The aim of this paper is to situate temples as destinations for worship, and to consider the consequent implications for our understanding of ancient sacred landscapes as a form of network of religious expression. A case study area of Chichester and Hayling Island in southern Britain is used to explore these aims, with the additional factor of a political dimension (in the sense of the ancient polis) linking the temple location to the Iron Age oppidum and Roman civitas capital of the region.

In the title of this paper, ‘processional route’ has been preferred over any direct mention of pilgrimage. The reason for this is largely due to the archaeological nature of the evidence from Iron Age and Roman Britain, with little by way of written source material to draw upon. As the extensive modern literature on ancient pilgrimage makes clear, there is a wide variety in the scale, motivations and purposes in travelling to a religious destination (see Elsner and Rutherford 2005; Kiernan 2012 for syntheses of the literature). Going on pilgrimage transported, both literally and metaphorically, individuals or a community out of their normal environment (Brown 1981: 43) and thus transformed the pilgrims as a result of their experience. As Coleman and Elsner (1995: 15) comment: ‘All ancient pilgrimage not only celebrated identity, but did so by linking it with a special place’. In the archaeological and largely unscripted sacred landscape of Iron Age and Roman Britain, the ‘celebration of identity’ is difficult to investigate, but the special places can be characterized through their material culture and physical structures as well as the processional routes used to access them.

A typology of pilgrimage has been laid out by Elsner and Rutherford (2005: 9–30), and it is possible to draw on this in relation to the case study. For instance, in journeying from Chichester to Hayling Island, or from the hinterland to the island, were there official processions, equivalent to the Greek theoria (Rutherford 2000), and were they at specific times (e.g. festivals for the deities or civic commemorations)? Or were processions more private affairs, for groups of cultores to make offerings and, perhaps, to seek healing? The material culture of the Hayling Island temple and its environs may assist in answering these questions.

In terms of networks of religious expression, we may usefully begin with a perspective from a later period. A model for the sacred landscape of medieval and modern Christian worship encompasses a series of processional levels, each increasing in complexity and difficulty. At the local level are parish churches, designed to cater for daily worship within a defined physical zone: the parish (Orne 2021). Cathedrals had a wider and more hierarchical function, but nevertheless still operated largely within the boundaries of their dioceses or archdioceses. Alongside this hierarchy was another, gauged by the prestige of their destinations. These were the pilgrimage shrines, often regional in character, e.g. Walsingham (Coleman 2023) or Bardsey, but frequently international in their catchment areas, notable examples being Canterbury, Santiago de Compostela, Rome and Jerusalem. These latter have attracted the bulk of literature and discussion concerning pilgrimage, but the notion of journeying must also be considered at a more local and personal level (cf. Collar and Kristensen 2024).

Going to a local place of worship regularly, e.g. once a week on Sundays, engaged worshippers in a journey, both literally and spiritually. Special clothes might be worn (‘Sunday best’), timings of departure and arrival finessed, and perhaps summoned by a call-out (bells), mental preparations made and at certain times of year, physical preparation in the form of fasting (Lent). There may also be processions involving priests, images of saints or other sacred symbols, and followers, on special days. These could include the transport of offerings to the church (e.g. the Harvest Festival). In essence, longer journeys, usually called pilgrimages, were the same, writ large, and designed to be more arduous mentally and physically.

How does this Christian model, which will be familiar to anyone steeped in the western Christian tradition, relate to paganism, and Iron Age/Roman Britain in particular? Immediately, we can see that our archaeological knowledge is deficient, especially at the local level. A typical Iron Age or Roman rural settlement does not usually have an identifiable equivalent to a parish church, except in the case of the larger villages and roadside settlements, such as Stansted, Heybridge or Higham Ferrers (Lawrence and Smith 2009: 325–334; Black 2015). Most rural places of worship were in fact relatively isolated, and consequently had potential as destinations of pilgrimage or processions, a matter to which we will return below in relation to the Hayling Island case study. Towns, however, often had a number of small temples, Silchester being the best example in Britain. Were these neighbourhood places of worship, or should we seek other explanations for their existence? Could they have been part of a circuit of intramural foci, such as has been suggested for Pompeii (van der Graaff and Poehler 2021), and attested to in festivals such as the Feriae Martiae organized by the Salian priests in Rome? Simon Esmonde Cleary (2005) has proposed this for Britain, using the towns of Silchester, Colchester and Verulamium as his basis. It is a suggestive and convincing approach, as it links buildings housing different cults and creates a sense of civic identity and cohesiveness, which can be connected to the notion of ‘polis religion’, but in a holistic geographical framework (see Woolf 1997; Scheid 1999; Rüpke 2016; Van Andringa 2017).

