Roscher’s Lexikon of Mythology as Linked Open Data: Starting a Project on FactGrid

In the age of Linked Open Data, the humanities have increasingly turned their attention from the mere collection of data to its modelling with ontologies and data models. In the field of Greek mythology, these approaches have started in very recent times. Alongside the creation of new databases like MANTO [1], Mythoskop [2], [3] and ToposText [4] (to name but a few), at least three attempts have been made to create an ontology for Greek mythology. In 2018, C. Syamili and R. V. Rekha outlined an ontology for mythical heroes with reference to Wikipedia and the Theoi project [5]. From a classicist’s point of view, R. Scott Smith of the MythLab team has written a series of blogposts with great scholarly detail, which not only address the modelling of mythical characters but also of events, objects and locations [6]. Another team of computer and information scientists have tried their hand at creating an ontology based on and Wikidata in 2021; they built on Syamili & Rekha’s study but significantly expanded its scope [7].

However, none of these projects have worked with the most copious source on Ancient Mythology which has been produced to date: Wilhelm Heinrich Roscher’s Ausführliches Lexikon der griechischen und römischen Mythologie [8]. Published from 1884 to 1937 in more than 100 fascicles which make up 6 volumes, it reports on individual deities, characters, creatures and (to a lesser extent) concepts, events, objects and places from Greek and Roman mythology. The result is an unparalleled collection of records and source material on more than 15,000 subjects, a number larger than that of any of the databases listed above. This vast knowledge base has to this day not been used to its full potential aside from serving as a reference work. While a full digitisation of the 14,000 colums of text interwoven with illustrations and tables is a daunting project (pending approval by De Gruyter, the publishing company which holds the rights to this work), it is certainly worth considering.

To pave the way for such a project, I have created an index file for Roscher’s Lexikon (to be published on Zenodo) featuring headwords, location (volume, fascicle, column extent), date of publication, author name(s) and subject classification – hence the tangent on ontologies for Greek mythology. This file can be not only be used as a reference in and of itself, but also to create basic items for more than 15,000 mythical entities which can then become part of a Linked Open Data hub for mythology. As a venue to create this dataset I have chosen FactGrid.

What is FactGrid?

FactGrid is a public research environment founded by Olaf Simons in 2018. Hosted by the Gotha Research Centre, it runs on MediaWiki with the Wikibase extension, which was developed by Wikimedia Deutschland in its creation of Wikidata, the RDF-based hub which connects all Wikimedia projects. Both Wikidata and FactGrid are licensed under CC-0, meaning data can be shared between the two platforms without running into legal issues. But aside from the identical licensing and familiar User Interface (which can be enhanced by using the custom FactGrid Viewer [9]), there are some striking differences between the two websites.

First of all, FactGrid not only allows original research (which is forbidden or at least strongly discouraged across Wikipedia and Wikidata) but presents itself especially to this sort of use. Second, it can only be edited by registered users, and signing up for an account is not possible without former approval. Third, users are required to use their real names (whereas in Wikidata and its sister projects, pseudonyms are the norm). Points #2 and #3 means FactGrid can afford to allow original research without risking being swamped by unsolicited data input (to avoid the term ‘spam’). The result is Linked Open Data from various genealogical and other research projects which are not only visible to the public but also widely reusable thanks to liberal licensing.

Integration with Wikidata is achieved in two ways: On FactGrid, sitelinks to the Wikimedia projects (including Wikidata) can be set to items; and like with Wikidata, these sitelinks need to be unique and distinct, meaning one and the same sitelink cannot be present in two items at the same time, and each item can have exactly one link to the German Wikipedia, the French Wikisource and so on. This feature is hard-coded in the Wikibase structure and it prevents data duplication. Wikidata on the other hand links to FactGrid Items and Properties with two custom properties (P8186 and P10787).

Goals for a mythical dataset on FactGrid

What can be achieved by bringing data from Roscher’s Lexikon to FactGrid? Many things come to mind, but I want to highlight just two, one achievable in short-term the other as a long-term goal.

Without too much effort, a Knowledge Base of entities (predominantly characters, but also events, objects, places and concepts), which exist or have existed in Greek and Roman (and neighbouring) mythologies, can be created on FactGrid. These items can then be checked against their source (Roscher’s Lexikon) which will be linked in every item.

In the long term, this dataset can be enriched by creating additional entries not covered by Roscher’s Lexikon (as well as marking pseudo-entities which have been disproven by later scholarship), and more importantly, by adding custom statements to the individual items about (including, but not limited to): name variants, genealogy, primary sources, depictions, reception, participation in events. In theory, anyone could contribute to this process, and it would be an ideal playground for students who want to learn about LOD or mythology (or both).

