It was with considerable hesitation that I acceded to the request from the journal's Review Editor to review all nine volumes of this remarkable corpus. This will necessarily be a partial and personal view, since I am not competent to evaluate all of its aspects or the full extent of its impact. For this reason, the review is written unashamedly in the first person. I can, however, write from the very particular perspective of someone who has brought to fruition a similar, if much more modest, project: a second edition of the Corpus Vasorum Arretinorum (OCK). For those who do not know it, this is a catalogue of names on Italian terra sigillata, compiled by August Oxé between 1896 and 1943, finally brought to publication by Howard Comfort in 1968 and revised, enlarged and brought into the digital age by myself between 1992 and 2000.Footnote ¹

The contrast between the makers’ marks on Italian and on Gaulish sigillata is instructive. The formulae on the Italian stamps are more complex (often naming more than one person), likewise the ductus (with extensive use of ligatures and abbreviation) and the actual frames and decorative characteristics of the stamps. For this reason, it was apparent to Oxé at an early stage that in order to classify and understand the Italian stamps, facsimile drawings would be essential. They could not be supplied where his information derived from the pages of the Corpus Inscriptionum Latinarum, which had never contemplated that necessity, but where Oxé examined collections in person he took great trouble to make accurate drawings of the stamps (or rather, to have them made: he did not draw himself) and these appeared in his published catalogue. Felix Oswald, to whose work the volumes under review represent the successor, also collected a vast amount of material in facsimile;Footnote ² but the necessity of setting up and printing the work himself by hand excluded the possibility of using this, which he did hope ‘eventually’ to publish in an album. (I wonder what happened to these drawings?) Oswald contrived to mitigate this deficiency by the ingenuity of using the type available to him in unconventional ways, setting letters upside down when this approximated more closely to the character of the potter's signature.

Over the years, the means of reproduction of graphic material has been revolutionised. The advent of the photocopier around 1960 made a huge difference with regard to reproducing drawings that had already been published, and it seems already surprising in retrospect that when in 1992 I embarked on the revision of Oxé's catalogue — despite a conviction that it should be converted into a digital database — I was still initially pasting photocopies of stamps onto cards. The scanning (and processing) of bitmap images at 300 dpi did not become viable until a year or two later, while now scanners are relatively inexpensive personal possessions and we think nothing of managing high-resolution images in full colour.

Brian Hartley's work on the Gaulish sigillata stamps, begun in 1962, was subject to this evolution in the digital environment. His enormous index, assembled on index cards with rubbings and associated drawings, was in due course converted by him into a digital text document. Yet even then, the possibility of incorporating images with the ease of today was not available and, alongside the drawings, he devised a special font which gave greater flexibility in representing the signatures in print. As anyone who has lived through these years of digital evolution will know, the constant development of new software and more sophisticated computers — which often choose to leave behind data in earlier formats — is itself an unwelcome headache for big projects of long duration. Another comparison between the OCK work on the Italian stamps, comprehending in its latest edition some 36,000 vessels signed by 2,500 potters, and that of Hartley and Dickinson on the Gaulish material, comprehending 210,000 vessels by 4,900 potters, readily explains the almost insuperable challenge of bringing the latter to completion. While many of us will remember the multiple occasions on which H. asserted that the catalogue was ‘nearly finished’ (Michael Fulford records one such occasion on p. vii of the introduction to vol. 9), it was surely not least this optimism which carried it as far as it went in his lifetime.

The conversion of this immense volume of work into a published document in nine volumes — itself a daunting task for those named as authors in addition to Hartley and Dickinson — is an achievement which deserves the praise of all users and reviewers, whatever other opinions they may have upon the material. I had occasion myself to discuss it with Geoffrey Dannell, even before the death of H., and the dedication and enthusiasm which he gave to the project (again duly recognised by Fulford in the introduction to vol. 9) were surely crucial to both the inception and the successful completion of this exemplary piece of teamwork. At the same time, it is important not to underplay the constant presence and participation of Brenda Dickinson, both in H.’s work and in that of the ‘rescue team’.

