The Two Frameworks
Each framework asks a distinct question. No single framework is decisive; agreement across both carries the most analytical weight.
The criteria were fixed and published before any candidate was scored. If a particular result looks wrong, the place to start is the evidence notes for that specific row, not the framework. Adjusting criteria after seeing results is the thing this site is designed to avoid. See A Proposal for Advocates.
How This Site Establishes Criteria
Before examining any candidate, this site agrees on what counts as evidence. Two independent frameworks, each drawn from existing scholarship, are applied to every candidate equally. The criteria are fixed before the results are examined. That is what separates assessment from advocacy.
Each framework asks a different question, and together they provide two independent tests. A candidate who does well on both has a stronger claim than one who scores on only one. A candidate who fails both, as Shakspere does, presents a problem that no single framework alone could establish.
This site applies both to fourteen candidates, including Shakspere himself.
| Framework | What it asks |
|---|---|
| Literary Paper Trail | Does primary source evidence of a literary career exist for this person? (10 categories) |
| Convergence Analysis | Do 18 methodologically diverse evidence lines point to this candidate? |
Three types of evidence in the authorship debate
Arguments for any candidate tend to fall into three types. The type matters as much as the content.
Type 1, Documentary. A contemporaneous primary source recording a fact directly: a signed letter, a legal record, a manuscript in the candidate's hand. The literary paper trail asks whether evidence of this type exists for a literary career. It is the only type that can close the question.
Type 2, Circumstantial. Documented facts converging on a candidacy without recording authorship directly: biographical parallels, topographical knowledge, verbal overlaps, career connections. Individually suggestive, cumulatively substantial, but available to motivated researchers across multiple candidates. Cannot substitute for Type 1.
Type 3, Interpretive. Readings of literary texts that require prior acceptance of the attribution to work as evidence: decoded sonnets, cryptographic arguments, allegorical identifications. Not worthless as literary analysis, but carries no evidential weight in authorship determination until the attribution is established independently.
Note: a literary text can be a Type 1 document and still generate only Type 3 evidence: the sonnets exist as primary sources, but a reading of them that requires accepting the attribution first is Type 3. The document and the interpretation are different things.
The test: could this evidence have come out differently? A Type 1 document could have named someone else. A Type 3 interpretation, if it didn't fit, would simply be set aside.
A note on texts as evidence of their own authorship. A document can be primary evidence of what it describes. It cannot be primary evidence of who wrote it.
This distinction matters because a recurring move in authorship arguments is to treat first-person statements in the works as self-testimony, as if the speaker of a sonnet or the narrator of a poem were a witness identifying himself. On this reading, a sonnet describing the speaker’s relationship with a patron, or a rivalry with another poet, becomes direct evidence of the author’s biography.
The problem is structural, not interpretive. A text is primary evidence of its stated content: a sonnet describing a rival who writes hymns is primary evidence that its speaker heard such a rival. It is not primary evidence of who the speaker was, because the identity of the speaker is exactly what is under dispute. Using the text to answer that question is circular; the conclusion is doing the work of the premise.
This does not mean textual readings are worthless. A statement describing a verifiable external situation (a rival who writes hymns, a patron who supports that rival’s published work) can be tested against the external record. If it holds, it confirms the reading. What it cannot do on its own is confirm the authorship. That step always requires a document from outside the works connecting a named person to their composition, the standard Price’s framework was designed to enforce.
Any argument that treats the works as self-testimony about their own author has not produced primary evidence. It has produced an interpretation that primary evidence would need to confirm.
Diana Price's Literary Paper Trail
In Shakespeare's Unorthodox Biography (2001, revised 2012), Diana Price surveyed the documentary records left by twenty-five Elizabethan and Jacobean literary figures. She logged every original document she encountered for the various poets and playwrights; adopted the criteria found in the biographies, histories, and journal articles consulted; and let the categories emerge by collapsing the qualifying evidence into ten manageable categories, so the scheme was built up from the surviving records themselves. The categories were not established in advance. Evidence must meet three gatekeeping tests before it qualifies: it must be contemporaneous (from the person's lifetime), personal (connecting directly to the individual), and literary (connecting to writing activity, not commerce or property). Only primary source documents count; secondary claims, later tradition, and inference do not qualify for a ✓.
