Reading the Roman house: the social interpretation of spatial order

Mark Grahame

Ifwe wish to study the Roman house, how should we go about it? If
we were to follow the tradition within which most of the literature on
Roman houses has been produced, our concerns would be “painstaking
recording and analysis of the minutiae” (Ling 1993, 331). The persistent
aim of such art historically driven research has been to create ever more
refined chronologies and typologies of the architecture of the house itself
(e.g. Richardson 1988) and of the decoration and objects found within it
This has had the unfortunate effect of fragmenting the archaeological
evidence, causing it to become divorced from both its physical and social
contexts (Wallace-HadrilI1988, 48).
More recently, however, some have voiced criticisms. Andrew
Wallace-Hadrill (1988, 1994) has perhaps been the most notable. He has
argued that the quality and quantity of decoration within the Roman
house was closely linked to social standing. The nature of aristocratic
competition and the role of patronage in Roman political relations re-READING TIIE ROMAN HOUSE 49
quired members of the elite to differentiate space within the house. Social
status was judged in relation to the head of the household, the paterfamilias, and this detennined how far visitors were permitted to penetrate
the house and where they were received. The more intimate the relationship the more intimate the setting for their reception. Oients were
reputedly received in the atria of aristocratic houses, while” friends” were
entertained at dinner in triclinia and oed. Differently decorated space
therefore acted as a sign system, linked to the political necessity for the
Roman elite to signal their social status to different categories of visitor.
Through analysis of the decorative schemes and their relations in space a
means is thus provided for the interpretation of spatial order (see for
example Clarke 1991; Gazda 1991; Thebert 1987; Wallace-Hadrill 1988
and 1990).
Although preferable to the art-historical tradition, this reorientation in
approach may still be criticised as not being adequately grounded in
theory. To consider the house as a “social unit” (Wallace-HadrilI1988, 48;
1994, 7) is one thing, but to understand how social life is constituted is another. This issue, which lies at the heart of contemporary social theory,
has not been given due attention by ancient historians or classical
archaeologists. In this respect Thebert’s (1987, 408) remark about “our
lack of knowledge concerning the way in which the social elite reproduced itself” is instructive. He appears to think the problem originates
from a lack of the right sort of evidence: details of property ownership
through time and family genealogies. However, even if Thebert did possess such information any understanding of the reproduction of the elite
would founder unless it was grounded in an adequately developed
theory of power. The elite only remains the elite while it can replicate the
social relations of dominance and dependency which define them and the
non-elite alike. We might then ask if the Roman house played any role in
the reproduction of such unequal social relations? To answer this question we need a theory of the role material things play in the constitution
and reproduction of society. The lack of theorising is perhaps the most
serious omission from attempts to interpret the Roman house. Although
some recent studies have used theory to provide a framework for interpretation, these are exceptional (e.g. Grahame 1998; Lawrence and
Wallace-HadrilI1997).
The lack of theory sterns ultimately from the dependency upon the literary texts to provide an adequate basis for the interpretation of archaeological material. But the vision of social life documented by the literary
evidence is not representative. Laurence (1995, 64) has already pointed so M. GRAHAME
out that the ancient authors wrote predominantly about Rome. Rome
was, in terms of population, size, and nature, very different from other
Italian cities like Pompeii, Herculaneum, and Ostia, which are our main
sources of evidence for houses. Even more problematic, from a theoretical standpoint, is the view advocated by Wallace-Hadrill (1988, 48; 1994,
7) that through judicious use of the texts we can begin to understand how
lithe structures and artefacts we disinter spoke to the Roman user” and
thereby gain an insight into how the Roman user responded to them
(1988, 48). The assumption here is that literary evidence enables us to
define the meanings which the physical remains had for their users in an
unproblematic way. In other words, it opens a vista on to the past,
allowing us to view the house as if through II Roman eyes”.
To what extent can this be said to be the case? A text is a sign system
and literary theorists from Saussure (1960) onwards have pointed out
that the relationship between the components of the sign, the signifier,
and the signified is arbitrary. Hence, there is a role for the reader in the
interpretation of texts. Meanings are read into texts, they do not emerge
from them. It follows that the meanings which historic texts have are thus
created in the present. They are not the unique preservation of the past’s
own self-awareness. Hence, we cannot employ the literary evidence as a
means of empathising with the past Furthermore, contemporary archaeological theory advocates that material culture is itself a sign system,
structured by, and structuring of, social practice (see e.g. Hodder 1991;
Shanks and Tilley 1987; 1992). If this is so, then the ways in which Roman
material culture “spoke” to the user and the ways in which the user responded to them will have been contingent upon the user’s exact social
position. We might suppose that Roman material culture may have held
very different meanings for those individuals not part of the elite than
those espoused in the ancient sources. It was certainly in the interests of
the Roman elite to represent their viewpoint as universal and to “fix” the
meanings that things had. This should not be taken to mean that elite
meanings were not contestable, nor contested by others within Roman
society. As a means of representing the social and cultural order literacy
was primarily an elite activity (Harris 1989). However, as a sign system,
Roman material culture was also a powerful means by which to represent
the social world. The way society expresses itself goes beyond the verbal
and therefore, by implication, the written. Texts and objects can be
thought of as forming competing “discourses” which may not always
have been in agreement. The consequence of this for the study of Roman
material culture should be clear; the literary texts do not provide a READING TIlE ROMAN HOUSE 51
unique insight into the past which permits us to use them to interpret
archaeological material.
THE HOUSE AND SPACE!
If we are to study the Roman house with recourse to the texts, we must
begin from first principles. We should think about what sort of a thing a
house actually might be. A house can be many things. It is, after all, a
complex object. At one level the house is very much a “container” for activities and objects, but do these alone define what a house is? The house
is itself a physical object with a definite structure. It is a “building”. A
building may be defined simply as an object with a continuous architectural boundary that delimits a discrete region of space and is punctuated
with one or more entrances. Such diverse architectural forms as Neolithic
henges and Iron Age enclosures thus also qualify as “buildings”. The
purpose of a building is to create and order empty volumes of space
(Hillier and Hanson 1984, 1). Space is not itself an empirical object. As a
void, it cannot have any properties or possess causative powers (Werlen
1993, 1-8). We might therefore be tempted to think of space as simply a
“stage” or “setting” for human action.