At Colchester and Verulamium, Esmonde Cleary (2005) included the suburban shrines of Gosbecks and Folly Lane, respectively, into his processional routes. For both these sites, roads or similar routeways have been identified linking the shrines with their urban centres, as can also be seen at several of the Gallic suburban shrines discussed by Péchoux (2010). This leads us to extramural circuits, sometimes encompassing sanctuaries relatively close to towns that could, in Gaul, be as large as or more extensive than the temples in the urban centres themselves, such as at Evreux and Vieil-Evreux, or Trier (Péchoux 2010). Towns such as Rennes seem to have had two or more extramural temple sites, a few kilometres from the centre. Together with the temples in the town itself, these provided worshippers with a range of deities, mainly of Iron Age (Celtic) derivation, but including official and mainstream classical cults (see King 2023a for further discussion). This sacred landscape model proposes a regional circuit or processional route(s) as a primary element. As such it differs from the Christian model outlined at the beginning of this paper.1

The evidence for supra-regional pilgrimage destinations along the lines of Santiago de Compostela or Jerusalem appears to be limited, but there are some pointers to their development during the Roman period (see Kiernan 2012; Roymans and Derks 2015; Graham 2020). For instance, the massive stone temple on Puy-de-Dôme suggests an investment for significant numbers of worshippers, at a difficult mountain-top location (Paillet and Tardy 2012). Similarly, the provision of baths and a hostel at Lydney, on an equivalent scale to urban baths but in a rural location, points to receiving large numbers of visitors for worship and rituals (Wheeler and Wheeler 1932; King 2023b). Perhaps the best example in the western provinces was Cadiz, where the island temple to Melqart-Herakles had an international reputation over many centuries (Fear 2005).

It is the regional rather than the supra-regional scale that concerns us for this paper. Hayling Island temple developed in the Late Iron Age on the western periphery of the territorial oppidum of the Regini/Regnenses. As such it may have marked the limit of the tribal territory and therefore was entirely within the orbit of the oppidum. Any supra-regional links seem to have been seawards to the south and continental Gaul, a matter which will be discussed further below.

With these general issues in mind, it is possible to propose a processional route that links the urban centre of Chichester to the large temple site of Hayling Island, via Fishbourne ‘palace’, a smaller Romano-Celtic temple, two springs and the jeopardy of a tidal crossing to the final destination (Figure 1). The route proposed here may not have been the only way to reach the Hayling Island temple (a maritime route across Chichester Harbour is also possible), but it follows the main logical route that encompasses the waypoints that are of significance between the Roman town and the temple.

Hayling Island may have been a ‘sacred island’ for the people of Chichester and its territory, with origins in the Late Iron Age and its dynastic politics. It is notable for its relative lack of Iron Age and Roman settlement compared with the adjacent mainland (Bloomfield 2024: figures 14 and 15). As such, it may have been an island set apart from normal settlement activity due to the presence of the temple and its precinct.

A preliminary point to make is that this is not an exercise in tracing a Roman road, although there was of course one linking Chichester to Havant and then westwards to Winchester via Wickham with a branch to Bitterne (Margary 1973: RR420, RR421; Soffe and Johnston 1974). The archaeological material from Chichester, Funtington and Hayling Island strongly suggests Late Iron Age origins for this route, and therefore there was more likely to have been a generalized alignment rather than one that was as precise as a Roman via (Garland 2016: 86–87; see also Bekker-Nielsen 2009; Engelbogen 2023).

To start at Chichester, the internal civic ‘circuit’ of cults, in the Esmonde Cleary sense, encompassed the famous central temple to Neptune and Minerva, only known through an inscription (RIB I.92; Russell 2006: 211–212; Henig forthcoming), Jupiter Optimus Maximus, the matres domesticae and a Genius (Down 1988: 66–71; Russell 2006: 207–219). On the face of it, all these seem Roman or at the least shaped by Roman influence, to the extent that local names have been lost. More local cults seem to have existed out of town, with their temples part of an extramural circuit. This could include two downland Romano-Celtic temple sites, Bow Hill to the north-north-west (Bedwin 1980: 191–192; Rudling 2008: 110) and possibly Red Copse to the north-east (Mark Roberts, pers. comm.). Selsey (to the south), with its substantial Late Iron Age coin evidence and other finds, may also be included (PAS; Heron-Allen 1911; Bean 2000: 269–271).

The journey to Hayling Island was more substantial than these local circuits or processions. Its putative course went from the west gate of the Roman town to Fishbourne, thence to the Romano-Celtic temple at Ratham Mill, Funtington, then to the spring site at Havant, turning south at that point to cross the Wadeway to Hayling Island. This was a total of c. 18.5 km (12.2 Roman miles), so was likely to have required an overnight stay on the island before returning, given the assumption of 20 Roman miles per day as an on-foot travelling rule of thumb (Kiernan 2012: 87–88; Graham 2020: 17). The reasons for it being a sacred destination are explored in the next section (below), but for the moment we need to explore the route itself and the waypoints along it.

