Materialising Roman Histories

Introduction

Chapter 1

What did objects do in the Roman world? Beyond representation

Astrid Van Oyen and Martin Pitts*

The problem with representation

Archaeologists often remark on the massive and widespread changes in the material environment in the Roman imperial period. There were more ‘things’ around, which impacted on the lives of the many as well as the privileged few. The volume of traded goods increased, networks of circulation expanded, and local production intensified (e.g. Greene 2008; Bowman and Wilson 2009). With quantitative increase came qualitative innovation. Objects became ever more differentiated in terms of style and function. For many communities, especially in northern and western Europe, the Roman period heralded the first appearance of genuinely standardised material culture, as opposed to objects that belonged to a more generally shared stylistic continuum. Despite the deep and far-ranging implications of these observations for current understandings of the Roman past, there have been relatively few attempts to explain or come to grips with their implications (a notable exception is Wallace-Hadrill 2008). Inquiries into the causes of such profound material changes have often involved methodological leaps of faith, connecting the plethora of new objects and styles to top-down models of imperialism, economic growth and Romanisation. But behind these empirical observations lurks another historical question: what were the historical consequences of these changes in the material environment? Did these changes actually alter people’s relations to things, and through this, to each other?

In order to address these questions, we believe it is necessary to free up conceptual space to reconsider the issue of how we write history from artefacts. The culture-historical equation between pots and people that underpinned (for example) the earliest models of Romanisation is now rightly frowned upon. From changes in artefacts and assemblages we cannot confidently deduce the arrival of new people (e.g. Eckardt 2014). We argue that in the wake of the discredited culture-historical paradigm, Roman archaeology has neglected the opportunity to rethink its model of material culture. Instead, it has merely refined a representational approach: if objects no longer represent people, they have come to stand for or reflect motives external to them, such as status or other facets of group identity.

What we describe as a ‘representational approach’ is firmly ingrained in the ways that artefacts figure in major narratives in Roman archaeology and history. To illustrate this further, let us take the example of pottery – the most ubiquitous and numerous class of artefacts that survives from the Roman world. If pottery plays a role in Roman history at all, it does so in an intrinsically representational manner. This practice is perhaps most explicit in studies of the Roman economy. Since the absence of equivalent data means that the Roman economy cannot be directly measured in ways analogous to modern economies, pottery falls into the category of ‘proxy data’, in which distribution patterns allow otherwise archaeologically invisible economic phenomena to be studied, such as market integration or economic growth (e.g. Brughmans and Poblome 2016b). Despite the general success of this approach, Kevin Greene (2005, 43) has drawn attention to problems associated with a representational way of thinking:

The ‘Roman economy’ is not a natural phenomenon or set of variables analogous to climate. Unlike weather and tree-ring growth, no direct causal connection exists between the workings of an economy and the deposition of potsherds on archaeological sites. Thus, the term ‘proxy evidence’ may promote an unduly optimistic expectation that material evidence can be used directly for ‘reconstructing’ the economy.

Greene’s cautionary observation highlights the essential disconnection in representational thinking between the specific circumstances of individual artefacts and the bigger ideas they are often made to stand for. In addition, extensive data-mining to illuminate understandings of Roman trade entails problems of biases in large datasets, uneven quantification, and comparing data recorded and classified according to different regional traditions (Wilson 2009, 245–6). Nevertheless, the core assumption that artefacts may stand as proxies for economic activity remains unchallenged (but see Scheidel 2007). Let us be clear: there is nothing wrong with this method per se. Studies of pottery as proxy evidence offer much potential for insights into ancient economies. Reduction is inevitable in an approach that largely divorces pottery from the specific contexts in which it was produced, consumed and discarded, so that it may stand for overarching phenomena such as ‘trade’. Likewise, it is commonplace in such studies for important characteristics of the data to be ignored, such as stylistic innovation and functional variation. The problem is not that the use of pottery as proxy data is reductive, but that it silences alternatives. Indeed, the fundamental question of why some pottery travelled long-distances when it could be produced locally in most areas of the Roman world is seldom considered in Roman economic studies.

