<![CDATA[The Voynich Ninja - Voynich Talk]]> https://www.voynich.ninja/ Thu, 17 Sep 2026 01:20:47 +0000 MyBB <![CDATA[The Voynich Manuscript’s positional entropy collapse]]> https://www.voynich.ninja/thread-6068.html Tue, 08 Sep 2026 22:08:36 +0000 Torsten]]> https://www.voynich.ninja/thread-6068.html Register or Login to view. at Cryptologia: "The Voynich Manuscript’s positional entropy collapse: comparative evidence from languages and ciphers"

You are not allowed to view links. Register or Login to view.

Quote:The Voynich Manuscript exhibits unusually low conditional character entropy and rigid positional constraints that have resisted explanation within linguistic frameworks for more than a century. In this study, I apply a leakage-free, folio-segmented, character-level analysis to the Voynich Manuscript and to a set of natural-language, constructed-language, and cipher control corpora. The analysis measures bigram and trigram bits per character, the marginal gain from higher-order context (ΔBPC), positional cross-entropy within words, and local lexical overlap under minimal edit distance. All experiments employ strict recto–verso train–test separation, transcription-layer isolation, and identical preprocessing across corpora. The results show that Voynichese exhibits a pronounced collapse of medial positional entropy and a consistently small ΔBPC. These properties persist across manuscript sections and transcription layers and are not reproduced by natural or constructed languages under identical conditions. Applying the same analysis to a verbose substitution cipher derived from a Latin source text demonstrates that this cipher reproduces the Voynich statistical profile with high fidelity. Shuffle controls confirm that this profile is destroyed when internal adjacency is randomized, indicating structured generation rather than sampling artifacts.
]]> Register or Login to view. at Cryptologia: "The Voynich Manuscript’s positional entropy collapse: comparative evidence from languages and ciphers"

You are not allowed to view links. Register or Login to view.

Quote:The Voynich Manuscript exhibits unusually low conditional character entropy and rigid positional constraints that have resisted explanation within linguistic frameworks for more than a century. In this study, I apply a leakage-free, folio-segmented, character-level analysis to the Voynich Manuscript and to a set of natural-language, constructed-language, and cipher control corpora. The analysis measures bigram and trigram bits per character, the marginal gain from higher-order context (ΔBPC), positional cross-entropy within words, and local lexical overlap under minimal edit distance. All experiments employ strict recto–verso train–test separation, transcription-layer isolation, and identical preprocessing across corpora. The results show that Voynichese exhibits a pronounced collapse of medial positional entropy and a consistently small ΔBPC. These properties persist across manuscript sections and transcription layers and are not reproduced by natural or constructed languages under identical conditions. Applying the same analysis to a verbose substitution cipher derived from a Latin source text demonstrates that this cipher reproduces the Voynich statistical profile with high fidelity. Shuffle controls confirm that this profile is destroyed when internal adjacency is randomized, indicating structured generation rather than sampling artifacts.
]]> <![CDATA[The Voynich Manuscript: Understanding the Untranslatable with Lisa Fagin Davis]]> https://www.voynich.ninja/thread-6065.html Tue, 08 Sep 2026 20:39:41 +0000 Torsten]]> https://www.voynich.ninja/thread-6065.html Register or Login to view. about the Voynich manuscript with Lisa Fagin Davis: You are not allowed to view links. Register or Login to view.

Quote:The Voynich Manuscript has been a curiosity and mystery for hundreds of years – what can its strange history teach us about worldbuilding? This week, we’re joined by Lisa Fagin Davis, one of the world’s experts on this enigmatic text. Come along as we discuss the history of the text, its creation and contents, and how the sciences and humanities can come together using tools like latent semantic analysis to discover the nature of untranslatable texts.
]]> Register or Login to view. about the Voynich manuscript with Lisa Fagin Davis: You are not allowed to view links. Register or Login to view.

Quote:The Voynich Manuscript has been a curiosity and mystery for hundreds of years – what can its strange history teach us about worldbuilding? This week, we’re joined by Lisa Fagin Davis, one of the world’s experts on this enigmatic text. Come along as we discuss the history of the text, its creation and contents, and how the sciences and humanities can come together using tools like latent semantic analysis to discover the nature of untranslatable texts.
]]> <![CDATA[An experiment with phonetic writing]]> https://www.voynich.ninja/thread-6052.html Wed, 02 Sep 2026 20:07:08 +0000 ololololo]]> https://www.voynich.ninja/thread-6052.html This experiment was inspired by Chinese theory, and in particular by Stolfi’s theses about the scribe’s mistake and writing by ear.

I'm very skeptical about linguistic theories, particularly about attempts to map VMS symbols onto other writing systems, and even more so, onto sounds.

However, I became interested in the thesis that the author might have recorded the language by ear, as he either did not know or underestimated the existence of any writing system among the people (if there was one at all).

I tried to create such a “phonetic” script for the Ingush language. I used not only letters (but not all of them) as symbols, but also very common syllables in words in the text. At some points, I left out problems because I didn’t want to create too many syllables. So, one of the syllables is written the same as another, but it differs from it by one sound (specifically, in that it lacks that sound).

I asked the AI (sorry, I don’t know how to speak Ingush Cry ) to write a short text, and I recorded it using this letter. Here’s what I got:

.jpg   photo_2026-09-02_22-46-02.jpg (Size: 116.42 KB / Downloads: 258)
AI text:
Хьянна саак хьятмад талх лоаат-таш да
Цэннабоа баш сабаара хьякха кхораш ба
Бэрда гтя гъуйра хьявса хьялха лъогlаш яц
Шура тlъянна цаlохк тхоа тэхка мара баккха дац
Къант човнаш тэхьа латт, цунгъала ра морх хьяшар да
Сак лъусташ ваа, амма цунзи ларглац

Will someone be able to match the symbols with letters/syllables from the text and reconstruct the writing? How difficult will this be?

