Cipher Arguments: A Quantitative Test

Of all the arguments in the authorship debate, the cipher arguments make the boldest promise: that the true author concealed his or her name inside the works themselves, in code, for a later reader to uncover. If even one of them held, it would settle the question outright, not a balance of probabilities but a signature read straight off the page. That is why the claims keep returning, and why they earn a careful test rather than a wave of the hand.

 

They come in several forms, and for most of the leading candidates: numerical and page-number ciphers built on the Earl of Oxford’s title as the “17th Earl”; Francis Bacon’s biliteral cipher and various acrostics; letter-ciphers advanced for Marlowe, Sir Henry Neville, and others. Most of the major candidacies have acquired a cipher claim of some description. This page works through the most fully developed examples on either side (the numerical and letter-sequence ciphers offered for Oxford, and the biliteral cipher and acrostic offered for Bacon), and holds every one of them, whoever proposes it, to a single standard. The same test would apply, and apply the same way, to a cipher advanced for Marlowe, Neville, or anyone else. Before the worked examples, three short sections lay the groundwork: why the cipher tradition has genuine traction, what a valid cipher has to do, and why the astronomical odds these arguments quote prove nothing.

 

One point should be made before the analysis begins. The Elizabethan cipher tradition is real, and dismissing it as a crank obsession misreads the period. Writers of the era built numerical and letter devices into serious work as a matter of course: Spenser structured the Epithalamion around twenty-four stanzas and three hundred and sixty-five long lines; Sir John Davies built his entire Hymnes of Astraea as twenty-six acrostics on the Queen’s name; even Francis Meres, so often cited against the Oxfordian case, wrote a devotional tract called Gods Arithmeticke. The idea that a concealed author might leave a coded trace in this milieu is not inherently absurd. The cipher arguments examined below take their appeal from something genuine. The question is not whether such devices were real (they were) but whether these particular claimed examples can carry evidential weight. They cannot, and the reason is precise: a genuine cipher has a finish line that everyone recognises when it has been crossed. These do not.

 

Why the cipher tradition has genuine traction

Elizabethan England ran on codes. Francis Walsingham’s intelligence bureau employed professional cryptographers, Thomas Phelippes decoded Mary Queen of Scots’ Babington correspondence, and she went to the scaffold for it. John Dee used “007” as his personal cipher signature in letters to Elizabeth I. Francis Bacon described genuine cipher systems in De Augmentis Scientiarum (1623), including an original invention, the biliteral cipher. Christopher Marlowe served as a government intelligencer. Numerological and Hermetic traditions (Kabbalistic letter-values, Pythagorean number symbolism) were serious intellectual concerns among the educated elite. The idea that a concealed author might leave a coded trace is not inherently absurd in this context.

 

The anagram has its own genuine tradition, too. In 1610 Galileo announced a discovery to Johannes Kepler in a Latin anagram (Haec immatura a me iam frustra leguntur, o.y.) rearranging to Cynthiae figuras aemulatur mater amorum (“the phases of Cynthia are imitated by the mother of love”: Venus, like the moon, shows phases). Two letters, o and y, are left over in the announcement and appear nowhere in the solution, the opposite of Ballantine’s method examined below, which adds letters where the source falls short. A genuine anagram cipher declares its leftovers rather than solving around them, and its plaintext is never less clear than its ciphertext.

 

What matters is the distinction between these real practices and what the arguments below claim. Walsingham’s codes, Dee’s signature, and Bacon’s biliteral cipher were all operational: they transmitted messages between specific parties who shared a key, could be decoded by a known method, and produced a unique solution. The SAQ cipher arguments claim something structurally different (a hidden authorial signature embedded in published texts, with no contemporary recipient, discoverable only by selecting whichever list or grid yields the desired result). That is not how Elizabethan ciphers worked. The tests applied in each entry below ask the same questions any genuine cipher must answer: was the rule fixed before the search began, does it yield a unique solution, and are the odds calculated over every text examined, not just the one that produced a hit?

 

What a valid cipher argument requires

Claims about hidden codes, acrostics, and number patterns in the Shakespeare texts have been advanced for Bacon, Oxford, Marlowe, and others. Mainstream textual scholars and cryptographers have consistently found these methods to be subjective and non-reproducible; they generate hits for multiple incompatible candidates and add nothing to a serious evidential case.

 

The root problem is not the size of the search. Running millions of permutations through a ciphertext is exactly how codes get broken, and it is no objection to the authorship ciphers that their proponents have tried many grids, many spacings, many letter-paths. Cryptanalysts do the same. The difference is the finish line.

 

When the Enigma traffic was broken at Bletchley, the finish line announced itself without ambiguity: exactly one key setting turned the whole intercept into fluent, continuous German. The same setting decrypted the next day’s messages. Any operator who applied it recovered the identical text. Nobody had to decide whether the output “counted”, the solution was unique, and everyone recognised it as unique regardless of what they had hoped to find. No authorship cipher reaches that finish line. The same 144 letters of the Sonnets dedication yield “Henry” at fifteen columns, “Wriothesley” at eighteen, and “De Vere lies here” at nineteen, three researchers, three messages, each finding the name he carried into the search. The search is not the problem. It is the finish line that does not exist, and without one the result is in the eye of the beholder and cannot be reproduced.

 

The underlying cognitive hazard has a name: apophenia, coined by the psychiatrist Klaus Conrad in 1958 to describe the “unmotivated seeing of connections,” the brain’s tendency to perceive meaningful patterns in random or unrelated data. It is not a pathology; it is a normal feature of human pattern-recognition, and it operates with particular force when a motivated searcher is given a very large text. The Shakespeare canon runs to roughly 900,000 words. Given that volume of letters, almost any target name can be extracted as an acrostic, a reversed sequence, or a numerical cipher if the search rules are adjusted until a hit appears. The tell is always the same: the same method, applied without adjustment, produces hits for Bacon, for Oxford, for Marlowe, and for candidates the researcher does not favour, which means it is detecting noise, not signal.

 

The definitive demolition is not sceptic opinion but mainstream cryptology. William and Elizebeth Friedman, two of the twentieth century’s foremost codebreakers, examined every cipher system proposed for the Shakespeare works and showed that none meets the elementary standards of valid decipherment. Their study, The Shakespearean Ciphers Examined (Cambridge University Press, 1957), won the Folger Shakespeare Library’s own literary prize. (For the Baconian charge that the Friedmans faked this book, and why it does not hold, see Were the Friedmans frauds? below.) The Friedmans established three criteria that any valid cipher claim must satisfy:

 

• Rule pre-specified. The decoding rule must be fully stated before the search begins, not adjusted until a hit appears.
• Unique solution. The same rule, applied independently by a different operator, must yield the same message, not a different name or a different reading.
• Odds over all trials. The probability of the result must be calculated across the full set of texts examined, not just the one that produced a hit.

 

These are the conditions that define the finish line. None of the seven entries below meets all three. Each shows how a specific version fails against the Friedman standard.

 

One scope note. The finish-line standard governs any claim that a hidden message or name was deliberately encoded and can be recovered, which is what every authorship cipher below asserts. Whether an early modern text also carries deliberate non-message structure—symbolic, mnemonic, or emblematic—is a separate question, of the kind raised in the Colombo section at the end; such a claim is judged on historical motivation, prior specification, controls, and robustness, and even when granted it does not by itself identify an author. The entries here are assessed as recoverable decipherments, which is what their proponents claim them to be.

Why the astronomical odds prove nothing

Cipher arguments almost always arrive with a colossal number attached: one in a hundred million, one in 10²⁶, one in a thousand billion. The number is meant to end the discussion: surely a coincidence that unlikely cannot be a coincidence. It can, and it fails the same way every time, so let me explain it once, here, before the entries. Several of them quote a figure of exactly this kind.

 

The figure always answers the wrong question. It calculates the odds of finding this exact result in one fixed place, read one fixed way, as though the searcher had named the target in advance and pointed to a single spot. That is never what happens. The result is found: by searching a long text, under rules the searcher is free to adjust, accepting any of many outcomes as a success. The probability that matters is not “how unlikely is this one hit?” but “given everything I was free to try, how likely was I to find some hit?”, and that probability is usually close to certain.

 

It is the difference between two lottery questions. “What were the odds that this ticket would win?” is astronomical. “What were the odds that some ticket would win?” is roughly one, once millions of tickets have been sold. A cipher hunter computes the first number and shows you the second situation: he holds up the winning ticket and quotes the odds against it, without mentioning the millions of losing tickets (the other passages, the other reading rules, the other names) bought and discarded along the way. Announce the winning number after the draw and any winner looks like a miracle.

 

Four freedoms inflate the true odds, and an honest calculation has to pay for all of them:

• Where to look. The whole text is in play, every line, every page, every column. The more places you may search, the more near-misses turn into hits.
• How to read. A single letter or a whole word; up, down, or across; every second letter or every fifth; skipping the lines that don’t cooperate. Each rule you are allowed to add multiplies the strings the text can be made to yield.
• What counts as a hit. If Bacon, Vere, Tobey, Neville and a dozen others would all have been accepted, and if spelling is elastic (“Vere,” “Ver,” “Vear”), you are not seeking one target but any of many.
• What to ignore. Leftover letters that spell nothing are quietly dropped. A method that never has to account for its misses can simply wait for its hits.

 

It helps to make that denominator concrete. Ask not “what were the odds of this name?” but “how many results would the searcher have hailed as a hit?” For an Oxfordian, “DE VERE” would do, but so would “VERE,” “E. VER,” “OXFORD,” or simply 17. A Baconian would have stopped at “BACON,” “TOBEY,” or “FR. B.”; a Marlovian at “MARLOWE” or “KIT”; a partisan of one of the others at “NEVILLE,” “DERBY,” “STANLEY,” “SIDNEY,” “WRIOTHESLEY,” or “SOUTHAMPTON.” Each may be read forwards or backwards, in any of several Elizabethan spellings, and taken as a full name, a surname, a set of initials, or a title. Multiply the candidates by the spellings by the directions by the reading-units and the list of strings that would have “counted” is not one but many hundreds. Every one of them belongs on the bottom of the fraction, and each name you add divides the quoted odds again. A few hundred acceptable targets shave a couple of orders of magnitude off a figure like 10²⁶ straight away; the far larger factor is the search itself (the tens of thousands of lines and the freedom in how to read them), and multiplying the two together is what erases the number entirely.

 

And that is the generous version, because it assumes the list can be written down. It cannot. No searcher records in advance the full set of things he would have been delighted to find: the nicknames, the near-misses waved through as “close enough,” the targets he never thought of until the letters suggested them. A genuine probability needs its winning outcomes fixed before the draw; this one has no such list, and none could be drawn up honestly after the fact. So the astonishing figure is not merely too large by some factor we could compute if we tried; it is undefined: a numerator with no denominator, presented as a certainty.

 

A number that ignores these, that divides by one when it should divide by the size of the whole search, is not evidence, however many zeros it carries. This is the third Friedman condition stated above: the odds must be reckoned over every trial, not just the successful one. It is the same failure that, in a different setting, governs correspondence arguments, the denominator problem: a probability with no denominator is not a probability at all. So when an astonishing figure appears in a cipher claim, one question deflates it: over how large a search was it computed? If the answer is “one,” the figure carries no weight.

