OXFORD BY THE NUMBERS: WHAT ARE THE ODDS …€¦ · · 2009-10-20ODDS THAT THE EARL OF OXFORD COULD HAVE WRITTEN SHAKESPEARE’S POEMS AND PLAYS? ... Alan Nelson and Steven May,

* Ward E.Y. Elliott is the Burnet C. Wohlford Professor of American PoliticalInstitutions at Claremont McKenna College. Robert J. Valenza is the Dengler-DykemaProfessor of Mathematics and the Humanities at Claremont McKenna College. They would liketo thank the thirty-seven alumni of the Claremont Shakespeare Clinic who pioneered much ofthe research reported herein. They also wish to thank The Shakespeare Quarterly, Computersand the Humanities, Literary and Linguistic Computing, The Oxfordian, Chance, Notes andQueries, and CMC Magazine for publishing their prior articles on Shakespeare authorship, fromwhich the present Article is adapted. Figures 3.1 to 3.7 were prepared for us by StephanieHughes, editor of The Oxfordian. The tables, appendices, and graphics of this Article areposted separately at http://govt.claremontmckenna.edu/welliott/UTConference/.

323

OXFORD BY THE NUMBERS: WHAT ARE THEODDS THAT THE EARL OF OXFORD COULD HAVEWRITTEN SHAKESPEARE’S POEMS AND PLAYS?

WARD E.Y. ELLIOTT AND ROBERT J. VALENZA*

Alan Nelson and Steven May, the two leading Oxford documentsscholars in the world, have shown that, although many documentsconnect William Shakspere of Stratford to Shakespeare’s poems andplays, no documents make a similar connection for Oxford. Thedocuments, they say, support Shakespeare, not Oxford. Our internal-evidence stylometric tests provide no support for Oxford. In terms ofquantifiable stylistic attributes, Oxford’s verse and Shakespeare’s verseare light years apart. The odds that either could have written theother’s work are much lower than the odds of getting hit by lightning.Several of Shakespeare’s stylistic habits did change during his writinglifetime and continued to change years after Oxford’s death. Oxfordianefforts to fix this problem by conjecturally re-dating the plays twelveyears earlier have not helped his case. The re-datings are likewise ill-documented or undocumented, and even if they were substantiated, theywould only make Oxford’s stylistic mismatches with early Shakespearemore glaring. Some Oxfordians now concede that Oxford differs fromShakespeare but argue that the differences are developmental, likethose between a caterpillar and a butterfly. This argument is neitherdocumented nor plausible. It asks us to believe, without supportingevidence, that at age forty-three, Oxford abruptly changed seven to nineof his previously constant writing habits to match those of Shakespeareand then froze all but four habits again into Shakespeare’s likeness forthe rest of his writing days. Making nine such single-bound leaps froma distant, stylistically frozen galaxy right into Shakespeare’s ballparkseems farfetched compared to the conjectural leaps required to take theStratford case seriously. Note, for example, the supposition that theyoung Shakespeare, who was entitled to do so, might actually have

324 TENNESSEE LAW REVIEW [Vol. 72:323

attended the Stratford grammar school. It is hard to imagine any jurybuying the Oxfordians’ colossal mid-life crisis argument without muchmore than the “spectral and intangible” substantiation that it hasreceived. Ultimately, this argument is too grossly at odds with theavailable documentary record and stylometric numbers for Oxford tobe a plausible claimant.

I. INTRODUCTION: DESPERATELY SEEKING SHAKESPEARE . . . 325II. OUR METHODOLOGY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 332

A. Quantitative Internal Evidence Versus Qualitative Internal and External Evidence . . . . . . . . . . . . . . . . . . . 332

B. Clean, Commonized Baseline . . . . . . . . . . . . . . . . . . . . . 335C. “Silver-Bullet” Negative Evidence . . . . . . . . . . . . . . . . . 337D. Block and Profile: Sample Length Matters . . . . . . . . . . 341E. Finding the Comparative Odds of Shakespeare

Authorship . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 344F. Discrete Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 348G. Continuous Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350H. Some Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 352I. Strengths and Weaknesses of Discrete Analysis . . . . . . . 355J. Strengths and Weaknesses of Continuous Analysis . . . . . 356K. Accuracy in Distinguishing Shakespeare

from Non-Shakespeare . . . . . . . . . . . . . . . . . . . . . . . . . . 356L. Factors That Affect Accuracy . . . . . . . . . . . . . . . . . . . . . 358M. Five Tests of Validity . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360N. Set-Asides and Latent Variability . . . . . . . . . . . . . . . . . . 365

III. HOW OUR METHODS APPLY TO THE EARL OF OXFORD . . . . . 368A. Oxford Fails Too Many Tests to Be a Shakespeare

Could-Be . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 368B. Test Specifics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372C. Test Outcomes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 373D. Summary of Oxford Rejections . . . . . . . . . . . . . . . . . . . . 376E. A Comparison of Orthodox and Oxfordian

Chronologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 376F. Lack of Oxfordian Consensus . . . . . . . . . . . . . . . . . . . . . 377G. Problems with External Evidence . . . . . . . . . . . . . . . . . . 379H. Problems with Internal Evidence . . . . . . . . . . . . . . . . . . 382I. Feminine Endings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382J. Open Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385K. Midline Speech Endings . . . . . . . . . . . . . . . . . . . . . . . . . 387L. Conclusions on Chronology . . . . . . . . . . . . . . . . . . . . . . 388M. Possible Discounts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389N. Grub or Butterfly? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390

IV. CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396V. APPENDICES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397

2004] OXFORD BY THE NUMBERS 325

1. The plays and poems of William Shakespeare are often collectively referred to as the“Shakespeare Canon.” On our last visit to the World Shakespeare Bibliography in 2001, wecounted eighteen pages of references and 533 entries. Fifty-seven percent of these entries askedwhether Shakespeare was the Stratford man; 23% concerned his hand in the “ShakespeareApocrypha,” works outside the Shakespeare Canon which have been attributed to Shakespeareat one time or another; and 21% dealt with the “Shakespeare Dubitanda,” works inside theCanon for which Shakespeare’s authorship has been questioned. See World ShakespeareBibliography Online, at http://www.worldshakesbib.org (last updated July 9, 2004).

2. When we began the Shakespeare Clinic, the case for Oxford was the following:Shakespeare the playwright seemed not only the greatest writer of all time but also a man oftremendous erudition. He had a known vocabulary of more than 17,000 words, a vocabularytwice the size of Milton’s and often said to be the largest ever. He appeared educated in law,falconry, heraldry, French, Latin, Greek, and English history. He seemed to have traveledthroughout Europe, especially Italy, and to have known the ways of kings and courtiers, as ifby first hand. His sonnets were then believed to have been written in the 1590s. They suggestthe following about him at that time: (1) that he was in his forties, Sonnets 2, 62; (2) that he waslame, Sonnets 37, 89; (3) that he had borne an “outward honoring” canopy, Sonnet 125; (4) thathe had a Dark Lady, Sonnets 127-152, and a Fair Youth whom he hoped would have children,Sonnets 1-126; and (5) that he might be writing under a pseudonym, Sonnet 76. WILLIAM

SHAKESPEARE, Sonnets (All quotations from and references to the Shakespeare Canon in thisArticle are from THE RIVERSIDE SHAKESPEARE (G. Blakemore Evans ed., 2d ed. 1997)[hereinafter THE RIVERSIDE SHAKESPEARE].). Shakespeare writes:

Why write I still all one, ever the same,And keep invention in a noted weed,That every word doth almost tell my name,Showing their birth, and where they did proceed?

WILLIAM SHAKESPEARE, Sonnet 76. But see Sonnets 135-136 (containing ten plays on the wordWill, including “for my name is Will”).

In addition, he was extravagantly praised by his contemporaries and hailed by Ben Jonsonas the “Soule of the Age!” Ben Jonson, To the Memory of My Beloved, the Author Mr. WilliamShakespeare: And What He Hath Left Us, in the First Folio, reproduced in THE RIVERSIDE

SHAKESPEARE, supra at 97.Such an image seemed like a glowing Christmas tree of amazing resemblances to the Earl

I. INTRODUCTION: DESPERATELY SEEKING SHAKESPEARE

Who wrote Shakespeare? What did he write? Thousands of books andarticles have been written on the first question since the 1900s, doubting thatthe lowly William Shakspere, the Stratford glover’s son, London bit actor andtheater shareholder, could have written the poems and plays of WilliamShakespeare, the greatest writer of all time.1 The contrast betweenShakspere’s supposedly humdrum, grasping, mercantile documents and theall-surpassing sophistication and learning of Shakespeare’s plays has seemedtoo great for many to believe. Surely, as many have argued, a more credibleauthor would be a traveled, polished, and educated noble like Edward de Vere,the seventeenth Earl of Oxford, today’s leading claimant to be the “TrueShakespeare.”2 As Sigmund Freud stated, “The man [from] Stratford


of Oxford, who had the upbringing of an earl, held degrees from Oxford and Cambridge, hadmade the Grand Tour, had lingered long in Italy, was a prominent presence in the Court, borethe Queen’s canopy in royal processions, was a published poet in his forties, was lame, wasassociated with an obvious Dark Lady, Anne Vavasour, and a Fair Youth, Henry Wriothesley,the Third Earl of Southampton and the then-intended match for Oxford’s daughter, Elizabeth,from whom he naturally would have hoped for children. See CHARLTON OGBURN, THE

MYSTERIOUS WILLIAM SHAKESPEARE: THE MYTH AND THE REALITY (1984). The resemblances to William Shakspere of Stratford seemed much fewer and less striking.

Shakspere appeared in documents and anti-Stratfordian literature as provincial, obscure, andbarely literate. In his will, he devised his second-best bed but mentioned no books ormanuscripts. Id. at 33-35. His parents and his daughters signed with an “x.” His records spokeof births, deaths, marriages, deals in land, corn, malt, stones, and claret, and suits for debts,taxes, coats of arms, and the abatement of dunghills—everything, it seemed, but literarytransactions or interests. Id. at 30-31. They showed nothing in his life of knowing a Dark Ladyor a Fair Youth, being lame, making the Grand Tour, bearing a canopy, or receiving a fineformal education. Shakspere had just turned thirty, not forty, when the sonnets were thenthought to have been written, and he was not known to have been lame. Why all the moaningabout “forty winters,” Sonnet 2, “bare ruin’d choirs,” Sonnet 73, “decrepit father,” Sonnet 37,and “Death to me subscribes,” Sonnet 107, from one who could hardly have experienced suchthings first-hand?

If one were to judge solely from such selected evidence, Oxford does seem to be a morelikely candidate than Shakspere, if only because he was a more prominent person and had amuch fuller record to ornament his Christmas tree with a profusion of fancied resemblances.On the other hand, the same exercise has been performed on other prominent claimants withessentially the same results: One could ornament their Christmas trees with long, glowingstrings of amazing resemblances to Shakespeare, many more and more glowing than what theycould find for Shakspere of Stratford. See, e.g., ALFRED DODD, FRANCIS BACON’S PERSONAL

LIFE-STORY 339-47 (1986) (concluding that Francis Bacon wrote the anonymous playsaccredited to Shakespeare); CALVIN HOFFMAN, THE MURDER OF THE MAN WHO WAS

‘SHAKESPEARE’ (1955) (arguing that Christopher Marlowe wrote the works accredited toShakespeare). However, not everyone with a fully decorated Christmas tree can be Shakespeare(unless you subscribe to the group theory, which neither our evidence nor any evidence in thedocuments supports), and only a few blown circuits turn out all the lights. See infra Part III;see generally GILBERT SLATER, SEVEN SHAKESPEARES (1931) (espousing the theory that sevenwriters authored the works of Shakespeare).

Since we began our work, a steady trickle of new documents and studies have shown anumber of breaks in the circuit for the Oxford case. See e.g., IRVIN LEIGH MATUS,SHAKESPEARE IN FACT (1994); ALAN NELSON,MONSTROUS ADVERSARY: THE LIFE OF EDWARD

DE VERE, 17TH EARL OF OXFORD (2003) [hereinafter NELSON, MONSTROUS ADVERSARY]; AlanNelson, Shakespeare Authorship Pages, at http://ist.socrates.berkeley.edu/~ahnelson/authorsh.html (last visited Nov. 13, 2004). These breaks include Oxford’s strange, un-Shakespeareanspelling and prosody, his grossly overrated academic credentials, his all-consuming self-absorption, and the total lack of direct evidence connecting him with Shakespeare’s poems orplays. See MATUS, supra; NELSON, MONSTROUS ADVERSARY, supra. At the same time, thecase for Shakspere has been strengthened by assigning the bulk of the sonnets to theseventeenth century, see STANLEY WELLS & GARY TAYLOR, WILLIAM SHAKESPEARE: ATEXTUAL COMPANION 444 (1987); A. Kent Hieatt et al., When Did Shakespeare Write Sonnets1609?, 88 STUD. PHILOLOGY 69, 70, 98 (1991), by re-examining old documents linking him to


the plays, and by the continued appearance of new ones, see Steven W. May, The SeventeenthEarl of Oxford as Poet and Playwright, 72TENN.L.REV. 221 (2004); Alan H. Nelson, StratfordSi! Essex No! (An Open-and-Shut Case), 72 TENN. L. REV. 149 (2004) [hereinafter Nelson,Stratford], and also by the recent discovery by us and David Crystal that Shakespeare’serudition, as measured by the size of his vocabulary relative to others’ has been grosslyoverrated. See DAVID CRYSTAL, THE STORIES OF ENGLISH 322-29 (2004); Ward E.Y. Elliott& Robert J. Valenza, Was Shakespeare’s Vocabulary the Greatest? 21-23 (Oct. 10, 2004)(unpublished manuscript, on file with the Tennessee Law Review). Many things that might haveseemed mysterious about Shakespeare in 1987 seem less so now. The obverse of our notionthat all map-changing Shakespeare authorship documents may have been discovered andexhausted by E.K. Chambers’s time is the fact that new documents continue to appear that donot change the map. Instead, those new documents reaffirm the most basic proposition sinceShakespeare’s own time: Shakespeare was the core author and not someone else. Our internal,stylometric evidence is consistent with this long history of documentary evidence.

3. OGBURN, supra note 2, at 146 (quoting from Freud’s letter to author John ThomasLooney).

4. See Shakespeare Oxford Society, The Honor Roll of Skeptics: Questioning theOrthodoxy, at http://www.shakespeare-oxford.com/skeptic.htm (last visited Mar. 11, 2005)(listing prominent Shakespeare sceptics).

5. See, e.g., MATUS, supra note 2; Irvin Leigh Matus, Irvin Leigh Matus’s ShakespeareSite, at http://www.willyshakes.com (last modified July 26, 2004); The Shakespeare AuthorshipPage, at http://shakespeareauthorship.com (last visited Mar. 13, 2005); The ShakespeareAuthorship Roundtable: Forum for the Authorship of the Shakespeare Canon, athttp://www.shakespeareauthorship.org (last visited Mar. 13, 2005).

6. See, e.g., Roland Barthes, The Death of an Author, in IMAGE, MUSIC, TEXT 142-48

. . . seems to have nothing at all to justify his claim, whereas Oxford hasalmost everything.”3 Otto von Bismarck and Walt Whitman were Stratfordskeptics in the nineteenth century. Freud, Mark Twain, and John Galsworthywere prominent anti-Stratfordians of the early twentieth century. They werefollowed by a host of lawyers, members of Parliament, and Washingtonnotables, including Ambassador Paul Nitze, three U.S. Supreme Court Justices(Justices Stevens, Blackmun, and Powell) and Harvard Professor William Y.Elliott, prominent Tennessean and father of one of the present authors.4

The controversy still rages on the Internet, in the media, and among thegeneral public—everywhere, in fact, but in modern English departments.These departments have always considered the question of who wroteShakespeare to be of interest only to hobbyist amateurs, suitable for theNational Enquirer or possibly for Harper’s but hardly for the ShakespeareQuarterly. As a result, until Alan Nelson and Steven May entered the debate,most of the work on this question, not only attacking Stratfordian orthodoxybut also defending it, has been done by amateurs.5 Some English departmentsdo retain a healthy, informed interest in the second question: “What didShakespeare write?” However, most trendy, postmodern ones do not. Sincethe “Death of the Author” was proclaimed in the 1960s, these departmentshave shunned not only Shakespeare, but every other kind of authorshipquestion.6 Their search for Shakespeare these days is less than desperate.


(Stephen Heath trans., 1977) (1968) (discussing the relationship between an author and hiswork); Michel Foucault, What Is an Author?, in TEXTUAL STRATEGIES: PERSPECTIVES IN POST-STRUCTURALIST CRITICISM 141-60 (Josué V. Harari ed., 1979) (discussing what it means to betermed an author). But see SEÁN BURKE, THE DEATH AND RETURN OF THE AUTHOR (2d ed.1998) (questioning the “Death of the Author” theory); BRIAN VICKERS, SHAKESPEARE, CO-AUTHOR: A HISTORICAL STUDY OF FIVE COLLABORATIVE PLAYS 506-41 (2002) (challengingFoucault’s theories on authorship).

7. In Re Shakespeare: The Authorship of Shakespeare on Trial, 37 AM. U. L. REV. 609-826 (1988).

8. NELSON, MONSTROUS ADVERSARY, supra note 2; see also Alan Nelson, OXDOX:Earl of Oxford Documents, at http://socrates.berkeley.edu/~ahnelson/oxdocs.html (last visitedJan. 18, 2005).

9. Steven W. May, The Poems of Edward DeVere, Seventeenth Earl of Oxford and ofRobert Devereux, Second Earl of Essex, 77 STUD. PHILOLOGY 1, 5 (1980).

10. Nelson summarizes his case as follows: In sum, the First Folio informs us that the playwright William Shakespeare hailed from

Stratford-upon-Avon, spent part of his life in Stratford and part in London, served underboth Elizabeth and James, belonged to the same fellowship of players as John Heminges

The general public and especially its lawyerly elites have never boughtinto postmodernism and are avidly interested in both the who and the whatquestions. Dozens of debates have been staged on the who question in theUnited States and Great Britain, many of them at law schools. Before the June2004 symposium at the University of Tennessee College of Law, the mostambitious debate was a 217-page symposium issue published by the AmericanUniversity Law Review in 1988.7 The Tennessee symposium, however, hada range of talent not available to the 1988 debate. The anti-Stratfordianpanelists included three of the Shakespeare Oxford Society’s top speakers andwriters: Roger Stritmatter, an actual practicing literature professor, andRichard Whalen and Diana Price, both published by respectable presses. LikeFreud, these panelists were convinced from years of study that the sublimeShakespeare looks more like the sublime Oxford (or someone else like him)than the lowly Stratford man. The anti-Oxfordian panelists did not includeShakespeare specialists, who shun the topic, but did include the two topOxford documents specialists in the world, Berkeley’s Alan H. Nelson, authorof a new Oxford biography, Monstrous Adversary,8 and Georgetown College’sSteven May, editor of Oxford’s collected poems.9 Both argued at thesymposium, as shown herein, that the Stratford man of record looked morelike Shakespeare than the Oxford of record. Oxford, more than Shakspere,was the one whose letters were grasping and mercantile. His own formaleducation fizzled out without distinction at age thirteen; his Oxford andCambridge diplomas turned out to be ceremonial souvenirs, not earneddegrees; he always signed himself “Oxenford,” never Oxford, as Shakespearecalled his ancestors. Nothing in Oxford’s record shows any connection withShakespeare’s poems or plays, and his known poems do not look and soundlike Shakespeare’s at all.10


and Henry Condell, and was one of twenty-six principal actors of the company which firstperformed his own plays. Moreover, the First Folio tells us that Shakespeare had a limitededucation (he knew some Latin and less Greek), was a friend of Ben Jonson, was at leastan acquaintance of Leonard Digges (also of James Mabbe and Hugh Holland), was buriedat Stratford where a funeral monument was erected in his honor, and, finally, was thesubject of a eulogy by the poet William Basse, who placed Shakespeare in the companyof, but thought him finally superior to, Geoffrey Chaucer, Edmund Spenser, and FrancisBeaumont.

By the same token, the First Folio is entirely incompatible with the assumption thatOxford was the author of its thirty-odd plays. Under no circumstances would an earl becalled “Maister,” not even on the basis of a university degree, earned or unearned.Heminges and Condell could not have publicly or accurately described Oxford as their“fellow.” To call Oxford a “servant” of the Herbert brothers would have been an outrageto his rank, not only because all three were earls but because the earldom of Oxfordoutranked the earldoms of Pembroke and Montgomery. Except in the wildest Oxfordianfantasies, Oxford was not one of the twenty-six “principal Actors in all these Playes.”Oxford had nothing to do with Stratford. He had no monument there and had noconnection to the Avon except that he had once owned, but never visited, an estate namedBilton on the Avon, which he sold by the mid-1580s. Finally, Jonson would not havequalified his admiration for a man so socially superior, nor would he have accused an earl,whether true or not, of having had “small Latine, and lesse Greeke.”

Nelson, Stratford, supra note 2, at 156-57.Likewise, May provides the following argument:

Let me conclude by summarizing what we must believe if we are to accept theOxfordian hypothesis. Since nothing in Oxford’s canonical verse in any way hints at anaffinity with the poetry of William Shakespeare, we must believe that Oxford made theleap from his mid-century poetic style to the late Elizabethan style without leaving behinda trace of transitional writing. We must next believe that, after publishing both verse andprose under his own name, the Earl was suddenly afflicted with a manic compulsion foranonymity. This compulsion did not lead him, however, to protest published referencesto him as a playwright or the on-going publication of verse under his name or initials. Hethen enlisted, not someone among his own players, but the obscure actor from Stratfordto set forth his own creative writing as Shakespeare’s work and under his name. We mustnext believe that De Vere and William took this secret to their graves, that they fooledeveryone. We must believe that as Oxford’s finances deteriorated in the last decades ofhis life, he nevertheless permitted his accomplice Shakespeare to enjoy the profits thataccrued from the popular and courtly success of both the Earl’s plays and his non-dramaticwritings. Finally, to be Oxfordians, we must believe that the Earl left Shakespeare asubstantial corpus of plays and poems before he expired in obscurity and poverty.Shakespeare then parceled out his patron’s work, successfully representing it as his ownto the end of his professional career some eight years after Oxford’s death. As a result,Shakespeare died well-to-do, if not wealthy, and highly esteemed to this very day for hisalleged literary accomplishments. What a tale of clandestine intrigue, bizarre passion, plusthe wholesale outwitting of friends, family, colleagues, and acquaintances for over twentyyears. Can you believe it?

May, supra note 2, at 242.


11. THE READER’S ENCYCLOPEDIA OF SHAKESPEARE 115 (Oscar J. Campbell & EdwardG. Quinn eds., 1966).

We have come to the same conclusion about Oxford. Neither he, nor anyother claimant we tested, is the True Shakespeare. However, we arrived at thisconclusion by a very different route, based not on documents but on aquantitative comparison of Shakespeare’s stylistic habits to those of Oxfordand others. Our evidence draws on the original work of the ClaremontShakespeare Clinic, a series of student-run teams that worked from 1987 to1994 and was originally funded by the Sloan Foundation. With much effortand ingenuity, the students put together what could still be the largestcommon-spelling, computer-ready Elizabethan poem and play archive inexistence and developed new computer techniques to shorten the list ofcredible, testable claimants. Table 1.1 shows their starting point: a list offifty-eight full and partial claimants, primarily taken from The Reader’sEncyclopedia of Shakespeare.11

Table 1.1. Fifty-Eight Full or Partial Shakespeare Claimants

Alexander, William (1568–1640)*Bacon, Francis (1561–1626)*Bacon, Anthony (1558–1601)Barnfield Richard (1574–1627)Barnes, Barnabe (1571–1609)Bernard, Sir John (1605–1674)*Blount, Sir Charles (1563–1601)Burbage, Richard (1567–1619)*Burton, Robert (1577–1640)Butts, William (d. 1583)Cecil, Robert (1563–1612)Chettle, Henry (1560–1607)Daniel, Samuel (1562–1619)*de Vere, Edward (Oxford) (1550–1604)Dekker, Thomas (1572–1632)Devereux, Walter (1541?–1576)*Devereux, Robert (Essex)(1566–1601)*Donne, John (1572–1631)Drayton, Michael (1563–1631)*Dyer, Sir Edward (1543–1607)Ferrers, Henry (1549–1633)Fletcher, John (1579–1625)Florio, John (1554–1625)Florio, Michelangelo

Kyd, Thomas (1558–1594)Lodge, Thomas (1557–1625)*Lyly, John (1554–1606)*Manners, Roger (Rutland)(1576–1621)Manners, Elizabeth Sidney (d. 1615)*Marlowe, Christopher (1564–1593)Middleton, Thomas (1580–1627)Munday, Anthony (1560–1633)Nashe, Thomas (1567–1601)Paget, Henry (d. 1568)Peele, George (1556–1596)Porter, Henry (fl. c. 1596–99)*Raleigh, Sir Walter (1554–1618)*The RosicruciansSackville, Thomas (1536–1608)Shirley, Sir Anthony (1565?–1635)*Sidney Herbert, Mary (1561 –1621)*Sidney, Sir Philip (1554–1586)Smith, Wentworth (William) (fl. c.1615)*Spenser, Edmund (1552–1599)*Stanley, William (Derby)(1561–1642)Stuart, Mary (l542–1587)*Tudor, Queen Elizabeth (1533–1603)


12. See THE RIVERSIDE SHAKESPEARE, supra note 2, at 1893-95.

Greene, Robert (1558–1592)Griffin, Bartholomew (d. 1602)*Heywood, Thomas (1574–1641)The Jesuits*Jonson, Ben (1572–1637)

Warner, William (c. 1558–1609)Webster, John (1580?–1625?)Whateley, AnneWilson, Robert (>1572–1600)Wolsey, Thomas Cardinal(1473?–1530) *Wriothesley, Henry (1573–1624)

Table 1.1. Twenty-one “full” claimants identified for us by the Francis BaconLibrary in Claremont, California are preceded by an asterisk. Thirty-seven testedclaimants, fifteen full and twenty-two partial, are italicized. The remaining twenty-one claimants, not italicized, six of them full, have left no known poems or plays totest.

The Claremont students succeeded beyond anyone’s expectations. Theyshortened the plausible, testable claimant list from thirty-seven to zero, andthey eliminated every play and poem of the Shakespeare Apocrypha asShakespeare’s. Among the rejects was A Funeral Elegy by W.S. (“A FuneralElegy”), the great “Shakespeare find” of the 1990s which was touted in allthree U.S. Complete Shakespeare Works editions of the decade as “possiblyShakespeare’s”12 and is only now getting un-touted. When the studentsannounced that their tests eliminated Oxford, Bacon, and Marlowe, theyreceived worldwide media attention. ABC, NBC, BBC, and several othernetworks covered their report, and Science magazine and more than onehundred newspapers here and abroad reported on their findings.

We were co-advisors to the clinic. When the students left, we developedand extended their work into a dozen articles on Shakespeare authorship inleading journals, and we defended it successfully against years of heavyattacks by critics. We are now updating and consolidating the articles into abook, Shakespeare by the Numbers.

Our main findings are the following: Shakespeare’s writings do showconsistent, countable, profile-fitting patterns, suggesting that, whoever he was,he was a single individual, not a committee. He used more hyphens, feminineendings, and open lines than most others, and fewer relative clauses. Hisknown poems were written between the tenth- and fourteenth-grade level.Other authors’ poems fit some, but not all, of these profiles. However, fittinga Shakespeare profile does not prove your poem is by Shakespeare any morethan fitting a size-four slipper proves that you are Cinderella. You could aswell be Tiny Tim. But not fitting the slipper profile jeopardizes your claim,and your claim is even weaker if you do not fit two, three, or four identifyingprofiles—not just shoe size but hat size, belt size, and eye color for example.

Our calculations show that the odds of not fitting six profiles in fourteentests, like A Funeral Elegy, or seven profiles, like Oxford’s poems, are


13. 509 U.S. 579 (1993).14. 526 U.S. 137 (1999). 15. Michael J. Saks, The Legal and Scientific Evolution of Forensic Science (Especially

Fingerprint Expert Testimony), 33 SETON HALL L. REV. 1167, 1167 (2002-2003) (footnoteomitted).

infinitesimal compared to the farthest outlier block from Shakespeare’s ownbaseline, which has only one narrow rejection. Unless Oxford’s writing habitschanged abruptly, miraculously, and simultaneously in seven to nine differentways in the 1590s, he cannot be Shakespeare. None of the other claimants’work that we tested could be Shakespeare’s work. Additionally, some of thelong-disputed parts of the Canon, such as A Lover’s Complaint and much ofTitus Andronicus, could not be Shakespeare’s work. One or two of theOxfordian panelists argued that, if writing Hamlet and the sonnets were acrime, Oxford would surely be convicted beyond a reasonable doubt. Ourfindings prove exactly the opposite. Neither the Earl of Oxford nor any otherclaimant we tested could possibly be guilty of such a thing.

II. OUR METHODOLOGY

Readers of this journal are aware that, since Daubert v. Merrell DowPharmaceuticals, Inc.13 and Kumho Tire Co. v. Carmichael,14 it is no longersufficient to wave an expert witness’s credentials in front of the jury andexpect them to swallow his or her conclusions whole. As Michael J. Saksstated:

[I]t would make sense to admit . . . expert evidence only if it meets theseconditions: (a) the opinions and conclusions of the expert are accompaniedby information that enables the factfinder to evaluate the likely accuracy ofthe expert’s opinion, and (b) the information is presented in such a way thatfactfinders will not be fooled into excessively overvaluing the testimony.15

The authorship controversy has not been litigated much lately, but we wouldthink that the Daubert rules as Saks restated them should also apply to alandmark debate on a nontrivial topic. In Part II, we present here a generaldiscussion of the validity of our methods. Part III follows with a specificdiscussion of how these methods apply to the seventeenth Earl of Oxford, theleading claimant. In Part IV, we set forth our conclusions.

A. Quantitative Internal Evidence Versus QualitativeInternal and External Evidence

Our methodology has five main distinctive features: (1) quantitativeinternal evidence; (2) clean, commonized baseline; (3) negative evidence;(4) block and profile; and (5) comparative authorship odds. “Quantitativeinternal evidence” means that our primary scholarly concern was countable


16. External evidence for Shakespeare includes the works of Henslowe and Meres, theStationer’s Register, and the Folios and Quartos. See also S. SCHOENBAUM, SHAKESPEARE’S

LIVES (new ed. 1991) (discussing Shakespeare documents); WELLS & TAYLOR, supra note 2(evaluating external and internal evidence of Shakespeare authorship of various works).

17. Among internal-evidence people, we are the ones who like to count things easy tocount—grade-level, hyphenated compound words, feminine endings, and the like—rather thanthings that are hard to count—literary parallels, echoes of other writers, shared imagery andallusions, or use of “distinctive” rhetorical figures. In terms of our conclusions, we should beclassified as “disintegrationists,” people who doubt that Shakespeare wrote everything in theCanon, rather than as “integrationists,” those who think Shakespeare wrote it all, and as“Stratfordians” rather than “Anti-Stratfordians.”

18. HARALD WESTERGAARD, CONTRIBUTIONS TO THE HISTORY OF STATISTICS 14 (1932).19. See, e.g., S. SCHOENBAUM, INTERNAL EVIDENCE AND ELIZABETHAN DRAMATIC

AUTHORSHIP: AN ESSAY IN LITERARY HISTORY AND METHOD (1966).20. See, e.g., Arthur Sherbo, The Uses and Abuses of Internal Evidence, in EVIDENCE FOR

AUTHORSHIP: ESSAYS ON PROBLEMS OF ATTRIBUTION 559 (David V. Erdman & Ephim G.Fogel eds., 1966).

stylistic markers that might give us a clearer notion of authorship.“Stylometry” is a word sometimes used to describe work like ours. We oftentook our bearings from conventional, qualitative scholarship based on soft,internal evidence—perceived “borrowings” and literary parallels between onewriter and another—and on external documentary evidence.16 We performedour tests as consumers, not producers, of conventional qualitative judgmentsand documentary research.

In one sense, this could be a terrible disadvantage. Even we believe thatfull authorship analysis requires a look at both right-hand (qualitativejudgments and external documents) and left-hand (internal, countable stylisticpatterns) evidence, as well as some admixture of interpretive skills andintuition before one can claim to have exercised due diligence. From thisperspective, if from no other, we are incorrigible lefties.17 By the same token,we have encountered many lawyers and literature department people, perhapsthe very ones we are trying to reach in this Article, who are incorrigiblerighties; they find counting of any kind uninteresting and serious countingwith statistics and computers perverse, soulless, abhorrent, and unprofessional.They are like the elders of the University of Göttingen in the eighteenthcentury who described the Tabellenknechte as “slaves to the tabular form ofpresentation, who g[i]ve only the dry bones of statistics without clothing themwith the flesh of descriptive reality.”18 Their model of excellence is JohnHenry, the legendary steel driver who outperformed the steam drill and diedwith his hammer in his hand.