However, to reduce space to a mere physical setting is to take too pessimistic a view. Because of the routine nature of social life, all societies
will habitually undertake certain activities in particular places. These
“places” Giddens (1984) has termed locales. A locale is a physical region
of space with definite boundaries. Although it provides a setting for interaction, the demarcation of the locale as a distinct region of space also
helps to concentrate it in some way (ibid., 375). As all human action is
social action, the presence or absence of specific individuals is required in
order to ensure that any particular activity is successful. The differentiation of locales from one another is fundamental in ensuring that the appropriate levels of “presence-availability” are maintained. Presence-availability is the degree to which social others are available for encounters
(ibid., 73). The boundary to a locale may be symbolic, but the use of
architecture is more powerful in that it physically constrains movement
Architecturally divided space thus regulates the degree of presenceavailability by constraining or enabling movement into different spaces.
SPATIAL ORDER, PRINCIPLES, AND PRACTICE
The manner in which architecture orders space will not be arbitrary. In
their seminal work on this subject, Bill Hillier and Julienne Hanson (1984) 52 M. GRAHAME
demonstrate that even apparently random agglomerations of structures
are the consequence of “rules”. This position receives further support
from Fletcher (1977, 49-53), who points out that order in architecture has
often been equated with conscious planning, causing the dismissal of
anything which cannot be measured according to “standard units” or
“harmonic series”. This creates a false dichotomy between “planned” and
“unplanned” buildings, with the latter often being associated with
vernacular architecture.
Houses are built according to culturally specific principles about how
space should be appropriately ordered These principles may become
represented in a particular theoretical discourse and this discourse may
alter the practice of building. However, it does not transcend the practical
knowledge upon which the discourse is based. Such practical knowledge
is acquired during the processes of socialisation, through which individuals gain an intuitive understanding of how to “go on” in social life
(Bourdieu 1977; Giddens 1984). Habitual practices instil dispositions into
individuals, which lead to certain expectations. If something does not
match these expectations, individuals may “know” that it is “wrong”
without necessarily being able to, or having to, discursively formulate
why. This non-discursive knowledge has been described by Giddens
(1984,41-45) as “practical consciousness”. Individuals gain a practical, or
intuitive, understanding of spatial order by moving into and through different locales during socialisation. This understanding forms the basis for
future action and will structure both the location and internal arrangement of houses. The manner in which space is arranged within the house
is thus the very physical embodiment and indeed the representation of
spatial order. Because the house structures movement into and through
different spaces, the principles upon which the ordering is based are reinstalled into the next generation. In this way a particular spatial order is
reproduced through time.
For this reason we must not think of the house as simply a given
physical object, helping to generate certain patterns of social relations and
not others. On the other hand we should not believe that the principles of
spatial order are somehow mechanically reproduced The building of a
house literally means the reinvention of spatial order anew each time.
This provides opportunities for the established ordering to be contested.
Spatial order is not fixed, but is analogous to a “discourse”, continually
being reconstituted through practice. Even in cultures where the use of
masonry architecture seems to fix the spatial order for long periods, there
is usually evidence for doorways being blocked and opened and rooms READING THE ROMAN HOUSE S3
being added, demolished, or reshaped.
However, architecture does help conserve the principles of spatial
order and provides a means for those principles to be retrieved and reproduced. Shanks and Tilley (1992, 133) have argued that because of
material culture’s “fixation of the practice embodied in it,” it permits “a
relative pennanence and efficacy in the structuring of subsequent practice”. From this standpoint architecture provides human societies with
durability and persistence by permitting a stabilisation of social relations.
The fixing of a certain order in space limits the possibilities of a transformation of social practices through redeployment in space. This adds a
further interesting dimension to the ordering of space. To construct
masonry architecture, especially if highly segmented, requires the mobilisation of both labour and material resources. Those individuals who
have the means with which to construct such buildings have the power to
“fix” in space an order which works largely to their benefit This results
in the replication of the structures of inequality upon which the powerful
depend. Since the physical form of the house has a high degree of permanence it appears entirely ” natural ” and limits the likelihood that the
existing spatial order will be contested. The house then provides a powerful means of perpetuating the existing social order.
The fixation of practice implies the fixation of meaning. If this is so,
then it suggests that the relationship between the signifier and signified
may not be as arbitrary as has been suggested by post-modernist literary
theory. Indeed, post-modernism has been criticised by both Marxists and
feminists (e.g. Harvey 1990; Wolff 1992) for a failure to take into consideration the real material conditions of existence within which individuals
have to operate. Such conditions, although represented and mediated
discursively, cannot be reduced to mere discourses. We might then postulate that the meanings which things have arise, in part, out of the
context of their use. For example, we may formulate a meaning for a particular space as ” public “. However, if in reality this space is relatively
inaccessible then its public meaning will simply dissolve. This, as Hillier
and Hanson (1984) point out, is precisely what occurred in modern town
planning. The meaning ascribed to spaces by planning theory were often
contradicted by the actual pattern of movement generated by the spatial
order. Spaces created to be public ended up being deserted. Support for
this proposition might be found in Gibson’s (1979) work on perception.
Gibson argues that there is a universal basis for human perception.
Because information is received by the brain already structured by the
environment, it means that material objects permit, or “afford,” certain 54 M. GRAHAME
UNDIFFERENT1AlED SPACE
*
WBOund”‘ll
t * y
* Segregmed Space
*
Figure 1. The effect of the boundary on the level of presence-availability. It will be reduced
for the individual X, as compared to the individual Y, due to the relative reduction in the
probability of an encounter caused by the boundary.
possibilities to the perceiver rather than others.