The first of these is Fishbourne (Figure 2). It was an important site from before the Roman conquest (Manley and Rudkin 2005), and may well have been an original focus for the territorial oppidum prior to the development of Chichester. It could, therefore, have been the Late Iron Age starting point for any processional route to Hayling Island. Following military use at the time of the conquest, Fishbourne apparently developed as a regal centre linked to Togidubnus, hence its popular designation as a ‘palace’, and by the Flavian period was a large Mediterranean-style Roman villa with some significant additions (Cunliffe 1998). One of these additional features is the large basilica or aisled hall at the north-east corner (Cunliffe 1971: 106–110; 1998: figure 52; 2013). It appears to have had an outward-facing use as an assembly hall (Cunliffe 1998: 83–84; Manley 2003: 129–130; Russell 2006: 114, 119), with an external entrance onto a gravelled road; this in turn led (a) to the main porch and thence to Chichester’s west gate, and (b) in the other direction round the Fishbourne main building to the north and then the west. In other words, it may have been a waypoint on the route, which had to navigate round the apparent obstacle of Fishbourne in order to get further west. The function of the aisled hall in relation to a putative processional route may have been to allow cultores to be received by Togidubnus or other members of the dynasty, to give refreshment, or even to mark the starting point of the event or festival associated with the procession. The high status of Fishbourne and the possibility of an official function for the aisled hall can be used to suggest that at least some of the processions towards the Hayling Island temple were akin to the official theoria of Greek religious practice (Rutherford 2000; Elsner and Rutherford 2005: 12–14).

An alternative to this route via Fishbourne is one further north, running alongside one of the Chichester Late Iron Age entrenchments (Bradley 1971: 24, entrenchment E-W C and the Hook Dyke; Allen et al. 2018: figure 1; Garland 2020: 117–118), and along the boundary of the coastal brickearth and the overlying gravel deposits. Such a route was suggested long ago by the late Hugh Toller (1981: 216; Down 1988: 50–51), in part because it was a plausible route supported by aerial photographic evidence, and also because it avoided the ‘palace’ altogether. To Toller’s reasoning can be added that it ran fairly directly in a straight alignment to the next waypoint, Ratham Mill temple.

A small Romano-Celtic temple is known at Ratham Mill, Funtington, from aerial photography and geophysics (King and Soffe 1983; Allison 2006; Mountstephen 2007; Rudling 2008: 109–110). It has a number of Late Iron Age coin finds, including of Tincomarus and the Commian dynasty (PAS), and there are a number of coin types attributed to this site in Ancient British Coins (Cottam et al. 2010: ABC 557, 569, 578, 596, 611, 614, 1070). It lies at a crossing point of a small stream, and also at the point where the route may have changed course to the south-west to run c. 1 km to meet the known alignment of the Roman road running east-west from Chichester to Winchester. The temple site was almost certainly in existence from the first century BC, and therefore predates any Roman road system in the area. It may originally have been on an Iron Age routeway along the south coast. The parish name, Funtington, is also relevant, as it is derived from Old English *funta and late Latin fontana, meaning ‘spring’ (Hawkins 2015; 2020: 56–57). It is an unusual place-name element confined to south-eastern England, and two of them lie on the route described here. The other is Havant with the Old English Hama + *funta, meaning ‘Hama’s spring’ (Hawkins 2015; 2020: 57; Oscroft 2015: 308), where the spring was likely to have been a key waypoint along the route, together with the others in the immediate vicinity (see Ghey 2005 for a discussion of temple landscapes, springs and water).

Whatever route the pre-Roman link from Chichester to Havant/Hayling Island took, by the early Roman period we are on surer ground in an alignment that was followed by the old A27 from Nutbourne to Havant, where there was a crossroads, a vicus and a perennial spring, known as the Homewell (Pile 1986). There were as many as 20 springs in the vicinity of the vicus, and the Homewell’s water was regarded in historic times as good, pure and potentially medicinal, and had a constant temperature of 10.5 degrees Celsius. Evidence for its use in the Iron Age and Roman periods is unfortunately circumstantial rather than direct, but there are Roman walls and occupation debris in the immediate vicinity. This would very likely have been an important stop on the way to and from Hayling Island, where cultores could have sought refreshment and perhaps cleansed themselves prior to (or after) crossing the salt-water Wadeway to Hayling Island. It should be noted that the temple site on Hayling has not yielded evidence for a bathhouse, so Havant may have been the nearest locations for bathing and ablutions, if needed (see King 2023b).