For some purposes they are right. The handcrafting man still can do sometasks better; on the other hand, how many of our handcrafters still write outtheir thoughts in longhand with a quill pen? In 1966, just before the “Deathof the Author” was proclaimed, there was a major exchange between theexternal-evidence righties19 and the internal-evidence lefties20 over whose


21. Perhaps significantly, when moderator Professor Judy Cornett asked the panelists ofthe Tennessee symposium what kind of evidence they thought might clinch the case for oragainst the Stratford man, almost everyone answered external evidence, such as a signedmanuscript or, in our case, a signed, sealed, sworn, notarized affidavit from Shakespeare thathe did or did not write the poems and plays. As we state in our conclusion, we doubt that evensuch an improbable smoking-gun document would end the controversy.

22. See, e.g., E.K. CHAMBERS, 4 ELIZABETHAN STAGE 9 (1923) (“The theory [thatShakespeare wrote Edward III] has received much support, largely owing to the assent ofTennyson, against whose authority, however, may be set that of Swinburne.”).

23. Mark Twain once compared the Shakespeare record to a much-reconstructedbrontosaur skeleton: “We had nine bones, and we built the rest of him out of plaster of paris.”MARK TWAIN, 1601, AND IS SHAKESPEARE DEAD? 41 (Oxford Univ. Press 1996) (1882, 1909respectively).

24. CHAMBERS, supra note 22. 25. WELLS & TAYLOR, supra note 2, at 450-55.26. See supra text accompanying note 12.27. P. LEVI, THE LIFE AND TIMES OF WILLIAM SHAKESPEARE 345 (1988).

evidence was more important, and each side issued a fatwa downgrading theother. We do not think such fatwas settle much. Some issues are best settledwith documents, others with counting. The normal expectation is that youneed to consider both. External evidence and qualitative judgments have beenthe default for many years, have received most of the serious scholarlyattention, and have normally provided the starting point for quantitative,internal-evidence exploration like ours, not the reverse.21

On the other hand, qualitative, aesthetic judgments tend to be subjectiveand inconclusive and can only take you so far before someone of equalreputation disagrees with you.22 Documentary evidence, where available, canbe more telling. For Shakespeare, however, the documentary map still hasmany blank and gray spots and has not been changed much lately by newdiscoveries.23 Stated differently, possibly because it has been the default forso long, external evidence appears much closer to being mined out thaninternal evidence. Do today’s Shakespeare scholars have a single majordocument bearing on what belongs in the Canon that was not available to E.K.Chambers, who completed his magnum opus in 1923?24 It is easy enough torecall the excitement over “discoveries” such as Shall I Die?,25 A FuneralElegy,26 and the Levi Poem,27 with or without the provisos that they all werereassessments of already-known documents and that they all turned out to bewrong. It is much harder to recall a document discovered since Chambers’stime that has actually changed the Shakespeare authorship map for the better.

Hence, there is a nagging feeling that most of the easy questions thatdocuments could settle conclusively were answered in Chambers’s time, andonly the harder, gray-area questions that the documents cannot settleremain—just enough to fuel either endless reassessment or a general boredomwith authorship among the cognoscenti but never quite enough to fill in theblank spots. Therefore, anyone who wants to make headway has only three


28. Foster recommended the following cuts from the Shakespeare Canon: Macbeth: act3, sc. 5; act 4, sc. 1, lines 39-43, 125-32; (2) Pericles, Prince of Tyre: acts 1-2; (3) Henry VIII:act 1, sc. 3-act 1, sc. 4, line 64; act 1, sc. 4, line 64-act 2, sc. 2; act 3, sc. 1; act 3, sc. 2, lines203-349; act 4, sc. 1; act 4, sc. 2, lines 83-108; act 5, sc. 2, lines 1-182; act 5, sc. 3-act 5, sc. 4;epilogue; (4) Two Noble Kinsmen: act 1, sc. 5; act 2, sc. 2-act 2, sc. 6; act 4, sc. 1-act 4, sc. 2;act 4, sc. 3, lines 1-57; act 5, sc.1, lines 1-33; act 5, sc. 2; epilogue; (5) Timon of Athens: theentire play; (6) Henry VI, Part 1: the entire play.

29. Edward III now appears in several recent Shakespeare anthologies. However, it wasnot in the Canon when we started, and we do not think it belongs in a clean baseline now. TheHand D section of Sir Thomas More, see THE RIVERSIDE SHAKESPEARE, supra note 2, at 1775-94, like The Phoenix and Turtle, is too short to put in a baseline; we also doubt that it isShakespeare’s.

choices: (1) wait and hope that something new and promising will turn up forShakespeare, as it occasionally still does for other authors; (2) try yet anotherreassessment of the same old inconclusive external evidence; or (3) attempt toaccomplish with computers what Galileo and van Leeuwenhoek did with theimproved optics of their time—that is, use them to examine previouslyunobservable internal evidence and substitute observation for what had at thattime been nothing but conjecture. We chose the third course, and it hasproduced some observations that could provide us with a much sharper viewof certain questions that were gray or blank spots before our new optics camealong.

B. Clean, Commonized Baseline

As consumers of Chambers via Donald Foster and The RiversideShakespeare (“Riverside”), we did our best to start with pure Shakespeare,externally and conventionally defined, as our basis of comparison and to editit carefully to modernize and commonize the spelling, strip the text of speechheadings and stage directions, separate prose and verse, and otherwise to makeit fit for computer analysis. With the help of Donald Foster, our initialadvisor, we first purged our baseline of every play or passage he (relyingchiefly on the Riverside) considered suspect.28 Roughly speaking, thiscompletely eliminated three and a half jointly authored plays from baseline:Henry VI, Part I ; Timon of Athens; Two Noble Kinsmen; and Henry VIII. Thisreduced the baseline from thirty-eight to thirty-five plays. Two of these,Shakespeare’s parts of Two Noble Kinsmen and Pericles, were partial.29

Subsequently, several developments compelled us further to cleanse andshorten our baseline list. We gained further confidence in our tests and wefound cases that showed strong internal evidence of co- or other-authorship,notably A Lover’s Complaint; Titus Andronicus; Henry VI, Part III; andportions of co-authored plays that Foster’s authorities had ascribed toShakespeare. With a bit of prompting from modern screenwriters, we lostconfidence in the convenient notion that co-authors always divide their workneatly by act and scene to make it easier for stylometrists to tell who wrote


30. WELLS & TAYLOR, supra note 2.31. The most arguable exception to our list of disputed texts is Henry V. As far as we

know, no one doubts from any external evidence that Shakespeare wrote the English parts. Onthe other hand, no one knows whether someone, perhaps Christopher Mountjoy, the Huguenotlandlord of his lodgings on Silver Street, Cripplegate, helped Shakespeare with the Frenchsections, see ANTHONY BURGESS, SHAKESPEARE 170 (1970), or whether Shakespeare evenwrote them at all. Hence our term “set-aside.” Common sense, consultation with modernscreenwriters, and our own experience preparing Latin orators for Claremont McKenna Collegegraduations all support the conclusion that collaboration is especially likely where foreignlanguages are involved.

Thomas Merriam believes from internal evidence of “intertextual distance” that the versepassages of Henry V differ significantly from the rest of Shakespeare, including the prosepassages of Henry V. He concludes that “[t]his feature reinforces previous doubt cast on theauthorial integrity of the play.” Thomas Merriam, Intertextual Distances Between ShakespearePlays, with Special Reference to Henry V (Verse) 9 J. QUANTITATIVE LINGUISTICS 261, 261(2002). However, he cites no external evidence supporting this view.

32. See Ward E.Y. Elliott and Robert J. Valenza, So Many Hardballs, So Few of ThemOver the Plate, 36 COMPUTERS &HUMAN. 455 (2002) [hereinafter Elliott, So Many Hardballs];

what. We also adopted two new systems for calculating the relative odds ofShakespeare authorship of any given text, neither of which makes much senseabsent a clean baseline. Finally, our baselines get oven-baked after theumpteenth layer of our analysis and cannot easily be unmixed retroactively.All of these factors supported the decision to start with a short, clean baselineand discouraged the use of a long baseline filled with an indiscriminatemixture of doubtful and not-so-doubtful texts.

In every case where our computer told us a Shakespeare-ascribed text wassuspect, we rechecked the suspect passages against conventional scholarship,especially Wells and Taylor’s William Shakespeare: A Textual Companion.30

With one easily distinguishable exception,31 we found that conventionalscholars also held the Shakespeare ascription in doubt, and again we resolvedthe doubts in favor of a cleaner baseline. We moved all of these newly-questionable plays—Henry VI, Parts II & III; Titus Andronicus; Pericles; TwoNoble Kinsmen; and Henry V—to a category we call “Dubitanda and Set-Asides,” further cutting the baseline from thirty-four to the twenty-nine playslisted in Appendices One through Five. The final result is that we kepttwenty-nine plays in the clean, single-authored baseline, and we moved nineto Dubitanda and Set-Asides.

Using a similar process, driven by strong internal evidence, lack of strong,contradictory external evidence, and the softness of the scholarly consensusto the contrary, we also removed A Lover’s Complaint from our Shakespearepoem baseline. We have found no reason to move any claimant or apocryphapoem or play we tested in the opposite direction—into the Shakespearebaseline. In particular, we are not persuaded that Shakespeare wrote any partof Shall I Die?, A Funeral Elegy, or Edward III. People may legitimatelydiffer as to how clean the baseline should be,32 but especially with our new


Donald W. Foster, The Claremont Shakespeare Authorship Clinic: How Severe Are theProblems?, 32 COMPUTERS & HUMAN. 491, 497-501 (1999) [hereinafter Foster, AuthorshipClinic]; Donald W. Foster, A Funeral Elegy: by W[illiam] S[hakespeare]’s “Best-SpeakingWitnesses,” 111 PUBL’NS MOD. LANGUAGE ASS’N AM. 1080, 1082 (1996) [hereinafter Foster,“Best-Speaking Witnesses”]; Ward E.Y. Elliott & Robert J. Valenza, So Much Hardball, SoLittle of It Over the Plate: Conclusions from Our “Debate” with Donald Foster (Oct. 26,2002), at http://govt.claremontmckenna.edu/welliott/hardball.htm.

33. See William Boyle, Smoking Guns and Silver Bullets, SHAKESPEARE OXFORD NEWSL.(Shakespeare Oxford Soc’y, Malden, Mass.), Summer 1997, at 20.

calculations of relative authorship odds, we believe that it is much easier, bothexpositionally and in terms of sound methodology, to start with a truly firm,clean baseline and consider adding problematic texts one at a time, than tostart by baking the problematic texts irretrievably into a contaminated,distorted baseline pie.

C. “Silver-Bullet” Negative Evidence

None of our individual tests are perfect in the way that fingerprint andDNA evidence are considered perfect for having zero false negatives and zerofalse positives. Our tests are more like comparing shoe sizes, blood typing, oreye color than comparing fingerprints. If our tests are defined and measuredproperly, they will show tons of false positives but no more than ounces orpounds of false negatives. We believe that negative evidence normallyoutweighs positive evidence by many orders of magnitude. As noted, fittingthe tiny slipper does not prove you are Cinderella nearly as conclusively as notfitting the tiny slipper proves you are not Cinderella. If you are a size four,you could just as well be a false-positive Little Miss Muffet or Tiny Tim; but,if you are a size ten, your claim to be Cinderella is in trouble. The trouble iscompounded, and the disproof stronger, for every additional profile you do notfit—hat size, height, eye color, or blood type—making it easy to eliminate aCinderella claimant even if uncanny numbers of other measurablefeatures—sex, ring size, hair color, inseam, resting pulse rate, cholesterollevel, or blood pressure—seem to match exactly. Hence, our distinguishingstock-in-trade has been “silver-bullet” negative evidence that tends to disprovecommon authorship by showing differences, rather than “smoking gun”positive evidence used by most other analysts to prove common authorshipwith similarities such as “borrowings” or “echoes.”33

In practice, this means that if one of our tests produces 70% false positivesfrom a given set of non-Shakespeare text blocks, it does not affect our results.Obtaining true rejections for the remaining 30% is more than enough payoffto justify the test for our purposes because we have set our baseline profiles toinsure that false negatives are kept to a reasonable minimum. As a rule, wehave tried to set our profiles to result in a “could be” for at least 95% ofShakespeare (or other) baseline blocks tested. Ninety-five percent is a floor,


34. See Appendix One (Shakespeare): Core Shakespeare Play Baseline DiscriminationSummaries.

35. See Appendix One (Claimants): Claimant Play Discrimination Summaries;Appendix Two: Claimants Versus Shakespeare Baseline, Three-Round Composite Scores.

36. Here are the calculations, aimed at the modal, math-challenged layperson. For“Discrete” odds, Shakespeare’s most distant core-baseline outlier is The Tempest, with two“Discrete” rejections in forty-eight tests. See Appendix One (Shakespeare); Appendix Two:Shakespeare Play Baseline Data, Three-Round Composite Scores. The odds that any test blockwould get two “Discrete” rejections in forty-eight tests, given a 1.9% error rate among all tests,are about 23%, or, in scientific notation, 2.316 times 10-1. The nearest “Claimant” match tobaseline Shakespeare is Marlowe’s Edward II, which gets ten “Discrete” rejections in forty-eight tests. The odds of this, with the same 1.9% Shakespeare error rate, are 0.00000001337or 1.337 times 10-8. The value of 0.2316 divided by 0.00000001337 equals 17,322,364, whichmeans that Edward II is 17.3 million times less likely to have come from Shakespeare’s pen bychance than The Tempest.

For “Continuous” odds, The Tempest is also Shakespeare’s most distant core baseline outlier,with a Continuous Composite Shakespeare Probability of 3.689 times 10-3, or 0.0037. Theclosest “Claimant” match by this measure is Thomas Heywood’s A Woman Killed WithKindness, with a Continuous Composite Shakespeare Probability of 1.6337 times 10-6, or0.000006337. The value of 0.0037 divided by 0.000006337 equals 2,258. This means that AWoman Killed With Kindness is 2,258 times less likely to have come from Shakespeare’s penby chance than The Tempest. For further details on how we calculate “Continuous” odds of

not an average. When we count every test run, on every core baselineShakespeare play, we find 98% true positives and only about 2% false-negative Shakespeare rejections. Applying the same process to fifty-one playsmostly by claimants produces 65% false-positive results. We ignore this resultbecause false-positive results do less to prove a Shakespeare ascription thanfalse-negative results disprove it, and 35% true negatives, with a bit ofaggregation, are more than enough to rule out all fifty-one plays asShakespeare’s solo work.

Our first aggregation is to count our results play by play, rather than testby test. When we do this, we find that no core Shakespeare baseline play hasmore than two individual rejections (shaded aqua) in forty-eight tests,34 whileno “Claimant” play has fewer than ten rejections.35 Our second aggregationis to use the two methods discussed under “Comparative Odds” below toestimate the composite odds, given what we now know about Shakespeare’scountable traits, that he could have written by chance a play like Marlowe’sEdward II that has only ten rejections. It is important to note here that,whenever we say “written by chance” in lay language, we are actually talkingabout something more technically qualified for numerate readers. We refer tothe odds that the specific features for which we test could have arisen bychance assuming the statistics and modeling that we have imputed to thebaseline. The odds for the least improbable Claimant play are somewherebetween 2,258 and 17.3 million times worse than the odds that Shakespearecould have written the farthest outlier in his own core baseline.36


common authorship, see infra Part II.E.37. Or, as R.W. Chambers put it, “‘You have to meet in a crowd a Mr. Harris, hitherto

unknown to you, but who, you are informed, has red hair, wears a monocle, and walks with alimp.’” MACD. P. JACKSON, DEFINING SHAKESPEARE: PERICLES AS TEST CASE 192 (2003)(quoting R.W. Chambers, The Expression of Ideas—Particularly Political Ideas—in the ThreePages, and in Shakespeare, in SHAKESPEARE’S HAND IN THE PLAY OF SIR THOMAS MORE 142(1923)). In 2003, MacDonald Jackson offered this piercing critique of Chambers’s argument:

We might even reckon probabilities—one in ten men is left-handed, one in eight wears amoustache, and so on—and enlarge the list to the point where multiplying the separateodds would produce a billion-to-one coincidence. Harris must have met his doppelgänger!No, the passer-by is a stout Caucasian and Harris is a slim West Indian. The total absenceof constraints on our search for resemblances renders the calculation meaningless.

Id.38. See Saks, supra note 15, at 1170 (suggesting that fingerprint evidence does not meet

Daubert standards and noting that “the field of fingerprint identification has thus far failed tosystematically test its underlying assumptions and its claims of expertise”).

39. See e.g., DODD, supra note 2; WILLIAM P. FOWLER, SHAKESPEARE REVEALED IN

We are aware that such aggregation, which can make non-Shakespearelook decisively unlike Shakespeare, might arguably be invoked in reverse toshow that a given Shakespeare play or poem looks decisively likeShakespeare. For instance, with a bit of aggregation, you could find enough“unique similarities” to light up a Christmas tree and thereby proveShakespeare authorship almost as conclusively as you could disprove it. Forexample, what if you find someone who has a shoe size and twelve othermeasurable features that all fall within Shakespeare’s ranges, and he andShakespeare are the only ones found who fit all of those profiles? Would thatnot be about five coincidences too many for it not to be Shakespeare? Wouldit not almost mean that he has to be Shakespeare? And if you could say thatabout aggregating the physical particulars of Shakespeare the man, could younot also say the same about aggregating the stylistic particulars of ahypothetical newly discovered Shakespeare play, such as a manuscript inShakespeare’s handwriting bearing over 18,000 supposedly uncannyresemblances to Shakespeare’s and Fletcher’s lost play, Cardenio? Would thatnot reveal that it was his?37

The short answer is that it depends on how far you looked before youdecided that Shakespeare was the only multiple-fit to be found. If you havelooked at millions of fingerprints or DNA samples and found no two alike, youcan talk about perfect, positive evidence; however, there are those who thinkthat even fingerprints are open to question because the science behind themhas been less thoroughly examined in court than DNA testing.38 But we haveseen too many overreaching claims of exhaustive due diligence that wiltedwhen the sun came out, and too few that have not, for us to be comfortablewith such claims.

Examples of such “Christmas trees full of unique quirks equals proof”claims abound in anti-Stratfordian tracts,39 but they are also the principal


OXFORD’S LETTERS (1986); HOFFMAN, supra note 2; OGBURN, supra note 2. But see May,supra note 2, at 223; Carol Boettger, That Way Madness Lies: Elegy Conference in LA StillLeaves Questions, SHAKESPEARE OXFORD NEWSL. (Shakespeare Oxford Soc’y, Malden, Mass.),Summer 1996, at 2.

40. See DONALD W. FOSTER, ELEGY BY W.S.: A STUDY IN ATTRIBUTION (1989)[hereinafter FOSTER, ELEGY BY W.S.]; Foster, “Best Speaking Witnesses,” supra note 32;Posting of Donald Foster, [email protected], to [email protected] (June 13, 2002), athttp://www.shaksper.net/archives/2002/1484.html (on file with the Tennessee Law Review)[hereinafter Posting of Donald Foster].

41. BRIAN VICKERS, ‘COUNTERFEITING’ SHAKESPEARE: EVIDENCE, AUTHORSHIP, AND

JOHN FORD’S FUNERALL ELEGYE (2002).42. See CHARLES HAMILTON, WILLIAM SHAKESPEARE AND JOHN FLETCHER: CARDENIO

OR THE SECOND MAIDEN’S TRAGEDY (1994).43. Id. at 139.

evidence in more respectable tracts, perhaps because they do not require thevast exercises of building, blocking, and profiling baselines that we favor andcan only be done conveniently with computers. For example, consider DonaldFoster’s now-abandoned claims that hendiadys, incongruous who’s, andredundant comparatives and superlatives found in A Funeral Elegy were like“thumbprints” of Shakespeare. Foster also claimed that A Funeral Elegy fitShakespeare’s rare-word peaking patterns identified by Shaxicon so perfectlythat it “cannot have been written by anyone other than Shakespeare.”40 Theresemblances between A Funeral Elegy and Shakespeare were many,remarkable, and often real, and they seemed unique at first glance. But itturned out that they were not, and the smoke from the supposed smoking-gun“proofs” vanished when Brian Vickers’s mighty ‘Counterfeiting’Shakespeare41 loomed on the horizon and the ascription itself collapsed.

The claimed Cardenio, as it happens, is not just a hypothetical. CharlesHamilton, a respected handwriting analyst, found what appeared to beShakespeare’s handwriting and dozens of other Shakespeare “echoes” and“borrowings” in the manuscript to The Second Maiden’s Tragedy andpronounced it the lost play.42 Consider, for example, the remarkableresemblance in Figure 2.1 between the top example of each pair, whichHamilton took from Shakespeare’s will, and the bottom example, taken fromThe Second Maiden’s Tragedy manuscript.43


44. Figure 2.1 is adopted from HAMILTON, supra note 42, at 139. From Cardenio, or TheSecond Maiden’s Tragedy, by Charles Hamilton, with permission of Glenbridge Publishing Ltd.

45. See generally FREDERICK SCHAUER, PROFILES, PROBABILITIES, AND STEREOTYPES

ix (2003) (defending “the morality of decisions by categories and by generalizations”).

Figure 2.1. Purportedly Uncanny Resemblances Can Deceive44

Figure 2.1. The manuscript to The Second Maiden’s Tragedy has strikingly similarhandwriting to that found in Shakespeare’s will, but Shakespeare did not write TheSecond Maiden’s Tragedy.

Despite its many remarkable Shakespeare resemblances, The SecondMaiden’s Tragedy is anything but a likely Shakespeare play. It receivedtwenty-two rejections in forty-eight tests, one of the worst scores of any play.The likelihood of Shakespeare authorship, by both of our comparative oddsreckonings, was too low to compute with standard-issue, double-precisiondesktop computer software.

Hence, we avoid the use of terms like “fingerprints” or “thumbprints” todescribe our work because their implications of conclusive, positive proof aremisleading when applied to imperfect tests—the only kind we have seen forShakespeare. Imperfect tests remain much better at disproving than proving.

D. Block and Profile: Sample Length Matters

Profiles are what we use to disprove common authorship. You cannotmake a valid exclusion unless you have valid standards of inclusion. Profilesimply such standards, and lawyers, in particular, spend their lives arguing overwhich ones to apply and with what degree of consistency and rigor.45

Quantified stylometric profiles make little sense unless they are wellcontrolled for text length because of the law of large numbers: longer texts aremore orderly and predictable because they average out more variance. For


46. See Figure 2.2.47. See Appendices One through Nine.

instance, no one can predict what an individual or small group of people willeat on a given night, but in order to anticipate how much food and how manyemployees they will need to meet the demand, restaurants can predict whatlarge numbers of people on large numbers of nights will eat. It is the same forgrade-level, hyphenated compound words, new words, and other stylisticindicators. Patterns that are hidden in small numbers become more and moremanifest as the numbers get larger. If a population of test outcomes forms anormal distribution curve, or bell curve, the curve should be tighter and moresymmetrical for a larger sample than for a smaller one and, as a result, be moreuseful for distinguishing the measured trait.46

Figure 2.2. Longer Blocks, Narrower Ranges

Figure 2.2. Shakespeare Poem Grade Level Profiles for 3,000- and 750-Word Text

0

2

4

6

8

10

12

14

Frequency

7 8 9 10 11 12 13 14 15 16 17 18 19 20 21Grade Level

Shakespeare Grade Level: 3,000- vs 750-word Blocks

3,000-word blocks

750-word blocks

Blocks. Three thousand-word text blocks (light bars) yield less sprawling, moreuseful profiles than shorter, 750-word ones (dark bars).

This means that we received our clearest and best results with our longesttext blocks, and it explains why we have organized our data47 by block lengthfirst and by genre second. We were able to validate forty-eight tests with play-length text blocks of 15,000 to 30,000 words. For text blocks of around 3,000


48. See Appendix Six for fourteen of these tests.49. See Appendix Nine.

words, we could only validate fifteen tests.48 For text blocks of 470 words, wecould validate only eleven tests.49 Close examination of the ranges for thesame test, such as grade level, shows that the ranges become broader and lessexclusive with each diminution in block size. Hence, we routinely block textby size before profiling, referring to the processes as “block and profile.”Table 2.1 summarizes the ranges for our most important small-block tests,sorted by block length and genre.

Table 2.1. Profile Ranges for Selected Tests,Sorted by Block Length and Genre

who

lep l

a y

3 00 0

wo r

dp o

e m

3 00 0

wo r

dp l

a yve

rse

1 50 0

wo r

dp o

e m

1 50 0

wo r

dp l

a yve

rse

7 50

wor

dpo

em

7 50

wor

dpl

ayv e

rse

4 70

wor

dpo

em

grade levelmin 4 10 3 10 4 8 3 8

max 7 14 8 16 9 16 10 18

HCW/20kmin 52 31 31 24 24 24 26 0

max 180 153 153 243 243 268 236 240

Rel. Cl.min 7

max 17

Fem. End.min 8 7 7 3 3 3 3 3

max 22 25 25 29 29 28 28 40

FE earlymin 8

max 17

FE middlemin 8

max 20

FE latemin 17

max 22

Open linesmin 11 9 9 7 8 6 6 7

max 50 57 57 24 55 51 51 28

OL earlymin 11 9 9 7 8 6 6

max 23 33 33 24 33 32 32

OL middlemin 16

max 32

OL latemin 31 15 15 13 13 12 12

max 50 57 57 23 55 51 51


50. RICHARD. J. HERRNSTEIN & CHARLES MURRAY, THE BELL CURVE: INTELLIGENCE

AND CLASS STRUCTURE IN AMERICAN LIFE 46 (1994).

Encliticsmin 27 27 18 18 10 10 17

max 89 89 123 123 137 137 196

Procliticsmin 265 265 235 235 152 152 183

max 476 476 561 561 510 505 589

with (2lws)min 9 4 4

max 21 36 36

no/no+notmin 242 167 167 100 0

max 358 536 586 667 800

BOB 5min 159 116 116 93 93 59 63 55

max 487 556 556 761 761 750 712 805

BOB 7

min 278 136 136 0 0 -146 -146 0

max 779 944 944 1000 1000 1000 1000 1000

BOB 8min -867 -867 -889 -889 -929 -929 -1000

max -265 -265 -209 -209 -142 -83 -167

TE slopemin -0.13 -0.22 -0.22 -0.22 -0.22

max 0.06 0.15 0.15 0.15 0.15

TE NWmin -14 -32 -32 -24 -24

max 5 21 21 12 12

TE RWmin -2 -33 -40

max 89 218 116

Bucketmin -2 -72 -77 -69

max 2 29 100 81

Modalmin 281 79 -11 -35

max 1149 407 189 154

Discretemax

2.9

E-01

3.1

E-01

2.9

E-01

2.7

E-01

2.5

E-01

3.6

E-01

3.4

E-01

4.4

E-01

Continuousmax

8.9

E-03

9.0

E-02

1.8

E-01

1.2

E-01

3.8

E-01

6.2

E-02

1.2

E-01

3.2

E-01

Table 2.1. Profile ranges tend to widen as block size decreases.

E. Finding the Comparative Odds of Shakespeare Authorship

After blocking, testing, and profiling, as Figure 2.2 shows, this processyields a set of ranges more or less resembling bell curves. Figure 2.3 showsthree overlapping, idealized bell curve examples.50 The curves represent therelative number of college graduates and non-college graduates plotted againsttheir corresponding standard deviations from the IQ mean of the wholepopulation. A normal population is distributed as follows: about two thirdsof the population fall within one standard deviation from the population’saverage, or mean; 95% fall within two standard deviations from the mean;


51. Id. at 44. 52. Figure 2.3 is adopted from HERRNSTEIN & MURRAY, supra note 50, at 46. Reprinted

with permission of The Free Press, a Division of Simon & Schuster Adult Publishing Group,from The Bell Curve: Intelligence and Class Structure in American Life by Richard J.Herrnstein and Charles Murray. Copyright © 1994 by Richard J. Herrnstein and CharlesMurray.

53. Id. at 46. 54. Figure 2.4 is adopted from Ward E.Y. Elliott & Robert J. Valenza, Smoking Guns and

Silver Bullets: Could John Ford Have Written the Funeral Elegy?, 16 LITERARY & LINGUISTIC

COMPUTING 205, 208 (2001) [hereinafter Elliott & Valenza, Smoking Guns]; see also WARD

99.7% fall within three standard deviations from the mean; and so on.51

Figure 2.3. Some Perfect Bell Curves52

Figure 2.3. Almost all college graduates have above-average IQs. Graduates of thetop dozen universities in 1990 had IQs almost three standard deviations aboveaverage. Conversely, very few people with an IQ more than one standard deviationbelow average graduated from college.53

Figure 2.3 shows that having an IQ one standard deviation above average,or even two, is no guarantee that you have graduated from college. Likefitting Cinderella’s shoe, such positive evidence of a higher IQ is inconclusiveas to graduation status. But if your IQ is one standard deviation belowaverage, the odds that you have graduated from college are very low, and theodds are lower yet if your IQ is two standard deviations below average. Inother words, a low IQ is much more closely associated with not havinggraduated from college than a high IQ is with having graduated from college.

Figure 2.4 shows two actual profiles of of’s per 4,300-word block.54


E.Y. ELLIOTT & ROBERT J. VALENZA, SHAKESPEARE BY THE NUMBERS (forthcoming). Thisbook in progress will present our findings, not just on the Earl of Oxford, but on all the otherclaimants and most of the disputed poems and plays of the Shakespeare Apocrypha andDubitanda.

55. Elliott & Valenza, Smoking Guns, supra note 54, at 207-08.

Shakespeare’s combined range from all ten blocks of his poems, and nineteenmore from his late play verse, is 46 to 108 of’s per block; Ford’s range for fivepoem blocks is 143 to 220. The frequency of of’s per 4,300-word block foreach author is less than perfectly symmetrical, but more importantly, they donot overlap at all. On average, Ford uses twice as many of’s per block asShakespeare. A Funeral Elegy, with 145 of’s, falls far outside ofShakespeare’s range, but just within Ford’s. Common sense and a host ofother such indicators suggest that it is much more likely to be Ford’s thanShakespeare’s.55

Figure 2.4. Shakespeare’s and John Ford’s Profiles of of’sper 4,300-Word Block

0

2

4

6

8

10

12

14

Frequency

20 40 60 80 100 120 140 160 180 200 220 240Count per 4,300-word Block (bin width = 20)

Shakespeare vs Ford: 'of' Ranges

Shakespeare

Ford

Figure 2.4. Shakespeare’s and Ford’s of ranges of 46 to 108 (darker bars) and 143to 220 (lighter bars), respectively, do not overlap. A Funeral Elegy, with 145 of’sseems much less likely to be Shakespeare’s than to be Ford’s.


56. Comparisons given here are between the original-punctuation Oxford poems, seeMay, supra note 9, and the modernized Riverside, supra note 2, used in our baseline. If we hadcompared with original-punctuation Shakespeare, whose grade-level scores are, on average, agrade higher than the Riverside, the Shakespeare baseline curve would have moved one gradelevel to the right, and the differences from Oxford would have been even more striking.

Figure 2.5, comparing all of the Earl of Oxford’s poems with all ofShakespeare’s poems by grade level, shows a similar result. Oxford’s poemsare written at the seventh-grade level and fall completely outsideShakespeare’s tenth to fourteenth-grade level range.

Figure 2.5. Shakespeare’s and the Earl of Oxford’s Profilesfor Grade Level: 3,000-Word Blocks

0

1

2

3

4

5

6

Frequency

5 6 7 8 9 10 11 12 13 14 15 16Grade Level

Shakespeare vs Oxford: Grade Level (3,000-word blocks)

Shakespeare

Oxford

Figure 2.5. Almost half of Shakespeare’s fourteen 3,000-word poem blocks werewritten at the twelfth-grade level (light bars). All of Shakespeare’s blocks werewritten between the tenth- and fourteenth-grade levels. The blocks show a standarddeviation of about one grade level with no block more distant than two standarddeviations from Shakespeare’s mean grade level. Oxford’s poems (dark bars) test atthe seventh-grade level, five grades and five standard deviations below Shakespeare’smean.56

What are the odds that poems like Oxford’s, which are five standard


57. See Appendix Six: Shakespeare Play Verse Baseline Data, Blocksize = 3,000. Ineighty-two 3,000-word blocks of Shakespeare play verse, we found thirty-five Shakespearerejections in 1,150 separate test runs, a 3% false-negative rate. With individual tests at a 3%false-negative rate, the theoretical odds of getting more than one rejection in the thirteen orfifteen tests we used are 6-7%. Four blocks, 5% of blocks tested, had actual occurrences ofmore than one rejection. Although we rejected slightly fewer blocks than expected, two blockshad an unexpectedly high number of rejections, three and four. All four blocks are red-shadedas composite rejections.

deviations removed from Shakespeare’s mean grade level, could have been soby chance? Before 2003 we classified each test as simply either a Shakespearerejection or non-rejection and then aggregated and counted all of therejections. We supposed, roughly but plausibly, that a poem block likeOxford’s with six or seven Shakespeare rejections is much less likely to beShakespeare’s than one with only a single rejection. Since then we haveworked out two new methodologies, one called “Discrete,” the other“Continuous.” Together, they have added a new layer of precision to ouranalysis.