To continue our example of pottery, its representational treatment in historical narratives of the Roman empire is not limited to the field of economics. The representational lure of pottery for cultural and social analysis is neatly summarised by Greg Woolf (1998, 186):

All but the very poorest had access to some kind of pottery, and those who could expressed their social position and tastes through selection within the variety of ceramics available. Pottery thus makes manifest a series of social categories and claims about status that are inaccessible through most other sources.

Here pots are not so much equated with people, but are viewed as conduits to revealing conscious choices made by different Roman socio-economic groups. In other words, pottery may be used as ‘proxy evidence’ for social differentiation and cultural process. At one level, this realisation is to be welcomed since it has encouraged wider consideration of pots and potsherds as social and cultural indicators, in addition to their well-established use in charting economic patterns and as a dating tool for archaeological structures. There are, crudely speaking, two kinds of major study that have harnessed this approach – the big picture historical narrative, and the more specialist account of consumption patterns. Both tend to be implicitly representational in their treatment of artefacts, which introduces similar problems to those associated with the use of material culture within economic history.

For an example of the study of pots and culture in big picture Roman history, the elegant discussion of Italian-style terra sigillata in Andrew Wallace-Hadrill’s Rome’s Cultural Revolution provides an excellent example (Wallace-Hadrill 2008, 407–21). In many ways this is a rare case in which pottery is given treatment beyond fleeting reference to the archetypal distribution map. Wallace-Hadrill gives detailed thought to the cultural influences acting on Italian-style sigillata, its material properties (e.g. colour and decoration), its origins and the reasons for its boom in the Augustan period. However, his conclusion focuses on a representational issue – what was the meaning of terra sigillata to the consumer? Wallace-Hadrill answers this question by equating the circulation of Italian-style terra sigillata with the blanket concept of ‘luxury’. While this achieves a satisfactory outcome for the reader in connecting the origins of sigillata to other important innovations of the Augustan age, there is an uncomfortable gulf between this high-level generalisation and a lack of detailed consideration of sigillata across various contexts of production and consumption.

If bigger picture historical studies can lack the space to do justice to the complexities of artefactual data, this is less a problem for more dedicated syntheses of Roman pottery. Successful approaches in this vein have connected pottery to its role in the social practices of eating and drinking, as everyday arenas in which routine use informed the formation of changing and contrasting cultural identities (e.g. Cool 2006; Roth 2007; Dietler 2010; Perring and Pitts 2013). However, despite the rejection of blunt representational analyses of ceramic changes under the umbrella of Romanisation, and the increased sensitivity of these studies to sample size and context, most studies continue to use an implicitly representational approach. For example, changes in ceramics and cuisine are variously attributed to internal community dynamics (Cool 2006, 168), an emphasis on local identity (Roth 2007, 201), the presence of ‘native wives in the households of early settlers’ (Dietler 2010, 253), and participation in ‘Gallic styles of consumption’ (Perring and Pitts 2013, 245). While analysis of material culture has become more nuanced, what appears to be most at stake is what ever more complex patterns represent.

At the root of the problems with this representational model is the partial methodological engagement with pottery (and other artefacts, for that matter) as material culture. Attributes of data not deemed essential to reconstructing social and cultural phenomena are often excluded at the outset of analysis, since the primary objective tends to be to understand an abstract concept or process (e.g. cuisine and identity formation) that is external to the object of study (pottery sherds). Crucially, little energy is expended on tracing the broader range of genealogies, associations, continuities and changes in the collective histories and biographies of Roman pottery. For example, in the tradition of Romano-British archaeology, the construct of ‘imported pottery’ forms a familiar and seemingly well-understood category that tends to be used as one of a suite of materials to shed light upon themes ranging from economic networks to urban/rural relations or eating and drinking. In this way, research jumps straight from labelling something as ‘imported’ to broader themes, rather than directly questioning the roles of ‘imported pottery’ as material culture itself. While such leaps were necessary in order not to lose sight of the big picture, archaeology now has tools to build a more continuous path from objects to historical process. A thorough consideration of ‘imported pottery’– including thinking about its associations, genealogies and biographical pathways in neighbouring parts of the Roman world – has considerable potential to inform representational readings at local and pan-regional levels (see Pitts, this volume, for further discussion).

