The Echo of Past Choices: The Roman dining table under twenty-first century scrutiny

Steven Willis

Cite this as: Willis, S. 2018 The Echo of Past Choices: The Roman dining table under twenty-first century scrutiny, Internet Archaeology 50.

1. Aims and Research Foci of the 'Big Data on the Roman Table' Network

Unlike many gatherings and publications of subsequent proceedings where Roman pottery has been the subject material, this volume has matters of dates, fabrics and techniques of production as the background. The aims of the Network, Workshops and this volume have focused upon the means by which the vast and growing information on Roman pottery can be harnessed to provide answers to enduring questions and suggest new avenues of enquiry.

Tablewares were omnipresent across the Empire, with the output of certain manufactories widely distributed and emulated. Relations between producer and consumer and the contexts of their use (what the people of the Empire in their own circumstances did with them) are lines of enquiry pursued in the articles here.

First it is necessary to ask why was there such a vast output of tableware, take-up and widespread emulation of pottery types? Production of certain tablewares in Italy and Gaul, for instance, was on a truly industrial scale, amounting to kiln loads of tens of thousands of vessels, with, in turn, provincial sources such as Cologne seeing very wide distribution of their products, while at a 'local level' numerous potters were replicating forms, though whether we should call these 'copies' is perhaps a matter for debate. Thirty years ago Romanisation theories suggested this was because communities 'wanted in' on what Rome could offer; there are general doubts about such a view these days. Were copies simply cheaper affordable versions (cf. Leleković, this volume), or did scarcity lead to a local response to fill the shortfall. For Leleković, this imitation and evident appetite for tableware could be showing change in foods and consumption choices, which, if correct, is a most significant development. More broadly we can propose that such shifts were a local elective choice: agency was in operation individually and among communities, but within a wider framework of circumstances shaping possibility, decisions and expectation (what Allison (quoting Dietler), this volume, refers to as the 'social and cultural logic underlying choices'). For Bermejo Tirado this means 'consumers of Roman tableware in provincial households managed their own selection criteria of consumption, probably explained by specific local or regional traditions' (Bermejo Tirado, this volume). Local copies were an expression of the 'demand of the market' but the nature of the responses and reception varied across geographies, site types and between status groups. So while there is often a certain similarity or familiar general appearance to Roman pottery, province to province, just as food types were shared, within and beside this was diversity in practice to do with location, ethnicity, community and status; mapping these elements is a firm outcome of the projects reported in this volume.

Function, rightly given the themes of the network, is a matter around which contributors circle and swoop and hence, unsurprisingly, pottery form is closely considered in several studies. This is no abstraction, nor simply an erudite matter, but relates to key questions in the phenomenon that was the Roman world: culture and consumption, dining and practice, how they ate what they ate, burial rites and beliefs. Over twenty years ago we were reminded that in the Roman world 'You are what you eat' (Meadows 1994), but of equal significance was how you consumed it (the relations of consumption). As Allison highlights in her introduction to this volume, this goes to the heart of the relationship between the object and the mind, the interplay of understanding, action and the material form.

Many of the spheres for enquiry examined in this volume have, hitherto, hardly been explored. Large catalogues, spreadsheets and data records coupled with fresh ideas, advances in technology and computer applications are bringing forth significant results as the contributions to this volume demonstrate. There is an experimental heuristic element to some of the new work but driven by the desire to address research questions with the expectation that by applying new techniques enlightenment will follow.

A major theme across the projects presented for discussion at the Workshops was their attempts to reconstruct how and why the data is as it is. These studies rightly raise all sorts of questions about the Empire and what people within it were doing, about the organisation of production and distribution, consumption and turnover, burial and site formation.

'Big Data' affords innovation through using older records and in generating new data resources; as the articles show, by revisiting existing empirical records, authors find that it pays to call on these 'old friends'. It is often acknowledged that contemporary accomplishments in research build from the work of others rather than arrive out of the blue. The BDRT initiative, and the project work it showcases, has the strong shoulders of previous research into Roman pottery to build on. Commentary upon this therefore needs to include some element of contextualisation that places the BDRT initiative within the framework of past and existing challenges, approaches and experiences.

2. Big Data are Served

There is no question that Roman pottery study has generated Big Data. This is in terms of 'large numbers' but also in the sense that its variables can be logged so as to be amenable to forms of empirical analysis, increasingly via digital technologies and application of computer programmes. Hence at the Leicester and Exeter Workshops Rien Polak and Ryan Niemeijer discussed their work mapping pottery finds from the Hunerberg site, Roman Nijmegen (the Netherlands – Appendix C), Nijmegen being a multi-faceted site where analysis of c. 2 million plus sherds has been undertaken, while the Römisch-Germanisches Zentralmuseum in Mainz (RGZM) samian database which holds nearly a quarter of a million records of stamped vessels from 3700 sites, was called upon in the presentations of Allard Mees and Geoffrey Dannell (see Appendix C). These numbers are small compared to, say, present-day supermarket data collection on when, where and in what quantities its apples and bananas are being purchased or analytics and algorithms relating to customer buying habits and inclinations at the swipe of a loyalty card. There is power in numbers and useful knowledge if the information has been appropriately collected, analysis structured and the right questions asked. Like Big Data generated and scrutinised by contemporary companies and organisations following purchasing and activity habits, Big Data studies for Roman ceramics are more than the sum of the parts fed in. This is because, like the study of banana purchases, pottery uptake was not amorphous acquisition but comes with a complex configuration that will in some way reflect: the need or wish to acquire it, intended functions, the possibilities available at 'the market', etc. (cf. Millett 1987; Willet this volume). Approaches and methodologies are then needed to distil the multifarious interplay of structure and agency behind the deposits of discarded pottery fragments that have 'come down to us' as archaeologists.