F. Discrete Analysis

“Discrete” is Elliott’s favorite. “Discrete Composite Probability,” avariant of simple, discrete rejection counts, involves setting up Shakespeareprofiles, rejecting anything outside of their boundaries, and then counting therejections. Our working motto is, “If the shoe don’t fit, you must give it arejection.” Because every profile is set to result in a “could-be” for at least95% of Shakespeare blocks, that is, yield no more than 5% false Shakespearenegatives, the theoretical odds of a given Shakespeare block with one testrejection being a false Shakespeare negative should be less than 5%. Thetheoretical odds of a Shakespeare block receiving one rejection in fourteentests, each with 5% false negatives, are much higher, closer to 50%.

On the other hand, many of our tests in practice have far fewer than 5%false Shakespeare negatives and result in a “could-be” for 99-100% ofShakespeare blocks tested. Appendix Six shows that the actual rejection totalin 196 separate runs of fourteen individual tests on fourteen 3,000-wordShakespeare poem blocks was only two. In essence, the baseline produces 1%false negatives, not 5%. Also, only 14% of Shakespeare blocks had onerejection in fourteen tests, not 50%. None of these Shakespeare blocks hadmore than one rejection on individual tests (marked in aqua in Appendix Six).None had a composite rejection, which would have been highlighted in red orgray in the yellow composite-score columns. The same process on eighty-two3,000-word blocks of Shakespeare play verse produced a 3% rejection rate onindividual tests, with 5% of the blocks receiving false-negative compositerejections.57

The same process, as applied to others’ poems and play verse, in 3,000-


58. See id.59. See id.60. Most of our readers, we surmise, will be horrified to find that all of the composite

figures, except raw rejections, are in scientific notation. We use scientific notation, not justbecause Valenza likes it that way, but because Elliott, the more math-challenged of the twoauthors, has finally admitted that, whenever you are dealing with lots of zeroes, it is muchneater, less confusing, and easier on copy editors to use scientific notation than to try to countout all those zeroes, one by one, with the point of a pencil and your lips moving. Our rule ofthumb for math-challenged readers is the following: a number in scientific notation, forexample 3 times 10-14, can be written as a decimal point, thirteen zeroes, and then the three or.00000000000003. To use scientific notation on a $14.95 entry-level scientific calculator, dothe following: (1) enter the number; (2) enter the exponent with the EE key, which results in

word blocks, yields much higher rejection rates for individual tests (shaded inaqua): 34% for others’ poems58 and 30% for others’ play verse.59 Only fourof ninety-six Shakespeare poem and play verse blocks display a darker-shadedcomposite rejection (gray or red) for having more than one individual rejection(aqua); the baseline contains 4% false Shakespeare negatives. Only three ofthe others’ poem and play verse blocks received as few as one rejection (alsogray or red).

Note that darker shading does not mean “rejection” but something moreakin to “anomalous outlier.” For our Shakespeare baseline, rejections areanomalous false negatives and get shaded. For everything else, includingdubitanda, non-rejections are anomalous false positives and get shaded.Hence, the total of composite anomalies for our 3,000-word Shakespeare poemand play verse baseline is 4%—four dark-spot false negatives in ninety-sixblocks. For all such verse by others, we get 2% anomalies—three dark-spotfalse positives in 125 blocks. For counting rejections, the dark spots are gray,not red, to avoid two kinds of confusion. First, we do not want to double-count with Discrete Composite Probability, which is nothing but a bettervariant of the same test. Second, we want to avoid the inviting, but not quiteaccurate, notion that the odds of getting two rejections from twelve tests arethe same as getting two rejections from fourteen tests. The odds are actuallya little lower. The individual test rejection odds are the same, and the grayshading serves as a reminder that because there are fewer tests, there are feweropportunities to get a rejection.

We get from the “Discrete Rejections” column to the “Discrete CompositeProbability” column of Appendix Six by starting with a blanket averageindividual-test rejection rate of 2.6% for all of Shakespeare’s verse blocks of3,000 and 1,500 words (the actual rejection rates are 2.7% and 2.4%,respectively). Assuming that rejections occur randomly at a fixed rate, onecan calculate the precise rejection odds from a given number of tests at thegiven average rejection rate. The precise rejection odds are the numbers in thecomposite column, “Discrete Composite Probability.” To simplify theprocess, we have prepared a user-adaptable spreadsheet, available from usupon request.60


something like 3 x 1014, or 3E14 as Valenza puts it; (3) press the +/- key to change the exponentfrom positive to negative; (4) 3 x 1014becomes 3 x 10-14; and (5) now you are ready to start yourcalculation. Press INV and EE buttons to reconvert from scientific to normal notation and toreturn to the tedious and often inaccurate struggle to count those otherwise gratifyingly longstrings of zeroes. Scientific notation makes it easier, with minimal preparation, to skim throughour columns of rejection odds without stopping to count zeroes. You can recognizeimmediately which ones are the easy rejection calls, the ones with probabilities of any numbertimes 10-3 (that is, E-03) or lower. The ones that say E-01 or E-02 are not. The exponent isusually much more important for such quick scanning than the number it modifies.

G. Continuous Analysis

The other composite odds method, “Continuous,” is Valenza’s favorite.The Continuous method uses aggregated distances from Shakespeare’s meansas opposed to the Discrete method which uses profile boundaries. Thismethod (1) aggregates every Shakespeare mean on every test into a multi-dimensional composite mean; (2) then measures a given text block’s distance,in standard deviations, from Shakespeare’s mean on every test; and (3) thenaggregates the “Shakespeare distance” on every test into a “ContinuousComposite Error” (“CCE”) score. A high CCE score indicates that thecomposite distance from Shakespeare is great. A low CCE score indicates thatit is small. A “Continuous Composite Probability” (“CCP”) score is calculatedfrom CCE scores. The higher the CCE, the lower the CCP. The CCP of asample play or poem block, though not useful as an absolute indicator of theodds that Shakespeare wrote the block, can nonetheless be compared to theCCP of a Shakespeare far outlier. Such comparisons can be very telling.Appendices One through Seven provide a complete list of CCPs calculated forevery Shakespeare baseline and comparison play, and for selected 3,000-wordand 1,500-word poem and play verse blocks by Shakespeare and others.

The virtue of Continuous analysis, with one sizeable exception, is that itdoes not throw information out. The sizeable exception is that Continuousdoes not account for time periods for traits like line endings, whereShakespeare’s style changed over the years. Discrete analysis distinguishesbetween early and late profiles; Continuous does not. In general, Continuousmeasures not only whether a sample text falls inside or outside a givenShakespeare range, but also how far it falls from the sample’s compositemean. In principle, Continuous should outperform Discrete at describing andpredicting Shakespeare’s patterns in the same way one forecaster with the fullresources of the National Weather Service should outperform one with nothingbut a K-Mart home weather station. Moreover, unlike Discrete, Continuousis in no way dependent on human judgment for making boundaries. In theOxford versus Shakespeare grade-level example in Figure 2.5, Oxford receivesa Discrete rejection for being over two standard deviations of distance fromShakespeare’s mean. Oxford’s poems, however, are actually five standarddeviations from Shakespeare’s mean—or six, if the comparison were toShakespeare in original punctuation. A Funeral Elegy, written at the twenty-


61. The Continuous method also registers “hyper-rejections” in ways that countingDiscrete rejections misses. For example, in Figure 2.4 above, what if A Funeral Elegy had had230 of’s, instead of its actual 145, twice Shakespeare’s maximum, but only 5% over Ford’s?Unadjusted, mechanical rejection counting would register it as both a Ford and a Shakespearerejection, spuriously evening the rejection score at one-one. The Discrete method misses thecrucial fact that the already gross rejection for Shakespeare is even grosser, while the rejectionfor Ford is narrow.

For both the Discrete and Continuous tests, we compute composite threshold probabilitiesthat a given block could be Shakespeare’s (Appendices One through Nine), for exampleShakespeare’s 3.1 E-01 Discrete Composite Probability in Table 2.4. Such “boundary block”thresholds help us to divide Shakespeare from non-Shakespeare. Other things equal, when atext sample of unknown authorship scores appreciably below Shakespeare’s boundarycomposite probability threshold, our best guess is that it is not Shakespeare’s. The bigger thedifference the better the guess. Thresholds also furnish a basis for a figure of merit reflectingon how well our tests separate Shakespeare from others for known samples. See Table 2.5.When we declare that our tests have a perfect score (100%) in net discrimination, we mean thatthe chosen threshold separates known Shakespeare from known non-Shakespeare perfectly,with no false negatives or false positives. If the discrimination figure was zero, as it probablywould be for very short samples, it would mean that our outcomes are no better than randomguesses.

62. Valenza would have used a multi-dimensional, sun-centered solar system to illustratethis point.

second-grade level, is ten standard deviations from Shakespeare in grade-levelscores—or nine, if we compared it to original-punctuation Shakespeare. Thesescores are not just out of the Shakespeare ballpark; in terms of the probabilitythat Shakespeare wrote works at these grade levels, they are in a different stateor on a different continent. Continuous scoring, with room to spare, takesaccount of such differences between gross and narrow rejections as well as thedifferences between narrow non-rejections and “firm” ones.61

It is important to stress again that composite probability scores, whetherfrom Discrete or Continuous analysis, are not indicators of the absolute, actualprobability that Shakespeare wrote the block in question. Compositeprobability scores are markers from which composite Shakespeare ranges arederived. The scores permit comparison of the block in question, not with theclosest theoretically imaginable Shakespeare match, but with an actualShakespeare block at the edge of his range.

A handy, comfortably geocentric way to visualize the idea of comparativeprobability distance is to start with a ballpark and work out.62 All ofShakespeare’s core-baseline 3,000-word poem blocks and plays, and 95% ofhis 3,000-word play verse blocks, would fit into the ballpark, most of them inthe infield. Ninety-five to ninety-eight percent of shorter Shakespeare verseblocks would also fit into the ballpark, most of them in the infield. For everyorder of magnitude difference from Shakespeare’s profile-boundary block forpoems or play verse, the Shakespeare distance would increase a notch:ballpark, city, county, state, country, continent, planet, moon, outer planets,solar system, galaxy, and another galaxy. If our tests are reliable, anything


inside the ballpark would be a Shakespeare could-be. Anything within the citylimits would get a relatively narrow, tentative rejection because our tests canvary somewhat in reliability and because we know that up to 5% of our coreShakespeare baseline blocks are false negatives which fall outside the ballparkbut still inside the city. For each extra notch of distance past the city limits,the rejection’s narrowness and the tentativeness are drastically reduced, andthe composite probabilities eventually fall so low that no amount of tinkeringor rationalization can avoid the conclusion that Shakespeare did not author thetested block.

H. Some Examples

To recapitulate, illustrate, and complete this process, we now turn tosomething we hope will be more accessible than the comprehensive, jam-packed, high-resolution, small-printed, scientifically-notated, Valenza-gradetables provided in the Appendices. These Appendices are appropriatelydesigned to present as much interrelated information as we can squeeze ontoa page and to allow an interested mathematician to follow our every step.Passing muster with mathematicians is the bedrock of our case, and oursubstantial Appendices reflect it. On the other hand, digestibility for a not-so-mathematical audience remains a high priority for us because we expect theseto be ninety-nine percent of our readers. For these, we now offer a less tightlypacked series of examples with fewer decimal places, larger type, and lessfumbling to consult the Appendices.

Table 2.2 is a cut-down, large-type, Elliott-grade version of Appendix Six:Shakespeare Poems Baseline Data, Blocksize = 3,000. Much of theinformation found in Appendix Six is omitted, and this table shows only sixof the fourteen tests we used and only six of the fourteen 3,000-wordShakespeare blocks we tested. But the visual impression is the same as inAppendix Six. This table includes a great deal of white space for non-rejections and only one dark space for a rejection, a fit portrayal ofShakespeare’s 1% individual rejection rate in this category.


Table 2.2. Seven Tests on Six 3,000-Word Shakespeare Poem Blocks

GradeLevel

HCW/20K

Fem.Endings

(%C)

OpenLines(%C)

Enclitics/1000 lines

Proclitics/1000lines

with (2lws)

10-14 31-153 7-25 9-57 27-89 265-476 4-36

14 98 12 19 77 334 1513 68 8 15 61 367 2213 88 3 18 43 316 712 50 8 15 48 321 712 56 12 19 87 360 712 104 7 17 81 476 12

Table 2.2. Highlights of Appendix Six: Shakespeare Poems Baseline Data,Blocksize = 3,000 with a 1% rejection rate (one example shaded here).

Table 2.3 shows the same seven tests on four blocks by Oxford, Bacon,Marlowe, and A Funeral Elegy. Many more dark spots meet the eye.

Table 2.3. Seven Tests on Five 3,000-Word Non-Shakespeare PoemBlocks

Author,Block

GradeLevel

HCW/20K

FemEndings

(%C)

OpenLines(%C)

Enclitics/1000lines

Proclitics/1000lines

with(2lws)

Sh. Range 10-14 31-153 7-25 9-57 27-89 265-486 4-36

Oxford 7 32 0 7 13 115 5

Bacon 12 21 2 6 18 149 11

Marlowe 1 9 42 8 18 51 298 16

Marlowe 2 9 13 9 19 30 206 7

Elegy 22 101 12 46 24 211 8

Table 2.3. Highlights of Appendix Six: Other Poets Versus Shakespeare Baseline,Blocksize = 3,000 with a 35% rejection rate (examples shaded).


63. That is, Discrete composite testing yields the following result: 3.1E-01 divided by5.2E-03=5.96E01, or 59.6 times less probable. For Continuous composite testing, the result is9.0E-02 divided by 1.6E-02=5.62E00, or 5.62 times less probable.

Table 2.4. Shakespeare’s Farthest Outlier 3,000-Word Poem BlocksCompared with Oxford, Marlowe, and A Funeral Elegy

Author Tests RejectionsDiscrete

CompositeProbability

ContinuousCompositeProbability

Shakespeare 14 1 3.1E-01 9.0E-02

Oxford 14 6 7.7E-07 <1.0E-15

Bacon 14 7 2.3E-08 <1.0E-15

Marlowe 1 14 3 5.2E-03 1.6E-02

Marlowe 2 14 4 3.7E-04 3.1E-04

Elegy 14 6 7.7E-07 <1.0E-15

Table 2.4. Blocks from the three leading claimants and A Funeral Elegy are all farless likely than Shakespeare’s own farthest outlier blocks to have come fromShakespeare by chance. Shakespeare authorship of the closest block, Marlowe One,is five to sixty times less probable than Shakespeare’s own outliers. Shakespeareauthorship of the most distant block, Bacon, is between 13 million and 90 trilliontimes less probable than Shakespeare’s own outliers.

Table 2.4, adapted from Appendix Six, identifies Shakespeare’s outlierblocks on both Discrete and Continuous composite testing. For Discrete, theoutlier is the first block of Venus and Adonis, with a rounded score of 3.1E-01=.31. For Composite, it is the second block of Venus and Adonis with arounded score of 9.0E-02=.09. The closest other block in the table is MarloweOne, which is between five and sixty times less probable than eitherShakespeare outlier by rounded calculation.63 Marlowe One is not very closeto Shakespeare—maybe in the same city or county, but this block is muchcloser to Shakespeare than Oxford’s or Bacon’s poems or A Funeral Elegy.These three are not even in the same galaxy.

In passing, it is important to note that although one Marlowe block ismuch closer to Shakespeare than Bacon or Oxford, Marlowe is not a credibleShakespeare match. Marlowe One is in the same city or county withShakespeare, not the same ballpark; Marlowe Two is in the same county orstate. The closest of seven Marlowe plays might be on the same continent or


64. See Appendix One (Claimants); Appendix Two: Claimants Versus Shakespeare’sBaseline, Three-Round Composite Scores.

planet; the other six are in different galaxies, some too distant to compute.64

Taken individually, most of Marlowe’s works are indeed very distant fromShakespeare. When Marlowe’s works are taken as a group, however, the oddsthat Shakespeare could have strayed by chance so persistently and soextravagantly far from his customary boundaries are far lower than the oddsof getting hit by lightning or winning the Irish Sweepstakes. Fortunately, wehave more than 100,000 words of Marlowe’s writing to compare withShakespeare. The law of large numbers kicks in much more conclusively forhim than for Oxford or Bacon, who each have about 3,000 words of poems totest. It buries Marlowe’s claim in an ocean of rejections, and despite hishaving one arguable near-miss, leaves him on a more distant galaxy thanOxford or Bacon.

We are aware that stylometry itself is a novel, ill-mapped field where allthe explorers are on the steep part of the learning curve. Not everyone gets itright on the first or even the tenth try, and many extravagant claims ofcertitude have been made by others who later came to grief. We are alsoaware that our target audience of Shakespeare lovers has more than its shareof numbers-skeptics. Although our latest odds-calculators are new and havenot been extensively reviewed, they seem to promise many zeros worth ofextra certitude. These facts seem like fuel for a bonfire of skepticism. Willour claims also come to grief? We doubt it because our findings are muchbetter validated than most. Nevertheless, the situation calls for some furtherwords regarding the strengths and weaknesses of our composite odds-making.

I. Strengths and Weaknesses of Discrete Analysis

The good thing about simply counting rejections is that the process issimple and the results are easy to understand, compute, and present. Theinformation can be easily organized on one page and contains no decimals orscientific notation, yet it gives a clear, usable first impression as to what couldbe Shakespeare’s—plays with one or two rejections—and what couldnot—plays with ten or twenty-six rejections. The results even hint that theplays with twenty-six rejections are less likely Shakespeare’s than the oneswith ten rejections. Unfortunately, the downside to counting rejections is thatit only gives one the crudest of notions of the odds of Shakespeare authorship,and it allows little variance for the number of tests applied.

On the other hand, the upside to calculating Discrete CompositeProbability is that it is a simple, logical supplement to rejection counting. Byhaving both rejection counts and average Shakespeare rejection rates, one canmake certain plausible and simplifying assumptions and calculate themathematical likelihood of getting the observed count from a block of a givensize. After making the proper allowance for number of tests, this figure is then


compared to that of Shakespeare’s boundary-block threshold. Mathematicallikelihood by itself may be nothing but an abstraction, but comparativemathematical likelihood is not an abstraction at all. It is real, subject tovalidation, and highly meaningful. This process adds only one column to thepage and does not require massive recomputation. On the other hand, this newcolumn is correspondingly harder to follow than the old because it has moreinformation, signalized by more decimal places and scientific notation.Moreover, like rejection counting itself, the process is all boundaries and nodistances. It is chunky and stepwise, and it discards or ignores a great deal ofinteresting information.

J. Strengths and Weaknesses of Continuous Analysis

The strength of Continuous Composite Probability is that it is bountifullyinformation-rich. It ignores nothing (except dates and boundaries), focuses allof its attention on measuring and aggregating distances, and measures andcompares all test results more fully and precisely than either of the other twomethods. But because it is bountifully information-rich, it adds two additionalcolumns of data, which make comprehension more difficult. Worst of all (oris it best of all?) to minimize human error, Discrete requires a massive,complicated, interdependent apparatus of cross-computation and a network ofautomated rejection markers. Together, these refinements push size andcomplexity to the outer limits of what can be done with a spreadsheet, makingtampering with it exceedingly difficult and risky. These factors limit, to someextent, our ability to add navigation aids, and they stringently limit our abilityto change ranges or add new or corrected data. For practical purposes, afterten years of endless tinkering, the tinkering is over. Appendices One throughSix are now cast in concrete.

The good thing about using both Discrete and Continuous analysis is that,though they have different starting points and travel very different analyticalpaths, one much more reliant on human judgment than the other, they areremarkably, reassuringly convergent at the bottom line—especially withpassages of 1,500 words or more—and remarkably, reassuringly consistentwith each other, and with available external evidence, as to what isShakespeare’s and what is not.

K. Accuracy in Distinguishing Shakespearefrom Non-Shakespeare

How good are the results? More important than the theoretical andpresentational strengths and weaknesses of the two new methods is theirability to correctly distinguish Shakespeare from non-Shakespeare. Table 2.5gives an overview of each test’s accuracy, validated for 96-167 blocks ofShakespeare and 70-125 blocks by others.


Table 2.5. Discrimination Accuracy of Discrete and Continuous Testing

Text Shakespeare OtherShakespeare/OtherNet Discrimination

Discrete Continuous Discrete Continuous Discrete Continuous

Full-length

Plays 100 100 100 100 100 100Poems 3,000 100 100 100 100 100 100Poems 1,500 100 98 100 96 100 94Poems 750 93 98 71 68 64 66Poems 470 92 75 73 75 65 48Play Verse

3,000 95 84 100 100 95 84Play Verse

1,500 96 76 88 98 84 74Play Verse

750 97 89 75 66 72 55

Table 2.5. “Net discrimination” is 100% minus total errors, that is, 100% minus thesum of false negatives for Shakespeare (=3% for Play Verse 750, Discrete) and falsepositives for other authors (=25% for Play Verse 750, Discrete) equals 72%. Ahigher number indicates a greater degree of accuracy. Both Discrete and Continuoustesting have high net discrimination for whole plays and verse blocks over 1,500words but much lower net discrimination for blocks less than 1,000 words.

Table 2.5 shows what looks like perfect net discrimination for whole playsand excellent, but not perfect, discrimination for 3,000-word blocks. Thecomputer is right at least 95% of the time. But “perfect” and “near-perfect”may not be quite the right words to describe this discrimination accuracy whenwe are dealing with only a limited number of blocks and when, unlike ourexperience with full plays, there are a few close calls and false negatives.Generally, if a sample block comes in above our threshold, the odds stronglyfavor the proposition that it is Shakespeare, but we cannot rule out exceptionsto this rule. It hardly seems likely that, if someone submitted a 3,000-wordblock from Joseph Kesselring’s Arsenic and Old Lace, and it tested inside ourShakespeare ballpark, we would conclude that it had to be Shakespeare. Solidnegative external evidence rules it out. On the other hand, if it tested on adifferent planet, we would probably and properly conclude that it was notShakespeare. What if it were a block from The Wisdom of Doctor Dodypoll,written by an anonymous author in 1600? Scoring inside our Shakespeareballpark would still make it no more than a Shakespeare could-be, though itwould probably send us scurrying to take a closer look at the play. We wouldnot scurry quite so fast if it did not sound like Shakespeare, and we would notscurry at all if it scored on a different planet from Shakespeare.

Table 2.5 also clearly shows that our accuracy diminishes as the blocks getshorter. For blocks of 750 words or less, the computer is right no more than


65. See supra Part II.B.

two times out of three. This result may be better than chance and may bebetter than nothing. But is it better than pure, aggregated intuition? Althoughwe have found some evidence that a class of Claremont McKenna Collegeundergraduates can get it right almost nine times out of ten with 150-wordpassages, we will not know until we do more work on aggregated intuition.But on present evidence, we doubt it.

L. Factors That Affect Accuracy

The following are five standard warnings for people who use our tests:(1) our tests work better on long texts; (2) they work better on poems thanplays; (3) they work much better on single-authored than on co-authored texts;(4) they can be thrown off by confounding factors such as editors, time ofcomposition, genre, and prosody; and (5) they are novel. The first twowarnings, especially the first, should be clear from Table 2.5. The thirdwarning is one part common sense, as discussed in our description of cleanbaselines,65 and one part observation of the Dubitanda section of AppendixOne (Apocrypha): Shakespeare Dubitanda and Apocrypha PlayDiscrimination Summaries. Without exception, and unsurprisingly, wholeplays conventionally deemed co-authored—Henry VIII, Pericles, Timon ofAthens, Two Noble Kinsmen, Titus Andronicus, Sir Thomas More, and muchof the Henry VI series—and the parts of these plays not assigned toShakespeare test outside of Shakespeare’s ballpark or worse.

More surprisingly, the parts of these co-authored plays that are assignedto Shakespeare also test outside the ballpark, although generally much closerthan the supposed non-Shakespeare. Are our tests oversensitive to even a traceof non-Shakespeare? Or are the conventional assignments wrong? We couldnot say for sure in 1994 when we first encountered this problem, and wecannot say for sure now. But we can say that we are now more willing toentertain the possibility that the conventional assignments are wrong for tworeasons. First, our methods, highly novel and untested by outside challengesin 1994, are not nearly so novel now. Because they have been on the marketfor ten years and they have survived many heavy-looking challengesunscathed, their accuracy on single-authored texts is much more validated.Second, a number of contemporary Hollywood screenwriters, who are inapproximately the same business now as Shakespeare was then, are uniformlyskeptical that Shakespeare and his co-authors neatly divided their writingscene by scene to make it easier for latter-day stylometrists to decipher whowrote what.

We shall consider our fourth warning—some of our tests can be sensitiveto time, editors, genre, prosody, and so on—in greater detail when we examinethe Oxford claim in Part III. For now, let us consider our least expected andmost problematic result: Continuous Composite Testing puts Henry V in a


66. For Henry V’s vanishingly low Continuous composite probability scores, seeAppendix One (Apocrypha): Shakespeare Dubitanda and Apocrypha Play DiscriminationSummaries and Appendix Two: Dubitanda and Set-asides, Apocrypha Plays VersusShakespeare Baseline: Three-Round Composite Scores.

67. See Merriam, supra note 31, at 270.

different galaxy from baseline Shakespeare while Discrete Composite Testingputs it in the same ballpark.66 Although there is some recent opinion that theverse passages of Henry V differ from the rest of Shakespeare,67 we, like mostpeople, would guess that Shakespeare wrote it, especially given its in-the-ballpark score by Discrete. However, Continuous caught two gross anomaliesin Henry V that Discrete missed: (1) its superabundance of words new toShakespeare and (2) words ending in –ish.

Neither of these anomalies should be particularly surprising for those whoknow Henry V “by the numbers.” Normally, having too many Shakespeare-new words is a sign that the play was co-authored. In Henry V thisphenomenon is easily explained by the fact that large portions of it (unlike anyother Shakespeare play) are in French. Such words are indeed new toShakespeare, and counting new words makes Henry V stand out, with 42%more new words than the runner-up, The Merry Wives of Windsor. Also,Henry V is about a war between the French and the English; hence, the word“English” appears three or four times more frequently than it does in therunner-up, King John, which explains the superabundance of words ending in–ish. Continuous analysis caught and fully counted these two giant, aberrantspikes and correctly identified Henry V as a gross Shakespeare outlier.Discrete filtered out the full dimension of the spikes, missed the grossanomalies by light years, and, perhaps by happy accident, correctly identifiedHenry V as a Shakespeare could-be. Was this just an accident? Is theresomething to be said for having test regimes with clipping filters similar tothose found on amplifiers to avoid circuit damage from information overload?We do not know. We would want to know both kinds of results, and wewould certainly want our readers to know both kinds of results.

It is useful to recall that most statistical tests do not actually measurewhether something was written by Shakespeare, but only whether and howmuch they depart from Shakespeare’s norms. It is helpful to keep these limitsin mind when considering whether to use statistical tests because, despite theirimperfections, they may be a better guide than other alternatives. For actuallydetermining whether a text block was written by Shakespeare, our tests so farappear to be 100% accurate for whole, single-authored plays and very accuratefor a limited selection of 3,000-word verse blocks. But we consider them lessconclusive as the blocks get smaller and more variable or where otherconfounding factors, such as co-authorship, are present.


68. See supra notes 13-15 and accompanying text.69. See MACD. P. JACKSON, SHAKESPEARE’S A LOVER’S COMPLAINT: ITS DATE AND

AUTHENTICITY (1965); KENNETH MUIR, SHAKESPEARE THE PROFESSIONAL AND RELATED

STUDIES (1973). 70. What if the inputs are different? It could make some difference with four of our tests.

We have a long list of caveats and discounts for tests like hyphens, which can vary as much byeditor as by author. Four of our tests out of fifty-three (8%) are sensitive to editors and are somarked in our Appendix Keys. These four tests are grade level, open lines, hyphenatedcompound words, and whereas/whenas. Such variances are not a great problem when, as withanything in the Riverside, comparisons are made within a large corpus with the same editors.Where different editors are involved, there are a number of ways to correct for this problem at

M. Five Tests of Validity

In keeping with the Daubert rules,68 we recommend five questions to askin evaluating our claims.

1. Do our assumptions make sense? Is a clean baseline preferable to a dirtyone? Is negative evidence stronger than positive? Is blocking andprofiling a reasonable way to find Shakespeare’s customary norms? Doesit make sense to compute comparative odds of departure from norms? Allof these assumptions seem plausible to us.

2. Do our findings square with the facts? This answer depends heavily onwhat the facts are, which in turn depends heavily on whose ox is gored.We believe that there are no glaring clashes between our findings andwhat is suggested by generally accepted documentary evidence. Forexample, we have not found that Hamlet or the sonnets must have beenwritten by someone other than Shakespeare, or that ChristopherMarlowe’s Hero and Leander or Edmund Spenser’s Amoretti areShakespeare could-be’s. Our closest brush with an outright clash withreality was the massive but discounted “rejection” of Henry V discussedabove. The next closest clash might be our finding that A Lover’sComplaint tests well outside Shakespeare’s profile and is probably not his.This finding is at odds with the prevailing scholarly consensus. Theprevailing scholarly consensus overturned the previous consensus that ALover’s Complaint was not Shakespeare’s.69 On the other hand, scholarlyopinion on A Lover’s Complaint has always been divided, suggesting thatit is one of many areas where neither the external evidence nor the internalevidence has been considered conclusive. Our numbers show that theolder consensus was closer to the truth; indeed, if there is a broad-brushsummary of our findings, it is that the consensus on authorship inChambers’s time had it about right.

3. Are our tests replicable? Thirty-five out of the fifty-one tests we used aremachine tests. With the same inputs they should be perfectly replicable.Fourteen tests, for example, counting hyphens, are manual but are simple,fast, and easily replicable with the same inputs.70 Two of the tests, enclitic


retail as illustrated in our discussion of the Oxford claim in Part III. It is correctable wholesale,if at all, only by a complete re-editing of the millions of words of text in the comparisonarchive, a process too ambitious and hazardous for us to contemplate. Donald Foster, forexample, re-punctuated A Funeral Elegy to increase its average sentence length by 44%, andthen he concluded that its resulting long sentences were strong evidence that Shakespeare musthave written it. See FOSTER, ELEGY BY W.S., supra note 40, at 24-67. We believe that thehazards of such practices far outweigh the benefits.

71. See JACKSON, supra note 37; VICKERS, supra note 41.72. When we started this project, the default desktop was the then-new DOS PC. The

default “minicomputer” for serious crunching was the VAX, then at the peak of its power andpopularity. The then-new e-text Riverside, readable only with a program called Wordcruncher,was the only available complete Shakespeare on disk. Elliott had just celebrated his fiftiethbirthday. Now, DOS is passé, and so are many of the programs that we ran on it. VAXcomputers will soon be museum pieces. When they go, will they leave Intellex, one of our twosignature analytical programs, without a platform? Wordcruncher is in abeyance right now, andwith it easy public access to the electronic Riverside. And Elliott is now sixty-seven; hisplatform is aging, too. The tools we have used for fifteen years, and have freely offered to sharewith others, have been fading away faster than most of the members of the computer-shyauthorship community have been willing to try them out on their own. All of thesedevelopments will create replicability problems not far down the road, so therefore, we desireto publish our results now, despite further tinkering. Almost all of our testing problems are stillsoluble for now, and whatever replicability problems may arise will have more to do withproduct cycles and market availability than with their intrinsic reproducibility.

and proclitic microphrases, are slow and judgmental and may be only 90-95% replicable, but they have very high powers of discrimination. Tomaximize replicability, we use them sparingly and, wherever possible,with strong controls.