Vol. 1 of the corpus necessarily included a lengthy introduction (1, 1–28) which discussed the authors’ approach to a wide variety of topics relevant to the evaluation of the material (1–2, the difficulty of reconciling information derived from sources of very different quality; 3, identifying the origins of the vessels; 4–8, evaluating dating evidence; 8–9, 22–5, understanding why the vessels were stamped and what the names signify) as well as summarising the incidence of various characteristics and explaining (25–8) how to read the catalogue. This volume was reviewed in this very journal by Peter WebsterFootnote ³ and I do not wish to revisit any of what he said, apart from commenting on the continuing enigma of who the individuals named might be, and why the stamps are there at all. The same question arises in respect of the Italian material, and is equally incapable of any single convincing answer. The famous firing lists from La Graufesenque complicate the discussion of the Gaulish stamps (1, 22–4) — but there are tantalising fragments of such lists from Italy also.Footnote ⁴ The on-line catalogue of Gallo-Belgic pottery by Jane Timby and Val Rigby, with stamps that are sometimes impossible to interpret, adds a further dimension to this difficult question.Footnote ⁵ Up-to-date corpora of stamps on other vessels such as mortaria would surely also contribute to the debate, but might not bring any greater clarity!

Vol. 1 signalled various lists which were intended to be published in the final volume. One of these, a concordance of sites and their related finds repositories, does indeed appear in vol. 9 (9, 367–73). This is followed by five pages of errata in earlier volumes. Considering the nightmare complexity of copy-editing and cross-referencing such an enormous work, printed in successive volumes over an extended period of time, these are commendably few; the redirected cross-references are generally the fruit of final revisions further down the alphabet. This is followed by the usual bibliography (specific, it would seem, to each volume, since they vary widely in length) and by an alphabetical index of potters. The index is differentiated internally in the sense that cross-references are printed in grey. This emphasises, curiously, the fact that these are actually the entries of interest, since the entire catalogue is rigorously alphabetical in any case and the ‘primary’ entries in the index are therefore largely redundant.

Another forward reference in the first volume was an expressed hope (1, 4) to include colour photographs of sample fabrics; there is no further mention of these in vol. 9. This is a pity, but an omission of greater import is surely that of the ‘sites providing reliable or potentially useful dating evidence at certain periods in their occupations’ (1, 4; 27). I have a strong recollection of a discussion of useful dated contexts in relation to Italian SigillataFootnote ⁶ and the surprise expressed when our volume was published that we had not been able to construct a longer list. Many archaeologists take too easily for granted the weight that they place on these fine wares for dating purposes and do not realise that once this prop is removed (since we are seeking to validate the dates of those fine wares themselves), the sites and contexts which are dated more rigorously by other means (clear association with events known from the historical record; dendrochronology; destruction deposits with big coin hoards and the like) are few and far between. Such contexts, once identified, have a utility for dating their artefacts which, of course, extends far beyond fine wares. For a variety of reasons, therefore, it was a disappointment not to find the promised list at the conclusion of the series.

This encyclopaedic work is a product of the early years of its conception, inasmuch as it takes the form of a work printed on paper, with all the limitations which that implies in terms of consultation. (What indexes could reasonably be compiled and included? How might the user search within it in order to identify a fragmentary stamp?) Of course, in the twenty-first century relational databases are supremely equipped to provide all manner of search mechanisms, together with the means to summarise data, generating statistics, distribution maps and the like. Paul Tyers in Britain has been an exemplary exponent of the techniques.Footnote ⁷ He has also been a member of the rescue team. The second edition of the Corpus Vasorum Arretinorum (OCK) was conceived primarily as such a database, and was greeted by reviewers with enthusiasm for its flexibility and its utility. In that database, I used very simple rules for the transliteration of the stamp-readings into characters which would present no particular difficulties for standard searching and sorting algorithms, and these proved gratifyingly powerful when trying to identify fragmentary stamps. Very few characters needed to be recognised in order for the possibility to be narrowed down to a dozen potters or less. The discipline of entering data into a well-constructed database also enforces a much higher level of consistency than one can rely upon when simply writing text. (An aspect of the material under consideration which is not susceptible of exclusively digital dissemination is the facsimile reproduction of the stamps. It is, to my mind, a fundamental necessity that these should be printed on paper, in quantity, so that one may compare them directly — using dividers, if need be — with a new find to be identified. To have the images on screen only would not tackle the issue of scale, and to print them out for each enquiry would be extremely laborious and wasteful. Hence, the book cannot be entirely dispensed with, though it could be reduced essentially to an album of images.)

