ch a p ter 5

Seeking the domus behind the dominus in Roman

Pompeii: artefact distributions as evidence for the
various social groups

Bankers and tax collectors should have larger and more beautiful
houses, safe from burglars. Public figures and speakers should have
elegant and spacious accommodation to receive their visitors. For the
true aristocracy who hold office and magistracies, and who must take
on state roles, we must build high and stately anterooms, and very
spacious atria and peristyles, along with wide groves and walkways
completed in a majestic style; in addition we must build libraries,
galleries and basilicas fitted out with a magnificence similar to that of
public buildings.
Vitruvius, On Architecture 6.5.2–3

All members of a society are contributors to the matrix of actions

that eventually becomes the archaeological context, and the variety of
features is a product of this activity . . . many archaeological features
are diffuse, mixed and difficult to interpret. Some, though, are more
likely than others to provide information on small group activities;
among these are features with rather structured event and/or episode
sets, such as houses . . .
Brooks 1982, 68–69

introduction
In relative contrast with the houses discussed in the other chapters of
this volume, those from the Roman town of Pompeii have been intensively
studied for more than 200 years, and it is easy to see why they have attracted
so much attention. The Campania region of southern Italy in which they
lie was covered in volcanic material by the eruption of Vesuvius in August
79 ce. Today, excavation has revealed about two thirds of the entire town
of Pompeii, and it is possible to walk through the streets, wandering in and
out of both public buildings and private houses. Walls frequently rise at
least one storey and often more, and although the roofs and upper floors

89

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 90 Artefact distribution and social groups
have collapsed, the ground floors are frequently relatively complete. The
larger houses are adorned with wall paintings, mosaic floors and marble
furnishings. Excavators have been able to locate an extensive array of the
kinds of everyday items of pottery and metalwork normally present in
archaeological contexts, and they have also been able to learn much about
some of the fixtures and objects made of organic materials such as wood
which are not usually preserved on archaeological sites. The covering of
volcanic material at Pompeii is full of voids left behind as those materials
rotted away, enabling their original forms to be reconstructed by filling
those voids with plaster – a technique which has revealed not only items
of furniture such as cupboards and doors, but also the grisly impressions
of some of the unluckiest final inhabitants, animal and human, who failed
to escape the town’s destruction.
The earliest excavations in the houses here and at the neighbouring
town of Herculaneum were primarily concerned with recovering mosaics,
wall paintings and sculptures, and this, together with the richness of the
material itself, encouraged a tendency to study the different architectural
elements in isolation. But the Campanian evidence has also been used
differently, as a tool for studying social relationships in Roman households,
and this is the line of inquiry I want to focus on in this chapter, building
on what previous scholars have already achieved. In a landmark paper
first published in 1988, Andrew Wallace-Hadrill was one of the first to
recognise that ancient houses could be viewed as occupied spaces rather
than simply as architectural complexes (Wallace-Hadrill 1988). Since his
article was published a generation of scholars have developed and elaborated
on many of Wallace-Hadrill’s ideas. His discussion builds on the work of
earlier researchers who drew a connection between the layout of the larger
excavated houses at Pompeii and the kinds of elite residences described by
the architect Vitruvius (quoted above), who was writing in Rome during
the first century bce. Among the points Vitruvius makes, two in particular
stand out. First, he claims that the type of house a man lived in was
expected to be appropriate to his social status. A member of the elite who
played a role in public life was expected to receive clientes (followers of
lower social status, referred to loosely below as ‘clients’), business associates
and political supporters at his house, and he required a suitable space in
which to do this. Second, Vitruvius distinguishes between two different
areas of an elite house, one of which was the sole domain of the residents,
while the other could be entered by outsiders who had not specifically been
invited.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Introduction 91

Figure 5.1 Plan of the House of the Ceii (House i.6.15)

Assisted by additional ancient textual sources, Wallace-Hadrill sketched

a detailed model for the way in which Vitruvius’ description may have
worked at Pompeii, arguing that domestic space can be read as a physical
map of social status distinctions. There is much variation in the scale and
layout of the individual excavated properties, but the interiors tend to be
arranged around two circulation areas, an outer hall and an inner garden
(Figures 5.1 to 5.3). Like other scholars before him, Wallace-Hadrill viewed
the hall as what Vitruvius called an atrium, but he also looked in detail
at how it might have functioned as a space for the formal reception of
clients, whose numbers at their patron’s house for the morning salutatio
or greeting were a mark of status and influence. The architecture of the
atrium gave it an aura of wealth and power: the ceiling was high, and
the decoration and architecture conjured up the impression of a public,

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 92 Artefact distribution and social groups

Figure 5.2 Plan of House vi.16.26

rather than a private, building (Plate 5.1). At the rear it gave onto a sec-
ond room, traditionally referred to by scholars as the tablinum, which had
a lower ceiling and sometimes also a raised floor. Here, Wallace-Hadrill
suggested, the dominus (master of the house) would have been positioned
to receive clients in an almost theatrical manner, framed by the architec-
ture. At its rear the tablinum opened into the second circulation space, a
garden which often took the form of a colonnaded peristyle. This, Wallace-
Hadrill argued (again following Vitruvius), was a more intimate area into
which only higher-status associates would have been invited to stroll in
the porticoes or dine in decorated triclinia along its sides. He pointed
out that service facilities such as storerooms and cooking spaces, which
would have been used largely by slaves or servants, were pushed outwards
towards the boundaries of the domestic complex and away from this main
atrium–peristyle axis, thus marginalising their habitual users.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Introduction 93

Figure 5.3 Plan of the House of the Menander (House i.10.4)

This interpretation of the atrium house-form is elegant and has held up

well to scrutiny. It is, nevertheless, questionable whether approaching the
archaeology through the framework of the texts produces a reliable picture.
Several problems with this methodology have been pointed out. One is
that, while archaeologists customarily label excavated spaces with Latin
names, we cannot be sure that our definitions match those the Roman
authors had in mind. This is certainly a difficulty with most room labels,
although the identification of atrium and peristyle, on which Wallace-
Hadrill’s basic argument rests, is fairly secure, thanks to Vitruvius’ detailed

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 94 Artefact distribution and social groups

Plate 5.1 View into the atrium of the House of the Menander (House i.10.4), Pompeii

