The Lullian Apophatic Machine
Walter Mandron > Yesterday, 09:44 PM
Another new theory About the Voynich Manuscript - oh God --- no!!!!
What follows is an attempt to write down an insight before it dissolves back into the diffuse mass of speculation from which it came. I make no claim to completeness. But I do claim that what I present here is more coherent than anything I have read about this text in recent years — and that is saying something.
The Problem Nobody Will Admit, really?
If you are honest — really honest, without the reflexive modesty that passes for good manners in academic circles (haha) — then you have to admit: we have no idea. After more than a hundred years of intense engagement, after Friedman's anagram excesses in the 1940s, after Feistely's Middle Eastern theories, after Stolfi's China stumble, Zandbergen, Montemurro, after Cheshire, who sent the world into uproar for about two weeks in 2019 before his approach crumbled apart at a leisurely pace like a stale cake — after all of that, the research stands roughly where it stood in 1912, when Wilfrid Voynich pulled the thing from a dusty box at the Villa Mondragone — well, or traded it with some Jesuits for cake for the abbot.
That is not nothing. Although one has also learned a great deal. We know that the vellum carrier was in all likelihood killed between 1404 and 1438. The hotly debated ink will presumably date from the same period. We know that a common typeface in the 1970s correctly identified two distinct writing styles, which he called 'Language A' and 'Language B'. We know that the statistical distribution of the words follows a Zipf curve — one of the few things that actually bothers me about my own theory — roughly as real languages are distributed. We know that certain words cluster in certain sections — differently in the herbal part than in the astronomical part, differently than in the bathtub part — it must be a language, surely.
What we do not know, however, is what any of it means. And that is where the real problem begins: most attempts at explanation implicitly assume that the manuscript contains a message. A content. A recipient. A key. One searches for a language because one believes there must be one. And when you cannot find it, when you have had enough of the endless failing, you arrive at the only conclusion that remains — the whole thing must be a hoax. And if not, you suddenly find yourself examining marginal annotations, because those can apparently be read, which is amusing: if scholars need years to decipher words written in Latin letters, should it surprise anyone that the main text has not been cracked? And the dedicated long-term student of Voynich turns to crowns, noble marks, cloud bands and page numbers; letters are folded in public to present evidence; YouTube videos are produced and forgotten. Every week countless (well, between one and five) AI theories appear, from which one can learn how AI is getting ever better at being bad.
The long story, which is itself only a brief glimpse into the cruelty of being a Voynich student, cut short:
What if the wrong question has been at the root of all of this?
Ramon Llull and the Wheel That Connects Everything
To understand what I mean, one must briefly travel to Mallorca — specifically to the thirteenth century. Ramon Llull, born 1232 or 1233, died 1316, is one of the strangest figures in European medieval history. Troubadour, mystic, missionary, philosopher, possibly stoned to death on a voyage to North Africa (are ladies present?), possibly also simply dead of old age, or done in by exasperated contemporaries — the sources contradict each other. What is beyond dispute: he developed a system he called the Ars Magna, the Great Art, which had an intellectually explosive effect for the following two centuries — much like AI today — or as one tends to say, the LLMs — which already bear a curious resemblance in name alone, and one wonders why nobody made that connection sooner — but I digress.
The core idea was strikingly simple. Llull believed that all truth — theological, philosophical, natural — could be derived from a finite set of divine attributes. Goodness, Greatness, Eternity, Power, Wisdom, Will, Virtue, Truth, Glory — nine fundamental properties of God, which he designated with the letters B through K (for reasons best known to himself). These attributes could now be mechanically combined by arranging them on concentric discs of parchment or wood and rotating these discs against each other. Each position of the wheel generated a new combination, a new theological statement, a new question. One no longer needed to think, in the sense that no creative inspiration was required — one turned, and the system thought for you.
That sounds simple to the computer-accustomed, light-switch-softened, civilisation-sick Voynich student. It is not. What Llull had actually built was one of the first fully mechanical combination machines in history. A system that could generate a potentially infinite corpus of text from a defined vocabulary and defined combination rules. Every phrase the wheel produced was grammatically correct within the system — it followed the rules — but its semantic content was not determined by the author, but by the position of the wheel.
Lullism in the Early Fifteenth Century
Llull was dead in 1316, stone dead, but his ideas were not. Quite the contrary. Lullism (would not LLMism be a fine term?) — the reception, further development and occasional condemnation of his method — experienced a flowering in the fourteenth and early fifteenth century that remains vastly underestimated in the literature. Paris had a chair for Lullism! That alone is strange enough. Padua and Bologna discussed his method within the framework of scholasticism (hear, hear!). In Catalonia, Valencia and the Balearic Islands, it was simply part of the intellectual basic equipment of any scholar who thought more of himself than Cacofonix.