So far, we have made the case for why the representational model of material culture can be problematic for bigger picture narratives and synthetic artefact research in Roman archaeology, in large part by failing to get to grips with objects’ context and stylistic and material specificities. But what of the implications for the study of Roman finds more generally? In a sense, traditional specialist artefact reports are better placed vis-a-vis synthetic studies to deal with such problems. Compared with the writing of big picture history, experienced artefact specialists often have a greater depth of knowledge of their material gained through routine handling of artefactual assemblages. Here, however, the problem of representation is less concerned with issues of interpretation and research questions (although the same issues are present in specialist reports), but rather with the descriptive languages used to categorise and classify artefacts.

To explore further, let us resume our consideration of pottery. Representational models of material culture are implicit at multiple levels of the description of ceramic wares. At a general level, descriptions of pottery fabrics and vessel shapes are frequently equated with inherently fuzzy cultural concepts, such as ‘Roman’, ‘Romanising’, ‘Romanised’, ‘Belgic’, ‘native’, etc.; the terminology of long since discredited archaeological cultures is persistently used, e.g. site-type names such as La Tène and Aylesford-Swarling; wares may be described in terms of historically attested regional groupings (e.g. Durotrigan ware) or modern administrative boundaries (e.g. South Devon ware, North Kent grey ware) with equal likelihood; and in some cases be associated with specific social groups, e.g. Legionary ware. In the majority of examples, these labels no longer carry explicit representational meaning among the practitioners that use them – they are instead a form of short-hand that has been retained for practical reasons (i.e. the need to ensure compatibility with older reports). Nevertheless, the problems caused by retaining such labels arguably go beyond those associated with mere clumsy terminology. These include the perpetuation of hierarchies of preconceived value and importance (from widely circulating ‘Roman’ terra sigillata and ‘imported pottery’ to inferior ‘Romanising’ coarse wares), which runs the risk of conditioning interpretation at the level of individual site narratives and regional studies. Categories of objects are treated as known quantities or passive indicators, so that all the analyst needs to do to scrutinise a social process in a given period and region is to build up a big enough database of objects.

Confronting representation

A representational model of material culture is not inherently wrong. This is true especially since material culture’s representational role is increasingly accepted as complex, context-dependent and fragmented. Concepts like ‘discrepant identities’ have proven useful in their emphasis on the multivocality and situatedness of material culture (Mattingly 2004). Moreover, a representational reading of artefacts can be an appropriate strategy in response to certain questions. There is no doubt that a tombstone communicates and reflects at least some aspects of identity (if not of the deceased, definitely of those commissioning it), or that statues rely precisely on a representational mechanism (although this may be more complex than hitherto acknowledged, see Trimble 2011). Our charge against representational approaches is not that they are methodologically unsound, as can be argued of the culture-historical approaches, but that they are partial. The mechanism of representation does not exhaust how material culture works. Study of artefacts centred on representation tends to privilege certain things in specific contexts: often the special, the new, or the visible; and mostly focused on distribution patterns and consumption practices, which, on analogy with our modern experience of being cut off from production, seem more directly ‘expressive’. If, for instance, a certain type of artefact has its ‘core’ distribution area in the Danube region, and one example is found in a burial in Britain, then this seems like a particularly ‘meaningful’ case in representational terms.

But even a tombstone or a statue does not only act as a signifier for an identity, value, or memory. Both tombstones and statues are also visible, solid, relatively durable, difficult to move around, etc. These aspects are the kinds of information that artefact specialists master so well. Specialists see, handle, measure, weigh, and touch artefacts in all their detail and specificity. Materials are provenanced, their properties noted and even experienced ‘first-hand’. Those material properties tend to be approached with a largely implicit, ‘common-sense’ instrumentalist model of material culture, according to which artefact properties are marshalled by people as befits a specific goal. Someone could for instance choose to harness the hardness of a particular rock in order to use it as a grinding stone. But when it comes to writing historical narratives, a representational model is used, in which artefacts’ material qualities do not contribute to historical interpretation. At their best, representational readings are cursorily informed by artefacts’ material properties. For example, the rarity of a particular resource may be seen as linked to a certain object’s exotic value. But reconstruction of the regime of value in which this object was set to work tends not to be based on the object’s material properties. For example, when Dressel 1 amphorae are lined up in a Welwyn tomb in Britain, their interpretation as prestige goods draws first on their scarcity and that of their assumed contents (wine), and much less on their attendant material properties and contextual associations with other objects (Millett 1990, 29–33; but see Poux 2004; Pitts 2005; Dietler 2010).