I hope you won’t have any difficulties with Cyrillic characters Smile . The text itself looks different, but it reads as it’s written here.
If this remains unsolved, I will send you my alphabet and a transcription of the text in accordance with that alphabet Big Grin
I would be interested to see your theories, “transcriptions,” attempts, and anything else.]]> This experiment was inspired by Chinese theory, and in particular by Stolfi’s theses about the scribe’s mistake and writing by ear.

I'm very skeptical about linguistic theories, particularly about attempts to map VMS symbols onto other writing systems, and even more so, onto sounds.

However, I became interested in the thesis that the author might have recorded the language by ear, as he either did not know or underestimated the existence of any writing system among the people (if there was one at all).

I tried to create such a “phonetic” script for the Ingush language. I used not only letters (but not all of them) as symbols, but also very common syllables in words in the text. At some points, I left out problems because I didn’t want to create too many syllables. So, one of the syllables is written the same as another, but it differs from it by one sound (specifically, in that it lacks that sound).

I asked the AI (sorry, I don’t know how to speak Ingush Cry ) to write a short text, and I recorded it using this letter. Here’s what I got:

.jpg   photo_2026-09-02_22-46-02.jpg (Size: 116.42 KB / Downloads: 258)
AI text:
Хьянна саак хьятмад талх лоаат-таш да
Цэннабоа баш сабаара хьякха кхораш ба
Бэрда гтя гъуйра хьявса хьялха лъогlаш яц
Шура тlъянна цаlохк тхоа тэхка мара баккха дац
Къант човнаш тэхьа латт, цунгъала ра морх хьяшар да
Сак лъусташ ваа, амма цунзи ларглац

Will someone be able to match the symbols with letters/syllables from the text and reconstruct the writing? How difficult will this be?

I hope you won’t have any difficulties with Cyrillic characters Smile . The text itself looks different, but it reads as it’s written here.
If this remains unsolved, I will send you my alphabet and a transcription of the text in accordance with that alphabet Big Grin
I would be interested to see your theories, “transcriptions,” attempts, and anything else.]]> <![CDATA[[Article] What We Actually Know ( about the VMS )]]> https://www.voynich.ninja/thread-6048.html Sat, 29 Aug 2026 10:42:28 +0000 bi3mw]]> https://www.voynich.ninja/thread-6048.html
You are not allowed to view links. Register or Login to view.]]>
You are not allowed to view links. Register or Login to view.]]> <![CDATA[What if it's a "Lossy Encoder" for 15th-Century Latin?]]> https://www.voynich.ninja/thread-6041.html Mon, 24 Aug 2026 13:18:19 +0000 Pointless..]]> https://www.voynich.ninja/thread-6041.html
By “lossy,” I mean that the source text may originally have been written in Latin, but that only part of its linguistic information was preserved in writing. Such a system would assume that the intended reader already possessed enough practical knowledge, contextual information, or interpretive tools to reconstruct the meaning. Where the result might otherwise be ambiguous, the system could provide additional cues to guide the reader toward the correct interpretation.

Shannon’s information theory suggests that natural languages contain a considerable amount of redundancy, sometimes estimated at around 50 percent. Latin, as a highly inflected language, carries much of its grammatical information in word endings. Nouns, adjectives, and verbs are marked according to their roles in the sentence, meaning that some structural information is communicated more than once.

In a practical text, however, much of this information may not be strictly necessary. If a practitioner already knows how to prepare a decoction, grammatical case endings and many function words may add relatively little to the basic operational meaning.

To see how this might work in practice, consider a short apothecary-style recipe in Latin:

"Ad dolorem capitis: Recipe radices hellebori, et tere in mortario. Misce cum aceto et oleo rosarum. Ungue frontem patientis ante somnum."

In ordinary English, the passage means roughly:

"For headache: Take hellebore roots and grind them in a mortar. Mix with vinegar and rose oil. Anoint the patient’s forehead before sleep."

If we reduce the Latin words to their basic dictionary forms, we get something like this:

“ad dolor caput recipere radix helleborus et terere in mortarium miscere cum acetum et oleum rosa unguere frons patiens ante somnus”

We can then remove prepositions such as ad, in, cum, and ante, conjunctions such as et, and predictable instructional verbs such as recipere and miscere. What remains is the raw referential payload:

"dolor caput radix helleborus terere mortarium acetum oleum rosa unguere frons patiens somnus"

The result is an awkward, almost “caveman-style” form of Latin. To a modern linguist looking for conventional morphology and syntax, it appears broken. To a trained fifteenth-century apothecary, however, much of the intended meaning would still be recoverable. The practical context constrains the possible relationships among hellebore roots, a mortar, vinegar, rose oil, headache, and the patient’s forehead.

This is important because practical knowledge does part of the interpretive work that grammar would normally perform. The reader does not have to consider every theoretically possible relationship among the words. Existing knowledge of the craft or some other external knowledge narrows the range of plausible interpretations. The system expects that the receiver has sufficient information to handle the context recovery, transmitted beforehand.

The difference between the information carried by ordinary Latin and the apparently smaller capacity of the encoded text might therefore be explained, at least in part, by removing inflectional endings, function words, predictable instructions, and other contextually recoverable material.

Once Latin is reduced in this way, however, the result no longer looks much like Latin. Ambiguities also inevitably arise. For example, simply listing an ingredient and an action does not always tell the reader what is being acted upon, in what order, or under which conditions.

To preserve meaning without ordinary inflections, the encoding system would therefore need an alternative way of organizing the remaining roots or semantic units. One possibility is a rigid positional framework in which an element’s location determines its function. A position might indicate whether a root refers to an ingredient, an action, a quantity, a body part, a preparation method, or some other category.

