Big Data Analyses of Roman Tableware: information standards, digital technologies and research collaboration

Sarah Colley and Jane Evans

Cite this as: Colley, S. and Evans, J. 2018 Big Data Analyses of Roman Tableware: information standards, digital technologies and research collaboration, Internet Archaeology 50.

1. Introduction

The 'Big Data on the Roman Table' (BDRT) research network was established to promote collaborative research into Roman socio-cultural practices by studying ceramic tablewares associated with eating, drinking and serving food. An important BDRT aim was to promote research into how such tablewares and their assemblages were used and the possible social, cultural and symbolic value of such use (Allison, this volume). Many BDRT studies published in this volume involve large-scale intra- and inter-site analyses of vessel forms and functions, quantitative methods and digital data visualisation.

Material remains from the ancient Roman world are found in numerous countries and archaeologists have excavated and recorded information about vast quantities of pottery, including tablewares. Making best use of and comparing existing data from across the ancient Roman world is a BDRT objective that requires local, national and international collaboration. Designing new ways of collecting and analysing future data, and incorporating existing data, requires knowledge of how pottery, tableware specifically, is currently classified, indexed and contextualised and knowledge about data formats, quality, accessibility and preservation (i.e. archaeological information standards). This includes understanding how and why current data have been collected and are organised, including the use of particular typologies, 'type series' (see below) and any existing standards and guidelines. Understanding and applying information standards are crucial for online data sharing and research collaboration and for developing and applying 'big data' methods discussed elsewhere in this volume such as ontologies (van Helden et al., this volume) and quantitative analyses and digital visualisations (e.g. Cooper et al., this volume; Tyukin et al., this volume; Christmas and Pitts, this volume; Sterry, this volume).

Ways of collecting, analysing, publishing and archiving information about pottery in general, and tablewares specifically, are variable and have changed over time (see e.g. Timby 2016; Evans et al. 2016), yet archaeologists need to be able to share data that is at least minimally comparable regardless of its context of production to better understand socio-cultural practices and their commonalities and diversities across the Roman world. Digital information also needs to be accessible to researchers in formats that are interoperable, scalable and sustainable (Richards 2008) to ensure preservation for the future regardless of technology change. Flexible and open data structures can accommodate change and expansion of the scope and scale of projects (Kansa et al. 2010), yet open access and open data approaches in archaeology raise their own technological, economic and professional issues that also require consideration (Huggett 2015; Moore and Richards 2015). Accessing information and being able to assess its reliability, significance and context also requires attention to metadata (Mason 2007; Huggett 2015, 20-21). People and organisations involved in producing and using digital information need resources, skills, knowledge and motivation to make best use of it (Wenger et al. 2009; Colley 2015), the motivation springing from an understanding that applying well-chosen methods will produce better results.

Our article discusses such topics as they relate to the BDRT research network, drawing on articles published in this volume, discussion at two BDRT workshops (Allison, this volume), a BDRT questionnaire survey, and with reference to recent strategic initiatives in British archaeology. The BDRT network mainly concerns tablewares. To gain a fuller and more reliable understanding of Roman socio-cultural practices, such tablewares need to be studied in the context of evidence for consumption more generally; other associated vessels and objects used for food preparation and cooking, the context in which tablewares are found or not found, the associated environmental evidence and residues revealing the types of food consumed. Many of the issues raised here therefore have implications for wider areas of study.

2. Typologies, Type Series, Standards and Guidelines

A typology is a set of rules to categorise archaeological artefacts into meaningful groups based on attributes. Typologies embed particular theories and methods that may be implicit or explicit, narrow or broad. A type series is a collection of examples of each 'type' (as defined by a typology), useful for categorising artefacts by direct visual comparison or by mathematical analysis of attributes (cf. in this volume Tyukin et al. and Christmas and Pitts). Standards are agreements developed by professionals or groups of scholars to, for example, collect information based on a particular typology and to a minimum level of quality. Guidelines assist practitioners to implement standards by providing more details and information about how to use them.