H. and D.’s work had gone too far and had become too large for the rescue team to contemplate converting the whole thing into a database in the first instance. On the other hand, the price tag on the entire set of printed volumes (currently £740) was bound to restrict its academic life to libraries (and seriously interested libraries, at that). It was therefore obvious that a digital presence (preferably accessible without charge) was desirable. Hence arose the collaboration with Allard Mees of the Römisch-Germanisches Zentralmuseum at Mainz, another impressive evangelist of archaeology in the digital age, to set about the creation of such a database in parallel with the preparation of the print version of the catalogue. The very process of doing so helped to identify errors and weaknesses in the data (see Fulford in the introduction to vol. 9, vii). The Mainz database is intended to be fully functional and publicly accessible in 2016, but already has some freely accessible functionality.Footnote ⁸ It offers a means to identify fragmentary stamps, but the returns from the process may be puzzling (they do not name the potter until you click on the returned findspot) and it is not very flexible in terms of identifying what you have. (Some 350 special characters may be selected in order to make up the reading — but if you mistake one of them slightly, you are unlikely to be led to the right result.) However brief experimentation is unlikely to reveal the power that is, even now, available through this resource. One has to have the patience to learn slowly and to explore it fully.

The major point about both the printed volumes and the database is that this vast resource is now widely available to scholars, who can make their own identifications of their material and can embark on a wide range of different researches. The point was emphasised by the holding at the University of Reading in 2011 of a conference entitled ‘Seeing Red: New Economic and Social Perspectives on Terra Sigillata’. This brought together the members of the rescue team and numerous other scholars in the field, in a display of the directions which research based on Gaulish sigillata may now take on the basis of this resource. The proceedings, edited by Michael Fulford and Emma Durham, appeared in 2013 under the same title as the conference and are reviewed elsewhere in this issue.

The stated longer-term intention of the NOTS team (9, ix) is to create a dynamic resource, so that the catalogue can continue to be updated, enlarged and refined rather than being a static document completed in the year 2012. I have repeatedly been pressed to ensure a similar future for the catalogue of Italian stamps. But this is not easy. ‘Why not provide a means for other people to enter data into your database according to strict rules?’ The pitfalls are obvious enough when you read H. and D.’s strictures (1, 1–2) on the quality of (published) information that they were not able to check for themselves. The strength of their catalogue is that it has been built upon the observation and judgement of two people alone. Likewise, the strength of OCK is that, over a period of slightly more than a hundred years, it has depended essentially on the judgement of three individuals. That kind of consistency cannot be replicated in this field (requiring acute visual judgement) by any set of rules. The addition of new data, if they are not to weaken the whole structure by meeting lower standards, will need somehow to be monitored to the same level as before. I do not say that it is impossible, but while its desirability is surely not in doubt, it will be a big challenge.

There is one further aspect to digital publication which does not affect the printed work (for all its limitations) and which causes me personally considerable anxiety. When the catalogue of the Italian stamps was published in 2000, I considered it premature to seek an on-line platform, on the basis that it would be needed both by practitioners in the field where — in many countries — there was not at that time ready availability of hard-wired internet connections, and by researchers in countries which had poor and slow internet availability at best. For this reason I chose to create an application on disk which could be installed on computers running Windows. (I had at one stage been anxious about the necessity of creating a separate version which would run on Macs, but a demonstration of a Mac successfully running the program with a Windows emulation relieved me of that additional burden.) The database was at the time (as I have already said) greeted with enthusiasm. However now, ‘only’ some twelve years later, it is verging on extinction because of the constant upgrading of Windows, which is content to abandon any attempt at compatibility with software that is ‘so’ old. Through the sympathetic interest of the Oxford Roman Economy ProjectFootnote ⁹ and specifically of Andrew Wilson, moves are now well advanced to make the OCK database accessible on-line and to replicate the functionality which the disk-based version possessed. This is a laborious process, since all of the code has to be rewritten for another platform. I am hopeful that it will soon, however, be readily usable once again. None the less, there remains the threat — for the NOTS database as much as for OCK — that some entity wholly independent of the archaeological community (i.e. the manager of a server system) may one day decide that this body of material is no longer of sufficient interest — and that it will then cease to exist as if it had never been. This simply cannot happen in the same way to a printed work; but while the enhanced utility of a digital resource is enormous, its vulnerability appears to be so too.