In effect, position would take over part of the work normally performed by Latin morphology and syntax. The resulting text would not simply be abbreviated Latin. It would be Latin material reorganized within a different information system, one that relies on shared professional knowledge, predictable context, and positional structure to recover what has been omitted.

Any views or comments on this?

Thank you.

-Pointless]]>
By “lossy,” I mean that the source text may originally have been written in Latin, but that only part of its linguistic information was preserved in writing. Such a system would assume that the intended reader already possessed enough practical knowledge, contextual information, or interpretive tools to reconstruct the meaning. Where the result might otherwise be ambiguous, the system could provide additional cues to guide the reader toward the correct interpretation.

Shannon’s information theory suggests that natural languages contain a considerable amount of redundancy, sometimes estimated at around 50 percent. Latin, as a highly inflected language, carries much of its grammatical information in word endings. Nouns, adjectives, and verbs are marked according to their roles in the sentence, meaning that some structural information is communicated more than once.

In a practical text, however, much of this information may not be strictly necessary. If a practitioner already knows how to prepare a decoction, grammatical case endings and many function words may add relatively little to the basic operational meaning.

To see how this might work in practice, consider a short apothecary-style recipe in Latin:

"Ad dolorem capitis: Recipe radices hellebori, et tere in mortario. Misce cum aceto et oleo rosarum. Ungue frontem patientis ante somnum."

In ordinary English, the passage means roughly:

"For headache: Take hellebore roots and grind them in a mortar. Mix with vinegar and rose oil. Anoint the patient’s forehead before sleep."

If we reduce the Latin words to their basic dictionary forms, we get something like this:

“ad dolor caput recipere radix helleborus et terere in mortarium miscere cum acetum et oleum rosa unguere frons patiens ante somnus”

We can then remove prepositions such as ad, in, cum, and ante, conjunctions such as et, and predictable instructional verbs such as recipere and miscere. What remains is the raw referential payload:

"dolor caput radix helleborus terere mortarium acetum oleum rosa unguere frons patiens somnus"

The result is an awkward, almost “caveman-style” form of Latin. To a modern linguist looking for conventional morphology and syntax, it appears broken. To a trained fifteenth-century apothecary, however, much of the intended meaning would still be recoverable. The practical context constrains the possible relationships among hellebore roots, a mortar, vinegar, rose oil, headache, and the patient’s forehead.

This is important because practical knowledge does part of the interpretive work that grammar would normally perform. The reader does not have to consider every theoretically possible relationship among the words. Existing knowledge of the craft or some other external knowledge narrows the range of plausible interpretations. The system expects that the receiver has sufficient information to handle the context recovery, transmitted beforehand.

The difference between the information carried by ordinary Latin and the apparently smaller capacity of the encoded text might therefore be explained, at least in part, by removing inflectional endings, function words, predictable instructions, and other contextually recoverable material.

Once Latin is reduced in this way, however, the result no longer looks much like Latin. Ambiguities also inevitably arise. For example, simply listing an ingredient and an action does not always tell the reader what is being acted upon, in what order, or under which conditions.

To preserve meaning without ordinary inflections, the encoding system would therefore need an alternative way of organizing the remaining roots or semantic units. One possibility is a rigid positional framework in which an element’s location determines its function. A position might indicate whether a root refers to an ingredient, an action, a quantity, a body part, a preparation method, or some other category.

In effect, position would take over part of the work normally performed by Latin morphology and syntax. The resulting text would not simply be abbreviated Latin. It would be Latin material reorganized within a different information system, one that relies on shared professional knowledge, predictable context, and positional structure to recover what has been omitted.

Any views or comments on this?

Thank you.

-Pointless]]> <![CDATA[AI Companies destroying rare books to train their AIs]]> https://www.voynich.ninja/thread-5980.html Wed, 29 Jul 2026 20:17:40 +0000 Mark Knowles]]> https://www.voynich.ninja/thread-5980.html <![CDATA[How to create a pseudo language in 13 easy steps.]]> https://www.voynich.ninja/thread-5963.html Sun, 26 Jul 2026 18:01:34 +0000 Dunsel]]> https://www.voynich.ninja/thread-5963.html Step 2:  Split those words into phonetic prefix/midfix/suffix.  Only keep the prefix and suffix.
Step 3:  Pick 20 random prefix and suffix from that normalized list.

PrefixSuffix
mensa
magnus
grana
corcor
miror
deles
monsdo
mucem
moda
narmons
merus
gensdus
maro
bouam
caue
aa
ee
ii
oo
uu

Step 4:  Create a list of vowels, a,e,i,o,u,y then, create a substitution table by shifting them one to the right.
Step 5:  Create a list of random consonants then, create a substitution table by shifting them one to the right.
Step 6:  Substitute those shifted vowels and consonants into your prefix and suffix table.  Make sure to pick suffixes and prefixes that do not contain k, t, p or f.

Shifted PrefixShifted Suffix
buqgo
bozqag
zyoqo
wiywiy
beyiy
xudug
biqgxi
bawub
bixo
qoybiqg
buyag
zuqgxag
boyi
riaob
woau
oo
uu
ee
ii
aa

Step 7:  Combine your new prefixes and suffixes randomly in order to create words.  Note, you can stack up to 3 prefixes onto a suffix.
Step 8:  Insert a k, t, p or f into the first letter position of the first word and randomly into other words.
Step 9.  See what your first test output looks like.

pbuqi aupa bkeo bowub i rii buua bazyoaob oiqo zyowogo uexag xu bozxag woixo babarixi woyiy ebue buqqag bia bkiwiy babiqg abiko baboyiy beyi zyodug ebawpiy eebaxo eobawub a uzyoyi ogo buxag buau ibozxag bozzyokego qoybee buuxi zyobebiqg woo bobouaob bau wokbuqbixo biqtag wkiyuwub yag rti zyobipqgqo xubodug zpyogo bebfebui.