A specific note on title page attribution. Title pages carrying the name 'William Shakespeare' are contemporaneous and literary, but they fail the 'personal' gatekeeping test; they connect to the name on the work, not independently to a specific biographical individual. Elizabethan and Jacobean publishers also routinely attached recognisable names to books without the author's authority or involvement. Price documents two instances in Shakespeare's Unorthodox Biography: The Puritan (1607, p. 96) and The Yorkshire Tragedy (1608, p. 135) were both published under 'W.S.' or 'William Shakespeare' and are not accepted as works Shakespeare wrote. The same attribution mechanism that produced those titles produced the canonical ones, which is why title page evidence is treated as circumstantial rather than probative, and carries no ✓ on the paper trail table.
The full table (all ten categories applied to all nine candidates, with primary source citations and scoring notes) is on its own dedicated page.
Objections to the framework, and Price's replies
Price's framework has attracted four principal objections from mainstream scholars. Each is worth stating precisely.
1. The social class objection. Critics argue that Price's categories favour writers who generated correspondence with aristocrats or institutions, types of record that a middling-sort commercial playwright would not typically produce. The absence, they argue, reflects social position and documentary survival rates, not non-authorship. This is the most persistent Stratfordian objection, stated at length by Tom Reedy and David Kathman, and briefly by James Shapiro in Contested Will (2010, pp. 243–4).
Price's reply: she deliberately included commoner playwrights (Thomas Kyd (son of a scrivener), John Fletcher, Thomas Dekker) precisely to test this hypothesis. All left at least one qualifying item. Category 8 is a catch-all specifically designed to capture any personal reference to the individual as a writer, without requiring aristocratic connections. More pointedly: Shakspere's other commercial records did survive abundantly, dozens of property, legal, and trade documents. The argument that records of his class did not survive is contradicted by the evidence of his own dossier. His middling-sort paper trail exists; it simply contains nothing literary.
2. The Hand D objection. Three pages of the collaborative manuscript Sir Thomas More, the "Hand D Additions", are widely attributed to Shakespeare's handwriting; the British Library called them "Shakespeare's only surviving literary manuscript." If the attribution holds, it is the one literary paper trail Shakspere would possess.
Price's reply: identifying a questioned document requires a control sample in "the same general class" as the disputed writing, dramatic manuscript to dramatic manuscript, but the only handwriting known to be Shakspere's is six irregular signatures on legal documents, and not all of those are certainly his. The Folger Shakespeare Library concedes the point: "we don't really know what Shakespeare's handwriting looks like ... too small a sample size to make any sort of reliable comparison." Arguments from style and spelling cannot supply the missing control sample; as Paul Werstine notes, Hand D is assigned to Shakespeare "only through an argument from 'cumulative evidence' – all of which evidence has been dismissed as inconclusive by Shakespeareans themselves." Even a stylometric match to the author of Hamlet would not show the pages were penned by Shakspere rather than by, say, a playhouse scribe. The British Library has since quietly downgraded its catalogue description from Shakespeare's "only surviving literary manuscript" to "possibly" and "probably" his. Full analysis of the Hand D attribution, with Diana Price's video presentation (July 2026).
3. The posthumous evidence objection. Price requires contemporaneous evidence, from the person's lifetime. Critics argue this is artificially strict: for most historical figures, posthumous testimony is routinely admitted as evidence. The First Folio (1623) and the Stratford monument are legitimate testimony.
Price's reply: she does not discard posthumous evidence; she treats it as requiring independent corroboration. The anomaly is not that posthumous evidence exists; it is that Shakspere is the only alleged professional writer from the period for whom you must rely entirely on posthumous evidence. Every other writer in her survey of twenty-four, including the worst-documented, left at least one contemporaneous personal record touching on literary activity. That uniqueness is the finding.
4. The circular reasoning objection. Some critics argue that Price selected her ten categories to produce the result she wanted, that the framework is designed to make Shakspere fail.
Price's reply: she did not choose the categories in advance. The three gatekeeping tests (contemporaneous, personal, literary) were drawn from the biographers and historians consulted, as well as historians and critics including Richard Altick, Paul Murray Kendall, and Harold Love, none of whom had any involvement in the authorship debate. The ten categories themselves emerged only afterwards, by collapsing the documentary evidence she had logged for twenty-four writers, e.g., Books Bought, Books Given, annotations and the like folding into a single “Books” category. The one category she created deliberately, Category 8, is a catch-all whose purpose is to ensure that the scheme is comprehensive, to give any surviving qualifying document a home on the chart, which is the opposite of a filter built to exclude. She created the framework by examining the documentary source material left behind by twenty-four writers and then comparing the evidence of “literary paper trails” to Shakspere's documentation.