A different way of looking at replicability might be to ask,“Replicable compared to what?” Compared to MacDonald Jackson andBrian Vickers, two of the greatest masters of authorship studies, ourreplicability is higher because our evidence and findings are much lessdependent on the astonishing feats of learning and qualitative judgment,which is their trademark. Our evidence is more homely, morequantitative, cut-and-dried, and hence, much simpler for ordinary mortalswho are not the greatest of masters, to follow and retrace step by step.Stated differently, it is not hard to imagine that anyone else who had ourtexts and tools could come up with results very much like ours, and testany of our propositions empirically with enough precision to bet for oragainst them and know for sure who won or lost the bet. But it is hard toimagine anyone but Vickers writing a book like ‘Counterfeiting’Shakespeare or anyone but Jackson writing a book like DefiningShakespeare, or anyone challenging their conclusions in a way that couldbe settled easily or objectively with a bet.71 If we have problems ofreplicability, they are practical ones having to do with aging software,platforms, and texts that are not freely available, not ones involving theintrinsic reproducibility of our methods.72


73. See Foster, Authorship Clinic, supra note 32; Donald Foster, Response to Elliot [sic]and Valenza, ‘And Then There Were None,’ 30 COMPUTERS & HUMAN. 247 (1996); W. RonHess, Shakespeare’s Dates: Their Effects on Stylistic Analysis, in 2 OXFORDIAN 25 (1999); JohnShahan, Letter to the Editor, Reply to Elliott and Valenza, “Can the Oxford Candidacy beSaved?,” in 4 OXFORDIAN 154 (2001).

74. See Ward E.Y. Elliott & Robert J. Valenza, Can the Oxford Candidacy Be Saved?A Response to W. Ron Hess, “Shakespeare’s Dates: Their Effect on Stylistic Analysis,” in3 OXFORDIAN 71 (2000) [hereinafter Elliott & Valenza, Oxford Candidacy]; Ward E.Y. Elliott& Robert J. Valenza, Glass Slippers and Seven-League Boots: C-Prompted Doubts AboutAscribing A Funeral Elegy and A Lover’s Complaint to Shakespeare, 48 SHAKESPEARE Q. 177(1997) [hereinafter Elliott & Valenza, Glass Slippers]; Ward E.Y. Elliott & Robert J. Valenza,The Professor Doth Protest Too Much, Methinks: Problems with the Foster “Response,” 32COMPUTERS &HUMAN. 425 (1998); Ward E.Y. Elliott & Robert J. Valenza, Letter to the Editor,Reply to John Shahan, in 6 OXFORDIAN 154 (2003) [hereinafter Elliott & Valenza, Reply toJohn Shahan]; Elliott & Valenza, Smoking Guns, supra note 54; Elliott & Valenza, So Many

4. Is there a margin of error? Individual test sensitivities, such as discountsfor editing, time of composition, or genre, are marked in the keys to theappendices. Seventeen of our fifty-three tests, about a third, have suchsensitivities. As far as we can tell, all but perhaps the four editor-sensitivetests discussed previously are not hard to control by comparing likes withlikes, and we have done so. Such sensitivities, and possible others that weand our critics have not yet detected, can be important when the compositerejection is narrow. But where there are so many individual rejections thatthe composite probability is so low that it has to be written in scientificnotation (as is true of all our Claimant and Apocrypha plays and maybe85-95% of our 3,000-word Other Poets’ blocks), the composite rejectionis so redundantly strong that every conceivably weak individual link couldfail and the rejected sample still would not be in the same county withShakespeare’s farthest outlier. Our tests in the aggregate show a greatdeal of redundancy, and it matters.

5. Have our tests held up under adverse scrutiny? Yes. Over the years, ourcritics have fallen into two pairs of categories: (1) the favorable and(2) the not-so-favorable. The favorable pair is comprised of people wholike our conclusions (such as Don Foster before 1996 and the Oxfordiansbefore 1990) and people who like our methods (such as most of the 30-odd outside scholarly readers of our journal articles). The not-so-favorable pair consists of people who do not like our conclusions (such asFoster and the Oxfordians after our evidence turned against them) andpeople who do not like our methods (such as literature departmentnumerophobes who think that crunching Shakespeare is as gauche andperverse as drinking from the finger bowl). Foster and the Oxfordiansloved our methods when they liked our conclusions, but they attacked uscategorically when they did not like them.73 Fortunately for us, the attackswere ill-substantiated and did no damage to our evidence or ourconclusions.74 For people schooled in the adversary process, this is good


Hardballs, supra note 32. 75. SHAKSPER: The Global Electronic Shakespeare Conference, at http://www.shaks

per.net (last visited Jan. 2, 2005). 76. See Posting of Seán Lawrence, [email protected], to [email protected], at

http://www.shaksper.net/archives/2003/1127.html (June 6, 2003) (on file with the TennesseeLaw Review).

77. Literature Online: The Home of Literature and Criticism, athttp://lion.chadwyck.com (last visited Jan. 2, 2005).

news. You do not know how strong your bunker is until someone bombsit. Ours was deluged with bunkerbusters, but the damage was negligible.Either our bunker was strong, or the bombs were duds, or both. Theattacks amounted to a series of massive, highly adversarial audits that wepassed with flying colors.

6. Are we or our critics willing and able to bet on it? Let us conclude ourDaubert duties with a thought experiment and an offer. We claim that ourtests have been 100% accurate in distinguishing core Shakespeare playsfrom non-Shakespeare plays. Would we be willing to bet on it? As ithappens, this is not just a hypothetical. A Canadian literature departmentnumerophobe and hockey fan proclaimed categorically on Shaksper, theleading Shakespeare news and discussion group,75 that our statistics,indeed all statistics, are circular and tell you nothing that you do notalready know. He announced that although we had admittedly foundsome tests that could separate a few known Shakespeare plays from a fewknown plays by others, these results tell us precisely nothing about playswe have not tested.

We thought he was wrong. If the sun rises in the east for eighty daysin a row, and not in the west, we would take it as a sign that it would doso on the eighty-first day as well. It did. If the New Jersey Devils playedthe neutral-zone trap all year in 2003, we would take it as a sign that theywould do so likewise in the playoffs. They did. If all twenty-nine ofShakespeare’s core plays test inside the ballpark and all fifty-one otherauthors’ plays test outside the solar system, we would take it as a sign thata hypothetical thirtieth pure-Shakespeare play would probably also testinside the Shakespeare ballpark and that a fifty-second play by someoneelse would probably test outside it. Barring a miraculous discovery of alost Shakespeare play, the second proposition is much more testable thanthe first. We thought it was testable enough to bet on it with a stronglikelihood of recognizing who won and who lost. We offered our critic aone thousand dollar even-odds bet that our tests would reject any hitherto-untested other-authored play he might choose. He declined, and webelieve he was wise to do so.76

What would have happened if he had accepted the bet; gone to LiteratureOnline (LION);77 called up all English plays between 1550 and 1620; cast all


78. See Appendix One (Claimants): Claimant Play Discrimination Summaries;Appendix Two: Claimants Versus Shakespeare Baseline, Three-Round Composite Scores.

79. On the other hand, Sir Thomas More, thought to be mostly written by AnthonyMunday, has thirty lines attributed to Dekker and 148 lines commonly attributed toShakespeare, yet it has only seven Shakespeare rejections, the same as Titus Andronicus, whichmost people believe was at least co-authored by Shakespeare. But these are only, respectively,about 1% and 6% of the play, and most people besides us think the 6% is Shakespeare. The 1%is probably too small to generate rejections on top of Munday’s, and the 6%, if it isShakespeare’s, would tend, if anything, to dilute Munday’s rejections, not add to them.Munday’s other tested play, John à Kent and John à Cumber, generated fourteen rejections.

80. The eighty works tested include twenty-nine core Shakespeare baseline plays andfifty-one identifiable other-authored plays. We did not count nine plays from the Dubitanda,whose results are qualified by varying degrees of co-authorship, nor twenty-seven plays of theShakespeare Apocrypha, whose results are unequivocal but barely appear on our LION list.

counting and calculation to the winds; simply picked one out on a whim,perhaps Sir Giles Goosecap: Knight, written in 1606, anonymous but ascribedby LION and others to George Chapman; and thrown it at our feet like a glovethat just might fit? In one sense, this play would be a better-than-averagechoice because, as far as we know from LION, it is single-authored. Our testsare quick enough to reject even one non-Shakespeare co-author; imagine whatthey would do with two! But in another sense, it would be a worse-than-average choice because we have already tested two other Chapman plays, TheGentleman Usher, written in1602, and Bussy d’Ambois, written in 1607, andboth tested in a different galaxy.78 We have not tested Sir Giles Goosecap, oreven read it, and it is certainly conceivable that, unlike the seventy-nine otherother-authored plays we tested, it would land right inside the ballpark and notoutside the Shakespeare solar system. But we would not want to bet on it, andneither, we suspect, would he.

On the other hand, nothing but his professed numerophobic principlesrequires him to pick at random or on impulse. With one thousand dollars atstake, our critic might swallow his contempt for systematic counting andcomparing, return to the LION list, and try to beat us at our own bean-counting, Sabermetric, Moneyball game. He would screen LIONsystematically, searching for a winner, using the very tests he professes toscorn and which we would happily supply. When we went to LION toconsider how big a job this might be, the site listed for us 361 plays performedbetween 1550 and 1620. Forty-nine of the plays (14%) were multiple-authored and therefore below-average candidates for the same reason that SirGiles Goosecap was an above-average candidate: Multi-authored plays wouldprobably have more Shakespeare-rejectable authorial quirks than single-authored plays.79

Of the remaining 312 available single-authored plays, we found that wehad already tested eighty of the first, most accessible layer and found none ofthe other-authored plays on the same planet as any Shakespeare play.80 Wedid not test the second layer, the 146 plays like Sir Giles Goosecap, because


they were believed to be written by authors like Chapman whose other playsor poems had been tested by us and landed outside the solar system. Thissecond layer amounts to half of the single-authored plays in LION, and wesuspect that any sensible person looking for a Shakespeare non-rejectionwould put these plays aside as well. If there is any regularity or consistencyto an author’s writing habits, and our tests show abundantly that there is, it isnot necessary to test all fourteen of Chapman’s plays, nor all seventeen ofFletcher’s, nor all thirty-eight of Jonson’s, to know when you have lost thescent of the True Shakespeare.

That leaves the third layer, the last quarter of LION’s single-authoredplays, eighty-six plays by forty-six authors we have not tested, such as JohnPhillips’s Patient Grissel, George Ruggle’s Ignoramus, or Nicholas Udall’sRalph Roister Doister. Screening eighty-six plays would be time-consuming,even with our wondrous toolbox of fast tests, and we would be surprised if itproduced a single play that fit our Shakespeare profile. After all, these playsare the leftovers from our claimant and apocrypha lists, the residue of centuriesof scholars picking over the same 300-odd plays to find another play writteneven partially by Shakespeare. No anti-Stratfordian nor any orthodox scholarsearching dusty libraries for lost Shakespeare works could identify in any ofthem a plausible, marketable resemblance to Shakespeare. Who would wantto spend any further time seeking a Shakespeare match in such barren-lookingtailings? Certainly not us.

On the other hand, if we could pay someone else one thousand dollars todo the necessary, massive, tedious, and unpromising due diligence, we wouldconsider it money well spent. It would validate our tests on the last, least-promising quarter of LION’s population of available plays for a tiny fractionof what it had cost us for the first quarter. It could confirm or qualify ourconfidence in our new techniques, help spread its use to others, and maybe,just maybe, it could turn up that long-overlooked Shakespeare treasure that somany have sought so long in vain. What is not to like? We would be willingto revive our bet offer, donate what is left of our best software, and help withthe crunching. We would enjoy the windfall win or lose. Are there anytakers?

Just as important as our willingness to bet on the predictive powers of ourfindings is the fact that our rules are so tight, quantified, and hence, replicablethat our prediction would be eminently testable and falsifiable. If anyonetakes us up on our bet, with or without screening, it will not be difficult to tellwho won or lost. Can this be said of any other composite authorship-identification system now on the market? We would not bet on it.

N. Set-Asides and Latent Variability

Before taking leave of the original challenge to try to prove that ourmethods would work on a new, untested Shakespeare or other-authored play,we should mention “set-asides”—the need to compensate for latent variabilityin small or partial baselines—and how these bear on the robustness of our


81. See Appendix Five: Shakespeare Play Baseline Data, Round Three Tests (Hamlet).82. See id.83. See id. (Julius Caesar).84. See Elliott & Valenza, Smoking Guns, supra note 54, at 210-11 (discussing safety

allowances with regard to determining the importance of rejections in tests comparing a smallJohn Ford baseline with a large Shakespeare baseline).

85. See id. at 209-11 (explaining the need for more safety allowances when testingsmaller baselines).

findings. What if no one had ever heard of Hamlet or Julius Caesar, but bothwere found hidden in the rafters of Shakespeare’s house or in the ruins ofOxford’s Castle Hedingham? Would our tests recognize them as Shakespearecould-be’s? More plausibly, what if, instead of testing all of Shakespeare’splays, we had purposely set these two plays aside from our baseline andfollowed exactly the same rules with the remaining plays that we did with ouractual core baseline? Would they still come out as could-be’s?

The answer, with a small safety allowance for the partial baseline, is yes.It is perfectly possible to do this exercise retroactively by examining theShakespeare baseline scores on each test in Appendices Three to Five and bylooking for profile-defining outlier scores. In Hamlet’s case, there would bethree of these in forty-eight tests: (1) un- words, with sixty-five per 20,000;(2) -ment words with thirty-six per 20,000; and (3) very’s with forty-two per20,000.81 The runners-up had, respectively, sixty-four, thirty-five, and forty-one—one less of each across the board.82 In Julius Caesar’s case, there wereonly seventeen ex- words per 20,000; the runner-up had nineteen.83

A small safety allowance, such as a 5-10% expansion of the profile rangeor a loosening of the Discrete Composite rejection threshold, or both, couldeasily accommodate such small differences and still easily say “could-be” toall known core Shakespeare plays and “couldn’t be” to all known, tested non-Shakespeare plays. Such ease of accommodation is due largely to the colossaldistance between the nearest of these non-Shakespeare plays andShakespeare’s farthest known core baseline outlier. We have not added suchallowances to our Shakespeare ranges, which are already too complex anddensely packed for most readers, but we have used them in the past to separate“firm” rejections from “narrow” ones.84 We would not hesitate to use suchallowances in the future, where appropriate.

The partial- or smaller-baseline consideration is of less concern forShakespeare, whose actual poem baseline is sizeable and whose play baselineis huge, than it is for others such as Marlowe, Chapman, and Ford, whoseavailable baselines are smaller. Accordingly, if the starting baseline is small,the safety allowances for non-inclusion should be large.85

In a sense, every other-authored play we have not tested is a kind of set-aside, and it should not be hard to test any one, ten, or all of the 232 or sountested LION plays simply by taking us up on our bet. But is that evennecessary? We never planned it that way, but many natural trials of the set-


86. See Elliott & Valenza, Oxford Candidacy, supra note 74, at 72 (listing our oldranges).

87. See infra Part III.

aside idea have already taken place with smaller blocks. Whenever weextended our testing from a small number of Shakespeare poem blocks to alarger number of Shakespeare play verse blocks, we were testing a very largeset-aside. Table 2.1 above hints at the outcome for several block sizes. Exceptfor grade-level, where Shakespeare’s use of much shorter sentences for a massaudience is hardly surprising (everyone else who wrote poems and plays didit too), most of the play-block ranges turned out to be all but identical to thepoem-block ranges.

Likewise, when Marina Tarlinskaja sent us a bonanza of her enclitic andproclitic counts in 2002, she tripled our holdings of reliably-countedShakespeare play verse, doubled our holdings of all Shakespeare verse, andled us to enlarge our total Shakespeare play verse baseline from fifty-six3,000-word blocks to eighty-two, a 46% increase. However, our ranges barelybudged. Table 2.6 shows the comparison between the relevant old ranges,86

before the “set-aside” bonanza, and the new ones in Table 3.187 thatincorporate the “set-aside” into the new baseline.

Table 2.6. What happened to our ranges whenwe increased the baseline by 46%?

Test, 3,000-word blocks Old 2000 range New 2004 rangeHyphenated CompoundWords

31-153 31-153

BoB7 136-944 136-944Open Lines 7-24 7-24Feminine Endings 5-23 5-23Enclitics/1,000 lines 31-87 27-89Proclitics/1,000 lines 265-476 265-476

Table 2.6. Increasing our Shakespeare verse baseline by 46% made no differencein four of our Oxford-relevant Shakespeare ranges. Doubling our Tarlinskaja-counted Shakespeare verse baseline enlarged our enclitic range by 6%, not enoughto make the Earl of Oxford a Shakespeare could-be, and left all other rangesunchanged.

We would assume that our tests that work on 3,000-word blocks aregenerally more robust than those that only work on whole plays. However,these very high levels of robustness in the small blocks, when tested againstlarge set-asides, are strong indicators that we would find high levels ofrobustness also in large blocks. Thus, our conclusions are no less reliablebecause we tested all of Shakespeare’s known or suspected plays at once.


88. The text and footnotes of Part III are largely derived from Elliott & Valenza, OxfordCandidacy, supra note 74.

89. See OGBURN, supra note 2, at 393-97 (reproducing and discussing Bénézet’s text).90. Id. at 396.91. See May, supra note 9, at 79-84.

They are also scarcely less reliable because we tested only a quarter of otherworks, where the other three-quarters of works are either by people we havealready tested and firmly rejected or by authors who have been passed over bygenerations of scholars desperately seeking Shakespeare.

III. HOW OUR METHODS APPLY TO THE EARL OF OXFORD88

A. Oxford Fails Too Many Teststo Be a Shakespeare Could-Be

Our methods are strong and predictive enough, for play-length samples,for us to bet a sizeable sum that they will be at least a tenth as good at rejectingthe next non-Shakespeare play as they were at rejecting the other seventy-eightplays that we tested. The odds are not always as uniformly overwhelming for3,000-word verse blocks, like our entire Oxford corpus, or for 1,500-wordverse blocks, like the half of the Oxford corpus that is in iambic pentameter.Such blocks require closer attention to the validity of individual tests in aspecific retail application, where various discounts for things like time,editorial practice, genre, and prosody can be considered. But still, even afterevery discount has been applied, the odds that someone of Shakespeare’sknown writing habits could have written Oxford’s poems by chance, or viceversa, are much lower than the odds of getting struck by lightning.

Oxfordians have long argued that Oxford’s writing sounded just likeShakespeare. They recall with pleasure the trick that Oxfordian scholar LouisBénézet, Sr. liked to play on Stratfordian English literature professors in the1940s. He would give them a seventy-line mixture of passages fromShakespeare and Oxford, defy them to tell one author from the other, find thatthey had great trouble in doing so, and conclude that his experiment showedthat Oxford’s style was barely distinguishable from Shakespeare’s.89

Much has changed since those days. In 1980 Steven May showed fromexternal evidence (and over Charlton Ogburn, Jr.’s objections that “[h]e isunconcerned with stylistic criteria”90) that some of the passages Bénézetthought were Oxford’s were in fact written by Robert Greene and that fiveother poems confidently assigned to Oxford by Thomas Looney, and otherOxfordian scholars (following A.B. Grosart), were not Oxford’s work.91 In1989 the students in the Claremont Shakespeare Clinic found mismatchesbetween Shakespeare and twenty-seven testable poet claimants, including thefrontrunners Oxford, Bacon, and Marlowe. Oxford’s poems failed five of thesix new tests and seemed particularly different from those of Shakespeare. A


92. See Ward E.Y. Elliott & Robert J. Valenza, A Touchstone for the Bard, 25COMPUTERS &HUMAN. 199 (1991) [hereinafter Elliott & Valenza, A Touchstone for the Bard];Ward Elliott & Robert J. Valenza, Was the Earl of Oxford the True Shakespeare? A Computer-Aided Analysis, 236 NOTES & QUERIES 501 (1991) [hereinafter Elliott & Valenza, TrueShakespeare].

93. See Elliott & Valenza, True Shakespeare, supra note 92, at 503-04. 94. Id. at 504. 95. Id.96. Id. 97. See Hess, supra note 73, at 25-39.

round of “refutations” ensued in Oxfordian publications and in privatecorrespondence with Oxfordians, followed by a long period of silence. Some“refutations” made worthwhile points; others did not. Invitations to respondto these charges in Oxfordian publications or at meetings stopped forthwith formany years. During this time, we made a few revisions in our tests andpublished our general findings in mainstream journals.92 We noted thatthe stylistic trends in Shakespeare’s plays, by conventional dating, mightprotect Oxford from a Shakespeare rejection for having too few feminineendings or open lines because most of Oxford’s poems were written beforeShakespeare’s,93 but that the trends in Shakespeare, conventionally dated,lasted for years after Oxford’s death.94 We also noted that if we usedOxfordian dating, all of Shakespeare’s dates would be dated ten or twentyyears earlier. In that case, no trends would have outlasted Oxford, but thetrends themselves would disappear entirely, leaving Oxford a gross mismatchwith Shakespeare.95 We thought that either interpretation would gravelydamage the case for Oxford, although we did not exclude the possibility thatOxfordian dating could “be reshuffled somehow to fit Oxford at both ends.”96

Eight years later, Oxfordian Ron Hess finally accepted our challenge andreshuffled Oxford’s dates to maintain the upward stylistic trends until 1604when Oxford died.97 The Oxfordian invited us to reply, and we did, startingwith a table of Oxford’s Shakespeare rejections, updated and reproduced herein Table 3.1.


98. Elliott & Valenza, Oxford Candidacy, supra note 74, at 72.99. See Appendix Six: Shakespeare Poems Baseline Data, Blocksize = 3,000 for fourteen

of these tests. The fifteenth test is the use of relative clauses, a well-validated test, which,however, is manual and too slow for us to apply wholesale outside of Shakespeare.

Table 3.1. Oxford’s Poems Compared to Shakespeare’sMost Discrepant Poem Block98

Shakespeare Range

MostDiscrepant

ShakespearePoem Block

OxfordPoems

ShakespeareBaseline

Remarks

Grade level, 10-14 10 7 A g, e

HCW/20k, 31-153 153 32 A, B e

Rel. clauses, 7-17 7 20 A

BoB7, 136-944 625 1000 A, B t, s/m

Modal distance, 281-1149 467 2892 A g

Open lines %, 7-24 8 7 C t, e, pFem. endings %, 5-23 11 0 C t, pEnclitics/1,000 ln., 18-143 22 13 C pProclitics/1,000 ln., 235-561 199 115 C p

Total Shakespeare rejections 1 7 A

Discrete Comp. Prob. 3000

Sh.: Oxford 400,000 to 1

3.08E-01 7.75E-07 A

Continuous Comp. Prob. 3000

Sh.: Ox. 150 trillion to 1

1.50E-01 <1E-15 A

Table 3.1. The Shakespeare Clinic’s fifteen authorship tests show strong similaritiesamong ninety-six 3,000-word Shakespeare core verse blocks tested, and strongdissimilarities between Shakespeare’s verse and Oxford’s. The first 3,000-wordblock of Venus and Adonis, though it least resembles Shakespeare’s other poemblocks, has only one Shakespeare rejection in fifteen tests.99 Oxford’s poems haveseven rejections in the same fifteen tests (shaded), far more than any Shakespeareblock tested. Oxford’s poems have many more relative clauses than Shakespeare andfar fewer feminine endings, enclitics, and proclitics. His grade-level scores are farbelow Shakespeare’s, his BoB7 scores above Shakespeare’s, and his modal distancefrom Shakespeare’s mean is far greater than that of any like-sized Shakespeare block.Note that the four-verse tests (between the thick horizontal lines) are for Oxford’siambic pentameter verse only, which is less than half of his 3,042-word sample. Forthese verses, the relevant comparison (shown) is to the first 1,500-word block ofVenus and Adonis. The odds that Shakespeare could have produced Oxford’s testpatterns by chance are between 400,000 to 1.5 quadrillion times worse than the odds


100. These odds do not exactly match those in Appendix Six because we use one more testhere; we separate iambic pentameter more explicitly; and we compare Oxford to a single, most-discrepant Shakespeare block under both composite-odds tests, rather than to the two or moremost-discrepant blocks under each test separately. Even so, the overall outcomes are similar;the odds of Shakespeare authorship are vanishingly low. In comparison, the odds of one personnot just getting hit by lightning, but dying from it, have been reckoned at about 1 in 2.4 milliona year, 1 in 32,000 in a seventy-five-year lifetime. See Heterosexual AIDS Risk Versus BeingStruck by Lightning, at http://www.righto.com/theories/lightning.html (last visited Oct. 4,2004).

for Shakespeare’s own most discrepant block. These odds are also worse than theodds of getting hit by lightning.100

Key to Table 3.1HCW: Hyphenated Compound Words per 20,000 words.Rel. Clauses: Relative clauses per 3,000-word block.BoB7, Modal distance, open lines, feminine endings, enclitics, and proclitics per1,000 lines: see text. Relevant Shakespeare ranges are set in bold type and italicized.

Shakespeare Baseline: A: Fourteen 3,000-word blocks of Shakespeare’s poems; all but A Lover’s

Complaint (Shakespeare authorship doubtful) and The Phoenix and Turtle (tooshort);

B: Eighty-two 3,000-word blocks of verse from selected Shakespeare plays;C: Twenty-eight 1,500-word blocks of Shakespeare’s poems, minus A Lover’s

Complaint and The Phoenix and Turtle; same as A, but with 1,500-word blocks.

Remarks on Test Sensitivities:g: results can be sensitive to differences of genre (poem verse v. play verse);e: results can be sensitive to differences in editing, such as spelling and

punctuation;p: results can be sensitive to differences in prosody, that is, verse structure, meter,

stanzaic structure, and rhyme schemes;s/m: results can be sensitive to differences in subject matter;t: results can be sensitive to differences in time of composition;

All ranges and results except those in the area between the thick lines are basedon comparisons between the entire Oxford poem corpus, per Steven May, and 3,000-word Shakespeare poem and/or play verse blocks. Ranges and results in the areabetween the thick lines are based on comparisons with 1,500-word Shakespeare poemblocks, and compare only the 1,304 words of Oxford’s poems that are in iambicpentameter with like-sized iambic pentameter Shakespeare poem blocks. See“Shakespeare Baseline” column, above, for details. Shakespeare’s most discrepantpoem block, both 3,000-word and 1,500-word, is the first block of Venus and Adonis.


101. See May, supra note 9, at 25-42. 102. See id. at 79-83. 103. We tested some of the A Hundreth Sundrie Flowres “Oxford” poems in 1990 and did

not find a Shakespeare match. See Ward E.Y. Elliott & Robert J. Valenza, And Then ThereWere None: Winnowing the Shakespeare Claimants, 30 COMPUTERS & HUMAN. 191, 201-02(1996) [hereinafter Elliott & Valenza, There Were None] (finding too many Oxford and“Meritum Petere Grave” Shakespeare rejections for either to be a credible True Shakespeare);Elliott & Valenza, True Shakespeare, supra note 92, at 502-04 & n.6 (rejecting both Oxford and“Meritum of Petere Grave” as credible Shakespeare claimants).

B. Test Specifics

For Oxford, as for Shakespeare, we followed our standard methodologicalguidelines— clean baseline, block and profile, and silver-bullet evidence—andused our standard methods for calculating comparative odds for Shakespeareauthorship. In Oxford’s case we used a clean comparison sample: the poemsSteven May assigned conclusively to Oxford in 1980,101 and not his “possiblyOxford” poems,102 nor the A Hundreth Sundrie Flowres poems claimed forOxford by some Oxfordians.103 We compared like-sized Shakespeare blockswith both relevant Oxford blocks—Oxford’s whole, 3,042-word corpus with3,000-word Shakespeare blocks, and Oxford’s 1,304-word iambic-pentametercorpus with 1,500-word Shakespeare blocks.

Besides size, we also tried to control for other relevant variables littlediscussed in our broad-brush, wholesale Part II but pertinent to detailed, retailanalysis of whether a given sample, such as Oxford’s poems, is a Shakespearecould-be. These variables are the following: genre (whether a work is playverse, prose, or poem), time of composition, subject matter, editorialconventions (spelling and punctuation), and prosody (for example, meter orstanzaic structure).

The most common of these variables are listed in the “Remarks” columnof Table 3.1. It is seldom possible to match perfectly all of these at once, butthere are often ways to try one combination against the other and see howmuch difference it makes. In Oxford’s case, we have matched our 3,000-wordblocks for genre (poem v. poem, or poem v. play verse) and spelling(Riverside spelling, including hyphenation) but not for prosody or time ofcomposition where Oxfordians take strenuous issue with orthodoxShakespeare dating. Ninety-nine percent of Shakespeare’s verse, but less thanhalf of Oxford’s, is iambic pentameter, and most of Oxford’s poems farpredate Shakespeare’s plays, as conventionally dated. For our four verse-tests,which are considered sensitive to prosodic variations (between the thick linesin Table 3.1), we used 1,500-word blocks matched for genre (poem v. poem),spelling, and meter (iambic pentameter v. iambic pentameter) but again, notfully matched for time of composition by conventional reckoning.


104. See Appendix Six: Other Poets Versus Shakespeare Baseline, Blocksize = 3,000;Appendix Seven: Other Poets Versus Shakespeare Baseline, Blocksize = 1,500.

105. After re-editing Oxford’s poems to follow spelling conventions found in theRiverside, we found five arguable hyphenated compound words—oft-times, late-done, good-liking, salt-sea, and tennis-knit—well below Shakespeare’s minimum eight in fourteen 3,000-word poem blocks. See Elliott & Valenza, There Were None, supra note 103, at 198, 237. Twoof our fifty-five blocks of Shakespeare’s play verse (2.9% of our total Shakespeare versebaseline) had as few as four HCWs. The figures given are HCWs per 20,000 words to facilitatecomparison with plays. As far as we can tell from the Riverside, Shakespeare’s hyphenationranges did not vary much between poems and plays, nor between early and late works.

C. Test Outcomes

Our emphasis on silver-bullet, negative evidence means that the eightstrong, validated Shakespeare tests that our Oxford poem sample passed104 aremuch less interesting than the seven that it failed. The former are nothingmore than could-be’s. Only the latter (plus hyphenated compound words, aclose-call Oxford pass, and open lines, a time-sensitive verse test passed byOxford’s iambic-pentameter poems and not segregated in our earlier work toinclude iambic pentameter only) are listed in Table 3.1. Table 3.1 comparesOxford’s poems with Shakespeare’s least typical core poem blocks, the first1,500 or 3,000 words of Venus and Adonis. In all but one case, the mostdiscrepant Shakespeare block fits (sometimes barely) within the Shakespeareprofile we used, while in every case but two (hyphenated compound words(“HCW”) and open lines), the matched Oxford block does not. Let us look atthe Oxford outcomes.

Grade Level. Shakespeare’s poems have much longer sentences and/orlonger words than Oxford’s, testing no lower than the tenth-grade level.Oxford’s poems test at the seventh-grade level. This test, which comparesOxford’s lightly modernized punctuation with that of the Riverside, issensitive to editorial preference, but comparing original-punctuation Oxfordwith original-punctuation Shakespeare would make the gap a full grade wider(Figure 2.5). It therefore seems to be a clear rejection.

Hyphenated Compound Words. Oxford’s poems have fewer HCWs perblock than any like-sized Shakespeare poem block and fewer HCWs than 97%of Shakespeare’s like-sized play verse blocks. But, to be cautious, we re-edited Oxford’s poems to mark every arguable Riverside hyphenation; weexpanded our Shakespeare verse baseline to include plays, as well as poems;and we broadened our Shakespeare profile to include the highest highs andlowest lows found in either genre (Table 2.1).105 This was just enough tosqueeze Oxford’s poems into the expanded Shakespeare range and turn anarrow rejection into a narrow pass, though it still hardly makes a close matchwith Shakespeare.

Relative Clauses. In the phrase, “the evil that men do,” that men do is arelative clause. Oxford’s poems have twenty relative clauses, many more thanShakespeare’s maximum of seventeen per 3,000-word block. We found this


106. See Elliott & Valenza, A Touchstone for the Bard, supra note 92, at 201-07 (providinga detailed discussion of modal analysis); Elliott & Valenza, True Shakespeare, supra note 92,at 502 (discussing modal analysis in general).

107. See Elliott & Valenza, There Were None, supra note 103, at 196-97 (discussing“Semantic Bucketing”). For our updated modal scores, see Appendix Six: Shakespeare PoemBaseline Data, Blocksize = 3,000.

108. See supra text accompanying note 45. 109. See Appendix Three: Shakespeare Play Baseline Data, Round One Tests; Figure 3.4.110. See Figure 3.4.

test useful, but we used it sparingly outside the Shakespeare baseline becauseit is slow and manual. Here, it produces another clear Oxford rejection.

BoB7. This is a ratio of occurrences of the word “is” to occurrences of thecontractions ‘tis, there’s, I’m, here’s, she’s, that’s, and what’s. It is validatedfor the entire range of Shakespeare plays but, because Shakespeare used fewercontractions in his earliest plays, it reasonably could be questioned as appliedto works, such as Oxford’s poems, composed before the earliest ofShakespeare’s plays, as conventionally dated. In other words, as noted, thistest is sensitive to time of composition. However, the Oxfordian re-datings,which backdate Shakespeare’s earliest plays by a decade or more, would makesuch sensitivity a moot issue.