kbebawiyau ewiy bebopbiqg bubobozau wiy buzyoku buqyag bozyiy euyiy exki uwtiy bozwub bobiqg be ewfiy beyag wobawkub btaeyag zuqgtwou u exuxi abibuqxakg ebiqgyi a baaob buqoyxo rikxo zuqgewub zuqgqag xugo bozwtiy axag biupbozau zuqgwiyxi exuwiy ou riqoyku ie bayiy ebozriwiy bobaxo o zuqgyi rioi ebiqgu biqgogfo boxi biqgykag zyoea ribiqgxo.

kbibiqgu bawub bteqo wofe bewoxag bexag etzyoxi ia wobiqg bozbpuqyag riyapg riexi bubewiy bozyi boukqag biwoyiy yiy zyowobexo bogfo abae bozbozau qoywub eqto aob bkozbuqxag zuqgwiy bizuqgoxi o ribuqtwiy bebeaob bozbabuxo riqo atu zyoau zuqgueyi xuau biupxo zyoo ubuqaxag buwowiy uribozwiy bobiqgu boubiqg xubozixag babeyiy qoybeeyiy buqxag biqgu zuqgeqfo eebuxi.

fbiqoywub xuxuxi biqgqo bebuqbiqg wopxag riyiy zyoxi boadug wto baaotb ria iabiqg wity xfubaqo u xag babiqag uboxuqag baiau qoybiqg i bekbuqyi e biabiqg wiybozwoa baxi biqgabae ebiqgqag iboqo qoyieqag wiyxo biqgibue owobiqg buqbadug zyoeadug biawou bobawub riqoydug e xurixo bayi riwfoeyag zyoyi biaxi rizyoqag aobiqag zuqgbekyiy xudug riboaob boufxo.

ICK! garbage.

Step 10: Restrict word lengths to 10 chars with an attempt to average around length 6.
Step 11: Create copy/mutate rules.  15% of words are created from the shifted prefix/suffix table or just the prefix or just the suffix. 30% are copies of words you've already put on the page.  50% are mutations of words by adding, deleting or swapping prefixes in words that have multiple prefixes. And the last 5% is just singletons.
Step 12:  Check your new output.

fbozqo beqo i bozqo i bozqpo beaob babozqo beqo beqo beqo oaob beaob xubozqo ibeqo beqo xubozqo beqo qoybeqo woaob obeqo babozqo zyowoaob beaofb rikbeqo wiyu zyowoaob woaob wiyykag buwiy qoybeqo oribeqo a beqo bubeqo bozqo qoywiyyag zuqggko bfeoaob buqwoaob bozyiy woaob bebeoaob bibuwiy wiyte bibuwiy bibeqto ribuwiy btabeqo wobeqo.

twiydug zyowtoaob buwity beqo i beaob beaob btozqo aaob zyowoaob biqgaaob buqbofaob xuzyowoaob bebeqo zuqggo buqbiqkg bozwoaob e obeqo a ubeqo ebuwtiy wiyi bozbibeqo ebutwiy biqpgbeqo bawobeqo exfag uibeqo bkozqo wiyu beqo bepwiye ubikbeqo oaaob ufwiye woaob ufyiy abeaob woxo wiyyiy biqgqo aaaob bozwiydug ibeqo xuaaob buqbetaob bawifyyag biqgaaob zyowobkeqo.

kzyowoaob ebeqo oabeaob beqo buwiy xkuxi ibeqo oubeqo rkibaqo bozqo awoaob aaob biqgbeqo u obeqo beqo buqwtoaob buqwiyfi bozqo riwiy bfoi i bozkxo bezyoe bozaob iaaob beqo bei bozbezyoe biqgbeqo buqwoaob rixuqko ariwiy bibeqo bfexag uiqo buwoaob bowiyu exuqo bowiy uriqo opbozqo bozwiyi wifyxi i oxki iebuwiy woaob ebeaob iebuwiy.

twoaob qoyafob biqgwiy zyooubeqo bozbozxo wofaob oiebuwiy buquwiye bozqo bowiyu euibeqo woaob baobeqo riqpoybeqo exukqo buquwiye xuwkiyyag uqpag zyowoaob woyiy bobebeqpo bobuwiy ibowiyu buqwoaob bowiy boztaaob zuqggo iebuwiy xuaaob beqo iqoyaob qoykuriqo zyoyi iwiyu ibobau oribeqo woyag u btawobeqo zyou bkeaob boyag wiywiy bebei aobeqo qoybiqg biqgaaob zyoki bozpqo bozexuqo.

Much better!

Step 13:  Substitute Latin characters for a cool looking alphabet you made up.

fbozqo beqo i bozqo i bozqpo beaob babozqo beqo beqo beqo oaob beaob xubozqo ibeqo beqo xubozqo beqo qoybeqo woaob obeqo babozqo zyowoaob beaofb rikbeqo wiyu zyowoaob woaob wiyykag buwiy qoybeqo oribeqo a beqo bubeqo bozqo qoywiyyag zuqggko bfeoaob buqwoaob bozyiy woaob bebeoaob bibuwiy wiyte bibuwiy bibeqto ribuwiy btabeqo wobeqo.

twiydug zyowtoaob buwity beqo i beaob beaob btozqo aaob zyowoaob biqgaaob buqbofaob xuzyowoaob bebeqo zuqggo buqbiqkg bozwoaob e obeqo a ubeqo ebuwtiy wiyi bozbibeqo ebutwiy biqpgbeqo bawobeqo exfag uibeqo bkozqo wiyu beqo bepwiye ubikbeqo oaaob ufwiye woaob ufyiy abeaob woxo wiyyiy biqgqo aaaob bozwiydug ibeqo xuaaob buqbetaob bawifyyag biqgaaob zyowobkeqo.