One further point belongs here. The mainstream has not published a systematic, category-by-category rebuttal of Price's chart. Shapiro's dismissal in Contested Will occupies two pages and does not engage with the framework at the level of individual categories. That absence is part of the methodological landscape, not a concession of the argument.
Why This Site Does Not Use an Author Profile Framework
The most widely used author profile in the Shakespeare authorship debate is J.T. Looney's eighteen characteristics, derived from close reading of the works in 1920. This site does not adopt it as a scoring framework. The reasons are examined in full on a dedicated methodology page.
In brief: several of Looney's criteria embed social and biographical assumptions not derived from the works, particularly characteristics 10–12, which pre-select for aristocratic candidates before any evidence is examined. More fundamentally, the characteristics that survive scrutiny are already captured, more rigorously, in the convergence analysis below (Italian knowledge, legal knowledge, musical knowledge, Greek dramatic knowledge, lyric and narrative poetry, court familiarity, genre range). A profile that largely duplicates a more rigorous instrument while adding unreliable criteria weakens rather than strengthens the methodology. The convergence framework is the site's instrument for asking what the works require of their author.
Why author profiles don't settle the questionThe Convergence Analysis
Distinct lines of evidence (Italian topographical knowledge, court access, Southampton connection, legal knowledge, the Sonnets, musical knowledge, Greek dramatic knowledge, and others) are each assessed independently for all ten candidates. Where multiple lines converge on a single candidate, the combined weight increases; but convergence of circumstantial evidence is not the same as primary source documentation. The table also explains why Greene's Groatsworth of Wit (1592) is not used as a convergence line.
More broadly, stylometric methods lack a coherent probabilistic framework for assessing probative value; they produce a number without the likelihood ratios needed to say how much that number should shift prior belief.
Two structural problems compound this. First, the circular corpus: stylometric attribution works by measuring an unknown text against a known author's authenticated corpus, but in the Shakespeare case, the canon itself is what is under dispute. The method cannot bootstrap its own premise. Second, the thin-corpus problem: stylometry can suggest a match to a known corpus; it cannot exclude a candidate whose surviving authenticated writing is too sparse to support a reliable model. For most alternative candidates, the surviving corpus of undisputed writing is a handful of letters and occasional verse, far below the threshold needed for meaningful comparison. A negative stylometric result in these conditions is uninformative.
Stylometric claims will be encountered in the authorship debate; this site notes them where relevant but does not score them.
Objections to the framework, and this site's replies
Three principal objections to the convergence framework arise in the literature. Each is worth stating precisely.
1. The strand selection objection. Critics argue that the strands were chosen to favour particular candidates, that the selection is not neutral but steers toward certain outcomes before any scoring begins.
This site's reply: the strands are drawn from the existing scholarly literature on what the canon demonstrably shows (documented Italian topography, verifiable legal terminology, traceable classical sources, and so on). They were listed on the Two Frameworks page before any candidate was assessed. A reader who suspects a strand was added after the scores were tallied can check this against the stated criteria, which do not change between updates. More pointedly: several strands produce results that are inconvenient for the leading candidates. The 1604 chronology strand creates a direct problem for de Vere. The Stratford connection strand creates a problem for every alternative candidate. A selection designed to steer would have excluded both.
2. The "stacking" objection. Multiple weak strands converging on a candidate do not add up to a strong result; you are accumulating maybes, not evidence. A candidate who scores ◑ on twelve strands has twelve soft matches, not one hard finding.
This site's reply: the objection would be decisive if all strands were of equal evidential weight and the scores were treated as proof. Neither is the case here. The framework intro states explicitly that convergence of circumstantial evidence is not the same as primary source documentation; it is a different kind of finding, not a substitute for one. What convergence across methodologically diverse strands does is narrow the field: a candidate who fails most strands is a weaker fit than one who passes most, even if no single strand is conclusive. That is triangulation in the ordinary scientific sense, the same logic used in geolocation, medical diagnosis, and archaeological dating, where no single measurement settles the question but the pattern across instruments does. The point about diversity of method is not incidental: more measurements of the same kind do not approach a fix. A navigator who takes twenty bearings on the same headland still has only one line of position. The framework does not claim to prove authorship. It claims to report where multiple different types of evidence point, and to do so consistently for all candidates.