At Havant, the route turned south at the crossroads and led directly to Langstone, where there was a traditional crossing point known as the Wadeway to Hayling Island (Morley 1988; Bloomfield 2024: figure 34). Excavations have shown that this tidally dependent crossing dates back to the medieval period, and quite possibly earlier (Satchell 2014). It still exists, and can be easily seen on satellite photographs, but was breached in 1820 by the New Cut for the Chichester Canal, prior to the construction of a road bridge, so it is now unusable (Green 2005). Bronze Age wooden stakes have been recovered from the south end of the Wadeway (Williams and Soffe 1987), which may suggest an early date, but the question can reasonably be asked as to whether Hayling was actually an island in the Iron Age and Roman periods (Bloomfield 2024). It lies in a drowned landscape, and there is good evidence of geological sinking and progressive inundation from the end of the last Ice Age onwards (Allen and Gardiner 2000; King and Soffe 2013). By the Iron Age, when the first temple was built, Hayling was very likely to have been an island, perhaps with extensive salt marsh and mud flats separating it from the mainland. Allen and Gardiner (2000: 215, figure 68) map Roman Hayling with a physical link to the mainland in their environmental survey of Langstone Harbour. This is by no means certain, however, and at the very least, there is likely to have been an extensive salt marsh at the north end of the island, with the consequence that a crossing could well have required guidance along a causeway. It was a locus of jeopardy in the journey from Chichester to Hayling Island.

The exploitation of salt was an important activity on Hayling Island at this time (Bradley 1975; Allen and Gardiner 2000: 214–215), and may have been one of the reasons why a temple site developed here.2 Salt working would have been a seasonal activity, during the summer and autumn, and pilgrimages to the temple site may have been organized to coincide with this, in part to celebrate the winning of a vital commodity for the community. There were other, more political reasons for the Hayling Island temple, too, connected with the Iron Age tribe of the Atrebates and its royal dynasty (see below).

The temple itself was constructed on a slight but obvious rise in the landscape, at six metres above sea level. The first Iron Age temple was a new foundation (Figure 3), and we do not know why it was positioned where it was. Possibilities are: the insular location, mentioned as significant by Greek and Latin sources (Pomponius Mela, Chorographia 3.6.8; Strabo, Geographica 4.4.6; Tacitus, Annales 14.30; Webster 1995: 451); its elevation within a generally flat coastal area; and thirdly its relationship to the extent of the territorial oppidum centred on Chichester and Fishbourne. In the Roman period, the temple had a tower-like appearance at least c. 10 m high, and thus visible for many Roman miles (Garland 2013: 185–186; 2016: figure 4).3 The island may also have marked the western limit of the Chichester territorial oppidum and later Roman civitas (see below). These factors gave the location a prominence in both physical and contemporary socio-political terms (Figure 4).

In terms of its sequence (Table 1), there are two Iron Age and an early Roman phase that are of relevance to this paper (Figures 4 and 5). Although there are traces of earlier Iron Age predecessors to many ‘Romano-Celtic’ temples in north-west Europe, this is one of the best examples from Britain. In addition, the second Iron Age phase continued beyond the AD 43 conquest. Its stone replacement in the AD 60/70s overlaid it exactly in terms of architectural layout. The Roman stone temple was one of the earliest temple buildings in the new province. As discussed further below, it had associations with the royal house of the Regini/Regnenses, and with Fishbourne ‘palace’ (Creighton 2000: 192; King and Soffe 2013).

The first Late Iron Age temple was a wooden construction of the early to mid-first century BC. It had three main elements: an enclosure (c. 25 × 25 m) with an eastern side that had an alignment exactly north-south, and an eastern entrance; within was another enclosed area, also opening to the east, and the third element was a pit on its west side (Figure 5a: Phase 2a). It was probably an open-air sanctuary. The outer enclosure was made with upright posts, and planking or wattles in between. A similar appearance also applied to the inner enclosure, bounded by well-preserved traces of upright planks and large square posts.4 The votive pit, c. 2.5 × 1.7 m, and 0.65 m deep, was respected by the fence line and therefore originated in this phase.

The second temple transformed the appearance of the shrine, and made it look much more like a building than an open-air place of worship. It used the earlier outer enclosure, but the rest of the interior, except for the pit, was replaced by a circular building that retained the pit in its centre. The building was 9.2 m in diameter, and had a concentric ‘drip-trench’ around the structural trench for its foundations. Architecturally, it was a typical Iron Age roundhouse, but at Hayling Island it was a house for the deity (see Kiernan 2020, ch. 4), along lines familiar to people living in its domestic equivalent. The closest parallel for the site at this period is the enclosure with roundhouse-like structures at Thetford (Norfolk) (Gregory 1992).

The pit was a significant feature in the sequence of development at the site, and it was not a feature that would be found in a domestic roundhouse. It yielded radiocarbon dates of calAD 20–330 (King and Soffe 2013), a wide date range, but compatible with the second temple continuing in use until the Roman temple was built, followed by the covering of the pit with a paved floor in the AD 60s/70s.