In order to escape the culture-historical model archaeologists felt the need to break the direct representational link whereby ‘pots’ stand for ‘people’. The alternative, however, has not been to rethink representation as the sole mechanism for linking artefacts and historical process, but to turn to a more complex kind of representation on the model of text (Hodder and Hutson 2005; Buchli 1995). The relation between a textual sign and its meaning is arbitrary and depends entirely on its position in a grammatical and semantic structure. The word ‘dog’, for instance, has no a priori relation to the concept ‘dog’ – hence the great variety in vocabulary between languages, so that words with spelling and sounds as different as ‘dog’, ‘hond’, and ‘chien’ can all refer to the same concept. On those premises, there would be nothing intrinsic to a particular artefact that would steer its meaning. Things accordingly get divided into physical matter plus social meaning (Van Oyen 2013, 87–8). Following this model, the ‘meaningful’ aspects of material culture are located outside the objects themselves, and added to their passive, stable, in itself ‘meaningless’ physical substrate. To repeat, our aim is not to disprove this model on theoretical grounds (for which, see Hicks 2010). Instead, we want to focus attention on what falls through the cracks of a representational reading of artefacts.

The denying of historical significance to artefacts’ material qualities is an important issue, not only because it impoverishes historical interpretation, but also because it severs the link between artefact studies and bigger narratives. The textual model of representation does not grant conceptual space to the specificities of objects, and, therefore, to the key parameters of artefact study. In the linguistic analogy, neither the spelling or typography nor the pronunciation of the word ‘dog’ steer meaning in any way (except through their arbitrary relations with other words). Moreover, texts are on the whole not handled, measured, weighed and touched as part of their interpretation process in the same way as archaeological artefacts. As a result, it is hard to find a space for the contribution of specialist artefact analyses in historical narratives predicated upon a representational template. The continued labelling of objects along categories and principles long critiqued on theoretical grounds is a case in point of the resulting incompatibility.

In its search for a more complex representational model than ‘pots equal people’, Roman archaeology thus turned to a textual analogy loosely inspired by post-structuralism. Structural concepts like socio-cultural ‘identity’ or economic ‘growth’ became the mould wherein artefacts are analysed and made to speak – e.g. the ‘expression of social identity’ (Eckardt 2002, 26). The reification of such postulated invisible phenomena has been repeatedly critiqued on theoretical grounds in the last decades (most notably by Latour 2005). While anchored in a more general tendency in social theory, the issue reveals itself particularly acutely in representational studies of material culture. If objects are assumed to represent external meanings, these meanings have to be attributed to and grouped by some external causal force. This leads to a form of reverse engineering by the analyst, whereby ‘structural’ categories are inserted at the start of research instead of being the outcome of analysis, e.g. ‘elite’ (Millett 1990), ‘Italians, soldiers, Gauls’ (Woolf 1998), and ‘the military community, the urban population, and the rural societies’ (Mattingly 2011, 223). This is not to say that ‘elites’ or ‘soldiers’ were not historical realities. Rather, it is an urge not to reify such postulated structural phenomena a priori, but to dissect what they were made up of and how they worked in order to avoid the danger of circularity.

Closely related to the reification of social categories is a final issue with the representational approach to artefacts, which is about the kinds of causal forces it invokes in history. The problem is neatly demonstrated by the perpetual bone of contention of how to make historical inferences from distribution maps: does the presence or absence of a certain artefact on a certain site reflect supply mechanisms or choice (e.g. Gardner 2007, 91)? In the former case, big economic structures seem to take over in history-writing, whereas the latter interpretation creates a past made up of human agents making conscious choices along the lines of popular discourse about identity today. While few scholars are fully invested in either model, the problem lies in the difficulty of choosing between them. This is not merely an issue of the limited detail of our data; instead it is symptomatic of how a curtailed concept of how artefacts work forces the analyst into an unrealistic choice between two extremes. Simplification of how material culture works, then, is likely to result in a reductive causality in history-writing. To avoid this, rendering the mechanism of representation more complex is only one step, which needs to be paired with an acknowledgement and exploration of other possible ways in which things can be said to be involved in history.

