Decrypting the Michtonese on f116r

I previously talked about the marginalia on f116r in this blog post and in this pdf document. In brief, there are four lines of text here, , the first up against the cut of the vellum, the other three in a paragraph below, mainly written in a Latin alphabet with a couple of Voychinese words at the beginning of the third line. Here’s the image:

The Michtonese on f116r (colour adjusted)
The Michtonese on f116r (colour adjusted)

The first two words on f116r have been thought to be “pox leber”, or goats liver in German – some say it’s a recipe, other that it’s a swear.

Note: I have superseded this attempt to decrypt the page with a new analysis and no longer think it is encrypted text. This is left here as a record of my attempt. I now believe this page is a medieval charm spell, read the article here.

My conclusion on f116r is different and accounts for all of the readable text of the first line. I have read the first line to read in the Spanish of the 16th / 17th century:

Por seber vm cn autentico afecto
Translation: I cede this to your esteemed magnificence with authentic affection.
Read this post to understand why.
Why Spanish? Well, remember that Spain was the dominant empire at the time in Europe; the native language of Rudolph II was Spanish, and he often corresponded in it; and the House of Hapsburg, the Spanish Kings to which Rudolph belonged, were producing a succession of Holy Roman Emperors who ruled Europe during this century. During the 16th and 17th centuries Spain would rule over Holland, parts of Germany, the Holy Roman Empire and other vassal states throughout Europe.
Back to the text.
The alphabet here used is certainly not Voychinese, and arguably just a standard Roman script. There are three lines. The first two have crosses between each word (with the exception of the space between the first two words). The last word has a cross squashed in above it half way – it looks like it was added afterwards.
A cross stuffed into this word
A cross stuffed into this word

The third line starts off with two Voychinese words then seem to lapse back into the Roman script.

Is it plaintext?

If it is, nobody has been able to read it. There are hundreds of comments on it in the mailing list archives (see on voynich.ninja). I couldn’t read it, apart from the first line. Transcribing the words and trying to find any words that match in any language I speak just didn’t work. I gave up on the effort – if it is plaintext, it’s not for me to read it.

Is it a code?

Ah-ha! First, an explanation of Spanish codes from the period (courtesy of Professor Juan Carlos Galende Díaz, professor of palaeography and diplomatic writing at Madrid’s Complutense University, whose fascinating tomes on the subject seem to be the only in depth work carried out).

In essence, the Spanish administration (and by extension, the ruling families of Spain) loved codes. Carlos V, the Holy Roman Emperor, insisted upon using a series of official codes in all messages which went abroad.

The Spanish were aware of basic frequency analysis, and hence opted for a simple code / cipher substitution model. A multiple substitution alphabet was setup, with the most common letters having multiple homophones. Common words or phrases would have their unique cipher as well.

The earliest official code was that of the Catholic Kings Isabel and Fernando, who had cipher books written which as well as encoding every letter of the alphabet represented some 2,400 words and syllables as well. This model was quickly copied by senior members of the clergy, the military, diplomats, senior ruling families and the like. A number of these fascinating code books are still preserved in Spain’s National Library.

This model would be used for the next four hundred years by Spain. They were confident it couldn’t be broken (although it usually was quite quickly), and saw no reason to change it. Vast numbers of scribes in government offices around the world were employed just to encode and decode the official messages. Carlos V was so confident in his messages that he didn’t change the code for two decades, despite the fact that everybody around the world had broken it by then. His successor, Felipe IV, did change the codes – but for new ciphers using exactly the same mechanism. He attributed the fact that foreign powers could read his messages to people having sold copies of his uncle’s code books, not because of any flaw in the code.

I’ll refer to this as a Patristocrat cipher, a simple substitution cipher in which word boundaries are not respected. If word boundaries are respected, it’s an aristocrat.

Now, look at any of the secret writings of this era and you’ll see exactly the same template as we’re seeing here: a plain text introduction followed by the encrypted bulk of the message. Often we see a mixture of plain text and encrypted portions of the text.

So, my suspicions are aroused. Is this a simple encryption?

Letting the machines do the work – part 1

Well, if it’s a simple substitution cipher, then the computer should be able to break it in a matter of minutes. Let’s think about this.

First off, we need to transcribe the letters. Easier said than done – look at the very first character. Is it an M, an AN, AM or ANI ?

The M of the "Michtonese".
The M of the “Michtonese”.,

Of course, assuming it is a code, we don’t have to transcribe the words character for character. Instead, we have to identify different glyphs and assign a random yet unique letter of our alphabet to them. We then end up with a transcribed standard alphabet reading of the code. I ended up with 21 different characters, excluding the two Voychinese words but including the crosses.