The BDRT volume editors have outlined the main tablewares discussed in this volume and used throughout the Roman world (Appendix A). They have also provided a glossary of some of the terms and abbreviations used by volume authors for Roman tableware pottery typologies, 'type series', and some specific tableware shapes, with relevant resources (Appendix B). Additional terminology and resources are described and referenced in individual articles. Some typologies and classification systems, mainly those used to describe internationally traded wares such as terra sigillata (e.g. Dragendorff, Curle, Conspectus and other examples in Appendix B), are widely known and used internationally. Some are so embedded in research practice they have entered a lexicon of shared terms that may require no further explanation or bibliographic referencing among some groups of pottery specialists. Archaeologists also use pottery terminology and taxonomies that are not widely used. These may be project specific, describe pottery with a limited geographical distribution or be based on modern languages that archaeologists from other countries may not understand without translation and concordances (e.g. Symonds and Haynes 2007). This volume has articles published in English from native speakers of at a minimum French, Italian, German, Spanish and Croatian, as well as English. Typologies and terminologies also continue to expand and be developed (e.g. for typologies of 'use' in this volume: Van der Veen, Bermejo Tirado and Marshall and Seeley).

Shared and differing terminology and typologies used by the BDRT volume authors demonstrates some of the complexity of understanding how different kinds of Roman tableware are currently classified at local and international level (see Symonds and Haynes 2007; Perrin 2011, 9, for discussion of UK issues). In the UK typologies, type series and reference collections exist in different formats including as material collections of pottery in museums, commercial field-units and universities, as hard-copy publications and as unpublished limited circulation 'grey-literature' reports and as online resources (e.g. Allen et al. 2015; Tomber and Dore 1998; Tyers 2014; Worcestershire Archive and Archaeology Service 2017). In some areas such resources do not exist at all (Perrin 2011).

3. A Standard for Pottery Studies in Britain

The development of a multi-period standard for pottery studies in Britain (PCRG et al. 2016) sprang from a recognition that many aims, methodologies and concerns were shared by specialists researching all periods of pottery; prehistoric, Roman, medieval and post-medieval. The three period specialist groups were particularly worried that existing standards were being driven down by competitive tendering in commercial archaeology, concerns articulated by Paul Blinkhorn (2014). While separate guidelines had previously been produced by each period group (for Roman pottery see Young 1980; Darling 2004), it was hoped that a single document would be more accessible for those involved in teaching, research, and community projects, but particularly for those designing and monitoring commercial projects whose expertise often does not cover pottery studies.

The Standard aims to:

These aims support research generally and overlap with those of the BDRT network, whose participants contributed to the consultation period for the Standard. It is hoped, therefore, that this British standard could provide a model for internationally agreed standards, or at least the basis for wider, international discussion. The use of these standards would potentially deliver the quality of data for the kinds of research the BDRT network is promoting. Could, or should, the British standards be adopted more widely, or would they need to be updated to take account of local circumstances and standards elsewhere? They will anyway need to be updated in the future as new techniques, approaches and questions are developed. The standard is shared on the website, as well as the pottery group websites (e.g. Study Group for Roman Pottery 2017). Use statistics show that to June 2017 it has already been accessed from 19 countries, across Europe and including Canada, the USA and Puerto Rico. Even where excellent resources exist, such as the UK National Roman Fabric Reference Collection (Tomber and Dore 1998), they are not necessarily used (Timby 2016). A recent survey has shown that although use of these reference collections is considered essential to professional practice (CIfA 2014), they were only referred to in 45 per cent of the specialist reports reviewed (Cattermole 2017). So standards and guidelines are needed to remind or mandate people to use them. The challenge is ensuring that agreed standards and guidelines are used and implemented; they need to be widely supported. The British Standard has formal recognition from Historic England (HE, formerly English Heritage), who funded the project, and the three specialist research groups who contributed to its development. It is hoped that it will be more formally endorsed by the Chartered Institute for Archaeologists (CIfA), when they develop a process to do this. Importantly, CIfA set professional standards for those who design and monitor archaeological projects, who will ultimately determine the level of work undertaken. Within the requirements of British commercial archaeology, there needs to be a process to ensure that they are referenced in the briefs sent to contractors and in the resulting 'Written Schemes of Investigation' that define the scope of the work. There then needs to be quality assurance when the project is completed, to ensure that standards have been complied with.

Issues concerning standards and standardisation of Roman pottery data in Britain have been evident for many years (e.g. Fulford and Huddleston 1991; Allison 1997; Evans 2001) but have been highlighted by a number of recent initiatives, including two government-funded synthetic studies. The first of these, undertaken by a commercial unit (Archaeology South East) attached to a university department (University College London), focused on town and country in Roman Essex, south-east England. This project drew on a wide range of archive data, including pottery, and aimed to give impetus to improvements in the ways that archaeological data are collected, described and disseminated; the resulting publications demonstrate the potential and problems of using pottery data (Perring and Pitts 2013; Doherty 2015). The second, was a collaborative project between the University of Reading, The University of York (Archaeology Data Service) and Cotswold Archaeology, a commercial unit, and was supported by both government (Historic England) and research (Leverhulme Trust) funding. This project explored the rural settlement of Roman Britain, examining regional and chronological variation. Strategically, it aimed to realise the research potential of developer-funded Roman archaeology in England. The project collated a vast body of data (Allen et al. 2015) and produced results that will inform both research and commercial practice in Britain (e.g. Smith et al. 2016; Allen et al. 2017). However, the complexity of pottery data and the lack of standardisation within the datasets made detailed analysis of the pottery very challenging within the funding and time constraints of the project, and one outcome of the project was a critical review of current pottery practice (Timby 2016).