description of these spaces.1 In addition, the fact that the model maps out
a series of conceptual distinctions as well as spatial ones makes the precise
identification of functional spaces less crucial to the overall argument.
There is, however, a further consequence of using Vitruvius’ description as
a point of departure in this way that has been less widely acknowledged,
but is more troubling, and that is that his particular viewpoint has tended
to shape our whole understanding of elite Roman houses. His perspective –
that of the dominus – is so overwhelming that it is difficult to find evidence
anywhere in his text for the roles played by other members of the household,
1 The dangers of using textual references, from Vitruvius or from any other ancient author, to deduce
how rooms may have been used, have been discussed in detail (see Riggsby 1997; Allison 2004,
161–177). See Chapter 1 for further discussion of the potential problems in trying to use ancient texts
to understand the use of household space.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Introduction 95
including its women (compare Milnor 2005, 107). Of course, this bias is
part of a wider cultural and social universe in which elite male discourse
predominated. One strength of taking a material cultural approach to the
social dynamics of Pompeian houses should be that we can look beyond
that dominant discourse to explore alternative viewpoints and other social
roles. The few attempts that have been made to investigate the activities of
particular social groups such as women and slaves in these households have
had mixed success. Viewing the household through the lens of Vitruvius,
and hence from the perspective of the dominus, has led to the conclusion
that the influence of these groups was relatively unimportant since they
had little effect on the physical organisation of the central axis of the house
(for example Wallace-Hadrill 1996; George 1997). From this perspective
other members of the household are detectable only in the interstices of
the dominus’ world. But do we get a different view if we try to approach
the archaeological evidence directly?
A major study by Penelope Allison was aimed at clarifying the way in
which individual spaces in Pompeian houses were used (Allison 2004). A
particular strength of Allison’s work is that, as well as looking at the archi-
tecture and decoration of the houses, she also explored in detail the types
and distribution of the many objects found in different rooms. Where
sufficient information was available, she recorded where in the room each
item was located and how deeply it was buried, important details which
can reveal that objects may have been used or stored together, or whether
they may have fallen from an upstairs room as the upper storeys of houses
collapsed under the weight of accumulating volcanic material. Excavators
must have missed some items and been selective about what they recorded,
so that this kind of database can offer only a partial picture of what the
houses originally contained. Even so, putting together a comprehensive list
is a daunting task: in Allison’s sample of thirty houses there were 865 rooms
which together contained over 16,000 artefacts. Many problems and incon-
sistencies in the material emerged which are routinely overlooked: some
houses showed evidence that redecoration or construction work was under
way, hindering any attempt to understand how those properties functioned
as living spaces; elsewhere, rooms with elaborate wall decoration contained
mundane domestic artefacts, or no artefacts at all; and many spaces con-
tained motley assortments of items with a wide range of different uses.
For those accustomed to Wallace-Hadrill’s dominus-centred model, Alli-
son’s study revealed some interesting results: for example, the atrium was
shown to be not only a monumental reception space for the use of the
dominus and his visitors, but also a location where a variety of domes-
tic items such as loom weights and table vessels were kept, revealing the

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 96 Artefact distribution and social groups
presence of other members of the household here as well. Ultimately, how-
ever, the volume and diversity of information resulting from this study are
difficult to digest, and Allison’s complete renunciation of textual evidence
made questions about patterns of social behaviour impossible for her to
address (Allison 2004, 154–158).
A question that remains, then, is whether it is possible to use the material
from Pompeii to build a model of domestic organisation which gives more
weight to the other members of the household besides the dominus, and
which combines the evidence of texts and archaeology in a more balanced
manner. Recent studies in the archaeology of other periods and regions
have pointed out that the best place to search for the influence of less
empowered groups like women, servants and children is in the kind of
artefactual material Allison has brought together, since this is evidence for
the kinds of small-scale, short-term activities over which such individuals
are likely to have had most control (Gilchrist 2000, 325–326). In both
Greek and Roman contexts the focus of most artefact analysis has been on
isolating the functions of different rooms, but to date studies of artefact
distributions have often been seen as frustratingly inconclusive (for example
Foxhall 2000; Cahill 2002, 70–72). Even where both architecture and finds
have been well preserved and relatively well documented, as at Pompeii,
the task of pinpointing a single dominant activity in a specific architectural
space has proved difficult, and this has led some scholars to suggest that
the individual room is too small as a unit of analysis (Berry 1997, 194),
while others have questioned whether much at all can be learned about
Roman society from studying artefact distributions (see for example the
contributions in Vanhaverbeke et al. 2008).
One response to this problem has been a more critical approach to the
natural and human processes which have shaped the archaeological site
since the houses were left by their occupants (known as site formation
processes). Even at a site such as Pompeii, which was rapidly abandoned,
we cannot expect the distribution of artefacts to be unaffected by the cir-
cumstances surrounding the site’s destruction. In fact detailed study of
the architecture has revealed extensive evidence for re-entry to buildings
during or after the eruption. Holes in many walls reveal escape routes or
the passage of intruders, and messages scrawled on house exteriors indi-
cate the activities of rescue parties or looters.2 In addition, for some years
before the final eruption in 79 ce there were earthquakes, including a

2 For example CIL iv 2311, on the exterior of house vii.2.20, which reads ‘House tunnelled through’:
Cooley and Cooley 2004, 40.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Flexible and multi-functional domestic space 97
serious one in 62 ce which is known from preserved textual sources to have
caused severe damage to public buildings and presumably damaged private
houses as well. It has often been argued that earthquakes led to the repair,
modification and even the abandonment of some houses prior to the final
eruption, although Allison’s analysis suggested that at least some of the
properties in her sample were still inhabited at the time of the destruc-
tion (Allison 2004, 192–196). Whenever the houses were abandoned, it is
likely that their inhabitants would have had an opportunity to store or
to carry away some prized possessions, and we should not discount the
potentially disruptive influence of these kinds of activities. The effect of
the eruption in distorting (rather than reducing) the artefact assemblages
should therefore not be underestimated, and study of a variety of houses
at other ancient sites has found similar levels of apparent disorganisation,
even though the circumstances of their abandonment were different. As
with domestic structures whose occupants may have departed for other
reasons and sometimes less hastily, some possessions may have been carried
away from Pompeii by its fleeing population while others may have been
gathered together or hoarded in the hope of an eventual return. Before
using artefact distributions as evidence for patterns of activity, it is there-
fore important to ask whether we can make more sense of the objects that
were found and recorded by thinking systematically and in detail about
these and some of the other processes which may have structured their
deposition.

flexibility and multifunctionality in the

use of domestic space at pompeii
The common focus on the role of the dominus in shaping the domestic
environment at Pompeii has tended to emphasise a number of aspects
of elite domestic culture which appear to be uniquely ‘Roman’. Indeed
it has been explicitly argued that the architecture of Pompeian houses,
and others outside Italy in a number of the provinces, deliberately pro-
jected an image of the ‘Romanitas’ or ‘Roman-ness’ of their owners (Hales
2003, passim). In a variety of practical ways, however, Roman housing
from Pompeii and elsewhere has much in common not only with that of
Classical Greece but also with dwellings from a range of pre-modern and
non-western societies which have been studied through archaeological, his-
torical and ethnographic work. Such studies suggest that the present-day
western concern for providing specialised rooms for particular activities
and for allocating separate spaces to different members of the household

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 98 Artefact distribution and social groups
is a relatively recent phenomenon which has developed as a consequence
of the rise of ‘individualism’ in western thought. In most societies, at most
times, a single space has tended to be multifunctional and to be used by
a number of different household members. At the same time, analysis of
significant numbers of Latin, and also Greek, texts making reference to
domestic activities suggests that in both Greek and Roman societies space
was perceived as potentially flexible, its use changing dependent on a vari-
ety of factors. Among these seasonality was particularly significant since
changes in sunlight and temperature had implications for how warm or
cool rooms facing in different directions could be (Nevett 1997, Nevett
1999, 36). These characteristics of ancient domestic behaviour have been
acknowledged by classical archaeologists looking at artefact distributions in
both Greek and Roman contexts (for example Berry 1997; Foxhall 2000).
Nevertheless, the extent, nature and implications of flexibility in spatial
use in ancient houses have remained largely unexplored.
Here, then, my goal is to investigate evidence for flexibility in the use
of space at Pompeii by identifying systematic patterns of variation which
may underlie artefact distributions. These are used as a basis for detecting
some of the relatively small-scale, short-term patterns of domestic activity
which should tell us most about the less privileged members of these
households. In the process I highlight some of the assumptions that need
to be made about how artefact distributions are likely to relate to patterns
of behaviour, and I make explicit use of some of the models for interpreting
these relationships outlined by the anthropologist Michael Schiffer based
on ethnographic studies (Schiffer 1996). Although the artefacts are the main
focus, they are viewed as part of a wider complex of information which also
includes the provenience (the location in which they were found) together
with any architectural and decorative features which help to provide a
context.
Three different aspects of the evidence potentially show multifunc-
tional or flexible use of space: the first, and most commonly observed
phenomenon in both Roman and Greek contexts, is the occurrence of
artefacts and/or architectural features used for two or more different activ-
ities in the same physical area; this is balanced by a second, closely linked,
phenomenon, more rarely discussed, which is that evidence for the same
activity is sometimes found in two or more different areas and may indicate
flexibility in spatial usage; a third and final pattern merits separate consid-
eration, and that is an apparent incompatibility in the activities suggested
by the architecture and decoration of a space, and the objects found there.
These configurations result from the kinds of small-scale activities carried

