What Is Actually Known About the Script Nobody agrees how many signs there are, and the disagreement spans a factor of four. That single unsettled number decides what kind of writing system this could be.
Set aside every proposed translation and a surprising amount is left. The direction of reading is established. Sentence boundaries are identifiable. Numbers have been located. None of that required knowing the language, and all of it survived being checked by someone else. This page is what is left when the readings are removed.
I. How many signs, and why it decides everything
The size of a sign inventory is the most informative thing about a writing system before anything is read. An alphabet runs to a few dozen signs, because it encodes sounds and there are not many. A syllabary runs to the low hundreds. A system with many hundreds or thousands of signs is doing something else, most likely giving symbols to whole words.
For this manuscript the count is unsettled by an enormous margin. Kálmán Némäti, working between 1889 and 1892, catalogued about 792 distinct symbols, and observed that this is roughly ten times the size of any known alphabet. Other treatments work with figures in the low hundreds. The published range spans a factor of four or more.
The disagreement is not carelessness, and it will not be resolved by counting more carefully. When two marks differ slightly there is no way to tell from the marks alone whether they are two signs, or one sign written twice by a hand that varies. Split generously and you get a large inventory and a word-based system. Merge generously and you get a small one and something closer to a syllabary. The choice is made before the analysis and it determines the analysis.
Némäti drew the right inference from his own count in 1892: whatever this is, it is probably not an alphabet. More than a century later, Tokai and Király's conclusion that small groups of symbols stand for whole words is the same shape of answer, reached by a different route.
This is not a Rohonc peculiarity. The dispute over whether the Indus script is logosyllabic turns mainly on its inventory size, and that number is unsettled for the same reason: nobody can say from the marks alone which variants are the same sign. Any claim about what a script is, made before its inventory is settled, is standing on the count.
II. What has held
Four things about the writing have been established and independently confirmed. All four were found without knowing what the text says.
| Property | Established by | Confirmed by |
|---|---|---|
| Runs right to left, top to bottom, pages right to left | Gyürk, 1970 | Locsmándi, 1990s. Consistent with the lines being justified against the right margin |
| A symbol acts as a sentence delimiter | Locsmándi, 1990s | Király's sentence segmentation, 2011 |
| Number-like patterns, later a system using powers of ten | Gyürk, 1970 | Tokai and Király, 2010 onward |
| No case endings, which argues against Hungarian | Locsmándi, 1990s | Consistent with a code that does not represent word structure |
The last of those deserves a note. Hungarian is heavily inflected, so if the underlying language were Hungarian written in an unusual script you would expect to see recurring endings. Their absence is evidence, and it is evidence against the reading most people would reach for first given where the book was found.
III. The 87 drawings
The illustrations are the one part of the codex that anyone can read, and they were the way Tokai got in. They show religious scenes, secular ones and military ones. The religious imagery is not from a single tradition: crosses, crescents and sun motifs appear together, which is either a striking piece of religious syncretism or a sign that whoever made the book was not working from one settled iconography.
They are also, by common consent, crude. That observation cuts both ways and both are worth stating. A professional scriptorium did not make this. Equally, a forger producing an object to sell would have had every reason to make the pictures impressive, and these are not.
IV. The book has been disturbed
Tokai and Király record that bifolios are missing, that leaves have been swapped and reordered, and that a later foliation misleads rather than helps. Anyone who has followed the Voynich Manuscript will find that familiar.
It matters more than it sounds. Any statistical claim about a text assumes you know what order it goes in. If the leaves are out of sequence, measurements that span a page boundary are measuring an accident of rebinding. It is one more reason that a transcription of this manuscript needs to come with its codicology attached, and one more reason that the eight higher-resolution pages photographed in 2015 are eight rather than 448.