Modal Distance. Modal analysis tests the extent that authors use, or avoidusing, certain words together.106 Our version was sensitive to genre, workingwell on poem-poem comparisons, like this one, but not so well on play-verseor song-verse comparisons.107 Shakespeare’s updated modal range for 3,000-word blocks runs from 281 to 1,149. Oxford’s poems received 2,892, almosteight standard deviations removed from Shakespeare’s mean and six beyondShakespeare’s maximum. By this test, the two authors were on differentplanets.

Open lines. These are lines that do not end with punctuation. They aresensitive to time of composition, editorial practices (punctuation), andprosody. Therefore, one should compare iambic pentameter with iambicpentameter, not with any of the un-Shakespearean meters that characterizemost of Oxford’s verse. Hence, all the verse tests in Table 3.1 (between thethick lines) compare only Oxford’s iambic pentameter poems, amounting to1,304 words, and written between 1576 and 1593, with Shakespeare’s iambicpentameter poems, written between 1593 and 1609 by conventional dating,and divided into like-sized 1,500-word blocks.108 Oxford’s poems barely fitwithin Shakespeare’s 1,500-word profile, with no effort to allow for strongupward trends in Shakespeare’s play verse, where, by conventional dating, thepercentage of open lines quadrupled between 1590 and 1613.109

Discounting for trends in open lines in the plays would be necessary underconventional dating because most of Oxford’s poems far predated most ofShakespeare’s plays.110 But discounting trends in open lines would not benecessary under Oxfordian re-dating of the plays, which makes many ofShakespeare’s plays appear contemporary with Oxford’s poems and makes the


111. See Figure 5.112. EVA LEE TURNER CLARK, HIDDEN ALLUSIONS IN SHAKESPEARE’S PLAYS: A STUDY

OF THE EARLY COURT REVELS AND PERSONALITIES OF THE TIMES (Ruth Loyd Miller ed.,Kennikat Press 3d ed. 1974) (1930) (first published as SHAKESPEARE’S PLAYS IN THE ORDER

OF THEIR WRITING).113. Hess, supra note 73; see Figure 3.5. 114. See Figure 3.1.115. See Figure 3.3. 116. See Figure 3.2. 117. See id. 118. See Elliott & Valenza, True Shakespeare, supra note 92, at 504-05. 119. See Figure 3.3 and accompanying text.

discrepancy between Shakespeare’s open-line play-verse range from 1579 to1591 (11-32%) and Oxford’s iambic pentameter 1576-1593 poem score (7%)start to look suspicious.111 All Oxfordian re-dating efforts, from the oldestones by Eva Turner Clark112 to the most recent by W. Ron Hess,113 threaten toturn Oxford’s narrow pass for open lines into a rejection. They weaken, ratherthan strengthen, the argument that Oxford fits within the Shakespeare profile.

Feminine Endings. These are lines ending on an unstressed syllable, withwords such as “gotten” or “running.” They are not sensitive to editing, butthey are considered sensitive to prosody and require iambic pentameter toiambic pentameter comparison in like-sized, 1,500-word blocks. Oxford failsthis test decisively. Compared to Shakespeare’s poem range, in which thelowest percentage was five and highest was twenty-five, Oxford has nofeminine endings at all. But this test, too, is sensitive to time of composition.Feminine ending percentages increased by a half or more in Shakespeare’splays, conventionally dated,114 as they do by the Hess dating,115 but not by theold Clark dating.116

As with open lines, if conventional dating of the plays is correct andallowance is made for the clear upward trend in play verse, one couldplausibly argue that Oxford’s lack of feminine endings fits comfortably belowthe bottom of a long upward trendline which is documented from 1590 on andmight well have started earlier. Conventional dating calls into question thistest’s rejection.

But such questions would disappear under Oxfordian re-dating. The Clarkdating117 crams almost every Shakespeare play back into the 1570s and 1580s,obscures the trend line, and makes Oxford’s lack of feminine endings a glaringmismatch with Shakespeare. This is essentially what we reported in our 1991Notes and Queries article, in which we used Charlton Ogburn, Sr.’s dates,which are similar to Clark’s.118 The Hess re-dating119 is less radical about thetime rollback and better at preserving trendlines in the plays. But the trendlineis still less distinct than with conventional dating, and Oxford’s poems, withno feminine endings at all, are still glaringly out of line with Shakespeareplays that Oxfordian re-dating portrays as contemporary with Oxford’s poems.


120. See Elliott & Valenza, There Were None, supra note 103, at 201 (discussing encliticand proclitic microphrases); see also MARINA TARLINSKAJA, SHAKESPEARE’S VERSE: IAMBIC

PENTAMETER AND THE POET’S IDIOSYNCRASIES 208-22 (1987) (providing general informationon proclitic and enclitic phrases and discussing Shakespeare’s use of the phrases in his works).

121. See Appendix Seven: Shakespeare Poems Baseline Data, Blocksize = 1,500;Appendix Seven: Shakespeare Play Verse Baseline Data, Blocksize = 1,500. The eight playsare Richard II, Richard III, Titus Andronicus, Romeo and Juliet, Troilus and Cressida, Antonyand Cleopatra, The Tempest, and A Winter’s Tale.

122. See Appendix Seven: Shakespeare Poems Baseline Data, Blocksize = 1,500;Appendix Seven: Shakespeare Play Verse Baseline Data, Blocksize = 1,500.

123. See Appendix Seven: Other Poets Versus Shakespeare Baseline, Blocksize = 1,500.124. See Hess, supra note 73.

Again, the Oxfordian re-dating weakens, not strengthens, the case for a matchwith Shakespeare.

Enclitic and Proclitic Microphrases. These tests count instances in whichcertain “clinging monosyllables,” stressed in natural speech, lose the stress formetrical reasons.120 Oxford’s iambic pentameter poems fall below the bottomof our Shakespeare 1,500-word-block profiles for both of these tests, andhence, suffer two more clear Shakespeare rejections. A generous sampling of108 1,500-word blocks from all of Shakespeare’s core poems and eight of hisplays provided no indication that his rates on either test increased or decreasedduring his lifetime.121 Of our 108 Shakespeare samples, only three had encliticreadings as low as our one Oxford sample.122 None had proclitic readingsanywhere near Oxford’s abysmal 115.123

D. Summary of Oxford Rejections

After all the refining and updating, the Oxford candidacy fares no bettertoday than it did in 1990. His poems now have seven Shakespeare rejectionsin fifteen tests, far too many to look like Shakespeare to us or to our computer,which calculates the Discrete odds against so many rejections arising bychance from Shakespeare as 400,000 times worse than those for his own mostdiscrepant block. The Continuous composite odds are 150 trillion timesworse. These comparative probabilities reveal that Shakespeare and Oxfordbelong in different galaxies; the odds that either could have written the other’swork are far lower than the odds of their having been struck and killed bylightning. Three of Oxford’s rejections could be time-sensitive. However,trying to run Oxford’s poems against Oxfordian-backdated plays only makesthe mismatches more glaring, not less.

E. A Comparison of Orthodox andOxfordian Chronologies

To compare orthodox and Oxfordian chronologies, we consulted the mostevolved Oxfordian chronology, that of W. Ron Hess,124 and the earlier ones it


125. Id. at 28. 126. See Elliott & Valenza, True Shakespeare, supra note 92, at 504 (identifying problems

with conventional Oxfordian dating).127. We and a few others have long doubted that A Funeral Elegy was written by

Shakespeare, and hence, it was no obstacle to the Oxford claim. We turned out to be right. Aswith Oxford’s poems, our tests say that A Funeral Elegy is on a different galaxy fromShakespeare, but in the same infield with Ford. See VICKERS, supra note 41; Elliott & Valenza,Glass Slippers, supra note 74, at 187-201; Elliott & Valenza, Smoking Guns, supra note 54, at205-10; Posting of Richard J. Kennedy, [email protected], to [email protected] (Mar.1, 1996), at http://www.shaksper.net/archives/1996/0152.html (on file with the Tennessee LawReview); Appendix Six: Other Poets Versus Shakespeare Baseline, Blocksize = 3,000. In 2002Donald Foster conceded that our ascription of A Funeral Elegy was right and his was wrong,and A Funeral Elegy has not been heard from since. See Posting of Donald Foster, supra note40.

replaced. Commendably, Hess acknowledged the conjectural aspects ofassigning dates to plays125 and attempted to respond to our 1991 challenge toreshuffle Oxfordian dating to reflect what looks like clear stylistic trends inShakespeare’s plays, by conventional dating, while ensuring that the trendsstop after Oxford’s death in 1604.126 As we have seen, the stylistic trends helpthe early Oxford “fit” by showing that some of his rejections are just what onemight expect from extrapolating Shakespeare’s trendlines backwards. Butthese trends do not fit Oxford at all if they continued after his death—unlessOxford’s death was faked, as some Oxfordians briefly tried to argue afterA Funeral Elegy (clearly dated in 1612) was mistakenly attributed toShakespeare.127

The Hess chronology was an improvement over the older chronologies ofEva Turner Clark and the senior Ogburns. Like the Clark/Ogburn dating, theHess chronology backdates the plays far enough to squeeze them intoOxford’s lifetime but not enough to obliterate the trends. Hence, it looks tous like a forthright, first-cut response to the challenge we issued in our 1991Notes and Queries article. If nothing else, it provides us with the opportunityto take a closer look at the question of play chronology, to see how it looks onthe numbers, and to examine how much changing the dating would influencea major ascription controversy. This may be of pertinence not only to theOxfordian controversy but also to efforts by mainstream scholars like AlfredCairncross, much cited by Charlton Ogburn, Jr., to push back the dating ofShakespeare’s plays.

F. Lack of Oxfordian Consensus

Although both Oxfordian and mainline Stratfordian dating have alwaysbeen speculative, Oxfordian dating seems even more speculative and lesssettled than Stratfordian. Examining successive editions of the Riverside orcomparing the Riverside chronology to other leading contemporaryShakespeare chronologies, E.K. Chambers’s chronology, or nineteenth-century


128. Peter R. Moore, The Abysm of Time: The Chronology of Shakespeare’s Plays,5 ELIZABETHAN REV. 24, 25 (1997).

129. Compare Hess, supra note 73 at 40-57, with Appendix Ten. 130. Compare DOROTHY & CHARLTON OGBURN, THIS STAR OF ENGLAND: “WILLIAM

SHAKE-SPEARE” MAN OF THE RENAISSANCE (1952), with CLARK, supra note 112. 131. This means that the “roaring gaffes” that Hess claims we felt “obliged to correct

without comment or apology” on our website (i.e., by substituting Clark’s supposedly superiordates for the Ogburns’ supposedly inferior ones) would be less than roaring even if they weretrue, see Hess, supra note 73, at 39, and they are not true. “Clark” is the name of Terry Ross’sserver and has nothing to do with Eva Turner Clark. Terry Ross, not us, did the posting to hisown website, not ours, using our original Ogburn dates, not our amended Clark dates or anyoneelse’s. Clark’s and the Ogburns’ dates, in any case, are so close to each other that substitutingeither for the other could not have made any visible difference on our chart.

132. See OGBURN, supra note 2. 133. See Hess, supra note 73. 134. See Moore, supra note 128, at 36-60. 135. Id. at 24-60.136. Id at 27-28.

chronologies, reveals that the resemblances between different estimates seemhighly persistent. The order and dates of individual plays may differsomewhat from one estimate to another, but the same plays appear repeatedlyin the same broad classifications: early, middle, and late. As Peter Moore putit, “Chambers dead is stronger than his successors alive.”128

Thanks to Mr. Hess’s collection of different versions of Oxfordian datingover the years, it is possible to compare Oxfordian estimates with oneanother.129 Today these dates are much more scattered than they were in theearly days. There are gaps of ten to twenty years between some alternativeversions. The senior Ogburns’ dating turns out to be almost a carbon copy ofEva Turner Clark’s dating,130 with no two dates for the same play more thantwo years apart.131 However, the senior Ogburns’ tight consensus was littleheeded either by Charlton Ogburn, Jr.,132 Ron Hess,133 or Peter Moore,134 onwhom Hess relied for about half of his seventeen “anchor” dates.

Moore challenged conventional Stratfordian dating (specifically that ofChambers) as speculative and inconclusive and offered new backdates for tenplays, which he also acknowledged to be speculative and inconclusive.135 Hecriticized Francis Meres’s 1598 list as incomplete and Philip Henslowe’s “ne”entries, which affect two Shakespeare plays, as probably meaning somethingother than “new.”136 Both points seem plausible, though we doubt thatMeres’s known early-play omissions, The Taming of the Shrew and the HenryVI series, from a list of twelve to thirteen plays are enough to justifyabandoning him entirely as a point of reference. Considering the even-more-speculative alternatives, we think it makes more sense to take account of theimperfect play list than to ignore it just because it missed a play or two.Ultimately, Moore proposed the following tentative backdates: “TitusAndronicus, circa 1585; Comedy of Errors, 1587-88; King John, circa 1590;


137. Id. at 55. The Riverside provides the following dates for the plays listed by Moore:Titus Andronicus, from late 1594; Comedy of Errors, from 1594; King John, 1596; Romeo andJuliet, 1596; Henry IV, Part I, 1597; Henry V, 1599; As You Like It, 1599; Hamlet, 1601;Macbeth, 1606; Pericles, 1608. THE RIVERSIDE SHAKESPEARE, supra note 2, at 78-86.

138. Moore, supra note 128, at 55; see also id. at 43-44, 46 (discussing the difficulty ofdating King John, Henry IV, Part I, and Henry V).

139. Id. at 28.

Romeo and Juliet, 1591; Henry IV, Part I, by 1592; Henry V, 1592-99; AsYou Like It, 1593-94; Hamlet, ?1594; Macbeth, perhaps 1600-01; Pericles, by1604.”137 He immediately, and appropriately, added that “some of the piecesof evidence underpinning this list are strong, others are weak.”138 He alsoproperly observed, as Chambers had, that evidence of earliest possible datestends to be “weak stuff” compared to evidence of latest possible dates.139

Our inclination here, for the moment, is not to examine his evidence indetail. Instead, we shall take him at his word, note that he attempted only tenbackdates for thirty-eight plays, all tentative, and note also that the spreadbetween his tentative dates and those of the Riverside averages only five or sixyears, not the twelve or more years found in other Oxfordian dating. Ingeneral, we find Moore’s external evidence more cautious and morepersuasive than Hess’s or Clark’s, though no more persuasive than theconventional, Chambers-derived evidence that Moore criticized. We wouldalso guess from looking at Figures 3.1, 3.4, and 3.6 that some of his proposedbackdates, such as Titus Andronicus, The Taming of the Shrew, and TheComedy of Errors, might fit the conventional trendlines every bit as well as theconventional dates for these plays, maybe better. Major differences remainbetween Hess’s bold, comprehensive estimates and Moore’s cautious, limitedones, and between Hess’s and Moore’s estimates and the older Oxfordianestimates. Current consensus among Oxfordians, after eighty years of tryingto get the dates right, still seems to be anything but tight.

G. Problems with External Evidence

We have no idea whether Oxfordians will ever be as agreed on chronologyas they once were or as mainline Stratfordians seem to be now. Surely theanswer will and should depend largely on external evidence, a subject onwhich we have never pretended to be authorities. But we do not believe thatthe Hess chronology will be the last word on the subject. For example, evenan amateur studying Eva Turner Clark’s actual evidence that Oxford wroteRichard III in 1581 might have misgivings about making it an “Anchor Play”in any chronology, as Hess has done. Clark’s basis for imagining this date isthat Oxford was in the Tower of London in 1581, that there are similaritiesbetween Richard III and a letter Oxford wrote to his father-in-law in 1581, andthat Richard III makes more references to the tower than Shakespeare’s other


140. CLARK, supra note 112, at 416-17, 425.141. WELLS & TAYLOR, supra note 2, at 121.142. Id. at 133. 143. See Hess, supra note 73, at 50; Moore, supra note 128, at 55.144. See Moore, supra note 128, at 46-48.145. See id. at 47-48. 146. Id. at 47. 147. Id.148. WELLS & TAYLOR, supra note 2, at 132-33.149. Id. at 30 (providing a reproduction of Sir Henry Wotton’s letter).

history plays.140 Such “evidence” seems skimpy and far-fetched compared to,for example, the mainline dating of As You Like It at 1599. As You Like It isnot mentioned in Meres’s compendious (though not quite exhaustive) 1598 listof Shakespeare plays known to him, but it was “stayed” in the Stationer’sRegister on August 4, 1600, and its song, It was a lover and his lass, waspublished in Thomas Morley’s First Book of Airs, 1600.141 This evidencesuggests that As You Like It should not be dated before 1598 and certainly notafter 1600. Also convincing is the evidence for the 1613 dating of Henry VIII(All Is True), which was being performed when the Globe Theater burneddown on June 29, 1613 and reportedly was performed no more than two orthree times previously.142

Hess’s “Anchor Dates” appear in boldface in our Appendix Ten; we havealso bolded dates we consider better substantiated than most in our RiversideDate Late column. The most interesting cases are As You Like It and Henry V.We would anchor both plays in 1599 for essentially the reasons mentioned forAs You Like It: the play is not mentioned in Meres’s list, but it was found inthe Stationer’s Register and in other convincing references (such as a “badquarto” of Henry V) in 1600. Hess would anchor them, respectively, in 1593and 1592, following Moore.143 Moore’s evidence for both plays144 turns outto be extremely speculative. He argues that because As You Like It seems torefer to Marlowe’s death in 1593, it therefore must have been written shortlyafterward.145 We do not think this necessarily follows. Moore also argues that“Shakespeare’s reference to Essex in Ireland in 1599 [in Henry V] bears themarks of revision of an earlier text” and shows the play to have been arevival.146 He concedes, however, that he has “offered no positive evidencefor an earlier date for the play.”147

Hess and the other Oxfordians pay little attention to Shakespeare’s threereported collaborations with John Fletcher around 1613: the lost Cardenio,Henry VIII, and The Two Noble Kinsmen. The dates of the first two seembetter substantiated: recorded payments to the King’s Men for two 1613performances of “Cardenno,”148 and Sir Henry Wotton’s letter to Sir EdmundBacon describing the burning of the Globe Theater during the performance ofa “new Play, called All is true, representing some principal pieces of the reignof Henry VIII.”149 If these relatively well-documented collaborations withFletcher were done while Shakespeare was alive, as we believe collaborations


150. See Appendix Ten: Chronological Indicators in Shakespeare’s Plays, Riverside LateDating (“1st clear”-“Rdate Late” columns).

151. See id. (“1st clear”-“Oxfd Hess” columns). 152. These long lags do not ring true to us. It is almost like asking us to suppose that

Oxford, because he wished to hide his authorship of Shakespeare’s plays, must have hidden theplays for twelve years. Showbiz people do not often do that. Most people who write or produceshows want them performed for an audience, measure their success by how their shows performat the box office, and do what they can to get their plays noticed on opening night. Here inClaremont, just up the road from Hollywood and Disneyland, we hear tons about this season’sreleases, ounces, at most, about last season’s, and nothing at all about whatever was the ragetwelve years ago. Who would suppose that Elizabethans in showbiz, even if they were tryinghard to hide the authorship of new plays, would routinely keep their plays under wraps fortwelve years before the word leaked out and someone managed to get them registered, printed,reviewed, or recorded? It would be as if none of the dozens of films with screenplays covertlywritten by blacklisted Hollywood screenwriters in the Eisenhower years, such as the 1957 filmBridge on the River Kwai, received any lasting mention until Richard Nixon was president. Noone would expect such a thing to happen, and it did not.

153. Hess, supra note 73, at 34 (emphasis added).

generally were then and are now, it poses a grave problem for the claim forOxford, who died in 1604. If the collaboration were posthumous, it raises thesame question as Oxfordian ascriptions do generally: If Oxford wrote his halfof these plays in 1601, 1603, or “not later than 1592,” why did Fletcher waitten or twenty years to get the other half written?

Besides listing in Appendix Ten three of the Oxfordian chronologies tocompare with the Riverside, we have also included a column called “FirstClear Mention.” This column records the date when we consider a play tohave first been clearly identified, whether by Meres, by an entry in theStationer’s Register, by publication of a quarto version, or by a report bysomeone having attended the play or having heard about it. Comparing “firstclear mention” dates with the Riverside and Hess chronologies makes astriking contrast. Of the thirty-eight plays listed, thirty-four had a “first clearmention” other than the pertinent, sometimes decades-later Folio edition. Forthese thirty-four plays, the average lag between the Riverside estimated dateand the first clear mention is a year and a half.150 For the thirty-three of thesethirty-four plays dated by Hess, the average lag is eleven and a half years!151

For the Clark/Ogburn dating, the lag would be even longer.152

Where did Hess get all those eleven-and-a-half-year lags? In most cases,out of a hat, as he freely admits: “So, to preserve the monotonic stylisticcontinuum for [the half of the plays he did not “anchor”], wherever there areno better reasons to date a play we simply subtract twelve years from theElliott/Riverside date for that play.”153 His most remarkable feats ofbackdating—the backdating of most of the plays conventionally dated afterOxford’s death—are typically accomplished by this disarmingly simpleexpedient, without even a nod to external evidence. He simply marks the play


154. Id. at 55 (suggesting the likeliest dating for Coriolanus).155. WILLIAM SHAKESPEARE, THE TRAGEDY OF MACBETH act 2, sc. 2, line 64.156. See generally WELLS &TAYLOR, supra note 2, at 104-05 (tabulating “colloquialisms”

and “negative”-trending words with endings like –eth, which we call “archaisms” in AppendixTen).

“[s]tylistically, transferred from 1608 to 1596,”154 and the job is done. AsLady Macbeth said, “A little water clears us of this deed.”155

H. Problems with Internal Evidence

But has it? Let us acknowledge, again, that external evidence is oftenskimpy, tricky, and inconclusive; that we do not pretend to follow it closely;and that we have rarely been picky about claimant-advocates’ external-evidence suppositions, no matter how far-fetched. What then does the internalevidence say? In addition to summarizing the various chronologies, AppendixTen also provides a summary of the following various stylistic chronologicalindicators: feminine endings (FEs), open lines (OLs), midline speech endings(MLEs), light endings, weak endings, most’s per 10,000 words,colloquialisms, and archaisms.156 Every stylistic indicator except archaismsincreased during Shakespeare’s writing lifetime, conventionally reckoned.

I. Feminine Endings

Figure 3.1 illustrates the upward trend of feminine endings byconventional dating, from as low as 5% in the 1590s to as high as 35% in theearly 1600s. It is neither ruler-straight nor ruler-thin, but it is smooth enough,tight enough, and consistent enough that our Excel spreadsheet drew a nice,straight, slanting trendline (not pictured) that could be extrapolated downwardto cross zero at 1580. Such a line, as we have seen, might arguably spareOxford’s iambic pentameter poems from a rejection, though they do not havea single feminine ending because they were mostly written before the earliestof Shakespeare’s poems and plays. But conventional dating shows that theupward trend in FEs continued for almost a decade after Oxford’s death. Aswe noted in 1991, if true, this fact would be the worst of news for the Oxfordclaim.


157. The exception was Henry VIII, not marked in Figure 3.2 but Oxford-dated at 1601 or1603.

Figure 3.1. Feminine Endings, Shakespeare Plays, Riverside Late Dating

Figure 3.1. By conventional, Riverside dating, Shakespeare’s feminine-endingspercentages increased throughout his lifetime and continued to do so after the deathof the Earl of Oxford in 1604.

The old Clark/Ogburn Oxfordian backdating solved the posthumous trendproblem by trying to push every Shakespeare play but one157 into the 1570sand 1580s. As long as one suspends skepticism of their external evidence, thisbackdating would more than solve the problem of posthumous trends becauseit would obliterate every sign of a trend. Excel draws a horizontal trendline(not pictured) halfway up the cloud of FE percentages, and as shown in Figure3.2, the mismatch with Oxford’s rock-bottom FE percentage becomesimpossible to dodge.


Figure 3.2. Feminine Endings, Shakespeare Plays, Clark Late Dating

1575 1580 1585 1590 1595 1600 1605 1610 1615

0

40

35

30

25

20

15

10

5

wivoth

tmpwt

cym

cor

aylant mac

adotro

ham

movshrerr

r3 jc h5

aww

mfm

tim

r22h6

tit rommndjn 1h4

1h6

tn

lr

2h4

Oxford 1-5 Poems, 1572-1594: zero feminine endings

Figure 3.2. By early Oxfordian (Clark) dating, the rising trend in feminine endingsdisappears, and Oxford’s poems, with no feminine endings at all, become a clearmismatch with Shakespeare.

Likewise, the new Hess backdating solves the posthumous trend problemand creates something that looks like two trend lines: one slanting up, fromHenry VI, Part II to The Tempest; and one slanting down, from Henry VI, PartII to Henry IV, Part I. Again, the only trendline that Excel could manage todraw from these contradictory impulses (not pictured) is perfectly flat. Onceagain, Oxford’s mismatch with Shakespeare becomes impossible to discountwith trendline arguments. Like the Clark re-dating, the Hess re-datingstrengthens, not weakens, our Oxford rejections by this test.

Figure 3.3. Feminine Endings, Shakespeare Plays, Hess Dating

1575 1580 1585 1590 1595 1600 1605 1610 1615

0

40

35

30

25

20

15

10

5

wiv oth

tmpwt

cymcor

aylant mac

adotro ham

movshr err

r3 jch5

aww

mfm

r2

2h6

tit rommndjn

1h4

1h6

tn

lr

2h4

Oxford 1-5 Poems, 1572-1594: zero feminine endings

Figure 3.3. By late Oxfordian (Hess) dating, the rising trend in feminine endings is


obscured, and Oxford’s poems, with no feminine endings at all, become a clearmismatch with Shakespeare.

J. Open Lines

Similar conclusions might be made from open-line trends. As shown inFigure 3.4, Oxford’s iambic pentameter poems barely pass our Shakespearethreshold for open lines, and their low percentages are about what you mightexpect from backward-extrapolating Shakespeare’s open-line trendline byconventional dating. If Hess’s re-dating affects this conclusion at all, asFigure 3.5 illustrates, the backward-extrapolation overshoots Oxford’s poemsand makes them seem anomalously low compared to Shakespeare’s playssupposedly written at the same time. Again, if anything, it weakens theinternal evidence of possible common authorship. In this case, the Hess trendlooks clear to the eye, but the Excel-drawn trendline (not pictured) is still flat.

Figure 3.4. Open Lines, Shakespeare Plays, Riverside Dating Open Lines

1575 1580 1585 1590 1595 1600 1605 1610 1615

0

60

50

40

30

20

10

wiv

oth

tmpwt

h8sh

tnkshcymcor

ayl

ant

mac

ado

trohammov

shr

err

r3jc

h5

awwmfm tim

r2

2h6 titrommnd

jn

1h4

1h6

tn

lr

2h4

Oxford 1-5 Poems, 1572-1594: 7% open lines

Figure 3.4. By Riverside Shakespeare late dating, Oxford’s poems barely fit earlyShakespeare’s open-line profile; the rising trend in open lines continues afterOxford’s death.


158. See Hieatt et al., supra note 2, at 98.159. See Donald W. Foster, SHAXICON 1995, SHAKESPEARE NEWSL. (Iona College, New

Rochelle, NY), Summer 1995, at 1, 30, 32; Donald W. Foster, Shaxicon and Shakespeare’sActing Career: Reply to Diana Price, SHAKESPEARE NEWSL. (Iona College, New Rochelle,NY), Fall 1996, at 57-58; Donald W. Foster, Abstract of Shakespeare’s Career as Reconstructedby SHAXICON 2.0 (1994) (unpublished manuscript, on file with author). But see Charles W.Hieatt et al., Attributing A Funeral Elegy, 112 PUBL’NS MOD. LANGUAGE ASS’N AM. 429, 429-32 (1997) (questioning the reliability of Foster’s Shaxicon program); Diana Price, SHAXICONand Shakespeare’s Acting Career, SHAKESPEARE NEWSL. (Iona College, New Rochelle, NY),Summer 1996, at 27-28, 46 (challenging the accuracy of Foster’s Shaxicon dating forShakespeare by examining the external evidence of Shakespeare’s acting career).

Figure 3.5. Open Lines, Shakespeare Plays, Hess Dating

1575 1580 1585 1590 1595 1600 1605 1610 1615

0

60

50

40

30

20

10

wiv

oth

tmpwt

h8sh

tnkshcymcor

ayl

ant

mac

ado

tro ham mov

shrerr

r3

jc

h5

awwmfm

movr2

2h6 titrommnd

jn

1h4

1h6

tn

lr

2h4

Oxford 1-5 Poems, 1572-1594: 7% open lines

Figure 3.5. Hess’s re-dating makes Oxford’s 7% open lines look anomalously low.

We shall spare the reader plots of comparable tests using light endings,weak endings, mosts, colloquialisms, and archaisms. They generally repeatthe lessons taught by FEs and OLs—clear upslanting trends (downslanting forarchaisms) by Riverside dating, most of them continuing after Oxford’s death,and not-so-clear trends by Oxfordian dating, none continuing after Oxford’sdeath. We shall also pass over some very interesting studies by Kent andCharles Hieatt and Anne Lake Prescott158 and by Donald Foster159 thatconclude from vocabulary overlap that some of Shakespeare’s poems musthave been written well into the seventeenth century. Instead of examiningthese studies, we shall close with some charts of midline speech endings,which by conventional dating increased from as low as 1% in the 1590s toover 90% in the 1600s. In Figure 3.6, the Riverside MLE upward trend lookstight, smooth, and steep; Excel drew us a nice, steep, slanted line (notpictured). In Figure 3.7, the Hess trend looks clearly upward to the eye, butmuch less tight and smooth. Excel again drew a flat, equivocal trendline (notpictured).


Figure 3.6. Midline Speech Endings, Shakespeare Plays,Riverside Late Dating

1575 1580 1585 1590 1595 1600 1605 1610 1615

0

100

90

80

70

60

50

40

30

20

10

wiv

oth

tmpwt

h8sh

tnksh

cymcor

ayl

antmac

adotro

ham

mov

shrerr

r3

jch5

aww

mfm

tim

r2

2h6 tit

rommndjn 1h4

1h6

tn

lr

2h4

Figure 3.6. Under Riverside dating, midline speech endings form a tight, steep,smooth uptrend throughout Shakespeare’s life, continuing after Oxford’s death.

Figure 3.7. Midline Speech Endings, Shakespeare Plays, Hess Dating

1575 1580 1585 1590 1595 1600 1605 1610 1615

0

100

90

80

70

60

50

40

30

20

10

wiv

oth

tmpwt

cymcor

ayl

ant mac

ado

tro

ham

mov

shrerrr3

jc h5

aww

mfm

timr2

2h6 tit

rommndjn 1h4

1h6

tn

lr

2h4

Figure 3.7. Midline speech endings trends under Hess dating are more scattered andthus less clear than under Riverside dating.

K. Midline Speech Endings

What if we had no external evidence at all but wanted to guess thesequence of the plays purely from one strong stylistic trend? Simply countingand ranking each play’s percentage of MLEs would produce a sequence that


160. Two of the three exceptions are from jointly written plays. The three exceptions areAll’s Well That Ends Well (MLE percentage is six places later than Riverside), Shakespeare’spart of Pericles (MLE percentage is four places earlier), and Shakespeare’s part of Henry VIII(MLE percentage is seven places earlier).

161. Compare Appendix Ten (Riv. Seq. column), with Appendix Ten (MLE Seq. column).162. Compare Appendix Ten (Riv. Seq. column), with Appendix Ten (Hess Seq. column).

in only three cases differs from the Riverside sequence by more than threeplaces.160 In other words, only 8% of the 38 MLE-percentage-ranked playsdiffered from the Riverside sequence by more than three rank places.161 Thesame exercise relative to the Hess sequence would produce eighteen suchanomalies,162 about half of the thirty-four plays Hess dated.

Such comparisons, of course, rest on the conjecture that one apparenttrend, under one set of assumptions, can actually serve as independentevidence of a sequence. We consider this conjecture more plausible thanHess’s conjecture that arbitrarily lopping off twelve years from half theRiverside dates and calling it a “stylistic transfer” will improve their accuracy.Neither we, nor any Oxfordian scholar we know, has found as tight, steep, orsmooth a trendline for any other indicator, under any set of Oxfordianassumptions, as we have found for the eight indicators treated here underStratfordian assumptions.