3. The approximate scores objection. Convergence totals are marked approximate throughout. Critics ask: approximate by how much? If the margin of uncertainty swallows the differences between candidates, the rankings are meaningless.
This site's reply: "approximate" marks two specific sources of uncertainty that cannot be reduced to a single number. First, several strand assessments involve judgment calls where a score of ✓ or ◑ is defensible but not compelled by the evidence; a different careful reader might score the same strand differently for the same candidate. Second, the strands are treated as equally weighted for the reasons stated above, but a reader who assigns higher weight to, say, Italian topographical knowledge than to musical sensibility would get a different ranking. The approximate label is honest about those limits rather than concealing them behind a false precision. It does not mean the differences are negligible: de Vere's convergence lead over Shakspere is large enough that no plausible rescoring of individual strands closes it. The label means the exact number should not be treated as a measurement; it should be treated as an indicator.
4. The peer-review objection. Price's literary paper trail is a peer-reviewed monograph. The convergence framework is this site's own construction and has not been independently reviewed before use.
This site's reply: the asymmetry is real and should be acknowledged rather than explained away. Two structural points reduce but do not eliminate the concern. First, the individual strands are not invented here: Italian topographical knowledge, legal expertise, the Southampton connection, and the others are lines of evidence drawn from existing scholarship. The site's contribution is consistent application across all candidates, not the identification of the strands. More precisely: each strand was identified by scholars working on questions other than authorship. A strand whose primary proponents have an authorship stake does not qualify; the risk of motivated selection cannot be separated from the argument. Second, the pre-commitment problem that peer review is designed to address, criteria selected after the results are known, is addressed by stating all the strands on the Two Frameworks page before any scores appear. That is the structural equivalent of pre-registration. What peer review adds beyond pre-registration, specialist scrutiny of the strand definitions and scoring judgments, is not present here. That is why convergence scores are marked approximate, why a convergence lead alone does not close the question on this site.
The full table (all the lines applied to all nine candidates side by side, with scoring rationale and methodology notes) is on its own dedicated page.
See the full Convergence tableThe Evidence Standard
The same four-symbol system is used in all three tables. The symbols represent evidentiary status, not quality judgements about candidates. Hovering over any symbol in any table shows a tooltip explaining exactly what evidence, or absence of evidence, that symbol represents for that specific candidate and category.
A primary source document directly supports this criterion for this candidate.
Suggestive but incomplete; interpretation may be contested; or evidence that cuts against this criterion where the contrary reading is itself disputed.
No primary source evidence located, in either direction.
The Evidential Landscape: what exists, what doesn't, and what the frameworks are designed to navigate
The only evidence that would definitively settle the authorship question is a primary source document directly linking a specific person to the writing of the works (a manuscript in their hand, a payment record naming them as author, a contemporary letter identifying them explicitly). Nothing of that kind has surfaced. What has surfaced are documents connecting the name "Shakespeare" to the title pages, documents connecting Shakspere of Stratford to the theatre business, and documents showing other candidates were active writers. None closes the gap between a named individual and the authorship of specific texts. This gap is what the two frameworks on this site attempt to navigate.
A second category of evidence, neither primary source documentation nor correspondence argument, exists and should be distinguished from both: quantitative and forensic methods. Statistical clustering analysis (Martin, 2023, applied to name and role patterns in the works for Bassano) and forensic document examination (Stritmatter, 2023, on the Audley End annotations) occupy this ground: testable, falsifiable in principle, stronger in kind than coincidence arguments. The limitation is asymmetric. These methods are significantly better at ruling candidates out than at establishing authorship positively. They narrow the field; they do not fill the primary source gap. Computer-assisted stylometric attribution was considered and excluded as a convergence line for methodological reasons explained below.
In the absence of primary source documentation, and alongside rather than above the quantitative methods, the authorship debate has largely been conducted through correspondences and coincidences: resemblances between a candidate's life and the works, incidents in the candidate's biography mirroring scenes in the plays, thematic resonances between documented preoccupations and the texts. This is the most abundant category of argument and the least probative. Its structural weaknesses are demonstrated in full on the Of Course It Was Him/Her page. The two frameworks on this site are designed as a check on that tendency: the Price literary paper trail asks what primary source documents actually exist; the convergence analysis asks whether multiple methodologically diverse lines of evidence (documentary, textual, and historical) point in the same direction. Neither of them replaces primary source documentation. Together, they provide the most systematic basis currently available for evaluating the question without pretending to resolve it.
