The Rohonc Codex

Using AI to extend and test the Király and Tokai dictionary

Sources and data

This page states where every file came from. It credits the work of others. It also states what may be done with each source. The data files are listed near the end, under The data.

The manuscript is out of copyright. Király and Tokai's dictionary and transcription are theirs. The readings, tests, English and editorial matter are this project's own.

What is cited

The corpus

The codex is a compilation, and most of what its compiler wrote fits inside a small shelf of books. Every folio of the translation was read against the passage it retells, because an elliptical text does not become English on its own: the manuscript leads, and the source only ever says what one of its phrases means. This is that shelf. Every book on it is public domain, and each links to where it can be read.

How each was fetched and under what terms is in the provenance file; the reading itself is in the book, and every departure from these sources is named in its notes.

King James Bible

Project Gutenberg ebook 10, public domain.

Read it: Gutenberg 10

Douay-Rheims Bible (Challoner revision)

Project Gutenberg ebook 8300, public domain. Added 2026-09-20 because the codex is a Catholic compilation working from the Vulgate, and pooling candidate words off a Protestant English translation manufactures near-misses: a third of the King James candidate pool turns out to be wording no Vulgate reader would have had. It also supplies the deuterocanon — Tobias, Judith, Wisdom, Ecclesiasticus, Baruch, Machabees — which the King James file here does not contain at all. Challoner's explanatory notes are set as separate paragraphs between the verses and are stripped, not pooled.

Read it: Gutenberg 8300

Wake, The suppressed Gospels and Epistles of the original New Testament

Gutenberg ebooks 6507–6515, public domain. Supplies the Protevangelium, the infancy gospels and the Gospel of Nicodemus.

Read it: Gutenberg 6507 · Gutenberg 6508 · Gutenberg 6509 · Gutenberg 6510 · Gutenberg 6511 · Gutenberg 6512 · Gutenberg 6513 · Gutenberg 6514 · Gutenberg 6515

York, Towneley, Chester and N-Town (Ludus Coventriae) play cycles

archive.org scans of 19th- and early-20th-century editions, public domain.

Read it: archive.org

Caxton's Golden Legend

archive.org, public domain.

Read it: archive.org

Barlaam and Ioasaph

attributed to St John Damascene, Woodward and Mattingly's translation — Project Gutenberg ebook 749, public domain. Added 2026-09-20. It carries the apologue of the man fleeing a unicorn who falls into a pit and clings to a tree gnawed by two mice, which folio 148v tells, and its moral names the falling man as "the race of Adam", which is why the codex writes him with Király & Tokai's Adam sign.

Read it: Gutenberg 749

The Roman Missal, English, 1865 and The Complete Office of Holy Week, 1875

archive.org scans, public domain. Added 2026-09-20 for the liturgy: a 1593 Catholic reader is at least as much missal and breviary as it is scripture, and the Reproaches of Good Friday that folio 048v tells live there and in no Bible. Both are Google scans of two-column books, so the Latin and English columns interleave and the OCR invents words; any token absent from the 1.75M words of period English in ALL.txt is dropped before use.

Read it: archive.org

R. H. Charles, The Apocrypha and Pseudepigrapha of the Old Testament in English, vol. 2

(Oxford, 1913), from the Internet Archive scan apocryphapseudep02charuoft. Published 1913 and out of copyright. The Vita Adae et Evae in it is the source of the codex's opening -- Michael, the making of Adam, Seth at the gates of Paradise for the oil of mercy -- and nothing in the corpus had covered it. Only the Books of Adam and Eve are indexed: four fifths of that volume is critical apparatus whose dense rare vocabulary outscored every real parallel.

Read it: archive.org

Flavius Josephus, The Wars of the Jews

tr. William Whiston, Project Gutenberg ebook 2850. Public domain. The codex tells the siege of Jerusalem, Vespasian and Titus, and the mother who eats her child.

Read it: Gutenberg 2850

References for the endnotes

Some endnotes in Book One name texts outside the corpus. None was downloaded; each was read once at the address given, and nothing from any of them is reproduced. On 23 September 2026 every source behind all 55 endnotes was checked again, and the notes that had to be corrected say so on the page. What each supplied, and its copyright position, is in the provenance file.

What is not here

Király and Tokai's dictionary, transcription, page records and the library scans are not on this site. The signs are traced from the manuscript's own ink. The shapes are the sixteenth-century scribe's.

Their site is rechnitzer-kodex.hu, and it is the place to read them.

Where the scans are

The page images this project reads are the Hungarian Academy of Sciences Library’s low-resolution monochrome scan of K 114: 227 spreads at about 100 ppi. The Academy’s repository offers it for academic and educational use and marks it “do not pass on”, so it is not copied here. The places below hold the scan, the microfilm it descends from, or a few public-domain page images.

The data

The data is in the repository at github.com/styopr613/rohonc-codex. This section names the books the codex is read against and says where everything else goes.

The books every folio is read against are listed above, under The corpus.

The outside review

The two reviewers' specifications are quoted whole: Gemini 2.5 Pro and Grok 4.7. Their replies as re-glossers and blind readers are files, and go up with the rest.

Everything else

The dictionary of added readings with its evidence, the folio English, the rendering, every saved run, the outside readers' replies, the working documents and the programs are posted to GitHub, with their history, so that nothing here links to a copy that can go stale. The standing orders are already published whole at The standing orders.

The working documents

The project's working documents are described here. They are working files written for whoever runs the project next. They are kept accurate rather than tidy. Anyone who wants one can ask for it.

Ask for any of them at [email protected].

The reading

The English summary of the gloss, folio by folio. Scroll for the folios on either side.

Open the whole reading →

Licence · How to cite · Sources and data · Repository · Contact