The Digital Enhancement of a Discipline

1.1 Scholarly Editions in Byzantine Sigillography

It is important to highlight that, aside from a few attempts to give some very general guidance (Oikonomides 1986; Tsougarakis 1999; Cheynet 2008, 1-82), there are no manuals for Byzantine sigillography, and edition standards have been developed over the last century in a somewhat haphazard fashion in individual publications, never being systematically and unanimously established. Moreover, there is a lack of consistency in the use of edition criteria among scholars and their schools: the features that differ most are those of epigraphical rendering, normalisation of language variants of Byzantine Greek, consistent use of the Leiden conventions, iconographical analysis, and historical commentary.¹

Paper publication is the rule for Byzantine sigillographers. This materialises mainly in ponderous, limited-edition corpora² or small articles scattered in journals and miscellaneous books, the Studies in Byzantine Sigillography being the only journal that specialises in this field.³ Existing paper publications are not readily available, rather expensive, and do not allow for updating, amending, or improving. There are also significant issues of image quality (even if greater efforts have been made recently).⁴

1.2 Byzantine Sigillography and Digital Humanities

In the last few years, sigillography has received increasing attention from experts in Digital Humanities, Western Mediaeval seals being at the forefront.⁵ But this increased attention is far from being sufficient to narrow the gap with sister auxiliary disciplines, such as numismatics, epigraphy, and papyrology, which, during the last 20 years, have developed their own digital approaches, stimulating the growth of online corpora. Unlike these disciplines, Byzantine sigillography has not yet benefited from the experience gained within the Digital Humanities. Within the scope of this paper, we discuss EpiDoc, an international, collaborative effort using a subset of the Text Encoding Initiative (TEI) that provides a standard, guidelines, and tools for creating and encoding scholarly editions of ancient texts and documents produced on various text-bearing objects other than parchment or paper, such as stone or papyrus.⁶ In the field of Byzantine sigillography the only consistent online presence is the ongoing Online Catalogue of the Dumbarton Oaks Research Library and Collection (Trustees for Harvard University) in Washington, DC, which holds the largest collection of Byzantine seals worldwide, with around 17,000 specimens.⁷

SigiDoc, the new encoding standard for Byzantine seals that has been developed since 2015 between Paris and Cologne by Alessio Sopracasa and Martina Filosa, is the first attempt to bridge this gap: the discipline is now ready to benefit fully from the method-oriented interdisciplinarity of the Digital Humanities, leveraging standards, tools, and practices developed in neighbouring disciplines and adapting these to its specific needs and materials.

SigiDoc is an XML-based and TEI-compliant encoding standard for producing digital editions of Byzantine seals and digitally enhanced versions of printed editions.⁸ It is largely based on the ongoing experience of TEI, EpiDoc, and EFES (EpiDoc Front-End Services) – a highly customisable platform for the online publication of ancient texts in EpiDoc XML, as will be illustrated in the examples below.⁹ The developers plan to release version 1.0 in the spring of 2021: a GitHub repository is already available with an in-progress version of the code¹⁰ and a website (http://sigidoc.huma-num.fr/, empty for now) is ready to host documentation, guidelines, and general information about the project, as well as a test corpus consisting of a digitally enhanced version of a printed article (Sopracasa, Prigent 2017).

SigiDoc 1.0 will finally provide something that has been long awaited and discussed since 2006 amongst the community of Byzantine sigillographers, being something of a chimera for Byzantine studies at large, often announced but never seen.¹¹ However, we are aware that it can be improved and we have therefore tried to anticipate future developments and needs: a second round in SigiDoc’s development is currently already seeking funding.

As already stated elsewhere (Sopracasa, Filosa 2020, 241), SigiDoc is:

• a schema, compatible with the EpiDoc and TEI All schemas;

• a template, i.e. SigiDoc’s edition structure (partially reproduced below);

• a stylesheet for HTML transformation;

• a set of stylesheets for scholarly editions of the legends on seals;

• a highly customised version of EFES;

• a set of encoding guidelines;

• a set of files intended to be shared among all future SigiDoc projects (ID lists, controlled vocabularies, authority lists, ontologies etc.).