Following from this we might ask if there is anything about spatial
order which might afford certain types of meaning. To provide an answer
to this we need to return to give further consideration to the nature of the
boundary.
THE CONCEPT OF THE BOUNDARY
The concept of the boundary features prominently in Hillier and
Hanson’s social theory of space (1984, 73 ff.). The purpose of the boundary is to enclose a definite region of space and segregate it from what
would otherwise be undifferentiated space. This segregation affects the
level of presence-availability within the space, in that the probability that
an encounter will occur by chance alone is significantly reduced (fig. 1).
The boundary thus creates a “fundamental discontinuity” in the principles of spatial organisation (ibid., 144). However, as Hillier and Hanson
point out, no space is ever totally enclosed. There is not much point in READING TIlE ROMAN HOUSE 55
creating a space which can never be entered! Although it might seem
contradictory, to be effective a boundary must be permeable. The degree
of presence-availability within a space will therefore depend upon the
number and nature of the relations of permeability which it has with
other spaces.
This means that building space obtains its meaning through its relational order. This order creates and reproduces a particular pattern of
permeability characterised by the juxtaposing of spaces with different
levels of presence-availability. What activity is undertaken in which
space will depend upon how that activity is interpreted in terms of the
necessary degree of presence-availability. For example, in most societies
sex is restricted to the private zone, whereas the consumption of food can
range from being a very intimate, private affair, to a very public and
communal one. Hence, although the order which space assumes in any
cultural context will be specific to that culture, each group will make different uses of the same logical properties of space. This then provides a
basis for the analysis and interpretation of the spatial order preserved in
archaeological remains.
Access analysis
To grasp this order we require a means by which to represent it. The map
or plan is, of course, the archetypal representation of spatial order. However, it is often difficult to grasp from a plan alone, particularly if we are
considering a large number of cases. Hence, we need to move beyond the
plan. Fortunately, Hillier and Hanson’s access (gamma) analysis (1984,
82-175) provides us with the means to do precisely this. Because their
theory and method have been deployed in a number of recent archaeological studies (Chapman 1990; Fairclough 1992; Foster 1989a and b;
Grahame 1995, 1997; Laurence 1991, 1994; and various papers in Boast
and Yiannouli 1986), only a brief outline need be given here. Analysis
begins with the construction of an access map (fig. 2a-c). This is done by
representing each unit of space within the building as a small circle and
then linking these with lines when there is permeability between them
Transitional spaces such as passageways may be represented as a black
dot, while the exterior space is usually denoted by a circle with a cross
through it The map may then be “justified” with respect to the exterior
space. This is done by making the exterior the “root” of the system and
then arranging all spaces at a similar depth from this root above it, while
ensuring that the links of permeability are maintained.
The level of presence-availability in each space will depend upon its 56 M. GRAHAME
A
.J c
B 21
c
Figure 2. Diagram A shows the ground plan of a house from Regia VI, insula block
Pompeii. B shows the access map for the same house. C is the access map justified with
respect to the exterior. READING TIIE ROMAN HOUSE 57
Exterior
Interior …… 1—Boundary
Figure 3. The simplest fonn of building consisting of a single cell with one entrance.
exact location with respect to all other spaces. Three aspects will define
the space: the global, the local, and its depth from the exterior. The global
aspect is the relationship a space has with all other spaces within the
building. This aspect may be captured by considering how accessible that
space is. The more accessible a space the fewer the number of boundaries
that will have to be crossed in order to reach it. The local aspect may be
understood by considering the number of irrunediate neighbours any
space has. If it has a large number of neighbours then we can say that it
controls access to those neighbours. The final aspect, depth from the exterior, will indicate how accessible any space is from the outside (table 1).2
However, “units of space” are not equivalent to “rooms”. We must be
aware of how space is actually constituted by the architecture. To understand this we need to examine the concept of the boundary in more
detail. The simplest form of building consists of a cell, which delimits a
region of space and is permeated by a single entrance (fig. 3). The boundary itself consists of an “interior” and an “exterior”. It is the interior of the
boundary which delimits the space within the cell. However, if we were
to aggregate a number of cells together, or arrange them one within another, the exterior of boundaries would also delimit regions of space. In
fact, it is useful to think of buildings as being composed of cells within
cells. In other words, building space may be defined by both the interior
and the exterior of boundaries. When considering the characteristics of a 58 M. GRAHAME
Table 1. The defining aspects of space within buildings.
Aspect
global
local
exterior
Characteristic Description
accessibility average number of boundaries which have to
be crossed in order to reach a space from any
start point
control number of immediate neighbours that a space has
depth number of boundaries to be crossed in order to
reach a space from the exterior
c c
0 c
Figure 4. A simple building which may be thought of as being of the three “closed” cells
(e) and one “open n cell (0).
particular space we need to consider whether it is being moulded by the
interior or exterior of a boundary. We might therefore think of a building
as consisting of combinations of two different types of cell: “closed” and
“open”. A closed cell contains a space delimited by the interior of a
boundary, while an open cell is a region of space which is defined by the
exterior of boundaries or a combination of both (fig. 4). In effect an open
cell is that which is “left over” after distinct regions of space have been
enclosed within closed cells. We should not, however, think of these open
cells being less important to the ordering of space. As we shall see below
open cells have properties which make them fundamental to the mediation of the level of presence-availability within buildings. READING THE ROMAN HOUSE 59
INHABITANTS AND STRANGERS
Before we can assess the significance of spatial order we need to ask who
exactly is involved in the “corning together” of co-presence? In the case of
the Roman house, we might be tempted to tum to the literary evidence
and overlay an idealised family or household structure onto the physical
remains. However, we cannot be sure how representative this structure
is. To avoid making simple empathetic assumptions we need to consider
instead the social categories which are likely to be differentiated by space.