Particularly interesting: Nicholas of Cusa — surely known to all of you — Cusanus, 1401–1464, one of the most important thinkers of the fifteenth century full stop — was demonstrably a Lullist. He not only owned several Llull manuscripts, he studied and annotated them. These manuscripts are held today in the Hospital Library in Kues and have been subjected to intensive investigation, which has been about as successful as Voynich investigations generally. Because nobody understands them — but that is an entirely different story. Why I mention Cusanus is important: he was not sitting on a small stool at the margins of the scholarly world watching the sky. He was a Cardinal (!), a papal legate, at the centre of intellectual Europe in the 1430s through 1460s — precisely the period to which the VMS is dated — interesting, or? Fine, then not.
Still, one must be clear: whoever produced the Voynich Manuscript around 1420 lived in an intellectual world in which Lullian combinatorics was not an esoteric niche phenomenon. It was a tool. It was a method. It was the horizon of thought — it was state of the art.
The wheels, nobody revcognized as wheels
This is where it gets interesting. And I do not mean that in the sense of the academic courtesy filler, but actually physically interesting — so: try to follow me.
The Voynich Manuscript contains several large, multi-page fold-out sheets — so-called foldouts — with circular diagrams. (Do not say you did not know that.) One of those at which every Voynich student, at least once in their Voynich-ruined life, wants to see a cipher key, is Folio 57v: four concentric rings of words and symbols, arranged around a centre with four naked female figures, 17 signs, and much else. Astrology? Horoscopes? Code?
But look at the structure more carefully. They are not horoscopes. Horoscopes of the fifteenth century have a very specific iconographic grammar — twelve houses, zodiac signs, planetary positions. The diagrams in the VMS do not follow this grammar.
What they do have are concentric rings with segmented text divided into sectors. That is exactly the visual structure of a Lullian combination wheel. Exactly. Down to the radial segmentation and the arrangement of units at equal intervals on each ring.
I have spent many hours comparing this and the other circles with the representation of the Ars Magna from. The structural similarity is not coincidental. It is too systematic to be coincidental.
The decisive difference: the Llull wheels are recognisable as what they are. The VMS wheels are in disguise. They are embedded in a botanical, astrological, balneological context that makes them superficially resemble something else. If that was intentional — and I believe it was — then the creator was someone who knew very precisely what he was doing and why he wanted to hide it.
The Statistics That Arise by Themselves
This brings me to the point that has occupied me longest and which in my opinion is the key to everything. Why does the VMS look statistically like language, if it is not one?
The usual answer is: well, it is one — we just have not found the key yet. Or: it is a hitherto unknown natural language. Or, in the more desperate variant: it is meaningless scribble deliberately designed to look like language — a forgery. The forgery argument is interesting, but breaks precisely because the statistical properties of the VMS show things that a forger of the early fifteenth century simply could not have known — that languages follow a Zipf distribution. That knowledge was still a few centuries away (1935). No medieval forger could have replicated that intentionally.
This is where the combinatorial explanation comes in, and it is mathematically elegant. If you assemble a finite vocabulary — say 40 to 60 basic units — into sequences according to fixed combination rules, the language-like statistical properties arise automatically. Not because they were intended. But because they are a mathematical consequence of the rule structure.
Benoit Mandelbrot demonstrated this in a mathematical analysis in 1953: a Zipf-like distribution arises not only in natural languages, but in any rule-governed combinatorial system with certain basic properties. A Lullian wheel with a medium vocabulary and hierarchical combination rules would generate exactly these properties — including the differences between 'Language A' and 'Language B', which could be explained by different wheel configurations or different sets of combination rules.
And most importantly: the scribe would not have needed to plan this. He turns the wheel. He writes down what the wheel shows. The statistics happen to him — as accidentally as a cell that suddenly defined itself as a cell. It is not his work. It is the work of the combinatorics itself — nothing more.
Apophatic Theology as Context
Now the question that many interpreters skip over, because they cannot grasp it, so foreign is medieval thinking to us today — although large parts of our brain are still fighting over the base instincts of millions of years ago.
Why, why and for what purpose would someone around 1420 disguise a Lullian combination machine as a herbal?
The answer lies, I believe, in apophatic theology — negative theology — as it was intensively discussed in the fourteenth and early fifteenth century especially in mystical circles in northern Europe and Italy. The core idea: God is not knowable through positive thinking, through statements like 'God is good' or 'God is wise'. Because every such statement limits. God transcends all categories. The mind reaches the highest knowledge — or at least the closest approach to it — not in the moment of understanding, but in the moment of failure. When thought encounters its own boundary and recognises it.
Meister Eckhart had taught this even before Llull's death. Tauler, Seuse, the Frankfurter Mystik of the fourteenth century — all of that still lived in the libraries and minds of monasteries in 1420. Pseudo-Dionysius the Areopagite, the principal witness of the apophatic tradition, was standard reading. Nicholas of Cusa, our Lullist, developed from this his concept of docta ignorantia — learned ignorance, knowledge through deliberate not-knowing.
