Consumption and Roman Archaeology: Beyond Pompeii

Nick Ray

Introduction The potential for the application of archaeological data to illuminate aspects of the ancient economy is immense. Such evidence provides the opportunity to ask questions that cannot be broached through other means, as it produces a significant component absent in most other areas of study: quantifiable data. As a result, there is greater capacity for investigation into unexplored realms, such as consumption. The use of this concept in Roman archaeology is limited, with TRAC being one of the few publications to offer papers demonstrating the possibilities of actively applying consumer theory in systematic and valid ways; for example, Fincham (2002), Martins (2003), Monteil (2004), and Pitts (2004).

‘Consumption’ as a model is seldom recognised as a feasible investigative tool for the ancient world, primarily due to its strong associations with economic studies. More specifically, as acknowledged by Fincham (2002), the conceptions of ‘consumers’ and ‘consumption’ have modernist connotations and have become synonymous with the idea of capitalism and the modern economic atmosphere. Thus, they are perceived as anachronistic in the realms of archaeological studies; a view that largely remains, despite an ever increasing presence in the literature across the social sciences (Fine 2000: 1). I believe these perceptions to be misplaced and, moreover, will argue in this paper that if correctly applied, consumption can be a useful lens through which the ancient world can be viewed. This will be achieved through correcting the misinterpretation that the concept is solely capitalistic in reach, and subsequently by creating a working definition that also considers social factors.

The increasing use of the term ‘consumption’, as opposed to the concept, brings its own problems however. Traditionally, consumer theory has been rejected for being contentious and unsuitable for archaeological investigation, but improper use without understanding it, or its limits, could be just as harmful. Therefore, it has become necessary to address these concerns. As such, one of the main purposes of this paper is, through a theory-oriented approach, to outline definitions and the framework’s capacities. Further aims are to address its limitations, and subsequently review the literature on some of the current theories about consumer behaviour and its relation to the field of archaeology. To aid in this, I will outline how such theory can be effectively applied to archaeological situations by discussing my ongoing research into private consumption behaviour evidenced through material culture at the household level, with analysis directed toward wider economic implications (such as market economies).

What is ‘Consumption’ Theory? The economic sense of the word ‘consumption’ had developed by the mid-nineteenth century with regard to the bourgeois, with consumption becoming a contrast to production (Aldridge 2003: 2; Williams 1988: 78–9). This was witnessed by Adam Smith who wrote that ‘[c]onsumption is the sole end and purpose of all production’ (Smith 1976: 660). Today, the act of consumption occurs in various forms and is not seen merely an economic behaviour (Bocock 1993, 2–3). It is an act with numerous socio-cultural contributing factors, cumulatively representing a human activity that is frequently taken for granted (de Grazia 1996). As such, consumption is not limited to advanced capitalist societies, but also occurs in societies that are predominantly rural and agriculturally-based. Once determining social or cultural influences have been experienced by an individual, the desire for goods or behaviour can transcend the economic factors, irrelevant of the capability of that person to acquire them (Bocock 1993: 3). Furthermore, consumption is not restricted to material goods, as it can also encompass political or social activity. For example, education has been included by some as an act of consumerism, although Douglas and Isherwood (1996: 109) argue that this is not the case, but is instead an investment in human capital. Gibb (1996) extends the list for consumer behaviour to include farm animals, as well as burial practices. Holt (1995) uses baseball spectators as an example of consumers, through their consumption of the sport; what he calls ‘consuming as play’. A positive consequence of widening the concept of ‘consumption’ in such social science studies has been that it is less often approached through a supply and demand theoretical framework, reducing its ahistorical character and serving to remove the focus from a primarily material-content view of acquired commodities (Fine 2000).

Fine (2000: 9) describes consumption as a conduit for cultural influences, reinforcing the necessity for creating distance from traditional supply-demand frameworks and producerconsumer oppositions. Anthropologists have applied this symbiotic material-social double-role perspective to consumption extensively, resulting in it being practically axiomatic (Douglas and Isherwood 1996: 39). By widening the context within which consumption is studied, in particular incorporating the concept of identity, the scope for its application becomes ‘unlimited’ (Fine 2000: 11). Regarding the ancient world, Smith (1999: 116) and Vaughn (2004: 63) state that the study of consumption is a logical step, as the archaeological record is composed of innumerable physical traces of consumption activities; ‘archaeological evidence for consumption is ubiquitous’ (Smith 1999: 116).