Hmm. 21 characters. It’s a short phrase -just 19 or 20 words long – so it might be enough. The Spanish alphabet of the modern era varies between 24 and 28 characters, as characters such as “ch”, “rr”, “ll” and ñ fell in and out of fashion. We also have the Latinate confusion between i/j and v/w to worry about.

Running the transcribed text though Cryptocrack as a Patristocrat cipher didn’t produce anything that leapt out at me. I’ll omit the details.

I considered the following points:

  1. What are the crosses?
  2. Are there cipher codes (letters that stand for words or phrases) in there?

1) What are the crosses? Well, they could be signs of the Christian cross, added in the two lines to denote the need for Christian help in the recipe or spell, as was common. (In which case, as someone pointed out in the past, the word with a cross above it could stand for “maria”, the virgin Mary). Or they could be part of the cipher (but only appear on two of the three lines, and would be the most frequent letter in the code). Or they could denote word boundaries (but again, they only appear on two lines). I reckon they’re the Christian cross, added in to either show how pure the writer was in his thoughts or as a protection against the magic contained in the words.

2) Are there cipher codes in there? Potentially, yes. It’s a very short message. I’ll come back to this. But if there are, it makes the breaking of the code even more difficult, as we have a shorter piece of text to work with. Which is why the software can’t crack it, it’s assuming some of the codes are simple letters.

Letting the machines do the work (part 2: damn the machines, I’ll do it myself)

Getting out pencil and paper, I attacked the code anew. Where to start? First off, I’m going to assume that the spaces do mark word boundaries (so it’s now an aristocrat code instead of a patristocrat), otherwise what’s the point.  Well, there’s a lovely word, the second one, transcribed as oladaba8. A long word with three repetitive letters (a). I assume a substitution cipher and look for words that fit this pattern in Spanish. Turns out there are quite a few, especially verbs. I take a to be the encryption for a vowel v and run searches on the pattern ??v?v?v?. Here are the results by vowel, ordered in the frequency of the usage of the word during the Spanish modern era (I have omitted repetitions that include misspelling such as trataban/tratavan).

  • ??a?a?a? (over a hundred of these, I list only the most frequent) trabajar/n/d,llamaban,llamadas,acabadas,trataban,granadas,fragatas,acababan,atalayas,tratadas,chacaras, llamaran,acabaran,llanadas,axatabas, alababen,calavadas,clamaban, bravatas,guanacas, llamaran, apagaban, amasaban…..
  • ??e?e?e? (over a hundred of these, I list only the most frequent) obedecer/n, queredes, preceder, apetecen, creyesen, rretener, preterea, fueredes, preceden, fuesedes, vieredes, arehenes, quereres, creyeren,
  • ??i?i?i? (over a hundred, but many of them varients so in reality a lot fewer) edificio, primicia, orificio, inimicis, omicidio, criminis
  • ??o?o?o? (47 here) amorosos, llorosos, olorosos, promotor, enojosos, teologos, troxolos, prologos, isocolon, onororos, adorolos, aforolos, globosos, apologos….
  • ??u?u?u? (12 here, all Latin sounding in name) paululum, grumulus, paululus, pauculum, scutulum, etc etc.

Now I need a word with a again. I take the penultimate word, which reads alra.

  • a??a agua, alma, avía, alta, anda, alla, asia, arma, alba, arca, aora, asta, abra, alua, asna……
  • e??e este, ende, esse, ente, ecce…
  • i??i only Latin words and placenames
  • o??o otro, ocho, oido, olio, odio, ocio, oigo, opio…
  • u??u very rare misspellings

I eliminate u/i from readings of a. I take the seventh word, por_ta8. (_ : possible a space, I hope not). The glyphs are distinct here, so the pattern is ????a?. Of course, this matches hundreds of Spanish words. I’m running out of words containing a, no other word contains two examples. Let’s change tack.

I take the first word, first paragraph. anchrton. The pattern is this [a/e/o]x?????x, where x has to be the same in both cases.

  • ax?????x alguacil, asturias, asientos, advertid, astucias, asombros.
  • ex?????x espaldas, esclavos, escritos, estudios, estorias,escuelas,espuelas,entrasen,
  • ox?????x nothing found.

I now have word lists for the first two words in the sentence. So… I need to start thinking about possible sentence structure. subject+verb+sentence it looks like. Which words to choose?

The most likely candidates for word 1: alguacil (a magistrate); advertid (from advertir , to warn. This is the imperative, vosotros advertid -they warned- which doesn’t really make sense; or misspelling?); escritos (writings); estudios (studies); estorias (common misspelling of historia, history); entrasen (subjunctive imperfective, ellos entrasen, they entered).