If the Hess dates were any stronger on external evidence than the Riversidedates, sequencing comparisons might reveal that MLE percentages, thoughtighter, smoother, and steeper with Riverside dates, nonetheless make for a badranking. Where the Riverside dates seem (1) stronger and more consistentwith other Stratfordian chronologies, as they do to us, and (2) freer from suchgross counterindicators as the Fletcher collaborations and the supposed eleven-year gap between the average play’s debut and the first clear mention of it,MLE percentages are yet another indicator that Oxfordians still have work todo to date the plays satisfactorily.

L. Conclusions on Chronology

The new Hess dating seems less formless and relentlessly confined toantiquity than the old Clark/Ogburn dates, and somewhat more systematic andattentive to comparative perspectives and external evidence. But Oxfordiandates still seem more scattered than Stratfordian, less well founded in externalevidence, and much more loosely and haphazardly sequenced, as measured bytightness, smoothness, and steepness of internal indicators. The blank spot forthe Fletcher collaborations, the wholesale lopping off of twelve years from theRiverside dates, and the long implied gap between opening night and firstmention, all seem like severe drawbacks and imply that there is much room forfurther improvement in Oxfordian dating. Taken at face value, Oxfordianbackdating does avoid the problem of play trends continuing well past


163. See Elliott & Valenza, Oxford Candidacy, supra note 74. 164. See Karl N. Llewellyn, Remarks on the Theory of Appellate Decision and the Rules

or Canons About How Statutes Are to Be Construed, 3 VAND. L. REV. 395, 401-06 (1950)(listing the “thrusts” and “parries” for statutory interpretation arguments).

Oxford’s death, but only by compounding the dissimilarities between Oxford’spoems and Shakespeare’s backdated “contemporary” plays.

After years of augmentation and refinement, our stylometric tests stillshow that Louis Bénézet’s inference that Oxford’s style was all butindistinguishable from Shakespeare’s was dead wrong. When one uses acomputer to test sizeable blocks, it is anything but indistinguishable. Oxfordfailed four of our six available tests in 1990. Now he fails seven of fifteentests, many more than the most errant like-sized block in our core Shakespearepoem baseline, and enough to put him and Shakespeare in different statisticalgalaxies. Four of the seven Oxford rejections are not time-sensitive and arenot affected at all by the proposed re-dating. The other three are time-sensitive, but the new Oxfordian backdating (though generally better than theold ones) still makes for stronger rejections than Stratfordian dating becausethe rejections make Shakespeare’s plays look more contemporary thanOxford’s poems and Shakespeare’s poem-mismatches with Oxford appearmore glaring. We think the Shakespeare Clinic has removed one seriousobjection to the Oxford candidacy by showing that A Funeral Elegy was notwritten by Shakespeare,163 but the Clinic’s overall effect has been much moreto show differences between Shakespeare and Oxford than to showresemblances.

M. Possible Discounts

Are there any counter-arguments left for Oxfordians or others to deflector discount our findings? Did not Karl Llewellyn, in a well-known article inthe Vanderbilt Law Review, counsel that for every thrust of statutoryinterpretation there must always be a parry?164 If, like the two Oxfordianpanelists, we were pretending to prosecute Oxford for being the TrueShakespeare, we could easily think of some parries we might use, but we arenot so sure that any of them would convince a jury. The prosecution couldalways plead novelty. Our methods were new and experimental; they still are.However, they have evolved over the years, and we have continued to discardsome tests and modify others. We would be the last to suppose that thisprocess has stopped or that what we have arrived at today will be the last wordtomorrow. On the other hand, after fifteen years of availability for refutation,the first rounds of criticism knocked out only one of our tests (among many),and subsequent rounds have barely changed our results at all. Under thecircumstances, we can hardly help feeling a bit less tentative than we did in1990 or 1994.

The prosecution could also ask for what amounts to a dirtier baseline than


165. See supra note 103 and accompanying text.166. None of these “Oxford Apocrypha” come close to fitting Shakespeare’s profiles. See

Elliott & Valenza, There Were None, supra note 103, at 214, 240; Elliott & Valenza, TrueShakespeare?, supra note 92, at 503.

167. See Elliott & Valenza, There Were None, supra note 103, at 207-10 (listing additionaltypes of tests that could have been utilized and noting the “cautions” and “caveats” associatedwith the testing process).

ours, one that includes material that we consider doubtful or co-authored. Ithas been done. If we went along with it, it would probably expand some ofour needle’s-eye Shakespeare profiles enough to get a camel or two throughthem. Or maybe they could try to shrink the camel by demanding theinclusion of more “Oxford Apocrypha,” such as the play Horestes, or thepoems of “Meritum Petere Grave” or other such posies in George Gascoigne’sA Hundreth Sundrie Flowres, claimed as Oxford’s by one Oxfordian scholaror another, in hopes that one of them might pass our tests; but our tests revealthat such hopes are pretty dim.165 They lengthen the already long chain ofspeculative evidence and, more often than not, make the camel bigger, notsmaller.166

Departing from the prosecutorial model in which prosecutors normallycarry the burden of proof beyond a reasonable doubt, critics could ask us tomeet a heavier burden of proof, citing the same limitations of our tests that weacknowledge in Part II and the “Baseline” and “Remarks” columns of Table3.1, and rejecting our proof unless and until we do more tests on more texts.More appropriately, though, they could perform the extra tests themselves andpossibly justify different conclusions. Are Shakespeare’s 45,000 words ofpoems enough of a baseline for a convincing test (“A” and “C” under“Shakespeare Baseline” in Table 3.1), or do you also need the 246,000 wordsof play verse (“B”)? Or is an even larger play verse baseline needed because“B” does not include all of Shakespeare’s play verse? “Time” has not turnedout to be a big discount relative to Oxford, so far at least, but what about“editing,” “prosody,” or even “subject-matter?” These are only a few of thepossible limitations. We can think of more, and we can think of other tests wemight have tried.167 It would take more time and work than we considerappropriate, considering the one-sidedness of the evidence we already have,but any of these tests might be a good starting point for a critique of our work.We note, however, that these, too, have been available for fifteen years andthat none of our critics has pursued them far enough to begin to make asuccessful challenge to our results.

N. Grub or Butterfly?

The last-resort parry, and in our view, the closest thing to a defense left forthe Oxford candidacy against evidence like ours, would be to do what at leastone Oxfordian writer has already done: concede our evidence, that Oxford’s


168. See Shahan, supra note 74, at 154-55. “Grub” was a Shakespeare equivalent of“Caterpillar.” WILLIAM SHAKESPEARE, THE TRAGEDY OF CORIOLANUS act 5, sc. 4, line 12.

169. See OGBURN, supra note 2, at 390-93 (advancing a similar argument). 170. Email from Stephanie Hughes, Editor, OXFORDIAN, to Ward E.Y. Elliott (Jan. 8,

2000) (on file with author). 171. Shahan, supra note 74, at 154. Oxfordians have argued, in effect, that Shakespeare

had a Blue Period, exemplified, for example, by Oxford’s “Who taught thee first to sigh, alas,my heart?,” poem Number Fifteen, see May, supra note 9, at 37, and a Guernica period,exemplified by Venus and Adonis (1593), The Rape of Lucrece (1594), and the Sonnets (1590s-1600s). But see Terry Ross, The Verse Forms of Shakespeare and Oxford, athttp://shakespeareauthorship.com/verform.html (last visited Jan. 21, 2005) (questioning theadequacy of this type of argument with respect to these and other poems). But the argumentweakens when one examines the actual timing. Picasso’s Blue Period ran from 1901 to 1904,when he was in his early twenties, and his Blue work, while far distant in style from what camelater, was still manifestly the work of a master. Unlike Oxford’s “help fish, help fowl” lines,poem Number Four, see May, supra note 9, at 28, The Old Guitarist could never be describedas a stumbling, apprentice work. By the time he did Guernica, in 1937 at fifty-six, Picasso hadnot done anything blue, good or bad, for thirty-three years. By contrast, Oxford’s great leap tostylistic maturity and master-level work, if there was one, had to take place in his mid-fortiesand virtually overnight, another case of drastically punctuated evolution, from stumbling baby

style, as we know it, is indeed in a different galaxy from Shakespeare’s, butreject our conclusion that he therefore probably is not a Shakespeare “could-be.”168 Why not instead suppose that the observed differences are moredevelopmental than essential, more like those between a grub and a butterflythan like those between a sow’s ear and a silk purse? Would not Oxford’swobbling baby steps be just what we would expect of someone as young andold-fashioned as Oxford was before blazing forth from his cocoon as theimmortal Shakespeare? Would it not, in fact, help solve the mystery of howShakespeare managed to start out a full-grown butterfly with no sign of everhaving taken baby steps or having been a journeyman grub? In sum, the verybaby-step stylistic discrepancies with Shakespeare that we observe in Oxford’swork should enhance his claims, rather than diminish them.169

As Stephanie Hughes, editor of The Oxfordian, stated:

We’re not dealing with just any writer here, but a genius on the scaleof Leonardo da Vinci or Mozart. Early Mozart can be confused withHaydn, late Mozart with Beethoven . . . . If we didn’t know for a factthat Picasso had a Blue Period, if all we knew of him was his workfrom Cubism on, we’d never believe that those early works were his.When we read anything from the seventies and compare it withShakespeare, and note the immense growth and changes in onlytwenty years, we can hardly expect that tests that compare early workswith late works will give a meaningful result.170

In John Shahan’s words, “developmental factors may account for any stylisticdifferences between Oxford’s early poems and Shakespeare’s mature ones.”171


steps in 1593 to practiced giant’s strides afterward.It is worth mentioning that, in response to 1990 Oxfordian assertions of the grub-butterfly

argument, we did try out our then-new modal test (which Oxford had failed badly then, as later)on two other writers with large, firmly dated bodies of poetry: Milton and Spenser. Milton’searliest poems (before 1633) and his later poem, Samson Agonistes (1670-71), both fit withina profile set by Paradise Lost (1658-65). Spenser’s Epigrams and Sonnets (1569) and hisAmoretti (1595) closely matched his Shepherd’s Calendar (1579), but his Faerie Queene (1590,1598) tested very distant from the other four works mentioned. One might expect this resultbecause Spenser took pains to write it in a contrived, archaic style as different on its face fromthe rest of his poems as the French passages in Henry V are from the rest of Shakespeare’splays. See Elliott & Valenza, Reply to John Shahan, supra note 74, at 157-58. “As far as wecan tell from these improvised tests [which used Shakespeare-optimized keywords, notkeywords optimized for Milton or Spenser], Milton was a butterfly all his life, and so wasSpenser—except when he wrote the Faerie Queene.” Elliott & Valenza, A Touchstone for theBard, supra note 92, at 206.

172. See Ross, supra note 171.173. May, supra note 9, at 37 (poem Number Fifteen).174. Ross, supra note 171.175. Id.176. May, supra note 9 (poems Six, Nine, Ten, and Twelve).

Such arguments that differences, no less than similarities, can help provecommon authorship—or at least stave off disproof—are hard to refute directly.Certainly there is no shortage of differences in Oxford’s case. By someestimates (other than Steven May’s), Oxford could have been as young asfifteen when he wrote the eight poems eventually published in the 1576 work,The Paradyse of Daynty Deuises. Any or all of them could be song lyrics, notpoems proper, and, hence, not suitable for comparison with poems. TerryRoss has noted that more than half of Oxford’s known poems are in meters notfound in the Shakespeare Canon.172 Only one of his poems, “Who taught theefirst to sigh, alas, my heart?”173 is a sonnet, and even that has an “echo” foundnowhere in Shakespeare.174 None of his poems are in blank verse,Shakespeare’s favorite verse form, or “rhyme royal” (ababbcc seven-lineiambic pentameter (I-7)), the form used in The Rape of Lucrece.175 Strictlyspeaking, only four of Oxford’s sixteen poems176 match anything inShakespeare’s known work (Venus and Adonis). Stated differently, two-thirdsof Oxford’s known verse has no structural parallel at all in Shakespeare; theother third matches no more than 2-4% of Shakespeare’s verse. In structuralterms, the two poets have as much in common as Picasso’s 1903 The OldGuitarist and his 1937 Guernica.

What could more firmly demonstrate Oxford’s primitiveness and lack ofsuitability for comparison with the mature Shakespeare? If one wanted aclincher, one could do what we almost never admit to doing—not just crunch


177. Id. at 28 (poem Number Four). 178. WILLIAM SHAKESPEARE, THE TRAGEDY OF KING LEAR act 5, sc. 3, lines 122-23. 179. The Oxford passage has three further stylistic quirks that we suspect distinguishes him

from Shakespeare: (1) his wailfulness, (2) his occasional odd combination of plural subject andsingular verb, and (3) his heavy doses of alliteration. We have spot-checked these against thefirst 3,000 words of Venus and Adonis and found perhaps ten times as many wailful passagesin our Oxford baseline as in Shakespeare, twice as much alliteration, and three instances ofplural subject, singular verb. We found no such plural-subject/singular-verb usage in the Venusand Adonis block but would be interested if Oxfordians could find any such examples elsewherein Shakespeare. This evidence is no more than suggestive because alliteration and wailfulnessare not always easy to count and because we have compared Oxford’s work to only one of ourfourteen Shakespeare poem blocks. But if differences from Shakespeare help prove commonauthorship, under the grub-butterfly argument, the pickings for Oxfordians could be very, veryrich, especially the articles by Alan Nelson and Steven May mentioned in these pages.

Oxford’s poems, but actually read them. Consider, for example, this passagefrom Oxford:

Helpe gods, helpe, saintes, helpe sprites and powers, that in theheaven doo dwell,Helpe ye that are to waile aye woent, ye howling hounds of hell;Helpe man, helpe beastes, helpe birds and wormes, that on theearth doth [sic] toile,Helpe fishe, helpe foule, that flockes [sic] and feedes [sic] upon the salte seasoil;Helpe eccho that in ayre dooth flee, shrill voices to resound,To waile this losse of my good name, as of these greefes theground.FINIS E.O.177

Contrast this with Shakespeare’s treatment of the same subject—loss ofgood name:

Know, my name is lost,By treason’s tooth bare-gnawn and canker-bit . . . .178

How could anyone suppose that the two passages were written by the sameperson? We seldom rely on such comparisons because the texts (just likeLouis Bénézet’s) are seldom selected at random, but more often are chosen toillustrate whatever point the writer is trying to make. Bénézet chose forsimilarity with Shakespeare; we chose for contrast. In this case, the evidenceis exactly opposite to Bénézet’s. The styles seem to be worlds apart, withShakespeare’s manifestly more polished and mature. Shakespeare managedto capture in eleven tight, vivid, lapidary words of iambic pentameter much thesame thought that took the struggling young Oxford seventy-nine sprawling,repetitious, overwrought, ungrammatical words of rhyme royal to convey.179

But Oxfordians like Ogburn drew the same conclusion from the apparent


180. Shahan, supra note 73, at 164.181. See generally STEPHEN JAY GOULD, THE STRUCTURE OF EVOLUTIONARY THEORY

(2002) (arguing that under his theory of punctuated equilibria, evolutionary trends do not resultfrom gradual transformations but from rapid spurts of evolutionary development).

stylistic mismatch that they once drew from a perceived stylistic match; thevery immaturity of Oxford’s writing is evidence that Shakespeare thereforemight have been Oxford after all, only grown-up. If you unconditionallyaccept the premise that the young Shakespeare must have been a grub,Oxford’s many and great differences from Shakespeare do not damage hisclaim to be the True Shakespeare at all. On the contrary, they support theclaim by showing that the young Oxford looks every bit the journeyman grubthat the True Young Shakespeare must have been.

But there are serious problems with this argument. One is that, even morethan the current farfetched Oxfordian play dating, this argument is pulled froma hat and rests not on any actual evidence on the record, but on the hope thatthe jury would take judicial notice of a wholly conjectural, and to us wildlyimprobable, scenario. Oxfordian John Shahan came all too close to theessence of this problem when he reproached us for our fancied resemblanceto the drunk who cannot find his keys, looks for them under a streetlight, andexplains to a passerby, “I don’t know where I lost them . . . [b]ut this is wherethe light is.”180 If looking where the light is amounts to a sin, we and AlanNelson and Steven May, along with the longstanding Oxford skeptics, IrvinMatus, Terry Ross, and David Kathman, are the greatest of sinners. Indeed,we are worse sinners than the others for using all that fancy night-vision gearso scorned by lit-department technophobes. However, our tests do help yousee where the light would otherwise be dim.

By the light of the documents, Shakespeare looks much more like theStratford man than the Earl of Oxford. Oxford’s poems do not scan likeShakespeare’s nor sound like Shakespeare’s. By the numbers, Oxford’spoems are in a different galaxy from Shakespeare’s, and they show no signsof testable stylistic change from his earliest poems at age twenty-two or earlierto his latest, at age forty-four. Unlike the young Mozart or the young Picasso,Oxford tests like a grub (and, to us, sounds like a grub) from beginning to end,showing no stylistic changes at all on any of our tests. If he abruptly leaptfrom a distant galaxy into Shakespeare’s infield and morphed from acaterpillar into a butterfly the very year he adopted the name Shakespeare in1593, it would be an extraordinary case of what Stephen Jay Gould calls“punctuated development,”181 not just in one of his previously-frozen stylistichabits, but in seven of them at once. This number would increase to nine, ifwe count the two close-call tests, where what we know about Oxford is amismatch with only 94% of what we know about Shakespeare, not the 95-100% we required for a full rejection. It would be, beyond doubt, thegrandmother of all mid-life crises. What jury would believe it withoutsufficient evidence, especially if six out of the nine changed habits froze


immediately into fixed new profiles for the rest of Shakespeare’s life, and thethree that continued to change went on doing so for years after Oxford’sdeath? Under the streetlight, Oxford was a caterpillar from beginning to end,and Shakespeare was a butterfly, as different from Oxford as the Beatles werefrom Vic Damone.

Finally, even if a few ounces of discrepancy could enhance a claim forcommon authorship, it does not follow that a ton of it would make the claimeven stronger. The grub defense does two awkward things for the Oxfordclaim: (1) it applies equally to other older-generation Shakespeare claimants,such as Sir Edward Dyer, and (2) it marks a huge tacit shift from LouisBénézet’s old-line Oxfordian arguments that Oxford and Shakespeare werestylistic look-alikes. The new Oxfordian argument is that maybe they were notstylistic look-alikes at all, but it does not matter because Oxford might easilyhave grown into a look-alike. What counts is no longer what Oxford wrote,which you can see all too clearly in the lamplight, but how his life experiences(the very ones that Alan Nelson and Steven May, the world’s leaders inOxford documents, say are mismatches) compared to those depicted inShakespeare.

As noted in Part II, we have generally stayed out of such controversies,which fall under the heading of conventional, literary, and historical“smoking-gun” evidence. But surely the net effect of the grub parry is tomove the defendant, Oxford, out of the category of “testable” claimants likeMarlowe, who have actually written something legitimately comparable toShakespeare. As a grub-designate, Oxford falls instead into the category of“untestable” claimants like the Rosicrucians and the Earls of Derby andRutland, from whom no poems or plays have survived. No amount ofstylometric testing can confirm or deny claims based on what the untestableclaimant might have written. But such claims, absent any comparablesupporting writing, seem far more speculative than those of the despisedWilliam Shakspere of Stratford.

Unlike Alan Nelson, Steven May, and the three named Oxford skeptics,we have done nothing to prove directly that William Shakspere wrote the playsprinted under William Shakespeare’s name. Instead, we have brought forwardstrong statistical evidence that the claims of the Earl of Oxford and the otherclaimants we tested are astronomically improbable. How this might withstandstrong, contrary documentary evidence—if there were any—can only beimagined. Late in the Tennessee symposium, moderator Judy Cornett askedthe panelists what kind of hypothetical evidence they would considerirrefutable. Everyone answered “a manuscript,” except Elliott, the incorrigiblesmoking-gun skeptic, who facetiously (and anachronistically because notariesdid not exist in Shakespeare’s time) suggested a signed, sealed, sworn,notarized affidavit by Shakespeare that either he or some other named partywrote the plays. If such a document materialized, would it be irrefutable?What if it named Mark Twain or the Pope as the “True Author”? Wouldanyone believe it? What if it named William Shakspere of Stratford as theTrue Author? Would any Oxfordian believe it? Or would they see in it fresh


evidence of the virulence of the conspiracy to hide the True Author, adding,if rightly understood, yet another conspiracy-confirming ornament to theirChristmas tree? We would not bet against it. We remain mindful ofLlewellyn’s talk of thrusts and parries, and we suspect that the wait forconclusive, smoking-gun documentary evidence to settle this dispute will bea long one.

IV. CONCLUSIONS

In the meantime, we have chosen to cast down our buckets where we areand pay attention to what you can see under the streetlight. We would urge thesame to any court or individual trying Oxford or anyone else for being theTrue Shakespeare. Surely the contours of what we know about Shakespeareand Oxford should be well considered before we venture too deeply intospeculation about what we do not know. What we know of them fromexternal evidence, from the two leading Oxford documents scholars in theworld, is that they do not match. What we know just from looking at Oxford’spoems is that they do not match Shakespeare’s. Oxford used different metricalstructures, different rhyme schemes, different imagery. On a cursory readingand scanning of their respective poems and on the numbers, our tests clearlyshow that Oxford and Shakespeare are light years apart. The odds that eithercould have written the other’s work by chance are much lower than the oddsof getting hit by lightning. Oxfordian efforts to fix discrepancies by pullingnew, older dates for the plays out of a hat have not helped. Nor has theOxfordian’s heroic, unsubstantiated conjecture of Oxford’s nine great stylisticmid-life leaps from a distant frozen galaxy right into Shakespeare’s ballparkat a single bound helped Oxford’s case. The thought of anyone making somany intergalactic conjectural leaps in a single year seems wildly implausiblecompared to whatever conjectural leaps are required to take the Stratford caseseriously—such as the notion that, being legally entitled to do so, the youngShakespeare might actually have attended the Stratford grammar school. It ishard to imagine any jury buying the Oxfordians’ colossal mid-life crisisargument without much more substantiation than it has received. In the end,the grub to the butterfly argument remains too grossly at odds with what youcan see, all too plainly, under the streetlight: far too many things “don’t fit”for Oxford to be a plausible claimant. We would acquit.


APPENDICES

APPENDIX ONE PLAY DISCRIMINATION SUMMARIESAPPENDIX TWO THREE ROUND PLAY TEST SUMMARIESAPPENDIX THREE ROUND ONE PLAY TESTSAPPENDIX FOUR ROUND TWO PLAY TESTSAPPENDIX FIVE ROUND THREE PLAY TESTSAPPENDIX SIX 3,000-WORD POEM & PLAY VERSE TESTSAPPENDIX SEVEN 1,500-WORD POEM & PLAY VERSE TESTSAPPENDIX EIGHT 750-WORD POEM & PLAY VERSE TESTSAPPENDIX NINE 470-WORD POEM TESTSAPPENDIX TEN CHRONOLOGICAL INDICATORS IN SHAKESPEARE PLAYS

398

APPENDIX ONE: KEY TO OVERALL PLAY DISCRIMINATION SUMMARIES

Column Heading Meaning Shakespeare Profiles*

Play Long title of play.

Short Title Short title of play.

Date Latest supposed date of composition.

Number of Words Number of words in play.

Discrete Rejections Number of Shakespeare rejections in 46 to 48 tests.

0 to 2

Discrete Composite Probability

Mathematical probability that the observed num-ber of rejections would occur by chance at Shake-speare’s baseline rejection rate. Profile minimum, right, is for Shakespeare’s most discrepant block.

>2.316E-01

Continuous Composite Probability

Mathematical probability that observed continu-ous composite error would occur by chance. Pro-file minimum, right is for Shakespeare’s most discrepant, least probable block. See text.

>3.6895E-03

Highlighting Conventions

The right three columns of Appendix One show composite results only and are shaded yellow = composite. Outliers—that is, rejections for plays in the Shakespeare baseline, non-rejections for all others—are highlighted in red or gray. That is, for Shakespeare baseline: Red or gray = outside Shakespeare profile. For all others: Red or gray = inside Shakespeare profile. See text.

* Composite Shakespeare profiles may also be found at the bottom of most tables in the Appendices.

SUMMARY OF RESULTS

Individual Discrete Rejections: Of 29 plays in our Shakespeare core baseline, only 7 have as many as 2 “dis-crete” Shakespeare rejections in 48 individual tests. Of 51 plays by Shakespeare claimants, none has fewer than 10 Shakespeare rejections. Of 27 plays in the Shakespeare Apocrypha, none has fewer than 7 rejections. Discrete and Continuous Composite Rejection Scores: By either of these scoring systems, all of the Claimant and Apocrypha plays are many orders of magnitude less likely to have been written by Shakespeare solo than any Shakespeare baseline play (see text and Key to Table Two). No play in the Shakespeare core baseline has even one rejection by any of the three composite tests used. No red- or gray-shaded anomalies are found in any of the baseline Shakespeare, Claimant, or Apocrypha plays tested. Dubitanda and Set-asides: The only red or gray shading anywhere in Appendix One is found, unsurprisingly, in the Shakespeare Dubitanda and set-aside section, which consists of plays which Shakespeare is suspected of co-authoring. Most of these selections, including several conventionally ascribed to Shakespeare, also seem very distant from Shakespeare. These look co-authored or other-authored. But a few are only slightly outside Shakespeare’s composite boundaries. These could well be by him alone or almost alone: 2H6; H5; Per (3-5); and maybe STM (Sh) fall into this category.

399

Appendix One (Shakespeare): Core Shakespeare Play Baseline Discrimination Summaries

Play Short TitleLatest

Supposed Date

Number of Words

Discrete Rejections



Richard III R3 1593 28473 0 1.000E+00 8.9791E-01The Taming of the Shrew (c) Shr 1594 20496 1 6.018E-01 4.4275E-01Two Gentlemen of Verona (c) TGV 1594 16952 1 6.018E-01 6.2283E-02The Comedy of Errors (c) Err 1594 14438 1 6.018E-01 4.2266E-01Richard II R2 1595 21896 2 2.316E-01 3.3822E-02Love's Labor's Lost (c) LLL 1595 21168 2 2.316E-01 4.2905E-01King John Jn 1596 20462 0 1.000E+00 1.3478E-01A Midsummer Night's Dream MND 1596 16164 2 2.316E-01 6.7391E-01Romeo and Juliet Rom 1596 24070 0 1.000E+00 9.9445E-01Henry IV, Part I 1H4 1597 24140 1 6.018E-01 9.5223E-01The Merry Wives of Windsor (Wiv 1597 21279 2 2.316E-01 1.2563E-02The Merchant of Venice MoV 1597 21002 1 6.018E-01 9.0359E-01Henry IV, Part II 2H4 1598 25829 1 6.018E-01 6.1270E-01Julius Caesar JC 1599 19187 1 6.018E-01 4.6432E-01Much Ado About Nothing (c) Ado 1599 20861 1 6.018E-01 9.9880E-03As You Like It (c) AYL 1599 21381 2 2.316E-01 3.9824E-01Hamlet Ham 1601 29673 2 2.316E-01 5.0606E-02Twelfth Night (c) TN 1602 19493 1 6.018E-01 6.8504E-01Troilus and Cressida Tro 1602 25640 0 1.000E+00 8.4691E-01Measure for Measure MFM 1603 21361 0 1.000E+00 9.6541E-01All's Well That Ends Well (c) AWW 1603 22585 0 1.000E+00 8.8350E-01Othello Oth 1604 25982 1 6.018E-01 2.9374E-01King Lear Lr 1605 25371 0 1.000E+00 9.0747E-01Macbeth Mac 1606 16194 1 6.018E-01 5.3985E-01Anthony and Cleopatra Ant 1607 23815 0 1.000E+00 9.5866E-01Coriolanus Cor 1608 26639 1 6.018E-01 4.2183E-01Cymbeline Cym 1610 26861 1 6.018E-01 9.9501E-01The Tempest Tmp 1611 16149 2 2.316E-01 3.6895E-03A Winter's Tale WT 1611 24680 0 1.000E+00 9.7333E-01

Shakespeare Core Profile Bounds 2 2.316E-01 3.6895E-03

Note : (c) indicates light comedy

Summary of Results . None of our 29 core Shakespeare plays has more than two rejections in 48 tests or scores outside either of our composite probability ranges. No claimant play or Shakespeare Apocrypha play has fewer than seven rejections or scores anywhere near our composite probability ranges. Results for our Shakespeare dubitanda and set-asides are not so clear, possibly because co-authored passages have not been completely separated, and probably also because some samples are too short and variable for standards based on whole, presumptively single-authored plays. Some Dubitanda samples are closer to Shakespeare than others, but most show some sign of joint authorship.

400

Appendix One (Claimants) -- Claimant Play Discrimination Summaries

Playwright PlayShort Title

Latest Supposed

Date

Number of Words

Discrete Rejections




Beaumont, Francis* The Knight of the Burning Pestle PESL 1607 21006 16 <1.000E-15 <1.0000E-15Chapman, George* The Gentleman Usher USHR 1602 20956 15 8.438E-15 <1.0000E-15Chapman, George* Bussy D'Ambois BUSS 1607 19787 16 <1.000E-15 <1.0000E-15Daniel, Samuel Cleopatra CLEO 1593 13612 18 <1.000E-15 <1.0000E-15Dekker, Thomas The Whore of Babylon WBAB 1607 20267 18 <1.000E-15 <1.0000E-15Dekker, Thomas Honest Whore HNWR 1608 23240 19 <1.000E-15 <1.0000E-15Fletcher, John The Woman's Prize WPRZ 1604 22983 14 2.092E-13 <1.0000E-15Fletcher, John Valentinian VALN 1610 24634 13 2.380E-12 <1.0000E-15Fletcher, John Monsieur Thomas MTOM 1616 20238 13 4.355E-12 <1.0000E-15Fletcher, John Chances CHNC 1617 16211 18 <1.000E-15 <1.0000E-15Fletcher, John The Loyal Subject LOYL 1618 25458 20 <1.000E-15 <1.0000E-15Fletcher, John Demetrius and Enanthe DEMT 1619 24130 14 1.075E-13 <1.0000E-15Fletcher, John Sir J.V.O. Barnavelt BARN 1619 21537 13 2.380E-12 <1.0000E-15Fletcher, John The Island Princess ISLN 1619 22483 19 <1.000E-15 <1.0000E-15Greene, Robert Alphonsus ALPH 1587 15072 19 <1.000E-15 <1.0000E-15Greene, Robert Friar Bacon & Friar Bungay FBFB 1591 16227 19 <1.000E-15 <1.0000E-15Greene, Robert James IV JAM4 1591 19872 16 <1.000E-15 8.3495E-15Heywood, Thomas A Woman Killed with Kindness HEYW 1603 16242 11 1.376E-09 1.6337E-06Jonson, Ben Sejanus SEJA 1603 25954 13 4.355E-12 <1.0000E-15Jonson, Ben Volpone VOLP 1606 26528 13 4.355E-12 <1.0000E-15Jonson, Ben The Alchemist ALCH 1610 26944 18 <1.000E-15 <1.0000E-15Jonson, Ben Bartholomew Fair BART 1614 35859 17 <1.000E-15 <1.0000E-15Jonson, Ben The New Inn NINN 1629 22028 14 2.092E-13 <1.0000E-15Jonson, Ben A Tale of a Tub TTUB 1633 8315 22 <1.000E-15 <1.0000E-15Kyd, Thomas The Spanish Tragedy SPTR 1589 20316 17 <1.000E-15 <1.0000E-15Lyly, John The Woman in the Moon LYWM 1597 12300 21 <1.000E-15 <1.0000E-15Marlowe, Christopher Tamburlaine TAM1 1588 17205 26 <1.000E-15 <1.0000E-15Marlowe, Christopher Tamburlaine, pt. 2 TAM2 1588 18122 21 <1.000E-15 <1.0000E-15Marlowe, Christopher Doctor Faustus, 1616 DF16 1588 16140 11 1.376E-09 <1.0000E-15Marlowe, Christopher The Jew of Malta JEWM 1589 17994 13 4.355E-12 <1.0000E-15Marlowe, Christopher Edward II EDW2 1592 21104 10 1.337E-08 1.2780E-07Marlowe, Christopher The Massacre at Paris MAPA 1593 10353 14 1.075E-13 <1.0000E-15Marlowe, Christopher Dido, Queen of Carthage DIDO 1586 13726 19 <1.000E-15 <1.0000E-15Middleton, Thomas The Phoenix PHOE 1604 19347 13 4.355E-12 <1.0000E-15Middleton, Thomas Michaelmas Term MICL 1606 19357 20 <1.000E-15 <1.0000E-15Middleton, Thomas A Chaste Maid Cheapside CHST 1611 16906 22 <1.000E-15 <1.0000E-15Middleton, Thomas No Wit Like a Woman's NWIT 1613 25463 17 <1.000E-15 <1.0000E-15Middleton, Thomas More Dissemblers MDIS 1615 18845 21 <1.000E-15 <1.0000E-15Middleton, Thomas The Witch WITC 1616 15859 18 <1.000E-15 <1.0000E-15Middleton, Thomas Hengist/Mayor of Queenboro HENG 1618 19507 13 4.355E-12 <1.0000E-15Middleton, Thomas Women Beware Women WBWM 1621 25135 18 <1.000E-15 <1.0000E-15Middleton, Thomas A Game at Chess GAME 1624 17670 18 <1.000E-15 <1.0000E-15Munday, Anthony John a Kent and John a Cumber JKJC 1594 13412 14 2.092E-13 1.1574E-10Nashe, Thomas Will Summer's Last Will & Testa. WILL 1592 16577 15 8.438E-15 <1.0000E-15Peele, George The Arraignment of Paris ARPA 1584 10209 19 <1.000E-15 <1.0000E-15Peele, George David and Bethsabe DBET 1594 14748 23 <1.000E-15 <1.0000E-15Pickering, John Horestes HORE 1567 11841 17 <1.000E-15 <1.0000E-15

401

Appendix One (Claimants) -- Claimant Play Discrimination Summaries

Playwright PlayShort Title

Latest Supposed

Date

Number of Words

Discrete Rejections




Porter, Henry Two Angry Women of Abingdon ANWO 1598 25473 17 <1.000E-15 <1.0000E-15Sidney Herbert, Mary Antonius (extract) ANTO 1590 2234 29 <1.000E-15 <1.0000E-15Smith, Wm. (Wentworth) The Hector of Germany HECT 1615 15224 10 2.072E-08 8.1949E-15Wilson, Robert Three Ladies of London 3LDY 1581 16949 12 4.724E-11 4.3662E-07

Discrimination Summary Non-Shakespeare discrete rejections mean 17

Rejections of 51 plays tested 51 51 51 Rejection percentage 100 100 100

*Note : Beaumont and Chapman are not claimants; all others are.Yellow indicates a composite score or standard.