From a SigiDoc point of view, we consider Byzantine seals as compounds of three different, intrinsically intertwined, aspects: the object, the text, and the image. As objects, seals in many ways resemble coins, thus enabling sigillographers to leverage the experience gained in the neighbouring field of numismatics. As the study of text-bearing objects, sigillography utilises standards and criteria developed for editions of papyri and inscriptions, such as the Leiden conventions, whilst customising them to best suit the different kinds of material being treated. Finally, seals are also treated as images, both as digital facsimiles and as bearers of iconographic depictions. This paper, however, will only deal with the first two aspects. In addition, as a minor but relevant point, this paper will devote special attention to some aspects of the relationship between print and digital publication, whilst addressing the topic of the degree of digital enhancement of the discipline offered by the current state of SigiDoc.

2.1 Manuscripts with Seals and Seals without Manuscripts: TEI Standards and Byzantine Seals

In the ‘digital genealogical tree’ of SigiDoc, the oldest ancestor is represented by TEI, whose aim is “to develop and maintain guidelines for the digital encoding of literary and linguistic texts”.¹⁴ Even though this definition embraces a very wide range of texts, some are only marginally or approximately represented: this is true, e.g. for inscriptions and seals’ legends. The privileged supports of the texts considered by the TEI guidelines are those of paper or parchment, much less stone or lead: this becomes extremely clear upon reading the guidelines about the description of the text-bearing object, explicitly defined as “Manuscript description”, and, more broadly, the names chosen by the TEI consortium for the elements.¹⁵

This remark leads directly to the next generation of SigiDoc’s genealogy, i.e. EpiDoc. As a matter of fact, EpiDoc has been the starting point for the development of several aspects of SigiDoc, mainly because this standard, now long established, represents a selection and a semantical adaptation of TEI’s standards in a direction very suitable for – even if not entirely coincident with – sigillography. Hence, placed under the large umbrella of TEI, SigiDoc greatly benefited from the experience of (digital) epigraphy.

Whilst the elements chosen by SigiDoc are TEI, they no longer refer to a manuscript:

As it shows, the basic structure of a SigiDoc edition structure is extremely close to that of EpiDoc, and it is organised around three main elements: the <teiHeader>, designed to collect the metadata; the <facsimile> for the images of the edited seals; and the <text> for the edition of the seals’ legends, apparatus, and commentary, including bibliography and footnotes.¹⁶

A TEI element <sealDesc> does exist and, according to the TEI standards, it is designed to describe “the seals or similar items related to the object described, either as a series of paragraphs or as a series of seal elements”;¹⁷ this element may contain only a handful of other ones, such as <p>, <ab>, <condition>, <decoNote>, <summary>, and, of course, <seal>, the latter being suitable for “a description of one seal or similar applied to the object described”.¹⁸

These elements are manifestly not sufficiently detailed to properly encode a Byzantine seal, because – as the previously quoted definitions clearly show – the seal is considered as a part of the very object of the description, i.e. the manuscript, <sealDesc> being nested inside <physDesc>. This approach implies a strict interdependence between the document and the seal, which is neither consistent nor realistic with the material available to Byzantine sigillographers: as we saw before, due to the nearly complete loss of Byzantine private and public archives, finding a Byzantine seal still attached to the document it authenticates is extremely rare (Cheynet 2008, 13-14 with references).

The adoption of the TEI standards by SigiDoc involves, therefore, a semantic ‘betrayal’ of some of them or, in a more positive way, their semantic extension.¹⁹ This is particularly true for the metadata, all nested inside a <sourceDesc> element which, at least by its name, can be adapted to multiple materials, but which is extremely stretched by EpiDoc’s and SigiDoc’s use of it if we consider the role that TEI assigns to it.²⁰

