Beyond the temple: blurring the boundaries of ‘sacred space’

Eleanor Ghey

Introduction
Landscape archaeological approaches have a contribution to make to the study of temple sites,
demanding a shift of focus from the temple itself to the wider surrounding area. However, such
approaches also have relevance to our understanding of the architecture of temple sites, here
considered using Gallo-Roman examples. In particular, a breaking down of the opposition of
nature to culture and the problematic definition of ritual in the archaeological record have
implications for the concept of ‘sacred space’. The idea of discrete sacred spaces may conflict
with interpretation of the landscape as a lived environment. It will be argued that the location
of temples is highly significant for their interpretation and that their architecture is rendered
meaningful only when situated in its landscape context. I wish to emphasise the problematic
nature of this ‘landscape context’ and the cultural and individual specificity of its interpretation
and experience, now and in the past.

Current approaches to landscape
Until the appearance of more critical approaches to landscape in archaeology, studies of
landscape had tended to be descriptive, with interpretation being undertaken primarily as a
means of setting the scene for monuments and human activity. The relationship between
humans and their environment was seen as one of mastery, in which humans gained control
over the environment (a theme found in discussion of later prehistoric agricultural
intensification), or as one of determinism, with behaviour influenced and restricted by the
availability of natural resources (for example processual theories of culture change in Andean
civilisations). In both approaches, there remained a conceptual divide between nature and
culture. Recent developments in human geography and philosophical approaches to the study
of space have introduced a reflexive understanding of the relationship between people and
land. These developments stemmed from rejection of the Cartesian duality of mind and body,
leading to models of embodied experience of the landscape and the indivisibility of culture and
nature. ‘Nature’ may be seen as a culturally constructed category of varying meaning (Daniels
and Cosgrove 1988: 1). Interpretation of the meaning of landscape has moved from the analogy
of landscape as text to a more reflexive understanding of the relationship between people and
land. Landscape is created by the repeated actions of those who work within it and move
through it. However, it also has an active role in the reproduction of society and culture. In the
words of Ingold, ‘through living in it, the landscape becomes a part of us, just as we are a part
of it’ (Ingold 2000: 191).

This concept of landscape as understood through inhabitation rather than as a setting in
which action occurs, has had an impact on archaeological narratives (particularly in prehistory)
which have sought to describe the landscape as experienced. The emphasis here has moved
away from the ‘site’ to a wider spatial context for human action, as monuments are understood
by their location and interrelationship. However, these earlier post-structural narratives can be
criticised for their under-theorisation of the body. The experience of landscape is necessarily
subjective and cannot be accurately represented by a ‘disembodied’ body. There has also been
an emphasis in these accounts on power and the agency of elites, which does not always allow
for other experiences of landscapes (Brück 1998: 32; Thomas 1993: 29).

Although these ideas are frequently found in prehistoric archaeology, archaeologists
working with the Roman period have less readily taken up the challenges that they present. In
the study of temple sites, such challenges are to situate sites within their wider context in such
a way that does not create an artificial distinction between the site and its surrounding area, and
to attempt to convey the multiplicity of meanings that may be invested in a temple site. The
subjectivity of landscape may be an opportunity to investigate culturally and individually
specific interpretation and experience.

Approaching temple sites and architecture
I would suggest that there has been a past bias towards a certain understanding of ‘ritual
architecture’ that has been based on classical and modern concepts of ritual structures. The
historical accounts of pre-Roman shrines by classical authors tended to emphasise the use of
natural places such as groves and islands. It may be that the lack of a familiar ‘architecture’
meant that indigenous shrines went unnoticed, and these accounts perhaps influenced early
ideas of a lack of ritual structures in the Iron Age (Lewis’ ‘aniconic and atectonic’ religion
(Lewis 1966: 4)), one that still persists in more recent work. This understanding of architecture
rests on the need to have a conceptual division between sacred and profane represented in
material form (Venclova 1993: 60). This has also had an effect on our understanding of the
Romano-Celtic temple, emphasising the typology of the form. The evidence suggests that in
reality there was less rigidity in what was perceived as an appropriate place for ritual action.
For example the material culture associated with the spring site at La Douix in Châtillon-sur-
Seine, Côte-d’Or (Coudrot 2002), is very similar to that of the nearby temple sites of Essarois,
Côte-d’Or (Daviet 1967) and Le Tremblois, Côte-d’Or (Paris 1960). The site at Chamalières,
Puy-de-Dôme (Romeuf and Dumontet 2000) is no less a ‘shrine’ although there does not
appear to have been an architectural element to the site. It appears that the place itself (or
perhaps the act) was more important than the associated architecture.