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Flexible and multi-functional domestic space 99
out by individuals or small groups over a short period of time (Brooks 1982,
68). As examples through which to explore these indicators of the flexible
use of space, I discuss in detail three specific atrium-type houses, the House
of the Ceii (i.6.15), House vi.16.26 and the House of the Menander (i.10.4).
In all three cases the excavators not only recorded many of the finds but
also attempted to distinguish items that were in storage from those that
may have been in use at the time of the destruction.3 Although they are all
atrium-type buildings they represent a range of different sizes, with ground
areas of roughly 300, 600 and 1,800 square metres respectively. While these
houses obviously do not constitute a statistically significant sample for
generalising about the use of domestic space across the town of Pompeii
as a whole, they do provide insights into multifunctionality and flexibility
of spatial use, and there are some similarities between them which may be
indicative of broader patterns.
In the House of the Ceii (named by modern scholars after two brothers
whom they thought once occupied it) the major phase of construction is
dated to the second half of the second century bce, although the house
incorporates earlier walls and also shows signs of some later modification
(Figure 5.1). Eleven rooms are arranged around an atrium (b) at the front
and a small garden at the rear (o) (Figure 5.1). Holes dug through the walls
in several places seem to indicate the movements of escapees or looters
at the time of the eruption. Allison comments that the assemblage is
very mixed, containing relatively few movable finds and much evidence of
empty storage cupboards, all of which she suggests are the result of events
during the abandonment of the town, when items were carried away or
salvaged (Allison, date unknown).
House vi.16.26 consists of twenty-two rooms organised around an atrium
(b) and a compact peristyle (Figure 5.2). The atrium appears to have had
an unusual design, with a large open area to the south-east, and porticoes
to north and west. The peristyle has a deep ambulatory to its north side
and shallower ones to its east and west, of which the western one was
walled in to create a corridor. On the fourth, southern, side the central
garden space is delimited by a wall, but engaged columns create the visual
3 I build directly here on the information gathered and summarised by Allison, whose work provides
the best picture we have of the artefactual contents of Pompeian houses and their spatial distribution.
However, I do occasionally differ from her in interpreting the use to which an object may have been
put (for example, I take small glass bottles/flasks to be toilet or table items, rather than transport
vessels). Assigning functions to specific objects is not a scientific process since they may not always
have been used for the purpose for which they were made, or indeed may have been made to cater
to a variety of uses. Assessments of whether an item was being used, stored or discarded are also
subjective, depending on interpretation of contextual information.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 100 Artefact distribution and social groups
effect of a full peristyle. Contrary to the stereotypical view, in this house the
atrium and peristyle lie side-by-side rather than one behind the other. Also,
somewhat atypically, there is no evidence for elaborate wall painting. Allison
suggests that the area around the peristyle may have been abandoned since
it appears to have yielded few finds. A pile of roofing tiles in the atrium
suggests that renovation work may have been planned or under way here
at the time of the eruption. Moving between these two areas would have
necessitated passing through one of the rooms facing onto the peristyle,
since the large room l, which is reminiscent of a transitional tablinum
space, seems to have been walled in at the back with only a window
looking out towards the peristyle. Both northern and southern areas of the
building had their own street entrances, and it is therefore a possibility
that the two were occupied separately, at least for part of the life of the
house.
The well-known House of the Menander (named by modern scholars
after a painting of the famous fourth-century Greek comic playwright on
the wall of one of its rooms) is unusually complex in its pattern of organisa-
tion (Figure 5.3), owing to its large size, but its relatively recent excavation
means that it provides particularly detailed information about movable
finds. Like the House of the Ceii, in-depth study of the architecture sug-
gests that this house developed into its present form over a period of several
centuries. It also grew considerably, incorporating parts of neighbouring
properties (Ling 1997, 52–59, 223–237). The original nucleus of the house,
the atrium (b) and surrounding rooms, perhaps with a garden behind, has
been dated to the late third to mid second century bce. By the time of
the destruction the property comprised some fifty different rooms, includ-
ing additional courtyard and garden spaces to east, south-east and west of
the main atrium–peristyle axis (b–c), a second atrium to the east of the
peristyle (41), and a bath suite along with four basement rooms on the west
side (46–49). In his published summary of the architecture and history of
the insula Ling commented that there was extensive evidence of earthquake
damage (Ling 1997, 234–237). Redecoration was evidently taking place in
several rooms, and work seems to have been under way to modernise the
furnace in the bath suite (Ling 1997, 135). There were, nonetheless, large
numbers of artefacts in the house at the time of the destruction, suggesting
that any interruption in occupation was not anticipated to be lengthy. In
fact during excavation of the corridor south-east of the peristyle, skele-
tal remains were recovered from a group of at least nine individuals who
included three children under five (Lazer 1997). It is unclear whether the
owner of the house and his family were in residence at the time of the

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Flexible and multi-functional domestic space 101
eruption, whether these were slaves or servants taking care of the property
while it was renovated or whether they were outsiders seeking refuge.

Evidence for two or more activities taking place in a single space, and
for the use of several different spaces for the same activity
The characteristic mixing of material associated with a variety of activities
taking place within a single architectural space is widespread in all three
houses. As the work of Michael Schiffer suggests (Schiffer 1996), a first
step in interpreting the artefactual evidence is to try to establish, as far
as possible, whether individual items were left behind where they were
actually used, or whether they were in storage or had been set aside for
discard or recycling.
Refuse is the category of evidence most difficult to identify for a variety
of reasons, not least because definitions of rubbish are subjective. Today
an item of rubbish could be seen as something no longer needed or
wanted by its owner. In pre-modern societies, though, rates of discard
were much lower. This is partly because in such societies people had fewer
possessions – particularly disposable ones (Deetz 1977, 59). But it is also
because a scarcity of resources meant that items tended to be repaired,
reused for other purposes or recycled to make new objects. Evidence of
this in Greek and Roman contexts can be seen in the number of excavated
ceramic vessels which have been mended in Antiquity. A couple of criteria
can be used to define refuse: first, its context, which may suggest that
objects had been gathered together to be discarded or that they had already
entered a rubbish deposit; and second, its condition, which may have been
damaged or fragmentary at the time it was deposited (although it should
be borne in mind that ethnographic work has shown that broken items are
sometimes reused for a different purpose: Schiffer 1996, 30–32).
At sites such as Pompeii, where many structures were excavated a long
time ago and there was a wealth of architectural and other information
to be recorded, fragmentary items and rubbish deposits have often tended
to be overlooked or given low priority. Nevertheless, given the likelihood
of the preferential recording of valuable items, one practice to look for
is the hoarding of discarded items, which in societies where resources
are scarce are often kept as potential raw materials for future reuse – a
phenomenon that Schiffer refers to as ‘clutter refuse’ (Schiffer 1996, 66).
The fact that such pieces should be categorised as discarded, rather than in
use, is not necessarily always obvious because identification relies on precise
and accurate information about their context and on a judgement about