What is a thing and what does it do?

The last decades have seen a surge of interest in material culture across disciplines as varied as literature, history, art history, anthropology, and sociology. Given its particular expertise and long history of dealing with objects, archaeology could have a major contribution to make to this material turn (Olsen et al. 2012). While each discipline comes to material culture studies with its own historically constructed goals and questions, all emphasise that material culture does not just work in a representational way.

In order to balance the dominance of the question of what objects mean or represent, the recent material turn has focused on asking what objects are and what they do. Historians and anthropologists have shown that the distinction between humans and things, or society and nature that seems self-evident in the modern West is a historical construct, and is not universally shared (Fowler 2004). Meanwhile, cognitive and ecological studies emphasise the difficulty in pinning down the boundary between humans and their environment: brain, body and world do not just ‘collide’ (Clark 2008), but may well turn out to be inseparable (Malafouris 2013). Human perception cannot exist except through material mediation (Gibson 1979; Ingold 2000). These insights challenge our ontological categories of ‘humans’ and ‘things’, and push for an inquiry into what exactly things bring to the ontological scene. Ingold’s call for not writing away the material qualities of objects strikes a chord with the above analysis of how representational approaches risk neglecting the specific properties of artefacts (Ingold 2007; Murphy, Van Oyen, this volume).

The question of ‘what objects are’ is answered more radically by studies subscribing to the so-called ‘ontological turn’ (e.g. Henare, Holbraad and Wastell 2007; Holbraad 2007; Viveiros de Castro 1998; critique by Heywood 2012). Anthropologists recounted how in some societies rocks can be people, or jaguars can be people. The traditional response of the western analyst is to revert once again to a representational model, ascribing these different categorisations to culturally specific, constructed meanings overlaying a universally shared and singular reality in which people, jaguars, and rocks are ontologically distinct. In this view objects may well represent people, spirits, or immaterial things, but this is not really what they are. The ontological turn questions this representational solution on political and methodological grounds and urges the analyst instead to turn to the fundamental ontological categories in her research. In contrast to our Western ontology in which nature is stable and culture varies, in other ontologies the same essence can be shared by different natural forms (e.g. Descola 2013). However, the ontological turn in archaeology has yet to transform from a critical into a constructive project capable of generating new knowledge (for steps in this direction, see special issue of Cambridge Archaeological Journal 19.3, 2009).

A more productive alternative seems to be to move from the question of what objects are to what objects do. As agency is no longer considered an inherent propensity, be it of humans or of things (Robb 2010), some other criterion has to be defined for ‘what things do’. This criterion can take different forms, but one of the most workable variants proposes that agency (sensu what things – or humans – do) be linked to effects on the course of action (Latour 1999; 2005, 71). Ontological status then features at the end rather than the beginning of the analysis, and is predicated upon the shape and modality of these effects (cf. Van Oyen 2015). The link between effect and agency is only one possibility, and different criteria have in the last decades led to a whole spectrum of views on material agency, some more ontologically radical than others (Van Oyen 2016a, 1–3). It is important to note here that, while many frameworks start from the ontological uncertainty just identified (i.e. not a priori assuming that things are what we understand them to be from a modern Western perspective), none so far has argued for a resultant material agency imbued with intentionality and reflexivity, and most stay close to fundamental principles of Western ontology in their interpretations (cf. Robb 2004). Concerns about the terminology of material agency, as raised for instance by Andrew Gardner at the seminar that led to this publication, are valid, but they cannot always be taken to indicate fundamental incompatibility (e.g. compatibility between a Latourian material agency and the dialectical framework of Bourdieu has been suggested in archaeology by Maran and Stockhammer 2012).