A fourth category of assumption runs through every framework on this site without always being made explicit: the conventional dates of the plays. Dates like Macbeth c.1606 or The Tempest c.1611 are typically presented as settled facts, but they rest primarily on topical allusion and inference, not on primary source documentation such as Stationers' Register entries, payment records, or confirmed performance dates in manuscript. Kevin Gilvary's Dating Shakespeare's Plays (2010) examines each play individually and argues that no play can be definitively dated after 1604 on primary source evidence alone. That argument has not been refuted on primary source grounds.
The honest approach to Kevin Gilvary is neither wholesale endorsement nor dismissal. His methodological point is legitimate: conventional dates vary considerably in their evidential support, and the distinction between inference and documentation is rarely made explicit in standard references. The Tempest's 1611 date depends partly on whether the Strachey letter was a source, which is itself contested. Macbeth's 1606 date rests on convergent topical inference rather than a document.
The honest limit of Kevin Gilvary's argument is equally important: absence of primary source evidence for a late date is not evidence of an early date. Uncertainty about dating cuts both ways. It does not establish that the plays were written before 1604; it establishes that the conventional chronology is more inferential than is typically acknowledged. This site treats the dating question as genuinely open, neither endorsing the orthodox chronology nor assuming the plays must pre-date 1604. Where a candidate's chronological compatibility is scored in the convergence framework, the note of uncertainty applies to both sides of the ledger.
Why Primary Source Evidence?
A common objection to this site's methodology is that primary source documentation is too strict a standard given how sparse Elizabethan records are. The objection turns out to answer itself.
The same standard is applied to all nine candidates here, not to any single one. When a candidate scores ~ on a category, it means no primary source evidence has been found, not that the candidate is disqualified. When a candidate scores ✓, it means a specific document exists and can be cited. The framework does not penalise candidates for the general sparseness of Elizabethan records. It records what the record actually shows, consistently, for everyone.
If the records of the period were generally too sparse to produce primary source literary documentation, we would expect to find writers with no surviving literary trace. Diana Price examined twenty-four Elizabethan writers of the period and found that every single one left at least one primary source document connecting them to their literary work. Not most, all of them. She then applied the same framework to William Shakspere of Stratford. He did not pass a single category. The standard is not calibrated to the twenty-first century; it is calibrated to what the period actually produced. This is why the same standard applied consistently across all nine candidates here is meaningful rather than arbitrary.
The bar is also deliberately low. Price's ten categories include things as minimal as a single letter mentioning a literary project, a payment record for any piece of writing, or a contemporary note identifying someone as a writer. One category passed is enough. The requirement is not a complete archive; it is a single trace.
If the primary source standard is abandoned because records are scarce, there is no principled basis for preferring any evidence over any other. Aesthetic impressions, biographical correspondences, and cumulative coincidences all become equally valid, and that is exactly the condition that has produced 150 years of unresolvable advocacy on all sides. The standard is not a burden imposed on the evidence. It is what keeps the question honest for every candidate equally.
A sharper version of the objection asks whether primary source evidence is itself reliable. Documents can be forged or misread; Ireland and Collier are the canonical examples in this field. The methodology's answer is structural: it does not rest on any single document. The paper trail framework requires a pattern across ten independent categories applied to twenty-five writers. If one category is in doubt, nine remain. If one writer's record is unreliable, twenty-three others provide the comparative baseline. A single wrong document changes one cell; it does not overturn the pattern. The framework was designed with exactly this redundancy, which is why Price applied it to twenty-four contemporaries rather than asking only about Shakspere.
The Argument from Coincidence
The structural weaknesses of coincidence arguments (the baseline rate problem, selection bias, the "biography in the works" assumption, and the comparative test) are demonstrated in full on the Of Course It Was Him/Her page. What follows is a reference taxonomy of correspondence argument types, for use when evaluating specific claims.
Types of correspondence argument
Not all coincidence arguments have the same structure. Recognising the type helps to assess its weight.
