Papyri.info, launched in 2010,The launch took place at the XXVI International Congress of Papyrology (Geneva), with a talk by Joshua Sosin (Sosin 2010); this information is reported in the "About" page of Papyri.info, https://web.archive.org/web/20170603015536/http://www.papyri.info/docs/about. is an extensive digital text collection of Greek and Latin documentary papyri, that is, ancient documents (accounts, petitions, tax receipts, letters, etc.) dating from the IV century BC to the VIII century AD, mostly written on papyrus, and mainly discovered in archaeological excavations in Egypt.

The collection is a general purpose one, based on the principles of completeness and of representativeness, within its domain of documentary papyrology. In fact, it has become a point of reference for documentary papyrology, as it aggregates three major databases of documentary papyri: the Duke Databank of Documentary Papyri (DDbDP), a very large collection of texts of documentary papyri realised by the Duke University; the Heidelberger Gesamtverzeichnis der griechischen Papyruskunden Ägyptens (HGV), created at the University of Heidelberg, which provides the metadata of all documentary papyri; and the University of Michigan Advanced Papyrological Information System (APIS), focused on images, as well as metadata, of papyri from the North American institutions, plus a few small federated collections worldwideFor the physical collections that have been digitised in the APIS database, see the overview of this project outlined in Bagnall-Gagos 2007, in part. 66-67. Roger Bagnall and Traianos Gagos, founders of APIS, soon realised the potential for digitisation to bring to light almost forgotten collections; this led them to digitise not only the papyri of the Michigan and other American collections, but also some unpublished papyri of European institutions (Bagnall-Gagos 2007, 66-67).; also important is the integration of an essential tool for papyrological research, namely its main bibliographical database, the Bibliographie Papyrologique (BP). The aims of completeness and representativeness are not explicitly stated on the Papyri.info website. We find a mention of the collection content on the home page, but this presents no information on its extension: "The Papyrological Navigator (PN) supports searching, browsing, and aggregation of ancient papyrological documents and related materials". Notwithstanding, we can infer that completeness is a principle of the collection from the words of one of the founders of the DDbDP, John F. Oates (1993, 62): "We are aiming to construct a machine-readable data bank comprising all published Greek and Latin documentary papyri". Also, the contributions published by William H. Willis, co-founder of the DDbDP with Oates, on the plan of the entering of the texts and on the updates on this progress demonstrate the intention that the collection may be as complete as possible: firstly they digitised the papyri published in the more recent volumes of papyrological collections, secondly the remaining volumes "in reverse chronological order until all are done", thirdly "all texts published separately in journals", and lastly the most significant corrections gathered in the dedicated papyrological tool, the Berichtigungsliste (Willis 1984, 172; updates on the progress of the data entry are provided in Willis 1988, 1992, 1994). The collection currently amounts to 55,080 texts, which have been merged from the DDbDP, each of them provided with metadata exported from the HGV and also, as to papyri from American collections, from the APIS. In addition to these records with the papyrus texts, there are 31,936 records with metadata only, exported from the HGV and APIS; these records pertain to published papyri whose texts have not been digitised yet, or to unpublished papyri whose metadata and images have been put at disposal by the APIS researches to enable the scholarly community to study them.Gagos 1996, 19-20; Gagos 1997, 155. Finally, as just mentioned, there are images of some papyri, exported from the APIS database, and therefore pertaining to fragments mostly housed in American institutions.These figures are drawn from the search form on the PN home page: at the bottom, the drop-down menu of the "Has Transcription" selection criterion shows the values "true (55080)" and "false (31936)". As one can infer, this indicates the records complete with texts and those which so far exclusively contain metadata.

The collection is meant to be used in research, especially by papyrologists or by ancient historians who need to consult papyrological primary and secondary sources. Besides the fields of papyrology and ancient history, the collection contributes to historical linguistics, helping scholars to examine the features of the language of the papyri – texts that constitute an important body of evidence for Greek and Latin in their postclassical phases, and for their contacts with other languages of the Mediterranean.Cf. Evans-Obbink 2010, 1-3, for an overview on the significance of the papyri for the study of the ancient languages. The authors also explain how the research on the language of the papyri is one of the areas that mostly benefits from technological advances. Through the citations of the DDbDP in this volume (found by searching for "DDbDP" in the index) one can find instances of linguistic researches conducted with the aid of this database, now merged into Papyri.info. Papyri.info intends to meet scholarly standards by submitting the texts and the changes entered by users to an editorial vetting, and by reporting the emendations and the supplements (both already published and newly proposed by users) of the text transcriptions.