A pragmatic decision was made not to include a pottery specialist in the core team collecting and analysing data, and only to record basic information about the pottery. This proved sufficient for general discussions but not for detailed analysis, and pottery evidence was rarely presented or discussed during the regional or national seminars. The only initial output of the project related to pottery was Timby's (2016) review of practice. Researchers with a specialist interest in Roman pottery felt this to be a missed opportunity; the great potential of Roman pottery data is widely recognised (e.g. Woolf 1998, 186, and reflected in the articles in this volume), and it provides the largest body of evidence for most Roman sites. The project is redressing the balance and has commissioned pottery papers for the second volume; covering Gloucestershire and Avon (Timby 2017), the north (Bidwell 2017) and the east (Rippon 2017). These, however, are based on more detailed data collated by, and the local knowledge of, the various specialists involved. The project does make the pottery reports included more accessible (Allen et al. 2015, downloads), and there is an interest and intent among pottery specialists to mine these more effectively.

The issues raised by the Roman Rural Settlement Project highlight a wider problem relating to research funding for pottery studies in Britain generally. Major pottery corpora used today, for example for the Oxfordshire (Young 1977) and New Forest (Fulford 1975) industries result from PhDs undertaken forty years ago, when there was sufficient funding to allow researchers time to visit collections and collate their own data. These have not been updated, despite the subsequent accumulation of vast quantities of data from commercial projects and the development of GIS, which makes distributional studies far easier. It is informative to compare this situation to that for small finds. In Britain finds made by members of the public, often using metal detectors, are reported to national finds advisors working for the Portable Antiquities Scheme (PAS 2017), who then make this data freely available and encourage research to justify their funding. At the time of writing the data has been used in 579 projects including 122 PhDs, 147 Masters degrees, and 6 international projects, among which are a number focusing on Roman research. This bias in research funding needs to be addressed.

The need to develop and promote appropriate professional standards in the UK was already being discussed by specialist groups such as the Study Group for Roman Pottery (Perrin 2011; Evans et al. 2016), but these projects added further impetus to the need to define standards, discussed further below. Improvements in practice are being supported by government and professional bodies. The Historic England research strategy includes 'supporting and improving the heritage sector' and 'improving and developing heritage information management' (Miles 2015). With these aims in mind they have funded a number of projects relevant to the BDRT network aims. The first is a review, as yet not completed, of 'Archaeological Reference Resources' used by archaeologists in the study of artefacts and ecofacts, including form and fabric type series for Roman pottery. Another project, commissioned by the Chartered Institute for Archaeologists Finds Special Interest group, has reviewed standards of reporting across archaeological artefacts more generally (Cattermole 2017). Historic England are also supporting the development of online resources, for example the National Roman Fabric Reference Collection (2017) and the Worcestershire online Fabric Series (Worcestershire Archive and Archaeology Service 2017), and is looking to make these even more widely accessible through the Arches open source data management platform. Further support has gone to the development of archive standards within Britain (Brown 2011) and internationally (Perrin et al. 2014).

4. Archaeological Information about Roman Tableware: a BDRT survey

Auditing existing Roman tableware data and current archaeological practice is no simple task, especially at international level. A survey was conducted as part of the BDRT network as a research quality auditing exercise. Twenty-two people, not all were network participants, replied to Survey 1 (before Workshop 1 in September 2015), eighteen network participants to Survey 2 (after Workshop 1) and eighteen network participants to Survey 3 (after Workshop 2 in July 2016). Survey participants worked on archaeological projects located in, or organised from, the United Kingdom, Ireland, Austria, Italy, Netherlands, Luxembourg, Croatia, Spain, France, Canada and the United States, among others. The survey was small-scale and conclusions are constrained by sample size and methodology. To assess current practices and the potential impact of this network outside UK academia, Survey 1 deliberately excluded archaeologists employed as researchers in UK universities in favour of archaeologists working for heritage consultancy companies, as consultants and freelancers or in museums, research institutions and other public, private and community organisations. Most respondents were from the UK so the results strongly reflect British perspectives. Nevertheless the survey provides useful general insights about Roman pottery studies and digital information in an international context. BDRT survey participants answered questions about their project aims, research questions, the kinds of data typically collected and their use of digital methods, quantitative analysis and visualisation. This provides insights into what might be involved in actually developing a viable system capable of supporting sustained large-scale international collaboration, for example using ontologies (van Helden et al., this volume).