At the weakest end: diffuse biographical correspondence. The general claim that a candidate's life experience matches the emotional or thematic range of the works. Every camp produces this type most prolifically, because the works are large, the themes are universal, and almost any rich documented life will resonate at some level. The problem is that no denominator can be specified: how many themes would we expect not to resonate? Without a baseline, the count of resonances is meaningless. This is the type that fills most advocacy literature.
Consider two well-documented Elizabethan figures who are not authorship candidates: Robert Sidney and Fulke Greville. Both travelled on the Continent, both moved in aristocratic literary circles, both had melancholy temperaments documented in their correspondence, both knew the law. A correspondence dossier for either could be assembled without difficulty.
But here is what the dossier would have to suppress. Sidney spent most of 1589–1616 as Governor of Flushing; the canon has no Dutch settings; his surviving poetry is stylistically unremarkable; his letters show no interest in theatre and no connection to any acting company; he knew little Italian. Greville's two surviving plays were closet dramas, never staged; his documented literary style is dense and Senecan, the opposite of Shakespeare's dramatic register; he wrote almost entirely in tragic and philosophical modes, leaving the comedies entirely unaccounted for; no Italian journey is documented; no connection to Southampton exists. The dossier for a serious candidate looks like the dossier for these two men, minus the suppressed column.
Neither man is a plausible candidate, but that's the point. The same biographical raw material that generates a correspondence dossier for a favoured candidate also generates one for people nobody seriously proposes. The dossier is an artefact of the method, not the person.
Stronger in structure, if not always in force: incident correspondences. A specific named event from a candidate's documented biography is said to mirror a specific scene or episode in the works. These are testable in principle; we can examine whether the correspondence survives scrutiny or dissolves, but the Texas sharpshooter problem applies as before. The correspondences below are drawn from the advocacy literature for different candidates; each illustrates how the argument works and where it breaks.
- Gadshill (de Vere). In 1573, a group of men including de Vere's servants committed a highway robbery on travellers at Gadshill in Kent. Henry IV Part 1 contains a robbery scene set at Gadshill, on the same road. Oxfordian writers cite the specificity of the location. The counterargument: Gadshill was a notorious robbery site throughout the period and appears in earlier literature for exactly that reason. The author needed no personal memory of the place; it was already a stock location.
- Lopez and Shylock. Roderigo Lopez, physician to Queen Elizabeth and of Portuguese-Jewish origin, was executed in 1594 on a charge of plotting against her life. The Merchant of Venice features a Jewish antagonist bound up in questions of contract and punishment. The connection is frequently made for Marlowe (who wrote The Jew of Malta, revived around this time) and for de Vere (who had connections to the Essex faction that prosecuted Lopez). Both versions are speculative at the coincidence level; neither produces a documentable link between the trial and the text.
- Elsinore. English players, including Will Kempe, later a colleague of the Stratford actor, were documented at the Danish court at Elsinore in 1585–86. Hamlet is set there. The correspondence is noted across several candidate cases. The counterargument: Elsinore appears in the acknowledged source (Historiae Danicae by Saxo Grammaticus, via Belleforest), so the author needed direct knowledge of this site only if specific topographical detail goes beyond what the source supplies. That is a more demanding and more specific argument, and a more useful one.
- The Gray's Inn Revels (Bacon). On 28 December 1594, The Comedy of Errors was performed at Gray's Inn as part of the Christmas Revels, a night so chaotic it was recorded in the Gray's Inn records as the "Night of Errors." Francis Bacon was a member of Gray's Inn, active in its revels, and later its Treasurer. Baconians cite his presence at the institution where the play was performed as a connection. The counterargument: Gray's Inn had hundreds of members, and the performance establishes proximity, to the same building, on the same evening, not authorship. Proximity without a documented link between the person and the text is coincidence at the institutional level, not evidence.
The strongest class: knowledge arguments. Evidence that specific, verifiable information appears in the works that goes beyond what any available printed source could supply, and that the candidate can be shown to have had access to that information through a documentable route. Richard Paul Roe's Italian topographical research is the clearest example: field observations of ferry routes and local geography in the plays that do not appear in any known novella source. One specific claim within Roe's work, the identification of sycamore trees near Verona's western gate as the grove in Romeo and Juliet (I.i), has been shown to be incorrect: the trees are plane trees, are modern civic plantings 40–50 years old, and a contemporaneous engraving of Verona shows no trees at that location (Hatinguais, The Oxfordian, Vol. 18, 2016). That argument should not be used. The remainder of Roe's topographical findings stands. George Coffin Taylor's documentation of Florio's specific English phrasings in the plays is another. These arguments identify knowledge with traceable provenance, not resonance between a life and a theme, and they are what earn ✓ scores on the convergence analysis.