Big Data has been with us for some while, as can be readily seen in some other domains of the Human Sciences. Cases in point include the statistical processing work of the Institute for Social and Economic Research, University of Essex, drawing on UK Census and other data, work focused on social trends and informing policy in recent decades, for instance by Profs David Rose and John Goldthorpe (Essex and Oxford Universities respectively) and the on-going Big Data project led by Prof. Eddy Higgs, also at the University of Essex, examining historical UK Census data. Equally Big Data have been part of archaeology for decades, though we did not call it such, nor process the data in quite the way we do today. Now engagements with Big Data are a very fashionable field (e.g. Creighton 2014; Cooper and Green 2017) and will presumably remain a popular frame of reference in the years ahead. Big Data past and present, in the sphere of Roman pottery, have a common root in that it will be the product of a great deal of organisation and collection in potshed or lab. So, for archaeologists very large numbers of sherds and issues around making sense of them are nothing new; excavation and survey of Roman sites has routinely yielded crates and crates of ceramics requiring much consideration about how to deal with the scale (e.g. Orton 1979; Fulford and Peacock 1984; 1994; Evans 1991; Faiers 2005, 57-8; the current project of the Soprintendenza Archeologia of Piedmont analysing the 800 boxes of pottery from the excavations at Augusta Taurinorum, as outlined at the Leicester Worksop by Stephanie Ratto and Alessandro Quercia; see Appendix C). In the first eleven years of its existence, a period of intense urban rescue excavation, the Colchester Archaeological Trust excavated an average of over one ton of Roman pottery from the town per year (Symonds 1982) but much more was to then emerge from the Culver Street Phase II excavations, 1984-5 (Symonds and Wade 1999). Pottery specialists were among the first archaeologists to harness computers to record, sort and analyse. This was the case, for instance, at Colchester where computer entry of pottery records and of quantified data by count, weight and EVES by type was in place from the beginning of the 1980s, though the publication was to appear (not atypically) much later (e.g. Symonds and Wade 1999). Now, however, we have both the huge cumulative records from past studious cataloguing and a recognition and drive to explore this resource via software applications in order to view it from new angles and ask fresh questions, many of which we hitherto have not been able to posit. The scale of the potential data to harness and investigate is enormous and the papers presented at the Workshops and in this volume show how scholars have taken the opportunity that these circumstances afford. Computer programmes are hungry for Big Data and a set of researchers will happily feed that need with 'raw data' on tablewares.

Archaeologists have been crunching numbers, via statistical analyses, particularly from the 1960s and 1970s (cf. Orton 1982), while the 1980s saw significant uses of Principle Components and Nearest Neighbour analyses in Roman pottery research in Europe. Accordingly, one question that might be asked at the outset, given that the subject matter has not altered in those 50 years, is whether looking at the presence of types and their potential uses and the quantities involved are just more of the same – or the same but in larger numbers. An important development has seen moves away from showing just numbers but using these as a stepping stone to new ways and forms of understanding, via visual media: visualisations such as those generated for this volume by Martin Sterry.

3. Applications to Study

Scrutiny of the articles published here reveals what the Network members regard as Big Data (or associated with Big Data), and its potentials for exploring pottery distribution, consumption patterns and practice. Several articles highlight research drawing on archive and accumulating data records and harnessing that information via new techniques to address undoubtedly big questions, beyond the purposes these catalogues were originally intended to serve. These might be seen as bridging projects, employing existing resources as a result of the possibilities new approaches and software enable and suggest, linked to the types of questions the individual researchers and the wider community of scholars are interested in. The contributions of Mees, using the RGZM samian database, Willet, on production and supply in and beyond Sagalassos, Sterry, mapping two types of finds data, and Marshall and Seeley, employing the MOLA 'spot-date row-records' are of this type. Certainly big numbers of items can be called upon for interrogation in these cases, the first mentioned having an interregional (interprovincial comparison), the second regional in scope and the latter for studying supply and consumption to and across the largest Roman city in Britain. In these instances Big Data is about the 'big picture' and the picture at 'Big Sites' (rural sites receive little attention among the contributions to this volume).

Centrally binding is the fact that the Roman world was not simply a political mega-state but at many levels a closely integrated economic and cultural entity such that it can be compared to contemporary expression in globalisation (Hingley 2005; Pitts and Versluys 2015). Unity and diversity were key features of the Empire: we see this in the ceramics, as in other material forms. A challenge is to map and conceive how that occurred given the long distances and the particularity of regions (cf. Christmas and Pitts this volume). Hence an integral part of the Big Data picture when applied to the Roman world is the matrix of interconnection of influences, emulation, deep-vein continuities and degree of 'sameness' that enable studies over such a broad landscape and time. Parallels exist in related fields, as with Ray Laurence's mapping of Roman stele data across the provinces (Laurence n.d.). Returning to the focus here, the family tree of Roman pottery types, so to speak, had many close cousins and lines of legitimate and less legitimate but associated descent that resulted in a series of connections, similarities and shared style; this has enabled studies of a 'compare and contrast' nature across the provinces, as seen in the articles I mention above in this section and in others such as those by Leleković and Luley. There are, however, limits to the likeness in vessel types and appearance in the Roman world, just as there was variation and regional character to food types and diet alongside some widely shared practice (consumption of olive oil, wine and fish sauces). As Willet observes, relating to globalisation, this 'does not lead to a convergent and homogenised singular culture alone, but to greater local heterogeneity as well … Instead, the resulting patterns of tableware at different places throughout Asia Minor show a complex interplay of similarities and differences in the shapes of vessels consumed' (Willet this volume).

While, therefore, Big Data may provide some view of the 'big picture', at a different level it provides the backcloth against which to gauge and characterise the particular. A case in point is Cooper et al.'s article on exceptional contexts from Roman Leicester: exceptional for their coherent integrity and 'tight dating' (that is to say, representing accumulated discard close to the point of consumption and use, with a structural association, over a relatively short period) shedding light on the character of catering and consumption 'hotspots' in the city. Their unusual character is emphasised by first comparing to Willet's observation of the 'relative paucity of well-studied closed archaeological contexts' his research was confronted with (Willet this volume; see also Luley this volume on this matter). Secondly, Big Data assists in providing an awareness of the normal picture at Leicester, of mixed assemblages arising from excavations at many other sites across the townscape, compared to which (for Cooper et al.) these particular cases present as abnormal but readily interpretable on the basis of their composition. As Richard Reece's work has demonstrated, big datasets can thus provide the touchstone for much archaeological enquiry: what is normal and what is exceptional, leading to questions and answers as to why (cf. Reece 1995; Willis 2005a, section 2.1).