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 102 Artefact distribution and social groups
whether or not any breakages occurred before the eruption. Nevertheless,
there are a few instances in which the inventories of our example houses
seem to include examples of this kind of clutter refuse.
In various locations in the House of the Menander and the House of the
Ceii fragments of marble sculpture were found, apparently already broken
before the eruption. Allison suggests that ‘unless their provenances were the
result of post-eruption disturbance, they are . . . evidence of pre-eruption
dislocation and disturbance’ (Allison 2004, 39). But this is not always borne
out by their contexts. For example, in the secondary atrium of the House
of the Menander (room 41) a marble arm seems to have been stored on a
shelf along with a variety of other objects. This location, together with the
fact that the room seems to have been a service area (see below), implies
that the arm had perhaps been here for a while, possibly salvaged at some
point by the users of this area for reuse or sale. A similar case in the House
of the Ceii involves two broken pieces of marble sculpture found in room
i, which seems to have been temporarily disused at the time of the eruption
and might therefore have been particularly suitable for keeping such clut-
ter refuse. A complete Egyptian-style marble figurine found in the fill of
room z of House vi.16.26 may also fall into this category. The association
between marble objects, often in fragmentary state, and cooking areas is
in fact more widespread in Allison’s sample. For example, the House of
the Vettii, another large, well-decorated house, also contained fragmentary
sculpture in hearth room w. The presence of such objects in these rooms
reinforces the suggestion that they were being hoarded as a potentially valu-
able resource, perhaps having been scavenged by lower-status individuals
working here.
A different form of recycling can be documented in relation to some
of the large amphorae which were used to transport and store foodstuffs.
Considerable numbers of these vessels were found in the House of the
Menander, some bearing dipinti (painted lettering) which noted their
contents. In one case a vessel whose original label stated that it contained
local wine had been relabelled indicating that it had been refilled with
vinegar. Another group of wine amphorae found in the secondary atrium
on the east side of the house contained stucco, lime used for plastering, and
signinum powder used for making floors (Stefani 2003b, 210–211). Even
broken amphorae could be useful: Allison suggests that amphora toes were
used to support round-bottomed cooking pots (Allison 2004, 101).
In addition to clutter refuse and recycling, there are also a couple of
instances where excavators do seem to have been dealing with, and to have
recorded, deposits of discarded refuse. These are a cistern beneath the floor

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Flexible and multi-functional domestic space 103
in the House of the Menander corridor l, and a walled-up ground floor
space in the House of the Ceii (n). These structures contained a number
of fragments of fine terra sigillata pottery and other ceramic vessels, as
well as sherds from lamps and glassware, all of which had presumably been
broken in use, cleared up, and discarded here. In the House of the Menander
these were combined with plant matter, and even a coin and four finger
rings.4
By contrast with the recycling and disposal of rubbish, which we can
only glimpse occasionally in the recorded evidence, storage is the most
widely represented activity: in the House of the Ceii, for example, evidence
for storage of one sort or another was noted in six of the thirteen spaces.
The atrium alone housed at least three separate storage structures: a set of
shelves under the stairs contained a mixed assemblage including tableware
and lamps; a cupboard in the south-east corner, preserved as a plaster cast,
held toilet articles, while a chest in the north-east corner contained a single
cup. A further concentration of stored items was located in room g, which
was lined with shelves and cupboards or chests. Table vessels were found
in this area, too, alongside iron tools and a casket for valuables. The varied
collection of items in this room is in keeping with Schiffer’s cross-cultural
observation that storage areas tend to be cleared out relatively rarely, and
that they therefore accumulate mixed assortments of possessions (Schiffer
1996, 68–69). In cases like the atrium of the House of the Ceii, such
stored assemblages can account for a particularly wide range of objects
with different uses occurring together in a single room. Storage patterns
are not completely random, however: within the atrium a distinction seems
to have been made between tableware and toilet items, and across the house
as a whole there is some degree of specialisation in terms of what is kept
where: tools are stored only in room g and toilet items only in the atrium,
while tableware is kept in the atrium and room g. A similar pattern occurs
in House vi.16.26, where evidence for food storage is found only in the
atrium and peristyle, storage of personal items is located only in rooms e
and k, and evidence for storage of table-ware is limited to rooms l and f.
4 Room n in the House of the Ceii may have been linked with a toilet in an upper storey and was
excavated in 1983 connection with the publication of the house (Michel 1990, 63). House of the
Menander corridor l is referred to by Ling as room P: Ling and Clarke 1997, 274. The presence here
of whole vessels as well as fragments led Allison to suggest that it was not a rubbish deposit (Allison,
date unknown). Whole vessels are sometimes found in wells and cisterns where they have been lost
in the process of drawing water. The inclusion of a bowl full of hazelnuts is admittedly curious,
but the amount of fragmentary material suggests that some of this deposit, at least, does constitute
refuse. It is perhaps possible that some of the other items were left behind during the construction
of the cistern or, as Allison suggests elsewhere, that it was some sort of foundation deposit (Allison
2004, 130).

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 104 Artefact distribution and social groups
The House of the Menander echoes these distinctions on a larger scale,
although tableware and personal items are recorded only as being stored in
a small number of rooms. An exceptional find came from the basement area
on the west side of the house, where the remains of a wooden chest were
found to contain 118 pieces of silver tableware which had been carefully
wrapped for storage in heavy cloth and were hence very well preserved.
In the chest with them was found a smaller wooden box containing some
gold jewellery and gold and silver coins, suggesting that this was a hoard
that had been hidden away with the hope of recovering it at a later date.5
A group of bronze table vessels were found in a second chest on the other
side of the room and additional bronze vessels seem to have been placed
directly on the floor of the room.
There are a number of locations in each house in which evidence was
recorded of storage structures but where only a few items are noted as
having been stored, and sometimes there is no note of any stored item
being recovered at all (for example, in rooms e and l in the House of the
Ceii).6 If we could safely assume that everything once kept in these houses
is likely to have been preserved, excavated and recorded, then the amount
of unused capacity would suggest that the occupants took away many of
their possessions. But such an assumption is unwarranted: the relatively
small assemblages of ceramic food preparation-, table- and cooking-wares
recovered from all three properties suggest the possibility that some of the
more mundane finds in these categories may not have been noted by the
excavators if they were in fragmentary condition. (In the House of the Ceii,
for instance, fewer than ten small, open vessels which could have been used
for drinking are recorded from the entire ground floor of the house.) Finds
of a limited number of complete vessels, together with the contents of the
refuse deposits in the House of the Menander and the House of the Ceii
mentioned above, reinforce the evidence for the use of fine terra sigillata
pottery and other ceramic vessels in the two houses, and the sherds of such
vessels, broken as the city was destroyed, may often have been disregarded
by the excavators. It is also relevant to think about what kinds of organic
materials are likely to have been in use in the house but will not have
survived at all, since these may account for at least some of this empty