The method

The worked examples begin with the page-number ciphers, which are specifically Oxfordian. The claim is that Edward de Vere, 17th Earl of Oxford, concealed his ordinal title, the number 17, in the page numbers of the First Folio’s table of contents. It only works for a candidate who has an ordinal number to hide, which is why it is made for Oxford and not for Marlowe, Bacon, or Mary Sidney: an untitled writer has no ordinal to encode, and a “17th Earl” is the whole of the trick. The most developed version, the “17 Solution,” applies digit addition to page numbers arranged in a grid, with paths summing to 17 taken as evidence for the 17th Earl.

 

The test has three steps. First, identify the method’s rules precisely. Second, apply those rules to a folio written by a known author who was not de Vere. Third, apply the same rules to the Shakespeare Folio with a different candidate’s number as the target. If the method is detecting authorship, it should find its target only in the right place. If it is not, it will find targets everywhere, which is what the data show.

 

The rules used in Entries 1–4 below:

• Take the page numbers as listed in a folio’s table of contents.
• Any non-empty subset of the digits in a page number may be combined by addition.
• Lay the entries in a fixed-column grid; digit values from adjacent cells (horizontally or vertically touching) may be summed across entries.
• A path “hits” if the digit values sum to the target number.

 

These rules are flexible enough to use any subset of digits and paths of two or more adjacent cells. The question is not whether the method can find the target (with sufficient flexibility it always can) but whether it finds the target at a rate above chance, and only in the right folio.

Entry 1: The number 17 as de Vere’s hidden signature

Because Edward de Vere was the 17th Earl of Oxford, Oxfordians argue that occurrences of 17 in the Shakespeare works (in page numbers, sonnet groupings, cipher grids, and scene counts) are the author’s concealed signature.

 

The argument is appealing in its simplicity: the author hid his identity but left his number as a trace. The problem is methodological, and it applies to every version of the claim.

 

The selection problem. The First Folio runs to over 900 pages. The Sonnets contain 154 numbered poems. A researcher choosing which numbers, which pages, and which arithmetic operations to apply has an enormous range before finding the target. The “17 Solution”, the most developed version of this argument, applies four rules: add individual digits of page numbers; reuse digits between sequential clues; digits must touch horizontally or vertically in a grid; sum must equal 17. Rules this flexible yield almost any target from almost any table of contents. The relevant question is how many candidate paths were tested before the ones summing to 17 were found. That number is never given, because recording the failures would expose the method.

 

The Bible Codes parallel. In 1994 a peer-reviewed paper in Statistical Science claimed the Hebrew Bible encoded names and birth dates of prominent rabbis in equally-spaced letter sequences. Brendan McKay and colleagues then showed the same technique applied to War and Peace produced equally “significant” results, and applied to Moby Dick it predicted the assassinations of Gandhi, Martin Luther King, JFK, Lincoln, and Princess Diana. Their conclusion: a method flexible enough to find a target in one long text will find it in any long text. The same conclusion applies here. The drferris68 website, which applies “Array 17, Row 17, Column 17” grid analysis to Elizabethan texts, finds 17 not only in Shakespeare but in Drayton, Sidney, England’s Helicon, and Cardanus Comforte, texts de Vere did not write. That is not evidence for authorship; it is evidence that the method generates positives everywhere.

 

The Sonnet grouping is editorial, not authorial. The “Procreation Sonnets”, Sonnets 1–17, are a modern critical label. The 1609 Quarto contains no section breaks, no titles, no numbered groupings. The sequence runs to 154 without internal division. The ending at 17 reflects a theme (the young man urged to have children), not a counted signature. If the grouping were authorial and deliberate, some mark in the original text would signal it. There is none.

 

The Folio’s page-numbering errors are printing errors. The First Folio is famously mis-paginated: the Comedies section skips from page 156 to page 257; Much Ado and Merry Wives have duplicate page numbers; Troilus and Cressida, inserted late, has no page numbers at all. The 17 Solution treats these as intentional cipher mechanism. Charlton Hinman’s The Printing and Proof-reading of the First Folio (1963), the definitive study, maps all these irregularities to identifiable compositor error patterns. No cipher is needed to explain what printing history already accounts for.

 

The same printer’s-house explanation applies to every comparable folio of the period. Ben Jonson’s 1616 Workes, the first English author to collect his own plays in folio, was printed by William Stansby, who was simultaneously running Ralegh’s History of the World and Purchas’s Pilgrimage through his press. The Cambridge Works of Ben Jonson records misprinted page numbers at pages 6, 7, 8, 34, 176, and 713 of the Jonson folio, the same class of compositor error, from the same period, from the same commercial pressures. If the Shakespeare Folio’s pagination errors are cipher, so are Jonson’s. If Jonson’s are printing errors (and no one doubts it), so are Shakespeare’s.

 

De Vere died nineteen years before the Folio appeared. De Vere died in 1604. The Folio was published in 1623. For the cipher to work, de Vere must have arranged the page-number scheme nearly two decades before publication, with reliable posthumous execution by complicit compositors. No documentation of any such arrangement exists.

 

An objection: the arrangement need not have been Oxford’s. Someone in his posthumous circle may have made the decision to encode the tribute after his death: the Folio editors, his son-in-law Philip Herbert (one of the dedicatees), or others who knew his identity. This removes the pre-arrangement problem. It does not remove the others. Documentation of any such scheme (by Oxford or by anyone else) is still absent. The motivation question becomes sharper, not easier: if the people assembling the Folio were prepared to encode a tribute to Oxford in the page numbers, why would they simultaneously publish front matter, dedications, and commendatory verses that say nothing about him? Encoding the identification in an obscure numerical pattern while suppressing it everywhere else is a stranger choice than encoding it nowhere. And the complicit-compositor problem remains regardless of who gave the instruction.

 

The base rate. 17 is a common prime that appears throughout Elizabethan literature in dates, measurements, list counts, and scene divisions. The probability of finding it somewhere significant in a 900-page book, once you have committed to looking, is high by chance alone.

Entry 2: The counter-demonstration, Derby’s “6” with grid

This section is a worked demonstration, not a Derby argument. It is constructed deliberately to show that the method applied to de Vere can be reproduced for any candidate using real text.

 

William Stanley was the 6th Earl of Derby, an alternative candidate. His number is 6, directly. No preliminary arithmetic required. Here is what the cipher method finds.

 

Step 1: Establish the key number. The 6th Earl’s number is 6. It is also the sum of the first three integers: 1+2+3 = 6. Pythagoras called it a perfect number. The author chose it with care.

 

Step 2: The Folio confirms it structurally. The First Folio contains 36 plays. 36 = 6 × 6. The 6th Earl encoded his number by squaring it in the total play count.

 

Step 3: The name contains the signature. “Stanley” has exactly 6 letters. The pen name “Shakespeare” has 11 letters: 1+1 = 2, the number of the two Earls most closely associated with the Folio’s publication (Pembroke and Montgomery, 3rd and 4th Earls respectively: 3+4=7; 7−1=6. The Derby number again).

 

Step 4: The grid. Assign each letter of “WILLIAM STANLEY” its ordinal position in the alphabet (A=1, B=2 … Z=26). Reduce any two-digit value by digit addition (so W=23 → 2+3=5). Lay the two words across two rows:

  W    I    L    L    I    A    M
  5    9    3    3    9    1    4

  S    T    A    N    L    E    Y
  1    2    1    5    3    5    7

 

Now trace paths through adjacent cells, horizontal or vertical, exactly as the 17 Solution does with the Folio page-number grid.

 

Path A (column 1, vertical): W reduces to 5. S reduces to 1. They are vertically adjacent. 5 + 1 = 6. The author’s initials, W and S, shared with the pen name “William Shakespeare”, reduce and sum to the Derby number.

 

Path B (columns 3–4, horizontal): The double-L at the heart of “wiLLiam.” Each L reduces to 3. They are horizontally adjacent. 3 + 3 = 6. Two threes, side by side, embedding the Derby number in the Christian name itself.

 

Path C (row 2, three adjacent cells, permitted by the method’s rules, shown for transparency): T=2, A=1, N=5. 2+1+5=8. Not 6.

(Path C fails. In a genuine 17 Solution presentation, it would not appear. The analyst shows only the paths that work.)

 

What the grid does not show you. The grid has 14 cells and 19 adjacent pairs. Three of those pairs sum to 6. The analyst presents the two successful paths as discoveries. The 17 unsuccessful pairs are not mentioned. The ratio of hits to attempts (roughly 2 in 19, or about 10%) is what you would expect by chance when the target is a single-digit number and the values range from 1 to 9. There is no discovery. There is only selection.

Entry 3: The Jonson folio test

Applying the 17 Solution to Ben Jonson’s 1616 Workes, and finding 17 there too

The decisive test of any cipher method is whether it also produces results in texts the alleged author did not write. If the 17 Solution’s rules find 17 in the Shakespeare Folio and also find 17 in other Elizabethan folios, the method is not identifying an author; it is generating positives everywhere.

 

The ideal test case is Ben Jonson’s 1616 Workes. Jonson’s collected plays were published by William Stansby in 1616, the first time any English author had collected his own plays in a folio. The volume runs to 1,015 pages and contains nine plays, two poetry collections, six entertainments, and thirteen masques.

 

It is the most pointed test for three reasons:

 

1. Jonson himself wrote the commendatory poem in the Shakespeare 1623 Folio, “To the Memory of My Beloved, the Author, Mr William Shakespeare.” If de Vere planted 17s in the Shakespeare Folio, either Jonson was complicit or de Vere also planted 17s in Jonson’s own Workes.

 

2. The Jonson 1616 Folio has documented pagination errors of the same kind as the Shakespeare 1623 Folio. The Cambridge Works of Ben Jonson records: pages 6, 7, 8, 34, 176, and 713 misprinted; additional errors in uncorrected states. Stansby was simultaneously printing Ralegh’s History of the World and Purchas’s Pilgrimage, scholarly consensus attributes the errors to the chaos of a busy press, exactly as Hinman’s analysis attributes the Shakespeare Folio’s errors to compositor confusion. The 17 Solution treats the Shakespeare pagination errors as intentional cipher. By the same logic, Jonson’s must be too.

 

3. De Vere died in 1604, twelve years before the Jonson Folio appeared. Even the 19-year posthumous gap for the Shakespeare Folio strains credibility. For de Vere to have pre-arranged cipher page-numbers in Jonson’s 1616 Workes, he would have needed to do so 12 years before the book existed, in a volume whose contents Jonson himself was still writing after de Vere’s death.