Big Data relates also to new theoretical approaches and new methodologies where the latter may not be about large assemblages of pots and sherds per se but the digital data collected by new technologies such as object scanning, drawn image scanning and analysis, photographic imaging and analysis, Reflectance Transformative Imaging (RTI), etc. (e.g. the articles in this volume by Banducci et al. and Christmas and Pitts). There are various potentials therein. Firstly, in some cases fairly big data is run through statistical and mapping packages to generate fresh ways of observing the data, its elements and variables. This may be as an aid to conveying information, for analytical purposes, for discerning patterning and as an experiment to see what useful outcomes arise (e.g. Sterry this volume; Polak and Niemeijer's investigative spatial analysis at Nijmegen presented at the Leicester and Exeter Workshops: 'Quantitative and spatial analysis of tableware from the canabae legionis at Nijmegen' and 'Tableware from the large Augustan camp on the Hunerberg, Nijmegen', see Appendix C). Secondly, this also points up the question of data management systems and accessing pottery records in the digital age. The possibilities raised by the photographic method outlined by Tyukin et al. (this volume) include this being employed as a search tool for finding sherds, such as rims of a 'similar to' nature, within a researcher's dataset or a museum collection; this will be something especially handy when assemblages are in their thousands or more.

4. Terms of Enquiry

How data are presented can both inform and influence thinking. Some standard approaches appear in this volume alongside newer tools. A heuristic and experimental aspect is apparent in several of the contributions. That is to be expected when new techniques are attempted. Luley, for instance, gives us an idealised set of types in an assemblage (Luley, this volume, fig. 11) reminiscent of the Weberian 'Ideal Types' approach, while Sterry is explicit in outlining some areas needing better resolution in the Correspondence Analysis he adopts. Banducci et al.'s wear analysis looks to a 'predictive model'. Van der Veen's 'flowchart to function' is a new means to assist in categorisation. This is a tool used by some of the most influential researchers of the 1970s and 1980s who framed much of the prevailing approaches to Roman pottery: as seen in the flowchart diagrams of David Peacock and Clive Orton. It is interesting to note, for instance, that an article I cited earlier by Orton includes an algorithm for paralleling and identifying fabrics that is a flowchart not a million miles from van der Veen's Figure 1 for functional type (Orton 1979, 71, fig. 34). There is, then, through the attention to quantities and scale, models and the socio-economic preoccupations, continuity on many levels in the methods and approaches with those of the empirical Processualist age. That is nothing to be shy of, as long as the studies and results are considered in the light of twenty-first century awareness. Indeed, van der Veen points out that while there was an 'ideal use' for the pottery form in the mind of the potter, its actual use is another matter (van der Veen, this volume) which is a rather Post-Processual point recognising the role of agency.

New computer power and applications are to hand but their consistent employment by day-to-day practitioners is more uncertain. Biddulph (this volume) highlights some of the challenges of developing his sample while the results achieved by Sterry (this volume) are doubtless the outcome of years of specialised study and development of expertise. Expertise and 'staffing' needs are a requirement before some of these works migrate from the grant-aided research project to more routine practice among a wider set of those who work with Roman pottery.

New questions are framed and addressed on this Big Data stage but clearly old questions too are suitably revisited by these means; not all can be answered yet, even with Big Data and the churn of such datasets. However, that, realistically, is not the purpose of the Network or its Workshops.

5. Data Collection: How was it for you?

A standard and often advisable entrée when discussing an archaeological subject is to outline the nature of the evidence, often under an explicit dedicated sub-heading. This approach is followed by several articles (as with Luley, Mees and Willet), where the aim is to recognise and sketch what one may be dealing with: areas of strength and weakness, lacunae, provisos, etc. Archaeology, as we know, is all about samples; samples are the raw material for another archaeological staple: comparative analysis. The latter should provide results, the strength of which is in proportion to the representativeness and character of the samples used. It is true that so often we use less than perfect samples but this is well recognised and reasons should be understood. It is not surprising too to see a concern among the contributors to the Network Workshops and this volume regarding the nature of the samples available to work with, given the composition of pottery samples is, itself, a key determinant of findings. Often this is a function of site formation processes and realising what lies behind them (cf. Luley this volume), and sample selection issues. Some causes are out of the control of archaeologists but advances can be frustrated when the sample available is not perfect owing to archaeological factors: the way the data were collected and recorded in the first instance. Issues of this sort arise with the locational recording of the finds from Vetera I (Sterry, this volume), or seen in the question of 'lumping' or 'splitting' (Cooper et al. this volume). Cooper et al. consider potentials and limitations of sample composition and taphonomy; Luley's groups lack the strong link to structural remains that are somewhat of a holy grail. Our search for answers is often complicated by the lack of primary associations since households were kept clean: they 'swept up'. We should expect a complicated picture: even well-collected samples will often have their drawbacks, and a certain 'fuzziness' might be anticipated. However, concerted research can find patterns: keep the faith with the archaeological record and it will often reward investigation with useful outcomes.

Big Data should not, ultimately, be about amalgams and it should not be regarded as a fuzzy cloud of stats: Big Data should not be at the loss of clarity. Big datasets need good structure if they are to be useful, with information recorded systematically to a set standard (cf. Marshall and Seeley this volume on 'qualitative entries' in the spot-date records for London); when existing datasets have not been designed for twenty-first century interrogation, restructuring and refinement is possible with care to capture (recapture) past archives for new uses (cf. van Helden et al. this volume). The key need is to be aware of the representativeness, reliability and veracity of the evidence to begin with.

An inherent stumbling block is one common in the sciences and social sciences, namely incompatibility: groups the researcher may wish to add or compare their data to are found to have been recorded by different methods, often methods that the researcher thinks are not the best. This is a fate that often bedevils the zealous first- and second-year PhD student, which might be summed up as 'sample unsuitable: return to go'. Biddulph points this up as a difficulty for his study, stating 'for the purpose of comparison, the data [available] must serve' (Biddulph, this volume). Sterry (this volume) refers to the 179 phased groups that I employed as a sample on the samian project: finding suitably recorded data in the subject literature was a significant challenge, as I noted in the original publication (Willis 2005a, 5.2.2). This important matter of standards and shared methodology was a focus of the first Workshop at Leicester (e.g. the presentation of Jeremy Levesley, Appendix C) and is addressed by Colley and Evans (this volume).