The then Centre for Computing in the Humanities, now Department of Digital Humanities, at King’s College London, was in charge of the first round of conversion from Beta Code into the Unicode character encoding, and from SGML into the EpiDoc XML standard (Bodard, Sosin, and Viglianti 2011). The Heidelberg Institut für Papyrologie mapped its HGV database with the other databases of papyri aggregated into Papyri.info, namely the DDbDP and APISSosin 2010; Papyri.info, "The DDbDP"; Integrating Digital Papyrology 3, 10-11, 18-21., and, also currently, works for continuing the integration of the HGV metadata of new papyri into the PNJ. Cowey and C. Lanz, personal communications.; also, James Cowey in Heidelberg is co-director of the DDbDP together with Joshua Sosin, as reported on the information page of the DDbDP (fig. 4). All this information reported on the DDbDP page is also very relevant to Papyri.info as a whole, as the texts, which it contributed, constitute the core of the PN. Furthermore, this page contains details on the editorial board who approves users' contributions, and it thus concerns the current state of the collection as well. As King's College London and the Institut für Papyrologie, too, gave a contribution to Papyri.info, and the latter still continues to collaborate, their role should be put in more evidence, and be as visible as the mention of the DC3 and the ISAW. Hence, King’s College London and the Institut für Papyrologie, too, should occur on the home page and in the history of the development of Papyri.info on the "About" page (fig. 3), and at least the HGV should be reported on the home page under the heading “Produced by” (at the bottom, on the right: see fig. 2). Other involved institutions mentioned on the "DDbDP" page, which should occur on the "About" page of Papyri.info, are the Columbia University Libraries, the University of North Carolina, the University of Kentucky and the New York University. Alternatively, the two pages about Papyri.info and the DDbDP could be merged into a single page.

Funding was provided by the Mellon Foundation and the National Endowment for the Humanities (NEH), as reported on the "About" page. Furthermore, from the "DDbDP" page we learn that financial support was granted to the Duke Databank by the Packard Foundation and the Packard Humanities Institute (PHI). As already noted, since Papyri.info has its roots in the DDbDP and therefore their histories are closely connected, it would be fair to mention the Packard funding in the "About papyri.info" page, as well: without the initial support from the Packard institutions, who also gave technical assistance, many of the texts that now populate the database could not have been entered.On the decisive contribution of the Packard Foundation and the PHI to the creation of the DDbDP, see the history of this database outlined in Oates 1993, 62-64, 69-71.

As for the long-term archiving of the data and institutional curation, information is provided, on the "About" page (in the "Stewardship" section) about the Duke Collaboratory for Classics Computing, based at the Duke University Library, reported as the unit who has the task of the "maintenance and enhancement of the papyri.info toolset and community". More details on the way the sustainability of Papyri.info is pursued are available in the related bibliography: open source development, through the use of a Git repository to store and update the data (Papyri.info IDP (Integrating Digital Papyrology) Data, https://web.archive.org/web/20180202152933/https://github.com/papyri/idp.data; standards-based formats, i.e., the Unicode character encoding, EpiDoc-conformant TEI markup, and the Resource Description Framework (RDF) model for the integration of the databases in the context of the Semantic Web.On the conversion of the texts in Beta Code and SGML into Unicode and EpiDoc XML, see Baumann et al. 2011, esp. "Integrating Digital Papyrology" and "Lessons from the conversion of the Duke Databank of Documentary Papyri from legacy formats into EpiDoc TEI XML"; Baumann (2013, 93-100) explains the choice of Git and its functioning; Baumann et al. 2011, esp. "The Papyrological Navigator: Project Integration with RDF", and Baumann 2013, 100 deal with the use of an RDF triplestore for integrating the source databases.

In conclusion, information about the creators and the participating institutions that provided expertise and funding is presented; on the other hand, it needs to be made more discoverable by integrating it into the information about the whole resource, rather than being referred to just in connection with one of its compounds. More details on the long-term data preservation, which are found in the bibliography on the project, could be added to the "About" page for any user interested in following up in the reasons of the choices made by the developers; as an alternative, users could be provided with the bibliographical references where this information can be found.