kzyowoaob ebeqo oabeaob beqo buwiy xkuxi ibeqo oubeqo rkibaqo bozqo awoaob aaob biqgbeqo u obeqo beqo buqwtoaob buqwiyfi bozqo riwiy bfoi i bozkxo bezyoe bozaob iaaob beqo bei bozbezyoe biqgbeqo buqwoaob rixuqko ariwiy bibeqo bfexag uiqo buwoaob bowiyu exuqo bowiy uriqo opbozqo bozwiyi wifyxi i oxki iebuwiy woaob ebeaob iebuwiy.

twoaob qoyafob biqgwiy zyooubeqo bozbozxo wofaob oiebuwiy buquwiye bozqo bowiyu euibeqo woaob baobeqo riqpoybeqo exukqo buquwiye xuwkiyyag uqpag zyowoaob woyiy bobebeqpo bobuwiy ibowiyu buqwoaob bowiy boztaaob zuqggo iebuwiy xuaaob beqo iqoyaob qoykuriqo zyoyi iwiyu ibobau oribeqo woyag u btawobeqo zyou bkeaob boyag wiywiy bebei aobeqo qoybiqg biqgaaob zyoki bozpqo bozexuqo.

Extra credit:  Tweak the output as needed to make it look good and add some ambiguous illustrations.

Now, before you say it: 

Do I really think this is how the Voynich was created? 

No.  Of course not. 

BUT... I'm going to ask you to look over that output a bit closer.  You are going to see a lot of 'linguistic' traits and Voynich traits.  I know we're used to looking at Voynich and seeing it's patterns, but give it a shot.  

Beqo.  It gets repeated.  Think daiin.

And you get words like ibeqo and obeqo and oubeqo without even really trying.  Then woaob and beaob and bozaob and oaoab.  

Think dain, ain, choldaiin, oldain.

Just sharing something I'm working on as a theory.   I'm pretty well into it and as soon as I have something formalized I'll be sharing it as well.]]> Step 2:  Split those words into phonetic prefix/midfix/suffix.  Only keep the prefix and suffix.
Step 3:  Pick 20 random prefix and suffix from that normalized list.

PrefixSuffix
mensa
magnus
grana
corcor
miror
deles
monsdo
mucem
moda
narmons
merus
gensdus
maro
bouam
caue
aa
ee
ii
oo
uu

Step 4:  Create a list of vowels, a,e,i,o,u,y then, create a substitution table by shifting them one to the right.
Step 5:  Create a list of random consonants then, create a substitution table by shifting them one to the right.
Step 6:  Substitute those shifted vowels and consonants into your prefix and suffix table.  Make sure to pick suffixes and prefixes that do not contain k, t, p or f.

Shifted PrefixShifted Suffix
buqgo
bozqag
zyoqo
wiywiy
beyiy
xudug
biqgxi
bawub
bixo
qoybiqg
buyag
zuqgxag
boyi
riaob
woau
oo
uu
ee
ii
aa

Step 7:  Combine your new prefixes and suffixes randomly in order to create words.  Note, you can stack up to 3 prefixes onto a suffix.
Step 8:  Insert a k, t, p or f into the first letter position of the first word and randomly into other words.
Step 9.  See what your first test output looks like.

pbuqi aupa bkeo bowub i rii buua bazyoaob oiqo zyowogo uexag xu bozxag woixo babarixi woyiy ebue buqqag bia bkiwiy babiqg abiko baboyiy beyi zyodug ebawpiy eebaxo eobawub a uzyoyi ogo buxag buau ibozxag bozzyokego qoybee buuxi zyobebiqg woo bobouaob bau wokbuqbixo biqtag wkiyuwub yag rti zyobipqgqo xubodug zpyogo bebfebui.

kbebawiyau ewiy bebopbiqg bubobozau wiy buzyoku buqyag bozyiy euyiy exki uwtiy bozwub bobiqg be ewfiy beyag wobawkub btaeyag zuqgtwou u exuxi abibuqxakg ebiqgyi a baaob buqoyxo rikxo zuqgewub zuqgqag xugo bozwtiy axag biupbozau zuqgwiyxi exuwiy ou riqoyku ie bayiy ebozriwiy bobaxo o zuqgyi rioi ebiqgu biqgogfo boxi biqgykag zyoea ribiqgxo.

kbibiqgu bawub bteqo wofe bewoxag bexag etzyoxi ia wobiqg bozbpuqyag riyapg riexi bubewiy bozyi boukqag biwoyiy yiy zyowobexo bogfo abae bozbozau qoywub eqto aob bkozbuqxag zuqgwiy bizuqgoxi o ribuqtwiy bebeaob bozbabuxo riqo atu zyoau zuqgueyi xuau biupxo zyoo ubuqaxag buwowiy uribozwiy bobiqgu boubiqg xubozixag babeyiy qoybeeyiy buqxag biqgu zuqgeqfo eebuxi.

fbiqoywub xuxuxi biqgqo bebuqbiqg wopxag riyiy zyoxi boadug wto baaotb ria iabiqg wity xfubaqo u xag babiqag uboxuqag baiau qoybiqg i bekbuqyi e biabiqg wiybozwoa baxi biqgabae ebiqgqag iboqo qoyieqag wiyxo biqgibue owobiqg buqbadug zyoeadug biawou bobawub riqoydug e xurixo bayi riwfoeyag zyoyi biaxi rizyoqag aobiqag zuqgbekyiy xudug riboaob boufxo.

ICK! garbage.

