deposits within the enclosure was also noteworthy and probably significant in ritual terms, being predominantly located in the south-eastern sector of the outer enclosure.

Introduction

The site at Hayling Island was first excavated at the beginning of this century (Ely 1908), but at that time the form and nature of non-classical temples was poorly understood and the temple not recognised for what it was. Lewis (1966), in his review of evidence for temples in Roman Britain, recorded the existence of the site as a possible temple, but left the question of its function open due to lack of detailed evidence. Interpretation of the site remained in this uncertain state until 1975/6, when new aerial photographs of high definition and clarity (Fig 1) enabled the present authors (together with the late R. Downey) to suggest tentatively that the form resembled a Romano-Celtic temple, and to mount a campaign of excavations (1976-82). It was immediately apparent that Iron Age material underlay the Roman phase, and structural details of an Iron Age shrine gradually came to light in successive seasons. The focus of this paper is on the Iron Age phases. Details of the Roman phases and finds can be found elsewhere (King and Soffe 1994a, 1994b) and in the final report, which is in preparation.

The Iron Age temple and its phasing

The first use of the site appears to have been in the early/mid 1st century BC, and it continued up to and beyond the Roman conquest, being replaced by the large stone-built temple in the 60s/70s AD. It is apparent from analysis of the excavated features that there was more than one structural phase, although the nature of the fine brickearth subsoil made it difficult to detect stratigraphic sequences in some parts of the site, and in general there was a dearth of intercut features. In view of the absence of clear stratigraphic indicators, the best approach to phasing the Iron Age temple is to propose a developmental model, which can subsequently be checked against the chronology as revealed by the artefactual evidence. The model adopted here is that of an increasing tendency to tectonisation, i.e. that religious structures early in the sequence are more likely to have been open-air in nature, while later structures will probably have been roofed. This is the type of sequence demonstrated at Gournay (Bruneaux et al. 1985) and hypothesized as a general sequence for many late Iron Age temple sites (King 1990). An additional factor that is specific to Hayling is that the circular structure (almost certainly a building) underlies the Roman temple, which takes the former's plan almost exactly. This strongly implies that it was in existence late in the sequence and still stood to within a short time of the construction of the Roman temple. Using these parameters, it is possible to propose the following general phases:-

Phase I Three main elements formed the focus of the ritual ensemble: an enclosure (c. 25 x 25 m) with its entrance aligned to the east, an inner enclosure, also with an eastern entrance, and a pit set on the western margin of the inner enclosure.

The chronology of phase I is best established from the stratified coins, since the pottery fabrics and forms are undiagnostic in this respect. The coins listed in Table 1 can all be dated to the early-mid 1st century BC, up to the 30s BC, and suggest that the phase can be placed in the mid 1st century BC or a little later.