Next there is <msDesc>: in order to meet sigillographic needs and to upgrade the place of a seal, assigning to it a central role, it should hypothetically be replaced by the already mentioned <sealDesc>. If SigiDoc’s use of <msDesc> does not strictly respect its semantics, one should also stress that the TEI consortium intends the msdescription module as “general enough” to be “extended [...] and [...] potentially useful for any kind of text-bearing artefact”,²¹ even if the original aim, i.e. working with Mediaeval manuscripts, is predominant. This clarification has been introduced in the current TEI Guidelines (P5) since the 1.0.0 version dating back to 2007; most recently, for the latest release (4.0.0), the consortium wanted to stress further and clarify that the <msDesc> element could be used for all text-bearing objects.²² A discussion took place about the possibility to expand the semantic area of <msDesc>, changing it to <TBODesc> (for text-bearing object description), but the idea was soon abandoned, because the priority was given to backward compatibility, concluding that any <ms*> element could and should be used for any text-bearing object.²³

2.2 Seals as Coin-Like Objects: Seriality vs (Relative) Unicity

Coins share some key features with seals, such as two faces, the presence of images and inscriptions, materials, and techniques of production. Coins are the closest objects to seals having already benefited from an important digital attention and they also have ‘their own TEI’, i.e. the Nomisma project.²⁴ Its Numismatic Description Schema/Standard (NUDS) is XML-based and influenced by the structure of, among others, TEI.²⁵ Looking at what neighbouring disciplines had found in order to represent their data in digital form, in one of SigiDoc’s previous development steps, Alessio Sopracasa, at that time the sole developer, tried to evaluate the possibilities of a close interaction between SigiDoc and NUDS, designing a template for the metadata of the seals based on NUDS, but largely adapted to the needs of sigillography, with its own schema to validate it, thereby seriously discussing the possibility of adapting NUDS for SigiDoc.²⁶ Traces of this attempt are still visible today in NUDS, where sigillographic elements (or elements used in a ‘sigillographic way’), have been included:

• <channelOrientation> is an exclusively sigillographic feature, as it represents the channel through which the thread attaching the seal to the document passed;²⁷

  27 See http://nomisma.org/nuds#channelOrientation.

• <date> and <dateRange> have been included inside both <obverse> and <reverse>, because seals may be struck by means of an assembled matrix, whose two sides date to a (slightly) different timeframe;

• <ab>, borrowed from TEI and added “for greater flexibility in the encoding of prose”: this is the element used by EpiDoc and, now, SigiDoc to encode inscriptions on stone and legends on seals;²⁸

  28 See http://nomisma.org/nuds#ab.

• more generally, NUDS now “allows for the optional namespacing of EpiDoc TEI elements for legends and descriptions for more complex tagging of inscriptions and prose”.²⁹

  29 See http://nomisma.org/nuds#toc-elements.

Beside some minor additions, it seems clear that the inclusion of the TEI-EpiDoc features in the numismatic description was intended to provide the user with much more latitude, particularly for tagging the legends, as explicitly stated in the description of NUDS’ element <legend>: “Typically, the legend will be a literal transcription [...]. Alternatively, one or more <tei:div> elements may be namespaced in for a greater degree of transcriptive accuracy, following the EpiDoc schema”.³⁰

This kind of addition is not accidental: in the editions of coins, the legend is in plain text without editorial intervention, reproduced as it appears on the coin, and followed (or preceded) by a short description of the iconography – the main focus of the edition being the numismatic identification by means of the typological description (so as to obtain a ‘coin-type’) and references to standard works.³¹ It is precisely here that coins and seals differ the most: the seriality of coin production is far greater than that of seals. Suffice it to say that coins are usually found in hoards, whereas seals are mostly found as single specimens. Thus, in Byzantine sigillography each specimen counts and every part of it needs to be thoroughly investigated, especially legends and iconography. This was one of the reasons why the rapprochement between SigiDoc and NUDS was not as effective as initially thought: the core elements necessary to the former were external to the latter.³²

However, far from being unhelpful, this attempt was interesting for SigiDoc with regard to the metadata. NUDS is a very good working example of a standard for the description of objects closely related to seals, and this had its share of influence on the future development of SigiDoc. However, since NUDS counts TEI among its inspirations, the difference between the two standards lies sometimes solely in the name of elements that have, after all, the same function – names more discipline-oriented and semantically consistent in NUDS than in a SigiDoc-adapted <teiHeader>. The clearest examples of this difference are in the NUDS elements <obverse> and <reverse>, for which SigiDoc uses <msPart> nested inside <msDesc>.