While such datasets in mythology already have been or are being created (MANTO is the most ambitious project in this regard), Roscher’s data may provide a useful corrective. And by matching the dataset against external databases like MANTO, Mythoskop,, ToposText and Wikidata, all these projects can partake in the information provided by Roscher. And while Wikidata has great potential in creating a rich well of LOD on mythology, it is prone to interventions due to its being non-restrictive in who can edit it. This is where FactGrid comes in handy: It is customisable and freely licensed, yet contained enough to provide stability for such a project.

Why use Roscher’s Lexikon for that?

Even though research in classical scholarship tends to have a longer half-life than STEM, using an encyclopedia that was for the most part published more than a century ago will raise some eyebrows. But Roscher’s Lexikon not only still useful as repository of primary sources and outdated literature, it is a unique and valuable resource and merits the attention of our age. The argument for this can be made from three aspects:

First, scope. Although Roscher’s Lexikon mainly deals with Greek and Roman mythology (as its title suggests), it also reports on the neighbouring cultures (Assyrians, Babylonians, Celts, Egyptians, Etruscans …). While it is by no means the only encyclopedia on mythology doing so, it was the first that could make use of the advances made by Cuneiform Studies, Egyptology, Epigraphy and Religious Studies which had seen an unparalleled surge in the 19th century.

Second, scale. While complete coverage is unlikely to ever be achieved, Roscher’s Lexikon came closer to it than any previous or later encyclopedia. Its more than 15,000 headwords not only take into account nearly all mythical characters featured in Greek and Latin literature, but also named characters and deities from inscriptions (votives) or artifacts (like mosaics, vases, statues). When compared to Pauly-Wissowa, the second installment of Pauly’s Realencyclopädie der classischen Altertumswissenschaft started seven years after Roscher’s Lexikon, Roscher almost always has more homonyms (characters with the same name) than Pauly-Wissowa. And as a mention of a character is enough to warrant recognition, Roscher has far more articles overall than the Lexicon Iconographicum Mythologiae Classicae (LIMC), its successor project published between 1981–1999 (with a supplement in 2009), even though the latter is more up to date, especially on artifacts.

Third, plurality of scholarly background. As already mentioned, Roscher’s Lexikon was not limited to classical Greek and Latin literature for its sources. It also relied on epigraphic and papyrological finds as well as art history and archaeology; in the latter department, the authors and editor often explicitly expressed dissatisfaction with their ability to do justice to their ambition: Scholarship in these areas was widely dispersed and had in many cases never been assembled. Considering their time and circumstances, the authors did a good job, even publishing some depictions for the first time in print; but in regards to depictions of myths, Roscher is most outdated and the LIMC should always be preferred as a reference work.

As any scholar who has researched literary sources for ancient myths will admit, Roscher’s Lexikon is still worth consulting. If this can be facilitated by shaping it into a LOD dataset, such an effort should be welcome to the scientific community.






[5] C. Syamili, R. V. Rekha, “Developing an ontology for Greek mythology”. The Electronic Library 36.1 (2018), 119–132,

[6] R. Scott Smith, “The Ontology of Mythical Entites: Part 1”. MANTO June 11, 2020:; id., “The Ontology of Mythical Entites: Part 2”. MANTO June 24, 2020:

[7] Juan-Antonio Pastor-Sánchez, Efstratios Kontopoulos, Tomás Saorína, Thomas Bebisc and Sándor Darányi, “Greek Mythology as Knowledge Graph: From Chaos to Zeus and Beyond”. Semantic Web Journal November 3, 2021:

[8] On Roscher’s Lexikon see most recently Udo G. Reinhardt, Hundert Jahre Forschungen zum antiken Mythos (1918/20–2018/20). Mythological Studies (MythoS) 5, Berlin/Boston: de Gruyter 2022, 19 sq. More information can be found on Wikipedia:


Iviron 812, a textual treasure-trove

Shortly after obtaining my PhD in late 2017, I stumbled upon a new topic which has since stuck with me: Paeanius’ Greek translation (Metaphrasis) of Eutropius’ Roman History, a rare surviving case of a near-contemporary translation from Latin to Greek.

The Metaphrasis has come down to us in five extant manuscripts whose relations I established three years ago in a paper published in the Greek, Roman and Byzantine Studies (online here:DOI). As I demonstrated, the parent of all other manuscripts is the most complete one, Athous 4932 = Iviron 812 (Diktyon 24407), a paper manuscript preserved in the library of the Monastery of Iviron on Mount Athos.