Phase II The elements that replace phase I use the same outer enclosure, but the inner enclosure was demolished to make way for the circular structure built around the central pit. The circular structure is made up of an inner gully, 9.2 m in diameter, with post-holes within it, presumably forming the foundation of the walling, and an outer gully of variable depth that appears to have served to drain water away from the structure's foundations. Both these features were better preserved on the south side of the circle. On the east side, the gullies terminated in post-holes forming the entrance into the structure. This seems to have been a simple, slightly projecting porch forming an entrance of c. 2.8 m, although it should be noted that the construction of the Roman temple may have truncated any eastward-projecting extension to the entrance. In the interior of the structure, post-holes were set out in a loosely shaped arc around the south, west and north sides, and probably served to support the roof. To all intents and purposes, this structure is a typical round-house, of the general type known from many Iron Age sites in southern Britain. Architecturally, nothing distinguishes it from a domestic structrure, and it can best be considered as a house for the deity, of a type that would have been familiar to inhabitants of the region in a secular context. What served to differentiate it from a primarily domestic building, however, is the central pit, which continued in use, and the votive deposits, which also continued to be placed in the courtyard around the structure (see below). The outer boundary of the courtyard was also reorganised in this phase, resulting ina deeper, more clearly aligned ditch, particularly on the east, south and west sides. The eastern entrance is of some interest, since the ditch terminals are slightly incurved, and in the case of the southern terminal, also significantly deeper than the ditch itself. Excavation of this terminal indicated that it had a flat rectangular bottom (c. 0.80 x 0.40 m) that may in fact not have been a sump, as might be expected, but the bottom of a post- or stone-hole. The fill of this feature consisted of a primary deposit that lined the sides, in the manner of a post or stone packing, and a secondary deposit (dated c. AD 50/60) that filled the central part of the feature. The interpretation of this terminal favours the notion that it contained an upright stone that served as a marker for the entrance to the temple. Support for this comes from the finding of large rectangular sarsen boulders in the foundations of the entrance porch to the Roman temple courtyard, indicating that large stones were available and indeed were not uncommon in the region, to judge from the results of the survey of sarsen stones in Hampshire (Gallup 1994). It is possible, but unfortunately not provable, that the sarsen boulders in the Roman foundations were derived from the Iron Age phase, and had been deliberately incorporated in the structure of the Roman temple. An additional factor in favour of this interpretation of the feature is that standing stones are known to have been in use in the Iron Age, principally in Gaul, for instance at Vieux-Poitiers (Lejeune 1988:70-82) and in Armorica (in probable funerary contexts; Daire and Villard 1996). A Romano-Celtic temple constructed around a standing stone is known at Triguères, Loiret (Boutet de Monvel 1863; Horne and King 1980:482-3). Dating of phase II is determined for the most part by the coins. The Iron Age coinage shows an apparent gap in the early 1st century AD (see below), and this may coincide with the change from phase I to phase II. Indeed, it is possible that the temple declined in use for a short period, but following the construction of the circular building in the early decades of the 1st century AD, activity picked up and continued to the construction of the Roman temple in the 60s AD. The coin assemblage reflects this, with a higher percentage of coins from the decades leading up to the Roman conquest. It is quite likely that Iron Age coins continued to be deposited up to the time of the construction of the Roman temple, and indeed possibly beyond, to judge from the high percentage of late coinage from stratigarphy associated with the Roman temple.