 

The page numbers. The Jonson 1616 Folio’s starting pages, reconstructed from its collation formula (signatures A–4Q4, in sixes):

 1. Every Man In His Humour ............... p.   1
 2. Every Man Out of His Humour ........... p.  73
 3. Cynthia’s Revels ...................... p. 177
 4. Poetaster ............................ p. 271
 5. Sejanus His Fall ...................... p. 355
 6. Volpone .............................. p. 439
 7. Epicoene ............................. p. 525
 8. The Alchemist ........................ p. 601
 9. Catiline His Conspiracy ............... p. 679
10. Epigrams ............................. p. 765
11. The Forest ........................... p. 819  ← digits 8+9 = 17
12. Entertainments ....................... p. 841
13. Masques .............................. p. 891  ← digits 8+9 = 17

 

Two entries independently contain the digits summing to 17, without combining digits across entries at all:

• The Forest begins at page 819: digits 8, 1, 9. Take 8 and 9: 8+9 = 17.
• Masques begins at page 891: digits 8, 9, 1. Take 8 and 9: 8+9 = 17.

 

In a 4-column grid, two adjacent-pair paths also emerge: The Forest (p.819, digit 9) and Entertainments (p.841, digit 8) are horizontally adjacent, 9+8=17; Catiline (p.679, digit 9) and Masques (p.891, digit 8) are vertically adjacent, 9+8=17.

 

Applying the same rules across the table. The 13 entries give 2 single-entry hits and 15 of the 18 adjacent pairs in the four-column grid summing to 17. That is already most of the table; the fuller picture comes from the whole book.

 

But the table of contents is only 13 entries. The 17 Solution does not confine itself to tables of contents; it draws on page numbers throughout the text. The Jonson 1616 Folio runs to 1,015 sequentially numbered pages. Applying the same digit-subset rule to all of them:

• 111 pages (10.9%) independently contain digit subsets summing to 17.
• 674 of 1,014 consecutive page pairs (66.5%) yield 17 under the same rules.
• 900 of 1,013 three-page runs (88.8%) yield 17.

 

Two-thirds of all adjacent page openings in the Jonson 1616 Folio encode de Vere’s number. This is not because de Vere arranged the pagination of a book published twelve years after his death. It is because the digit-subset rule extracts a target sum from nearly any pair of multi-digit numbers. The method is measuring its own flexibility, not detecting a signal.

Entry 4: The same method, the same Folio, but finding 6

Entry 3 applied the 17 Solution method to a different folio. This entry applies it to the same Shakespeare Folio, with target 6 instead of 17. The result proves that the method’s rules cannot distinguish between candidates: they find both numbers equally.

 

If de Vere encoded his number, 17, in the Shakespeare Folio’s page numbers, and if the 17 Solution reliably detects it there, then applying the identical method with target 6 should find nothing, or very little. Stanley’s number is not de Vere’s number. The Folio belongs to one author.

 

Here is what the method actually finds.

 

Solo hits, pages that independently contain 6 as a digit subset. Applying the same digit-subset rule to the Shakespeare First Folio Catalogue (35 plays, excluding the uncatalogued Troilus):

 p. 46  1 Henry IV              (digit 6 present ✓)
 p. 61  Measure for Measure     (digit 6 present ✓)
 p. 69  Henry V                 (digit 6 present ✓)
 p. 96  1 Henry VI              (digit 6 present ✓)
 p.145  Midsummer Night’s Dream (1+5 = 6 ✓)
 p.152  Hamlet                  (1+5 = 6 ✓)
 p.163  Merchant of Venice      (digit 6 present ✓)
 p.185  As You Like It          (1+5 = 6 ✓)
 p.346  Antony and Cleopatra    (digit 6 present ✓)
 p.369  Cymbeline               (digit 6 present ✓)

 

Ten of 35 plays (29%) independently encode the Derby number. Without combining digits across entries, without a grid, the method has already placed Stanley’s number in Hamlet, The Merchant of Venice, As You Like It, and Antony and Cleopatra.

 

Grid paths. In a 4-column grid the 35 entries produce 57 adjacent cell pairs. Applying the same digit-subset rule with target 6: 42 of 57 adjacent pairs (73.7%) yield a path summing to 6.

 

A selection of the adjacent-pair hits:

Tempest p.1 [1] + Comedy of Errors p.85 [5] = 6
Merry Wives p.38 [3] + Love’s Labour’s Lost p.122 [3] = 6
Merchant of Venice p.163 [1] + As You Like It p.185 [5] = 6
Romeo and Juliet p.53 [3] + Hamlet p.152 [3] = 6
Macbeth p.131 [1] + Hamlet p.152 [5] = 6
Antony and Cleopatra p.346 [3] + Cymbeline p.369 [3] = 6

 

Stanley’s number appears in vertically and horizontally adjacent cells throughout the grid (early plays, middle plays, late tragedies) exactly as the 17 Solution claims de Vere’s cipher does.

 

The complete matrix. The table below shows two Shakespeare measurements side by side: the ToC grid just demonstrated above, and a full-section test using the same consecutive-pair method as the Jonson column, all printed page numbers across the three sections (Comedies, Histories, Tragedies), 826 pairs in total. If the method is discriminating, only the true author’s number should produce high hit rates, and only in the folio he wrote.

 

Adjacent pairs summing to target	De Vere / 17	Stanley / 6
Shakespeare Folio (ToC grid, 57 pairs)	20/57 (35.1%)	42/57 (73.7%)
Shakespeare Folio (full sections, 826 pairs)	421/826 (51.0%)	561/826 (67.9%)
Jonson Folio (full book, 1,014 pairs)	674/1,014 (66.5%)	683/1,014 (67.4%)

 

The proponent’s evidence is the 35.1% figure, de Vere’s number in the Shakespeare ToC grid. It is the lowest cell in the table. Every other cell is higher. A discriminating cipher should produce near-zero in the wrong-candidate/wrong-text cells and a high value only in the true-author/true-text cell. This method puts its lowest figure in the cell it is supposed to identify.

 

Within the same ToC grid, Stanley’s number finds 73.7%, more than double de Vere’s figure in the works he supposedly authored. Expanding to the full sections of the Folio (826 consecutive pairs, the same method used for Jonson) changes the absolute figures but not the conclusion: de Vere’s number reaches 51.0% and Stanley’s 67.9%. De Vere’s number finds 66.5% in Jonson’s Folio, fifteen points higher than it finds in Shakespeare. Most tellingly, Stanley’s number finds 67.9% in Shakespeare and 67.4% in Jonson, a difference of less than one percentage point. The method cannot tell the two folios apart. It is not detecting authorship; it is measuring how freely multi-digit page numbers can be subdivided.

 

The method cannot tell the candidates apart, cannot tell the folios apart, and cannot produce a result below 35%. It is not detecting authorship. It is confirming that multi-digit page numbers routinely contain digit subsets summing to any target between 1 and 17, and that the hit rate is elevated everywhere because the rules are flexible enough to make it so.

 

Note on sources. Page numbers are as listed in the First Folio Catalogue (1623). Henry V is catalogued at p.69, which falls before 2 Henry IV’s p.74, one of the Folio’s documented pagination errors, which the 17 Solution treats as cipher mechanism. If those errors are intentional, they are Stanley’s cipher too.

Note on the table figures. Shakespeare ToC row: the 1623 First Folio catalogue lists 35 plays with section-specific page numbers (Troilus and Cressida has no catalogue entry). The 35 entries are arranged in a 4-column grid, producing 57 adjacent cell pairs (horizontal and vertical). For each pair, the digits from both page numbers were pooled and every non-empty subset tested: the pair counts as a hit if any subset sum equals the target. 20 of 57 pairs hit 17; 42 of 57 hit 6. Shakespeare full-sections row: all printed page numbers from the three main sections of the Folio, Comedies (pages 1–304), Histories (pages 1–232), and Tragedies (pages 1–293), verified against the Internet Shakespeare Editions facsimile of the National Library of New South Wales copy (SLNSW). For each of 826 consecutive page pairs within sections, the same pooled subset-sum test was applied. 421 of 826 hit 17; 561 of 826 hit 6. Jonson Folio row: the 1616 Workes (William Stansby, London) runs to 1,015 pages. The same test applied to each of the 1,014 consecutive page pairs (pages 1–2, 2–3, …). 674 pairs hit 17; 683 hit 6. Every figure in Entries 3 and 4 is regenerated by the script in the method appendix.

Entry 5: The Sonnets dedication, finding a name by letter skip

Proponents of Edward de Vere’s authorship have argued that the 1609 dedication to the Shakespeare Sonnets, a 144-letter document, contains the full name “Henry Wriothesley” encoded as equidistant letter sequences, and that a second cipher in the same letters yields “These Sonnets All By Ever”, read as “E. Ver.” The calculated odds against both findings arising by chance are claimed to be astronomical. The argument has surface appeal because the mathematics is precise. The precision conceals a series of undisclosed choices.

 

One preliminary point in the argument’s favour: acrostics and hidden-name devices were a genuine Elizabethan literary practice, not an anachronistic fantasy. Sir John Davies encoded “ELIZABETHA REGINA” in his poem Orchestra (1596) and built his entire Hymnes of Astraea (1599) as twenty-six acrostic poems openly structured around the Queen’s name. These were overt displays of technical craft; readers were meant to find them. The Friedmans themselves acknowledged that a genuine acrostic, properly established, “must be taken as conclusive.” The question is not whether hidden names can exist in Elizabethan texts (they can) but whether this particular claimed example meets the standard required to establish one.

 

Henry Wriothesley, 3rd Earl of Southampton, was the only person Shakespeare publicly dedicated his published poetry to: Venus and Adonis (1593) and The Rape of Lucrece (1594) both name him on their title pages. The 1609 Sonnets dedication, published by Thomas Thorpe and signed “T. T.,” is 144 letters long, set with a period after every word and an unusual inverted-triangle layout. Dr John M. Rollett, a British physicist, argued in 1997 that reading every 15th letter of the text yields “HENRY,” and reading every 18th letter yields fragments that can be assembled as “WRIOTHESLEY.” Since the only ‘y’ in the 144 letters appears in both strings, the odds against “HENRY” alone are roughly 1,000–to–1; the full name in combination is claimed to reach 100 million to one against chance. A separate word-skip pattern (reading the 6th, 2nd, and 4th words in sequence) yields “These Sonnets All By Ever,” taken as “E. Ver” = Edward de Vere.

 

Two different spacings for the same name. “Henry” is found at letter-skip 15. “Wriothesley” is found at letter-skip 18. If the dedication encodes a single person’s name, why does each half require a different key? The answer is that the analyst tried multiple spacings and reported the ones that worked. The 100,000,000–to–1 figure is calculated for the specific fragments actually found; it does not divide by the number of spacings tried before finding them.

 

Fragment flexibility. Rollett reads “Wriothesley” in three fragments at x18 spacing, reading some letters upward and some downward. An independent engineer who later reviewed the work found a different four-fragment reading at the same x18 spacing, reading exclusively downward, and preferred it on methodological grounds. Both readings yield the same name; neither is the unique, pre-specified solution a genuine cryptogram produces. When two analysts find different fragment combinations from the same spacing, the odds calculation becomes unreliable, because it was built around one specific arrangement.