Step 10: Restrict word lengths to 10 chars with an attempt to average around length 6.
Step 11: Create copy/mutate rules.  15% of words are created from the shifted prefix/suffix table or just the prefix or just the suffix. 30% are copies of words you've already put on the page.  50% are mutations of words by adding, deleting or swapping prefixes in words that have multiple prefixes. And the last 5% is just singletons.
Step 12:  Check your new output.

fbozqo beqo i bozqo i bozqpo beaob babozqo beqo beqo beqo oaob beaob xubozqo ibeqo beqo xubozqo beqo qoybeqo woaob obeqo babozqo zyowoaob beaofb rikbeqo wiyu zyowoaob woaob wiyykag buwiy qoybeqo oribeqo a beqo bubeqo bozqo qoywiyyag zuqggko bfeoaob buqwoaob bozyiy woaob bebeoaob bibuwiy wiyte bibuwiy bibeqto ribuwiy btabeqo wobeqo.

twiydug zyowtoaob buwity beqo i beaob beaob btozqo aaob zyowoaob biqgaaob buqbofaob xuzyowoaob bebeqo zuqggo buqbiqkg bozwoaob e obeqo a ubeqo ebuwtiy wiyi bozbibeqo ebutwiy biqpgbeqo bawobeqo exfag uibeqo bkozqo wiyu beqo bepwiye ubikbeqo oaaob ufwiye woaob ufyiy abeaob woxo wiyyiy biqgqo aaaob bozwiydug ibeqo xuaaob buqbetaob bawifyyag biqgaaob zyowobkeqo.

kzyowoaob ebeqo oabeaob beqo buwiy xkuxi ibeqo oubeqo rkibaqo bozqo awoaob aaob biqgbeqo u obeqo beqo buqwtoaob buqwiyfi bozqo riwiy bfoi i bozkxo bezyoe bozaob iaaob beqo bei bozbezyoe biqgbeqo buqwoaob rixuqko ariwiy bibeqo bfexag uiqo buwoaob bowiyu exuqo bowiy uriqo opbozqo bozwiyi wifyxi i oxki iebuwiy woaob ebeaob iebuwiy.

twoaob qoyafob biqgwiy zyooubeqo bozbozxo wofaob oiebuwiy buquwiye bozqo bowiyu euibeqo woaob baobeqo riqpoybeqo exukqo buquwiye xuwkiyyag uqpag zyowoaob woyiy bobebeqpo bobuwiy ibowiyu buqwoaob bowiy boztaaob zuqggo iebuwiy xuaaob beqo iqoyaob qoykuriqo zyoyi iwiyu ibobau oribeqo woyag u btawobeqo zyou bkeaob boyag wiywiy bebei aobeqo qoybiqg biqgaaob zyoki bozpqo bozexuqo.

Much better!

Step 13:  Substitute Latin characters for a cool looking alphabet you made up.

fbozqo beqo i bozqo i bozqpo beaob babozqo beqo beqo beqo oaob beaob xubozqo ibeqo beqo xubozqo beqo qoybeqo woaob obeqo babozqo zyowoaob beaofb rikbeqo wiyu zyowoaob woaob wiyykag buwiy qoybeqo oribeqo a beqo bubeqo bozqo qoywiyyag zuqggko bfeoaob buqwoaob bozyiy woaob bebeoaob bibuwiy wiyte bibuwiy bibeqto ribuwiy btabeqo wobeqo.

twiydug zyowtoaob buwity beqo i beaob beaob btozqo aaob zyowoaob biqgaaob buqbofaob xuzyowoaob bebeqo zuqggo buqbiqkg bozwoaob e obeqo a ubeqo ebuwtiy wiyi bozbibeqo ebutwiy biqpgbeqo bawobeqo exfag uibeqo bkozqo wiyu beqo bepwiye ubikbeqo oaaob ufwiye woaob ufyiy abeaob woxo wiyyiy biqgqo aaaob bozwiydug ibeqo xuaaob buqbetaob bawifyyag biqgaaob zyowobkeqo.

kzyowoaob ebeqo oabeaob beqo buwiy xkuxi ibeqo oubeqo rkibaqo bozqo awoaob aaob biqgbeqo u obeqo beqo buqwtoaob buqwiyfi bozqo riwiy bfoi i bozkxo bezyoe bozaob iaaob beqo bei bozbezyoe biqgbeqo buqwoaob rixuqko ariwiy bibeqo bfexag uiqo buwoaob bowiyu exuqo bowiy uriqo opbozqo bozwiyi wifyxi i oxki iebuwiy woaob ebeaob iebuwiy.

twoaob qoyafob biqgwiy zyooubeqo bozbozxo wofaob oiebuwiy buquwiye bozqo bowiyu euibeqo woaob baobeqo riqpoybeqo exukqo buquwiye xuwkiyyag uqpag zyowoaob woyiy bobebeqpo bobuwiy ibowiyu buqwoaob bowiy boztaaob zuqggo iebuwiy xuaaob beqo iqoyaob qoykuriqo zyoyi iwiyu ibobau oribeqo woyag u btawobeqo zyou bkeaob boyag wiywiy bebei aobeqo qoybiqg biqgaaob zyoki bozpqo bozexuqo.

Extra credit:  Tweak the output as needed to make it look good and add some ambiguous illustrations.

Now, before you say it: 

Do I really think this is how the Voynich was created? 

No.  Of course not. 

BUT... I'm going to ask you to look over that output a bit closer.  You are going to see a lot of 'linguistic' traits and Voynich traits.  I know we're used to looking at Voynich and seeing it's patterns, but give it a shot.  

Beqo.  It gets repeated.  Think daiin.

And you get words like ibeqo and obeqo and oubeqo without even really trying.  Then woaob and beaob and bozaob and oaoab.  

Think dain, ain, choldaiin, oldain.

Just sharing something I'm working on as a theory.   I'm pretty well into it and as soon as I have something formalized I'll be sharing it as well.]]> <![CDATA[Are there any VMS-type hoaxes?]]> https://www.voynich.ninja/thread-5951.html Thu, 23 Jul 2026 21:36:52 +0000 ololololo]]> https://www.voynich.ninja/thread-5951.html It's worth wondering if there's any hoax "VMS spirit" at all. For example, it could be written in a made-up, nonsensical language, or filled with nonsensical illustrations, or misleading, etc.