Above all three: source-text fingerprints. Distinct from all the categories above and should not be grouped with them. A fingerprint argument identifies a specific idiosyncratic feature, a translation error, a unique phrasing, a named passage from a specific page of a specific edition, that appears in the works and can be shown to exist in only one place in the prior literature. The argument is not "this fits my candidate" but "this specific thing exists in this specific place and nowhere else."
The test is demanding. The feature must be (a) idiosyncratic, not a commonplace or a phrase that might have been arrived at independently; (b) traceable to a specific source in the candidate's documented orbit; and (c) absent from alternative routes the author might have taken to the same material. When all three conditions are met, a fingerprint argument constrains the field in the same way a primary source document does; it narrows the space of possible authors to those with access to that specific text.
Three examples currently in the authorship literature meet this standard. The translation error in Thomas North's Dial of Princes that appears in The Winter's Tale and nowhere else in printed English before that translation (McCarthy and Schlueter, Thomas North's 1555 Travel Journal, Fairleigh Dickinson University Press, 2021). The verbal echoes of Florio's specific English phrasings documented by Taylor (1925), not the ideas of Montaigne, but Florio's particular choices of English words. Roger Stritmatter's identification of marked passages in de Vere's Geneva Bible linking the annotations to the Shakespeare canon (doctoral dissertation, 2001; Folger accession cs1472, call number STC 2106 copy 1; digital images). These are not correspondence arguments with better examples. They are a different kind of argument. They should be evaluated accordingly.
Notes on Candidate Assessments
Two methodological notes apply to specific candidates and are recorded here, after the frameworks have been introduced, to avoid associating them with any single framework.
As of this writing, the 2023 study has not been independently replicated by a second forensic document examiner. A single-specialist finding, however rigorous in methodology, carries different evidential weight from one that has survived independent scrutiny. This is part of why this site uses F.A. rather than ✓: the designation signals that the finding rests on specialist interpretation not yet confirmed by a second specialist, in addition to the general principle that expert attribution is differentiated from directly readable primary source documentation. If independent replication is published, the notation and its significance will be reviewed accordingly.
Some Oxfordian scholars argue that the F.A. designation should be scored as a full ✓, on the grounds that forensic document examination is a recognised specialist discipline and that Stritmatter's study meets its methodological standards. That argument is not dismissed here; the finding is treated seriously. The distinction between F.A. and ✓ is not a judgment on the quality of the scholarship but a consistent notational principle: any claim requiring specialist expert interpretation, rather than a directly readable primary source document, is differentiated from direct documentation. The same logic applies to all candidates equally. For a fuller discussion of what the Audley End findings establish, and what they do not, see the dedicated analysis.
The strongest exception to the general weakness of biographical correspondence arguments is the Polonius/Certain Precepts connection for de Vere. William Cecil, Lord Burghley, de Vere's guardian and father-in-law, wrote a letter of advice to his son Robert that survives as a primary source document. Its specific counsel against borrowing and lending, against quarrelling, and against extravagance corresponds to Polonius's advice to Laertes in sequence and language close enough to invite direct textual comparison. The documentary correspondence is real and available; the inference from it to authorship is not scored but is noted on the de Vere candidate page.
One Unresolved Condition
The two frameworks above are scoring tools applied equally to all candidates. A separate condition sits alongside them; one that neither framework directly measures, but that any complete theory of authorship must satisfy.
Why does William Shakspere of Stratford leave no literary trace of a writing life? Price surveyed twenty-four of his literary contemporaries under the same record conditions and found that every one left at least one primary source document connecting them to writing, a letter, a payment, a manuscript, a record of books owned, a reference to work in progress. Shakspere left none. His general biographical record is substantial: over seventy documents survive, covering baptism, marriage, property, litigation, a will. Not one touches his activities as a writer.
This is a condition of completeness, not a score. A Stratfordian theory must explain the absence; an alternative theory must explain both who the real author was and why Shakspere left no trace. No current theory has fully satisfied it. The What the Evidence Returns page draws together what the frameworks produce, including this condition.