A meditation tool built on these foundations would do the following: it would present the observer with sequences that look meaning-like — that activate all the cognitive patterns that normally signal meaning. Word lengths, word repetitions, sentence structures. But it would deliver no meaning. The mind would run against a wall again and again. And that wall, not despite but because of its impenetrability, would be the point.
The manuscript is not a message. It is an experience ex machina (quite the comparison) — it is a vision of failure at form. It served — today we would say — to meditate on the being of failure. Read the lines, try to understand them, drive your mind to its limits in order to know God. From today's perspective this sounds like madness. It is not.
The Illustrations as Lullian Combination Domains
What then about the plants? The naked women in the bathtubs? The astronomical diagrams?
In the Lullian Ars Magna, the combination wheels do not operate in a vacuum. Each wheel belongs to a subject area — a domain in which the combinations apply. Llull developed wheels for medicine, for theology, for law, for natural science. The illustrations in the VMS could be read as visual labelling of these domains. The herbal section not as a herbal, but as a marker: here the wheel operates in the domain of the natural, of plants, of healing. Which is why the roots and the blossoms and the stems are all mixed up. The astronomical section: the wheel in the domain of cosmology, of stars and of depth. The balneological section — the women in the pools — without question the strangest part, but even in the Lullian tradition there are wheels that treat the human, the bodily, the living as a domain.
The plants do not need to be identifiable. That is not a bug, it is a feature. Were they identifiable, the manuscript would be readable as a herbal — and with that, the disguise of the wheels would unravel. Only non-identifiable plants fulfil their function completely: they mark the domain without anchoring it to real objects that would invite comparisons and thereby questions.
Who Could This Have Been?
This is the question at which I am most cautious, because it is the most speculative. I will name no names I cannot substantiate; I have an idea, but no. But I can describe the frame.
The creator must have been: firstly, a trained Lullist who not only knew the Ars Magna but had applied it practically. That considerably narrows the circle of candidates. Secondly, someone with access to the apophatic mystical literature — Pseudo-Dionysius, Eckhart, possibly Cusanus, whose early writings were circulating from the 1420s onward. Thirdly, someone with a reason to hide it. Lullism was not without danger everywhere. The University of Paris had condemned Llull's teachings in 1374. Anyone using a Lullian system as a meditation tool for mystical purposes was operating in territory that could attract ecclesiastical attention.
Northern Italy, first third of the fifteenth century: that is the place of origin most frequently proposed on palaeographic and codicological grounds. Zandbergen and others have argued this convincingly. In northern Italy — Venice, Padua, Milan — Lullism was alive, mystical currents were active, and learned men existed who knew and combined both.
I will not speculate further here. Not because it is impossible, but because any specification at this point weakens the theory without strengthening it.
Why It Cannot Be Deciphered — and Why That Is Not an Objection
The obvious criticism of what I describe here: if that is true, why has nobody found the combination wheel? Why can the mechanism not be reconstructed?
The answer is simple: because it was not written down. In the logic of this project — building a meditation tool deliberately designed to be undecipherable — it would be self-defeating to include the key. The wheel perhaps existed physically, as a real object of parchment or wood. Or it existed in the head of the scribe, who moved through the combinations by routine. After his death: gone. No key, no wheel, no context. Only the product.
This is not a special case. There are medieval documents whose production mechanism could be reconstructed because accompanying documents survive. There are others where this is not possible because the entire production context was lost. The VMS has no known provenance before the early seventeenth century — the first secure evidence is the letter from Georg Baresch to Athanasius Kircher in 1639, in which Baresch describes the manuscript and asks for help with decipherment. Everything before that is hypothesis.
And here is the decisive point I want to emphasise: the absence of a reconstructable key is not evidence against this theory. It is consistent with it. A system designed to prevent the mind from understanding leaves, by definition, no trace of its mechanism. Undecipherability is not the symptom of a lost language. It is the result of a successful construction.
What Remains
I have no illusions. This is not provable. None of the theories about the Voynich Manuscript are provable, as long as no document from the fifteenth century surfaces that explicitly describes the context of its creation — and even then doubts would be possible. I have learned to live with this state. One must, if one engages with the VMS for any length of time — otherwise one ends up making pointless VMS YouTube videos, abandoning one's blog in disappointment, arguing about colours and plants, or defending hoax theories with the most baroque sophistry.
What I do claim: this explanation is more coherent than its alternatives. It requires no unknown language. It requires no anachronistic assumptions about how people thought in the early fifteenth century. It explains the statistical properties of the text without magic. It explains the structure of the diagrams. It explains the non-identifiability of the plants. It explains why the manuscript falls into several thematically distinct sections. And it explains why nobody has deciphered it in 600 years.
Not because it is too hard. But because there is nothing to decipher.
The manuscript is finished. It has served its purpose. Somewhere, sometime, someone opened this book, looked at the pages, let their mind drift across the words, and ran into the wall. Again and again. And in doing so, perhaps, briefly touched something larger than what words can say.
Whether that is theology or madness, I leave open.
In my own case, I know: it is madness.