The many votive deposits across the temple site and beyond also marked a distinct difference between Hayling Island and the average Late Iron Age settlement. Deposition was mainly in the outer courtyard, especially in a zone towards the south-east of its area. This was visible in most object categories (King and Soffe 1998: 42–44), and probably originated in the rituals of offering and deposition that were part of the cult practices of the temple. It can be linked to the right-hand directionality alluded to by Poseidonius (Athenaeus, Deipnosophistae 4.152D; Fitzpatrick 1994; 1997; Webster 1995: 460). Using this as a guide, sacrificial actions may have taken place in the northern (i.e. right-hand) part of the enclosure, followed by deposition to the south-east. Other temples in Britain, such as Bath and Wanborough (Surrey), display similar arrangements (King 2023c: 181), as does the Martberg in Gallia Belgica (Wigg-Wolf 2005).

The temple architecture and its enclosed spaces created an apparent hierarchy of ritual activity within the site. The outer enclosure may have marked the nemeton or sacred area (Piggott 1978; Webster 1995; see also Loth 1924 for etymology of the word nemeton), while the inner enclosure was a secondary delimitation, cutting off the central area with the pit as its focus.

Beyond the nemeton, many artefacts have been found in the surrounding area, up to at least 500 m distant from the temple itself. Many hundreds, if not thousands, of coins, brooches and other objects must have been deposited within a broader sacred area encompassing much of the north part of Hayling Island. This is the location where cultores appear to have performed ritual activities, celebrated festivals, and perhaps stayed the night before returning to Chichester or other settlements.

Analysis of the Iron Age coins has been undertaken for the temple site itself (Briggs et al. 1992; Haselgrove 2005), and is ongoing for the surrounding area.5 Of the c. 350 Iron Age coins analysed up to 2022, some 18% are of early British Iron Age issues, c. 39% from later issues, c. 18% are from Gaul, and c. 24% are Roman republican and imperial issues up to the reign of Claudius (Table 2). When analysed in more detail, there is an interesting difference between the temple and its surrounding area, in that a significantly higher percentage of early British issues was found in the surrounding area. In contrast, a significantly lower percentage of early Roman coins came from the surrounding area. This implies that a wide area of deposition existed from early in the temple’s existence,6 and that by the later Iron Age and conquest period, coin deposition had tended to concentrate at the temple itself. This did not, however, mean that the surrounding area ceased to be used for deposition, as many Roman coins up to the late fourth century AD have been found in the metal-detecting survey.

From both the temple site and its surrounding area, the Iron Age coin assemblage included a significant representation of continental Iron Age coins, mainly from northern, western and central Gaul (Briggs et al. 1992; Haselgrove 2005: 393–395; Garland 2016: 86). This reflects the cross-Channel links of the southern peoples of Iron Age Britain, and points to a potential supra-regional pilgrimage connection for the temple site.

The distribution of Iron Age gold coins can be seen in Figure 6. It reflects the fields available for the survey, in part, and it is very likely that the spread of coinage was up to 500 m or more in each direction from the focal point of the temple itself. In addition to coins, the survey has yielded many fibulae, bracelets, fittings and studs. Amongst these are pieces of broken-up Late Bronze Age/Early Iron Age gold torcs, several gold pellets and shaped flat billets, some of which look as though they were formed in a standard Iron Age pellet mould (Daphne Nash Briggs, pers. comm.). Although these objects are essentially undated, they point to some sort of gold working on site. This was possibly linked to testing or money-changing, and also to the detection and control of ‘coins’ with copper alloy cores within a gold outer casing. These forgeries of official coins are not uncommon on temple sites, and suggest that offerings of this sort were acceptable within the ritual framework of the cult.

The coins and other objects discussed above form a major aspect of deliberate deposition at the site. From the temple excavations, other classes of material, mainly pottery and animal bone, can add to our understanding of what the cultores were doing at the site and when the activity occurred. Most of the large pottery assemblage (from the Roman phase of activity) consisted of bowls and jars, mainly in the local coarse ware made at Rowlands Castle. This production centre lay to the north of Havant, and it is likely that the Havant waypoint on the route to the temple was a place where pottery for offerings could be obtained. Once an offering was made, the pottery vessel was discarded at the temple site, to become part of the very fragmented assemblage accumulated within the nemeton. The animal bones present a different picture, since a case can be made to suggest that they were already present on the island, as sacred flocks of sheep and herds of pigs (King 2023d). They may well have grazed or foraged within the outer surrounding area in the north part of Hayling Island.7 The age at death for the sheep and pigs in the bone assemblage suggests seasonality of slaughter/sacrifice, probably in the late summer or early autumn (King 2005: 338–339). This gives us clues for the likely timing of any processions to the temple as part of communal activity.8

The second Iron Age phase changed the temple layout physically, but retained the ritual arrangements from the first phase. Creation of a roofed structure in this phase may have indirectly reflected Graeco-Roman tectonizing influences, or was a locally-driven ‘domestication’ of the cult, and provision of shelter for increasingly valuable offerings (King 2007; Fauduet 2010: 144–151).