6. 'Ways of Seeing' and their Inhibitors

One of the many positives of the Workshops and articles here are the graphic presentations of information: stylised groups, diagrams, plots and maps, some of which have a novel and stimulating look. Readers can develop their own view as to how helpful they are. That will depend on what they show, how they show that information (supported by textual explanation) and how they are used by the particular authors. Some of the analyses display trends and information we are already aware of, or suspect, but form new perspectives that can confirm, clarify and lead thought (e.g. articles by Cooper et al., and Sterry). These may be tested via comparison with other significant indicators for verification (or otherwise) and be subject to statistical tests.

Conceptualisation and reconstruction of past practice (in the ancient world) are often-evoked goals of the archaeological researcher and in this volume there is accordingly attention to what functions the pottery forms were employed for and what their role was on the Roman table and in funerary rites (e.g. articles by Dannell, Dananai and Deru and Banducci et al.). Yet we cannot know the minds of the ancients, what they were thinking in the compartments of daily life when the vessels were used. We can, however, search for the ways – repeated ways - that they 'did things', such as to generate the 'objectscapes' that Martin Pitts spoke of at the Exeter Workshop ('Tablewares in funerary objectscapes. Styles of consumption in the Roman west', Appendix C). The situation is rather like that with Iron Age coins: there are plenty enough of these items recorded but what they were used for is a matter of much discussion; we have a pretty good idea that many had a final use as votive/religious gifts, but we do not know what their uses were as a 'life assemblage' prior to deposition (if any), though we can see some patterns in the locations where they occur. On the basis of such patterning in the material record (coins, pottery, etc.) inferences can be forthcoming (cf. Prof Christopher Hawkes' 'Ladder of Inference' (Hawkes 1954)). In the Roman world this was by habit and norm, often rule-governed and rehearsed such that the renowned Roman archaeologist and numismatist John Casey often (in a light manner) refereed to the 'rigor mentis' of Roman society: practice got 'stuck' in a cycle of replication with limited variation, or that was what the cultural framework often resulted in. While this is a caricature it catches something we see very often in Roman manifestations and materiality. Gidden's 'Structuration' theory (Giddens 1984) is relevant here.

7. Stop-press: 'Form and function in crisis talks'

One of the striking features of the articles in this volume, and something often raised in the workshops, was the question of how vessels were used, and the relationship between form and function. This is a particular focus of several articles in Section 2 in this volume (e.g. by Bermejo Tirado, Dannell and van der Veen) following presentations at the Leicester Workshop (such as those of Laura Banducci and Xavier Deru, Appendix C) but this uncertainty threads through many contributions: what the Drag. 27 form represents pops up frequently, with Dannell's 2006 paper common to the bibliographies of several articles. Perhaps there should be a sub-group formed: the society for the investigation of the Drag. form 27 (small bowl or cup; sauce holder or something to drink from; the same applies to Drag. 18/31, commonly regarded as a dish but which Luley not unreasonably classifies as a plate). We might wonder, however, if discussing the use of modern labels (with their associated ethnocentric baggage) is the best use of resources and time, and I am not convinced that trying to find the 'actual use' is what archaeology is primarily about, besides being as elusive as the Scarlet Pimpernel, or, to repeat Dannell's term, just abstract. This comes back to more fundamental questions around the 'types and categories' we create and understand as archaeologists and how these relate to ancient mind-sets: how valid is this? Mutability and the context of use are recognised by the contributors to the Workshops as key. Yet there is a tension here between, on the one hand, the astonishingly consistent standardisation of tableware forms and types of the Roman world and the strong discrete, confident, category classifications they give rise to (for us), and on the other the consequences for our cataloguing where this meets the imperative to speak of functions. The results of residue analysis, RTI and scanning may provide some answers to usage but there are inherent limitations in terms of feasibility of universal application of such techniques and the fact that high-fired gloss slip tablewares were not often absorbent and were resistant to marking with, say, knives, if indeed such potentially damaging instruments came into contact with them in use. Given these circumstances the approach developed by Bermejo Tirado (this volume) is all the more pertinent. Relevant too, the Exeter Workshop was reminded of progress made with residue analysis on Roman vessels via Lucy Cramp's presentation ('Organic residue analysis and Roman pottery: past, present and future applications for understanding vessel use', Appendix C; also see Cramp et al. 2011).

There is one samian ware form that we might have more confidence in knowing its use, the inkwell, Ritterling 13 (Willis 2005b); this is not a tableware for the dining room but it seemed like 'The last form standing' at the Exeter Workshop. Form and ink staining point to its normal use, but there has long been the question of why this form was made in samian ware when the latter was normally for dining vessels; then there is the question of their size, for they would have held a great deal of ink, with capacities much larger than Roman inkwells made of metal (Hella Eckardt, pers comm.); could they sometimes have held salt?

As Dannell's contribution, focused around the samian docket graffiti at La Graufesenque highlights, there are problems with what the ancients wrote down. This spotlight serves to point up a familiar pattern of silence in ancient documentation (for, as in so many areas there is little written), as well as uncertainty in this case as to the significance and degree of common usage etc., of the labels of pot types implied by the La Graufesenque graffiti. Where chance epigraphic records or references survive and surface from the Roman world they do not always seem to match the much fuller archaeological record. Yet there is a traditional deference in Roman studies to the written records, such as they are. Problems include attempting to find a general 'truth' by using fragmentary documents, the precise purpose of which is not understood (as in this case and, for instance, with the itineraries of places). Given the gap between the inadequate and small written record and past practice, there is, alas, no short cut for solidifying our understanding: archaeology will need to come up with its own answers based on context, association, attributes (such as vessel shape and wear) and close detailed study of its site-finds.