 

Two contradictory ciphers in 144 letters. The name cipher identifies Wriothesley as the “Fair Youth”, the “Mr W. H.” of the dedication. The word-skip cipher in the same text simultaneously identifies de Vere as the author. Oxfordians cite both as confirmation of the same theory. But if both are genuine independent encodings, their joint probability should be calculated as the product of both, and each encoding must be tested against the null hypothesis that the other already exhausts the available letters. That calculation is never presented. Two alleged discoveries in 144 letters, each treated as statistically independent of the other, is not stronger evidence. It is a sign that the search space is larger than the stated odds reflect.

 

The target is not independent of the search. Henry Wriothesley’s name appears in plain text on the title pages of both poems Shakespeare publicly dedicated. He is the most famous name already connected to Shakespeare’s published work. A researcher searching for names in the Sonnets dedication and finding Wriothesley has not made a surprise discovery from a blind search. The correct prior probability is not the probability of finding any arbitrary 16-letter name in a 144-letter text; it is considerably higher, because the most likely candidate name was chosen before the search began.

 

The dedication’s author is not Shakespeare. The 1609 dedication is signed “T. T.”, Thomas Thorpe, the publisher. Any cipher in the dedication was placed there by Thorpe, not by the author of the plays. Even if the cryptogram is genuine, it records what Thorpe believed or chose to encode. Assigning any structure in the dedication to Shakespeare, or to de Vere, would need independent evidence about who composed, authorised, and set the page, so the finding does not reach the playwright on its own.

 

The Friedman criteria. William and Elizebeth Friedman’s The Shakespearean Ciphers Examined (Cambridge University Press, 1957), the standard reference on Elizabethan cryptanalysis, established three tests for a valid cipher claim: the decoding rule must be fully specified before the search begins; it must yield a unique solution; and the probability calculation must account for all trials, not just the successful one. The Sonnets dedication analysis fails all three. The spacing rule differs between the two halves of the name. Two different analysts derive different fragment readings from the same spacing. The odds figure is calculated without disclosing the search space. The mathematical precision of the final number is real. The number it is attached to is not.

A direct test: the same method, run on all 154 sonnets

Entry 5 argues that the letter-skip method hides its choices. Here is the simplest way to see it, not as an argument, but as an experiment anyone can repeat.

 

A letter-skip cipher works like this: pick a starting letter, then take every Nth letter after it (every 15th, every 18th, whatever you choose), reading straight through the text once, never looping back to the start; string the results together and see whether they spell a name. We took the full text of Shakespeare’s Sonnets (72,734 letters) and had a computer try every skip, in both directions, searching not only for de Vere but for rival candidates and, as a control, for three people who lived long after Shakespeare died.

 

Start with the very name the cipher claims: HENRY WRIOTHESLEY. Point the method at it and the two halves behave completely differently. HENRY (five letters) turns up almost at once, at a skip of 6; but that is no feat, because a five-letter string is expected to occur about 180 times in this much text by chance. WRIOTHESLEY (eleven letters) appears nowhere: not at any of the 7,273 skips at which an eleven-letter name could fit in the Sonnets. An eleven-letter name is expected to occur roughly once in four million texts this size, so a clean letter-skip essentially never produces one. Which raises the obvious question about the original claim: if the name cannot be read off a clean skip in 72,734 letters, how was it read off 144? It wasn’t. Every 18th letter of the 144-letter dedication yields only eight letters, not eleven; “WRIOTHESLEY” was assembled from scattered fragments, the reading free to jump and double back. The half of the claim that is real (HENRY) is meaningless; the half that is impressive (WRIOTHESLEY) is manufactured.

 

First, a sense of scale. Allowing skips of only up to 120, the Sonnets already contain 308,662 of the 456,976 possible four-letter strings, 68 percent of them. More than two of every three four-letter “signatures” you could invent are already sitting in the text: VERE, yes, but also BASS and meaningless clusters like FAPP and MECK. Finding “VERE” is finding one straw in a haystack that is almost entirely straw.

 

Then we searched for full names. The smallest skip at which each one first appears:

string	letters	smallest letter-skip that spells it
VERE	4	1, appears in ordinary text (“sovereign”)
BASSANO	7	222
EMILIA	6	69
LANIER	6	104
BACON	5	5
MARLOWE	7	1,526
NIXON	5	4
OBAMA	5	28
ELVIS	5	141
OXFORD	6	not found (all 14,546 possible skips)
WRIOTHESLEY	11	not found (all 7,273 possible skips)

 

The method spells out Emilia Bassano Lanier in full (BASSANO at skip 222, with EMILIA and LANIER as well). It spells NIXON, OBAMA, and ELVIS, three people born centuries after Shakespeare; NIXON turns up at skip 4, sooner than any real candidate, ahead of BACON at 5 and DE VERE at 13. And it cannot find OXFORD at all, not at any of the 14,546 skips at which a six-letter word could fit in the text.

 

Notice OXFORD at the foot of the table, unfound, while the longer BASSANO and MARLOWE are found. That is not an error, and length is not the whole story: what a name needs is not to be short but to be built from common letters. OXFORD carries an X (only 60 of them in all 72,734 letters, against nearly five thousand each of A and S), so there is almost no X to thread an equidistant line through. BASSANO, though longer, uses nothing rarer than B. Whether a name surfaces is a lottery of the letters it happens to contain, not its length and certainly not its history, which is exactly why the method turns up Bassano, Nixon, and Elvis but never the Earl of Oxford’s own title.

 

A large skip does not mean the reading runs off the end of the poem and starts over; there is no wrapping. It simply spreads the letters farther apart: spelling a seven-letter name such as MARLOWE at its skip of 1,526 places its letters across about 9,200 letters of the poem, still well inside the Sonnets’ 72,734, far apart, but read in a single pass. (The largest skip at which a seven-letter name could fit at all is about 12,000; beyond that the text runs out.)

 

Which names emerge is decided by nothing more than how common their letters are in English. It is the same reason a large enough bowl of alphabet soup will, somewhere, spell your name, not because the soup knows you, but because there are so many letters. A code-detector that confirms Elvis Presley and misses the Earl of Oxford’s own title is reading noise and calling the pieces it likes a signal, not a hidden message.

 

“But it takes 880 skips to reach Bassano and only 4 to reach Nixon, surely the small skip counts for more?” The intuition is backwards, for two reasons. First, the skip at which a name first appears measures only how common its letters are in English; short names built from frequent letters surface at tiny skips automatically, which is just what “expected to occur many times over by chance” means. A small skip is the fingerprint of a common string, that is, of noise. Second, and decisively, apply the “fewer skips, more serious” rule consistently and it demolishes the case it was meant to save: in this very search NIXON appears at skip 4 and BACON at skip 5, smaller than DE VERE’s skip of 13, while OXFORD never appears at all. Ranking candidates by fewest skips therefore puts a twentieth-century president and a rival candidate ahead of Edward de Vere, and the Earl of Oxford dead last. A yardstick that crowns Richard Nixon has measured nothing.

 

The root of the illusion is what statisticians call the look-elsewhere effect. The advertised odds, “a hundred million to one”, are computed as though a single pattern were ever tried. But the search ranges over thousands of skips, both directions, several spellings of a name, and the freedom to keep whichever fragment looks best and quietly drop the rest. Count the chances actually taken and a hit at some small skip is not a miracle; it is the expected outcome, which is why the same procedure hands back Nixon, Obama, and Elvis.

 

This is the same demonstration Brendan McKay used to dismantle the Bible Code, carried out here on Shakespeare’s own Sonnets. The test is a few lines of code and the text is public; anyone can repeat it and check every number. The text used here is the Project Gutenberg edition (eBook #1041); the script and the text regenerate every figure above. (The reading is a single straight pass; letting the text wrap around would only produce more names, not fewer.)

The same method, run on the dedication’s own letters

The 154-sonnet test above can be run on the dedication itself. Taking the 1609 text as printed (146 letters in this transcription), a single-skip letter search, forwards or backwards, hands back VERE at skip 19 and HENRY at skip 15, the second being exactly Rollett’s published result, which confirms the procedure is his. The operative half of “de Vere” duly drops out.

 

So does almost everything else. The identical rule—one fixed skip, read in either direction—pulls 242 different words of four to seven letters out of those 146 letters, 34 of them five letters or longer, taken from a standard American-English word list (which includes common proper names). Alongside VERE sit ESPN, GREG, RENE, REID and VEER. VERE is one entry on a long list, and its appearance is exactly what that many letters predict. The transcription, the word list, and a short script that reproduces the count and the two named targets are in the reproducible script.

 

Two features of the exercise are worth stating plainly, because they are the point. The letter count is itself a choice: 146 here, 144 in Rollett’s version, the difference being whether one counts the “Mr W H” and “T T” initials, the hyphenated compounds, and U against V. And the long names never fall from a single clean skip; “Wriothesley” and a full “Edward de Vere” need the fragmenting and mixed directions their finders allowed themselves, which is where the method comes apart. Short targets drop out on their own, buried among the other 240.

Entry 6: The 1740 argument, de Vere’s composite number

Of all the cipher arguments considered on this page, Alexander Waugh’s 1740 analysis is the most elaborate. His Westminster decode (the grid that extracts “De Vere lies here” from the 144 letters of the Sonnets dedication, with directional indicators he reads as pointing toward Poets’ Corner) carries into the First Folio by the same method, and it is built on a genuine historical uncertainty about de Vere’s burial. The parts that are real prove nothing about authorship, and the parts built on them require a different rule at every step, which is the tell of selection, not discovery.

 

What 1740 is supposed to be. De Vere’s two numbers are 17 (his ordinal title as the 17th Earl of Oxford) and 40 (derived from de Vere’s “double V” signature: in the Latin alphabet V is the 20th letter, so VV = 2×20 = 40). Combined, they produce 1740. Waugh argues that de Vere used this composite number as a recurring signature, and that its appearance across the pen name, the monuments, Greek Christological symbolism, and the Sonnets dedication is too consistent to be coincidental.

 

The pen name. In the Classical Latin alphabet W does not exist; it is written as two Vs (VV, “double V”). V is the 20th letter in that alphabet, so VV = 40. The letters that follow, “illiam Shakespeare”, number exactly 17. So “WILLIAM SHAKESPEARE” reads: VV (=40) + 17 letters = 1740. The pen name, on this reading, is not a pseudonym concealing identity but a numerical signature advertising it.