A tendency to overlook natural features perpetuates a cultural evolutionary approach to the
study of religion, with the Roman ‘achievement’ of temples seen as having progressed from a
culture without permanent ritual structures. Other similarities and continuities between sites
may be overlooked. The apparent relative uniformity of the built form of the Romano-Celtic
temple also masks the fact that different landscapes will create different meanings for places.
Excavation has tended to reinforce a limited perception of architecture by focusing on the
building itself. I would suggest that a broader definition of architecture might be helpful when
looking at these sites. Architecture can be seen as constructed space, the division and creation
of space and manipulation of landscape. In a broader understanding of the architecture of
temple sites, the zone of the ‘building’ would include the surrounding environment,
watercourses, ephemeral structures, pathways, boundaries and open spaces. Apparently natural
elements such as water may be ‘constructed’ in some way; the diversion of watercourses
creates other foci and influences the passage of people through space. These things are integral
to an understanding of how this architecture functioned in providing a setting for action.
The relationship between landscape and ritual is an active and reflexive one; ‘landscape and
material culture do not provide merely a stage setting for human action but create a set of
locales integral to that action’ (Gosden and Lock 1998: 4). Shrines in the Roman landscape
were appropriate places for communication with the gods and brought their presence to mind.
These appropriate places may have been chosen because of their pre-existing associations, or
the suitability of their location to become part of regular ritual practice through the presence of
roads or route-ways.

The whole Roman period landscape could in fact be seen as a ritual landscape, and the
desire to distinguish specific ‘sacred spaces’ is arguably not universal. The layout of a Roman
town had religious signification, which extended by implication throughout the surrounding
area. In the rite of inauguratio, described by Livy (I.18), great importance was placed on the
priest’s lines of sight, which encompassed the landscape as far as he could see and allowed
landmarks and directions to be identified and fixed. The concept of the templum related to the
diagram drawn by the priest during this ceremony, inscribing these lines of sight into the soil
(Rykwert 1976: 45). It was therefore more of a conceptual space than a physical one, although
it was given spatial form in the layout of military camps and cities. The cardinal directions
determined the position of the roads that extended beyond the town and connected it to others,
as well as the division of the land itself. Ritual served not only to establish the town but also to
reinforce its boundaries through festivals and processions (ibid.: 27).

Ritual in daily life
The identification of sacred space touches on the same theoretical issues as the recognition of
ritual practice in the archaeological record. The issue of whether it is possible to distinguish
‘ritual’ from the ‘mundane’ in the archaeological record has been raised in work on the Iron
Age (for example Hill 1995), with the suggestion that such a distinction might be inappropriate
for the period. As Brück argues (1999: 317), one of the defining characteristics of ritual in
archaeological and anthropological theory has been its non-functional nature, leading to
difficulties of recognition when there is no clear distinction in the material record. Ritual may
not only be defined as something ‘out of the ordinary’, elements of ritual derive from the
context of daily life. Brück suggests that a distinction between ‘practical’ and ‘ritual’ actions
may not exist in all societies. She considers the concept of ritual to be of no value for the study
of archaeology:

‘it would seem unhelpful to apply a functionalist approach to certain aspects of prehistoric life (for
example subsistence) while admitting that others (such as religious beliefs) cannot be explained in
this way. Rather, prehistoric peoples applied an historically-specific logic to the world around
them. This comprised a set of culturally-specific values, aims and rationales which shaped their
practical interaction with the world’ (Brück 1999: 327).

The theory of ‘ritualization’ developed by Bell (1992) may be a more productive way of
looking at the relationship between what we perceive of as ritual and space, without creating a
rigid and permanent distinction between the sacred and the profane. Bell does not go as far as

Brück in abandoning the concept of ritual but considers the way in which action marks out
activity as ‘ritual’. Bell sees ritualization as a strategic form of social action through which
these distinctions are created and privileged (1992: 74). By emphasising the embodied nature
of ritualized action, this approach would suggest a dynamic interpretation of ‘sacred space’
through action and movement. ‘The strategies of ritualization are particularly rooted in the
body, specifically, the interaction of the social body within a symbolically constituted spatial
and temporal environment’ (ibid.: 93). This recognises the materiality and performative nature
of ritual practice. By the creation of certain settings and use of certain objects and gestures, the
ritual nature of the action is established. The material culture used in such circumstances is not
necessarily distinct from that found in a mundane context; indeed, it may gain power from its
domestic associations.