In the same way as consumption itself, the study of consumption is not restricted to the field of economics – it is multidisciplinary. Fundamental concepts from the disciplines of psychology, sociology, social psychology, and anthropology are all influential in the study of consumption and consumer behaviour (see Table 1.1). Some, however, separate its study into two polarised groups: formal, rational behaviour studied through economics; and other irrational cultural contexts studied through branches of the social sciences (Aldridge 2003: 7–8; Slater 1997: 51). Irrespective of the perspective from which researchers approach the subject of consumption, the emphasis is upon understanding and explaining why, and how, people acquire what they do (Bagozzi et al. 2002: 1). Recent pleas for more integrated recognition of both functionalism and symbolism (for example Miles et al. 2002; Smith 1999; Warde 2002) have helped broaden the concept. Subsequently, there has been a shift in attitude away from the mechanics of capitalist production and its resultant impact on consumption behaviour, towards cultural and communicative systems balanced with utilitarian use.

Defining ‘Consumption’ Within anthropological studies, consumption has been defined by Douglas and Isherwood (1996: 37) as the use of material possessions beyond commerce and free within the law, enabling the concept to be used for non-state societies that lack not only capitalism but also commerce. Similarly, the broad definition of consumption involving the selection, purchase, use, maintenance, repair, and disposal of any product or service (Campbell 1995: 102) has been applied in archaeological studies (for example Smith 1999; Vaughn 2004). It follows, therefore, that for the purposes of Roman archaeology, the term ‘consumer’ can be stripped down to its fundamental meaning allowing for significant application to economic studies: someone who consumes a certain thing, regardless of how or why (Fincham 2002: 34). In addition to the acquisition of objects/‘products’, Schiffman and Kanuk (1987: 6) define consumer behaviour as ‘the behaviour that consumers display in searching for, using, evaluating, and disposing of products, services, [practices,] and ideas which they expect will satisfy their needs’ (also see Bagozzi et al. 2002: 1, 171; de Grazia 1996: 3–4). Further to this, consumption is more often than not interpreted as individual rather than collective behaviour, neglecting the impact of ruling institutions and the state (de Grazia 1996: 9); an important point when applying consumption studies to the ancient economy and the Roman world.

Additionally, consumption provides an alternative perspective from which to investigate the ancient economy, especially when compared to the much-used approach of production (Mattingly 1988; Wilson 2001; 2002a; 2002b). This is not to say that the two are entirely separate entities within the same economic continuum, parasitical consumers versus functional producers, rather they exist through complex reciprocal relationships (Courtney 1997: 98). For example, the mode of production sets the parameters within which other activities (such as consumption) can occur, but it does not determine such activities (Bocock 1993: 6). As with production, consumption – and thus consumer behaviour – is embedded in social structure (Foxall et al. 1998: 213) and relates to issues of trade, transport, marketing, and sale, as well as the larger issue of the ‘ancient economy’. Therefore, to examine the dual concepts of production and consumption separately is to consider only half of an interrelated nexus (Belk and Ger 1994; de Grazia 1996), and they should consequently be seen as ‘imbricated’ in order to reach a full understanding (Aldridge 2003: 31). Further, it could be argued that the connection is more entwined than this, with the act of production only being possible through the consumption of raw materials and labour (Smith 1999: 116).