5 The latest coins date to 78 or 79 ce suggesting that they were placed here at the time of the eruption
or shortly beforehand, although this does not necessarily mean that the silver, which was on the
bottom of the chest, had not been there for longer (Painter 2001, 12).
6 These are identifiable as niches, remnants of fitments and fittings, or as casts, although items recorded
as door fittings are excluded from my consideration on the basis that they could indicate room doors
rather than pieces of furniture. My selection therefore represents a minimum number.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Flexible and multi-functional domestic space 105
space. A major category of bulky item rarely recovered at Pompeii is cloth:
woollen blankets would have been required for night-time use in winter,
and these are likely to have been stored away at the time of the eruption in
August, perhaps along with extra layers of warm clothing also for winter
use.
A second type of organic material which may be under-represented in
some of the houses is basic foodstuffs, both liquid, such as wine and olive
oil, and solid, such as flour. There is evidence for large amounts of storage
and/or consumption of liquids in the House of the Menander, where
numerous amphorae were found in a variety of different areas including
the stable-courtyard (34), rooms off the peristyle and the basement under
the bath suite. Many of the shapes are those of wine amphorae and they
show that the vessels had once contained both local wines and also imports
from Crete and Rhodes. The dipinti carried by some of them specified
that they also held other products including olive oil, honey and Spanish
liquamen (a form of garum or fish sauce) (Stefani 2003b, 210–211). Liquids
could also have been kept in smaller vessels, like the coarse ware jug, 25 cm
tall, found in the secondary atrium and labelled with a dipinto giving its
contents as vinegar (Fergola 2003, 165–166). These containers offer some
indication of the variety of foodstuffs which might have been consumed in
a household such as this, although the evidence for recycling, noted above,
makes it unclear how many full ones would have been kept on hand at any
one time. It is possible that at the time of the destruction some of them were
clutter refuse. A likely example is a group of forty-three amphorae which
were stacked together against the wall in a corner of the stable-courtyard
in such a way that most would have been inaccessible had anyone wanted
to get at their contents.
The household must have prepared and consumed staples such as grain,
since several hand querns for processing it were found in the house. But this
is likely to have been supplemented by pre-prepared foods, such as bread
from local bakeries which were equipped for large-scale production. We
currently know little about general consumption patterns and the balance
between home-produced and bought-in products is likely to have varied
between households. Direct evidence for storage of dry goods in the house
is lacking, and these were perhaps kept in sacks or cloth bags which were
not preserved. Comparable evidence for storage of foodstuffs is also present
in House vi.16.26 in the form of amphorae and large dolia or storage jars,
which were kept in the peristyle. In the House of the Ceii three amphorae
were found in room i, and smaller vessels and/or dry foods in sacks or
bags could have occupied some of the empty cupboard space in this room.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 106 Artefact distribution and social groups
The smaller size of this house is likely to have meant that there were fewer
people here to feed, and more limited space to accumulate clutter refuse
such as empty amphorae, although more amphorae were present amongst
material found above floor level in room d, which may have fallen from
upper-storey rooms.
To some extent, then, recycling and storage patterns help to explain some
of the mixture of activities apparently attested in many rooms. But this
raises a further question: where were these stored items used? Of course,
we cannot answer this for sure, but it is possible to make some suggestions
based on the proximity to each other of items used for associated tasks and
on the distribution of items which seem to have been left out in rooms,
rather than stored away.7 An example are the related groups of objects
associated with storing, preparing, cooking and consuming food and drink
already mentioned, which are relatively straightforward to identify even
if it is not always possible to be precise about the stage (or stages) they
facilitated in the journey of food and drink from store to table. Room
i in the House of the Ceii, where the three amphorae were found, was
also furnished with a fixed hearth which was probably used for cooking.
In the House of the Menander a room in a similar location and with a
similar structure (room 3) may also have been used this way. It is easy to
see that, given the size and location of these rooms, the atrium was the
convenient place to store bulk foods that could not be accommodated in
the hearth room itself. A similar arrangement seems to have prevailed in
House vi.16.26, where a hearth room (z) was located at the rear, opening
onto the peristyle. An additional hearth which may also have been used for
cooking was found in nearby room t.
A mystery is the lack of cooking pottery in or close to any of these
hearth rooms: this could reflect selectivity on the part of the excavators,
but, as Allison suggests, it could also indicate that none was actually being
used for cooking immediately prior to the eruption. If any of the three
houses was still occupied at that time, then cooking may have taken place
elsewhere in the building. The House of the Ceii was equipped with a
bronze brazier found in room l (off the garden) which could have been
used for this purpose. Seasonal variation in day-length and temperature
7 It is important to remember that the assemblages noted in the excavation reports as being in storage
are likely to represent a minimum: at least some of the objects whose contexts are not noted or which
were thought to have come from the floor may actually have been in or on pieces of furniture, the
presence of which was not detected or noted. I am excluding from consideration objects noted by
Allison as likely to have come from upper storeys: while these are important evidence of the kinds
of activities which may have been carried out in upstairs rooms, their lack of context makes them
more difficult to use for the questions I am trying to address here.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Flexible and multi-functional domestic space 107
must have been a major influence on patterns of activity, even in the largest,
wealthiest houses. In recent years mean temperatures in Campania have
varied between a minimum of close to 0◦ C in January and a high of almost
30◦ C in July and August, and individual days can be somewhat hotter (or
colder). The range in Antiquity is likely to have been similar. In the absence
of effective artificial heating and lighting, the most efficient strategy for
coping with this kind of variation would have been to use different rooms
at different times of day and during different seasons. This would have
enabled the occupants of the house to take advantage of the warmth and
light provided by a low sun in winter, or exclude the hot sunlight and
capture breezes in summer.
The importance of this kind of seasonality in both Greek and Roman
Antiquity is supported by a variety of references made in the ancient sources
to the need for adapting architectural design to take sunlight into account
at different times of year (for example, Varro, On the Latin Language
8.29.4 and Cicero, Letters to his Brother Quintus 3.1.2). Since Pompeii was
destroyed in August, it seems possible that hearth rooms were used more
in winter months, when the heat would have been an asset, whereas during
the summer food was cooked on braziers which could be carried to the
best-ventilated locations and would not have overheated the interior. In the
summer months, then, light cooking which did not require lighting a whole
oven may have been the norm. Such meals could have been supplemented
by baked goods bought from one of the many commercial bakeries in the
town. The evidence from the House of the Menander suggests that this
pattern may have been followed even in large houses where slaves may have
been able to prepare food well away from the apartments occupied by the
householder.
In the House of the Menander a decorative bronze brazier with lion feet
and lion head fittings was found in the peristyle itself, and a plainer iron one
in an alcove on its south wall. But the situation here was more complicated.
As well as the front hearth room (3), several other spaces in the house were
furnished with similar built-in hearth features (including rooms 20, 34a,
41, 45, 52 and 54). There has been some debate about how many were used
for cooking and whether others may have had heating or craft purposes,
but it may not make sense to try to separate these different uses too rigidly:
a single hearth could potentially have been used in different ways. In most
cases there was little trace of any cooking or food-preparation equipment
associated with these structures, suggesting that they, also, may not have
been in use in August 79 ce. Again, it is possible that during the colder
winter months several of the ovens/furnaces operated at once to warm