 

Where else 1740 appears. Waugh finds the number in several locations:

 

• The Westminster Abbey monument (1740). Erected that year, not earlier; the Dean and Chapter had granted permission in 1726 but installation was delayed fourteen years. Waugh argues the delay was deliberate, to align with de Vere’s number. The monument was partly funded by Alexander Pope, a Freemason.
• The Tempest inscription on the same monument. The base of the Westminster Abbey monument carries a passage from The Tempest (IV.i): “The cloud capt tow’rs, the gorgeous palaces, / The solemn temples, the great globe itself, / Yea all which it inherit, shall dissolve, / And like this insubstantial pageant faded, / Leave not a rack behind.” This passage is reordered from the First Folio text, and “towers” is contracted to “tow’rs,” making the first line exactly 17 letters. The four lines each begin with T (The / The / The / Tempest), which Waugh reads as four Ts = 40, again encoding the year 1740.
• The ChiRho symbol. Rho (ρ) is the 17th letter of the Greek alphabet, encoding de Vere’s title number. Waugh assigns Chi (χ) a value of 40 (imported from the VV Latin gematria argument already claimed separately), making the combined symbol a numerical encoding of 1740. The ChiRho is the early Christian monogram for Christ (the first two letters of the Greek χριστός) and appears throughout Elizabethan hermeticism.
• The Sonnets dedication grid. Arranged in 19 columns, the 144 letters of the Thorpe dedication yield “De Vere lies here” and directional information pointing to Poets’ Corner, Westminster Abbey. The word-skip code (6th, 2nd, and 4th words) yields “These sonnets all by Ever the forth T”, “Ever” being the Elizabethan pun on E. Vere, and “forth T” the fourth T = 40.
• The Stratford monument plaque. The anomalous space between “with” and “in” (“placed with in this monument”) signals that Shakespeare was “placed with” three poets named in the Latin couplet above, all buried in Poets’ Corner, not “within this monument.”

 

What is genuinely uncertain. Before the methodological critique, a word on what is real. Percival Golding, de Vere’s cousin, wrote around 1620 that de Vere “lieth buried at Westminster.” The parish register of St Augustine, Hackney records a burial there in July 1604, and Elizabeth de Vere’s 1612 will asks to be buried “as near to the body of my late dear and noble Lord and husband as may be.” The two primary sources are in tension. Golding’s statement is the Oxfordians’ strongest card on the burial question; the parish register and the countess’s will point to Hackney. The location of de Vere’s grave is a genuine historical uncertainty: St Augustine’s was demolished in 1798, removing the physical record, and no excavation beneath the Westminster monument has been conducted. This uncertainty is real. It says nothing about who wrote the plays.

 

A different rule for each location. The pen name argument uses Classical Latin alphabet values. The ChiRho argument uses Greek New Testament gematria. The monument inscription argument uses a letter count that depends on accepting a contracted form as intentional. The dedication argument uses a 19-column grid. The Stratford plaque argument is a syntactic re-reading. None of these is the same method. A system that requires a different rule for each positive finding is not a system; it is a collection of individually contrived observations that share only their conclusion.

 

The ChiRho argument has two internal problems. First, Chi does not equal 40 in standard Greek numerals (isopsephy): Chi (χ) = 600; Mu (μ) = 40. The value 40 is not derived from the Greek system at all; it is carried over from the VV Latin gematria argument (V = 20th letter, VV = 40), presented here as independent confirmation but actually the same assertion twice. Second, the symbol is ChiRho, not RhoChi: Chi comes first. If Chi = 40 and Rho = 17, reading the symbol left-to-right yields 4017, not 1740. To produce 1740 by concatenation you need Rho first (RhoChi), which reverses the order of the actual monogram. The famous early Christian symbol only “encodes” 1740 if you read it backwards.

 

The column count is unjustified. The dedication grid uses 19 columns. Rollett’s earlier analysis of the same dedication used 18 columns and found “Wriothesley.” A 15-column grid finds “Henry.” Different column counts produce different messages from the same 144 letters. The choice of 19 is not independently motivated; it is the count that produces the desired result. Waugh himself acknowledged trying “hundreds of different grids” before settling on 19 columns. That admission is the argument’s own refutation: it is the clearest possible statement that the finish line was drawn after the race was run.

 

The contraction is standard, not coded. “Tow’rs” as a contracted form of “towers” is routine throughout the Shakespeare verse plays, used for metrical purposes wherever a disyllable would disrupt the pentameter. Counting an elided letter as intentional cipher encoding is a method that will find almost any number in any Elizabethan verse text.

 

The 14-year delay has a documented explanation. The Westminster monument’s installation was delayed from 1726 to 1740 because fundraising was slow. The London theatre community mounted a benefit performance in 1741 to complete the financing. A fundraising timeline is not a coded date.

 

The pen name argument is circular. The construction VV (40) + “illiam Shakespeare” (17 letters) = 1740 requires that de Vere chose the pen name. But that is precisely what the argument is trying to prove. The pen name cannot simultaneously be evidence for the authorship and a product of the authorship. And the construction requires the Classical Latin alphabet, a choice that is not independently justified, and that yields VV = 40 only because V happens to be the 20th letter of that system.

 

The same flexibility works for Stanley. William Stanley’s title number is 6. S is the 19th letter of the English alphabet (1+9=10; 1+0=1) and W is the 23rd (2+3=5); together they reduce to 6. The Folio’s 36 plays equal 6×6. The Sonnets have 154 poems (1+5+4=10; 1+0=1... but six times that is 6). The Westminster monument has six columns of stone. This demonstration is deliberately absurd, but it is constructed using exactly the level of arithmetic flexibility the 1740 argument requires. When a method can prove anything, it proves nothing.

Entry 7: Positional placement, the name at 17. Or 18

A class of cipher argument holds that the name “Shakespeare” or “William Shakespeare” appears at the 17th position in certain texts, or immediately after a count of 17 items, making it the 18th item. Because 17 is identified as de Vere’s number, this positional appearance is offered as the author’s hidden counter-signature. The method can be applied to any ordered sequence: words, lines, titles, or names in a list.

 

The counting problem is structural. Any name in any sufficiently long list will occupy some position. “Shakespeare” will land at position 17 in some documents and not others, depending entirely on what else is in the list and how it is ordered, not on any authorial intention. A method that searches texts until it finds the target name at the target position is not deciphering a hidden message. It is selecting the document that happens to produce the desired result. The relevant question, how many sequences were examined before one yielded the name at position 17 or 18, is never given.

 

Accepting two adjacent positions doubles the hit rate. The argument claims both position 17 and position 18 (“after a count of 17 items”) as valid hits. A method that accepts either of two adjacent slots as confirming evidence has roughly twice the expected hit rate of one that accepts only one. As the target window expands, the probability of a hit by chance rises correspondingly, without any change in the underlying evidential value.

 

Counter-demonstration. Consider the most celebrated Elizabethan literary list: Francis Meres, Palladis Tamia (1598). Its most-quoted passage reads: “the English tongue is mightily enriched … by Sir Philip Sidney, Spenser, Daniel, Drayton, Warner, Shakespeare, Marlowe and Chapman.” By positional logic, Shakespeare is the sixth poet named, which would encode Derby (the 6th Earl), not Oxford. Yet the same document, in its comedy playwrights list, places “Edward Earl of Oxford” first and Shakespeare ninth; in the tragedy list Shakespeare is tenth; in the lyric poets list, fourth. The same source, in the space of a few pages, “encodes” four different earldom numbers. Any of them would furnish a plausible-sounding positional argument if selected after the search; none was specified before it.

 

A variant: the Meres symmetry argument. A more sophisticated Oxfordian reading holds that positional analysis of Meres is unnecessary, Meres himself signalled the identity by violating his own design rule. Throughout Palladis Tamia, Meres maintains a strict symmetry between the number of classical and English writers in each comparison: he explicitly says, in his paragraph on iambic poets, that he curtailed the Greek list to match the number of English writers he could find: “Among the Greeks I will name but two for Iambics … so amongst us I name but two iambichal poets … because I have seene no mo in this kind.” The comedy playwrights list (paragraph 34) breaks this pattern: it has 16 names on the Greco-Roman side and 17 on the English side, “off by one.” Since the asymmetry is deliberate (the argument goes), two of the 17 English names must stand for one person. Oxford and Shakespeare are both on the English comedy list. Therefore, Meres intended us to understand them as the same person.

 

The argument is more careful than the simple positional versions, and it deserves a precise response. Two problems are structural.

 

First, the rule “an off-by-one asymmetry signals identity” is derived from the same data it is used to interpret. The analysis shows that Meres violates his symmetry rule in paragraph 34, and he infers that the violation is intentional. But to conclude that the violation points to Oxford and Shakespeare specifically requires a further step: selecting those two names as the identity pair out of seventeen possibilities. The logic of the argument (any two names on the longer side could be the intended pair) does not itself pick Oxford and Shakespeare. That identification is supplied from outside Meres, from prior belief about who the real author was. The off-by-one fact constrains very little; it does not narrow to a single pair without external input.

 

Third, and most directly, the iambic passage cited as evidence of Meres’ commitment to symmetry actually demonstrates a different mechanism from the one the comedy argument requires, and on a close reading, it undermines it.

 

The Oxfordian reading of the iambic passage is this: Meres had more Greek iambic poets available but chose to list only two, to match the two English iambic poets he knew of. Symmetry was his governing design principle; he curtailed the longer side to achieve it.

 

A simpler reading is equally available. “Because I have seene no mo in this kind” refers to the English poets: Meres could only find two English iambic writers worth naming, so he listed only two Greek ones to correspond. The English side was the natural constraint. Meres was not imposing aesthetic balance on abundant material; he was acknowledging scarcity. On this reading the iambic passage is evidence of intellectual honesty, not of a design programme.

 

Both readings agree on one thing: the mechanism is reduction. Meres trims the potentially longer side down to match the shorter side. Now look at what the comedy argument requires:

 

Meres’ comedy playwrights list, Palladis Tamia, paragraph 34

Classical (16 names)	English (17 names)
Menander	Edward Earl of Oxford
Aristophanes	Doctor Gager of Oxford
Eupolis Atheniensis	Master Rowley
Alexis Terius	Master Edwardes
Nicostratus	John Lilly
Amipsias Atheniensis	Lodge
Anaxedrides Rhodius	Gascoyne
Aristonymus	Greene
Archippus Atheniensis	Shakespeare
Callias Atheniensis	Thomas Nash
Plautus	Thomas Heywood
Terence	Anthony Mundye
Naevius	Chapman
Sext. Turpilius	Porter
Licinius Imbrex	Wilson
Virgilius Romanus	Hathway
,	Henry Chettle

 

The English side has 17 names; the classical side has 16. For the symmetry argument to work, Meres must have deliberately added a seventeenth English name beyond a natural count of 16, engineering an imbalance as a hidden signal. That is the opposite of what the iambic example shows. The iambic example shows a writer who, finding himself with only two examples on the shorter side, said so plainly and matched the longer side to it. The comedy argument requires a writer who, finding himself with a naturally balanced set of 16 per side, reached for a seventeenth English name precisely to break the symmetry he otherwise maintains, and then said nothing about it.

 

There is also nothing internal to the list that marks either Oxford or Shakespeare as the added name. Any of the 17 English entries could be the supposed addition. The argument selects Oxford and Shakespeare as the pair because they are the names the theory needs, not because the list points to them.

 

And more fundamentally, the Meres counter-demonstration above already shows that Meres’ lists yield mutually contradictory positional signals. The comedy list puts Shakespeare ninth and Oxford first. The tragedy list puts Shakespeare tenth, with Oxford absent. The lyric list puts Shakespeare fourth. The enriched-English list puts Shakespeare sixth. Each of these is a different “signal” from the same document published within the same few pages. The symmetry argument adds one more: the comedy asymmetry. What it does not do is explain why this signal, out of all those the document produces, is the one to credit. Selecting the asymmetry argument after the positional arguments failed the wrong-candidate test is exactly the pattern the Friedman standard was designed to catch: a result is not valid if the rule that identifies it was chosen because it produced the desired output.