Here is an example for the obverse of the seal in fig. 1:³³

The seriality in numismatics and sigillography involves the use of tools in the production of these objects: the die (for coins) and matrix (for seals) can strike a series of identical objects. However, in numismatics this feature is not as prominent as it is in sigillography. In this lies a methodological difference: the seals struck by the same matrix represent a specific category in sigillography, so-called ‘parallels’ or ‘parallel seals’. In sigillographic editions, known parallels should always be cited: they help in the reading of a damaged seal, the number of extant parallels might represent an element discussed in the commentary, or, further, they allow for the reconstitution of the lost matrix from which they come.

2.3 A Digital Identity for Byzantine Seals

Matrices and parallels bring us back to the medium of publication. What is regrettably lacking in traditional sigillography is the unambiguous identification of the specimens. Whether belonging to a private collection or to a public archive, each seal has an inventory number, which is modelled according to changing criteria internal to each holding institution, thereby hindering cross-references across collections and publications. One of the major achievements of a digital scholarly edition in SigiDoc is that a unique identification number will be assigned to each seal. The importance of this feature becomes even clearer when it comes to parallel seals: aside from attributing a unique ID to each seal, it is of paramount importance to gather all the seals coming from the same matrix and bearing the same imprint. This is done by modelling an ID composed of a part shared by all the parallels and a part pinpointing each individual seal. The ultimate goal is to identify the matrices. One must note that, compared to the estimate of 80,000 extant Byzantine seals, only six matrices survive nowadays, hence a proper identification of the parallels would allow for a virtual reconstruction of the countless missing matrices. It is in this respect that the seriality of the seals’ production appears more clearly and that is the reason why SigiDoc also attributes a unique ID to the matrices, thus enabling the gathering of all the seals produced by a given matrix. Stretching this approach further, it is also possible to think that, in some cases with seals with a significant number of parallels, Byzantine sigillography, too, could develop the concept of ‘seal-type’, similarly to numismatics. This is the path taken by the project Sigilla, where a seal-type is virtually reconstructed from a number of specimens belonging to the same seal issuer.³⁴ Finally, the IDs are essential when it comes to overstrikes: just like coins, seals may be struck twice or more, leaving us with two or more imprints on the same support. This means that the ratio ‘one imprint equals one object’ no longer applies, and the IDs ought to express this new situation clearly enough to be understood.

3.1 The Editions of the Legend: Diplomatic, Interpretive, Digital

The diplomatic and interpretive editions are displayed by SigiDoc in two different tabs in our EFES visualisation and, consistently with all EpiDoc projects which use this platform, the XML markup of the legend is available in a third tab. Conversely, unlike EpiDoc, in SigiDoc the two editions are placed within two separate <div>s, as shown by the code snippet on page 107 and as discussed in the following sections.

3.1.1 The Diplomatic Edition and the Epigraphy of a Seal: Typological vs Visual Rendering of Lettering

The semantic content of the legend of a seal is not the only meaningful aspect. The transcription of the characters ‘as they appear’ on the seal is the first step towards a scholarly edition of a legend. Nevertheless, there is no established practice in the rendering of the lettering among Byzantine sigillographers, thus leading to a variety of outputs. Here follow examples of diplomatic editions of six different seals from publications dating from 1884 to 2019:

From its very beginning, one of the major questions raised by SigiDoc’s development was how to produce a diplomatic edition of the legends on the seals, i.e. how to render the lettering. Sigillography shares this issue with epigraphy, but the scarcity of the available space – and the problems it raises – brings the seals even closer to the coins. The need for a (digital) instrument allowing for a better understanding of the evolution of the lettering and, more generally, of the writing practices, as well as of improving and simplifying the scholarly activity, is already a desideratum for other sister disciplines. Scholars in numismatics now increasingly devote attention to the legends of the coins³⁵ – attention that was beneficial for projects in Western Mediaeval seals, too – by means of the development of dedicated fonts.³⁶

35 See Codine 2013 as well as an online report by the same author titled Epigraphie monétaire mérovingienne: nouveaux outils et nouvelles perspectives: http://www.archeologiesenchantier.ens.fr/spip.php?article161.