So ritual activity does not take place in a vacuum but is part of daily life. Ritual and secular
spheres may converge or at least overlap. However, this is not to suggest that the landscape is
experienced in the same way in the context of ritualized action. The same landscape can take
on a completely new set of meanings and be perceived as another place during religious
contemplation, pilgrimage or procession. The layout and topography may reflect aspects of
another place; its architecture may be replicated (Loosley 1999: 19), or the other place may be
alluded to by signification alone. Space can be used to represent aspects of historical or
mythical stories, for example in the Stations of the Cross in Christianity. Conceptual distance
therefore need not equate to physical distance (Coleman and Elsner 1994: 77). The
architectural similarity between places of worship within a religion can also deny place in a
sense, replicating a similar use of space in many different places. It is the differences between
these buildings that attest to their unique histories.

Sacred places are created through the process of pilgrimage and repeated visiting keeps
them in individual and collective memory. Places gain significance through stories built around
them and the practices into which they are incorporated. Sacred places are therefore linked to
the everyday world by means of this process, and renewed and transformed by it. The act of
pilgrimage extends the boundaries of sacred space, so that ritual landscapes exist as layers
within those lived in daily life. For example, in rural France people may visit a church on a
regular basis, but a ‘pilgrimage’ might be made only once a year on pre-determined festivals.
The act of ritualized movement brings additional meanings of a place to the fore through
visible performance.

Bounded space?
This emphasis on movement brings another dimension to archaeological study of landscapes.
Routes connecting locales can be as significant as the sites themselves (Stoddard 1987: 450), as
it is through these routes that people experience and interact with place. It may be helpful to
look at these routes away from a site-based archaeology. Although centred on a bounded space,
the cult may have had an impact outside of this space, perhaps with markers or stopping points
such as roadside shrines, which create a sense of purpose and destination and help to fill
journeys with meaning. Certain features appear prominent and significant either through their
visual impact or through pre-existing knowledge about them. The reactions of the visitor
depend on the visitor’s cultural background, social standing and physical attributes. Conceptual
connections may also have been created between relatively distant places, forming a wider
network of meaning. It could be argued that extramural temples extend their realm into the
city; there is mutual implication through routes, orientation and sight. The notion of a
discretely bounded sacred space is therefore at odds with the experience of a temple site,
whether or not there was a physical boundary or temenos.

In the case of Mâlain, Côte-d’Or, at least three sanctuaries have been identified, with that
at Les Froidefonds, Ancey (Roussel 1969) placed some distance away from the centre of
occupation and looking away to the east. This may however have been linked to the settlement
by means of an aqueduct and lines of sight might have been created between the theatre and the
temples on the higher ground to the east and west. The streets lead up to this place from the
area of theatre, suggesting that cult buildings are central to the articulation of the settlement.
Indeed, they may have been the reason for its existence, as they appear to predate the area of
settlement excavated at La Boussière (Roussel 1975) and go out of use in the same period.
The temple complex at Sanxay, Vienne (Formigé 1945) provides a good illustration of how
temple complexes can make use of landscape on a large scale. Here connections between the
temples, theatre and baths were made through lines of sight. The passage of water was
rendered visible by the use of open basins and the whole complex is situated above the river
Vonne. Although the theatre was separated from the remainder of the complex by the river
valley, it was clearly part of these same articulations of space. To reach the temple from the
theatre, a visitor would have had to cross the river. The experience of visiting the temple would
therefore not have taken place in isolation from its surroundings, and would have required
significant engagement with its topography.

Water and rivers
The most common location for a temple is on high ground on a spur of land overlooking a river
valley. In some cases, they appear to be oriented to face the river, as in the case of Crain,
Yonne, which faces southeast down to the river Yonne, and Montmartre, Yonne, which
overlooks the river Cousin (and its confluence with the Cure). There also appears to be a
preference for a position at the confluence of two rivers, typically with a view of both valleys,
as at Alésia. Temples are also situated at confluences when these are not on higher ground, as
in the case of the sites of Prégilbert, Sainte-Pallaye, Trucy-sur-Yonne and Cravant, all located
at the confluence of the Yonne and the Cure, Yonne. The confluence appears to have had
particular significance, possibly because of the way in which rivers were bound up with tribal
identity in the naming of peoples and gods (for example the Sequani and the Seine, with its
eponymous goddess Sequana). This is demonstrated by the positioning of the altar of the Three
Gauls at the confluence of the Rhône and Saône in Lyon, where imperial ideology appears to
have appropriated this emotive imagery.