Finds from the Iron Age temple

The objects deposited include a large number of coins, mainly celtic coins of the immediate area, but also those of the peoples to the west and a significant number from Gaul, primarily from Armorica and central/northern Gaul. There are also some Roman republican coins, unusual for a British Iron Age site. Analysis of the coins (Briggs et al. 1993:35-41; Haselgrove 1987:129-30) shows that they are relatively early in date, the majority being of the mid/late 1st century BC. and there may have been a gap in the practice of coin deposition in the early 1st century AD. A significant feature of the coin assemblage is the high percentage of plated coins (78% of the gold and 46% of the silver coins). This is argued to be a deliberate element of depositional practice, and appears to reflect the selection of plated coins for deposition at the temple. Presumably these must be irregular issues or coins deliberately manufactured to save valuable precious metal. As such, their symbolic value in the depositional context is enhanced, at the expense of the interpretation that they represent the destruction of wealth by its neutralisation in a votive locus. As a rider to this, Briggs et al. (1993:44-5) have suggested that the percentage of solid gold and silver coins may originally have been higher, if these were selectively positioned in or on the (phase II) temple structure (together with other valuable votive offerings). Their depositional history would therefore have been different, and it is unlikely that they would have been incorporated into the excavated deposits. Other votive material from the site includes two fragmentary 'currency bars', fibulae (a significant number being of Nauheim type), shield binding, iron spearheads, vehicle fittings and some fragmentary human remains. The horse and vehicle equipment from phases I and II includes a three-link bridle-bit of cast bronze with bronze-cased iron rein rings, datable to the 1st century BC. This bit, of Palk's 'double-jointed snaffle' category (Palk 1984, DJ17), has its most notable parallels from the Lady's Barrow and King's Barrow, Arras, Yorkshire (Stead 1979), Hengistbury Head, Dorset (Palk 1987:151-2), Otterbourne, Hants (Denford 1993:39-41) and elsewhere. The last-mentioned site is only a short distance from Hayling Island. A terret with knob decoration of Arras type (Leeds 1933, type I) was also found and is probably of similar date. One of the more remarkable objects found was a bronze yoke-terminal with inlaid red enamel decoration on its terminal knob (Fig 2) It is without parallel in Britain, but is almost identical to examples from Mont-Beuvray (Goudineau and Peyre 1993:115-7). There are several iron linch pins, paralleled at Worthy Down, Hampshire (Dunning et al. 1929) and Llyn Cerrig Bach, Anglesey (Fox 1946) and two iron nave hoops of a type found in the Yorkshire and Champagne vehicle burials (Stead 1979) and also at Llyn Cerrig Bach. A number of bronze and iron rings from the site may also have been horse or vehicle trappings, or from a warrior's equipment, as at Owslebury (Collis 1973), but there is also the possibility that they are specially manufactured votive objects, such as has been suggested for those from Uley temple, Alesia and elsewhere (Bailey and Woodward 1993:135-40). The martial equipment forms another important category of deposit on the temple site, and is also a characteristic of other Iron Age ritual sites in Britain (Woodward 1992:66-7). It includes a group of iron socketed spearheads. Most of them resemble examples from the hillforts of Hod Hill, Dorset (Brailsford 1962), Maiden Castle, Dorset (Wheeler 1943) and Danebury, Hants (Cunliffe 1984:361-6). The other significant weaponry find was of numerous fragments of edge binding and terminal knobs for shields, similar to that from Gournay (Brunaux and Rapin 1988:223). Recent finds have clearly demonstrated the purpose of this type of binding for shields rather than sword scabbards (Stead 1991). Three belt-hooks for baldrics were found, of which one winged example is only the second of its type to be found in Britain. The other example comes from Owslebury 'warrior' burial (Collis 1973), and both are probably continental imports. Another significant class of find is the flintwork. Mostly it consists of scrapers of Neolithic/Bronze Age type which appear to have been deliberately deposited with the other artefacts. There is no sign that the site had pre-Iron Age activity that was disturbed and redeposited. Also of probable significance in this respect are a Neolithic polished flint axe and a Mesolithic axe from the topsoil over the temple. Although less well stratified, these can probably be linked with the practice observed at Romano-Celtic temple sites (e.g. St-Aubin-sur-Gaillon (Eure); Poulain 1913; Horne and King 1980:457-8) and elsewhere (Adkins and Adkins 1985; Bradley 1986) of depositing axe-heads and other much earlier artefacts. Of additional relevance in this respect is the broken Middle Bronze Age spearhead from the northern post-hole of the entrance to the phase II circular structure. It seems likely that objects from earlier prehistory, perhaps found casually during the Iron Age as a result of activity that disturbed earlier material, were brought to the temple as votive offerings. The small assemblage of human bones from the temple, consisting of parts of a cranium, a mandible and broken limb bones from one or more young adults, obviously raises the vexed issue of whether or not they represent human sacrifice, due to the association of this practice with Celtic religion in ancient literature and the finding of apparent remains of victims at Gournay (Brunaux et al. 1985) and Ribemont (Cadoux 1982). The Hayling Island remains are too fragmentary to contribute significantly to this debate, sonce no forensic data were observable, nor were their depositional positions unusual or suggestive of sacrifice. There are two points to note: firstly, the bones are amongst the general assemblage of votive offerings in the temple area and as such probably represent votive offerings, and secondly, the scattered and broken nature of the bones indicates either the deposition of disarticulated (perhaps quite old) bones as offerings, or the disturbance or a burial or human corpse within the temple area. Whether that burial resulted from a sacrifice is entirely unknown. Unfortunately the number of bones is too small to detect zonation of parts of the body, as observed so dramatically at Ribemont. The animal bones from the Iron Age phases are of some interest in that they show an almost complete lack of cattle bones, with only a handful of specimens being noted, mainly from zones outside the temple enclosure. Sheep and pig bones formed the large mammal assemblage, with sheep predominating in terms of the number of fragments (sheep 59%, pig 41%, n = 2395). This degree of selectivity in the assemblage is also seen at other religious sites, such as Uley where goat and sheep are predominant (Levitan 1993), and is usually assumed to be a result of selection of animals for sacrifice according to precepts that were particular to the cult being worshipped. For Hayling Island, sheep and pig were apparently the two species regarded as worthy of forming votive offerings. The parts of the carcass represented was also a noteworthy feature: for sheep, upper limb bones and mandibles were most common, while horncores and phalanges were rare (the latter being an absolute lack since the sieving programme did not reveal a higher percentage); for pig, mandibles and maxillas were most common, followed by upper limb bones, while lower extremities were rare. In terms of purely dietary considerations, this represents the better-quality meat, together with parts of the crania which were not necessarily highly favoured. It is possible that the better cuts of meat were offered as sacrifices to the deity, as well as parts of the skulls. The poorer cuts of meat were probably consumed by those making the offerings and the remains disposed of elsewhere. The lack of horns is not easy to explain, but certainly represents a significant degree of selectivity in depositional practice.