The Roman temple had a circular limestone cella with a pronaos or porch on its east front. It was plastered and painted red externally, multicoloured inside and roofed with tiles. In the early second century the porch was enlarged. The circular cella was built surrounding the second Iron Age temple’s circular structure, implying clear continuity from the Iron Age phase to its Roman stone successor. The temple was set within a square temenos or sacred courtyard, which mirrored the layout of the Iron Age outer enclosure, on a larger scale and with a portico surrounding the courtyard (see Aldhouse-Green 2018: 77; Pliny, Epistulae 9.39 for the significance of porticoes at temples). Across the temenos, a gravel path linked the cella to the entrance hall, outer porch and other rooms along the east front of the temenos. There were adjustments made to enlarge both the cella porch and the temenos porch in the early second century, and a probable outer wall or ditch lying further out to the north and east (Downey et al. 1977).

The votive material deposited at the Roman temple contrasts somewhat with that of its Iron Age predecessor. There is an absence of military, equine and vehicle equipment, perhaps a result of the rapid decline of Iron Age ‘warrior culture’ after the Roman conquest. Civilians were also forbidden from carrying arms, once the new province had been taken into Roman control. Pottery, glass objects, bone pins and animal bones continued to be offered at the temple, together with metal objects, mainly coins and brooches, including enamelled horse and rider and hippocampus brooches.

The Roman temple lasted about 200 years before clear signs of decline became apparent. By the mid-third century AD, masonry had begun to collapse into the courtyard, and the tall cella was probably unstable due to lack of maintenance. This coincides with similar signs of abandonment at Fishbourne ‘palace’, where there was a fire, but no subsequent rebuilding. The south coast was becoming unsafe at this time, due to Frankish and Saxon raiding during the third century. Eventually the fort at Portchester was constructed to counter this threat, so that some signs of security in southern Britain had returned by the early fourth century. The Roman temple was not rebuilt, however, and Hayling Island as a whole has very little evidence of late Roman occupation. Interestingly, many third and fourth century coins have been found in the rubble of the collapse of the temple, and they suggest that worshippers may still have been visiting the temple’s site and leaving votive offerings.

To set the scene, a preliminary political history of the region in the Late Iron Age is needed.9 At some point before 60 BC, there was an immigrant Belgic colony, perhaps based at Selsey (cf. Bean 2000: 269–271; Sills 2017: 100–256). They struck a gold coinage (quarter staters British Ad2 and Insular Cf2; Daphne Nash Briggs, pers. comm.), which was the earliest in Britain, and the full sequence of which is well represented amongst the Hayling Island coin finds. These Belgae may have been of multiple origins, hence their generalized name by the time of the later Roman conquest in AD 43. There might well have been some Atrebates amongst them, but the coinages do not confidently support a coherent Atrebatic presence yet in Hampshire or Sussex.

Soon after this, Caesar’s incursions into Britain in 55 and 54 BC brought Commios/Commius the Atrebatan to Britain as an ambassador. His role in the south is unknown, but he helped Caesar defeat Cassivellaunus, after which the tribute that Caesar imposed on the island was paid in gold staters for three to four years (54–50 BC). For southern Iron Age communities this is represented by British B coinage (Daphne Nash Briggs, pers. comm.). Commios, as is well known in the ancient sources, had gone back to Gaul in 54 BC but fell out spectacularly with the Roman commanders after an attempted assassination. After a second attempt on his life Mark Antony let him retreat to Britain in 50 BC with a vow never to associate with Romans again (Caesar, Bellum Gallicum 8.48; Frontinus, Strategemata 2.13.11). He was one of the most experienced and able of the large number of Belgae and some Gauls who took refuge in Britain in the aftermath of the conquest of Gaul. As a result, the 40s BC were chaotic, but in the end they settled north and south of the Thames and in Kent, striking prolific coinages to meet ongoing needs. Chichester and Selsey seem to have been a major gateway (Hawkes 1977: 184, map 12; Trott and Tomalin 2003: 176), and in the local region, at least three principal territorial districts emerge: Regini in West Sussex and south Hampshire; Atrebates in central and northern Hampshire, later leaving its name at Calleva, and initially in cahoots with Regini but with an identity of their own; Belgae between the rivers Test and Meon (leaving their name at Winchester), with ties to Regini but also with an identity of their own until c. 40 BC (Cunliffe 1991: 108–110; King 2020); and an East Wiltshire group known only from coins, with a distinctive elite of its own, to be included as an unnamed fourth district of the same major regional grouping.