The question of material agency and its modalities is therefore not the main concern of this volume. Instead we are interested in how the creation of interpretive space for mechanisms other than instrumentalism (things as tools) and representation (things as signifiers) in dealing with things can improve historical narratives and can lead to new insights into the Roman world. The adoption of ontological uncertainty as a starting point for research cautions that the way humans relate to things is not uniform across time and space. Different kinds of persons and different kinds of things emerged from different ways of structuring this engagement (already in Foucault 1975; cf. studies on personhood, Brück 2001; Fowler 2004). By challenging the boundaries of bodies, identity, and memory, Emma-Jayne Graham (2009), for example, has shown how M. Nonius Balbus, major benefactor of Herculaneum, emerged as a literally larger-than-life person through social commemoration after his death. His personhood was no longer defined by the boundary of his skin, but dissipated across the city’s monuments and topography, including memorials literally and metaphorically indexing his transformation from patron to ancestor. An answer to the question ‘what did things do in the Roman world?’ promises to inform at the same time on the specificity of human-thing relations and on the kinds of persons in that world.

Such inquiries should, in turn, shed new light on that other aspect of how material culture works: what and how things represent. Current Western society, for instance, in which few people produce objects themselves yet many consume large quantities of objects whose links to production have been obliterated, creates certain kinds of things and specific kinds of people. As a result, some objects are more predisposed than others to representing for instance status, value, or power. We have already noted that consumption is traditionally considered an area in which the expressive role of material culture is particularly salient. While this may well be the case today, it is the historically specific result of a particular rendering of the human-thing relation, and cannot a priori be extrapolated to other times and places.

What different kinds of historical narratives can we expect to follow from a move beyond instrumentalism and representation as the sole modalities of human–thing relations? Starting from ontological uncertainty, the boundary between description and explanation becomes more porous. The case is well illustrated by considering the role of ‘practice’ in historical explanation (Bourdieu 1977; Giddens 1984; Ortner 1984). Ever since the development of post-structuralist practice theory, ‘practice’ has been a staple term in archaeology (Dobres and Robb 2005; Dornan 2002). In practice theory, it features as the intermediate-scale dynamic driver of the dialectic between structure and agency. In analytical terms, it is the ‘how’ of agency, but in Roman archaeology, it is often thought along functional categories, e.g. practice of ‘eating’, ‘cooking’, ‘constructing’, etc. (e.g. Gardner 2007). As far as artefacts are concerned, then, such a functional reading forces them into an instrumentalist role, where they are always already for something (for eating, cooking, constructing, etc.) (see organisation of the contributions in Aldhouse-Green and Webster 2002 and Allason-Jones 2011). Because such categories are pretty universal placeholders, both practice (the ‘how’ question) and objects are denied much explanatory value. Instead, an additional explanatory level is invoked over and above everyday practices, be it a habitus à la Bourdieu, status, power, or some other deus ex machina acting as prime causal mover.

Traditionally, then, objects can be causally involved in history, but only insofar as they slot into human schemes of purpose (instrumentalist) or meaning (representational). Once we widen the spectrum of mechanisms for how material culture works, however, causality changes and the question of ‘how’ (i.e. practice) is granted explanatory value. The point is neatly made by an example of ethnography of hospital practices (Mol 2002). In a general practitioner’s consulting room, the disease of atherosclerosis manifests itself as ‘pain when walking a certain distance’. Under a microscope, diagnosis of the same disease relies on ‘×% blockage of the arteries’. These different definitions are not wholly due to a different habitus of ‘general practitioners’ versus ‘surgeons’ for instance – some sort of external causal force. Instead, they are shaped by the practices of the settings: the possibility to talk to and touch the patient in a consulting room; the presence of a microscope to visualise the interior of the arteries in the laboratory. These differences in practice themselves have explanatory power: they explain discrepancies and negotiations, for instance in coming up with a diagnosis and deciding on the best treatment. The question ‘what did objects do in the Roman world’ is therefore no mere trivial addition to existing narratives; it can fundamentally change the dynamics, causality, and agency in historical explanation.

Representation refined: artefact biographies and networks

This book aims to convince the reader of the need to expand and diversify our interpretation of material culture, but also hopes to provide some tools for realising this theoretical move. Just as we are looking for a more complete theoretical model of how material culture works – adding to but not precluding instrumentalism and representation – we are not jettisoning the tried-and-tested analytical tools of archaeology. The fact that typologies long outlived the usefulness of the culture-historical paradigm shows how tools can be deconstructed and repurposed as interpretative models change (Van