Deposition Practice and Zonation

Deposition within the site was primarily within the outer courtyard area, and showed a marked bias towards the south-eastern sector, i.e. on the left-hand side as worshippers came within the entrance of the temple area (Fig. 3). This was very probably the result of the ritual used for disposing of votive offerings. It must, however, be acknowledged that the northern part of the enclosure appears to have been truncated stratigraphically during the construction of the Roman temple, probably in an attempt to level the ground surface. This has resulted in some of the features being relatively shallow on the north side of the phase II circular structure and in the enclosure area to its north. Having taken this into account, it still seems to be the case that the bias of deposition to the south-east sector of a real phenomenon of Iron Age practice, since enough of the stratigraphy survived over the whole of the Iron Age temple area to demonstrate this. Several classes of artefact show this zonation. The coins (Fig 3A) tend to be clustered near the entrance to the outer enclosure, between the outer and phase I inner enclosures and in features associated with the phase II circular structure. Amongst these are a couple of Roman republican coins from the circular structure (but otherwise no Roman coins in this phase within the enclosure as a whole) and a 'hoard' of six coins of the Carnutes found together in a deposit within the south-east corner of the inner phase I enclosure. Fibulae, items of personal adornment and other bronze objects such as rings and edge binding (Fig 3B) are located in approximately the same zones as the coins, except in the case of the phase II circular structure, where they tend to be found in the central pit rather than the circular gullies. The pit yielded pieces of bracelet, rings, fibulae, parts of a mirror and other dress items, which is probably a depositional bias of deliberate votive significance, relevant in particular to the period when the pit was infilled towards the end of phase II. Another concentration of bronze finds consisted of fibulae located adjacent to a mudstone block to the west of the south entrance terminal. The iron work (Fig 3C) clearly demonstrates that the south-east part of the outer enclosure, particularly on its eastern margin, was considered a focus for deposition. The great majority of the iron objects are small unidentifiable fragments, but also of significance are nails (associated probably with wooden artefacts now decayed) and the spear-heads and knives mentioned above. There were also two broken pieces of 'currency bar' from Iron Age levels (Fig 3D). Miscellaneous finds are plotted in Fig 3D. Of note is the human bone, which is located adjacent to the south-east and south-west corners of the outer enclosure and also in the main south-eastern deposition zone within the enclosure. Another significant depositional practice on the site was that many of the artefacts were deliberately broken or bent, including several of the coins (Briggs et al. 1993:2-3), and in addition, spearheads were often reused or sub-standard. The action of breaking or bending artefacts can be interpreted as indicating that the objects were 'killed' in an act of dedication to the deity by rendering them useless (Webster 1986:132). Whether or not this is the most appropriate explanation, it is clear that the high concentration of artefacts of many different classes indicates that objects, vessels and animals were brought to the site, 'sacrificed' in some form to the deity, and the physical remains left as votive offerings. These offerings appear to have been regarded as inviolate (as reflected in Caesar's reference (BG VI,13) to the heaps of spoils in the territory of the Carnutes), and were left in and around specific locations in the temple enclosure, to be disturbed and redeposited again and again in subsequent acts of votive deposition.