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 108 Artefact distribution and social groups
different parts of the structure, and perhaps to cook food for different
groups of occupants. Allison raises the possibility that the side atrium (41)
and surrounding rooms may have been occupied by a household separate
from the one living in main part of the house, although the overall pattern
of organisation suggests that at one time, at least, the whole house did
function as a single residential unit (Ling 1997, 105–118). Alongside the
core atrium–peristyle axis was the bath area to the west, together with
a hearth room which, during the winter months, may have served those
who worked here. To the east, further rooms are arranged around the
secondary atrium (41), the stable-courtyard (34), and the corridor which
connects the two (l). These make sense as service quarters: the secondary
atrium and surrounding rooms contained a variety of different artefacts,
some, such as a decorated couch, seemingly quite expensive. But these were
combined with utilitarian household items, suggesting that a few choice
pieces may have been scrounged from the main part of the house when
no longer required and hoarded by those whose lives were centred in these
rooms. The stable-courtyard in this area, with its cart, was perhaps used
for provisioning a large household, although Allison suggests that it may
have fallen into disuse prior to the eruption.
How do patterns of consumption relate to this picture of food prepara-
tion? Vessels will be found where they were used, giving an indication of
dining areas, only if the dishes from the last meal taken in a space were
left behind and not cleared away. This means that in order to get a fuller
picture it is helpful to look at how the distribution of any serving vessels
from the house floors may relate to those in storage. In the House of the
Ceii tableware stored both at the front of the house in the atrium and
at the rear, in room g, is likely to indicate patterns of dining rather than
food preparation: the shapes of the majority of vessels suggest they were
used for drinking, and they are therefore likely to have been filled where
dining was taking place. The binary pattern of storage suggests that, in
common with cooking, dining may also have moved seasonally, or even
at different times of the day, between the front part of the house and
the rear.
In large houses like the House of the Menander, rooms with comparable
dimensions which could have been used for dining are duplicated, with
examples leading off both atrium and peristyle, facing in different direc-
tions, and varying in the extent to which they are open to the outside. In
winter, those facing south across an open peristyle would have been lighted
and warmed by the rays of the low sun for much of the day, while others on

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Flexible and multi-functional domestic space 109
the north side would have avoided the sun and remained cooler for summer
use. Similarly, at midday during the summer the atrium may have acted as
a reservoir for cool air, shaded and with the double height ceiling allowing
hot air to rise. Rooms leading off here may have been cooler at this time
of day than those giving directly onto the peristyle. Vitruvius specifically
recommended using different rooms for dining at different times of the
year.8 A similar consideration probably also lies behind some of the more
uncommon names applied to domestic spaces mentioned in the texts: for
example, hibernacula (literally, little winter places), perhaps easier to heat
than larger ones which may have been used in summer. Such a wide range
of alternative venues would not have been available in smaller houses like
the House of the Ceii; nevertheless, even the occupants of this house may
still have had some choice of locations with different thermal properties
which they could have used for dining and other activities. Decorated
rooms (c) and (e), which opened off the atrium, perhaps provided warm,
sheltered spaces in winter; a further, smaller, decorated room (f ) faced west
onto the ambulatory and may have received more ventilation from the
garden; a larger, undecorated, room (d) faced north onto the ambulatory
and looked out over the garden beyond (o) through a wide opening.
In the House of the Menander evidence for consumption in the form of
serving vessels is widespread. Only in two rooms are caches of such items
noted specifically as being in storage, as opposed to in usage contexts. One
of these is the set of silverware and the bronze vessels from the basement
mentioned earlier. The other consists of two groups of serving vessels of
various types in glass, ceramic and bronze, which were stored on a shelf
and inside a chest in the secondary atrium (41), along with various other
kinds of objects such as toilet articles. A mixture of cooking pots and
further serving vessels was recovered from the floor of the same room and
may indicate the use of such vessels there. Serving vessels which appear to
have been found where they were used are also noted from a number of
other locations in the house. Of the rooms in the atrium complex facing
either inwards into the atrium itself, or south into the peristyle, half had
at least one such vessel: mostly these were small, isolated ceramic cups,
plates or lids, although room 2 held a cache of sixteen ceramic dishes.
A couple of bronze vessels also came from this area, but the amount of
pottery is striking in comparison with what is recorded from the other
two houses. Rather than representing a real difference, this may simply

8 Vitruvius, On Architecture 6.4.1–2.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 110 Artefact distribution and social groups
suggest that greater attention was paid to recording such items on the part
of the excavators of this house. It also seems surprising in the light of the
concentration of bronze vessels which, although they might be assumed to
be more valuable, were found in the secondary atrium (41), an area which
may have been associated with food preparation by slaves and servants.
One potential explanation for this apparent anomaly is that the cache of
silver and bronze in the basement represents a collection of more valuable
table items gathered up from the rooms around the atrium and peristyle,
where they may have been used instead of, or alongside, pottery. In his
publication of the silver plate, Kenneth Painter argues that the collection
represented eight place settings, and points out that this number is suited
to formal dining outdoors, rather than to indoor banquets, where couches
were normally set up to hold groups of six or nine diners. Nevertheless, he
also raises the possibility that there was a special dining set for the dominus
himself, which may have included vessels of gold and/or precious stones,
and which was not stored along with the silver (Painter 2001, 40–41). If
this were the case, then a set of eight place settings to complement it would
make perfect sense for dining indoors.
The small numbers and wide distribution of the ceramic items suggest
the possibility that some may have been the residue of food and drink
taken in different rooms and not cleared away before the abandonment
of the house. (Comparable untidiness before the abandonment of a house
has been observed in other cultural contexts: Schiffer 1996, 97–98.) Their
widespread occurrence raises a question about modes of consumption in
such large, wealthy households outside the formal banquet or cena. If they
were resident at the time of the eruption, is it possible that the owner
and his family may have eaten separately or in small groups, rather than
together? Or were the dishes left here by caretakers or workmen taking
refreshments as they redecorated or cleaned the rooms in which they were
found? Again, our knowledge of the context and condition of these items,
and of what may originally have lain alongside them, is not sufficient to
justify a conclusion, but this is a question that might be addressed through
future research into more informal patterns of consumption.

Cases where the evidence appears to indicate incompatible activities

taking place in a single space
Another phenomenon which creates an appearance of disorder is the appar-
ent incompatibility between different elements of the architecture of a space
and/or the objects found in it. This can take different forms, and there

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Flexible and multi-functional domestic space 111
are therefore a number of potential explanations. In general terms, dis-
cussion of this phenomenon has often been focused on the effects of the
eruption and preceding seismic activity, associating these events with struc-
tural damage and with consequent changes in patterns of use. While the
consequences of such events must, of course, be considered, there are also
other reasons why we might perceive the evidence as contradictory, some
of which result from our own cultural expectations about what activities
may share a single space, while others are linked to the way in which both
the houses themselves, and the composition of the households occupying
them, changed through time.
Perhaps the most commonly cited instance of this kind of contradic-
tory evidence is where rooms whose décor suggests that they may have
been intended as reception or living rooms contain items which are more
mundane and functional. A striking example is the atrium. Superficially, if
these spaces are viewed from the Vitruvian perspective as reception areas
designed to fit the needs of the dominus, then the discovery of a piece of
storage furniture together with an amphora in the atrium of the House
of the Menander, alongside its carefully executed wall paintings, seems
anomalous. Similarly anomalous are the three different storage features in
the painted atrium of the House of the Ceii. As argued earlier in this chap-
ter, however, storage was a major feature of these houses, and the presence
of this and other forms of domestic activity pervades the atrium–peristyle
axis just as much as the outer or rear ‘service’ areas. There must have been
a practical advantage, in that the wide space of the atrium was suitable
for accommodating large pieces of furniture and storage vessels. This must
have been even truer of the peristyle and garden spaces, which were the
only possible locations for keeping even larger storage vessels like the dolia
in the peristyle of House vi.16.26. Storage in such visible locations might
have been a deterrent against pilfering of food. The atrium was used for
storage even in the House of the Menander, where the numerous rooms
and variety of courtyards provided alternative space for such items.
It therefore seems that, contrary to what we might assume, amphorae,
chests and other storage equipment were not something to hide away
so as not to spoil the gracious effect of architecture and wall paintings.
Instead it is likely that they were considered not to detract from the overall
effect and possibly even to contribute positively to it in some way, perhaps
as evidence to visitors that the house was well supplied and organised
in an orderly fashion. Such an interpretation would be consistent with
some of the symbolism attached to the atrium in various textual sources,
where it is seen as the traditional core of the house, the main living space,