The only person who comes to mind is Samuel Literati Nemesh, and no one else...]]> It's worth wondering if there's any hoax "VMS spirit" at all. For example, it could be written in a made-up, nonsensical language, or filled with nonsensical illustrations, or misleading, etc.

The only person who comes to mind is Samuel Literati Nemesh, and no one else...]]> <![CDATA[Research that needs to be done / research ideas.]]> https://www.voynich.ninja/thread-5945.html Tue, 21 Jul 2026 13:13:52 +0000 Koen G]]> https://www.voynich.ninja/thread-5945.html
But maybe some of you guys have more inspiration? If someone wants to contribute to the research, what are things they can do? More generally, what are research topics that you can think of that you'd like to see anyone explore? 

(Disclaimer: all suggestions made in this thread should be "up for grabs", so don't share research ideas that you want to keep for yourself).]]>
But maybe some of you guys have more inspiration? If someone wants to contribute to the research, what are things they can do? More generally, what are research topics that you can think of that you'd like to see anyone explore? 

(Disclaimer: all suggestions made in this thread should be "up for grabs", so don't share research ideas that you want to keep for yourself).]]> <![CDATA[Singulion reorder based on Davis and Layfield article]]> https://www.voynich.ninja/thread-5940.html Sat, 18 Jul 2026 16:16:37 +0000 mgvtheus]]> https://www.voynich.ninja/thread-5940.html
Based on Davis and Layfield recent article about the order of the quires and pages, I tried to "reorder" the entire MS using their method, but using the ZL3b transliteration.

The results were:


.png   Captura de tela 2026-07-18 130748.png (Size: 14.88 KB / Downloads: 113)

This was done using an Held-Karp algorithm. Now, this is only an hypothesis, and something I noticed is that: Q20 matches the research, but the "edges" of Q13 change (Davis/Layfield propose 77|82, 78|81, 75|84, 76|83, 79|80, my algorithm proposes 79|80, 78|81, 75|84, 76|83, 77|82). This could be due to transliteration differences, or maybe the algorithm used (Held-Karp), but it's something pretty interesting.

By the way, the results that have a 0% improvement (Q15 included) may not be anyway useful, since there is no verified improvement in the reordering.

I ask you to consider the hypothesis and make your own assertions about it. I'm not an expert on the field, so your thoughts are really important.  Smile

Thank you.]]>
Based on Davis and Layfield recent article about the order of the quires and pages, I tried to "reorder" the entire MS using their method, but using the ZL3b transliteration.

The results were:


.png   Captura de tela 2026-07-18 130748.png (Size: 14.88 KB / Downloads: 113)

This was done using an Held-Karp algorithm. Now, this is only an hypothesis, and something I noticed is that: Q20 matches the research, but the "edges" of Q13 change (Davis/Layfield propose 77|82, 78|81, 75|84, 76|83, 79|80, my algorithm proposes 79|80, 78|81, 75|84, 76|83, 77|82). This could be due to transliteration differences, or maybe the algorithm used (Held-Karp), but it's something pretty interesting.

By the way, the results that have a 0% improvement (Q15 included) may not be anyway useful, since there is no verified improvement in the reordering.

I ask you to consider the hypothesis and make your own assertions about it. I'm not an expert on the field, so your thoughts are really important.  Smile

Thank you.]]> <![CDATA[Voynich statistical data]]> https://www.voynich.ninja/thread-5938.html Fri, 17 Jul 2026 16:29:06 +0000 stato]]> https://www.voynich.ninja/thread-5938.html
New user (sorry if I am not posting in the correct place) here.
Is there a collection anywhere of statistical data about the voynich? I am imagining CSVs of bigram frequencies etc. I want to make some original graphics to illustrate what we know about the text.
 
I tried to use the search bar and just browse around but couldn't find what I was looking for.

Thank you!]]>
New user (sorry if I am not posting in the correct place) here.
Is there a collection anywhere of statistical data about the voynich? I am imagining CSVs of bigram frequencies etc. I want to make some original graphics to illustrate what we know about the text.
 
I tried to use the search bar and just browse around but couldn't find what I was looking for.

Thank you!]]> <![CDATA[My "Attempt" to Solve the Manuscript.]]> https://www.voynich.ninja/thread-5932.html Fri, 17 Jul 2026 08:49:34 +0000 J_Voy]]> https://www.voynich.ninja/thread-5932.html
This solution is NOT meant to be taken seriously, and is meant to:
1. Test the confirmation bias hypothesis.
2. Parody other Voynich solutions.

I know that simple substitution is not the right answer, but I tried anyway! Wish you all the best.


.pdf   My Voynich Solution.pdf (Size: 545.85 KB / Downloads: 42) ]]>
This solution is NOT meant to be taken seriously, and is meant to:
1. Test the confirmation bias hypothesis.
2. Parody other Voynich solutions.

I know that simple substitution is not the right answer, but I tried anyway! Wish you all the best.


.pdf   My Voynich Solution.pdf (Size: 545.85 KB / Downloads: 42) ]]> <![CDATA[Hand Copying the Entire Manuscript]]> https://www.voynich.ninja/thread-5890.html Wed, 08 Jul 2026 21:49:07 +0000 seridane]]> https://www.voynich.ninja/thread-5890.html Almost a year and a half ago I registered on this site for the first time and made a hello post where I said I was going to make a copy of the entire manuscript by hand. Never posted or commented since, though I've done a lot more lurking and reading.
And I've done it! 
Almost. 
I have about 2 pages left to ink, and 20 pages plus the Rosette fold out to paint. And of course I have to bind the entire thing. I'm hoping I'll be completely done in two or three weeks.
I'm in the process of scanning and formatting the pages I've completed while I finish the last physical pages, and I think at least one person was interested in seeing it? Here's a link to a goole drive folder with a sampling of completed pages if that's you: You are not allowed to view links. Register or Login to view.