Summary of Results . No claimant play play has fewer than ten rejections or scores anywhere near our composite Shakespeare probability ranges, either by discrete or by continuous testing.

402


Supposed Date

Number of Words

Discrete Rejections




Dubitanda and set-asidesHenry VI, Part I 1H6 1590 20595 11 1.376E-09 <1.0000E-15

Henry VI, Part 2 2H6 1591 24533 3 6.309E-02 2.724E-01

Henry VI, Part 3 3H6 1591 23402 8 3.252E-06 3.411E-07

Henry V H5 1599 25788 2 2.316E-01 3.258E-11

Henry VIII (Fletcher's part) H8, (Fl) 1613 7158 15 8.438E-15 <1.0000E-15

Henry VIII (joint part) H8, (Jt) 1613 3986 18 <1.000E-15 <1.0000E-15

Henry VIII (Shakespeare's part) H8, (Sh) 1613 11953 9 2.768E-07 5.234E-07

Pericles, Acts 1-2 Per, 1-2 1608 7839 15 8.438E-15 <1.0000E-15

Pericles, Acts 3-5 Per (3-5) 1608 9907 3 6.309E-02 6.864E-03

Timon of Athens Tim 1608 17704 13 4.355E-12 <1.0000E-15

Two Noble Kinsmen (Fletcher's part) TNK (Fl) 1613 14668 17 <1.000E-15 <1.0000E-15

Two Noble Kinsmen (Sh's part) TNK (Sh) 1613 14528 4 1.305E-02 4.207E-06

Titus Andronicus Tit. 1594 19835 7 3.323E-05 1.840E-05

Titus Andronicus, early stratum Tit early 1594 10609 15 8.438E-15 <1.0000E-15

Titus Andronicus, late stratum Tit late 1594 7789 9 2.768E-07 2.267E-08

Sir Thomas More (Sh's part) STM (Sh) 1595 1382 21 <1.000E-15 <1.0000E-15

Discrimination Summary

Non-Shakespeare discrete rejections mean 11

Rejections of 16 works tested 15 15 14 Rejection percentage 94 94 88

ApocryphaHorestes HORE 1567 11841 17 <1.000E-15 <1.0000E-15Famous Victories of Henry V FVH5 1588 12496 16 <1.000E-15 <1.0000E-15Taming of a Shrew TOAS 1589 12214 15 8.438E-15 <1.0000E-15Ironside IRON 1590 15037 12 8.165E-11 <1.0000E-15Arden of Feversham ARDN 1592 19453 10 2.072E-08 5.316E-14Contention of York, Part 1 YRK1 1592 16149 14 2.092E-13 3.060E-10Contention of York, Part 2 YRK2 1592 17011 17 <1.000E-15 <1.0000E-15Guy of Warwick GUYW 1593 12731 19 <1.000E-15 <1.0000E-15Leir LEIR 1594 21062 8 3.252E-06 <1.0000E-15Richard III RCD3 1594 19506 15 8.438E-15 <1.0000E-15Sir Thomas More STMO 1595 19509 7 3.323E-05 <1.0000E-15Edward III EDW3 1595 19395 13 4.355E-12 2.639E-12King John, Part 1 KJN1 1595 14141 14 2.092E-13 2.063E-11King John, Part 2 KJN2 1595 9646 16 <1.000E-15 1.504E-09Locrine LOCR 1595 15440 22 <1.000E-15 <1.0000E-15Woodstock WOOD 1595 25009 20 <1.000E-15 <1.0000E-15Mucedorus MUCE 1598 11739 11 1.376E-09 <1.0000E-15Sir John Oldcastle OLDC 1600 20823 11 1.376E-09 4.862E-10Lord Thomas Cromwell CROM 1602 13866 13 4.355E-12 3.365E-11The Merry Devil of Edmonton DEVL 1604 11588 11 1.376E-09 <1.0000E-15The London Prodigal PROD 1605 15656 16 <1.000E-15 <1.0000E-15The Puritan PURN 1607 18477 19 <1.000E-15 <1.0000E-15

Appendix One (Apocrypha) --Shakespeare Dubitanda and Apocrypha Play Discrimination Summaries

403


Supposed Date

Number of Words

Discrete Rejections




A Yorkshire Tragedy YKSH 1608 5722 14 2.092E-13 <1.0000E-15The Second Maiden's Tragedy MAID 1611 18454 22 <1.000E-15 <1.0000E-15Double Falsehood FALS 1613 15429 11 1.376E-09 <1.0000E-15Faire Em FAIR 1631 11529 22 <1.000E-15 <1.0000E-15The Birth of Merlin MERL 1631 18393 11 1.376E-09 <1.0000E-15The Revenger's Tragedy* RVGR 1606 19690 19 <1.000E-15 <1.0000E-15

Discrimination Summary Non-Shakespeare Discrete Rejections Mean 15 Rejections of 28 works tested 28 28 28 Rejection percentage 100 100 100

*The Revenger's Tragedy is not considered part of the Shakespeare Apocrypha; the others are.

Yellow shading indicates a composite score or standard.Gray shading indicates a discrete rejection count that unexpectedly lies inside of the designated Shakespeare rejection profile.Red shading indicates a value that unexpectedly lies inside of the designated Shakespeare probabilities profile.

Summary of Results . All 27 Apocrypha plays, and The Revenger's Tragedy fall decisively outside Shakespeare's profile. Most of our Shakespeare dubitanda and set-asides show some signs of co- or other-authorship. Dubitanda results are less conclusive than results for Claimants and Apocrypha, probably because the passages are, in fact, co-authored or other-authored and the other author's work not perfectly segregated from Shakespeare's. Some samples, such as STM Sh andH8 (jt) are too short and variable for composite rejections based on whole, presumptively single-authored plays to mean much.

Appendix One (Apocrypha) --Shakespeare Dubitanda and Apocrypha Play Discrimination Summaries

404

APPENDIX TWO: KEY TO 3-ROUND PLAY TEST SUMMARIES

Column Heading Meaning Shakespeare Profiles

Play Short title of play.


Discrete Rejections Number of Shakespeare rejections in up to 48 tests. 0 to 2

Discrete Rejection Rate

Percentage of Shakespeare (Sh.) rejections (here 27) of all observed test scores on Shakespeare baseline (here 1,392).

1.9%


Abstract probability that the observed number of rejections would occur by chance at Shakespeare’s baseline rejection rate. Profile minimum shown at right is measured for Shakespeare’s most discrep-ant, least probable block.

>2.316E-01

Continuous Composite Error

Composite distance from composite mean on every test. See text.


Abstract probability that observed continuous com-posite error would occur by chance. Profile mini-mum at right is measured for Shakespeare’s most discrepant, least probable block. See text.

>3.6895E-03

Relative Shakespeare Probability

Shakespeare’s minimum Composite Probability score (see Shakespeare profiles, right column) di-vided by sample text’s Composite Probability score.

>1.0

Highlighting Conventions

Individual tests for all: no highlight = no rejection, inside Sh. profile. aqua = rejection, outside profile. Composite results for all: yellow = composite, not individual, score. For Sh. base: Red or gray = out-side Sh. profile, unexpected comp. rejection. For all others: Red or gray = inside Sh. profile, unex-pected non-rejection

SUMMARY OF RESULTS

Simple Rejections: Of 29 plays in our Shakespeare core baseline, only 7 have as many as 2 Shakespeare rejections in 48 tests. Of 51 plays by Shakespeare claimants, none has fewer than 10 Shakespeare rejec-tions. Of 27 plays in the Shakespeare Apocrypha, none has fewer than 7 rejections. Relative Discrete Rejection Odds: The odds of 7 rejections taking place by chance, at Shakespeare’s av-erage baseline rejection rate—3.323 x 10-5—are almost 7,000 times lower than the odds for Shake-speare’s own most discrepant baseline plays—2.316 x 10-1. The odds of 10 rejections are 11 million times lower. Relative Continuous Rejection Odds: The closest claimant play to Shakespeare by this test is HEYW, with a continuous composite probability (CCP) of 1.6337 x 10-6. 1.6337 x 10-6 is 2,255 times lower than Shakespeare’s lowest-probability play, The Tempest; hence, even HEYW is very unlikely to be by Shake-speare. Forty-nine of the fifty-one Claimant plays tested have CCPs too low to compute, <1 x 10-15. Their relative Shakespeare probability, at best, is about 370 billion times lower than that of The Tempest.

405

Shak

espe

are

Pla

y B

asel

ine

Dat

a, T

hree

-Rou

nd C

ompo

site

Sco

res

Rou

nd O

neR

ound

Tw

oR

ound

Thr

eeT

hree

-Rou

nd C

ompo

site

s

Play

Dat

e L

ate

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

nsC

ontin

uous

C

ompo

site

Err

orN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Con

tinuo

us

Com

posi

te E

rror

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

nsC

ontin

uous

C

ompo

site

Err

orN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ility

(1

.9%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

R3

1593

170

3.59

5316

03.

3005

150

3.49

6448

01.

000E

+00

6.00

378.

9791

E-0

1Sh

r (c

)15

9417

03.

3345

160

4.72

0315

13.

9176

481

6.01

8E-0

16.

9820

4.42

75E

-01

TG

V (

c)15

9417

13.

8947

160

5.20

6615

04.

6471

481

6.01

8E-0

17.

9920

6.22

83E

-02

Err

(c)

1594

170

3.50

3016

04.

1528

151

4.44

2948

16.

018E

-01

7.01

834.

2266

E-0

1R

215

9517

26.

6600

160

3.11

5715

03.

6498

482

2.31

6E-0

18.

2088

3.38

22E

-02

LL

L (

c)15

9517

15.

1498

160

2.91

4515

13.

7522

482

2.31

6E-0

17.

0067

4.29

05E

-01

Jn15

9617

04.

6784

160

4.17

9415

04.

4201

480

1.00

0E+

007.

6741

1.34

78E

-01

MN

D (

c)15

9617

14.

0330

161

3.85

8915

03.

4544

482

2.31

6E-0

16.

5642

6.73

91E

-01

Rom

15

9617

03.

3023

160

3.14

8915

02.

4353

480

1.00

0E+

005.

1721

9.94

45E

-01

1H4

1597

170

3.42

9816

13.

6228

150

2.83

5848

16.

018E

-01

5.73

859.

5223

E-0

1W

iv (

c)15

9717

05.

8133

161

5.10

5915

13.

5630

482

2.31

6E-0

18.

5182

1.25

63E

-02

MoV

1597

171

4.52

0116

02.

4490

150

3.05

8248

16.

018E

-01

5.98

189.

0359

E-0

12H

415

9817

02.

4853

160

4.91

9115

13.

7726

481

6.01

8E-0

16.

6788

6.12

70E

-01

JC15

9917

14.

2579

160

3.51

5415

04.

2099

481

6.01

8E-0

16.

9434

4.64

32E

-01

Ado

(c)

1599

170

5.42

0016

04.

0941

151

5.24

8848

16.

018E

-01

8.58

429.

9880

E-0

3A

YL

(c)

1599

171

4.33

7716

04.

2498

151

3.60

6948

22.

316E

-01

7.06

303.

9824

E-0

1H

am16

0117

03.

7082

161

5.43

7615

14.

6672

482

2.31

6E-0

18.

0685

5.06

06E

-02

TN

(c)

1602

170

3.38

0216

04.

2941

151

3.59

7348

16.

018E

-01

6.54

266.

8504

E-0

1T

ro16

0217

03.

0306

160

3.29

0415

04.

2502

480

1.00

0E+

006.

1705

8.46

91E

-01

MFM

1603

170

3.67

9516

03.

6298

150

2.25

5648

01.

000E

+00

5.63

939.

6541

E-0

1A

WW

(c)

16

0317

04.

3521

160

2.80

9015

03.

1369

480

1.00

0E+

006.

0557

8.83

50E

-01

Oth

16

0417

13.

8977

160

5.33

5715

03.

0237

481

6.01

8E-0

17.

2667

2.93

74E

-01

Lr

1605

170

4.06

4416

02.

6985

150

3.43

4448

01.

000E

+00

5.96

639.

0747

E-0

1M

ac16

0617

03.

8803

160

3.22

2115

14.

5750

481

6.01

8E-0

16.

8095

5.39

85E

-01

Ant

1607

170

3.48

4216

03.

7680

150

2.46

4648

01.

000E

+00

5.69

329.

5866

E-0

1C

or16

0817

03.

7449

161

4.67

7315

03.

6573

481

6.01

8E-0

17.

0198

4.21

83E

-01

Cym

1610

170

2.27

1516

02.

9010

151

3.59

5748

16.

018E

-01

5.14

839.

9501

E-0

1T

mp

1611

171

5.54

3116

15.

0321

150

4.72

3348

22.

316E

-01

8.85

203.

6895

E-0

3W

T16

1117

02.

7265

160

3.68

5015

03.

1534

480

1.00

0E+

005.

5639

9.73

33E

-01

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Dis

cret

e R

ejec

tions

27 C

omp

Rej

00

Perc

enta

ge2%

0%0%

Blo

cks

Tes

ted

1,39

229

29

406

Cla

iman

ts v

ersu

s Sh

akes

pear

e B

asel

ine,

Thr

ee-R

ound

Com

posi

te S

core

s

Rou

nd O

neR

ound

Tw

oR

ound

Thr

eeT

hree

-Rou

nd C

ompo

site

s

Play

Dat

e L

ate

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

nsC

ontin

uous

C

ompo

site

Err

orN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Con

tinuo

us

Com

posi

te E

rror

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

nsC

ontin

uous

C

ompo

site

Err

orN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ility

(1

.9%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Bea

umon

t: PE

SL16

0717

46.

1509

168

10.8

081

154

5.28

4048

16<

1.00

0E-1

513

.511

8<

1.00

00E

-15

Cha

pman

U

SHR

1602

176

7.34

0316

67.

5246

153

18.7

230

4815

8.43

8E-1

521

.472

1<

1.00

00E

-15

B

USS

1607

175

8.90

5016

98.

9708

152

4.64

2448

16<

1.00

0E-1

513

.465

8<

1.00

00E

-15

Dan

iel:

CL

EO

1593

178

10.2

965

163

15.8

001

157

12.3

658

4818

<1.

000E

-15

22.5

516

<1.

0000

E-1

5D

ekke

r

WB

AB

1607

176

8.43

0916

715

.585

415

58.

8969

4818

<1.

000E

-15

19.8

278

<1.

0000

E-1

5

HN

WR

1608

177

9.06

1816

530

.133

415

76.

6256

4819

<1.

000E

-15

32.1

565

<1.

0000

E-1

5Fl

etch

er

WPR

Z16

0417

610

.733

716

716

.672

815

14.

7280

4814

2.09

2E-1

320

.385

0<

1.00

00E

-15

V

AL

N16

1015

16.

2970

166

14.8

560

156

6.48

8046

132.

380E

-12

17.3

910

<1.

0000

E-1

5

MT

OM

1616

177

14.1

974

164

9.09

9215

26.

4928

4813

4.35

5E-1

218

.069

8<

1.00

00E

-15

C

HN

C16

1715

710

.453

716

715

.376

615

46.

2723

4618

<1.

000E

-15

19.6

229

<1.

0000

E-1

5

LO

YL

1618

158

8.73

5616

824

.769

815

46.

6647

4620

<1.

000E

-15

27.0

974

<1.

0000

E-1

5

DE

MT

1619

156

7.86

0716

515

.987

715

35.

4294

4614

1.07

5E-1

318

.624

6<

1.00

00E

-15

B

AR

N16

1915

37.

0467

167

11.8

648

153

4.96

3846

132.

380E

-12

14.6

652

<1.

0000

E-1

5

ISL

N16

1915

79.

4736

168

17.3

603

154

6.30

0746

19<

1.00

0E-1

520

.756

4<

1.00

00E

-15

Gre

ene

A

LPH

1587

178

10.7

996

166

29.7

590

155

18.6

488

4819

<1.

000E

-15

36.7

425

<1.

0000

E-1

5

FBFB

1591

179

10.1

298

163

8.04

1715

76.

8444

4819

<1.

000E

-15

14.6

331

<1.

0000

E-1

5

lAM

415

9117

56.

1470

164

5.31

0215

79.

9686

4816

<1.

000E

-15

12.8

591

8.34

95E

-15

Hey

woo

d: H

EY

W16

0317

36.

1582

166

6.81

1415

24.

8661

4811

1.37

6E-0

910

.392

21.

6337

E-0

6Jo

hnso

n

SEJA

1603

175

7.99

3516

56.

2245

153

9.44

7848

134.

355E

-12

13.8

528

<1.

0000

E-1

5

VO

LP

1606

174

8.17

5716

617

.472

015

311

.248

148

134.

355E

-12

22.3

301

<1.

0000

E-1

5

AL

CH

1610

176

13.4

228

166

42.2

479

156

6.95

2048

18<

1.00

0E-1

544

.870

7<

1.00

00E

-15

B

AR

T16

1417

510

.267

716

851

.558

515

46.

1398

4817

<1.

000E

-15

52.9

283

<1.

0000

E-1

5

NIN

N16

2917

513

.287

016

429

.544

815

55.

8452

4814

2.09

2E-1

332

.918

1<

1.00

00E

-15

T

TU

B16

3317

515

.727

316

878

.732

815

98.

0772

4822

<1.

000E

-15

80.6

935

<1.

0000

E-1

5K

yd: S

PTR

1589

178

8.64

9216

24.

6251

157

11.4

267

4817

<1.

000E

-15

15.0

589

<1.

0000

E-1

5L

yly:

LY

WM

1597

1710

10.2

884

166

5.35

0815

56.

3417

4821

<1.

000E

-15

13.2

174

<1.

0000

E-1

5M

arlo

we

T

AM

115

8817

1212

.116

216

55.

7899

159

10.2

345

4826

<1.

000E

-15

16.8

841

<1.

0000

E-1

5

TA

M2

1588

1711

11.4

670

164

5.38

1915

69.

5213

4821

<1.

000E

-15

15.8

465

<1.

0000

E-1

5

DF1

615

8817

36.

7075

162

10.2

607

156

11.2

148

4811

1.37

6E-0

916

.614

5<

1.00

00E

-15

JE

WM

1589

172

5.86

6216

512

.531

915

65.

8610

4813

4.35

5E-1

215

.027

1<

1.00

00E

-15

E

DW

215

9215

55.

9676

162

5.45

5615

36.

9311

4610

1.33

7E-0

810

.649

71.

2780

E-0

7

MA

PA15

9315

68.

0442

161

6.81

5315

79.

0498

4614

1.07

5E-1

313

.894

5<

1.00

00E

-15

D

IDO

1586

1510

9.16

5416

37.

5014

156

26.3

062

4619

<1.

000E

-15

28.8

495

<1.

0000

E-1

5M

iddl

eton

PH

OE

1604

173

7.80

2816

729

.185

715

36.

8466

4813

4.35

5E-1

230

.976

9<

1.00

00E

-15

M

ICL

1606

176

8.60

1216

1038

.809

915

45.

7446

4820

<1.

000E

-15

40.1

645

<1.

0000

E-1

5

CH

ST16

1117

711

.063

816

1126

.344

315

44.

9763

4822

<1.

000E

-15

29.0

033

<1.

0000

E-1

5

NW

lT16

1317

59.

6017

1610

30.5

349

152

5.12

6948

17<

1.00

0E-1

532

.417

0<

1.00

00E

-15

M

DIS

1615

178

8.34

4316

927

.316

815

47.

2850

4821

<1.

000E

-15

29.4

772

<1.

0000

E-1

5

WIT

C16

1617

49.

3154

168

20.1

227

156

7.21

5348

18<

1.00

0E-1

523

.318

6<

1.00

00E

-15

H

EN

G16

1817

36.

5766

167

10.1

719

153

5.76

6548

134.

355E

-12

13.4

154

<1.

0000

E-1

5

WB

WM

1621

176

8.69

4416

928

.960

715

35.

2429

4818

<1.

000E

-15

30.6

888

<1.

0000

E-1

5

GA

ME

1624

176

11.4

011

166

24.0

085

156

6.74

8648

18<

1.00

0E-1

527

.421

5<

1.00

00E

-15

Mun

day:

JK

JC15

9417

67.

0042

163

6.99

0215

56.

3440

4814

2.09

2E-1

311

.754

51.

1574

E-1

0

407

Cla

iman

ts v

ersu

s Sh

akes

pear

e B

asel

ine,

Thr

ee-R

ound

Com

posi

te S

core

s

Rou

nd O

neR

ound

Tw

oR

ound

Thr

eeT

hree

-Rou

nd C

ompo

site

s

Play

Dat

e L

ate

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

nsC

ontin

uous

C

ompo

site

Err

orN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Con

tinuo

us

Com

posi

te E

rror

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

nsC

ontin

uous

C

ompo

site

Err

orN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ility

(1

.9%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Nas

he: W

ILL

1592

1710

15.0

467

163

6.25

4515

23.

8703

4815

8.43

8E-1

516

.748

2<

1.00

00E

-15

Peel

e

AR

PA15

8417

1111

.314

816

26.

4076

156

9.70

9748

19<

1.00

0E-1

516

.228

4<

1.00

00E

-15

D

BE

T15

9417

1217

.334

016

56.

0020

156

7.26

5948

23<

1.00

0E-1

519

.730

3<

1.00

00E

-15

Pick

erin

g: H

OR

E15

6715

68.

6511

165

17.6

199

156

20.1

372

4617

<1.

000E

-15

28.1

213

<1.

0000

E-1

5Po

rter

: AN

WO

1598

174

10.9

344

164

15.0

509

159

8.82

1048

17<

1.00

0E-1

520

.588

9<

1.00

00E

-15

Sidn

ey: A

NT

O15

9017

1620

.610

816

68.

1244

157

11.7

404

4829

<1.

000E

-15

25.0

729

<1.

0000

E-1

5Sm

ith: H

EC

T16

1517

47.

8092

163

8.43

2515

35.

7713

4810

2.07

2E-0

812

.860

78.

1949

E-1

5W

ilson

: 3L

DY

1581

157

7.91

8516

24.

1843

153

5.40

5146

124.

724E

-11

10.4

607

4.36

62E

-07

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Dis

cret

e R

ejec

tions

856

Com

p R

ej51

51Pe

rcen

tage

35%

100%

100%

Blo

cks

Tes

ted

2,42

651

51C

ompo

site

Thr

esho

lds

2.31

6E-0

13.

6895

E-0

3

Sh D

iscr

ete

Rej

ectio

n Pr

ofile

(Se

e no

te in

key

) M

inim

um0

Max

imum

2

408

APPENDIX THREE: KEY TO TABLES FOR ROUND ONE PLAY TESTS NEW-TECH TESTS, HYPHENATED WORDS

Column Heading and Test Sensitivities

Meaning Shakespeare Profile

Buckets: G Semantic buckets score in standard errors from the Shakespeare mean.

-2 to +2

Slope Thisted-Efron Slope Test score. -0.13 to +0.06

Rare Words: G Thisted-Efron Rare Words Test score. -2 to +89

New Words Thisted-Efron New Words Test score. -14 to +5

Grade Level: G, E Reading grade level score. 4 to 7

Fem Endings: T, P Percentage of feminine endings of verse lines. All figures are computer counts, which are generally lower and less accurate than manual counts.

Early (to 1597): 8 to 17 mid: (‘97-‘04): 8 to 20 late: (1605 on) 17 to 22

Open Lines: T, E, P Percentage of open or run-on verse lines. All figures are computer counts, which are generally comparable to manual counts.

early: 11 to 23 mid: 16 to 32 late: 31 to 50

HC/20K: E Hyphenated compound words per 20,000 words. 52 to 180

no / (no + not) Ratio of the number of occurrences of no to that of no plus not combined, times 1000.

242 to 358

it (lws) Rate of occurrence for it as the last word of a sen-tence (per 1,000 sentences).

8 to 30

with (2lws) Rate of occurrence for with as the penultimate word of a sentence (per 1,000 sentences).

9 to 21

it (fw) Rate of occurrence for it as the first word of a sen-tence (per 1,000 sentences).

7 to 18

the (2lws) Rate of occurrence for the as the penultimate word of a sentence (per 1,000 sentences).

30 to 63

BoB 1–7 Bundles of badges, 1 to 7. See text for components. (see following table)

Round One Rejections Total number of rejections, this round. 0 to 1

NOTES

See Key to Appendix Two. For individual tests, all rejections are shaded aqua. Composite scores and ranges are shaded yellow except for outlier scores—rejections for core Shakespeare baseline, non-rejections for all other categories—which are shaded red or gray. The profile boundaries are given at the bottom of the tables.

SUMMARY OF RESULTS Only one of 29 Shakespeare core plays, The Comedy of Errors, has as many as two individual rejections in this round of 17 tests. Of 79 claimant and apocrypha plays, only two have fewer than two rejections. Test Sensitivities: G = genre; T = time of composition; E = editing; P = prosody.

409

Shak

espe

are

Pla

y B

asel

ine

Dat

a, R

ound

One

Tes

ts

Play

Dat

e L

ate

Gra

de

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

with

(2lw

s)no

/(

no+

not)

it (

lws)

it (

fw)

the

(2lw

)B

oB1

BoB

3B

oB5

BoB

7T

-E S

lope

T

est

T-E

Rar

e W

ords

T-E

New

W

ords

Agg

rega

teB

ucke

tsN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ilit

y (2

%)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abil

ity

R3

1593

585

1617

1231

116

953

425

-109

358

688

-0.0

137

5-0

.95

170

1.00

0E+

003.

5953

7.41

11E

-01

Shr

(c)

1594

482

1511

1728

713

951

641

8015

943

4-0

.05

55-3

1.49

170

1.00

0E+

003.

3345

8.50

32E

-01

TG

V (

c)15

945

7617

1616

287

129

4357

778

120

497

0.06

632

0.04

171

2.90

7E-0

13.

8947

5.83

33E

-01

Err

(c)

1594

575

1612

1426

124

1061

575

3123

646

6-0

.04

631

0.54

170

1.00

0E+

003.

5030

7.83

46E

-01

R2

1595

782

923

1031

79

746

269

-174

435

779

-0.0

422

2-3

.01

172

4.45

9E-0

26.

6600

3.03

49E

-04

LL

L (

c)15

955

133

914

1130

017

1570

549

108

308

726

-0.1

311

-11

1.29

171

2.90

7E-0

15.

1498

6.54

86E

-02

Jn15

966

939

2318

264

1612

4636

8-9

238

272

4-0

.04

251

1.97

170

1.00

0E+

004.

6784

1.89

09E

-01

MN

D (

c)15

965

124

815

1532

98

1147

520

5227

290

0-0

.10

19-1

0.66

171

2.90

7E-0

14.

0330

5.05

13E

-01

Rom

15

964

113

916

1428

311

1041

328

-419

260

2-0

.02

49-4

-1.6

717

01.

000E

+00

3.30

238.

6145

E-0

11H

415

975

137

1229

1230

719

957

386

1226

452

4-0

.05

37-5

0.36

170

1.00

0E+

003.

4298

8.14

24E

-01

Wiv

(c)

1597

418

017

1619

303

2312

4075

824

716

453

4-0

.03

63-1

4-0

.38

170

1.00

0E+

005.

8133

8.92

40E

-03

MoV

1597

564

1327

1229

422

1477

671

130

321

679

-0.0

165

0-0

.23

171

2.90

7E-0

14.

5201

2.52

77E

-01

2H4

1598

510

514

2717

324

1615

4656

246

337

632

-0.0

138

-60.

2217

01.

000E

+00

2.48

539.

9192

E-0

1JC

1599

443

1522

1326

617

1250

568

111

407

723

0.05

773

-0.4

417

12.

907E

-01

4.25

793.

8070

E-0

1A

do (

c)15

994

6519

1919

341

3012

4269

019

720

262

00.

0489

2-1

.44

170

1.00

0E+

005.

4200

3.12

09E

-02

AY

L (

c)15

995

6718

2319

358

1812

4061

216

128

669

20.

0065

0-3

.53

171

2.90

7E-0

14.

3377

3.39

22E

-01

Ham

1601

570

1527

1130

819

1653

605

4039

745

4-0

.09

28-8

1.43

170

1.00

0E+

003.

7082

6.84

66E

-01

TN

(c)

1602

479

1723

2135

117

939

598

147

215

486

-0.0

268

-30.

1017

01.

000E

+00

3.38

028.

3360

E-0

1T

ro16

024

8315

2614

298

167

4450

270

318

431

-0.0

915

-9-0

.16

170

1.00

0E+

003.

0306

9.34

29E

-01

MFM

1603

559

2030

2128

721

1248

700

156

348

429

-0.0

359

-1-0

.28

170

1.00

0E+

003.

6795

6.99

44E

-01

AW

W (

c)

1603

560

1932

1229

425

1736

656

144

267

397

-0.0

358

-1-1

.39

170

1.00

0E+

004.

3521

3.31

94E

-01

Oth

16

044

4717

2416

242

1514

4367

812

119

927

8-0

.01

53-2

-0.1

017

12.

907E

-01

3.89

775.

8165

E-0

1L

r16

054

106

1931

1735

611

741

449

929

532

7-0

.01

21-8

0.51

170

1.00

0E+

004.

0644

4.87

36E

-01

Mac

1606

410

418

3518

291

1112

5343

2-2

141

137

7-0

.05

1-9

0.13

170

1.00

0E+

003.

8803

5.91

40E

-01

Ant

1607

452

1741

1629

710

737

427

3133

039

7-0

.05

29-7

0.24

170

1.00

0E+

003.

4842

7.91

61E

-01

Cor

1608

557

1846

2031

814

847

552

5348

730

3-0

.02

35-4

-1.6

617

01.

000E

+00

3.74

496.

6538

E-0

1C

ym16

105

7520

4713

320

1711

4154

553

285

394

-0.0

218

0-1

.13

170

1.00

0E+

002.

2715

9.97

28E

-01

Tm

p16

114

126

2246

635

79

931

317

-17

180

441

-0.0

86

-6-0

.68

171

2.90

7E-0

15.

5431

2.15

61E

-02

WT

1611

511

821

4815

294

239

4461

766

278

375

-0.0

537

-50.

4317

01.

000E

+00

2.72

659.

7721

E-0

1

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ecti

ons

02

00

10

00

21

01

10

00

210

11

Per

cent

age

0%7%

0%0%

3%0%

0%0%

7%3%

0%3%

3%0%

0%0%

7%2%

3%3%

493

2929

Pla

ys B

asel

ine:

Sta

ndar

d St

atis

tica

l Pro

file

Com

posi

te T

hres

hold

s B

lock

s T

este

d29

2929

2929

2929

2929

2929

2929

2929

2929

2.90

7E-0

18.

9240

E-0

3 M

ean

4.69

88.2

815

.66

26.4

115

.10

305.

0016

.52

10.9

047

.14

537.

1459

.52

291.

4852

7.90

-0.0

341

.59

-3.1

4-0

.26

Std

Dev

0.70

31.1

73.

8410

.65

3.54

28.9

65.

322.

769.

6212

4.19

89.5

489

.41

157.

220.