Two categories are apparent in the separation of the space inside the
building from that exterior to it inhabitant and stranger. “Inhabitants”
have preferential rights to use the particular spatial domain defined by
the architecture of the house whereas “strangers” do not The identities
of “inhabitant” and “stranger” are both constituted and reproduced
through practices which routinely include some individuals while simultaneously excluding others. These habitually situated practices give
inhabitants preferential knowledge and therefore control over certain
spaces. Such social identities are implicated in practices and should not
be thought as somehow existing externally to them Furthermore, they
are not absolute categorisations: individuals are “inhabitants” in certain
situations and “strangers” in others.
The level of presence-availability varies according to whether relations
between inhabitants or between inhabitants and strangers are involved.
Spatial order may be thought of as being structured along two
dimensions (table 2). The first is that which mediates relations between
inhabitants. The position of a space along this dimension depends upon
the degree of accessibility and control it exhibits. The second dimension,
relating relations between inhabitants and strangers, depends primarily
upon how accessible any space is from the exterior. These two are not
independent of one another; they will overlap in all situations to some
extent The point is that we should not assume at the outset that the
spaces furthest from the exterior are always the most segregated.
Table 2. The dimensions structuring spatial order
Dimension
inhabitant-inhabitant
inhabitant-stranger
Relevant characteristics
access ibility
control
depth from the exterior 60 M. GRAHAME
In terms of the first dimension (inhabitant-inhabitant meetings), if a
space is accessible and has a high degree of control, it follows that the
level of presence-availability is likely to be high. The conjunction of these
two aspects results in a space which forms a “node” in the relational form
of the spatial arrangement (for example spaces 4 and 15, fig. 2c, clearly
form nodes). A node will tend to generate encounters, for routine movement through the building will be continuously directed towards it In
contrast, spaces which are highly segregated and have a low degree of
control exhibit the reverse characteristics. They are likely to be cells with
single entrances, removed from the routine flow of action (for example
spaces and 16-20, fig. 2). The level of presence-availability will tend
to be low, since there will be a reduced probability that routine movement will require entry to that space.
Along the second dimension (inhabitant-stranger meetings), it is the
exterior space which will be characterised by a probabilistic pattern of
encounters and hence a high level of presence-availability. In relation
to the exterior all interior spaces will have a much reduced level of
presence-availability, and the further from the exterior any space is, the
less likely it will be that a stranger will have access to it
FRONT AND BACK REGIONS AND DISCLOSURE AND ENCLOSURE
What are the social implications of these variations in spatial characteristics? Here I wish to introduce the concepts of “front” and “back” regions.
These were originally identified by Goffman (1959) who employed a
“dramaturgical model” in his interpretation of the nature of public Hfe.
Front regions are conceptualised as stages upon which individuals “act”
out the rOles prescribed for them by society. In these regions the norms of
society hold sway and the presence of others produces a powerful
incentive to conform to the accepted modes of practice. Back regions, by
contrast, are analogous to the “backstage”. These allow an escape from
the scrutiny of others and hence a relaxation of the demands to conform
A front region thus involves the disclosure of the body, which means
literally making oneself visible to others. By contrast, back regions allow
enclosure of the body, the hiding of oneself away. Front and back regions
are not absolute categories but relative ones. A public space within a
house, such as a passageway, may be a front region in comparison to the
rest of the house but a back region in comparison to the exterior.
The aspects of a space which enforce disclosure are high accessibility
and control. As we have seen, these aspects characterise nodes in the READING TIm ROMAN HOUSE 61
spatial order. The high probability of an encounter occurring within these
spaces constitutes them as front regions. These are spaces within which
individuals “see” and are “seen” by others. In short, they possess the
characteristics we associate with the “public” domain. The pressure to
conform to accepted modes of practice make front regions primarily
spaces of politics and negotiation. The presence of others enforces a degree of “tact” in interaction. This is a latent conceptual agreement between the participants in interaction of the requirement for the maintenance of mutual self-esteem (Giddens 1984, 75).
Spaces which are segregated and have low control are most likely to be
back regions: places which provide the means of enclosure. Such spaces
may be either places of escape or confinement Escape from the scrutiny
of others may be thought of as a fonn of “self” confinement To be confined, or to confine oneself, requires spaces that are removed from the
routine flow of action. In contrast to the politics and tact of front regions,
back regions are more ideological in character. By this it is meant that
they are more likely to be associated with particular individuals and
practices. The privacy of enclosure permits selfishness, self-indulgence,
and deviation from accepted norms of behaviour. If confinement is an
habitual part of the daily routine, then we might expect there to be a proliferation of such back regions in contrast to frontal contexts.
Between these two extremes are spaces which may exlubit contradictory characteristics, such as high accessibility and low control. Whether
these spaces are front or back regions will depend more upon their exact
relations with other spaces and also upon whether relations between inhabitants or between inhabitants and strangers are involved.
In terms of relations with the exterior, the front-back divide corresponds neatly to “inside” and “outside”, in that exterior space is after
all the archetypal “front” region.
THE SOCIAL INTERPRETATION OF SPATIAL ORDER
The above discussion has outlined how the location of a space within a
building affects the level of presence-availability of that space. This characterises the space as more or less public, or private, depending upon the
social categories involved. The formal analysis of spatial order thus provides a basis for its social interpretation. This is consistent with the
argument above that meanings arise, in part, out of the real material
contexts that individuals confront on a daily basis. Having established
this, we can move the discussion on to consider the Roman house itself. 62 M. GRAHAME
The primary arrangement
The research for this study was based on the ground floor plans of 144
“architectural units” from Regia VI in Pompeii (Grahame 1995). The
maps used to provide the basis for the analysis were the Corpus Topographicum Pompeianum maps (van der PoeI1984). The accuracy of these
maps has been the subject of favourable comment (Wallace-Hadril11990,
155) and they have corrected some of the errors of the previous survey
undertaken by Eschebach (1970). Each unit contained a minimum of five
spaces or more. This minimum was imposed by the method and no
further attempt was made to restrict the scope of the analysis. Although
there is evidence, in the fonn of stair wells, for upper storeys, these are no
longer present in Regia VI. However, from what we know of houses
where an upper floor has survived, it seems that the ground floor was the
main forum for domestic interaction and was almost certainly where
interaction with strangers took place.