Conclusions

Comparison of the votive material with other sites, combined with data from the Roman phase, indicates that the shrine was dedicated to a celtic Mars-type deity (Downey et al. 1980:300). There are significant links with Gaul during the Roman phase, particularly in the architectural form of the temple, which has its best parallels in the large circular temples of western Gaul, such as the Tour de Vesone, Périgueux. It is most likely that these cross-Channel links originated in the Iron Age The temple shows many points in common with Gournay and other northern French temples, in terms of the overall sequence, and was probably constructed as a result of the so-called Belgic influence on Britain in the first century BC. In this respect, it is pertinent to mention the reference by Caesar (BG VIII,48; Frontinus, Strat. II,13,11) to the flight of Commius to Britain in c. 50 BC, and the geographical location of this event by Hawkes (1977:184, map 12) in the Chichester/Portsmouth area. In addition, Cunliffe (1991:108-10) makes a case for the Belgae attested in the Winchester area in the Roman period to be a genuine reflection of a pre-Roman situation. Thus, there is a possible context for the importation of religious ideas, including those concerning temple construction, alongside political and other influences flowing into central southern Britain during the Late Iron Age. The phase I temple was probably set up at this time and indeed indicates a more continental style in its constructional methods than the succeeding phase II structures. The establishment of the phase I temple appears to have been on a new site, and it is not possible to be certain as to why this location was chosen. There are some clues: the site is on an island, a factor mentioned by ancient authors as significant (Strabo IV,4,6; Pomponius Mela III,6,8; Tacitus Ann. XIV,30; Webster 1995:451); it is on the highest point of the northern part of the island and would have been clearly visible from the mainland (assuming that the surrounding areas had been cleared of woodland, as seems likely); and it is close to the presumed oppidum centred on the Chichester/Fishbourne area (Bedwin 1984:50-1). It is therefore in a privileged location relative to both the physical geography of the region and the presumed tribal locus of power at that time. The architecture of the temple is clearly designed to enhance the differentiation of the sacred site from the surrounding area (cf Webster 1995), and as such the outer enclosure formed the most significant element of the complex and may have formed the nemeton in Celtic parlance (Piggott 1978). The inner enclosure appears to have served primarily as an additional form of differentiation within the enclosure, cutting off what was probably the main ritual area from the rest of the temple. The pit seems to have been the focus. Deposition, on the other hand, was carried out in specific zones of the site, notably on the south side, i.e. the left-hand side for worshippers approaching the temple and its focus from the entrance on the east side. This zonation may perhaps be linked with allusions by Poseidonius (quoted in Athenaeus IV,152D) to Celts paying respect to the gods by turning to the right (Webster 1995:460), apparently indicating a spatially significant element to ritual practice, that perhaps also had its counterparts in everyday life (Fitzpatrick 1994). If Poseidonius is taken literally, it could be that sacrificial actions took place on the right-hand (northerly) side of the enclosure, whilst the deposition of the votive remains took place on the left-hand side. Clearly the act of deposition was important during the making of votive offerings, and for Hayling Island (but not all Iron Age temples, e.g. Gournay) the locus of these actions was mainly in a particular south-easterly zone within the enclosure. Interestingly, a south-easterly concentration of artefacts was also detected within the large 7th-5th century BC round-house at Dunston Park, Berkshire (Fitzpatrick 1994). The organisation of space on both domestic and ritual sites in central southern Britain may have had similar symbolic referants through much of the Iron Age, and as such, is worthy of detailed further investigation to elaborate on this hypothesis. The phase II temple saw a change in the physical appearance of the site, but the use of space seems to have remained the same, namely an outer enclosure where depositional activity took place, an inner circular enclosure, probably roofed, and a central pit. The significant change was the probable round-house-like building that replaced the inner enclosure. As discussed above, this resembles a domestic structure in plan, and as such, can be regarded as a 'house' for the deity. The motivation for this phase of apparent tectonisation is not entirely clear: was it a dim reflection of the Graeco-Roman practice of building temples for the gods (Brunaux 1988:32; King 1990:223), or a 'domestication' of the deity arising out of local cultural changes and preferences, or merely the desire to build a shelter for valuable votive offerings that needed to be placed near the ritual focus?

Bibliography

Adkins L. and Adkins R. 1985 Neolithic axes from Roman sites in Britain. Oxford Journal of Archaeology 4:69-75

Bailey J. and Woodward A. 1993 Rings. In A. Woodward and P. Leach 1993:135-40.