Just as consumption is not solely an economic process, it also represents more than the utilitarian, material object, encompassing cultural signs and symbols. It is a system, or process, communicating social meaning through which people convey symbolic messages to themselves as well as to others on matters such as class, social status, and identity (Aldridge 2003: 10–24; Belk 1985; Corrigan 1997; Douglas and Isherwood 1996; Hodder 1982; McCracken 1986; Rook 1985). As an extension of this, consumer actions can be ends in themselves (autotelic actions), as well as a means to further ends (instrumental actions). Furthermore, diverse aspects of culture, including political, social, religious, military, and technological fields (MacKinnon 2004: 11–12) are pervaded by consumption, with numerous factors impacting upon the behaviour of consumers and affecting their subsequent actions. For example, identity is a powerful, fluid, context-dependent perception that influences consumption through the desire to belong to a group, or a social/cultural unit. This, in turn, is thought to be centrally influenced by gender, amongst other factors (Aldridge 2003: 10–24; Fine 2000: 10; Schroeder 2003). One example of how identity and consumption share common ground is the adoption of Roman-style material culture, such as ceramics, by indigenous populations. Frequently, such consumption of goods is interpreted by archaeologists as being representative of emulation, or an attempted association of identity with the Romans, but do not consider a cultural hybridisation resulting from a two-way process (Cooper 2000). The use of the term ‘reference group’ has been used in consumer studies to describe an individual or collection of people, whom the individual uses as a source of behaviour, values, or beliefs. Consequently, consumers appear to select collectives that are compatible with their self-concepts; the family often forms one of the most important social influences (Foxall et al. 1998: 218–19). Unfortunately, such intangible social factors are of questionable visibility in the archaeological record. A possible exception to this is the interpersonal mechanism of internalisation, which is one way of explaining the influence that reference groups can have on consumers. This is significant because it can contribute to understanding processes relating to adapted use of ‘Roman’ objects as an expression of indigenous culture, in other words the internalisation of goods and practices (see Cooper 1996). This may become more significant with identity being increasingly examined through the study of the consumption of material culture (for example Carr 2001; Fincham 2000; Gardner 2001; Revell 2000; Spradley 2001).

Looking Beyond Generic Models for the Ancient Economy: The Role of Consumption Despite advancement towards more specific examples and applications of archaeological evidence, there still remains an issue inherent to ancient economic history: the lack of an answer to the question posed by Davies (1998: 230), ‘[w]hat do we regard as a satisfactory framework for the description, analysis, and interpretation of economic activity in antiquity?’ It is only through detailed consideration of the archaeological record that attempts can be made to test the validity of a proposed model (Mattingly et al. 2001: 67) or enable a progression towards more carefully directed research questions. However, making generalisations based on the conclusions of one study and then extending them spatially or temporally to other contexts becomes dangerous. Rather, the value of conclusions drawn stems from a methodological framework that is transferable to other situations.

The focus of the significant body of work regarding theoretical frameworks and models for the ancient economy rests upon production. Within this, questions of scale continue to be the central issue. Consumption as a concept and an investigative tool has remained marginal to discussions, gaining little more than a casual comment. Within the confines of consumer behaviour and the identification of consumption patterns, the debate has shifted away from the usual ‘scale of production’ arguments to one of access and market availability, in addition to the numerous other cultural issues that have been addressed. As such, it creates an opportunity to tackle the topic of market economies from the other side: that of the consumer, the destination of goods. My study adopts and adapts such a perspective, not only making consumption the central subject, but on completion will provide a transferable framework enabling future application of the method to other sites and time periods.

The Application of Consumption to Historical and Archaeological Studies Fincham (2002) argues that the concept of discrepant experience can be extended into the study of consumption through ‘discrepant consumerism’, a process by which objects have social lives (refer to Appadurai 1986; Gosden and Marshall 1999; Kopytoff 1986) and the meaning of an object changes between social, cultural, political, or personal circumstances. These circumstances, or locations, within the object’s life thus represent discrepant consumerism. One example of such a process is highlighted by Werbner (1990) through gift economies that become embedded in capitalistic, market-commodity economies, and in plural societies. Commodities can be converted into gifts (becoming customary, ranked, personalised) depending of the social context of the transaction, although they can still be responsive to changes in the market (inflation, fashions etc.), and can also subsequently be reconverted into commodity-level objects (see Gregory 1982).

Douglas and Isherwood (1996: 131) state that for less developed countries it is not difficult to recognise different consumption patterns as they usually correspond to different levels of income, and to a very obvious social stratification. Such consumption patterns might also be evident in Roman archaeology as status was of great importance in Roman society.

Recognising consumer behaviour in archaeological contexts generally means a focus on rich, high status people. This is primarily because the surviving evidence is material culture, which is more conspicuous and ‘available’ in ‘elite’ houses. This does not mean that the poorer classes are ignored, rather that ‘elite’ consumption creates a relative scale with which comparisons can be drawn permitting a further understanding of poverty and consumption patterns (Douglas and Isherwood 1996: 108). However, this provokes the question of just how visible are the poorest classes in the archaeological record. For example, Barclay et al. (1990: 69) state that urban domestic sites from Medieval Winchester ‘did not compare unfavourably’ in terms of their consumption of goods compared to the greater houses. Although when the status of the residents from these houses is considered, it becomes evident that they were not ‘poor’, rather an impression of an occasionally wealthy inhabitant is presented (Hinton 1990: 34).