In the UK more Roman pottery data is produced by archaeologists working on developer-funded commercial archaeology projects than by researchers employed in academic positions in universities. Work is guided by local, regional or national research frameworks, developed as part of the heritage management process and informed by appropriate specialists (Fulford and Huddleston 1991; Willis 2004; Perrin 2011; Cooper 2007, 38-39). Use of these research frameworks is not mandatory but projects are, or should be, monitored with reference to them. Importantly, they provide a justification for the level of work undertaken and the funding required to achieve this. They are therefore relevant to commercial, university research, and community-based projects. At present there is only limited research using Roman pottery data in British universities, though where specialist knowledge and research funding are available, universities and affiliated organisations should be better placed to produce large-scale syntheses, develop research questions and take international perspectives. The best commercial projects draw on the expertise of pottery specialists to provide analyses, syntheses and discussions of pottery that are highly academic, representing 'leading work' that incorporates the latest ideas and methods. Some larger commercial units, for example Museum of London Archaeology (MOLA) have 'independent research organisation' status, allowing them access to academic funding for more research-based projects. However, most commercial archaeology focuses on very local mitigation projects which, while undertaken with reference to research priorities, are funded within the context of competitive tendering. As a result, commercial archaeologists increasingly do not have funded time to produce wider overviews, though they still want to produce data that contribute to synthetic studies that inform their work. The requirement for this is recognised in the 'Standard and guidance for the collection, documentation conservation and research of archaeological materials', produced by the Chartered Institute for Archaeologists (CIfA 2014). This defines the purpose of finds work as: 'to provide an understanding of societies and their environments, not only at a site-specific level, but also in a local, regional, national and international context'; to create 'a stable, ordered, well documented, accessible material archive which should act as a resource for current and future research'; and to contribute to 'local, regional, national and international research frameworks and policies'. This implies a professional expectation that commercial projects should produce data that could, where appropriate, be used for international comparisons. BDRT survey information from respondents working outside the UK, and information contained in non-UK based projects published in this volume and discussed in the workshops, suggests circumstances prevalent in the UK also apply in other countries though specific details vary and there are some major differences in some countries.

BDRT Survey 1 respondents answered the short open-ended question: 'Briefly describe your usual processes for analysing Roman finewares and tablewares. What questions are you addressing through these processes? '. Various research themes were apparent in the replies, although follow-up questions would be needed to judge if some research questions were more common or more significant than others and why. The research themes included determining the use or function of different kinds of pottery vessels and the activities indicated by pottery (e.g. eating and drinking and funerary practices). Functional analyses of Roman pottery are less common in published literature (Allison this volume). As they were a key BDRT research network focus, the survey results are likely biased in favour of respondents with particular interests in the topic. Other common research themes were chronology and the production, trade and exchange of pottery from an economic perspective, including identifying activities associated with pottery manufacture. Only a few respondents specifically mentioned research questions about site-formation (e.g. the general condition of the pottery, erosion, evidence for post-depositional breakage), which might indicate rubbish disposal methods and taphonomic or depositional processes that are important for understanding data reliability. Likewise, few respondents mentioned questions about social and economic status or broader socio-cultural interpretations about identity, for example. Overall the kinds of research questions nominated by the BDRT respondents concur with those addressed through British archaeological pottery studies more generally (Fulford and Huddleston 1991; Willis 2002, 2004; Perrin 2011; Evans et al. 2016).

Replies to the question 'What characteristics do you record (e.g. in terms of fabric, shape etc. and spatial referencing)?' produced examples of mixed data categories including broad research and/or methodological themes (e.g. chronology, production methods, vessel use, context information), more specific technical information (e.g. rim diameter in mm, matrix porosity, potters marks) and reference to various local, regional or national 'ware' or 'fabric' typologies. When asked 'What quantitative data do you record?' respondents nominated (in decreasing order of frequency): Sherd Count, Sherd Weight, Estimated Vessel Equivalent (EVE) and further variations (e.g. Estimated Vessel Number (EVN), Minimum Number of Vessels (MNV), Minimum Number of Individuals (MNI) of vessels) using different methods. Some other categories were mentioned (e.g. vessel size reconstruction). Terminology was confusing and opinions varied on the value of different approaches. Some respondents only measured e.g. rim diameter and calculated EVEs for 'large coherent samples', otherwise relying on sherd counts or weights. Such variations in standards and approaches to quantitative data recording are common, long standing and not obviously resolvable (e.g. Perrin 2011; Doherty 2015)