A traditional partner in such studies is measurement, which so often has been the safe 'fall-back' position of the archaeologist. Does it offer some salvation in this instance? The articles by Baddiley and Christmas and Pitts have this as a particular focus, but others also discuss functional categories in terms of ratio of vessel height and width etc. Christmas and Pitts' study results in a rather elegant, visually helpful scatter plot (this volume, fig. 8), while Baddiley concentrates on capacity study, something hitherto under-examined in Roman pottery study other than for amphorae. Baddiley (this volume) turns to Graham Webster's classic 1969 Romano-British Coarse Pottery: A Student's Guide (a Council for British Archaeology (CBA) booklet still having value some fifty years on) to consider the criteria for functional classification categories wherein height to width is a major defining element. All this attention to form-function reminds me of an inquiry long ago as to 'What is it?' when Clive Orton put pottery classification under the microscope (Orton 1982, 34-6), taking Webster's 1969 definitions and John Gillam's 350 types in Types of Roman Coarse Pottery Vessels in Northern Britain (originally published in 1957) and finding 'good agreement' between what might be termed the intuitive approach of Gillam with Webster's definitions. A rather helpful template expressing classification by the height-width ratio criterion was included in Orton's assessment (Orton 1982, fig. 2.8); these days this might be 'ghosted' over a pot drawing on screen to establish what classification the particular pot lies within by this measure. This is a part-solution in so far as it provides a standard for comparison, but actual function remains unresolved by this mathematical expression. Size measures are significant for understanding volumes and therefore proportions and servings at the table, though there remains the question of 'seconds please' and the 'topping-up' of beverages; do we really think that the option for more was not on the menu at formal and less formal Roman meals? Perhaps that is why Leleković (this volume) is sceptical on what some trends in size and form may suggest regarding use. Can such measurement provide a direct road to functional use/understanding of functional use?

The above notwithstanding, whether it was ever a good idea in the first instance for archaeologists to ascribe functional names to the pottery of 2000 years past is questionable. The typologies we use are artefacts themselves, in many cases with grafted additions, replacements, duplications, gaps, unused classes, mistakes, etc. such that if we were starting afresh we would almost certainly not start typologies along the same lines. That said we have nonetheless got to a place where we know or can look up the types say of Dragendorff, Stuart, Holwerda or the Conspectus that are 'useful' and have survived the testing years and weeding of types. The existing typologies we employ are at best neutral identifiers (providing convenient labels).

The point was made above that there is a deep, wide and long-lived aspect of standardisation, similarity and replication in forms seen across the Roman world. There are two qualifications here. One is that within norms there are variations, such as a typical samian bowl form being replicated in the provinces but fired or slipped to a black/dark grey finish rather than red (as with London ware and the PSW types characterised by Leleković) together with some variation in vessel sizes and devolved decorative schemes. Secondly, when we find items to do with weights and measures, capacities and sizes, item names (as with the La Graufesenque dockets; see above) we find that the peoples of the Empire were not quite the rigid formulaic Romans we imagined, routinely following imperial standards; as Baddiley points out (this volume) the capacities he records do not fit with known Roman measures just as studies of lead weights rarely conform to any known Roman standard (Hunter 2013, 103). Similarly, attempts to match ancient label names to actual pottery forms is far from straightforward (Dannell this volume etc.). This is important to know, and it raises many questions, for a mismatch between ancient word-label categories and actual manifestations to such a degree is not what we would perhaps expect. Ironically, this rather mirrors the long and on-going call for adherence to similar compatible methods of quantification in measuring quantities of Roman pottery and the desirability of standardised recording and approaches in general as seen in A Standard for Pottery Studies in Archaeology (PCRG 2016) and the article in this volume by Colley and Evans.

Readers will make their own judgements on the discussion of form and function in the articles of this volume. It must be noted here that much good work has been published that systematically records proportions of functional types (as conventionally understood), often within site monographs dealing with individual assemblages and where there is at least internal consistency as the pottery researcher year on year has followed an established method. The work of Jeremy Evans is a case in point (cf. Evans 1995). As long as it is clear what shapes are being recorded as what category, readers of those reports can rework the data if they wish.

More generally the Network learned that flagons might be employed as kettles (Dananai and Deru), which for many was a new realisation (though entirely understandable in the context of Temperate Europe!).

Of course size and capacity might be something of a red herring with decorated pieces, where the function is not primarily that of holding substances in a certain measure but rather display, where size may have been about display and fitting on a striking decorative scheme, though this may be true more normally for metal vessels rather than fine pottery.

8. Visualisation: the plotting

Archaeology is a highly visual realm. This extends to the display of its huge data elements, rich in variables and attributes, its spatial aspects and phases. These have long been illustrated by graphs and maps. Measurable attributes and spatial phenomena are conducive to the new visual display portals for presenting the results of digital processing. As we see with this volume, visualising Big Data is a key means by which authors can show findings (often the work of teams or processing or reprocessing of existing data) for others to comprehend, with significant possibilities for out-reach to wider communities (cf. Allison this volume). There is flexibility too in terms of layering, turning off and on certain information types, creating plots, discrimination and greater subtlety of what is/can be displayed once the data are in the database. Much of the visualisation and data capture in this volume is innovative, and comes out of relatively sophisticated expertise. An issue that arises is that the plots can be challenging for those not familiar with the metadata or perhaps the method and so the plot may seem abstract to the viewer (cf. Sterry this volume). However, archaeology remains a small subject by comparison, with many basic questions and gaps still in need of attention; in other words running is fine but in many areas we are barely walking. This being so the more fundamental need is often simply straightforward data presentation as a staple, as in a presence/absence plot or a histogram. That said, I greatly welcome Martin Sterry's work on the data I collected on samian ware which moves beyond, as he states, 'conventional graph types' (Sterry this volume). Or, to put it another way, his spatial plots provide an alternative and strikingly constructive set of patterns that Willis (2005a) otherwise used tabulated data and words to convey.

Simple solutions are often good solutions. Dananai and Deru's schematic figures (fig. 12) showing pots of different type coloured up makes them 'active', raising them out of the ordinary black-and-white line drawing familiarity that somehow now seems rather passive by comparison.