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 112 Artefact distribution and social groups
whose features accommodated some of the most important aspects of
domestic activity. This, the writers claim, was the original location of
the main domestic hearth and the room where the marriage bed of the
dominus was placed. In addition it was a centre for domestic worship at
the household altar, and the location in which a variety of events such
as religious festivals, births, marriages and deaths would all have been
celebrated or commemorated (Flower 1995, 194–203; Nevett 1997, 290).
These sources may not necessarily give an accurate account of the historical
origin of the atrium, particularly for this part of Italy, which became
Roman territory only after the construction of some of the atrium houses
standing in 79 ce, but the cultural attitudes they articulate and some of
the symbolism they attribute to this space may still be applicable here.
The functional character of the atrium in House vi.16.26, which lacks
elaborate wall or floor decoration, supports the association of atria with
more utilitarian aspects of domestic life as well as with reception of visitors.
Similarly, the mixture of cooking and serving vessels, personal effects and
clutter refuse in the secondary atrium of the House of the Menander may
also reflect such a vision. In fact, it is interesting that out of Allison’s whole
sample of thirty houses, which together include thirty-five atria, only a
minority – thirteen atria – have evidence for wall paintings.
In this case, then, it is our own cultural preconceptions that lead us
to see an incompatibility between architecture and movable furnishings
which the occupants of the houses would probably not have perceived
themselves. But there are also instances requiring a different explanation.
In the House of the Ceii, for example, the small room g, which had storage
shelves fixed around its walls, was once decorated with wall paintings.
These were damaged and obscured by the shelves, implying that what may
originally have served as a small decorated living room had subsequently
been converted for use as a store. Allison notes a similar instance of this
kind of change (which she refers to as ‘downgrading’) in room i of the
same house, where a fixed hearth suggests that the room was used for
cooking, while the walls carry paintings. We should not assume a priori
that such decoration was inappropriate to a cooking area, but the fact that,
out of forty-four rooms with fixed hearths in Allison’s sample, only one
other example has wall paintings (House viii.5.9) while the remainder are
finished with plain plaster suggests that the combination of hearth and
decoration was probably not intentional and that, like room g, it did result
from a change in the use of space within the house.9

9 In contrast, painted shrines are often associated with cooking areas: Foss 1997.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Flexible and multi-functional domestic space 113
Seismic damage might have caused the reorganisation of domestic activ-
ity, either temporarily while repairs were undertaken, or permanently if
parts of the house ceased to be used or the property was no longer con-
sidered viable for the needs of the original owners. But the adaptation of
domestic space for use in new ways is also an integral part of the process
of living, and given that the House of the Ceii was occupied for 200 years,
it would perhaps be surprising if evidence for changes in the configuration
and use of space were completely absent. Households grow and contract
as available resources fluctuate, children are born and leave and partners
marry, divorce and die (these patterns are all documented by Bradley for
high-status Roman families based on textual evidence: Bradley 1991, 125–
176). This domestic life cycle means that changes in the requirements
placed on living space are part of a normal process as the architectural
structure is reshaped to suit the requirements of new, or newly config-
ured, households. The easiest elements of the domestic environment to
adapt to these changing requirements are the movable finds and furniture.
Aspects of the architecture, such as the location of doorways, decoration
of the walls, and even the boundaries of the house itself, were altered dur-
ing the lifetime of the houses, as demonstrated particularly by the House
of the Menander. But this process would have been much more expen-
sive and disruptive, and is therefore likely to have been relatively slow to
catch up with the household’s requirements. Such changes are likely, how-
ever, to account for the contradictory impression given in rooms g and
i in the House of the Ceii by the décor and the fixtures and finds they
contained.
The relatively long lifespan of these houses also raises another consider-
ation, which is the need for on-going maintenance. In House vi.16.26 the
stack of roof tiles in the atrium seems to indicate that repairs were antic-
ipated or being carried out at the time of the eruption, and the various
building materials in the House of the Menander suggest that redecoration
was either planned or taking place here, too. It is possible that this work
was being undertaken to repair damage caused by earthquakes, but work
on the roof to remedy the effects of general wear and tear, and redeco-
ration of rooms, must also have been a normal activity for residents of a
building of this age. Some of the works which were being prepared for
or were in progress at the time of the destruction might be the result of
this kind of routine process. Evidence of the inhabitants thinking ahead
to the need for this kind of maintenance can perhaps be seen in the form
of three compluvium tiles found in basement room b of the House of the
Menander.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 114 Artefact distribution and social groups

conclusions: towards a dynamic and socially inclusive

model for households at pompeii
The discussion of the three houses presented above provides a starting point
for building a more complex model for understanding the distribution of
artefacts in ancient domestic contexts and a variety of conclusions emerge
which could be tested against a larger sample of houses. By distinguishing
between items which may have been in use and those more likely to
have been in storage or in refuse deposits, it is possible to explain some
of the variability indicated by the artefact distributions in these houses.
Selective discussion of examples of the three organisational patterns I have
highlighted here shows that all of them contribute in various ways to an
impression of confusion and ambiguity in the archaeological record and
that this is unlikely to have been entirely due to disturbance caused by
seismic activity. Even so, there is only limited spatial separation of objects
and architectural features which are likely to have been used for different
activities. Such a conclusion suggests not only that we cannot answer the
commonly raised question about precise room function, but also that this
is probably not the right question to be asking. Nevertheless, it does not
mean that detailed study of artefacts and their spatial settings is unhelpful.
On the contrary, by thinking in more detail about how our own cultural
perspective affects our interpretation, and about some of the processes
which may have created the archaeological record, it is possible to come to
a better understanding of what multifunctionality and flexibility in spatial
usage may actually have involved in a Roman household.
A major underlying feature of many of the patterns outlined here is
an interplay between activities operating on different timescales. Lin Fox-
hall has pointed out how in Greek contexts archaeologists have tended
to ignore short-term change which took place over days, weeks, months
or a few years, and think instead on the scale of events taking place over
the medium-term – a period of a generation or several generations (Fox-
hall 2000, 485). Indeed this timeframe has been seen as the one classical
archaeology is best fitted to investigate (Snodgrass 1991, 69). Yet what we
see in individual houses at Pompeii is the result of both medium-term
and also short-term activities, taking place both in linear and in cyclical
fashion. Rather than aspiring simply to characterise the house as a static
entity frozen in time at the moment of its destruction, paying attention
to these different scales can give us a richer, more dynamic and more
socially inclusive picture. Distinguishing between interlocking timeframes
of short duration is difficult or impossible to do based on the archaeological