The survey asked archaeologists approximately what proportion of their pottery data records were paper-based or digital, approximately when they became digital and about their compliance with any existing standards and guidelines for recording and archiving Roman pottery data. Sixteen respondents (73 per cent) said all their current Roman artefact catalogues were digital. Eleven provided dates when digital recording began (mostly from the 1980s and 1990s but some from the 2000s). A few mentioned software changes over time. Nine respondents (41 per cent) described having a mix of paper and various digital format records, including scans of older paper records. The term 'digital records' was variably interpreted by respondents as direct input of data into software, initially recording information on paper then manually entering this into e.g. database or spreadsheet software and/or scanning paper records into digital text and image formats. Nobody mentioned visual data collected by e.g. drawing, photography or digital scanning. Photography and drawing were only mentioned briefly by a few respondents when answering other questions.

The survey asked 'Do you use and/or have national/ regional/ organisational standards and guidelines for recording Roman finewares and tableware as part of your normal excavation or post-excavation process? If 'yes' please provide some details e.g. whether national, regional, institutional guidelines'. Of twelve UK-based respondents, eight said they used national, regional or institutional standards or guidelines, one only used institutional or in-house standards, one said that no prescribed standards applied to their work and two did not reply. Other respondents based in the Netherlands, Austria, Luxembourg, Croatia and Spain variably used in-house or institutional standards and guidelines, said that no such standards existed or did not answer the question. Some non-UK respondents said that national guidelines existed but they did not use them. Some used online systems with associated data standards that operate internationally e.g. SYSLAT (Py 2016) and the RGZM database. A separate question asked respondents to briefly describe their processes for collating and managing documentation about Roman tablewares (e.g. data and archives rather than physical storage of artefacts) and if particular standards applied. From 22 people replies, 18 made no mention of information management standards and 13 gave no information about archive deposition. Three respondents said they archived their data and documentation in private or 'in-house' collections. Two said their information management complied with current professional standards, and five lodged their archives with a museum, public archive or recognised repository service. Five respondents specifically said they made their archives public through scholarly or professional publications or reports.

In reply to the question 'What approaches do you take to mathematical modelling and visualisation?' 13 respondents did not mention mathematical modelling, three said their samples were too small or otherwise unsuitable, two mentioned 'statistical analyses' (not further specified), two said 'simple percentages' and one mentioned Correspondence Analysis. One survey respondent demonstrated a depth and breadth of knowledge about a range of mathematical approaches applied to their data. Twelve respondents made no comment or mention of visualisation. It is not clear if they never present their data visually or if they misunderstood the question. One person specifically stated they opposed any visualisation as it was 'not factual'. The remaining replies gave examples of particular types of maps, plans, diagrams, charts, visual and 3D reconstructions and image-based outputs created using different types of software (e.g. Excel, statistical software packages, GIS, CAD, 3D modelling software, photogrammetry etc.).

Only so much can be read into these results but a general impression was that a minority of respondents made any or extensive or explicit use of mathematical modelling. About half the respondents were familiar with and used at least some commercially available software to produce some kind of visualisation using archaeological pottery data, although practice varied.

5. Online Collaboration

In workshop discussion, and through survey replies, BDRT network participants recognised the need for minimum standards for pottery recording to share and compare information. However, provision of and adherence to standards was patchy. Some archaeologists also want to record things in different ways. Adherence to 'authorised' typologies and standards can act to lock in particular research approaches or encourage archaeologists to limit the kinds of information they collect and the way they think about 'data' and 'information'. This particularly applies to online systems (Huggett 2015).