This manuscript is a veritable treasure-trove for Roman history for several reasons:

  1. First, it has the most complete version of Paeanius’ text.
  2. Second, on four leafs it preserves parts of John of Antioch’s Chronological History, a work otherwise attested only in fragments.
  3. Third, it is the only witness of a short text called On the genealogy of Caesar.
  4. Fourth, it represents an independent strand of transmission for John XiphilinusExcerpts from Cassius Dio (edition in preparation by Kai Juntunen, Helsinki).
  5. Last but not least, it has been attributed by Inmaculada Pérez Martín to the workshop of Maximus Planudes in the Chora library (“The role of Maximos Planudes and Nikephoros Gregoras in the transmission of Cassius Dio’s Roman History and of John Xiphilinos’ Epitome“. Medioevo Greco 15 (2015) 175–193).

Pérez Martín found notes resembling the handwriting of Planudes and Gregoras in the margins of the manuscripts, and Planudes’ composition of a series of historical extracts corroborates this hypothesis. Consequentially, Pérez Martín opted for an earlier date and suggested the 12th century by comparing a hand in Vat. gr. 746 (ff.14–219v). However, another scribe’s handwriting bears much more resemblance with the Iviron scribe: Marc. gr. IV 58 (Diktyon 70442), dated to the late 12th century. I owe this information to Ciro Giacomelli.

For comparison’s sake and future reference, here is a photograph of the first page of the manuscript taken in 2018 by Father Theologos Iviritis, librarian at the Holy Monastery of Iviron on Mount Athos.

ms. Iviron 812, fol. 1r. Published with permission of the owner, the Holy Monastery of Iviron on Mount Athos. Copyright is held by the owner.

As can be seen, the manuscript is not in the best condition. Its text gets more legible as it goes on, but water stains, corrosion of the ink and the ink bleeding through the pages have done their part. The best description of the manuscript is still that of Panagiotis Sotiroudis (Untersuchungen zum Geschichtswerk des Johannes von Antiocheia, Thessaloniki 1989, 159–164). For reference, and as the work is hard to obtain, I quote here some parts of his description: “Der heutige Zustand des Codex ist sehr schlecht, besonders am Anfang, wo man ihn kaum durchblättern kann. Schuld daran sind die Würmer und die Nässe. Die Würmer haben an einigen Stellen das beschriebene Papier angefressen, so daß die Schrift fast unlesbar geworden ist. (…) Ein großer Teil dieser Schäden muß während der letzten 90 Jahren entstanden sein, da Lambros’ Mitarbeiter, welche den Text entziffert haben, auch solche Stellen lesen konnten, wo man heute nicht einmal Spuren erkennen kann.”

In private communication, an eminent Austrian scholar had already opted for a 13th century dating of Iviron 812. He vividly described the writing style thus: “deutlich dekadent und nicht auf dieser Höhe, daher vermute ich (und das Papier würde dazu passen) eine spätere Entstehung. Der Schreiber bringt einfach die Ordnung nicht mehr in die Schrift, torkelt herum, fol. 18 sind schöne Beispiele, wie er das Niveau nicht mehr erreicht, aber immerhin eine gut lesbare Schrift zustande bringt.”

A lot more remains to be done. A classification and assigment of the marginal notes, which are by far more than the three hands identified by Sotiroudis. The original condition of the manuscript must also be determined as it was rebound in or after the 14th century and is now missing substantial parts. Also, a new edition and translation of Περὶ τοῦ Καισαρείου γένους would be most welcome.

I have made a transcription of the Paeanius text in the manuscript in 2018/19, and I hope I will some day have the opportunity to edit the whole text with a translation and notes. Perhaps this blog can help achieve this goal?

Paianios, griechischer Übersetzer

Im Zusammenhang mit meinem Aufsatz “On the Transmission of Paeanius” habe ich einen Wikipedia-Artikel über Paianios geschrieben. Erst einmal auf Deutsch, denn in anderen Sprachen gab es das Stichwort schon. Wobei, den Englischen könnte man auch mal überarbeiten.

Warum ich ausgerechnet über diese unbekannten Autor schreibe, dessen Name selbst von Fachleuten als “Panaitios” oder “Pangasius” verballhornt wird? Weil ich das Werk des Paianios kurz vor meinem Referendariat als lohnenden Forschungsgegenstand entdeckt habe.