By c. 30 BC or slightly later, Commios seems to have emerged as the paramount leader out of the chaotic 40s, and (nearly) all of the other warlords stopped coining. He held sway over an area that included west Sussex, east and north Hampshire, and was founder of what became an effective dynasty of an initially more unified, later shrinking, territory with its heartland not amongst the Atrebates of Calleva, but amongst the Regini of Chichester. The western boundary (with the Belgae) may have extended to the valley of the river Meon, but remains ragged and ill defined. The coin finds at Hayling Island suggest that the first Iron Age temple was a focal point for Commian activity as warlord and king (cf. Brunaux 2004).

In this context, the first Iron Age temple at Hayling Island was created during the influx of Gaulish culture at this time, which seems to have included ideas concerning cult architecture and votive deposition (cf. Bataille 2011). Indeed, this phase shows more Gaulish elements than the second Iron Age temple (cf. King 2007). The military equipment, vehicle remains and related artefacts have parallels in terms of depositional practice in Late Iron Age Gaulish temples, both in western Gaul and Gournay-sur-Aronde (Oise) (King and Soffe 2013). This can be linked to a Roman inscription to Mars Mullo at Allonnes (Sarthe) (Brouquier-Reddé and Gruel 2004), to suggest that the Hayling Island Iron Age temple had a dedication to a Celtic Mars-type deity.

Another significant aspect of the temple site is the link to elite groups. Elsewhere, ancestor cults can be seen developing on the sites of high-status burials, notably at Folly Lane, St Albans (Niblett 1999; 2001). John Creighton (2000: 192–197) has discussed this in relation to Hayling Island, and in his scenario, the first temple is linked to Commios, perhaps his ancestral cult or his mausoleum/cenotaph.10 The second Iron Age temple can be linked to Verica, in a revival of the earlier ancestral, now dynastic cult. Finally, the Roman rebuilding was a reinforcement of these cultic ideas by Togidubnus, ‘Great King of the Britons’ or ‘Great King in Britain’. As such, it was not only a sacred sequence, but a landscape of commemoration as well (cf. Roymans et al. 2009; van der Schriek 2019). This interpretation is significant for giving a political and cultural context to the activity at Hayling Island temple. It places the cult as one of protection for the people, via elite warrior culture and a hero/ancestor focus, linked additionally to Celtic Mars (Thevenot 1968: 53–56; Green 1986: 103–110). It would also give the temple a territorial resonance, reflected in its use as a probable pilgrimage destination and locus for deposition of many objects within the wider sacred zone around the shrine. The Mars-type deity may also be of importance to worshippers in relation to Late Iron Age military service and oath taking.

Moving to the Roman phase of Hayling Island, why was such a large and (exceptionally for Britain) early Roman temple built so soon after the conquest, and in an island location that was not readily accessible? The answer seems to lie in the political alignment of the area at the time of the Roman conquest of AD 43. In the late first century BC the Commian dynasty had struck up cordial relations with Rome and when, in c. AD 25, the oppidum at Calleva (modern Silchester) fell into the (politically anti-Roman) hands of the sons of Cunobelin, King Verica retreated to his capital in the Chichester area. Eventually Verica was forced to flee to Rome to appeal to the emperor for help, and in the wake of the invasion the southern Atrebates were allied to Rome (Creighton 2000: 218–221).

This alliance resulted in the military supply bases at Chichester and Fishbourne serving the Roman invasion and liberation force, particularly Vespasian and the Legio II Augusta, in one of their main landfalls and their advance westward into Durotrigan (enemy) territory in AD 43–44 (Manley 2002). Soon after this in the Neronian and early Flavian period the Iron Age temple at Hayling Island was rebuilt in stone on a monumental scale, coinciding with the building of the proto-palace and later Flavian palace at Fishbourne, as well as other structures in the Chichester area. These events can be ascribed to the client king Tiberius Claudius Togidubnus (Cogidubnus), successor to Verica, and it cannot be happenstance that Gallic architects and builders seem to have been working on these enterprises, perhaps represented by the collegium fabrorum (guild of craftsmen) mentioned on Togidubnus’s dedicatory inscription from the Chichester temple of Neptune and Minerva (RIB I.92). Obviously, the Roman emperors were keen to concentrate expertise and expenditure within the area to reward Togidubnus for his help during the invasion, to promote Roman values, and perhaps also to perpetuate and enhance the Iron Age temple at Hayling Island, associated with the oppidum.

The island location of the temple is also of significance, both as a spiritually charged location in its own right and as a territorial marker. Plutarch (Moralia 419E) encapsulates the view of ancient authors:

Strabo (Geog. 4.4.6) also refers to an island near the mouth of the river Loire, where the women priests reroof their temple each year, by inference with thatch or reeds. This was a temple in the Ocean, at the apparent edge of the world, and Richard Bradley (2000: 25–26), in his discussion of Pausanias, alludes to this as a significant aspect of pilgrimage and processions. Hayling Island, of course, was not in the Ocean as such, but served as a stepping stone for crossing to and from the Gallic side of the English Channel. In this respect, the island location was not so much for spiritual isolation as for propitiating the deities of the crossing. Several Romano-Celtic temples are located either on islands, promontories or with a sea view (Woodward 1992: 17; Tomalin 2022: 306–314; King 2023c: 169, table 9.2), each probably with its own particular sense of protection and ritual practice.