Pilot studies by BDRT network members (van Helden et al. this volume), and other archaeological examples (e.g. Binding et al. 2008) suggest that ontologies offer advantages over more conventional relational databases and other software currently used for sharing and analysing information about Roman pottery and tablewares. Ontologies are particularly suited to information that is hierarchically organised, structured in different ways, uses differing terminologies and vocabularies, contains gaps, is ambiguous, and often needs to be interrogated and analysed to different degrees of detail and at different spatio-temporal scales. Ontology-based systems are also important for improving the efficiency of 'big data' analyses of large and complex datasets. The BDRT survey suggests that some pottery attributes recorded by project participants are more easily 'fixed' and quantifiable, while other attributes and data categories are more 'fluid' or interpretative and subject to change as new research impacts on the kinds of information people need and want to collect. There is an iterative relationship between archaeological data, method, theory and practice that ontology-based systems seem well placed to recognise and support. Building ontologies is also not 'just' about creating research infrastructure. The process of developing ontologies, especially as a group exercise, is a very useful and positive way for archaeologists to better understand the nature of their research practice and the relationship between research contexts, methods, theory, information and research questions. Ontologies are designed for networked information sharing and can, for example, enable archaeologists to 'bolt on' their own data structures and combine data much more easily than is possible with relational databases. Using ontologies does require learning new software and a slightly different way of thinking about information (van Helden et al. this volume), but open-source ontology platforms such as Protégé (Stanford University 2017) are freely available online and should present no more challenge to digitally literate archaeologists than implementing a standard relational database application. Thus, differently compiled data can be relatively easily integrated and analysed using ontologies, at little effort and expense. Copyright and intellectual property rights might need to be resolved, though, which can be a particular problem for sharing images and visualisations (Colley 2015; Gibbs and Colley 2012). Data creators need to be credited for their work as well as researchers who use other people's data for their own research (Moore and Richards 2015, 34-6). One issue for BDRT, given the current state and status of digital data resources indicated by BDRT survey participants, is how to fund and implement digitisation of valuable and useful older data that has never been recorded in digital format, and how to incorporate older digital 'legacy data' created with now out-dated technology. Taking a further step and building and maintaining large-scale multi-user ontology-based tools, services and infrastructure would require appropriate funding as well as dealing with design and other issues involved in comparable technology projects but it is feasible to share data ontologically without this further step.

6. Vessel Use, Vessel Form and Shape

The BDRT network aims include study of how tablewares were used. As discussed and demonstrated by many articles in this volume, vessel 'use' can be interpreted with varying degrees of reliability from direct and more 'objective' methods such as usewear; residue analysis; quantitative observations of shape, size, form, design, fabric and manufacture relevant to vessel affordance. More 'subjective' interpretative approaches to vessel use include using textual sources and indirect ethnographic and historical analogy (see Appendix A), and detailed analysis and cross-comparison of patterns in data from archaeological contexts in which tableware occurs.

Current data recording methods can affect the viability of such studies. For example, usewear and residue data about tableware from Roman London (Marshall and Seeley this volume) was recorded inconsistently as 'present' or 'absent'. Details were noted in a non-standard way in the general 'Comments' field of the Museum of London Archaeology (MOLA) recording system. The sample sizes were also small. Ambiguity about the 'precise location, character and intensity of wear' and residues limited what could be done with the data. At least the MOLA recording system allowed archaeologists to write 'open' comments about information that has not yet been formally codified. We cannot set out to codify every attribute about artefacts that anyone might ever want to study to answer every, as yet unknown, future research question. There are also significant barriers of cost and feasibility in applying scientific methods of materials analysis as a routine to large collections of pottery etc. (Perrin 2011, 10). As research questions evolve and new technologies make application of scientific materials analysis cheaper and more efficient, this becomes more feasible in future (Tyukin et al. this volume); Christmas and Pitts this volume).

A vessel's form and shape is used to understand its possible use. BDRT participants used various 'form' typologies (Appendix B) and some articles in this volume discuss links between form and use. There are problems with the way shapes and forms are incorporated and named in existing typologies (van der Veen this volume). Timby (2016) notes that for the UK there are limited form classification series for Roman pottery and that no overall form classification for UK Roman pottery types (and those from elsewhere) exists beyond specialist cases (e.g. Fulford 1975; Young 1977). Christmas and Pitts (this volume) present a pilot project for an innovative way of collecting qualitative information that aims to categorise vessel form using digital data capture and mathematical analysis of image content. Data were captured from digitised versions of standardised technical drawings of pottery vessels originally made on paper, reproduced in print format and subsequently digitised. Accurate capture of usable data relied on the quality and content of the original standardised pottery drawings and especially on information about the scale of the drawing in relation to the size of the original vessel. There are well-established standards to guide technical illustration in archaeology (Adkins and Adkins 2009) and for pottery (e.g. Collett 2012; Biddulph 2014), including ways of accurately representing the size of the objects illustrated. The digitisation process also impacted on the results in Christmas and Pitt's study. To apply such methods in future, archaeologists need to pay more attention to standards for digitising hard-copy and paper records into electronic formats (e.g. National Archives 2016), especially for drawings and photographs to be used as sources of visual data that can be analysed using new methodologies in addition to being illustrative, although distinctions between archaeological illustration and visualisation are indistinct (e.g. Cochrane and Russell 2007; Llobera 2011). Archaeological photographs and drawings always need to include a scale if size of the objects in them is important.