Those building temples appear to have manipulated watercourses in the immediate area,
possibly to allow access to them or to incorporate them into the temple site for ritual purposes.
A deliberate selection of marshy ground appears to have been involved for example in the
siting of temples at Entrains-sur-Nohain, in the marsh of Saint-Cyr, Nièvre. The site of Les
Fontaines-Salées, Yonne, may have been surrounded by water. It is possible that the baths and
spring catchments were on an island, with the ancient bed of the river Cure passing between
them and the building to the northwest. The whole area was almost an island, being bounded
by a watery landscape with marshes to the north (‘les marais de la Bazaine’) and south (known
as ‘les marais de la Morte’), and the river to the east (Louis 1944: 42–3). This area was drained
in the fourteenth century. Similarly, the temple at Les Bolards, Côte-d’Or, appears to have been
built on a low-lying gravel bed between two oxbow bends of the river Meuzin (Pommeret
2001: 9).

At some sites, there appears to have been considerable manipulation of the terrain in order
to maximise visual effects. The terracing at the sanctuary of Apollo Moritasgus at Alésia/
Alise-Sainte-Reine, Côte-d’Or (Rabeisen 1992), would have created distinct levels between the
baths and the temple, allowing maximum impact of the view eastwards upon reaching the
temple. This makes use of natural ruptures in the limestone plateau, but may also have been
enhanced artificially. The different levels of terracing at Sources de la Seine, Côte-d’Or, may
have also operated to control movement of people in stages of a ritual sequence or according to
social groups (Aldhouse-Green 2001: 68). Deliberate levelling of the site at Essarois was
proposed by the excavator and may have been intended to manipulate the relationship between
the temple and the stream (Daviet 1963: 4).

At Alésia, a spring was incorporated into the temple complex and the presence of water
flowing through the temple foundations appears not to have been disguised, giving a natural
character to its architecture. By their placement at the head of springs and the sources of
tributaries (as at Sources de la Seine), it could be argued that temples and shrines acted to
control or sacralise these sources (or vice-versa). The power of the associated deity was thus
extended over the whole landscape by means of this water, which traversed bounded space
(Ghey, forthcoming). Their impact therefore would have been felt over a wider area than may
be apparent.

A number of temples appear to have been visible from roads, for example, the temple at
Les Bolards was aligned with a long approach from an offshoot of the Agrippan way. This may
be significant given the ideological implications of roads outlined by Witcher, who emphasises
their colonial nature and impact on previously inhabited space, and suggests that they are ‘part
of an imperial dialogue articulated in time and space’ (Witcher 1998: 60). The roads facilitated
imperial domination and brought the landscape into the military arena. Even if they used preexisting
routes, they were named anew and made straight. Perhaps like the temples themselves,
the roads suggest an ambiguous attitude to past landscapes, ‘both a maintenance and as a
subversion of pre-existing conceptions of space and place’ (ibid. 1998: 67). The same could be
said of sanctuaries located on pre-Roman road networks, which would indicate the significance
of these places in the pre-conquest landscape. In the case of Les Bolards, the earliest phase of
the sanctuary was located at a juncture of pre-Roman roads that bore no relation to the later
urban plan (Pommeret 2001: 80), so the later road network could be said to have been
constructed with relation to this pre-existing cult site, even if this was partly complicit with a
new ideology.

The significance of rivers in Roman imperial ideology may have been similar to that of
roads. They were undoubtedly important communication routes, particularly in mountainous
and forested landscapes such as that of the Morvan. The visibility of temples from rivers may
thus have created an impact in the same way as temples visible from roads. However rivers
were also conceptual and physical boundaries, perceived as such from pre-Roman times.

Relationships between sites
In his work on Romano-British temples, Smith proposed the intentional visibility of groups of
temples in the Cotswolds and Southern England (Smith 2001: 150). This could have been
significant in reinforcing their interconnection and created visual goals for those travelling
between them. It may be possible to see groups of Burgundian temples clustering in the same
way. Even in a wooded landscape, it is probable that these temples were visible landmarks. The
visitor to a site such as Mont de Sène (Santenay, Côte-d’Or) may have been aware of several of
these. In addition, if temples themselves were not visible, distant peaks would have been
instantly recognisable and their religious significance is likely to have been known.