This research showed that one principle of spatial ordering was of particular significance in that it occurred repeatedly. This was a single,
centrally located space that fanned a node around which the other spaces
tended to group. Relations with the exterior were mediated by a space
which interceded between it and the node. We might think of this
structuring of space as being the primary arrangement. In terms of the
traditional labels applied to the Roman house, the most prominent node
turned out consistently to be the atrium. Segregated from the exterior by
the entrance passage of the fauces, this ordering of space is so apparent
that the atrium house has become conceived of as being the typical
Roman dwelling (fig. 5}.3
By returning to the notion of the building as being composed of cells,
we can begin to appreciate how the primary arrangement was formed. If
we arrange a series of “closed” cells so that they fonn a courtyard, while
still maintaining access to the space outside, the resulting structure is not
too dissimilar from that of the atrium house (fig. 6). The courtyard is, in
effect, a large “open” cell, formed by the exteriors of the surrounding
“closed” cells. Because it is the “open” cell which forms the node in the
arrangement we can conclude that “open” cells have the public properties of front regions, while “closed” cells have those of back regions.
In larger houses, the tendency was for this arrangement to be repeated,
rather than be elaborated. For example, the atrium house with peristyle
might be thought to contain two such primary arrangements, although
three or more are not unknown from Regia VI (fig. 7). When multiple READING THE ROMAN HOUSE 63
H
Tab F = Fauces
A = Atrium
Tab = Tablinium
H = Hortus
F
Figure 5. An idealised plan of the atrium house.
nodes were present, they were not all equal. Usually a hierarchy existed,
with the atrium generally being the most integrating space and having the
highest control overall. Only in the very largest houses, those with about
forty spaces or over, do we apparently see any elaboration of this basic
structure. Passageways which have rooms off them begin to appear, creating internal circulation routes mostly absent from smaller houses. In
these, there is usually only one, or two, possible routes between primary
arrangements. One route frequently passed through the space traditionally identified as the tablinium, but there was often also a separate passageway which by-passed it Given that the tablinium is usually fonned
by the interior of a cell, while the passageway is shaped by its exterior,
we might speculate that the passageway was the preferential means of
moving between nodes (for example spaces 12 and 14, fig. 2). Both spaces
are usually characterised by having a high degree of accessibility, but low
control. They may be thought of as front areas, forming part of the “public” space of the house but as back regions in comparison to the nodes.
What then is the social meaning of the primary arrangement? It should
be made clear that this arrangement does not exhaust the possibilities.
The spatial order to be found in Roman houses is more subtle, especially
in those houses that contain a large number of spaces. However, space
here precludes any further analysis and the rest of this paper will concentrate instead upon interpreting the primary arrangement 64 M. GRAHAME
—
I
I
I
I
– –
Figure 6. A hypothetical building formed by arranging single cells so as to form a
courtyard ref fig. 5).
Relations between inhabitants: the disciplinary institution
Firstly, let us consider the significance of this arrangement in terms of the
relations between inhabitants. The inhabitants of the house formed the
basic social unit by virtue of their collective rights to use the same spatial
domain. Social relations can only be established and maintained through
contact and frequent encounters between inhabitants would have served
to reinforce them. Encounters would have been generated because the
routine flow of action would have been constantly directed towards the
node. This would have forced disclosure of the body and increased opportunities for interaction.
However, this explanation does not fully account for the existence
of the primary arrangement, since we might wonder why, if “coming
together” was so important, there were so many possibilities for
segregation and confinement To create a high level of presence-availability a single space would have been sufficient There are certainly
numerous examples from the ethnographic record of families sharing a
single space. It is notable too that this juxtaposing of a front region providing, and therefore controlling, access to multiple back regions has
been recognised in the prisons, schools, and hospitals of the modern
period (Foucault 1979; Hillier and Hanson 1984). The common feature READING TIIE ROMAN HOUSE 6S
F = Fauces
A = Atrium
Tab = Tablinium
P = Peristyle
Tri = Trielinium
Figure 7. An idealised plan of the atrium house with peristyle (ef fig. 2a).
which all these buildings share is that they are disciplinary institutions.
Much of Foucault’s writings (e.g. 1977) have been concerned with the
origins of disciplinary power. Enclosure of the body is the generalised
basis for such power and, as Giddens notes in commenting upon
Foucault, this can only be achieved by “partitioning”, which ensures that
individuals are in their “proper” place at any particular time (1984,
145-146). Indeed, as Samson (1992) has recently pointed out, architectural
boundaries remove ambiguities; there can be no doubt if someone crosses
into a space to which they do not have any entitlement to enter. If caught,
any transgression cannot easily be explained away as accidental. The
primary arrangement thus provided a means of exercising disciplinary
power by promoting opportunities for surveillance. Because the node
enforces disclosure it enables the collection and collation of knowledge
about the movement of individuals and its deployment in their supervision. The relative lack of internal circulation routes within the Roman
house would have helped reinforce this.
Architecture segmented in this manner can be interpreted as relating to
the construction and maintenance of inequalities in power through control over the body and its location and movement As Foucault (1983,
219) reminds us, power only exists when put into action and is only
exerted by one to, over, or through another. This “reality” of power
would have been particularly salient in pre-capitalist societies which 66 M. GRAHAME
lacked developed institutionalised forms of dominance. As Bourdieu
(1977,190) has pointed out, in such societies the forms of domination are
elementary, in that the exploitation of labour, goods, service, or homage
can only proceed through the direct domination of one individual by
another.
However, the “disciplinary institution” analogy must not be stretched
too far. Unlike prisons and asylums, and to a lesser extent schools and
hospitals, the spaces of disclosure within the Roman house were not
always controlled by those in authority. This feature of modern institutions Hillier and Hanson (1984, 183-197) have termed the “reversal
effect”, in that the “inhabitants” of the building are the ones confined
whereas the “strangers” are allowed to move about without supervision.