36 The project Sigilla is planning to update its legends’ transcriptions thanks to the achievements of the PIM project (about which, see Codine 2013).

A different approach has been developed by Archetype for the study of palaeography.³⁷ Built on its predecessor DigiPal, Archetype is an open-source web-based suite of tools for the study of handwriting, palaeographical features, and iconography. The scholar defines a highly structured taxonomy of descriptions of characters, allographs, their components and features. These descriptors are then used to annotate images and further, to search, compare, and sort through scripts, characters, scribal hands etc., providing also a quantitative approach to, e.g. tracing the evolution of scripts or identification of scribes.

37 See https://archetype.ink/#top.

In epigraphic and papyrological editions a diplomatic transcription records the characters extant on the support, without any editorial intervention, as is the practice in EpiDoc. This diplomatic view, however, does not attempt to represent letterforms, ligatures, and decorative elements graphically.³⁸ Considering the wealth of inscribing traditions, scripts, variations of letterforms and ligatures within the wider EpiDoc community, it would be impossible and impractical to attempt faithful visual representation in the diplomatic view, not least because individual characters are identified in Unicode code points. Thus, the diplomatic transcription is considered complementary to photographs/drawings.

38 For further discussion, see the EpiDoc Guidelines https://epidoc.stoa.org/gl/latest/trans-diplomatic.html and the MARKUP-list.

Conceptually, with SigiDoc the choice has been between a typological and a visual rendering of the characters. With the former solution, what counts is the presence of the main types of letters, variant letters, and ligatures rather than the completeness of the palette, and it is aimed at creating broad categories of single letter types; conversely, the latter option aims at exhaustiveness and is intended to reproduce as faithfully as possible the appearance of all the characters, thus making the palette constantly grow as new editions appear. SigiDoc ultimately chose the visual rendering through a specialised font called Athena Ruby.

In Byzantine studies, a ground-breaking True Type font for polytonic Greek called Athena, suitable for the diplomatic edition of both coins and seals, was developed at Dumbarton Oaks under the supervision of the late Nicolas Oikonomides, historian of Byzantium and eminent sigillographer, and used by him for the first time in 1986 (Kalvesmaki 2015, 122): the goal was “to give sigillographers direct control over nuances in the typography” in order to, e.g. “choose variant letterforms and ligatures”.³⁹ It represents a typological, abstract rendering of the lettering found on Byzantine coins and seals. This font is used in fig. 6 above and it is nowadays the specialised font most frequently used by Byzantine sigillographers, despite it not being the standard, as figs. 4 and 5 clearly show.

39 See https://www.doaks.org/resources/athena-ruby.

Moreover, Byzantine sigillography can count on a font (used in fig. 7) which is an answer to the above-mentioned desideratum. This font is called Athena Ruby and it was conceived as both an evolution of Athena and its unification with other fonts for coins developed at Dumbarton Oaks under the project management of Joel Kalvesmaki. Athena Ruby – OpenType and Unicode-compliant – is specifically designed for the epigraphy of Byzantine coins and seals, and goes in the direction of what we called a ‘visual rendering’ of the legend. This font “has been designed to anticipate the needs of digital projects that use XML, JSON or other structured text formats”, and the wish of its developers is the inclusion of Athena Ruby in projects using EpiDoc standards.⁴⁰ Accordingly, SigiDoc encourages its use and the encoding for the diplomatic edition is as follows:⁴¹