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Conclusions 115
evidence alone, but ethnographic parallels and ancient literary sources
can assist by suggesting possible interpretations of the archaeological
material.
The multifunctional and flexible use of space indicated by patterns of
artefact distributions relate to the shortest timescales, and many of the
activities they reveal may have been cyclical, with the same tasks ceasing
and resuming in a particular location on a regular basis. At the scale of a
single day, a formal atrium like that of the House of the Ceii may have
seen a variety of activities at different times. Aside from the reception of
visitors, members of the household might have come here to draw water,
taking toilet articles from the cupboard in the south-east corner of the
room to wash, or a cup from the chest in the north-east corner, to drink
from. Here, too, tools may have been left as their users came in or out.
Different members of the household would have passed through on their
way to room i, perhaps to use the latrine, warm themselves by the hearth,
or prepare or fetch food. At nightfall, lamps may have been taken from the
shelves under the stairs in order to provide some light. On a seasonal basis
the pattern might have changed somewhat: on summer evenings the focus
of activity might have shifted more towards the rooms at the back of the
house to take advantage of breezes around the garden. Meals might have
been prepared on a brazier in a room adjoining the garden, and eaten off
vessels taken from the shelves of room g.
On a longer timescale, the state of repair of a house must have echoed
the fortunes, or the priorities, of its owners, as we see in the case of
House vi.16.26. The property contains relatively little decoration, most of
it consisting of styles of wall decoration traditionally dated to the second
and first centuries bce. The majority of the rooms, including the atrium,
had only plain plaster walls, and in contrast with the Vitruvian model, the
emphasis of this space seems to have been on functionality. The southern
area around the peristyle may have fallen into disrepair, or even been
converted into a separate unit, and work seems to have been under way on
the roof. Here, then, was a house which does not seem to have attracted
much care and attention from its owners for some considerable time before
the eruption, possibly even for several generations if the wall paintings can
be used as a guide. (Of course, the owners need not have been the same
family throughout this whole period.)
Multigenerational time is seen even more clearly in the House of the
Menander, at both small and large scales. On a small scale, in a wealthy
household like this one the possessions of one generation may have rep-
resented an accumulation of items, some of which may have been passed

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 116 Artefact distribution and social groups
down through the family, while others may have been acquired at different
times by the owners. The silver hoard found in the basement may be an
illustration of this process since it consists of vessels made over an extended
period between the mid first century bce and the mid first century ce in
different styles and with different designs, and is not a matched service
which could have been purchased together. On a larger scale the structure
of the house itself also shows the effects of change over time. The evidence
for expansion of the property boundaries suggests a gradual increase in
the wealth and status of successive generations of owners. The original
core, consisting of the atrium and surrounding rooms, was comparable in
ground area to the House of the Ceii, although there was also a garden
whose size and layout cannot now be determined (Ling 1997, 223–225).
In its final form, however, the house covered roughly six times that area.
But while the overall trajectory during more than two centuries of occu-
pation is one of dramatic growth, Ling’s detailed analysis of the insula’s
building phases shows that there were periods of contraction. For example,
for part of its history the house was also connected with its neighbour to
the west, the so-called House of the Smith. The House of the Menander
seems to have suffered, losing a room on its north-west corner, when the
two were subsequently separated (Ling 1997, 134). Similarly, the provision
of a separate stable-court, which may have served to bring large quantities
of provisions into the house to supply the household, or may even have
been part of a commercial or farming enterprise (as Ling argues: Ling 1997,
252), is an indication of a flourishing household, but there may have been a
period of recession immediately before the eruption since the stable-court
seems to have fallen out of use by then. While it is impossible to recon-
struct the individual domestic circumstances which led either to periods of
expansion or to setbacks, such changes are, again, an important reminder
of the intimate connection between the physical structure of a building, its
contents, and the fortunes of its owners. They also underline the way in
which different timescales can offer contradictory information: short-term
changes can move in a different direction from longer-term trends.
There is thus a tension between the conclusions to be drawn from
the more durable evidence of longer-term behavioural patterns, and the
less permanent material representing smaller-scale, shorter-term activities.
These two different aspects of the archaeological record are likely to coin-
cide to some extent with the perspectives of different social groups within
the household. While the literary sources emphasise, and the architecture
creates an appearance of, monumentality, this does not seem to have been
the only, or even the main, consideration governing the organisation of the

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 Conclusions 117
contents of the houses. The presence in the atrium of cupboards containing
household implements emphasises alternative uses made of this space by
different members of the household. Unlike the monumental architecture
which serves as a cue to signal its function as a reception space, these
other roles were not, as far as we can see, made permanent in the architec-
ture: aside from built-in shelves and cupboards they can be detected only
through the more ephemeral evidence of the finds. The social hierarchy
of the household is expressed through the durability of the material cor-
relates for the activities of the different household members, reinforcing
the dominance of the dominus. It is unclear whether the performance of
domestic activities in the atrium was part of the ideological construction
of the household as an institution, but this is a possibility. If the view of
Augustan culture as manipulating the virtuous elite wife as a public sym-
bol of a man’s domestic probity (Milnor 2005, passim) can be extended
to the Campanian context, one might infer that the presence of domestic
equipment in the atrium during the salutatio formed part of a single larger
rhetorical construction of domestic order.
Another component may have been a view into the atrium at other
times of day, with household members visible, busily engaged in domestic
chores. This possibility is perhaps supported by the artfulness with which
the view of the domestic interior from the street is constructed: as Shelley
Hales points out, while appearing to lay the house open to scrutiny, in fact
the layout hides the majority of space, including the entrances to most of
the rooms (Hales 2003,112 and 119). To date most attention has focused
on the way in which the fauces–atrium–tablinum alignment would have
displayed the dominus to visitors. It would, nevertheless, have been equally
possible to create a tableau of domestic industry in the atrium aimed at
passers-by, without exposing activity elsewhere in the house to scrutiny;
and in fact plaster casts of the main doors of the House of the Ceii show
that at the time of the eruption they stood open, presenting such a view
(Michel 1990, 18). (Of course, it is impossible to know whether the open
doors may have been a response to the events surrounding the eruption,
but if Pliny the Younger can be believed that the eruption began in the
afternoon, falling debris may have prevented closure of doors which were
customarily left open at that time of day.)
The patterns of flexibility and multifunctionality highlighted here show
that there is also an additional reading of the evidence, which offers the
perspective of the household as a whole, rather than that of the dominus
alone, and the two perspectives could have co-existed: the arrangement of
smaller-scale domestic activities suggests that in addition to keeping up

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press

 118 Artefact distribution and social groups
appearances, considerable attention was also given to practical issues such
as the availability of space for large storage vessels, keeping things at or
close to the places where they were needed, and ways of keeping warm
or cool at different times of year. Thus, the comfort and convenience of
the household as a whole were an important, integral part of its overall
pattern of organisation. Studying the finds, which represent the result of
short-term activities by a range of household members including women
and slaves, alongside the architecture therefore produces a more deeply
textured understanding of the household as a composite social unit with its
own logic and agenda, and offers a new perspective which to some extent
counterbalances the traditional image of the household as revolving around
the requirements of an elite male dominus.

https://doi.org/10.1017/CBO9780511780103.008 Published online by Cambridge University Press