Let’s try escritos. Follow it with… obedecen? escritos obedecen, “writings obey”. We’re missing an article (los), not much we can do about that.

The next word (3rd word 1st line):

3rd word 1st line of paragraph
3rd word 1st line of paragraph

malta8? It’s a difficult word. The tall glyph initially looks like its counterpart in the previous word (l = b) but the l curves over and crosses the stalk half way up. This doesn’t curve, it has a strong 90 degree right angle down almost to the bottom. It’s similar to the difference between a P and a D in handwriting. That penultimate glyph has been clearly marked with an open bottom. Did the scribe start to write the 8 there and stop? Did he write o but changed it to a? A search for the pattern ?e?t?n didn’t show much up, although it could be a name like Newton. But that very definite penultimate glyph is worrying me, it looks like a cipher to me – a glyph that means something. Maybe it’s a syllable, it’s before n? I rerun the search with the pattern ?e?t*n and don’t get anything I can slot in there. I jump ahead.

6th word 1st line. teve. All I have here is the letter t. Pattern t??? brings up dozens of words, naturally enough, but I’m missing vowels a/u, one which must be the glyph e.  ta?a: tasa,taza,tapa,tala. tu?u: tutu.

The 4th word, 1st line: a two letter word. Initially seems to be te. According to my reading, this would make it ta, which doesn’t mean anything. If we change the vocal, we could get te or tu. What’s the next word?

The 5th word, 1st line: looks like tav, with the a being corrected.

It’s at this point the line peters out. I tried starting afresh with new selections of words, but nothing jumps out at me.

I leant back in my chair and chewed my pencil

So, it doesn’t seem to be a direct substitution cipher. If the encoder is using a prearranged cipher with multiple homophones per plain text letter (so a could be d,h,o,p for example) then we don’t have enough text to crack it. My mind starts to wander.

Could it be an anagram? Friedman looked into this – but as he pointed out, anagrams of this length are essentially uncrackable unless you know what they’re about. They were popular in the 1500’s, and would explain the mixed up repeated word endings – but if they are, they’re probably lost. I quickly started to form the sentence “tronchan chalados” out of the first two words before giving up.

Is it a magic spell? That would explain the crosses. There are numerous examples of random incantations from western Europe which combine random repetitious text that you were supposed to sing or pray aloud. The insertion of a cross between words at important spots reminds the reader to cross himself at that point. What I’m not aware of is anything like that coming from Spain, or the southern European nations under its influence – a) the Inquisition came down like a ton of bricks on that sort of thing, and b) the tradition comes from the ancient pre-Catholic “pagan” magic rituals remembered in folk memory and incorporated into the Church. That was a Germanic thing, not a Spanish thing.

There are also strong differences between the first line of the paragraph and the second. The first line contains words of variable lengths with lots of apparent vocals and no repetitions with the exception of three ‘8’s. The second line reads much harsher, plenty of harsh consonants (q,x,r) with short words ending all in X or A. Take a look at the repeated word endings:

See the second line?
See the second line?

Jacques Guy suggested that this was an exorcism, scribbled on the last page by a fearful owner of the book.

More mysteries

And what of the ‘aror sheery’ Voynichese words on the third line? They are fluently written, indicating the author was familiar with Voynichese. And they swing nicely into the rest of the Michonese on that line, although the lettering seems more rushed and done with less care than the previous lines. The handwriting is the same – look at the way the a is formed on all lines – but the writing is far more cramped and rushed.

There’s also the “falling stone” mystery at the end of the final line. It looks like an “o” with four dots above it, like a falling stone.

Far left, amongst the images which I have so far ignored, is written in the same handwriting something that looks like Pab. The vellum is creased and folded here, it’s difficult to make it out.

Pab f116rAnd finally, that cut gets on my nerves. It’s a rip from the top of the page down, and there are little holes either side of the rip, as if someone had tried to sew it up. Look at the reverse of the page and you see it seems to have been patched with something, but I don’t know what.

f116v
What’s that rip been patched with? Fur?

So, it’s a long day of doing nothing, essentially.

3 thoughts to “Decrypting the Michtonese on f116r”

  1. Hi David,

    For “something that looks like Pab”, you may be interested in my note of a year’s ago: http://athenaea.net/index.php?id=53

    For the crosses, don’t know if you are right or wrong that they are “Germanic thing” exclusively, but if you are right, this is one more argument towards the overall German context.

  2. I think that is one of the so much significant information for me.

    And i am satisfied studying your article. However wanna observation on some normal issues, The website taste is
    wonderful, the articles is truly great : D. Good job, cheers

Leave a Reply

Your email address will not be published. Required fields are marked *