7. Visualisation

Articles in this volume use visual methods to collect, analyse and present information (e.g. scanning, moving images, photography, reconstruction drawings, pottery illustrations, data visualisation or 'infographics'), yet the BDRT survey data (see above) indicated limited use of 'visual methods' by pottery specialists beyond, for example, graphical data presentation tools (bar charts, pie charts) that come with Excel and other software commonly used by archaeologists. Allison (this volume) comments that Roman tableware studies have made less use of digital data visualisation than might be expected. One aim of the BDRT network is to promote the use of visualisation and visual methods. One barrier to implementation is publication and public dissemination of numerous, large and more complex digital images (BDRT Workshop 1 discussion), especially for archaeologists working in commercial archaeology who produce most pottery data and who still need to create hard-copy reports in black and white, although the internet and digital repository services are disrupting this trend. The BDRT network deliberately chose to publish with Internet Archaeology because it is possible to publish many more images and different kinds of images here. However we also experienced problems sharing digital images in running the BDRT network and editing this volume. These included practical limits on the size and number of large image files it was possible to transfer and circulate on e.g. Dropbox and Wordpress, difficulties opening and manipulating less common image formats and ensuring image filenames and in-text labels (as text is still needed to find and retrieve visual content) were accurate and conformed to Internet Archaeology publication and archive standards. None of these problems are unexpected and all are resolvable given adequate funding, better technical expertise and improved literacy about digital images and imaging among archaeologists. However, such practical day-to-day issues, which are not unique to the BDRT network, may partly explain why digital data visualisation is less commonly used than it might be.

Visual practice in archaeology (e.g. technical drawing, illustration, photography, 2D and 3D imaging and visual 'reconstruction') remains a specialist area, despite the advent of digital technologies that make some visual methods cheaper and easier for archaeologists to perform without technical training (e.g. photography, simple infographics, basic GIS mapping). Only some practitioners have visual, graphic design and technology knowledge and skills and the experience necessary to consistently produce archaeological visualisations to a high standard. But what is 'high standard' archaeological visualisation? It is possible to define standards for particular types of archaeological illustration (see above) and computer-based visualisation of cultural heritage, e.g. the London Charter (2009), but visualisation in archaeology raises many unresolved and likely irresolvable issues of theory and practice with implications for questions of quality and information standards (e.g. Perry 2009; Llobera 2011; Huggett 2015; Colley 2015, 24-26). For example, developing and using pattern recognition technologies based on automated mathematical analysis of big data captured from digital images (see Tyukin et al. this volume and Christmas and Pitts this volume) requires thinking further about the nature of 'visual data' contained in, derived from, and interpreted through images. Creating, reusing and sharing visual 'data' requires similar attention to standards, context and metadata as non-visual data. Such issues have already been well discussed for the role and use of photography as an archaeological research tool for 'data capture' and a means of communicating with professional and 'public' viewers and audiences (e.g. Bohrer 2011) and there is now a very large and growing body of archaeological and cultural heritage literature about interpretation, indexicality, 'data' and visual communication raised by digital media and visualisation technologies (e.g. Bonde and Houston 2013).

On a more pragmatic level, some members of the BDRT network (Workshop 1 discussion) were keen to develop further online photographic collections of Roman tableware types as 'reference collections' to aid comparative identification of pottery at an international level. Photographic images have significant advantages for communicating visual information about pottery types between archaeologists who speak different languages and are located in different countries. Such an image system would be very helpful in developing concordances for pottery terminology in conjunction with development and use of ontologies, for example. Building and curating online image collections that archaeologists can use to find, compare and share images in ways that are useful for research is perfectly possible but requires design and other issues to be resolved (e.g. Colley and Brownlee 2010). Photographs and other digital images need to conform to minimum technical standards and contain (or show) key information necessary for archaeological analysis and interpretation. For example, this is easier for flatter and simpler objects than those with more rounded and complex shapes.