On the result page, one may expect that entering a new word and hitting the "Search" button will perform a new search; the result is, instead, a refined search. Users should, in fact, click on the x-button in the black box on top of the results to remove the previous search value; however, this button is not evident, as was also pointed out by one of the interviewed users. As an alternative, one can perform a new search by clicking on the "New Search" button; users, though, need to remember that they should first click on "New Search", and only then enter the new value (and finally perform their search by hitting "Search"): if one types the new word first, and then hits "New Search", as it may happen, this will remove the new value. This aspect could be improved by allowing the "New Search" button to perform a search with the new value, rather than to go back to the PN home page. Moreover, the "Search" button could be renamed "Refine Search" to explain its function more clearly; even though this phrase is already present, on the top left, it would be more evident if it was positioned on the button, where users actually need to click.

Other aspects of the PN user interface can be better illustrated by way of a comparison with the interface of the HGV, the Heidelberg database with records of metadata. As we have seen, the PN offers more accurate results than the HGV because it does not contain multiple records for the same papyrus, as the HGV does whenever a dating is uncertain. On the other hand, a reason why one may still wish to utilise the HGV database is that its interface offers more possibilities of search than that of the PN. The HGV allows searching for a date according to a single year, and even a month or a day; moreover, one can search for any range of time (by entering any year into the "ChronMinimum" and "ChronMaximum" fields). Instead, on the PN one can perform a search by date exclusively by selecting a predefined value in the "Date on or after" and/or "Date before" fields, within a fifty-year range. Surprisingly, if one tries searching for a single year among the PN metadata (by entering it in the search field and selecting the "Metadata" option), no result is retrieved, although the help information (in the right-hand pane of the search) states that this search is possible, as one in fact expects.Cf. the following passage in the "More about string-search" section, about the metadata search: "Metadata: Searches the metadata associated with the document - for example, the identification number, location, and dating of the document".

It is also true that users look for types of documents not only by means of a metadata search, but also by searching the texts for significant words that are typical of certain documents; they are therefore able to find texts of a certain category with this other method. Two users I interviewed combine this kind of search with the previous one about the metadata: for example, they look up formulas or keywords, like "ἐξαυτῆς" ("at once"), a word typically occurring in summons, in order to retrieve all the texts of this category. Indeed, the HGV warns users that "there is no guarantee of completeness in the case of searching according to content as we did not develop a system of categorisation"HGV, "Search Tips and other Information", https://web.archive.org/web/20171004170724/http://aquila.zaw.uni-heidelberg.de/help/search/en.; it would be useful to point out this limitation in the PN search form, as well, in the help information in the right-hand part of the search page (among the information about "Metadata", in the "More about string-search"), so as to suggest users to use both search methods.

In conclusion, there are a few circumscribed problems that slightly limit the usability of the search interface, especially for first users, sometimes causing them to spend more time than necessary when performing a new search, looking for typologies of documents, and seeking for texts within a limited time-span.

The process of integration of the three main databases, accomplished in the Integrating Digital Papyrology project, has allowed the integration of different kinds of data, in different formats: the texts, coming from the Duke Databank and encoded in Beta Code and SGML, were converted into Unicode and EpiDoc TEI; they were integrated with the metadata and the images, on records encoded in FileMaker and exported to XML, from the Heidelberg HGV and the Michigan APIS databases. The merging of texts and metadata was facilitated by the addition, to the texts, of a unique identifier, i.e., the Trismegistos number. A great effort was put into this process, in the Integrating Digital Papyrology project which consisted in three phases and lasted six years, from 2007 to 2012 (Sosin 2010, Baumann et al. 2011). Papyrologists can thus now access their main resources from only one extensive platform. The conversion to EpiDoc, an open markup standard which is a subset of TEI, ensures that Papyri.info has a high degree of compatibility with other projects in the Humanities and a long-term sustainability.

To summarise the strengths and weaknesses of Papyri.info, we might highlight that the comprehensiveness of the material, the integration of the different databases (both of texts and metadata, and of research tools) and the ability of an integrated search across them explain why this is an indispensable resource for the study of documentary papyri. However, precisely to valorise the collection, that is, to enable users to take advantage of all its features, there should be a better communication about the characteristics of its content and tools, and about the procedure to contribute new data. Other useful additions would be pointing to resources where similar content can be found, as happens for the literary and medical papyri on the Digital Corpus of Literary Papyri, and more accuracy in the documentation relative to the involved institutions and staff.