If Hayling Island had a role as a seaward-facing liminal place, it also was one of a small group of islands in the Chichester/Langstone/Portsmouth harbour area. Collectively, they appear to have marked the western limit of Reginian/Regnensian territory, with the Belgae to the west and the Vectuarii to the south on the Isle of Wight. In addition, the processional link with Chichester may have acted as a seasonal reinforcement of this territorial marker, as a form of ‘beating the bounds’.11 Other temples, such as Harlow (Rippon 2016), appear to have served a similar liminal function by their positions on river valleys between tribal land units. An apparent confirmation of Hayling Island’s boundary function comes from the distribution of the local coarse Roman pottery, Rowlands Castle ware, which is found in large quantities in the temple’s pottery assemblage. Work on its distribution by Ian Hodder (1974) demonstrated a significant drop-off in its presence west of a line running roughly northwards from the Hayling and Portsea Island area, which appears to coincide with the putative boundary between the Roman civitates of the Regnenses and the Belgae. The temple assemblage, therefore, was at the south-western margin of this distribution area, and indicates its economic orientation towards the east, the coastal plain and Chichester.

The island significance of Hayling Island can be supplemented by the possible ancestor cult associations of the temple, as discussed earlier, in that the first Iron Age phase might be linked with Commios and the second phase with Verica who revived the site as an ancestral shrine and cult of Commios to legitimate his reign. The succeeding early Roman temple was a reinforcement of this cult by Togidubnus (Creighton 2000: 191–197; Haselgrove 2005: 399–400), to which official processions from Fishbourne and Chichester may have played an important role. Perhaps a Mars-type Celtic deity was the tribal god of the dynasty, with attributes of tribal protection, warriors and death. Mars may also have been a suitable deity for expressing loyalty to the Roman state, as one of the major gods in the Roman pantheon.12 Hayling Island Roman temple may have been a grand rebuilding of the second Iron Age temple in stone, to demonstrate an affirmation by the royal line of the Regini of their gratitude for Roman aid and intervention, while at the same time preserving a link back to their Iron Age ancestry at the traditional dynastic temple. It is probably significant that both Fishbourne and Hayling Island temple shared the same building techniques, and may have been constructed by the same (probably Gallic) building team. This highlights the potential high-status links that the temple had, which appear to have been upheld by the possibility of official processions from Fishbourne and Chichester to the temple.

The archaeological evidence at Hayling Island temple suggests that the processional route posited above was in existence from the Late Iron Age. The continuity of depositional practice at the site from this period into early Roman times indicates that its ritual practices, once established, were relatively unchanged by the events of the Roman conquest. As such, Hayling Island temple, its rituals, and the wider regional sacred landscape can be seen as a bridge across the Iron Age-Roman transition in Britain.

By way of summing up, it is evident that the typology of pilgrimage put forward by Elsner and Rutherford (2005), and outlined in the introduction to this paper, does have relevance to the suggested processional route from Chichester to Hayling Island. In particular, the Fishbourne element of the route can give an official and dynastic gloss to the processional activity, similar to the Greek theoria. Other large suburban temples in Britain and Gaul, such as Folly Lane (Verulamium), Gosbecks (Colchester), Vieil-Evreux (Evreux), Allonnes (Le Mans) or Le Haut-Bécherel (Corseul), almost certainly had similar functions as processional destinations (Esmonde Cleary 2005; Péchoux 2010: 262, 276, 303). What marks out Hayling Island as different from these examples is the greater distance between the temple and the oppidum/civitas central place. This is probably accounted for by the specific political and historical circumstances of the region (discussed above), and the extensive area of Chichester’s territorial oppidum. A related aspect is the continuity between the Iron Age and Roman temples, mainly in terms of the associated depositional practices. The processional route and the activity of the cultores had important roles to play in this, as they acted as a chronological bridge spanning the Roman conquest period. Despite the presence of Roman troops at Fishbourne and elsewhere in the region, the political and religious structures of the local people appear to have remained intact. Indeed, the high status of Hayling Island temple and its links with the ruling elite favoured it as a locus for display of the continuity of the lives of the local people during a period of profound change, and a means of negotiating that change through religious practice and pilgrimage.

This paper has focused on Chichester’s links to Hayling Island, giving a new perspective on polis or civitas religion as it applies to the territories of the Iron Age and Roman peoples of southern Britain. It must not be forgotten, however, that the rural population also engaged with Hayling Island temple, and that it was part of a sacred network of locations across the landscape, that included natural places such as springs, hilltops and islands.