The Roman house does not exhibit the reversal effect, which is tied to the
development of administrative power in the modern period. Except
under these circumstances, surveillance generally operates in both directions. In other words, there may be just as much necessity for individuals
in authority to escape scrutiny by social inferiors in order to “relax” and
participate in activities which might compromise that authority. The
disclosure enforced by the primary arrangement may have provided
opportunities for surveillance but it would also have prompted a familiarity which could have been detrimental to the maintenance of social
distinctions. An alternative strategy would have been to assign social
inferiors to particular spatial domains. Both these possibilities may be
interpreted as ways of creating social distance through physical separation. Of course, it is usually in the interests of those in subordinate
positions to escape surveillance if they can. Hence, the creation of social
distance through physical distance may have held advantages for them
Power is constituted dialectically: it rarely operates in one direction only.
However, if segmentation in architecture equates with control, we might
suspect that on balance the spatial order within the Roman house
operated to perpetuate, rather than ameliorate, inequalities.
Relations between inhabitants and strangers: the theatre
The organisation of space with respect to the exterior is necessary in
order to mediate relations with strangers. The integrity of any social unit
can only be maintained by keeping such social “others” at a distance.
However, it is not possible for any social unit to survive in total isolation
and this inevitably means that the house will have to be penetrated by
strangers.
An analogy of interest here is that of the theatre. A theatre depends on READING TIlE ROMAN HOUSE 67
the creation of a performance space, which is the domain of the inhabitant This is opposed to a space zoned off from the normal flow of
activities which is occupied by strangers. Theatre is the mimicking of
social life but it is also its negation. The performance must be zoned as a
back region phenomenon in order to avoid any possibility of confusion
with actual social life. If we examine the justified map of a Roman house
(fig. 2c), most of the spaces habitually group at a depth of three from the
exterior, while the nodes are usually shallower at a depth of two.
Assuming that action will always flow preferentially towards the most
accessible spaces then, from the exterior, strangers would have been directed into the node but would not have been permitted to penetrate the
house any further. The node, in most cases the atrium, is thus analogous
to the auditorium of the theatre, both enclosing strangers from the outside world, while disclosing them to the inhabitants. The spaces at a
depth of three or more would thus represent the “backstage”, a domain
away from the prying eyes of strangers. Encounters with strangers would
have involved the movement of inhabitants from back to front This
movement would have consequently involved an act of disclosure on the
part of inhabitants, signalling their special status in that they can
“appear” from domains denied to strangers.
In many cases it is the tablinium which acts as the linking space between the node and the back regions “deeper” within the house. The
tablinium also usually possesses the characteristic of being open to the
atrium. The combination of these features makes this space of some interest It is a space of disclosure but its juxtaposition with the atrium also
makes it segregated and thus the domain of the inhabitant, when considered from the point of view of the exterior. This space therefore has the
characteristic of a “stage”, linked both to the domain of strangers in front
and to the “backstage” region behind. Disclosure within this space would
have provided a powerful means of emphasising the distinction between
inhabitant and stranger, while simultaneously allowing contact between
the two. The inhabitant would have had the option either to maintain a
distance or to move into the atrium. These options provided the means by
which relations between individuals of differing social statuses could be
replicated by representing social distance as physical distance. Hence, the
arrangement of space with respect to the exterior was also intimately
intertwined with relations of power and control.
This ordering of space suggests that encounters between inhabitants
and strangers were ritualised. By this it is meant that certain forms of
talk, actions, and objects were deemed appropriate to the encounter. The 68 M. GRAHAME
metaphor of the theatre thus corresponds to the requirement to II act out”
the roles demanded by the ritual. The assumption here is that as the relative social distance increases between individuals, encounters become
more formalised. With a space intervening between the node and the
street it seems unlikely that interaction was casual; it suggests a certain
formality and sense of occasion.
Strangers are always social lIothers” and as such represent a potential
threat However, in a society where power lacked developed institutionalised forms, the reproduction of status depended upon the contradictory
requirement to extend the network of social relations far beyond the
immediate household group. The consistency with which the primary
arrangement occurs suggests a certain predictability from the point of
view of the stranger. Beyond the reception area the degree of variation in
spatial order increases. These backstage areas, by contrast, would have
been places of mystery to strangers, just as the backstage is a mystery to
the audience in the auditorium of a theatre. The predictability of the
primary arrangement would have encouraged strangers to penetrate the
house by reducing the mystery of what lies beyond the entrance. However, the separation of the reception area from the street would have also
clearly signalled the distinction between them and the inhabitants. This
ordering of space would have permitted the reproduction of relations on
the terms of the inhabitant Hence, relations between inhabitant and
stranger were probably antagonistic. Ritualised encounters provided a
context for the amelioration of possible tensions between individuals
who were not of the same basic social unit but were locked into a system
upon which both depended.
In houses which contain two or more nodes there is usually a distinction between the two, with the secondary node often being the space
within which the peristyle is located. The consequence of this is that the
peristyle space is usually less integrating and has less control than the
atrium. In houses with only one entrance the peristyle forms part of the
backstage but even when a second entrance is present its greater degree
of segregation suggests that it was still more of a back region from the
point of view of strangers. We might speculate that different categories of
II stranger” may have entered the house by different entrances, or possibly one entrance may have been reserved for the inhabitants only. If so,
then we might be able to extend the theatre metaphor by considering
secondary entrances to be analogous to IIstage doors”. READING 1RE ROMAN HOUSE 69
THE HOUSE AND SOCIAL CONTRADICTIONS
The analysis of spatial order undertaken here does not reveal details of
the activities undertaken in any particular space, nor does it reveal the
identities or statuses of those involved in the activities. As Bourdieu reminds us, practices can not be deduced from objective conditions (1977,
78). However, it does provide us with an indication of how power relations were constructed and maintained in space. As Giddens has noted,
power cannot be considered a “second order” consideration in the study
of social relations (1984, 283). We should acknowledge that power is intimately involved in all human relations, regardless of whether there is
actually an intention to exert dominance or not
From the ordering of space within the house we can conclude that
Roman society was constituted on the basis of a number of contradictions. A strong household identity may be inferred from the generation of
encounters implicit within the existence of nodes towards which action
within the house was continually directed. By contrast, the segmentation
of architecture suggests inequalities in power within the household
group. This ordering of space is a recurring theme in human societies.