Although status in the Roman world was a governing force, it is difficult to identify at an individual level through material culture. Within Roman houses there was a range of social standing, from the ‘owners’ at the top of the hierarchy to the servants at the bottom. Separating who owned what is an unattainable target and therefore individual status levels are not possible to identify with a high degree of confidence. This impacts on issues of cultural value, as occupants of the house will have ascribed value differently. One often-used method of consumption analysis consists of defining social classes in terms of spending habits, which are then linked to occupation and income groupings to create categorised guidelines. However, although goods could define social class, it has been argued that they could not then be used to explain consumption behaviour, as occupation categories are not a safe guide (Douglas and Isherwood 1996: 131–32).

Brewer and Porter (1993: 3) stated that one of the primary problems of the concept of consumption, when applied to economic history and material culture, was that it was ‘historiographically immature’. When this statement was made, the focus of the application of the concept had been placed upon the eighteenth and nineteenth centuries; for example Johnson (1988), LeeDecker et al. (1987), Shammas (1993), and Weatherill (1988; 1993). However, consumption has been applied more recently to other branches of archaeological study, such as archaeology of Native America (Bayman 1996), Latin America (Vaughn 2004), and the Bronze Age (Webb 1998). It has also received thematic emphasis in some areas, such as architecture as conspicuous consumption (Schoep 2004; Trigger 1990), and even archaeological sites have been seen as objects of consumption, such as Stonehenge (Hetherington 1992). Despite this, consumption largely received a tentative introduction to Roman archaeology, in studies such as Laurence (1994) and Paterson (1997), although its use has been slowly permeating further into Roman studies; for example, Cooper (1996), Crummy and Eckardt (2003), Fincham (2002), Funari (2002), MacKinnon (2004), and Martins (2003). Within this slow methodological development, the current trend in research appears to be a focus on food consumption, and in particular faunal remains as a representative function of this. Pottery assemblages, however, still form the predominant and most accessible forms of evidence from the archaeological record for investigating economic activity in past societies. The recurrent focus within this is upon supply and its relationship to productive processes rather than direct application of consumption analysis (for example Evans 1993; Going 1992; Marsh 1981; Willis 1997; 1998).

The capacity for answering questions about consumer behaviour in the Roman world is starting to be recognised, a fact highlighted by studies such as Monaghan’s (1995) in which a five-level gradation system of pottery demand by households is laid out. This work was carried out in order to contend with assumptions regarding free-market trade, and to demonstrate that consumer choice in Roman Britain was limited; Monaghan argues that the capacity for consumer selectivity of commodities was low and exercised at a very low level. Fincham (2002) uses quantified ceramic data to investigate whether native North African populations consumed material culture in an attempt to ‘become Roman’. The observed pattern indicates that the more that is consumed, the greater the number of ‘fringe items’ relative to the core mode, rather than a general increase across the range of pottery forms (Fincham 2002: 41). In addition to the pottery vessels themselves, ceramic evidence can also help to indicate trade in other commodities, such as those transported in amphorae. De Sena and Ikäheimo (2003) have applied the ceramic assemblage from the House of the Vestals in Pompeii (VI.i.7) to the investigations of chronological trends in the supply of wine, olive oil, and fish products, as well as for the domestic pottery itself. The evidence shows that the consumption of such commodities can be mapped, indicating a decline in regional amphora-borne products, less exportation of goods, and increased long-distance importation; a trend reflected by the pottery (2003: 314–15). Continual increase in the wealth of the city, combined with extended trade links, thus served to provide greater consumer choice – a factor exaggerated by the increased demand for the consumption and conspicuous display of luxury items, also mirrored by secondary products (De Sena and Ikäheimo 2003).

In reviewing the literature on the application of consumption studies in archaeology, several points have become evident. Firstly, there is a potential in the Roman ceramic evidence to identify and indicate changes in social stratification based on socio-economic circumstances, highlighted through consumption practices. Secondly, it is clear that this potential is equally prominent for artefactual evidence on the whole; a point that has been raised in several studies that call for comparable work based on other facets of material culture in order to identify patterns supportive (or otherwise) of the ceramic data (for example Baugher and Venables 1987; Fincham 2002; Klein 1991; Miller 1980).