 

Friedman test failure. The three Friedman criteria (rule pre-specified before the search; unique solution; odds calculated over all trials) are not met. No independent specification identifies which texts are to be searched or which ordering principle applies. The same text, ordered differently (by date, by length, alphabetically), places the name at a different position. And the probability of a “hit” is calculated for the single chosen list, not for the full set of lists that were examined and discarded before the successful one was found. A genuine positional cipher would produce the same name at the same position under any independent application of the stated rule. A positional search does not.

Acrostics: the “Bacon–Tobey” signature in The Tempest

A Baconian example, and the single move that manufactures it

The cipher claims rebutted above are Oxfordian, the ordinal “17” and its variants (Derby’s “6” appears only as a constructed counter-demonstration). This example is Baconian, and an acrostic rather than a number cipher: a different candidate and a different technique, and the same failure appears regardless of who is being sought or how. The mechanism that produces the “hidden name” sits right on the surface. The best-known example reads the first letters of the lines on the opening page of The Tempest in the First Folio and reports the message “SIT THE DIAL AT NBW, F. BACON, TOBEY”, Tobie Matthew being Bacon’s friend and, in his own words, his “alter ego.” The odds against it are put at 1.8 × 10²⁶ to one. Both the message and the number dissolve on inspection.

 

The reading unit changes from line to line. Watch how “F. BACON” and “TOBEY” are actually assembled from the lines of the passage:

  For thou must now know farther        → take one letter:  F
  Begun to tell me what I am           → take one letter:  B
  And left me to a bootless inquisition → take one letter:  A
  Concluding, stay, not yet            → take three letters: CON
  The howr’s now come                 → take one letter:  T
  Obey, and be attentive               → take four letters:  OBEY

  Claimed reading:   F · B · A · CON · T · OBEY   =  “F. BACON, TOBEY”
  Consistent rule:   F · B · A · C   · T · O      =  “FBACTO”

 

For three lines the method takes a single initial letter; for the other two it silently takes the whole first word, “Concluding,” “Obey”, exactly where a whole word is needed to complete the name. There is no rule that says when to take one letter and when to take the word; the analyst takes whichever spells the target. Apply one consistent rule (one initial per line, which is what the word “acrostic” means) and the same six lines give FBACTO. The name exists only because the unit is allowed to move. This is the letter-level twin of the page-number method’s freedom to choose which digits to add: the flexibility is the whole engine.

 

The rest is read out of order, with the leftovers ignored. The opening “SIT THE DIAL AT” is not read downward at all; the line-initials actually run T–A–D–I–A–L–THE–T–I–S, which are then anagrammed into a phrase, leaving “NBW” sitting in the middle of the plaintext as unexplained residue. Lines whose initials would break the string (“More to know,” “You have often”) are simply skipped. A genuine acrostic reads straight down, in sequence, with nothing discarded and nothing rearranged: Sir John Davies’s real acrostics spell ELIZABETHA REGINA cleanly, in order, and were meant to be found. This one requires reordering, tolerated junk, and selective omission before it yields anything.

 

The giant odds are the weakest part of the claim, not the strongest. Its headline is that a message like this could arise by chance only once in 1.8 × 10²⁶ tries. But that figure is computed on the one string that was found, as though the searcher had named it in advance and pointed to a single spot, when in fact he ranged over the whole Folio, mixed single letters with whole words, read in several directions, discarded the leftovers (“NBW”), and would have accepted any of a dozen names. Counted honestly, over everything he was free to try, a hit like this is not surprising but expected. (This is the general trap set out above under Why the astronomical odds prove nothing.)

 

And the number quietly backfires. To make it impressive, the argument borrows the Friedmans’ own letter-frequency table and quotes their one-in-a-billion standard, then, in the same breath, says the Friedmans never examined this passage and can be set aside. You cannot lean on their authority to build your odds and wave it away to keep your conclusion. And the standard it borrows is the very one it fails: the Friedmans required the odds to be counted over every attempt, not just the lucky one. (The claim invokes “William Friedman” of the NSA; the table is in fact from the 1957 book he co-wrote with Elizebeth S. Friedman, the cryptologist Fabyan first hired at Riverbank, in 1916, to hunt the Bacon cipher in the First Folio.) The Friedmans examined the method, and set the three conditions, rule fixed in advance, unique solution, odds over all trials, that this reading breaks on every count. (See Were the Friedmans frauds? for the standard itself and the charge that they faked it.)

Structural decoding: a hidden claim read from arrangement

A distinct class of argument claims a message was concealed not in the letters or numbers of a text but in the way its parts are arranged: the order of a list, which items are paired, what stands next to what, the shape of a layout. Because it works on organisation rather than on letter-values, it is usually defended as “not a cipher, just the structure.” It meets the standard set out above, and fails it for the same reason.

 

The claim has to be kept apart from a legitimate one it resembles. Describing a structure is ordinary scholarship, and checkable: Francis Meres does pair English poets with classical ones, and Thomas Thorpe did set the 1609 Sonnets dedication as an inverted triangle with a point after every word. Anyone can confirm those features on the page. Structural decoding is the further step of reading a specific hidden proposition out of the arrangement—“this pairing means Oxford is Shakespeare,” “this word-order spells E. Ver”—and claiming evidential weight for it. The test that divides the two is a single question: does the reading recover something the text also states openly, or that an independent source can confirm, or does it depend on the arrangement carrying a message available only through the decoder? Only the second is structural decoding.

 

No finish line. A genuine encoding has a key that any reader re-applies to reach the identical result. An arrangement-reading has none: the decoder chooses which feature to read (order, adjacency, position, symmetry), and a second decoder, choosing differently, recovers a different message from the same text.

 

A larger search space. A list of seventeen or so names, or a dedication of thirty-odd words, offers a great many describable arrangement-features: sequence, adjacency, pairing, position, mirror-symmetry. The more features on offer, the more surely a motivated search finds one that fits the answer already in mind. It is the base-rate problem set out under Why the astronomical odds prove nothing, with more knobs to turn than a letter-skip allows.

 

No control. The method is never run where it could be caught failing. Does the same reading pull a false name out of the same list, or a true one out of an unrelated commonplace book or dedication? Until it is tested that way, what it measures is the ingenuity of the reader. Like the letter and number ciphers, it can be aimed at any candidate, so the rule holds whoever it favours.

 

The class shows up twice in these pages. The Sonnets dedication: the word-order reading “These Sonnets All By Ever,” taken as “E. Ver,” is a structural decode of Thomas Thorpe’s word sequence. Describing the dedication’s lapidary layout is legitimate, and the plain-text reading of “our ever-living poet” is a separate, arguable matter of usage set out on the Ever-Living Poet page; the decode is the part that carries no weight. The equivalent letter-cipher on the same dedication is examined at Entry 5. Meres’s Palladis Tamia: the claim that the arrangement of names in the 1598 list encodes the identification of Oxford as Shakespeare is a structural decode of the same kind, discussed with the plain reading that answers it on the de Vere page.

Geometric pattern arguments: a further class

A variant of the cipher argument applies not to letter sequences or number grids but to the geometric properties of a printed page or engraving: hidden angles, proportions, diagonals, and symmetries claimed to encode authorial identity.

 

The Friedman criteria apply with the same force here as to letter-skip arguments, but with one additional problem. A letter-sequence search operates over a finite alphabet and a countable set of spacings. A geometric search operates over continuous space: every angle, every ratio, every alignment between image elements and text is a candidate relationship. The searcher’s eye is not sampling randomly from that set; it is drawn to whatever looks significant, which means selection is always post-hoc, even when the analyst is not aware of it.

 

The question that cuts through geometric arguments is the same one that cuts through letter-skip arguments: how many relationships were examined before the chosen ones were found? But here, unlike letter-skip ciphers, no number is actually germane. Even “I tried only three” does not help, because those three were not chosen by a pre-specified rule; they were chosen because they looked interesting. The only satisfying answer is not a number at all. It is a documented prior specification, written before the analysis began, committing to which geometric relationship would be tested and why.

 

A further problem arises when a geometric argument is explicitly built to reconcile two prior claims. If the project’s stated purpose is to synthesise findings A and B, then the relationships “discovered” are constrained to connect what A and B already found. The rule is not independent of the result; it was selected to bridge two known endpoints. That is selection bias built into the project’s purpose, not an inadvertent methodological error.

 

A common response to these objections is: “But look at the interesting relationship I discovered.” The word “interesting” is doing exactly the work it should not be doing. In a genuine cipher, the result announces itself without the analyst deciding it is interesting, everyone who applies the key recovers the same output, recognisable as meaningful independently of what they hoped to find. “Interesting” is what apophenia feels like from the inside. The response to any claimed geometric discovery is not to dispute whether the relationship exists (it may well exist) but to ask whether it would have been noticed if it had not pointed somewhere the analyst already wanted to go.

Text-embedded anagrams: the one class that would bear on authorship

Every argument above shares a weakness beyond the finish line: a page number, a typeface, a title-page grid, the proportions of an engraving, all belong to the printing house, not the pen. At most they show who saw a book through the press. A cipher can bear on authorship only if it sits where none but the author could have placed it: inside the dialogue and the verse. That distinction is decisive, because it disqualifies most of the tradition at a stroke, and because one body of work does clear it.

 

Roberta Ballantine’s Marlowe Up Close (2000) argues that roughly a third of the text credited to Shakespeare is Christopher Marlowe’s: linked anagrams read two pentameter lines at a time, the opening couplet a signed declaration of authorship, the technique adapted from the Attic tragedians (she cites Thompson and Padover’s Secret Diplomacy). Because the unit is fixed by metre (two lines of blank verse are the same twenty syllables however they are set), it survives changes of layout, and because the messages live in the spoken text, they are exactly the class that would, if genuine, speak to who wrote the plays. This is the strongest form the cipher argument can take.

 

It fails at the same finish line as the rest, and, unusually, its author says so. Ballantine grants that anagrams are unsuited even to the statecraft they descend from “because to a limited extent their syntax is determined by the analyst”: her own illustration is that “Smith shot Becker” could as readily be read “Becker shot Smith.” A genuine cipher clears one of two bars, and a literary one need not be Enigma to do it. If it claims a specific hidden text, that decoding must be forced and reproducible, the Enigma case, where one key turned the whole intercept into fluent German, the next day’s messages fell out under the same key, and every operator recovered the identical words, and no one was left to decide whether the output “counted.” If instead the claim is statistical, that a body of text carries encoding in aggregate, the pattern must exceed what the same procedure produces on genre-matched control text, and survive the ordinary explanations (a larger, more varied vocabulary spreads letters more evenly on its own). Ballantine’s method clears neither bar: by her own account the word-order is the decipherer’s choice, so no single decoding is forced; and an aggregate claim has yet to be tested against a matched control under rules fixed in advance.