40 Kalvesmaki 2015, 123; on Athena Ruby, see also Codine-Trécourt, Sarah 2012, 276-7.

41 The only technical infelicity is that Oxygen editor reads Athena Ruby only in Author Mode.

Should, then, Athena Ruby be used to replace the images? Its glyphs are intended to be “idealized replicas of letterforms” and are, more precisely, “meant to evoke, but not replicate, types of letters”.⁴² Nevertheless, one should note that Athena Ruby has, e.g. twenty-eight variants of the Greek letter alpha,⁴³ whereas Athena only six, and, further, that the former has an ever growing palette.⁴⁴ The trend is clearly that of replicating the letterforms as faithfully as possible without, however, aiming at photographic exactness. Using an ever-growing font like Athena Ruby will give SigiDoc (and Byzantine sigillographers) an increased and ‘atomic’ character-level searchability of the text’s lettering as well as a more accurate analysis of the evolution of the sigillographic epigraphy, one of the major criteria for dating a seal. Scholars of Byzantium have already delivered some important contributions to our knowledge of the epigraphy of Byzantine coins and seals (Oikonomides 1986, 165-9; Morrisson 1994; Oikonomides 2004), but the constant increase in the number of published seals results in a constant expansion of this knowledge. The font’s effectiveness will be enhanced by its inclusion in SigiDoc: the epigraphy will be contextualised thanks to the other data available, such as dating, places, social milieu of the seal’s issuer etc., all materialising in a search form that takes all these factors into account.

42 See https://www.doaks.org/resources/athena-ruby/users-manual.

43 Database of the Athena Ruby Glyphs (HTML): https://www.doaks.org/resources/athena-ruby/database-of-characters-in-html.

44 The font is now under the management of Colin Whiting at Dumbarton Oaks and periodical additions of new characters will henceforth take place (on a biannual rhythm) concurrently to the upcoming publications of seals in the Dumbarton Oaks collection – as Whiting told us per litteras.

The main difference between Archetype’s approach to the script and the one using Athena Ruby lies in their ultimate goal. Archetype provides a framework and methodology for palaeographical analysis that can be adapted to a wide range of alphabets, scripts, and decorative elements, based on the comparison of structured descriptions provided by specialists in their respective fields. Here the standardisation comes in the approach, methods, and tools, rather than in any specific vocabulary used for the distinct letter components.

The idea underlying the development of Athena Ruby is, however, to develop a shared editorial standard for the diplomatic edition. This is arguably beyond the scope of both palaeography and epigraphy, and intrinsic to Byzantine sigillography and numismatics. It aims at providing sigillographers with a single point of reference for the diplomatic transcription in a critical edition, which is both Unicode-compliant (and thus computationally solvent) and reflective of the disciplinary requirements in this particular publishing tradition. Nevertheless, Athena Ruby is still little used by Byzantine sigillographers, and sometimes, when used, its richness is not fully exploited. In order to be properly used, it is good scholarly practice to fully exploit its palette, otherwise – with an inaccurate character choice – there is a danger of providing the readers with the wrong idea of the epigraphy. Paradoxically, the wide range of choice offered by this font has constituted a deterrent to its use, and for a valid reason: Athena Ruby allows for more subjectivity in the diplomatic edition, thus multiplying the possibilities of mistakes. However, ‘subjectivity’ here stands for ‘editorial choices’ and every edition largely relies on them: a large palette pushes us to be more detailed and this is a desideratum, which, of course, makes the work harder. It is true that sometimes, e.g. with a damaged seal, the degree of speculation is very high. For this reason and to limit the subjectivity, SigiDoc allows character-level queries not only as Athena Ruby variant, but also under the normalised Greek version of the letter they represent.⁴⁵

45 Taking the above-mentioned example of the alphas, the variants wwill be also indexed and made searchable under the generic Greek alpha (A).

In light of these issues, the question asked above ought to have a negative answer: Athena Ruby cannot and will not replace the images.

3.2 The Seal as a Meaning-Bearing Object: Semantic Annotation

Alongside the peculiarities of a strictly sigillographic analysis, which focuses on the seal as an object and on the markup of the legend as explained above, what matters most is the information these data convey. They represent the historical information extracted from a single seal and transcending the seal itself to become a source, possibly related to others of non-sigillographic nature and thereby significantly feeding the historical debate.