Bedwin O. 1984 Aspects of Iron Age settlement in Sussex. In B. Cunliffe and D. Miles (eds) 1984:46-51.

Blagg T. and Millett M. (eds.) 1990 The Early Roman Empire in the West. Oxford, Oxbow.

Boutet de Monvel 1863 Nouvelle étude sur les ruines celtiques et gallo-romains de la commune de Triguères. Mémoires de la Société d'Agriculture, Sciences, Belles-lettres et Arts d'Orléans 2nd ser., 7:137-88.

Bradley R. 1986 Neolithic axes in Roman Britain: an exercise in archaeological source criticism. Oxford Journal of Archaeology 5:119-20.

Brailsford J.W. 1962 Hod Hill I. London, British Museum.

Briggs D., Haselgrove C. and King C. 1993 Iron Age and Roman coins from Hayling Island temple. British Numismatic Journal 62:1-62.

Bruneaux J-L., Meniel P. and Poplin F. 1985 Gournay I, Les fouilles sur le sanctuaire et l'oppidum (1975-1984). Amiens.

Bruneaux J-L. and Rapin A. 1988 Gournay II, Boucliers et lances, depots et trophées. Amiens.

Cadoux J.L. 1982 L'ossuaire gaulois de Ribemont-sur-Ancre (Somme). Revue Archéologique de Picardie 1982, 3:12-13.

Collis J. 1973 Burials with weapons in Iron Age Britain. Germania 51:121-33.

Cunliffe B.W. 1984 Danebury, an Iron Age Hillfort in Hampshire, II, the excavations 1969-1978, the finds. London, Council for British Archaeology.

Cunliffe B.W. 1987 Hengistbury Head, Dorset, I: the prehistoric and Roman settlement, 3500 BC-AD 500. Oxford, Oxford University Committee for Archaeology Monograph 13.

Cunliffe B.W. 1991 Iron Age Communities in Britain (3rd ed.). London, Routledge.

Cunliffe B.W. and Miles D. (eds) 1984 Aspects of the Iron Age in Central Southern Britain. Oxford, Oxford University Committee for Archaeology Monograph 2.

Daire M-Y. and Villard A. 1996 Les stèles de l'Age du Fer à décors géométriques et curvilignes. Etat de la question dans l'ouest armoricain. Revue Archéologique de l'Ouest 13:123-56.

Denford G.T. 1993 Some exotic discoveries at Silkstead Sandpit, Otterbourne, and the possible site of an ancient temple. Proceedings of the Hampshire Field Club and Archaeological Society 48:27-54.

Dunning G.C., Hooley W. and Tildesley M.L. 1929 Excavations of an Early Iron Age village on Worthy Down, Winchester. Proceedings of the Hampshire Field Club and Archaeological Society 10:178-92.

Downey R., King A.C. and Soffe G. 1980 The Hayling Island temple and religious connections across the Channel. In W. Rodwell (ed.) 1980:289-304.

Ely T. 1908 Roman Hayling: a contribution to the history of Roman Britain (2nd ed.). London, Taylor and Francis.

Fitzpatrick A.P. 1994 Outside in: the structure of an Early Iron Age house at Dunston Park, Thatcham, Berkshire. In A. Fitzpatrick and E. Morris (eds) 1994:68-72.

Fitzpatrick A.P. and Morris E.L. (eds.) 1994 The Iron Age in Wessex: recent work. Salisbury: Trust for Wessex Archaeology.

Fox C. 1946 A Find of the Early Iron Age from Llyn Cerrig Bach, Anglesey. Cardiff, National Museum of Wales.

Gallup P. 1994 The survey of sarsen stones - Hampshire. Hampshire Field Club and Archaeological Society Section Newsletters 2nd ser., 22:11-13.

Goudineau C., Fauduet I. and Coulon G. (eds.) 1994 Les Sanctuaires de Tradition indigène en Gaule romaine; Actes du Colloque d'Argentomagus. Paris, Editions Errance.

Goudineau C. and Peyre C. 1993 Bibracte et les Eduens: à la découverte d'un peuple gaulois. Paris, Editions Errance.

Green M.J. (ed.) 1995 The Celtic World. London, Routledge.