Figure I: Chassey seen from Mont de Sène, Santenay (Côte-d’Or)

From Mont de Sène it is possible to see Mont Beuvray, Mont Saint-Vincent (both with
known temples), Mont Rome, and Chassey (Saône-et-Loire), opposite across the Dheune
valley where another temple was located. The flint assemblages at Chassey and Mont de Sène
suggest contemporary pre-Roman occupation of both sites (Bulliot 1874: 139) and the
sculptures of Mercury found at both sites indicate a later connection. Artificial terracing
enhanced the position of the temples on Mont de Sène; a bed of rubble was found underneath
them, making them slightly higher than the other buildings situated just below the ridge
(Bulliot and Thiollier 1892: 155). Interestingly, the orientation of the two temples at Chassey
and Mont de Sène suggests that they were not primarily intended to face each other, although
they were intervisible. The unusual orientation of the temple on Mont de Sène gives it a wider
view of the Dheune valley further east of Chassey, into Santenay itself and an area known for
its springs; Roman coins and a piece of gold leaf were found in a Roman well in this area
(Thevenot 1971: 267). The temple at Chassey (de Coynart 1869: lxviii) is located in a position
that commands a view up the Cosanne valley towards Mont de Rème, and overlooks the
confluence of the Cosanne with the Dheune.

Past and present
Landscape is in part a creation of past human action, and experience of landscape involves
interaction with and negotiation of this past. The preservation of monuments and other features
will ensure that they enshrine and continue to create memories. However, these memories will
have been selected and socially sanctioned, while others may be forgotten or retained by fewer
individuals. Original meanings will be lost and features created by human activity may be seen
as ‘natural’, ‘part of the landscape’. The role of associated memories and oral histories in
lending significance to a place cannot be underestimated (Bradley 2000: 157; Meade 2004: 87–
88) and so it is necessary to bear in mind the invisibility of this aspect of the landscape.
While the continuing social significance of prehistoric monuments tends to be a standard
assumption, similar continuities in the Roman period are more often questioned, despite the
fact that the chronological scale is much shorter. A longer-term perspective is necessary if we
are to give meaning to the Gallo-Roman landscape and continuity of place is perhaps to be
expected rather than treated as something unusual. It does not necessarily indicate continuity of
an entire belief system. Cooney writes:

‘Continuity at the ideological level can be accommodated within a changed world view.
Continuity of place may be the essential ingredient, particularly in a world of narrative
tradition where connections with the landscape are regarded as history’ (1999: 60).

The landscape surrounding Alésia is a good example of a landscape that cannot be
considered as neutral space. The area surrounding the hillfort was an arena of war, the scars of
which are still visible today. The physical traces of this could have been extremely emotive
features of the Roman period landscape, with many associated memories that differed greatly
from Roman and indigenous perspectives. It is hard to be certain of what meanings were
intended to be read in the construction of a temple in this particular spot, but it was potentially
ideologically charged.

The careful placing of Gallo-Roman temple sites shows an awareness of the way in which
landscape informs the experience of ritual sites. In turn, the experience of landscape was
irreversibly altered by the construction of these distinctive monuments. By their presence, they
gave a new aspect to towns and journeys, and to the horizon seen on a daily basis by local
inhabitants over a wide area. Although on a local level, those constructing and visiting these
monuments may have been sensitive to the indigenous ‘gods of the place’ (Millett 1995: 99),
the conspicuous uniformity of the Gallo-Roman temple could represent linkages between
distant places, those visible and those beyond. The gods of these places were seen and looked
out over rivers, valleys and roads, attracting pilgrimage and inspiring awe. However, any
visitor to the temple would be experiencing it through its immediate context in a very physical
sense, and in a unique way, by the effort of walking up or down to the site, becoming aware of
new perspectives from a theatre seat, and hearing the sound of running water, for example. It
was through these experiences that the visitors made sense of their surroundings and in this
way that the gods of the place revealed themselves.

Department of History and Welsh History, University of Wales, Bangor

Acknowledgements
This paper is based on PhD research funded by the University of Wales, Newport (Ghey 2003).
I would like to thank Miranda Aldhouse-Green, who supervised this research, and Adrian
Chadwick for his comments on the text. I would also like to thank Élisabeth Rabeisen, Jean-
Louis Coudrot, Louis Roussel and Emily Hunt for their assistance in Burgundy, and Ben
Croxford for organising the session at TRAC in Durham.