The nature of our methods of displaying data in archaeology inevitably imposes itself on our thinking so this domain is one for serious consideration. Are methods used our analytic friends or foes? How we present the data visually can restrict or determine our thinking by insinuating itself on the interpretative options. Diagrams and visuals often need guiding captions and annotated text explaining their background, veracity and limitations, yet often readers will dip into reports and browse the images and leave with an impression, not a thorough digested read from beginning to end (more of a trend now than ever in terms of the use of digital media). Hence selection of options in visualising data are crucial, and for the consumer of the figures, the significance must be weighed.

9. Producers, Markets and Consumption

Marketing and the nature of production, supply and consumption were unsurprisingly a concern of the Network members. These relationships are specifically pursued by several contributors (e.g. the articles by Dannell, Luley, Mees and Willet). This is welcome as this area has received comparatively little attention in recent years, and one hopes this will be an incentive to fresh enquiries in this key field. Clearly the conclusions remain provisional and there is still work to do. Even where the name stamps on samian ware might be thought to provide an insightful 'tracer' from production site to market, shedding light on the relationship between the two, we actually see a complex scenario with enduring uncertainty around the nature of 'marketing' (Mees, this volume; for a recent discussion on such issues, for example, Van Kerckhove 2015).

One operating assumption among the contributors is questionable and a potential weakness. This relates to the seemingly universal implicit view that the pottery (all pottery types) was readily to hand for all potential consumers and to place in graves. Yet I wonder. Scarcity or shortfall in supply is not mentioned. Perhaps often pots were not something you could just take off your own shelf or a shop shelf to replace a breakage or to extend your service. This thinking of unproblematic acquisition assumes that 'the boat always came in'; in other words there were constant reliable supplies that met needs and that pottery was affordable. We know from several indicators that samian was not cheap (and that is why ownership marks are more common on samian than any other type, not least at military sites where the chances of loss in a large communal setting were high (cf. Evans 1987)). Was this sometimes or endemically an economy of scarcity (cf. Cooper 2007), which may have impacted on prices and explain why Luley (this volume), to his surprise, encountered more black-slipped fineware but less terra sigillata (samian ware). In other words, supplies of fineware were seemingly diminished; perhaps there was simply often not enough arriving to meet demand. It is important to remember how production of pottery occurred at this time, the comparatively small scale and limits of most workshops, even if there were many producers (cf. Peacock 1982). Locally produced Usk ware may have been made to meet requirements when imported supplies were insufficient (cf. Baddiley this volume). Counter to this perhaps is the infrequency of repairs to table or other wares away from truly remote locations and the curious fact that eras of high fineware importation are associated with the greatest levels of regional and local production of fine tablewares too (at least in the early Roman west as Greene (1982) demonstrated, implying peaks in fashion when one might have expected local industries to be filling the gap in periods when output from the samian etc., workshops was low, as in the first decades of the second century CE). Then there are issues of both differential rates of turnover of pottery (Orton 1989), touched upon by Cooper et al. and of curation (Willis 2005a, section 5.8).

Tablewares were in common use beyond the triclinium. The evidence from Leicester highlights these vessels in use at a specialised food dispensary (probably not that rare, for we can imagine there being several such enterprises in Leicester alone) and the likely tavern deposit (Cooper et al. this volume). Various other venues will have hosted feasts and festival events, dining and trade guild/club dinners. Fast food in the urban setting is well attested in London (e.g. Bow Bells House period 3 for a recent likely example (Howell et al. 2013, 19-21)) while the 2010 excavations at the Beaney Institute site, Canterbury, adjacent to the Forum area, had a similarly indicative structural, faunal and ceramic signature that could point to phases of fast-food service for those about their town business (Diack 2010). At Leicester the closely dated and integral nature of the deposits is particularly compelling (cf. above). Seemingly the catering trade was big business then, as now.

Studies of the inclusion of tablewares and settings in funerary contexts are familiar enough but are given fresh impetus by the work of Dananai and Deru and Biddulph reported here (see also Pitts, Appendix C). The on-going investigation of formal rites and ceremonies, including the consumption of food and alcoholic drink at the time of burial and included in the grave, carries much potential to shed light on practices that hitherto have been quite superficially considered. Not least, normative practice (tradition) meets agency (choice decisions) in such deposits, with ideology and status strong components of the grave furnishings equation: no wonder such studies attract interpretation!

With the discussions of pottery and burial rites we need to be cautious and avoid the assumption that those buried immediately outside Roman towns were solely the urban populace, rather than being from the civitas as a whole; that really is, at present, an open matter. Moreover, rural communities seem to have had some proportion of fine tablewares but seemingly were more diligently careful with them and did not use them for 'everyday' meals (as they are rare in normal site settlement layers) and thus there was less vessel turnover. What is surprising about the St Pancras cemetery, Chichester, is the lack of samian; while findings from rural contexts show that during the first and second centuries samian was comparatively well represented in graves and special deposits (Willis 2005a, section 9).

The BDRT has aimed to promote new ways of considering Roman tablewares and move out of conventional grooves. Turning the focus on the distinctive Lyon Ware pottery of the period c. 40-75 CE perhaps we may advance some new suggestions (cf. Allison, considering the site of Vetera I). Firmly associated with the Roman military and perhaps senior administration, as seen from its distribution in Britain (Willis 2003), one now wonders if these beakers and cups and occasional tripod bowls and lamps were dedicated accoutrements of imperial worship and festival celebrations, the cult of Augustus particularly, or relate to a military 'officer's club' style of use (see Dannell (after Birley), on the frequent ceremonial and religious festivities necessitating formal celebration).

10. Micro-scrutiny: something is stirring at the bottom of the pot; it's personal

There are undoubtedly significant aspects to be discovered in wear patterns seen on ceramic vessels. Provisional work is cited by Marshall and Seeley in their own important discussion included in this volume (section Various examples might be used to illustrate this recent progress such as Monteil's finding that internal wear on both large and small versions of the samian ware 'cup' Drag. 33 indicated both were used for 'mixing' (Monteil 2013, 367). As Marshall and Seeley state in the case of the MOLA records, where recorded this is often the simple record of presence or perhaps absence. Even when details are recorded regarding where the wear occurs (on MOLA 'rows' and in various publications where this is reported) this may be general and inconsistent in descriptive and locational terms. Clearly there is much more scope for standardised recording of location, character and intensity of wear (cf. Marshall and Seeley, this volume; Biddulph's paper at the Leicester Workshop 'Why quantifying Roman pottery matters', see Appendix C).