Haselgrove C. 1987 Iron Age Coinage in South-East England: the archaeological context. Oxford, British Archaeological Reports British Series 174.

Hawkes C.F.C. 1977 Britain and Julius Caesar. Proceedings of the British Academy 63:125-92.

Horne P. and King A.C. 1980 Romano-Celtic temples in Continental Europe: a gazetteer of those with known plans. In W. Rodwell (ed.) 1980:369-555.

King A.C. 1990 The emergence of Romano-Celtic religion. In T. Blagg and M. Millett (eds.) 1990:220-41.

King A.C. and Soffe G. 1994a The Iron Age and Roman temple on Hayling Island. In A.P. Fitzpatrick and E.L. Morris (eds.) 1994:114-16.

King A.C. and Soffe G. 1994b Recherches récentes sur les temples romano-celtiques de Grande-Bretagne. L'exemple de Hayling Island. In C. Goudineau, I. Fauduet and G. Coulon (eds.) 1994:33-48.

Leeds E.T. 1933 Celtic Ornament in the British Isles down to AD 700. Oxford, Clarendon Press.

Lejeune M. 1988 Recueil des Inscriptions Gauloises (RIG), volume II, fasc. 1, Textes gallo-étrusques, textes gallo-latines sur pierre. Paris, Gallia Supplément 45.

Levitan B. 1993 Vertebrate remains. In A. Woodward and P. Leach 1993:257-301.

Lewis M.J.T. 1966 Temples in Roman Britain. Cambridge, Cambridge University Press.

Palk N.A. 1984 Iron Age Bridle Bits from Britain. Edinburgh, Edinburgh University Department of Archaeology.

Palk N.A. 1987 Bridle bit. In B.W. Cunliffe 1987:151-2.

Piggott S. 1978 Nemeton, temenos bothros: sanctuaries of the ancient Celts. In Accademia Nazionale dei Lincei (ed.) I Celti e la loro Cultura nell'Epoca Pre-Romana e Romana nella Britannia. Rome, Accademia Nazionale dei Lincei.

Poulain G. 1913 Les fana ou temples gallo-romaines de St-Aubin-sur-Gaillon (Eure). Bulletin de la Société Normande d'Etudes Prehistoriques 21:48-82.

Rodwell W. (ed.) 1980 Temples, Churches and Religion: recent research in Roman Britain. Oxford, British Archaeological Reports British Series 77.

Schubert F. 1994 Zur Mass- und Entwurfslehre keltischer Holzbauten im Oppidum von Manching. Germania 72/1:133-92.

Schubert F. and Schubert M. 1993 Metrological research into the foot measurement found in the Celtic oppidum of Manching. Complutum 4:227-36.

Stead I.M. 1979 The Arras Culture. York, The Yorkshire Philosophical Society

Stead I.M. 1991 Many more Iron Age shields from Britain. Antiquaries Journal 71:1-35.

Webster G. 1986 The British Celts and their Gods under Rome. London, Batsford.

Webster J. 1995 Sanctuaries and sacred places. In Green (ed.) 1995:445-64.

Wheeler R.E.M. 1943 Maiden Castle, Dorset. London, Society of Antiquaries.

Woodward A. 1992 Shrines and Sacrifice. London, Batsford.

Woodward A. and Leach P. 1993 The Uley Shrines: excavation of a ritual complex on West Hill, Uley, Gloucestershire, 1977-9. London, English Heritage.

Captions

Fig. 1 A. Le temple de l'âge du fer, premier phase; B. Le temple de l'âge du fer, deuxième phase. A. The Iron Age temple, first phase; B. The Iron Age temple, second phase.

Fig. 2 Une exemple des objets déposés au site: un ornement de char (prob. de joug) d'un type gaulois, avec des parallèles à Mont-Beuvray. An example of the objects deposited on the site: a vehicle fitting (prob. part of a yoke), of a gaulish type, with parallels from Mont Beuvray.

Fig. 3 La deposition des objets dans l'intérieur du temple: A. monnaies; B. objets de bronze; C. objets de fer; D. autres catégories. Deposition of objects within the temple: A. coins; B. bronze objects; C. iron objects; D other categories.