Paianios’ Metaphrasis ist die einzige zeitgenössische griechische Übersetzung eines nichtchristlichen lateinischen Werks. Und die letzte Ausgabe, aus dem Jahr 1912, lässt noch viel zu wünschen übrig. Es lohnte sich also, in die mittlerweile digitalisierten (oder digitalisierbaren) Handschriften des Werks zu sehen. Und obwohl es unscheinbare, kaum verzierte und nicht besonders kunstvoll ausgeführte Papierhandschriften sind, war es doch eine Freude für mich, die Mühe jener byzantinischen Schreiber nachzuvollziehen, die Wort für Wort, Zeichen für Zeichen, Strich für Strich das profane Geschichtswerk der Nachwelt überlieferten. Eine alte Handschrift, und sei sie auch noch so schlicht, ist eine Kostbarkeit.

Cardinal Bessarion’s Composite Codex: Marc. Gr. 523

 This came in the mail today.

Prof. Carlo Santini, who edited Eutropius in the Biblioteca Teubneriana in 1979, most generously sent me photographs of a part of Bessarion’s composite codex of philosophical and historiographical Greek texts (donated to the Biblioteca Nazionale Marciana in Venice after his death 1472, Diktyon no. 69994). The Paeanius text, as already noted by Hans Droysen in 1879, is a direct copy of Laur. Plut. 70,5 (= L, Diktyon no. 16570).

As I already assumed in my most recent paper (“On the Transmission of Paeanius“, GRBS 63,2020,487-309, DOI 10.5281/zenodo.3960022) the Paeanius text is copied from the Laurentian ms. but independent of all younger copies. My assumptions were based on a few pages that Ciro Giacomelli kindly shared with me in February, 2020. What’s more, he promised me to photograph the rest of the text on his next visit to Venice. But that was before the plague hit.

Now, through the generosity of Prof. Santini, I can corroborate my earlier assumptions: The manuscript has the same defects as its parent, but does not copy all of its marginal notes. Because the other copies retain more of these notes, they must be independent of the Venice manuscript and are rather copied directly from L.

Two things are particularly interesting in this textual witness: Its text is in at least three places superior to its parent (avoiding needless gemination in proper names), in two of them it corresponds with its grandparent, ms. Iviron 812. And for the large gap in the middle of the text (book 6, chapters 9–11), Bessarion’s ms. leaves nearly two pages empty – apparently in hopes of eventually finding and restoring the missing text.

If only he had ventured to Mount Athos.

Monacensis Graecus 101: A collection of historiographical and rhetorical texts

Making research materials available online has become even more important in times of lockdown and reduced service from and access to public research institutions.

The Bavarian State Library, a long-time forerunner in making their valuable collections available online, recently uploaded another batch of digital images from their treasury of Greek manuscripts. One of them preserves some rare historiographical and rhetorical Greek texts and can now, through a digitisal copy of its black-and-white microfilm reproduction, studied from the comfort of your home.

Cod. Gr. 101, written around 1555 in Northern Italy, was described in detail by M. Molin Pradel, Katalog der griechischen Handschriften der Bayerischen Staatsbibliothek, vol. 2 (Wiesbaden 2013) 279–286. Information about its age, scribes, possessors and textual contens can also be found in the Pinakes database (Diktyon no. 44545).

My interest in this manuscript is due to the fact that it is one of only five extant manuscripts transmitting Paeanius’ Metaphrasis (Translation) of Eutropius’ Roman History. The Munich witness of this text was studied extensively by Ernst Schulze (1870), and later by Spyridon Lambros (1912) who both (indepent of one another) proved it to be a copy from an older manuscript, Laur. Plut. 70.5 (Firenze, Biblioteca Media Laurenziana, Diktyon no. 16570also available online). The Munich ms. thus has value not as an original textual witness but as a possible source of scholarly conjectures to the text.

So far, checking the first 51 folios of the ms. has proved fruitful as it has yielded half a dozen corrections to the text – granted, all are easy corrections and have no ground-shaking consequences, but they are testiment to the critical acumen of its scribe or his supervisors. And they predate the first printed edition by more than 30 years.

Of course, a black-and-white microfilm can only take us so far. I have not been able, for example, to ascertain or even locate two notes in the ms. reported by Lambros (1912) 114–115 saying that the manuscript had been lent to Gotha librarian August Beck in 1868 (who requested it on Schulze’s behalf) and to Theodor Mommsen in 1872 (who used it to prepare an edition of Paeanius). To this end one must consult the object itself, not a copy from a copy.

Hello World

Hello and welcome!

I’ve started this blog to share information on my research. A classicist by training, I find pleasure in discovering old visual artifacts (be it text, image or something else) which transport meaningful messages through space and time. I love extracting those messages (or, if that’s not possible, imagining a possiblem message) and sharing them with others.

This blog will predominantly be about my studies in Greek and Latin texts, but also on any other dusty topic that is just waiting to be brought to life.