Evidence retrieved from RTI, object scanning and residue analysis linked to more systematic recording of macroscopically observable phenomena such as cut marks, scrapes and incising through stirring holds much potential (Banducci et al. this volume). This might enable the imprint of an individual to be discerned through a distinctive wear pattern.

The custom of considering pottery at an assemblage level and the imperative to prepare a report for publication and distil the general patterns out of large numbers has perhaps closed down the scope for considering other more individual and personal interactions with pots by their ancient users. The idiosyncrasies seen occasionally on vessels and sherds that are sometimes even found repeated on other vessels, take us beyond the sherd to the human. In present times we can see that people become 'attached' to particular items of crockery, mugs and cups etc. and these can be highly personal relationships. More work may find similar results to those highlighted by Ellen Swift, referenced by Banducci et al.. To the individual user, familiar and less familiar pottery has personal, sensory dimensions, tactile and experiential, and may be evocative and part of their meaningful worlds. We can comprehend that probability but only rarely can we begin to approach that through the study of pottery fragments from antiquity as such traces are few (and we probably routinely overlook what is there) and at best may have to be inferred by, say, inclusion in burials.

11. Conclusion: The echo of past choices

This volume suggests something of the new scale researchers can work to in the immediate future: what it may be possible to ask of the data and, via digital software, see visualised. New tools enable us to approach questions anew with Big Data and to explore best use of existing records. That said, it is necessary to be realistic: Big Data itself will not provide answers per se, though it can be harnessed towards that end. As Mees' contribution points up, its exploration may not provide definitive answers, is not necessarily able to show the 'big picture', is not necessarily 'clearer' in outcomes simply because it is large scale, but the answers should be stronger than at present if the samples are robust and given the increasingly sophisticated awareness that can inform interpretations.

Will the type of processing work seen with the articles of this volume become more widespread and normal? It requires resources, including time and some specialist expertise in techniques. The uptake of Correspondence Analysis since the late 1990s is a sign of what may be possible. However, specific funding to pursue some of the potentials witnessed in the BDRT Workshop presentations and in this volume might be sought since most work on pottery is an individual specialist undertaking, so a wider team and resource will almost certainly be required on projects where there is Big Data to process.

In order to understand the past better and to convey such understanding to a wider audience, an idea or models of the social context of pottery use is likely to prove expedient (beyond the idealised reconstruction of the Roman kitchen). Allison sketches unfolding scenarios of the context and use of the ceramics at Kinchega homestead in her opening article. While there is a speculative element, the interpretation is related to the evidence apparent through other variables and information types. An absorbing picture emerges that might serve as a signpost to others in developing archaeological reconstructions: it is worth the risk to break out of the narrow 'hard empirical fact', chart and plot approach that is so standard for practitioners in pottery studies to a more socio-anthropological paradigm. Hence it is pertinent that this interpretative approach comes first in the volume but is not something matched in later articles, despite the relative licence afforded by the nature of the BDRT Workshop for exploratory approaches. Why is this? Is it because the historical proximity of the Australian case makes it more comprehensible to us? I doubt that there is so simple an explanation. The challenge is there to develop more interpretative explanations that provide a bridge to wider audiences to access this past via the common practice of taking food and drink. These narratives will do better justice to the immensely rich archaeological remains that have come down to us from the Roman world and to the innumerable hours researchers have spent cataloguing 'the record' as we have and employ it. This is the true legacy data: one of ancient past choices, agency and human practice, so promising to explore now.

12. Prospect: Current practice and future choices

As I mentioned in the introduction to this discussion, the focus of papers presented at the established Roman pottery conferences (such as SFECAG, the Aardewerkdag, the Fautores meetings and the annual SGRP conferences) have shown a strong tendency towards matters of reporting types, chronological understanding, production sites and distributions; in other words 'updates' pertaining to the 'nuts and bolts' for processing and reporting pottery assemblages. The BDRT Workshops have provided a platform for the presentation and discussion of new approaches and accordingly there is a freshness, excitement and challenge in what the ideas and methods hold. I mentioned above too, the tendency in the Roman world to 'inertia', often commented upon by specialists; moving to present times perhaps degrees of inertia are likewise seen in the way dealing with Roman pottery has become highly conventional. Articles in this volume invite the subject to move out of the constraints of these narrow conventions. Customary practice is embedded and perhaps practitioners are locked into methods that, while strong and successful, are at the same time channelled and limited. Recording a greater range of attributes and phenomena present or absent on tablewares is likely to prove beneficial (cf. Marshall and Seeley this volume). The prospect is raised to think differently about the potentials of the sherds and data; and for these resources to be mobilised in new ways and, via means such as visualisation, to traverse pathways to different (better?) understandings of the role and use of pottery vessels in and around the Roman table.

The BDRT initiative has at its core an imperative to engage and communicate: engage and communicate information and, as the various contributions here testify, to show how data can be worked with in fresh ways to transfer information and ideas, and, not a little, to inspire. On the one hand there are the rich assets of the culture of the past, in various forms, and on the other a series of opportunities that are now opening up for the study of that material.

Study of the Roman era is of course implicitly an international forum with so much that is similar (yet different!), shared and contrasting, region to region, site to site. Information exchange and efficient communications are essential. Contributors to the Workshops reflect that broad constituency of shared interest. The common themes and issues articulated through the Workshops and seen in these articles was a striking feature of the meetings, as was the manner in which this could be fully communicated between scholars from quite widely different backgrounds. This bodes well for the future. So in terms of current ideas, issues, aims and priorities there is a common vocabulary and narrative. This common 'language' will be further enabled if the subject, through its practitioners, responds to the call for shared practice in recording; the 'common standards' for these will unite and enable comparison by common denominators (as Colley and Evans re-state in their article). One can see how shared categories (taxa) and methodologies have permitted other subject areas to advance (as in the spheres of environmental archaeology). Common standards will enable the creation of international datasets, facilitate comparison and engender wider exploration of the subject. As Colley and Evans show (sections 3 and 4), there is some way to go even in Britain, but a move to confluence in approaches only requires a shift in degree, not kind. The adoption of shared practice in recording etc. is a choice to be encouraged. There is a community beyond the Network aware of the value of such developments and ambition to take studies forward.