More use could be made of images in collecting, recording and interpreting Roman tablewares (and see innovative examples in Banducci et al. this volume), in conjunction with textual and numerical information, if collecting and sharing suitable images became more routine. It is still not common practice for archaeologists to photograph all or most of the pottery they find. Formal pottery illustration drawing is even more restricted. For example, the UK Standards for Pottery Studies in Archaeology only mentions (e.g. PCRG et al. 2016, 5, 17, 39) that archaeologists should identify and separate pieces of pottery that 'require' illustration or photography and that illustrations and photography (including archiving) should comply with minimum technical standards. Most UK commercial archaeology units are not funded to pay an experienced archaeological illustrator or photographer to record more than a small sample of the pottery they excavate. It is more common to cross-reference with images of similar pottery types illustrated in existing reports. Such an approach risks losing the possibility of challenging interpretations of form, or subtle changes in forms identified after particular typologies were created. The imaging pilot studies presented by Tyukin et al. (this volume) and Christmas and Pitts (this volume), and which arose from interests of members of the BDRT network, provide tantalising glimpses of what might be possible in future.

8. Bigger and Smaller Data and Quantitative Analysis

Sample sizes and quality of pottery information (e.g. retrieval methods applied during excavation, availability of detailed spatial and stratigraphic and other information) analysed by archaeological projects reported in this volume are variable. Some BDRT survey respondents commented that smaller excavated pottery collections tended to get less attention when recording particular data attributes and applying quantitative analyses than larger ones. The UK Standards for Pottery Studies in Archaeology (PCRG et al. 2016) allows for different 'levels of analysis' to be applied to different archaeological samples. In some cases it may be decided that smaller samples do not merit the same level of recording as larger ones. A prospect raised by the BDRT network's interest in applying 'big data' methodologies to analysis of Roman tablewares is being able to 'join together' information about or deriving from smaller collections to get a 'bigger picture' (see Cooper et al. this volume; Sterry this volume). In considering information standards and building digital systems to support comparative analysis of Roman tablewares at international level, it is important not to exclude studies and data from parts of the ancient Roman world where less archaeological work has been done and where archaeologists may have fewer resources than elsewhere. What may work in the UK, Germany, France or the Netherlands, for example, may be less applicable or realistic in other countries. For example Leleković (this volume) presents analysis of smaller samples of inconsistently recorded Pannonian slipped wares excavated from sites in Croatia, that have produced interesting and useful results. Baddiley (this volume) uses a small sample of tablewares to say something interesting about use. Other BDRT contributors (e.g. Marshall and Seeley, Mees) faced challenges of making sense of very large samples of data. This is about data complexity, data quality, data granularity and scales of analysis, not just sample size.

9. Conclusions

We have outlined some recent developments in UK archaeology, including the Standard for Pottery Studies in Archaeology (PCRG et al. 2016) document which has implications for pottery studies elsewhere. In Britain Historic England, the government's adviser on heritage in England, are thinking strategically about research and how to make the most out of commercial data. They are funding both reviews of and the development of new reference resources and standards. This reflects the concern of other professional bodies, representing specialists (SGRP) and the archaeological profession generally (CIfA). It also reflects fruitful dialogue between commercial and academic archaeologists, as with the Roman Rural Settlement Project and Town and Country in Roman Essex. Communication and collaboration between all sectors is essential to progressing research, and is beneficial to all; commercial archaeologists want their data to be used and benefit from the new understanding that research produces, researchers want existing data that can be easily accessed and used. BDRT has expanded this dialogue; the needs of international research do need to be taken into consideration when developing standards and resources at local and regional level.

The BDRT survey has provided an insight into working practices internationally, with specific regard to the study of tablewares. This could be developed further, with tablewares viewed in relation to pottery assemblages generally. And are we sure we are identifying them correctly? There are copies of samian forms in coarse-ware fabrics, and other vessel types that may have been used to both cook and serve foods. In terms of wider strategic aims, the approach used by Historic England could be followed internationally, collating more detailed information about existing reference resources and standards across countries involved in research into the Roman Empire. A better understanding of the relevant bodies responsible for defining and monitoring standards for archaeological work across these countries, and how research agendas and methodologies for collecting and sharing information about Roman tablewares are developed and applied in different countries, might assist the BDRT network's aims in promoting research that focuses on socio-cultural interpretations of the ancient Roman world through analysis of tablewares, as well as having wider value for archaeologists working on Roman archaeology at local, regional and international level more generally. Pottery specialists need to communicate with those who curate and manage archaeology. CIfA have a number of special interest groups, for example covering finds, international practice and research, who could engage with these debates and facilitate cross-sector collaboration.

The BDRT network has conducted a successful international collaboration between archaeologists from different countries and representing both academic and commercial sectors, all sharing an interest in the socio-cultural aspects of the ancient Roman world through the study of tablewares in their broader archaeological and historical context. Our article has highlighted some key issues raised by developing and implementing 'standards' of various kinds. These will need to be considered in future developments of 'big data' approaches to Roman tablewares of interest to the BDRT network.


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.