Let me be precise about how this condition relates to the authorship question itself, because they are different kinds of unknown and they resolve asymmetrically. The authorship question, who wrote the works, can be closed by a single primary source document of the right type. The Stratford anomaly, why Shakspere's substantial documentary record contains nothing literary, and what his role actually was, closes with it in only one direction. A document connecting Shakspere himself to writing would resolve both at once: it would settle the attribution and supply the missing literary record in the same stroke. A document naming any other candidate would close the authorship question while sharpening the anomaly, his name is on the title pages, the monument, and the First Folio, so he had a role in the story whoever held the pen, and proof that the pen was elsewhere would convert that role from an unexplained absence into an unexplained presence. One discovery would end the enquiry; the other would end only half of it.
Framework Review Log
The convergence lines and the framework definitions above were fixed before any candidate was scored, and they do not change silently. When correspondence or further work raises a design objection that survives first scrutiny, it is recorded here with its status: under review, adopted (with a note of what changed), or declined (with reasons). Any adopted change applies to all fourteen candidates identically. Contributors are credited by the site’s usual convention, anonymously as “a correspondent” unless they ask to be named. Corrections of published errors are recorded separately, on the errata page; this log is for the design of the instrument itself.
| Raised | Issue | Status |
|---|---|---|
| 6 July 2026 | The Southampton connection line is framed as a “documented relationship,” which presumes what a dedication evidences. The printed fact is the dedications themselves; the interpretations (patronage, commission, parody, courtship) are contested, and Southampton’s documented finances in 1593–4 tell against the patronage reading specifically. Proposed remedy: re-anchor the line’s title and note to the printed dedications. Raised by a correspondent. Supporting citations on manuscript-access parallels (Weiss on Daniel and the Henry plays; Moffett’s Silkworms and Dream, with Duncan-Jones contra) received 7 July 2026; under audit against the discriminating-access standard. Adopted 12 July 2026. The title and note are re-anchored to a documented, personal, dedication-period tie to Southampton himself, and the note now states that shared membership of a wider patronage network is not scored, because a broad network discriminates among none. The audit of every candidate’s Line 1 cell found three scores, Marlowe, Neville, and Rutland, resting only on that network reading; all three were corrected ◉→~ and are recorded on the errata page. The supporting source-access citations were declined under the discriminating-access standard: Daniel’s Civil Wars was printed (1595) and separates no one, and the Moffett–Dream link is both contested (Duncan-Jones) and dependent on a pre-1599 composition date the site does not allow to do evidentiary work. | Adopted (with change note) |
| 6 July 2026 | The collaborative-authorship criteria overlap with the convergence lines: documented connection to Daniel, Kyd, Marlowe, and the Pembroke’s Men circle bears on single authorship at least as much as on collaboration. Proposed remedy: clarify or redraw the boundary between the two sets of criteria. Raised by a correspondent. Adopted 20 July 2026. The two frameworks are held to answer different questions, and a boundary note now says so: the convergence framework discriminates a single author, so network, competence, and posthumous connections do not score there; the separate collaborative-authorship framework characterises a documented collective, so those same connections are weighed there and are not added into the convergence total. The note appears on the comparison tool and is mirrored on the collaboration and convergence pages, and the “convergence of independent threads” wording on the collaboration page was tightened so that network density is not presented as identifying a sole author. No score changed. | Adopted (boundary clarified) |
| 6 July 2026 | A candidate’s connection to the First Folio carries evidential weight that may be muted by where it is scored. Proposed remedy: reconsider the placement and weight of Folio connections across the frameworks. Raised by a correspondent. Adopted 20 July 2026. The Folio connection is already scored, in the collaborative-authorship framework (its First Folio / Pembroke dimension), and is correctly excluded from the convergence framework, where line 17 records the 1623 Folio as posthumous and non-scoring for every candidate. Rather than promote it into the convergence total, which would re-import posthumous testimony and network proximity; its place is fixed by the new boundary note. A paragraph was added to the documentary problem page recording that the Folio’s front matter is generic and literary rather than personal. No score changed. | Adopted (weight located) |
A proposed new evidence line is judged by one test before all others: not “is the source documented?” but “does access to it discriminate between candidates?” Printed sources were available to every literate candidate and separate no one; unpublished or restricted material can. This is the same standard the site applies to Plowden’s law reports and Florio’s unpublished Montaigne translation.