For example, the primary structural feature of the Konkomba settlement,
studied by Fletcher (1977), is a courtyard delimited by huts and walls
linking those huts. Although the group occupying the settlement are related by kin ties there are inequalities between its members. It may be
seen therefore as an attempt to reconcile in spatial order the contradictory
requirements to “come together” to reinforce collective identities and to
create social distance through surveillance and separation.
The Konkomba settlement, however, is physically removed from other
similar settlements, whereas the Roman house is located within an urban
context. This positioning suggests that the mediation of relations with
strangers was of considerable importance. Urban environments generate
high levels of presence-availability. Such levels are required to support
institutions and services which exist beyond the household The social
networks which these presuppose could not be sustained if the ordering
of space did not permit at least a modest flow of activity between street
and house. The mediation of only a single space between the house and
the exterior is in marked contrast to the medieval castle at Edlingham,
studied by Fairclough (1992), or the broch at Gurness analysed by Foster
(1989a). In these the stranger had to pass through multiple spaces before
reaching a domain within which to interact with the inhabitants. By comparison the Roman house was relatively “tied in” to the wider network of 70 M. GRAHAME
the street system This may then be interpreted as an attempt to balance
the imperative to maintain household integrity with the necessity to participate in broader social networks.
All human societies face these contradictions to some extent. What
distinguished Roman society was the particular manner in which the
resolution of these contradictions was attempted. The spatial order preserved within the Roman house was the consequence of an intersection of
the requirement to mediate two distinct set of relations. The result, it is
suggested, is a structuring of space which exhibits a high degree of flexibility. The key role of the atrium in both inhabitant-inhabitant and
inhabitant-stranger relations testifies to this. By way of a contrast, the
Ashanti “palace” (Hillier and Hanson 1984, 167-175) contains different
domains for interaction between inhabitants and between inhabitants
and strangers. The flexibility of the Roman house would have enabled it
to accommodate a range of practices and permit a reasonable change in
such practices through time without necessitating extensive rebuilding.
The presence of multiple spaces which have identical sets of relations indicate what Hillier and Hanson have descnbed as “interchangeability”
(ibid., 214 ff.). This may be compared to the Swahili house (DonIey-Reid
1990), which essentially consists of a linear sequence of spaces. The high
degree of non-interchangeability exhibited here suggests that any transformation of practices would quickly come into contradiction with the
overall spatial order. This conclusion would seem to support Dwyer’s
(1991) case that, in contrast to the established view, the atrium house was
showing considerable vitality and was not a redundant architectural
form by the first century AD.
However, in stressing the flexibility of the Roman house we must not
lose sight of the fact that it “fixed” a particular spatial order. The high
degree to which household space was segregated, it has been suggested,
indicates the existence of inequalities in power. Such asymmetries are
threatened by practices which are too probabilistic in nature. They cannot
be established and maintained by casual contact and loose associations.
Bourdieu (1977, 190-191) reminds us that in pre-capitalist societies differences in power can only be maintained at the cost of repeated social
practices. To be repeated implies the habitual presence of particular individuals at a specified place and time. Segmentation in architecture makes
encounters more predictable and this would have helped ensure the continuance of the practices necessary for the perpetuation of social life. READING 1HE ROMAN HOUSE 71
CONCLUSIONS
In summary then, this paper has argued that the house is primarily a
building and the purpose of a building is to order space. As an object of
material culture the house therefore “speaks” of the necessary intersection of presence and absence required for the reproduction of social life.
Presence and absence implies space which encloses and discloses the
body. These may in turn be zoned into front and back regions. Hence,
although space has no power to cause action, its relational order is significant It is a form of structuration (Giddens 1984), that is, a means by
which practices are reproduced diachronically. The house has been interpreted as a means of resolving the contradictory demands of Roman
social life in order to perpetuate it These meanings were drawn out of
the structuring of space by employing the metaphors of the disciplinary
institution and theatre and by the application of social theory. This study
represents a departure from the traditional practice, discussed above, of
bringing the ancient sources to bear on the archaeological remains. The
historical vision of Roman society does not represent the “truth” in light
of which material culture should be understood. Material culture is not a
“passive” reflection of the social order. Objects are the products of social
practices and hence embody within their very forms social meanings that
contributed to the replication of practice. Such meanings may not have
been the subject of comment in ancient texts. The material culture of the
Roman period has its own story to tell and we should therefore develop
theories and methods that allow us to interpret it We should not
subordinate it to a historical discourse.
ACKNOWLEDGEMENTS
I would like to thank Simon Keay, Ray Laurence, Stephen Shennan, and
Andrew Wallace-Hadrill for their comments on this or earlier drafts of
this paper. Needless to say they cannot be held responsible for any errors
or omissions.
NOTES
1. For a full exposition of the theoretical framework outlined in this and the
subsequent sections see Graharne 1995, 1-{)1.
2. Hillier and Hanson (1984) provide numerous indices that quantify these
aspects of space. There is not the space to discuss them here. For further
details, in addition to the above references, see Grahame 1995, 62-78; 1997. 3. Because this paper was written prior to the completion of my thesis, the conclusions presented here are only initial findings. For a full statement of
results and their meaning, see Grahame 1995.
BIBLIOGRAPHY
Boast, Robin and E. Yiannouli (eds) 1986. Archaeological Review from
Cambridge 5(2).
Bourdieu, Pierre 1977. Outline of a Theory of Practice. Cambridge: Cambridge
University Press.
Chapman, John 1990. Social inequality on Bulgarian tells and the Varna
problem. In R. Samson (ed.), The Social Archaeology of Houses, 49-92.
