A History of the Rohonc Codex From a donated library in 1838 to a peer-reviewed reading in 2018. Who looked at it, what they concluded, and how often the answer was that it could not be done.

I. The Record

• 1743
  A Batthyány catalogue

  Lists Hungarian prayers in one volume, duodecimo. Possibly this book, possibly not. It is the only candidate for a record of the codex before it surfaces, and it is a description that would fit a great many books.

• 1838
  Count Gusztáv Batthyány

  Donates the family library at Rohonc to the Hungarian Academy of Sciences. The codex is in it. This is where its documented existence begins, and everything earlier is inference.

• c. 1840
  Ferenc Toldy

  The founding figure of Hungarian literary history examines it and publishes little. It is the first of many inspections by a serious scholar that end without a conclusion.

• 1844
  János Jerney

  Publishes the first substantial notice, in Tudománytár, describing the writing as belonging to no known system. Hungarian scholarship at this moment badly wants the codex to be an early Hungarian script, which shapes the next fifty years of attention.

• 1866
  Károly Szabó

  Proposes that the codex is a fabrication by Sámuel Literáti Nemes, a Transylvanian antiquarian and a known forger, active in the 1830s, who had died in 1842. No evidence has ever connected Nemes to this manuscript specifically. The attribution nevertheless became, and largely remains, the mainstream Hungarian position. The case, in full.

• 1884–85
  Josef and Konstantin Josef Jireček

  Father and son, both professors in Prague, study 32 pages between them. No result.

• 1885
  Bernhard Jülg

  The Innsbruck orientalist receives the codex for study. No result.

• 1890–92
  Mihály Munkácsy

  The painter, then one of the most famous Hungarians alive and working in Paris, takes the manuscript there so it can be examined. No result. By this point the codex has defeated a literary historian, an orientalist, two Prague professors, a paleographer and a national celebrity.

• 1889–92
  Kálmán Némäti

  The first systematic study, and the first genuinely useful one. He catalogues roughly 792 distinct symbols, notes that this is about ten times the size of any known alphabet, and draws the correct inference: whatever this is, it is unlikely to be alphabetic, and more likely a syllabary or a system in which signs stand for whole words. Published as Rohonczi Codex Tantétel in 1892. It takes another 126 years for anyone to confirm it.

• 1970
  Ottó Gyürk

  The turning point, and it is a change of question rather than a flash of insight. Instead of trying to read the text he measures it, working on sequences of symbols that repeat. He establishes the reading order: right to left, top to bottom, with the pages themselves running right to left. He identifies patterns that behave like numbers. Much of his statistical work was never published, which is its own small tragedy.

• mid–1990s
  Miklós Locsmándi

  Repeats the exercise with a computer and confirms Gyürk independently. He finds that one symbol behaves as a sentence delimiter, that the diacritic dots follow no system he can detect, and, importantly, that there are no case endings, which argues against the language being Hungarian. He concludes the regularity is far too strong for the text to be gibberish, while being unable to rule out a hoax.

• 1996
  Attila Nyíri

  Proposes a Sumero-Hungarian reading from two pages. Refuted.

• 2002
  Viorica Enăchiuc

  Publishes a complete translation as a Vlach war chronicle in a Dacian dialect of Latin. Refuted.

• 2004
  Mahesh Kumar Singh

  Publishes a reading of the first 24 pages as Hindi in a variant of Brahmi. Refuted in the same journal issue.

• 2010
  Benedek Láng

  Publishes Why don't we decipher an outdated cipher system? The Codex of Rohonc in Cryptologia, and later a full monograph, The Rohonc Code. The first treatment to work through the possibilities methodically, using historical analogies to the secret writing of the period. His conclusion is that the forgery verdict is implausible and that the codex must be a deliberately encoded text. His argument.

• 2010–11
  Gábor Tokai

  An art historian, working from the pictures. In three short articles in Élet és Tudomány he dates the imagery and finds structure in the text around it: coded INRI inscriptions on the drawn crosses, codes that behave like the names of the evangelists followed by numbers, and four-digit numbers that behave like years counted from the creation of the world.

• 2011
  Levente Zoltán Király

  Independently, and in parallel, working from the text rather than the pictures. He finds a way to segment sentences, a seven-page section with numbered headings that functions as a table of contents, the same evangelist codes Tokai had found, and a chapter system carrying biblical references. He also explains why the first quarter of the book has no chapter structure: it is one long Passion narrative.

• 2015
  University of Hamburg

  Rephotographs eight pages at higher resolution.

• 2018
  Tokai and Király

  Publish jointly in Cryptologia. The writing is a code rather than a cipher, the content is an ordinary Catholic prayer book of its period, and a date in the text points to 1593. Announced as the first of three or four papers. The reading, and the objections.

II. The Shape of It

Read as a whole, one thing stands out. For a hundred and thirty years the codex was given to eminent people who looked at it and gave it back. Toldy, Jerney, Hunfalvy, the Jirečeks, Jülg, a paleographer from Vienna, and Munkácsy carrying it across Europe. Between them they represent a good deal of the philological talent available in Central Europe in the nineteenth century, and not one of them produced anything that survives.

What broke the pattern was not a cleverer reader. It was Gyürk in 1970 asking a different and much smaller question: not what does this say, but which way does it run, and what repeats. That question can be answered without knowing the language, and unlike a translation, its answers accumulate. Locsmándi could confirm them with a computer. Király could build sentence segmentation on top. Tokai could match numbers to a calendar. Each step is checkable by the next person, which is exactly what a translation produced by resemblance is not.

The three readings that failed all did the opposite. Each began by deciding what language the text was in and worked backwards to make the signs fit. That is the difference between the two halves of this history, and it is worth more than any individual result in it.