 

The degrees of freedom are wider still. Ballantine’s working rules permit, inside the message, U and V to interchange, I to serve for J, Y and I to swap “sometimes,” two V’s to become a W, and a W or M to be turned upside down; and where the letters will not fit, she allows vowels to be added, on the theory that a compositor “ran out of e’s”, calling it “acceptable emendation.” Once letter-identities may be reassigned at need and missing letters supplied, a coherent message can be drawn from almost any stretch of text. Scale does not rescue this: a corpus of a million words does not discipline the method, it enlarges the room in which to manoeuvre. And the offered proofs of validity, that the result reads as coherent narrative, that it “comes alive” read aloud, are precisely the beholder’s judgement a finish line exists to remove.

 

Two further features fit the pattern this page describes rather than breaking it. The biography, Ballantine writes, is “not merely speculative”; its “sequence of events derives from conventional research bolstered by Kit’s own ciphered letters.” So by her own account the historical record offered to confirm the ciphers is partly derived from them, and the loop closes on itself. And the method finds its man almost everywhere, extending Marlowe’s hand past the canon into the apocryphal plays and a corpus of more than a hundred works, the signature of a technique that generates hits for the candidate it is looking for, which is the failure this whole page is about.

 

None of this turns on the author’s character, and the recurring dismissal of Ballantine as unwell is both irrelevant and beneath the question; the method answers on its own terms. What would change the verdict is not a cleverer specimen but a different order of evidence: a statistical test of the corpus against a proper null, the same technique run on length-matched text by other writers, and on shuffled text, with the decipherment rules fixed in advance (no reassigned letters, none added) and the readings reproduced blind by analysts not told whose name to seek. That is the standard this page holds out elsewhere as what rigorous cipher research looks like, and the one that would show whether there is a signal here or only the room to find one.

 

The same conclusion has been reached independently. Literary scholar Ros Barber, reviewing Ballantine’s method directly, found the same failure by a different route: a counter-anagram from the same lines of Love’s Labour’s Lost, produced by Peter Farey in half an hour with anagram software, reads as coherently as Ballantine’s own, a plain demonstration that the letters permit many messages, not one. Barber’s essay is worth reading in full for Ballantine’s own decoded sentences, which speak for themselves.

 

The verdict: the right place for a cipher, judged by the wrong standard of proof. Text-embedded anagrams are the only class that could bear on authorship, but on the method as its own author describes it, the case is not made.

Were the Friedmans frauds?

The charge against the standard this page relies on, examined by that same standard

Every test on this page leans, in the end, on one outside authority: William and Elizebeth Friedman, whose The Shakespearean Ciphers Examined (1957) set the three criteria a valid decipherment must meet. A recurring Baconian rejoinder does not contest those criteria. It attacks the people, the claim, advanced most fully by the anonymous author “A. Phoenix” in The Fraudulent Friedmans (2020), that the book was a deliberate lie, a piece of commissioned disinformation written to protect the “Stratfordian myth.” If that held, the foundation of this page would fall with it, so it gets a direct answer here, by the same rules applied to everything above.

 

The charge at its strongest. Two elements should be granted before anything is answered. First, the biographical premise is true: the Friedmans really did begin their careers, around 1916–1920, as young cipher workers at George Fabyan’s Riverbank Laboratories, the estate where Elizabeth Wells Gallup was employed to extract a hidden Baconian autobiography from the First Folio. They arrived not as sceptics but inside the Baconian project, and took it seriously. Second, there is a single documentary item, the Fermor report of 1977, below, in which a senior intelligence officer is said to record Friedman privately conceding ground. A fair assessment holds both in view rather than pretend the charge is made of nothing. The rest of the apparatus, Henrion, Patton, the anonymous “former member of Russian intelligence,” the eleven “intelligence papers”, is decoration on a single load-bearing inference.

 

The spine is a genetic fallacy. Stripped of rhetoric the argument is one syllogism: the Friedmans began as Baconian cipher workers; they ended by demolishing the Baconian ciphers; therefore the demolition was insincere and someone must have paid for it. The conclusion does not follow. It judges a 1957 argument by its authors’ 1917 origins rather than its content, and treats a change of mind on the evidence as proof of corruption. An investigator who starts as a believer and ends as a debunker is displaying the ordinary arc of someone who actually tested the hypothesis and watched it fail (the experimenter who abandons his own promising model once the measurement refuses to replicate, and who is a more reliable witness against it than someone who never held it). That the Friedmans came up through Riverbank is the grain of truth the charge is built on, but it points the opposite way from the conclusion drawn from it.

 

The sharper version (“they secretly endorsed it, then hid it”) reverses the record. Advocates note, correctly, that the young Friedmans did Baconian cipher work at Riverbank and that pamphlets such as The Greatest Work of Sir Francis Bacon (1916) went out under Colonel Fabyan’s name though his cipher department did the writing. All true, and none of it secret. Fabyan was notorious for printing his employees’ work under his own name; and the Friedmans are famous as the insiders who came up through the Baconian project and then dismantled it. They never concealed that history; it is the founding story of American cryptology, and The Shakespearean Ciphers Examined is built on it: we did this work ourselves, and here is why it fails. Declining to claim personal authorship of a 1916 tract that bore your employer’s name is not “lying to the world about your position.” Nor is it the “opposite position” it is made out to be: the biliteral cipher is a real system Bacon genuinely invented (the Friedmans always said so), and endorsing that is a different claim from endorsing Gallup’s decipherment of the Shakespeare Folio. The early tracts and the late book can both be right at once.

 

A closed citation loop. An extraordinary claim, that a Cambridge University Press book by two celebrated cryptologists was a knowing forgery, needs sources from outside the circle that wants it true. It has none. Baconiana, where Henrion’s critique appeared, is the house journal of the Francis Bacon Society; Fermor chaired that Society; Patton’s Setting the Record Straight is self-published; “A. Phoenix” is an anonymous pseudonym self-posted to Academia.edu (a file host with no peer review) and mirrored on the advocacy site sirbacon.org; and the “incontestable evidence” of disinformation rests on an anonymous “former member of Russian intelligence.” This is advocates quoting advocates in advocacy venues, each pass counted as fresh corroboration. Not one peer-reviewed cryptological or historical source appears.

 

The conspiracy defeats itself on scope. The thesis is not that the Friedmans erred, but that the two most decorated cryptologists in US history, the NSA, MI6, French cipher intelligence, and organised Freemasonry all colluded across decades to suppress Bacon’s authorship. Each institution the theory must enlist is another independent point of failure. Conjoining many low-probability conspirators does not strengthen a case; it multiplies small probabilities into a vanishing one, the way a chain of improbable events is less likely than any single link. The breadth meant to make the theory impressive is exactly what drives its prior toward zero. Attributing a motive (“someone with a vested interest commissioned it”) is not evidence the motive operated; it is the conclusion dressed as a premise.

 

The one item worth weighing: the Winterbotham recantation. The Fermor report of 1977 is the only piece that rewards a second look. In it, Fermor relays that Group Captain F. W. Winterbotham, the MI6 officer who ran distribution of the Ultra decrypts, understood Friedman to have admitted “that he had been wrong to condemn all Baconian ciphers,” and that at a London meeting Commander Pares demonstrated ciphers at the conclusion of Camden’s Remaines to a Cambridge professor’s satisfaction and without contradiction from Friedman. Weighed honestly it does not carry the load. It is third-hand and unfalsifiable, Fermor reporting what “we understand from” Winterbotham about what Friedman is said to have said, printed by the Society’s own chairman after Friedman was dead (1969) and could not answer. Even at face value the concession is narrow: “wrong to condemn all Baconian ciphers” is trivially true, since Bacon did invent the biliteral cipher and the Friedmans’ book says so, a universe away from conceding that Gallup read a hidden autobiography in the Folio. And the cipher Pares demonstrated was in Camden’s Remaines, not the Shakespeare Folio, which is the text at issue.

 

The one checkable claim: Gallup’s biliteral cipher. Behind the character case sits a testable question: did Gallup actually decode a biliteral cipher in the Folio? The biliteral cipher is real: Bacon designed it to hide a two-symbol message by setting text in two subtly different founts (a fount being a complete set of cast metal type in one design and size, what we now call a font), one standing for “a,” the other for “b.” The question is whether Gallup’s readings are decipherments or projections, and they fail on the property this whole page turns on: reproducibility. A valid cipher yields the same plaintext to any competent operator given the rule, the finish line announces itself, as one Enigma setting turned an intercept into fluent German for anyone who applied it. Gallup’s method required sorting the Folio’s type into two clean founts, but Jacobean type was worn, mixed, and set haphazardly from cases of battered sorts; there is no clean partition to recover. And the decisive fact is documentary: no one else at Riverbank could reproduce her assignments. Her own trained assistants, given the same pages, could not get her result; she said the work depended on recognising “minute differences” and on a faculty she called inspiration. A signal only one observer can ever extract, and only by feel, is not a measurement; it is the same apophenia this page has met at every turn, now reading type instead of page numbers. The Friedmans laid out the objective criteria and showed Gallup’s method meets none of them. “A. Phoenix” never rebuts the criteria. The attack is on motive, not method, and that substitution is the answer: when the argument cannot be beaten, the arguer is impeached instead.

 

The “independent reproductions” are not independent. The strongest-sounding rescue is that eminent cryptographers, Captain Powell of US cipher intelligence, and above all General François Cartier, head of the French cipher bureau, reproduced Gallup’s decipherments. But illustrating the method, or recovering the expected message once you already know it, is not what replication means. Cartier came to the cipher through Fabyan’s own correspondence and was steered to the very page to examine; he was a convert, not a control, and the Friedmans did not ignore him, they gave him a chapter. What would count is a neutral operator, handed only the rule and not the answer, sorting the founts and arriving at the same text. That never happens, because the Folio’s worn, mixed type cannot be split into two clean founts to begin with: sorting ambiguous letters toward a message you already expect is how you “reproduce” it. (That Bacon’s own books, such as the Novum Organum, may be set in two founts is beside the point: Bacon demonstrating his own cipher in a book he supervised says nothing about Jaggard’s 1623 Shakespeare Folio, printed in a different shop with no Bacon near the press.) Adding the Simple cipher, the Kay cipher, and anagrams only multiplies the failure: each finds its target everywhere, as this page’s own tests show, so six non-reproducible methods all pointing at Bacon are not convergence but the same noise counted six times.

 

Further reading. The two points on which this section turns (that Jacobean type cannot be sorted into two clean founts, and that a method which finds its target everywhere finds it nowhere) are the settled view of historians of cryptology, not a position peculiar to this site. David Kahn’s The Codebreakers (1967), the standard history of the field, treats Gallup’s biliteral readings in its chapter “The Pathology of Cryptology” and makes the same typographic objection, that early printers used varied type and no compositor could have held tiny letter-variations constant while setting it. Terry Ross, “The Code that Failed,” demonstrates the control-text test directly on a different Baconian cipher, recovering “Bacon” hundreds of times from Moby Dick, the Gospels, and Caesar’s Gallic Wars.