Shared methods will mean that not only can valid comparisons be made, say, between terra sigillata assemblages from forts on the continental Limes and the northern frontier in Britain, but between site assemblages as a whole. An earlier attempt at just this was the Caesar Project, comparing pottery from three Roman towns on the littoral of the north-west provinces: London, Bordeaux and Irun, with valuable findings that warranted a higher profile (Symonds 1998; 2003). More widely read has been Perring and Pitts' study showing the scope for such integrated study within a broad region (Perring and Pitts 2013). Now, across the geography of the Empire we see some huge assemblages being dealt with, with methods comparable to best current standards and considering questions that most will agree are pertinent for tablewares as well as other pottery types (e.g. from military sites such as the Kops Plateau, Nijmegen (e.g. Carreras and van den Berg 2017); modern industrial zone developments revealing Roman landscapes (e.g. Clotuche 2009); from Roman towns as at Baelo Claudia, southern Spain (Xavier Deru, pers. comm.)). Projects of equivalent scale may be anticipated in the course of the imminent High Speed 2 railway development through central Britain. The large scale of some of these projects coincides with the availability and application of new software tools, showcased in this volume, which can greatly assist in data management and presentation. The subject is finding ways of coping with the vast numbers being generated by modern development.

Expanding contract archaeology across Europe has been a response to infrastructure development, housing projects, suburban retail and business zones, etc. in places that are not of the subject's choosing, but reactive, recording sites that happen to be in the pathway, and deemed too insignificant to warrant protection. There is a degree of randomness to this sample (though modern development often lies where our ancestors also chose to live and work), which generates a comparative representativeness in the samples forthcoming. The findings of the New Visions of the Countryside of Roman Britain project have demonstrated the value of this work in transforming understandings, scooping data from innumerable fieldwork investigations from c. 1990 onwards: from many interventions Big Data grows. Mobilisation of what might otherwise become 'dusty archives' carries enormous potential with relatively recent work, such as the New Visions project has employed, and with other archives (cf. van Helden, this volume).

Futures and the past can be said to meet in the present. The Classical world, in its Roman manifestation, saw an extraordinary expression and material reality in the spread of pottery forms and consumption practice. This has been systematically studied now for some 150 years with the collected information documented in illustrations, catalogues and narratives. Much is now incompatible with present methods but as we see can be engaged to speak to us in a new tongue through ontological approaches. Long ago the prehistorian John Barrett, invited in the 1980s to comment on the future of Roman studies, observed that there was already (by that time) so much understudied Roman material in museums and archives that a generation of archaeologists might have research careers solely working on such material (Day Conference: New directions in the study of the Roman era in Northern Britain, Durham University, November 1985). In the age of digital humanities the ontological approach featured in the BDRT Workshops and in this volume signals the means to activate archives to new ends. A particular reason why this is important is because a century or more of Roman archaeological practice has, through direct choice, attended to key sites through excavation and survey, these being formative in the development of Roman archaeology. They were chosen because they were seen as crucial to address research questions. For sure some project archives and legacy data are more amenable than others because recording choices made in the past were often partial, unappreciative of the possibilities more systematic methods might hold. Now there is less choice where fieldwork can be conducted, due to preservation legislation and budgetary restrictions. Accordingly, legacy data revisited and released through media such as outlined by van Helden et al. holds great promise (cf. Colley and Evans, this volume). Two aspects are significant. First, these studies, clearly, can be enlightening. New methods come out of new thinking and in turn inform thinking; with an ontology, data can be conceptualised, plotted and mapped in ways that move beyond the conventional primacy of the object/artefact and specific context as units of assessment, but may instead, for instance, be deployed as values and probabilities over space and through chronological phase. There is a new spatial geography to finds study now and ahead of us. Yet, secondly, in order for this to become other than an occasional undertaking – the 'privilege' of the well-funded blue-skies research project - several elements will need to configure: appreciation of the value and relevance of such 'revisiting', time/funding, training and team building. Initiatives such as the New Visions project (cf. above) and the BDRT show funding can be won for innovative work on past datasets. The BDRT initiative speaks to the existence of shared research goals across a variety of types of practitioners and suggests a model for avoiding two archaeologies: that of the exceptional well-funded research project, and that of the workaday processor/pottery report. Matters are not quite that stark, for as articles such as that of Cooper et al. demonstrate, contract archaeology can provide headline insights on life and practice in Roman worlds, though again, here, collaborative working makes for enhanced findings. Workshops on skills acquisition in these areas of working with archives, visualisation and new software tools may be a suitable way forward, noting that Colley and Evans (this volume, section 9) highlight English Heritage's support of strategic projects aiming to make better use of commercially recovered data. Communication, collaboration, inclusivity and outreach are key themes of the BDRT Network and, going forward, this will remain vital for ceramic studies.

The discussion must end with the raw material of study. The Roman world, like our own, was composed of a myriad of small worlds, populated by individuals and groups making choices in the context of options, options influenced and circumscribed by the wider cultural world and economics. Consumption is a personal practice; as Dannell stresses, what happened in and around the household table will have varied, indeed greatly varied, through different spheres and identities within society. Nonetheless the variability occurred within bounds with many shared and repeated (cultural) practices. This formative background gave rise to some remarkable levels of homogeneity and patterning that run as encoded strands through the archaeological record. There is complexity to that record, as we see in this volume, but within it there lie strongly intelligible configurations that fall to us to make sense of.


Internet Archaeology is an open access journal based in the Department of Archaeology, University of York. Except where otherwise noted, content from this work may be used under the terms of the Creative Commons Attribution 3.0 (CC BY) Unported licence, which permits unrestricted use, distribution, and reproduction in any medium, provided that attribution to the author(s), the title of the work, the Internet Archaeology journal and the relevant URL/DOI are given.

Terms and Conditions | Legal Statements | Privacy Policy | Cookies Policy | Citing Internet Archaeology

Internet Archaeology content is preserved for the long term with the Archaeology Data Service. Help sustain and support open access publication by donating to our Open Access Archaeology Fund.