Let me know if there are any particular pages anyone is interested in and I'll add them.
Thanks for looking!
-michelle]]> Almost a year and a half ago I registered on this site for the first time and made a hello post where I said I was going to make a copy of the entire manuscript by hand. Never posted or commented since, though I've done a lot more lurking and reading.
And I've done it! 
Almost. 
I have about 2 pages left to ink, and 20 pages plus the Rosette fold out to paint. And of course I have to bind the entire thing. I'm hoping I'll be completely done in two or three weeks.
I'm in the process of scanning and formatting the pages I've completed while I finish the last physical pages, and I think at least one person was interested in seeing it? Here's a link to a goole drive folder with a sampling of completed pages if that's you: You are not allowed to view links. Register or Login to view.

Let me know if there are any particular pages anyone is interested in and I'll add them.
Thanks for looking!
-michelle]]> <![CDATA[[split] Voynich 2026 Conference QUESTIONS/COMMENTS]]> https://www.voynich.ninja/thread-5871.html Tue, 30 Jun 2026 03:16:35 +0000 BessAgritianin]]> https://www.voynich.ninja/thread-5871.html Note:  Proposed "solutions" will not be accepted."
 If the proposed solutions are not accepted, then what is the purpose of the conference?

This policy actually says :"We want to study the Voynich Manuscript, but we do not actually want anyone to solve it."
It protects the status quo and ensures that the mystery will continue indefinitely.
 Refer my thread:  You are not allowed to view links. Register or Login to view.
  Isn't it time on testing evidence, not just maintaining the mystery?]]> Note:  Proposed "solutions" will not be accepted."
 If the proposed solutions are not accepted, then what is the purpose of the conference?

This policy actually says :"We want to study the Voynich Manuscript, but we do not actually want anyone to solve it."
It protects the status quo and ensures that the mystery will continue indefinitely.
 Refer my thread:  You are not allowed to view links. Register or Login to view.
  Isn't it time on testing evidence, not just maintaining the mystery?]]> <![CDATA[Barbavara Ciphers]]> https://www.voynich.ninja/thread-5828.html Thu, 18 Jun 2026 12:50:07 +0000 Mark Knowles]]> https://www.voynich.ninja/thread-5828.html
His 6 nephews, the brothers->

Marcolino Barbavara's enciphered letters in the Bibliothèque Nationale de France in the Library of Paris

Francesco II Barbavara intercepted letter referring to ciphers that he sent in the Archivio di Stato di Firenze

Bishop Giovanni Barbavara's cipher key in the Archivio Diocesano di Tortona

Abbot Antonio Barbavara who I associated with the authorship of the Voynich manuscript on the basis of a cartographic theory long before I knew of any connection of his family to cryptography

Nicolino Barbavara, Mayor of Vercelli(No known surviving ciphers)

Pietro Barbavara, Apostolic Protonotary(No known surviving ciphers)

Son of Marcolino->

Scipione Barbavara's cipher key in the Tranchedino Cipher Ledger in the Austrian Library in Vienna

---

I am most focused now in locating ciphers written by Francesco II Barbavara.

Unfortunately, I haven't yet seen any surviving document in cipher or not in cipher written by Abbot Antonio. Although, I know a certain amount about him from various sources including his great interest in and collection of varieties of plants as the plaque in the Abbey where he was Abbot states.

----

Surviving evidence shows that the chancellery of Milan was responsible for producing the most advanced and innovative ciphers in the world at the time from which the Voynich manuscript was dated. See, in particular, the 1424 intercepted Milanese enciphered letters in the Archivio di Stato di Firenze.

And the Barbavara family were at the centre of Milanese cryptography. Francesco I Barbavara is credited with having created the first modern foreign office under Duke of Milan, Gian Galeazzo Visconti. Francesco II Barbavara was Ducal Secretary and Head of the chancellery of Duke of Milan, Filippo Maria Visconti. Marcolino Barbavara was a prominent diplomat in the government of Filippo Maria Visconti.]]>
His 6 nephews, the brothers->

Marcolino Barbavara's enciphered letters in the Bibliothèque Nationale de France in the Library of Paris

Francesco II Barbavara intercepted letter referring to ciphers that he sent in the Archivio di Stato di Firenze

Bishop Giovanni Barbavara's cipher key in the Archivio Diocesano di Tortona

Abbot Antonio Barbavara who I associated with the authorship of the Voynich manuscript on the basis of a cartographic theory long before I knew of any connection of his family to cryptography

Nicolino Barbavara, Mayor of Vercelli(No known surviving ciphers)

Pietro Barbavara, Apostolic Protonotary(No known surviving ciphers)

Son of Marcolino->

Scipione Barbavara's cipher key in the Tranchedino Cipher Ledger in the Austrian Library in Vienna

---

I am most focused now in locating ciphers written by Francesco II Barbavara.

Unfortunately, I haven't yet seen any surviving document in cipher or not in cipher written by Abbot Antonio. Although, I know a certain amount about him from various sources including his great interest in and collection of varieties of plants as the plaque in the Abbey where he was Abbot states.

----

Surviving evidence shows that the chancellery of Milan was responsible for producing the most advanced and innovative ciphers in the world at the time from which the Voynich manuscript was dated. See, in particular, the 1424 intercepted Milanese enciphered letters in the Archivio di Stato di Firenze.

And the Barbavara family were at the centre of Milanese cryptography. Francesco I Barbavara is credited with having created the first modern foreign office under Duke of Milan, Gian Galeazzo Visconti. Francesco II Barbavara was Ducal Secretary and Head of the chancellery of Duke of Milan, Filippo Maria Visconti. Marcolino Barbavara was a prominent diplomat in the government of Filippo Maria Visconti.]]>