Edinburgh: Edinburgh University Press.
Clarke, John R. 1991. The Houses of Roman Italy 100 B.C. -AD. 250: Ritual,
Space and Decoration. Berkeley: University of California Press.
Donley-Reid, Linda W. 1990. A structuring structure: the Swahili house. In
S. Kent (ed.), Domestic Architecture and the Use of Space: an Interdisciplinary Cross-Cultural Study, 114-126. Cambridge: Cambridge
University Press.
Dwyer, Eugene 1991. The Pompeian atrium house in theory and in practice. In E. Gazda (ed.), Roman Art in the Private Sphere: New Perspectives
on the Architecture and Decor of the Domus, Villa and Insula, 25-48.
Michigan: University of Michigan Press.
Eschebach, Hans 1970. Die Stiidtebauliche Entwicklung Pompejis. Heidelberg:
(R5rnische Mitteilungen suppl. 17).
Fairclough, Graham 1992. Meaningful constructions – spatial and
functional analysis of medieval buildings. Antiquity 66, 348-366.
Fletcher, Roland 1977. Settlement studies (micro and semi-micro). In David
1. Clarke (ed.), Spatial Archaeology. London: Academic Press.
Foster, Sally. M. 1989a. Analysis of spatial patterns in buildings (access
analysis) as an insight into social structure: examples from the Scottish
Atlantic Iron Age. Antiquity 63, 40-50.
Foster, Sally. M. 1989b. Transformations in social space. The Iron Age of
Orkney and Caithness. Scottish Archaeological Review 6, 34-55.
Foucault, Michel 1977. Discipline and Punishment. Harmondsworth:
Penguin.
Foucault, Michel 1983. How is power exercised? In H. L. Dreyfus and P.
Rabinow (eds), Michel Foucault. Beyond Structuralism and Henneneutics.
Chicago: Chicago University Press.
Gazda, Elaine. K. (ed.) 1991. Roman Art in the Private Sphere: New
Perspectives on the Architecture and Decor of the Domus, Villa and Insula. READING TIIE ROMAN HOUSE
Michigan: University of Michigan Press.
Gibson, J. S. 1979. The Ecological Approach to Visual Perception. Boston:
Houghton-Mifflin.
Giddens, Anthony 1984. The Constitution of Society. Cambridge: Polity
Press.
Goffman, Erving 1959. The Presentation of Self in Everyday Life. Hannondsworth: Penguin.
Grahame, Mark 1995. The houses of Pompeii: space and social interaction.
Unpublished PhD thesis, University of Southampton.
Grahame, Mark 1997. Public and private in the Roman house: the spatial
order of the Casa del Fauno. In R. Laurence and A. Wallace-Hadrill
(eds), Domestic Space in the Roman World: Pompeii and Beyond, 137-164.
Portsmouth: Uournal of Roman Archaeology suppl. 22).
Grahame, Mark 1998. Material culture and Roman identity. In J. Berry and
R. Laurence (eds) , Cultural Identity in the Roman Empire, 156-178.
London: Routledge.
Harris, W. V. 1989. Ancient Literacy. Cambridge, Mass.: Harvard.
Harvey, David 1990. The Condition of Postmodernity: an Enquiry into the
Origins of Cultural Change. Oxford: Blackwell.
Hillier, Bill and Julienne Hanson 1984. The Social Logic of Space. Cambridge:
Cambridge University Press.
Hodder, Ian 1991. Reading the Past: Current Approaches to Interpretation in
Archaeology. 2nd ed. Cambridge: Cambridge University Press.
Laurence, Ray 1991. The social structure of the Roman city in Italy: the
logic of space. Unpublished PhD thesis. Newcastle.
Laurence, Ray 1994. Roman Pompeii: Space and Society. London: Routledge.
Laurence, Ray 1995. The organisation of space in Pompeii. In T. Cornell
and K. Lomas (ed.), Urban Society in Roman Italy, 63-78. London:
University College London Press.
Laurence, Ray and Andrew Wallace-Hadrill 1997. Domestic Space in the
Roman World: Pompeii and Beyond. Portsmouth: Uournal of Roman
Archaeology suppl. 22).
Ling, Roger 1993. German approaches to Pompeii. Journal of Roman
Archaeology 6,331-340.
van der Poel, Halstead B. 1984. Corpus Topographicum Pompeianum. vol. 3.
Austin, Texas: University of Texas Press.
Richardson, Lawrence, Jr 1988. Pompeii: an Architectural History. Baltimore:
Johns Hopkins.
Samson, Ross 1992. Knowledge, constraint, and power in inaction: the
defenseless medieval wall. Historical Archaeology 26, 26-44.
de Saussure, F. 1960. Course in General Linguistics. London: Peter Owen.
73 74 M. GRAHAME
Shanks, Michael and Tilley, Christopher 1987. Social Theory and Archaeology.
Cambridge: Polity Press.
Shanks, Michael and Christopher Tilley 1992. Re-Constructing Archaeology:
Theory and Practice. 2nd ed. London: Routledge.
Thebert, Yvon. 1987. Private life and domestic architecture in Roman
Africa. In Paul Veyne (ed.), A History of Private Life: from Pagan Rome to
Byzantium,315-409. Cambridge, Mass.: Harvard.
Wallace-Hadrill, Andrew 1988. The social structure of the Roman house.
Papers of the British School at Rome 56, 43-97.
Wallace-Hadrill, Andrew 1990. The social spread of Roman luxury:
sampling Pompeii and Herculaneum. Papers of the British School at Rome
58, 145-192.
Wallace-Hadrill, Andrew 1994. Houses and Society in Pompeii and
Herculaneum. Princeton, N. J.: Princeton University Press.
Werlen, Benno 1993. Society, Action and Space: an Alternative Human
Geography. London: Routledge.
Wolff, Janet 1992. The real city, the discursive city, the disappearing city:
post modernism and urban sociology. Theory and Society 24(4), 553-560.