Another conclusion frequently reached in analyses of consumption patterns, irrespective of the chronological frame, is that stimuli for consumer behaviour are rarely the result of a single, easily explained, prime mover. Rather, complex inter-relating factors are the motivators, and consumption models should reflect this. Courtney (1997), Klein (1991), Leone and Crosby (1987), and Weatherill (1988), for example, all argue that simplistic models for consumption are ineffectual, citing factors including:

standard of living household size and structure social expression (communication of identity) income strategies purchasing power cost competing items durability process of acquisition personal (or group) customs and tastes external economic conditions object use and function emulation peer pressure cultural differences (such as diet) choice availability in the market place market impact market accessibility the role of women in the household

The Data One of the critical issues with the application of concepts such as consumption is the identification of key research questions. Through the examination of previous methods and studies, those particularly pertinent to this study have been identified:

1) Do patterns of consumption exist in a distinguishable way that can be observed within the Roman Empire?

2) Did the rich have more, or did they also have different?

3) Within any patterns that emerge, are there levels of consumption represented by subsets of the population?

The core data will come from Pompeii, as not only does the site present opportunity for investigation at the household level, but also examination at the level of urban centre through comparison of different households. Nonetheless, contrary to popular belief, Pompeii does not represent a moment frozen in time impervious to outside interference or influence, and is therefore not the ‘ideal’ site often expected, but one that carries its own inherent obstructions (refer to Allison 1992a: 2004). The data comprises artefact assemblages from a database of twelve Pompeian houses and their contents, collated by Dr. Penelope Allison. The data, consisting of approximately 10,000 artefacts, was collected from three main sources: the published reports in Notizie degli Scavi di Antichità, the original records from the Giornali degli Scavi, and the Pompeii archives containing the artefact inventories. In addition to these are the extant architectural remains of the houses themselves (Allison 1992b: 22–27; 1997a: 2004). The data is therefore taken from a single architectural group (that of atrium houses) for reasons of documentary availability: 19th century excavators focused their efforts on the larger, wealthier domestic residences, rather than the smaller households or commercial buildings (Allison 1992b: 21; Ellis 2004). Allison’s research concentrates on two principal topics: 1) the distribution of the room contents as indicators of habitual room use; 2) the living conditions prior to the final phase of the city as well as abandonment, and post-abandonment, processes (Allison 1992b: 3). Therefore she has not analysed the data from an economic perspective, although recognises the fact that the houses represent the consumption destination for the objects (Allison 1992b: 1997b).

Allison categorised the data in several ways, although those of artefact type and artefact function will be of most interest to this study, consisting of 239 and 35 categories respectively. Analyses to identify patterns in a household’s consumption behaviour – or to use an economic expression, their ‘consumption strategy’ – will be conducted by the author for both of these groupings, as well as for material, through the use of correspondence analysis (a multivariate statistical technique). Prior to this, preliminary investigation of the data is ongoing in order to identify any general, broad inter- and intra-assemblage trends (see Figs. 1 and 2). During this process Allison’s existing categorisation systems were found to be unsuitable for some aspects of this study, due in part to the large numbers of groupings. Therefore, the decision was made to re-categorise the data based upon Crummy’s (1983) groupings of artefacts from Roman Britain, so that the framework is based upon a more widely recognised structure. All of Crummy’s groupings are incorporated as well as some adjuncts to the system allowing all of the artefact types from the Pompeian data to be categorised. Even where some of these categories are not represented in the data being used in this study, the framework exists for others to make use of in the future. Not only will this create a more manageable and useful data set for the purposes of this study, it will also make the methodological framework accessible for those working on other sites, especially those in Britain already familiar with the system. The categories I will be using are listed in Table 1:

Table 1: Categories of artefacts Category No. Category Title 1 Objects of personal adornment or dress 2 Toilet, surgical or pharmaceutical instruments 3 Objects used in the manufacture or working of textiles 4a Household utensils 4b Household furniture 5 Objects used for recreational purposes 6 Objects employed in weighing and measuring 7 Objects used for, or associated with, written communications 8 Objects associated with transport 9 Buildings and services 10 Tools 11 Fasteners and fittings 12 Objects associated with agriculture, horticulture and animal husbandry 13 Military equipment 14 Objects associated with religious beliefs and practices 15 Objects and waste material associated with metal working 16 Objects and waste material associated with antler, horn, bone, and tooth working 17 Objects and waste material associated with the manufacture of pottery vessels or pipeclay objects 18 Miscellaneous (including Crummy’s category of ‘objects the function or identification of which is unknown or uncertain’) 19 Organics 20a Ceramic vessels 20b Other vessels 21 Coins