The first step in this direction is to allow the editors to extract what they consider to be the most important data coming from the seals they are working on: this usually takes the form of indices. With regard to sigillographic printed editions, indices are mostly available in larger corpora, but not in journal articles, with the praiseworthy exception of those published in the above-mentioned Studies in Byzantine Sigillography. All these indices are published to very different degrees of detail, and the choice of which terms should be indexed is variable. There are corpora which index separately proper names, dignities, and functions of the clergy and non-clerical ones (Zacos 1984; Zacos, Veglery 1972); corpora giving special attention to iconography and metrical legends (Jordanov 2006; Cheynet, Campagnolo-Pothitou 2016) and others which do not register either (Oikonomides et al. 1991-2009); whilst in the Studies in Byzantine Sigillography all terms deemed relevant are grouped under a single, broad category, named “Index of proper names and terms”, followed by an iconographic index. It is clear that such an approach, lacking common rules, hinders the analogical interoperability of these indices. Despite being time-consuming and not always as effective as hoped, this perusal of published material remains one of the first and unavoidable steps in studying a Byzantine seal, as it is necessary to ascertain whether the seal has already been published and whether parallels exist.

SigiDoc solves this shortcoming by means of the semantic encoding of the legends: this markup, aimed at extracting the most valuable information coming from the material, has been designed to work best with the EpiDoc Front End Services publishing platform (EFES). EFES provides SigiDoc with two major assets: web visualisation and data valorisation, which is expressed in customisable, automated indices and faceted federated search. Knowing that EFES allows for the creation of a potentially unlimited number of indices, the developers of SigiDoc encourage its users to prepare a common set of indices, to ensure optimal cross-referencing across different corpora.

The indices are generated on the basis of Authority Lists used in EFES, to which the markup points by using the attribute @ref. Here is an example of the legend of the seal in [fig. 1] with only the markup relevant for the indices:

The Authority Lists will be shared among all SigiDoc-based projects, which will have the responsibility of supplementing them with new entries arising from their own materials.

The approach used for indices is part of a broader objective of simplified data searchability, on the basis of developing a standard: creating a set of common and shared rules allowing for proper interoperability. Such interoperability will be realised primarily through a centralised search interface, which will allow for a virtual unification of all the corpora encoded in SigiDoc, thus going beyond the indices themselves and single search forms of each EFES corpus. The functionality of this interface will be based on the shared application of the principles currently employed in the creation of indices and search forms for individual projects. The unified search interface is not yet available and represents the main goal of the next round of SigiDoc’s development, currently seeking funding.

If appropriate indices and search forms are effective with large amounts of data coming from large quantities of seals, the meaning of each seal needs to be fully explained in its commentary. In printed sigillographic editions, commentaries greatly differ in structure among publications, especially when it comes to a proper historical analysis connecting the seal to people, sources, and events related to the broader context of Byzantine history. SigiDoc gives its users the freedom to choose the approach they judge best for their data. The commentary takes the form of a separate <div>, organised in several <p>s containing also bibliographical references, automatically generated by EFES on the basis of a TEI bibliographical list; bibliography, too, will be shared and available to all SigiDoc-based projects.

As it is also designed for the digitally enhanced edition of material already published in print, after a long and lively debate we decided that SigiDoc’s template should include a <div> specifically designed for footnotes. Footnotes play a major role in scholarly publications in the Humanities and represent more than a place to store bibliographical references: they show the sources of a statement written in the main text, they offer a space for further argumentation and demonstration, and offer a chance to discuss other publications without burdening the main text. In EpiDoc-based projects this feature is not explored, and the rule is to have short references for the bibliographical apparatus or written in brackets inside the commentary. However, if one wants faithfully to transpose a printed publication into a digital form, one should find a place for footnotes:⁵³ the amount of information they convey might be too rich to be incorporated into the main text without jeopardising its legibility, as has been shown clearly by the experience in progress of transposing a journal article into digital form with SigiDoc (Sopracasa, Prigent 2017).

Furthermore, thanks to the possibilities offered by EFES, the corpora encoded in SigiDoc will be multilingual by default. The accessibility of the data coming from seals will thereby be increased, reaching non-specialists, too: the user will find a multilingual version of the frame of the standard webpage, including the field names structuring the data, as well as the indices; however, the data themselves (including the text of the legend) and the plain text will have to be manually translated.