In the late 15th through 17th centuries, the line between a cryptography manual and a grimoire barely existed — the same scholars who catalogued angels also invented cipher systems, and the two pursuits used the same intellectual toolkit of correspondence, permutation and hidden structure. This chapter includes the most genuinely unresolved material in the entire cipher section: two manuscripts here remain only partially understood even today, and one has never been decoded at all.
Johannes Trithemius — the same Benedictine abbot whose 1518 Polygraphia preserved the Theban alphabet — wrote Steganographia around 1499, though it circulated only in manuscript for over a century before its first print edition in 1606. On its surface, the text reads as a manual for summoning spirits to carry messages across long distances, complete with angelic names, elaborate invocations and instructions for magical communication. For roughly a century, it was treated — and occasionally condemned — as exactly that: a work of practical magic, and the Catholic Church placed it on its Index of banned books in 1609.
The reveal — the "angelic" instructions in Books I and II are, in fact, elaborate cryptographic and steganographic techniques written in deliberately magical-sounding language — a disguise. Later scholarship (most thoroughly, cryptographic historian Jim Reeds in 1998) decoded these sections and confirmed they function as genuine, sophisticated cipher instruction dressed as demonology. The shorter Book III remained more genuinely ambiguous for longer and is still debated, though later analysis suggests it too encodes additional cryptographic content beneath its astrological surface.
Francis Bacon published his biliteral cipher in 1623, in De Augmentis Scientiarum. Rather than substituting one symbol for one letter, it hides a message inside the physical presentation of an entirely different, innocent-looking text: every letter of the true alphabet is represented not by a single character but by a unique five-character sequence built from just two symbols (conventionally "a" and "b"). With five positions and two possible values each, the system yields 32 combinations — comfortably enough to cover the 24-letter Elizabethan alphabet.
Also known as Aldaraia, the Book of Soyga is a 16th-century Latin manuscript consisting largely of dense tables of letters, extensively studied by John Dee, who considered it one of the most important texts he owned — important enough that he reportedly asked his angelic contacts, via Edward Kelley's scrying, for help understanding it. According to Dee's own diaries, the angel Uriel provided a partial key, though even that guidance left most of the book's structure unexplained during Dee's lifetime.
The manuscript itself was considered lost for centuries after Dee's library was dispersed, known only through references in his other papers, until historian Deborah Harkness and cryptographic historian Jim Reeds independently located two surviving copies in the 1990s — one in the British Library, one in the Bodleian Library at Oxford. Reeds subsequently reconstructed the algorithm generating many of the book's 36 letter-tables: each is produced from a keyword using a defined letter-shift procedure, a genuinely elegant piece of 16th-century systematic construction. What remains unresolved is less the mechanical generation of the tables and more their intended magical or divinatory purpose — why these particular 36 keywords, and what practitioners believed the resulting grids were for.
The Voynich Manuscript is the most famous undeciphered document in the world: roughly 240 surviving pages, carbon-dated to 1404–1438, written entirely in an unknown script — now called "Voynichese" — that has never been read by anyone, in any era, as far as any verified record shows. It takes its name from Wilfrid Voynich, the antiquarian book dealer who acquired it in 1912, though its documented ownership history reaches back at least to the court of Rudolf II of Prague in the early 1600s.
The manuscript is heavily illustrated in sections that appear botanical (unidentified plants), astronomical and astrological (zodiac-like diagrams), biological (small human figures in interconnected tubes and pools, sometimes called the "balneological" section), and pharmaceutical (labelled jars and plant parts), followed by a final section of continuous text with no accompanying imagery. Codebreakers of real distinction have attempted it — including cryptanalysts associated with both World War II codebreaking efforts and, later, computational linguistics teams — without producing a decipherment accepted by the wider scholarly community.
Where the evidence actually points — statistical analysis of Voynichese shows it follows patterns real languages exhibit, including a rough version of Zipf's Law (the predictable relationship between a word's frequency and its rank), which has led most serious modern researchers to consider a meaningless hoax less likely than it once seemed. That is different from claiming the text has been read: dozens of proposed solutions — an unknown natural language, a constructed language, an elaborate cipher of a known language, glossolalia transcribed phonetically — remain competing, unproven hypotheses rather than an accepted answer.
Trithemius and Bacon are solved and well-documented; Soyga is partially solved; Voynich is not solved at all. These four systems sit at genuinely different points on the spectrum from "fully understood" to "completely mysterious," and treating them as equivalently mysterious flattens real, important differences in how much is actually known.
The Bacon-Shakespeare biliteral claims are a useful case study in how pattern-matching produces false decryption. Given enough freedom to choose which typeface variations "count," a determined reader can find almost any message they are looking for — the same underlying issue as the digit-sum "angel number" pattern-matching discussed elsewhere in this section.
Voynich deserves particular caution against premature "solved" claims. Multiple decipherments have been announced publicly and covered in mainstream media over the decades, and every one so far has failed to hold up under scrutiny from specialists. As of the most recent serious scholarship, the manuscript remains genuinely, actively unread.