Figure 1: Summary of Assemblages by House – absolute values

Figure 2: Summary of Assemblages by House – relative values by percentage

It is predicted that, through detailed analysis of the data, general patterns of consumer behaviour will be revealed in the clustering of artefact type or material. For example, (relating back to question 2, above) do the larger, more elaborate residences exhibit greater quantities of goods, or the same number but of a higher quality, when compared to residences lower in the social hierarchy? Once the general patterns have been identified, particular categories can be investigated in greater detail. For instance, the categories of various vessel types can be analysed by their material to answer questions such as, is there a high proportion of glass or metal vessels in some of the houses (relative to the remainder of the sample)? If this is the case, does it indicate wealthier households? As this paper demonstrates, there are many ways in which consumption analysis can be applied to this form of data depending on what research questions are posed. For example, if the more ‘luxury’ items are taken out of the data sets, are there patterns or similarities still visible and, if so, are they significant? This could mean that, as identified by Fincham (2002: 32–33), elitist markers can be identified and used to potentially indicate social selection.

Figures 1 and 2 summarise the data for the twelve house assemblages in the sample, illustrating some distinct trends: certain categories demonstrate peaks that are consistent across the sample, such as numbers 4b (household furniture), 9 (buildings and services), 11 (fasteners and fittings), and 20b (other vessels), which all offer stronger patterns than other groups, such as 4a (household utensils). Similarly, troughs can be observed for categories 7 (objects used for, or associated with, written communications), 13 (military equipment) and 14 (objects associated with religious beliefs and practices); predictable for the first two given the nature of the groups concerned. Although these relationships do not represent absolute similarities in artefact frequencies, they do show relative correlations that can begin to be examined more closely, such as through the application of correspondence analysis. This will be performed on the data in multiple formats, including the re-categorised data as well as Allison’s original categories. By using different levels of grouping, patterns already identified can be confirmed or discounted – additionally serving to verify the format of the re-categorised data.

An further way of maximising use of the data is to consider fineware proportions in the assemblages. The principal problem, however, that arises from the application of such an analysis is that high status sites also utilised non-ceramic vessels (such as glass, silver, bronze) whereas low status sites had fewer alternatives to coarse pottery other than wooden vessels. Consequently, the fineware proportions may demonstrate a lower-status pottery assemblage for high-status sites (Evans 1993: 110), due in part to the bias of archaeological visibility. Furthermore, sites from the higher end of the socio-economic spectrum may use vessels in a different way to those lower down the spectrum. Not only could this relate to prestige items and their use for display, but also issues of disposal and subsequent deposition. Functional issues with vessel re-usage can be highlighted by an example from Catterick where a small enamelled flask, that would have once been an elegant item from a toilette, was found in a third-century context where it had been used as a container for adhesive (Cool 2002: 30–31). Most re-use, however, is not as recognisable as this because functional application of the object remains constant.

A further problem that is evident from the assemblages is that it is necessary to remove some of the original data from the records, as certain categories are inappropriate for this study. For example, human remains are present in the records for some of the houses in the database, although are not relevant to studies of goods consumed. Other categories, such as coins, can have a significant effect on the results (Allison 1992b) and represent a method of negotiating consumption, rather than themselves being consumed goods. To control such distorting factors, analysis will be conducted with coins included, and then repeated with them omitted. Similarly, silver will have a distorting influence on the results of the analysis because its presence is skewed toward one house, the House of the Menander. In effect, the distortion would be significant because more silver has been recovered from this one house than from all of the other excavated houses in the city combined.