What rigorous cipher research looks like: Colombo (HistoCrypt 2026)

Every entry on this page has examined what cipher arguments look like when they fail the standards a genuine cipher must meet. A 2026 peer-reviewed paper by Lyle Jennings Colombo (Tulane University), published in the proceedings of the International Conference on Historical Cryptology (HistoCrypt), is worth examining for a different reason: it is an example of what the research looks like when it knows its own constraints. The paper makes no authorship claim. It makes a historical and methodological argument, and it is unusually honest about what its findings do and do not establish.

 

The historical contribution. Colombo’s most secure finding is historical rather than cryptographic. George Fabyan’s Riverbank Laboratories, founded in 1913 partly to investigate Baconian cipher claims, was where William and Elizebeth Friedman received their cryptographic training. Their systematic demolition of those claims helped establish the methodological standards that eventually shaped the U.S. Army’s Signal Intelligence Service and, through it, the NSA. David Kahn’s The Codebreakers (1967) documents this lineage. The SAQ’s cipher debates, in other words, are not merely an antiquarian curiosity: they functioned as an institutional incubator for modern cryptography. The undisciplined Baconian methods that this page describes provoked a corrective rigorous enough to shape an intelligence agency. That is a genuine contribution to the history of the discipline.

 

A methodology that fits its object. Colombo argues that the Friedmans’ corrective, while necessary, defined legitimate cryptography too narrowly. Algorithmic, rule-based, communicative systems, Enigma, Walsingham’s coded dispatches, were designed for transmission between parties who shared a key. Renaissance Hermetic concealment systems were designed for something different: preservation, and the regulation of access. They were not intended to be decoded by a recipient; they were intended to enshrine knowledge for the initiated while withholding it from everyone else. Applying the Enigma finish-line standard to a Hermetic emblem is, on this account, a category error, measuring one kind of object with instruments designed for another. This is a serious methodological point. Elizabethan secrecy practices were real, varied, and not well served by a single modern definition.

 

Pre-specified constraints. The central methodological advance the paper demonstrates is a rule for determining grid dimensions that is fixed before interpretation begins. In the Sonnets dedication, the letter T occurs nineteen times; nineteen is also T’s positional value in the Latin alphabet (A=1, B=2…T=19). The paper states that no other letter in the dedication satisfies this condition. The 19-column grid is therefore determined by a rule that exists independently of any desired result, not by selecting whichever of hundreds of grids produces the most interesting output. The same procedure applied to Jonson’s prefatory poem “To the Reader” yields a 13-column grid (N occurs thirteen times; N is the thirteenth letter). The same method in two distinct texts, yielding a different grid dimension in each, is the kind of replication that distinguishes systematic analysis from pattern-seeking.

 

Statistical honesty. Colombo and Chambers (under review at the time of publication) report Kruskal–Wallis tests showing the proposed grids are statistically distinguishable from adjacent grids: p = 5.9×10⁻⁵ for the Sonnets dedication, p = 0.002 for “To the Reader.” These are not trivial numbers. More importantly, the paper does not overstate what they mean. It states explicitly: “Statistical analysis alone does not constitute proof of encryption; however, taken together, these results indicate that the structural features identified here are distinguishable from chance at a level that warrants literary-historical interpretation.” That is honest reporting. The claim is proportionate to the evidence. Distinguish this from the “100 million to one” figure in Entry 5, which was presented as near-conclusive while concealing the undisclosed search space that generated it.

 

Explicit separation of methodology from attribution. The paper is direct about what it is and is not arguing. Its central question is whether the paratexts exhibit “structured, non-random organisation consistent with historically attested practices of symbolic or concealment-based secrecy”, a question it treats as independent of authorship attribution. It notes that much subsequent work in this area has been conducted in an Oxfordian advocacy context, and it explicitly sets that context aside. The paper is not making a case for de Vere. It is making a case for a research methodology, and it is careful to keep those two projects separate. That separation is exactly what the entries above this one fail to maintain.

 

What remains open. The paper acknowledges what it has not established. The full statistical analysis is in a companion paper still under review. The gematric rule, while constraining, was itself selected from a field of possible gematric systems, and the paper does not fully establish that the rule preceded the discovery. The specific relationship chosen, a letter whose frequency in the text equals its positional value in the Latin alphabet, is one of many possible gematric relationships one could test. The paper argues this kind of letter-number correspondence has grounding in Hermetic practice, which may be true historically, but that does not show that this particular rule was committed to before the search revealed that T satisfies it in the dedication and N satisfies it in “To the Reader.” The residual question is the same one that defeats entries 1–7: how many gematric relationships were tried before this one produced a clean result? If that number is not given, the rule-selection stage carries the same vulnerability as the grid-selection stage it was designed to solve. And even if the paratexts contain genuine structured concealment, the Sonnets dedication was signed by Thomas Thorpe, the publisher. A concealment structure in that document records what Thorpe chose to embed, not what the author of the plays wrote. The paper does not try to close that gap. It recognises that closing it would require a different kind of argument entirely.

Method and data for the digit-cipher counts

Every count in Entries 1–4 comes from one rule, applied the same way throughout: pool the digits of the page numbers in a group—a single page, an adjacent pair in a grid, or a consecutive run—and count the group as a hit if any non-empty subset of those pooled digits sums to the target. Nothing is discarded and no digit is added. The figures below are regenerated by a short script, verify-digit-cipher.py, from the data named here; anyone can rerun it or change the inputs.

 

Data.

• Shakespeare table of contents: the 35 catalogued plays of the 1623 First Folio (Troilus and Cressida has no catalogue entry), with their section page numbers, laid out in a four-column grid in catalogue order, giving 57 horizontally and vertically adjacent pairs.
• Shakespeare full sections: every printed page number of the three sections—Comedies (1–304), Histories (1–232), Tragedies (1–293)—verified against the Internet Shakespeare Editions facsimile of the National Library of New South Wales copy; 826 consecutive within-section pairs.
• Jonson table of contents: the 13 section start-pages of the 1616 Workes (Stansby), from its collation formula, in a four-column grid.
• Jonson full book: the sequential page numbers 1–1,015.

 

Counts the script returns.

• Shakespeare ToC grid, 57 pairs: 20 sum to 17, 42 to 6.
• Shakespeare full sections, 826 pairs: 421 sum to 17, 561 to 6.
• Jonson full book: 111 single pages, 674 of 1,014 pairs, and 900 of 1,013 three-page runs sum to 17; 683 of 1,014 pairs sum to 6.
• Jonson ToC: 2 single entries and 15 of 18 adjacent pairs sum to 17.

 

The Shakespeare table-of-contents grid is the proponent’s own construction; the full-section and full-book counts extend the same rule to every page number, and the wrong-candidate and wrong-folio columns in Entry 4 come out as high as the intended one. The method does not discriminate.

Sources and notes

• Francis Meres, Palladis Tamia: Wits Treasury (London: Cuthbert Burbie, 1598). Primary source for Entry 7: lists of comedy playwrights, tragedy playwrights, lyric poets, and enriched-English poets; iambic poets paragraph (explicit statement of symmetry method). Digitized at Early English Books Online (EEBO).
• Charlton Hinman, The Printing and Proof-reading of the First Folio of Shakespeare, 2 vols. (Oxford: Clarendon Press, 1963)
• William F. Friedman and Elizebeth S. Friedman, The Shakespearean Ciphers Examined (Cambridge University Press, 1957), source of the three criteria used throughout this page; winner of the Folger Shakespeare Library literary prize.
• Elizabeth Wells Gallup, The Bi-literal Cypher of Sir Francis Bacon (Riverbank Laboratories, 1900), the decipherments the Friedmans later refuted (see “Were the Friedmans frauds?”).
• A. Phoenix, The Fraudulent Friedmans: The Bacon Ciphers in the Shakespeare Works (2020), Academia.edu / sirbacon.org, the fullest statement of the fraud charge.
• Pierre Henrion, “Scientific Cryptology Examined,” Baconiana Vol. XLIII, No. 160 (March 1960), pp. 43–63.
• Noel Fermor, report on the Pares–Friedman–Winterbotham meeting, Baconiana Vol. LX, No. 177 (November 1977), p. 76.
• David Kahn, The Codebreakers: The Story of Secret Writing (Macmillan, 1967), esp. “The Pathology of Cryptology”, the standard history of the field on Gallup’s biliteral readings and the two-type problem.
• Terry Ross, “The Code that Failed: Testing a Bacon-Shakespeare Cipher,” shakespeareauthorship.com, the control-text method applied to a Baconian cipher [shakespeareauthorship.com/bacpenl.html].
• David Gants, “The 1616 Folio (F1): Textual Essay,” in The Cambridge Works of Ben Jonson (Cambridge University Press Online, 2012) [universitypublishingonline.org/cambridge/benjonson]
• Witztum, Rips & Rosenberg, “Equidistant Letter Sequences in the Book of Genesis,” Statistical Science 9:3 (1994)
• McKay, Bar-Natan, Bar-Hillel & Kalai, “Solving the Bible Code Puzzle,” Statistical Science 14:2 (1999)
• “The 17 Solution,” Don’t Quill the Messenger podcast (Steven Sabel)
• The Workes of Benjamin Jonson (1616), digitized at Internet Archive [archive.org/details/workesofbenjamin01jons]; collation: [12] 1–1015 [1016] pp., 2°, signatures ¶6 A–4P6 4Q4
• William and Elizebeth Friedman, The Shakespearean Ciphers Examined (Cambridge University Press, 1957)
• John M. Rollett, “Secrets of the Dedication to Shakespeare’s Sonnets,” The Oxfordian, Vol. 2 (1999), pp. 60–75 [shakespeareoxfordfellowship.org/secrets-dedication-shakespeares-sonnets]
• Alexander Waugh, “Where Shakespeare is Really Buried,” video lecture series (YouTube, 2017–2018); “The Sonnets Code Deciphered,” De Vere Society publications and conference lectures
• L. James Hammond, “Shakespeare Codes,” Phlit newsletter (January 2019) [ljhammond.com/phlit/2019-01b.htm], summary of Waugh’s 1740 argument, Fourth T motif, and Sonnets dedication grid
• Percival Golding, manuscript note c.1620: de Vere “lieth buried at Westminster”, cited in Waugh’s lectures
• Parish register, St Augustine, Hackney: burial entry, July 1604
• Will of Elizabeth Trentham, Countess of Oxford (1612)
• Modern editorial grouping of Sonnets 1–17: see Helen Vendler, The Art of Shakespeare’s Sonnets (Harvard UP, 1997), pp. 1–2
• Lyle Jennings Colombo, “What Counts as a Cipher? The Evolving Role of Shakespearean Paratexts in Cryptographic History,” in Proceedings of HistoCrypt 2026, Tulane University. Historical argument on the Riverbank–Friedman–NSA lineage; methodological framework for non-algorithmic concealment systems; gematric pre-constraint for grid analysis of the Sonnets dedication and “To the Reader.”
• Lyle Colombo and Paul Chambers, “Hermetic Secrecy and Figurative Acrostics in Shakespearean Paratexts” (under review at time of publication). Full statistical analysis underlying the Kruskal–Wallis results cited in the HistoCrypt paper.