One approach to consumption analysis in Pompeii would be through the concept of ‘conspicuous consumption’ (refer to Veblen 1925; Warde 2002: 12), applied to any of the multiple examples of ‘elite’ houses in the city. This concept describes the actions of consumers who spend, or acquire, conspicuously in order to convey messages to others about their social status. However, this avenue of investigation does not provide suitable answers relating to the ancient economy (for a study that makes use of the concept of conspicuous consumption, see Martins 2003). It would only minimally offer explanation for what is a complex issue by focussing on a limited aspect of individual elitism, that of status orientated display and social emulation (refer to Simmel 1904 for his ‘trickle-down’ theory of emulation). For consumption analysis to be effective it is necessary to not concentrate solely on the luxury and statusorientated objects, but also on the mundane, everyday items (Warde 2002: 19). The sample to be used in this study does incorporate ‘elite’ aspects, but also integrates more mundane elements (for example, locks, keys, gaming pieces, and hairpins), serving to broaden the scope for interpretation of economic patterns and implications.

Interpreting ‘Consumer Behaviour’ In addition to problems with analytical methods, ways of interpreting the results and identifying consumption patterns are important. For example, Douglas and Isherwood (1996: 133–34) propose three distinctive patterns of scale for overall consumption, whereas Monaghan (1995: 153–55) proposes a five-level gradation system of household demand for pottery. Although not directly applicable to Roman Pompeii, they indicate methods of ranking observed patterns. Relating to this, it will also be necessary to consider methods used in modern consumption studies. For example, Fine et al. (1992a) examine data from the recent National Readership Surveys in two ways: firstly, they contrast frequencies of ownership across different sections of the population; and secondly, they order, or rank, the durables by frequency for the population as a whole, and for sub-samples. In addition to this, another study by Fine et al. (1992b) applies a method motivated by social choice theory using data from National Readership Surveys for five-yearly intervals between 1975 and 1990 in which consumption norms for consumer durables are theoretically identified, together with consistency across the population. Goods are ranked in order (accompanied by a measure of confidence), with those at the top of the poll representing standard items, whereas goods lower down in the average level of ownership indicate items consumed by relatively few people. By examining the consumption patterns expressed by the data, the authors attempt to identify varying patterns exhibited by different sections of the population. The methodology used in these two studies by Fine et al. is relevant to this research because it provides a technique for identifying ‘fringe’ and ‘core’ goods within the data set. Identifying such distinctions is an important stage in establishing consumption patterns, as demonstrated by Fincham (2002). Future work on consumption analysis in the ancient world could benefit from considering types of site other than urban domestic. For example, a Romano-British military assemblage would offer an interesting study. Corresponding to the observations by Millett (1980), Evans (1993) suggests that the high proportion of finewares within forts is a product of the aggregate purchasing power of the site serving as a redistributive centre. A mechanism such as this would be essential in order for finewares to be supplied to rural areas where there is low consumer demand (Evans 1993: 116–17); a feature also identified by Weatherill within the seventeenth and eighteenth centuries, with merchants and inhabitants of urban centres more capable of obtaining such goods (Weatherill 1988; 1993: 117). It is also possible, however, that soldiers’ money had a significant influence on the fineware presence in forts through an ability to acquire more readily, thus supporting Hopkins’ (1980) ‘taxes and trade’ model for the ancient economy.

Conclusion The principal aims of this study were to demonstrate that consumption is a valid concept to apply to studies of the ancient economy, and, having accomplished this, to establish a workable method with which to investigate data from Pompeii. The theory and working definitions behind ‘consumption’ have been outlined, demonstrating that it is not an anachronistic concept, but one that can be applied to archaeological, or ancient historical, investigations concerning economic and social questions. In addition, the examination of previous studies that have adopted the theoretical base has helped to identify methodological problems and ascertain which techniques are relevant to the ancient world and the available data. Although consumption analysis has hitherto remained marginal in studies of the ancient economy, with the subject of scale of production remaining central, application of the concept redefines parameters to focus on questions of access from the perspective of the destination of goods and the consumer. Such an approach will, as a result, necessarily consider conceptual issues of market economies.

The recognition of key research questions for this study has enabled a working method to be established that incorporates aspects from different academic fields. Application of this methodology to artefact assemblages from Pompeii is providing a means of identifying private consumption behaviour, evidenced through material culture at the household level, and relating these patterns to wider economic questions.

School of Archaeology and Ancient History, University of Leicester

Acknowledgments I would like to thank David Mattingly, Penelope Allison, Steven Ellis, and Lin Foxhall for their time, direction, and advice during the preparation of this work. Also, I owe thanks to Xanth Mallett for her contributions and support throughout my work. Finally, I would also like to thank the AHRB for funding my doctoral research programme.