0422

.57

4.54

1.22

Sh D

iscr

ete

Rej

ectio

n Pr

ofile

(Se

e no

te in

key

) M

inim

um4.

0043

.00

8.00

11.0

06.

0024

2.00

8.00

7.00

31.0

026

9.00

-174

.00

120.

0027

8.00

-0.1

31.

00-1

4.00

-3.5

3 M

inim

um0

Max

imum

7.00

180.

0022

.00

48.0

021

.00

358.

0030

.00

17.0

077

.00

758.

0024

7.00

487.

0090

0.00

0.06

89.0

05.

001.

97 M

axim

um1

Pla

ys B

asel

ine:

Lin

e E

ndin

g St

atis

tics

by

Per

iod

Ear

ly (

to 1

597)

Blo

cks

99

Mea

n12

.00

16.3

3St

d D

ev3.

624.

00 M

iddl

e (1

597-

1604

)B

lock

s13

13M

ean

16.2

325

.00

Std

Dev

2.36

4.28

Lat

e (f

rom

160

5)B

lock

s7

7M

ean

19.2

942

.00

Std

Dev

1.67

6.14

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leG

rade

L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

with

(2lw

s)no

/(

no+

not)

it (l

ws)

it (f

w)

the

(2lw

)B

oB1

BoB

3B

oB5

BoB

7T

-E S

lope

T-E

Rar

e W

ords

T-E

New

W

ords

Agg

rega

teB

ucke

ts

Glo

bal M

in4

528

119

242

87

3028

4-1

7415

927

8-0

.13

-2-1

4-2

.00

Glo

bal M

ax7

180

2250

2135

830

1863

758

247

487

779

0.06

895

2.00

Ear

ly M

in4

528

119

242

87

3028

4-1

7415

927

8-0

.13

-2-1

4-2

.00

Ear

ly M

ax7

180

1723

2135

830

1863

758

247

487

779

0.06

895

2.00

Mid

dle

Min

452

816

924

28

730

284

-174

159

278

-0.1

3-2

-14

-2.0

0 M

iddl

e M

ax7

180

2032

2135

830

1863

758

247

487

779

0.06

895

2.00

Lat

e M

in4

5217

319

242

87

3028

4-1

7415

927

8-0

.13

-2-1

4-2

.00

Lat

e M

ax7

180

2250

2135

830

1863

758

247

487

779

0.06

895

2.00

Blo

cks

Tes

ted

410

Cla

iman

t Pla

ys v

ersu

s Sh

akes

pear

e B

asel

ine:

Rou

nd O

ne T

ests

Pla

ywri

ght:

Pla

yD

ate

Lat

eG

rade

L

evel

HC

W

/2

0K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

with

(2lw

s)no

/(

no+

not)

it (l

ws)

it (f

w)

the

(2lw

)B

oB1

BoB

3B

oB5

BoB

7T

-E S

lope

T

est

T-E

Rar

e W

ords

T-E

New

W

ords

Agg

rega

teB

ucke

tsN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

P

roba

bilit

y (2

%)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Pro

babi

lity

Bea

umon

t: P

ES

L16

077

3319

2616

332

177

3357

755

150

164

-0.0

468

-10

0.31

174

3.09

0E-0

46.

1509

2.59

05E

-03

Cha

pman

U

SHR

1602

551

1431

833

525

2326

762

164

241

100

-0.0

166

-12

-0.0

517

66.

551E

-07

7.34

031.

0352

E-0

5

BU

SS16

079

5119

3611

322

2021

4257

3-2

032

010

0-0

.09

-2-1

4-2

.50

175

1.61

9E-0

58.

9050

5.10

70E

-10

Dan

iel:

CL

EO

1593

831

1226

935

32

430

276

-290

228

613

-0.0

35

-2-6

.82

178

5.29

6E-1

010

.296

56.

7794

E-1

5D

ekke

r

WB

AB

1607

544

1442

934

915

1336

474

-18

334

182

-0.0

1-6

4-2

0-4

.03

176

6.55

1E-0

78.

4309

1.40

44E

-08

H

NW

R16

084

8814

2311

396

1412

3965

015

912

010

0-0

.09

18-2

6-3

.64

177

2.08

7E-0

89.

0618

1.61

03E

-10

Flet

cher

W

PR

Z16

043

3534

2612

345

3610

3256

514

7-9

-113

-0.0

255

-80.

5817

66.

551E

-07

10.7

337

<1.

0000

E-1

5

VA

LN

1610

76

931

030

936

346

1318

660

50.

0667

-3-0

.02

151

2.61

4E-0

16.

2970

5.11

66E

-04

M

TO

M16

165

5339

237

409

247

2645

210

529

500.

0157

-11

-1.0

817

72.

087E

-08

14.1

974

<1.

0000

E-1

5

CH

NC

1617

210

1346

533

1126

451

147

4047

70.

0366

-28

-0.3

915

77.

153E

-09

10.4

537

<1.

0000

E-1

5

LO

YL

1618

316

1541

035

633

393

7911

932

0.11

72-1

40.

2015

81.

453E

-10

8.73

563.

2795

E-1

0

DE

MT

1619

21

1233

524

625

434

156

020

0.01

72-7

-1.0

615

62.

742E

-07

7.86

071.

2382

E-0

7

BA

RN

1619

70

1527

133

941

512

-18

383

635

0.00

29-2

10.

2615

33.

039E

-03

7.04

671.

3703

E-0

5

ISL

N16

194

1914

369

3729

2553

087

147

589

0.07

48-1

40.

9015

77.

153E

-09

9.47

361.

1054

E-1

2G

reen

e

AL

PH

1587

105

526

935

67

861

405

-187

461

862

-0.1

034

-11

-3.5

617

85.

296E

-10

10.7

996

<1.

0000

E-1

5

FBFB

1591

647

517

1621

73

462

324

-172

318

487

-0.1

7-1

1-3

2-2

.16

179

1.07

6E-1

110

.129

82.

9090

E-1

4

lAM

415

915

174

168

289

84

4943

3-6

025

872

7-0

.04

17-1

4-2

.42

175

1.61

9E-0

56.

1470

2.63

00E

-03

Hey

woo

d: H

EY

W16

038

2015

2315

339

1711

3563

9-3

139

278

-0.0

255

-1-0

.47

173

4.40

9E-0

36.

1582

2.51

80E

-03

John

son

SE

JA16

036

2613

3113

318

1722

5944

0-1

4750

975

5-0

.05

-20

-19

-1.1

217

51.

619E

-05

7.99

352.

3581

E-0

7

VO

LP

1606

349

2021

1035

326

1355

723

162

238

528

-0.0

87

-28

-1.2

817

43.

090E

-04

8.17

577.

4876

E-0

8

AL

CH

1610

390

2023

1138

919

1160

720

171

217

403

-0.1

5-3

-56

0.08

176

6.55

1E-0

713

.422

8<

1.00

00E

-15

B

AR

T16

144

153

1624

932

617

1355

703

208

265

422

-0.1

642

-40

-2.1

117

51.

619E

-05

10.2

677

8.68

58E

-15

N

INN

1629

510

122

2412

341

2918

6668

910

027

050

3-0

.16

-27

-54

0.57

175

1.61

9E-0

513

.287

0<

1.00

00E

-15

T

TU

B16

334

137

2023

736

520

1339

828

236

225

412

-0.1

09

62-1

.86

175

1.61

9E-0

515

.727

3<

1.00

00E

-15

Kyd

: SP

TR

1589

743

712

1536

25

954

275

-297

381

504

-0.0

72

-21

-3.8

217

85.

296E

-10

8.64

923.

1554

E-0

9L

yly:

LY

WM

1597

418

53

1819

64

649

880

6314

889

4-0

.08

35-1

70.

7917

101.

756E

-13

10.2

884

7.27

75E

-15

Mar

low

e

TA

M1

1588

955

722

1423

07

348

225

-407

509

709

-0.1

7-4

8-1

6-4

.13

1712

<1.

000E

-15

12.1

162

<1.

0000

E-1

5

TA

M2

1588

919

720

1027

710

864

221

-370

438

798

-0.1

4-4

8-2

1-2

.71

1711

2.77

6E-1

511

.467

0<

1.00

00E

-15

D

F16

1588

517

1221

1228

18

651

307

-108

306

600

-0.0

98

-23

-1.0

117

34.

409E

-03

6.70

752.

4430

E-0

4

JEW

M15

894

1210

1516

355

138

5838

5-9

121

830

8-0

.05

41-2

00.

3217

24.

459E

-02

5.86

627.

4245

E-0

3

ED

W2

1592

510

1425

74

668

385

-178

434

512

-0.0

245

-9-0

.40

155

8.12

6E-0

65.

9676

2.01

02E

-03

M

AP

A15

936

414

323

14

8340

2-2

5438

350

90.

0050

-17

0.26

156

2.74

2E-0

78.

0442

3.84

22E

-08

D

IDO

1586

526

1924

02

1040

214

-239

266

813

-0.1

5-1

6-1

9-5

.36

1510

2.85

3E-1

49.

1654

1.28

44E

-11

Mid

dlet

on

PH

OE

1604

458

1617

1029

814

636

702

113

179

-120

0.03

41-2

60.

4017

34.

409E

-03

7.80

287.

5059

E-0

7

MIC

L16

064

4415

1311

315

1812

3873

320

314

8-2

19-0

.01

43-1

60.

9317

66.

551E

-07

8.60

124.

4028

E-0

9

CH

ST16

114

3724

2915

391

159

3064

814

6-8

-430

-0.0

233

-33

-0.6

617

72.

087E

-08

11.0

638

<1.

0000

E-1

5

NW

lT16

134

7321

2115

430

1315

3374

612

081

-343

0.01

32-2

00.

3517

51.

619E

-05

9.60

172.

4322

E-1

2

MD

IS16

154

9023

2913

378

815

2968

085

121

-269

0.03

20-1

6-2

.09

178

5.29

6E-1

08.

3443

2.49

99E

-08

W

ITC

1616

583

2331

932

88

1235

632

103

-5-4

310.

0532

-22

-0.5

017

43.

090E

-04

9.31

542.

3434

E-1

1

HE

NG

1618

512

324

3119

305

2711

4461

611

417

8-7

20.

0130

-15

0.43

173

4.40

9E-0

36.

5766

4.41

14E

-04

W

BW

M16

214

6723

3010

323

1214

2770

211

234

-367

0.06

33-9

-3.4

917

66.

551E

-07

8.69

442.

3000

E-0

9

GA

ME

1624

588

2337

1436

613

1439

613

1125

7-1

67-0

.14

-60

-41

-0.7

917

66.

551E

-07

11.4

011

<1.

0000

E-1

5M

unda

y: J

KJC

1594

615

177

637

110

538

224

-62

337

709

0.01

60-1

00.

1417

66.

551E

-07

7.00

425.

9064

E-0

5N

ashe

: WIL

L15

929

987

53

438

915

5028

9-6

835

285

0-0

.20

-74

-45

-2.3

317

101.

756E

-13

15.0

467

<1.

0000

E-1

5P

eele

A

RP

A15

849

126

1114

336

28

6614

1-2

7138

586

8-0

.08

-19

-23

-3.8

717

112.

776E

-15

11.3

148

<1.

0000

E-1

5

DB

ET

1594

1160

1017

2427

60

279

-164

-541

593

932

-0.0

9-1

4-2

2-9

.71

1712

<1.

000E

-15

17.3

340

<1.

0000

E-1

5P

icke

ring

: HO

RE

1567

75

832

40

534

384

-106

292

914

0.06

80-2

2-0

.77

156

2.74

2E-0

78.

6511

6.04

77E

-10

Por

ter:

AN

WO

1598

1176

1519

1325

725

1034

455

172

223

80.

0110

5-1

10.

5517

43.

090E

-04

10.9

344

<1.

0000

E-1

5Si

dney

: AN

TO

1590

1154

639

742

90

029

-478

-608

563

100

-0.2

4-8

2-1

8-8

.92

1716

<1.

000E

-15

20.6

108

<1.

0000

E-1

5Sm

ith: H

EC

T16

156

5812

1314

369

1911

6557

0-1

134

437

70.

0227

-10

0.30

174

3.09

0E-0

47.

8092

7.22

50E

-07

Wils

on: 3

LD

Y15

818

218

306

46

4636

938

208

963

-0.0

966

-16

1.16

157

7.15

3E-0

97.

9185

8.60

16E

-08

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ectio

ns18

3223

2010

2019

2015

1211

2129

1214

3318

327

5151

Per

cent

age

35%

63%

58%

50%

20%

39%

37%

39%

29%

24%

22%

41%

57%

24%

27%

65%

35%

39%

100%

100%

Blo

cks

Tes

ted

5151

4040

5151

5151

5151

5151

5151

5151

5184

551

51

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

Gra

de

Lev

elH

CW

/20K

Fem

E

ndin

gsO

pen

Lin

esw

ith

(2

lws)

no

/(no

+no

t)it

(lw

s)it

(fw

)th

e (2

lw)

BoB

1B

oB3

BoB

5B

oB7

T-E

Slo

peT

-E R

are

Wor

dsT

-E N

ew

Wor

dsA

ggre

gate

Buc

kets

Dis

cret

eC

ontin

uous

Glo

bal M

in4

528

119

242

87

3028

4-1

7415

927

8-0

.13

-2-1

4-2

.00

2.90

7E-0

18.

9240

E-0

3 G

loba

l Max

718

022

5021

358

3018

6375

824

748

777

90.

0689

52.

00 E

arly

Min

452

811

924

28

730

284

-174

159

278

-0.1

3-2

-14

-2.0

0 E

arly

Max

718

017

2321

358

3018

6375

824

748

777

90.

0689

52.

00 M

iddl

e M

in4

528

169

242

87

3028

4-1

7415

927

8-0

.13

-2-1

4-2

.00

Mid

dle

Max

718

020

3221

358

3018

6375

824

748

777

90.

0689

52.

00Sh

Dis

cret

e R

ejec

tion

Pro

file

(Se

e no

te in

key

) L

ate

Min

452

1731

924

28

730

284

-174

159

278

-0.1

3-2

-14

-2.0

0 M

inim

um0

411

Dub

itan

da a

nd S

et-a

side

s, A

pocr

ypha

Pla

ys v

ersu

s Sh

akes

pear

e B

asel

ine:

Rou

nd O

ne T

ests

Pla

yD

ate

Lat

eG

rade

L

evel

HC

W

/2

0K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

wit

h

(2lw

s)no

/(

no+

not)

it (l

ws)

it (f

w)

the

(2lw

)B

oB1

BoB

3B

oB5

BoB

7T

-E S

lope

T

est

T-E

Rar

e W

ords

T-E

New

W

ords

Agg

rega

teB

ucke

tsN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ilit

y (2

%)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abil

ity

Dub

itan

da a

nd S

et-a

side

s

1H6

1590

663

1114

1434

25

762

226

-177

352

695

-0.0

4-2

1-2

02.

6017

66.

551E

-07

7.56

413.

0046

E-0

62H

615

916

8212

1414

331

79

6328

4-1

7936

951

4-0

.04

180

-1.5

417

24.

459E

-02

5.08

767.

6606

E-0

23H

615

915

107

1312

1834

75

556

229

-251

384

539

0.01

344

-1.9

017

43.

090E

-04

6.27

321.

5955

E-0

3H

515

997

106

1626

1229

424

1854

493

-21

463

686

-0.1

3-2

-20

-1.2

717

12.

907E

-01

7.12

513.

2085

E-0

5H

8, (

Fl)

1613

550

2933

1639

625

326

501

2324

960

0.07

48-1

51.

0017

85.

296E

-10

9.31

522.

3470

E-1

1H

8, (

Jt)

1613

535

2638

825

534

563

729

1728

425

40.

1257

-17

0.30

178

5.29

6E-1

08.

6080

4.20

06E

-09

H8,

(Sh

)16

136

5423

5312

346

2814

5076

839

382

376

-0.0

123

1-0

.44

173

4.40

9E-0

35.

3815

3.48

92E

-02

Per,

1-2

1608

641

1417

1225

032

1455

561

-72

389

275

0.04

25-1

20.

9017

51.

619E

-05

7.84

165.

9518

E-0

7Pe

r, 3

-516

084

6917

3721

333

157

4652

430

226

278

-0.0

724

40.

2117

01.

000E

+00

4.11

824.

5711

E-0

1T

im16

084

4716

3012

287

2013

2931

4-3

433

525

80.

007

-22

0.00

176

6.55

1E-0

76.

6275

3.51

48E

-04

TN

K (

Fl)

1613

486

3232

1433

98

523

604

111

7686

-0.0

452

-90.

3017

51.

619E

-05

9.65

571.

5698

E-1

2T

NK

(Sh

)16

135

8121

5011

348

115

3046

323

241

415

-0.0

45

-9-0

.76

171

2.90

7E-0

14.

6076

2.16

17E

-01

Tit.

1594

679

1015

2432

59

555

231

-204

286

468

-0.0

625

3-2

.99

175

1.61

9E-0

56.

3814

1.02

25E

-03

Tit

earl

y15

946

607

1526

295

54

5019

1-2

4927

165

4-0

.04

161

-4.2

017

72.

087E

-08

7.73

391.

1282

E-0

6T

it la

te15

946

100

1415

2436

013

260

245

-173

294

295

-0.0

821

1-0

.50

174

3.09

0E-0

46.

6440

3.26

29E

-04

STM

(Sh

)15

956

2914

4411

400

2222

8662

415

165

347

4-0

.04

28-6

0.00

176

6.55

1E-0

710

.945

5<

1.00

00E

-15

Apo

cryp

ha

HO

RE

1567

75

832

40

534

384

-106

292

914

0.06

80-2

2-0

.80

156

2.74

2E-0

78.

6523

5.99

57E

-10

FVH

515

883

329

2318

325

185

7152

231

275

624

0.04

139

-64.

2017

72.

087E

-08

9.27

183.

2814

E-1

1T

OA

S15

893

1311

1722

184

179

4353

9-4

317

628

90.

0185

-11

1.20

174

3.09

0E-0

46.

9390

8.14

60E

-05

IRO

N15

905

3610

2817

163

59

4841

9-8

635

164

8-0

.05

4-1

4-2

.10

175

1.61

9E-0

57.

5390

3.46

23E

-06

AR

DN

1592

328

1013

1931

211

1461

494

1826

544

2-0

.04

58-1

0-0

.20

172

4.45

9E-0

24.

4534

2.82

87E

-01

YR

KI

1592

314

1013

1632

88

767

315

-203

347

402

-0.0

165

2-0

.40

174

3.09

0E-0

46.

1236

2.87

89E

-03

YR

K2

1592

32

1211

2238

94

465

226

-283

393

565

0.03

489

-1.7

017

101.

756E

-13

8.70

852.

0830

E-0

9G

UY

W15

9310

619

929

314

75

3434

7-1

1615

748

1-0

.02

50-1

6-3

.60

178

5.29

6E-1

010

.572

1<

1.00

00E

-15

LE

IR15

934

1312

816

362

1217

6052

2-6

725

565

8-0

.01

7-3

0.40

173

4.40

9E-0

35.

3176

4.18

05E

-02

RC

D3

1594

53

1434

1943

89

369

421

-179

381

788

0.03

81-5

2.10

178

5.29

6E-1

09.

2776

3.13

81E

-11

STM

O15

956

3616

1816

309

1712

5940

818

296

487

0.00

52-1

31.

0617

12.

907E

-01

4.19

004.

1735

E-0

1E

DW

315

958

688

2410

374

99

5616

3-2

3739

276

3-0

.08

-23

-3-3

.40

177

2.08

7E-0

88.

8547

7.35

05E

-10

KJN

115

955

66

1414

252

95

7126

8-2

0742

667

4-0

.08

4-1

20.

7017

66.

551E

-07

7.18

682.

3335

E-0

5K

JN2

1595

56

515

1722

62

570

284

-226

479

669

-0.0

51

-5-0

.90

177

2.08

7E-0

87.

7324

1.13

82E

-06

LO

CR

1595

857

610

1537

33

542

285

-300

419

885

-0.2

3-7

3-3

5-3

.50

1712

<1.

000E

-15

13.3

374

5.42

10E

-19

WO

OD

1595

38

1832

935

817

445

201

-202

305

32-0

.02

14-7

-5.1

017

91.

076E

-11

9.32

342.

2025

E-1

1M

UC

E15

986

657

616

333

66

6234

72

151

401

-0.0

253

-11

1.70

175

1.61

9E-0

57.

8316

6.31

96E

-07

OL

DC

1600

524

1316

1333

013

1059

476

2637

364

6-0

.01

47-1

12.

4017

24.

459E

-02

4.87

531.

2586

E-0

1C

RO

M16

025

2017

820

315

1617

4861

322

351

589

0.04

90-3

0.20

173

4.40

9E-0

36.

0636

3.61

81E

-03

DE

VL

1604

588

1517

525

811

361

413

922

223

3-0

.06

32-2

40.

2017

43.

090E

-04

7.31

121.

2104

E-0

5PR

OD

1605

486

149

1730

117

826

747

265

159

403

0.03

89-2

61.

1017

51.

619E

-05

9.50

755.

1790

E-1

2PU

RN

1607

512

218

2813

289

176

3570

013

116

814

6-0

.05

45-1

91.

1017

43.

090E

-04

6.03

324.

0551

E-0

3Y

KSH

1608

510

1624

2527

015

1340

578

-520

543

-0.0

127

40.

4017

51.

619E

-05

6.63

593.

3844

E-0

4M

AID

1611

433

2123

1633

514

1221

393

-38

103

-164

0.10

39-3

-3.3

017

72.

087E

-08

8.14

798.

9429

E-0

8FA

LS

1613

551

2023

1029

518

927

701

7722

282

-0.0

14

-19

0.20

175

1.61

9E-0

56.

6602

3.03

22E

-04

FAIR

1631

623

919

1325

916

2052

521

-48

300

807

0.02

54-8

2.50

176

6.55

1E-0

79.

3179

2.29

84E

-11

ME

RL

1631

565

1919

1439

228

1229

476

-31

248

263

0.00

38-6

0.40

174

3.09

0E-0

46.

1616

2.48

49E

-03

RV

GR

1606

371

2026

1230

215

1631

534

-15

266

171

-0.0

111

-14

-1.0

017

34.

409E

-03

5.86

047.

5771

E-0

3

Apo

cryp

ha D

iscr

ete

Dis

crim

inat

ion

Stat

isti

csA

ggre

gate

sA

pocr

ypha

Com

p D

iscr

imin

atio

n St

ats

Rej

ecti

ons

1019

917

69

713

104

93

122

48

1015

227

24 P

erce

ntag

e36

%68

%33

%63

%21

%32

%25

%46

%36

%14

%32

%11

%43

%7%

14%

29%

36%

32%

96%

86%

Blo

cks

Tes

ted

2828

2727

2828

2828

2828

2828

2828

2828

2847

428

28

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

Gra

de

Lev

elH

CW

/20K

Fem

E

ndin

gsO

pen

Lin

esw

ith

(2lw

s)no

/(

no+

not)

it (l

ws)

it (f

w)

the

(2lw

)B

oB1

BoB

3B

oB5

BoB

7T

-E S

lope

T-E

Rar

e W

ords

T-E

New

W

ords

Agg

rega

teB

ucke

tsD

iscr

ete

Con

tinuo

us

Glo

bal M

in4

528

119

242

87

3028

4-1

7415

927

8-0

.13

-2-1

4-2

.00

2.90

7E-0

18.

924E

-03

Glo

bal M

ax7

180

2250

2135

830

1863

758

247

487

779

0.06

895

2.00

Ear

ly M

in4

528

119

242

87

3028

4-1

7415

927

8-0

.13

-2-1

4-2

.00

Ear

ly M

ax7

180

1723

2135

830

1863

758

247

487

779

0.06

895

2.00

Mid

dle

Min

452

816

924

28

730

284

-174

159

278

-0.1

3-2

-14

-2.0

0 M

iddl

e M

ax7

180

2032

2135

830

1863

758

247

487

779

0.06

895

2.00

Sh

Dis

cret

e R

ejec

tion

Pro

file

(S

ee n

ote

in k

ey)

Lat

e M

in4

5217

319

242

87

3028

4-1

7415

927

8-0

.13

-2-1

4-2

.00

Min

imum

0 L

ate

Max

718

022

5021

358

3018

6375

824

748

777

90.

0689

52.

00 M

axim

um1

412

APPENDIX FOUR: KEY TO TABLES FOR ROUND TWO PLAY TESTS CONTRACTIONS, METRIC FILLERS, SELECTED WORDS AND PHRASES, PER 20,000 WORDS



I’m: Occurrence rate of I’m (per 20,000 words). 0 to 1*

you’re: T Occurrence rate of you’re or y’are. before 1608: 0 to 2* from 1608: 0 to 6*

we’re Occurrence rate of we’re. 0*

I’ve Occurrence rate of I’ve. 0*

you’ve Occurrence rate of you’ve or y’have. 0*

‘em Occurrence rate of ‘em. 0 to 9*

Total 1 Total of preceding six columns. 0 to 9

on’t: T Occurrence rate of on’t. To 1600: 0 to 2 From 1600: 1 to 11

ne’er Occurrence rate of ne’er. 1 to 12

e’en Occurrence rate of e’en. 0 to 3

i’faith Occurrence rate of i’faith. 0 to 8

th’: T Occurrence rate of th’. (see following table)

i’th’: T Occurrence rate of i’th’. before 1600: 0 to 9 from 1600: 6 to 20

ha’ Occurrence rate of ha’. 0 to 5

Total 2: T Total of preceding seven columns. before 1601: 6 to 37 from 1601: 42 to 115

‘ll Occurrence rate of ‘ll. (Example: I’ll.) 31 to 90

‘d | ‘ld: T Occurrence rate of ‘d or ‘ld. (Examples: I’d or I’ld but not curs’d.)

before 1602: 0 to 2 from 1602: 3 to 11

‘tis Occurrence rate of ‘tis. 16 to 67

if that Occurrence rate of if that as a conjunctional affix or met-ric filler. (Example: “If that the world and life were young,” but not, “If that were so.”)

0 to 6

the which Occurrence rate of the which as a metric filler. (Example: “to the which place.”)

0 to 8

Other Fillers Combined rate for other metric fillers: when that, since that, sith that.

0 to 5

Total Fillers Total of all metric filler rates. 1 to 13*

I do Occurrence rate of I do, excluding I do not. 6 to 41

I do + verb Occurrence rate of periphrastic verbs with I do (such as I do weep), excluding I do not.

5 to 28

Round Two Rejections Total number of rejections, this round. 0 to 1

SUMMARY OF RESULTS

See Key to Appendix Two. * = not shown or counted separately. In 15 tests no Sh. play had more than one rejection, but 98% of the Claimant plays and 89% of the Apocrypha plays had two or more.

413

Shak

espe

are

Pla

y B

asel

ine

Dat

a, R

ound

Tw

o T

ests

Play

Dat

e L

ate

Tot

al 1

on't

ne'e

re'

eni'f

aith

th'

i'th'

ha'

'll

d | '

ld'ti

sif

that

the

whi

chO

ther

Fi

llers

I do

I do

+ v

erb

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

ns

Dis

cret

e C

ompo

site

Pr

obab

ility

(1

.3%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

R3

1593

01

10

07

00

410

222

24

1813

160

1.00

0E+

003.

3005

8.16

04E

-01

Shr

(c)

1594

10

71

38

44

791

520

00

75

160

1.00

0E+

004.

7203

1.34

34E

-01

TG

V (

c)15

945

14

00

40

573

246

20

019

1316

01.

000E

+00

5.20

664.

0293

E-0

2E

rr (

c)15

941

010

10

61

064

024

11

411

716

01.

000E

+00

4.15

283.

6988

E-0

1R

215

950

05

00

10

031

024

23

016

1316

01.

000E

+00

3.11

578.

8145

E-0

1L

LL

(c)

1595

00

60

211

10

400

191

31

2421

160

1.00

0E+

002.

9145

9.32

77E

-01

Jn15

960

06

02

121

031

017

54

217

1416

01.

000E

+00

4.17

943.

5599

E-0

1M

ND

(c)

1596

00

10

26

40

410

60

10

2419

161

1.88

9E-0

13.

8589

5.32

65E

-01

Rom

15

960

27

22

50

074

034

22

115

1016

01.

000E

+00

3.14

898.

7100

E-0

11H

415

970

04

06

00

085

023

23

115

616

11.

889E

-01

3.62

286.

6364

E-0

1W

iv (

c)15

972

24

02

179

310

60

460

31

178

161

1.88

9E-0

15.

1059

5.30

47E

-02

MoV

1597

01

51

110

10

540

192

40

2118

160

1.00

0E+

002.

4490

9.88

12E

-01

2H4

1598

01

41

521

30

410

200

82

1410

160

1.00

0E+

004.

9191

8.52

83E

-02

JC15

997

13

00

62

131

024

11

132

2216

01.

000E

+00

3.51

547.

1900

E-0

1A

do (

c)15

990

12

07

40

146

018

06

020

1216

01.

000E

+00

4.09

414.

0117

E-0

1A

YL

(c)

1599

10

30

410

50

600

314

55

2414

160

1.00

0E+

004.

2498

3.20

35E

-01

Ham

1601

15

17

149

133

510

491

31

2011

161

1.88

9E-0

15.

4376

2.03

76E

-02

TN

(c)

1602

47

20

812

80

786

333

10

2816

160

1.00

0E+

004.

2941

2.98

80E

-01

Tro

1602

02

52

327

81

903

391

03

2013

160

1.00

0E+

003.

2904

8.20

06E

-01

MFM

1603

11

30

033

70

495

420

10

2922

160

1.00

0E+

003.

6298

6.59

87E

-01

AW

W (

c)

1603

76

53

237

121

596

470

30

1913

160

1.00

0E+

002.

8090

9.52

05E

-01

Oth

16

040

53

04

326

057

467

22

041

2816

01.

000E

+00

5.33

572.

7766

E-0

2L

r16

057

24

00

6920

270

941

22

216

1316

01.

000E

+00

2.69

859.

6744

E-0

1M

ac16

065

52

01

6916

054

646

02

27

616

01.

000E

+00

3.22

218.

4593

E-0

1A

nt16

072

68

30

7320

369

350

22

014

816

01.

000E

+00

3.76

805.

8397

E-0

1C

or16

089

56

00

127

302

668

382

35

1711

161

1.88

9E-0

14.

6773

1.47

21E

-01

Cym

1610

711

50

083

141

776

401

21

158

160

1.00

0E+

002.

9010

9.35

51E

-01

Tm

p16

1120

104

00

7820

088

524

01

127

1716

11.

889E

-01

5.03

216.

4353

E-0

2W

T16

116

104

00

6915

376

1139

12

014

1116

01.

000E

+00

3.68

506.

3002

E-0

1

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ectio

ns1

00

10

11

01

01

00

00

06

00

Per

cent

age

3%0%

0%3%

0%3%

3%0%

3%0%

3%0%

0%0%

0%0%

1%0%

0%46

429

29

Pla

ys B

asel

ine:

Sta

ndar

d St

atis

tica

l Pro

file

Com

posi

te T

hres

hold

s C

ount

2929

2929

2929

2929

2929

2929

2929

2929

1.88

9E-0

12.

0376

E-0

2 M

ean

2.97

2.93

4.28

0.72

1.90

30.5

57.

591.

0361

.41

2.59

33.7

91.

342.

411.

2819

.34

13.1

7 S

td D

ev4.

303.

302.

101.

482.

2532

.10

7.85

1.43

19.1

93.

2413

.59

1.24

1.75

1.53

7.15

5.39

Sh D

iscr

ete

Rej

ectio

n Pr

ofile

(Se

e no

te in

key

) M

in0.

000.

001.

000.

000.

000.

000.

000.

0031

.00

0.00

6.00

0.00

0.00

0.00

7.00

5.00

Min

imum

0 M

ax20

.00

11.0

010

.00

7.00

8.00

127.

0030

.00

5.00

106.

0011

.00

67.0

05.

008.

005.

0041

.00

28.0

0 M

axim

um1

Pla

ys B

asel

ine:

Sel

ecte

d St

atis

tics

by

Per

iod

Cou

ntto

160

016

to 1

600

1616

to 1

602

17 M

ean

0.63

8.00

1.94

0.18

Std

Dev

0.70

5.30

2.44

0.51

Cou

nt16

00 -

160

46

6 M

ean

31.6

79.

00 S

td D

ev11

.10

2.58

Cou

ntfr

om 1

600

13fr

om 1

605

77

from

160

212

Mea

n5.

7781

.14

19.2

96.

00 S

td D

ev3.

0219

.36

4.98

2.27

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

le

Tot

al 1

on't

ne'e

re'

eni'f

aith

th'

i'th'

ha'

'll

d | '

ld'ti

sif

that

the

whi

chO

ther

Fi

llers

I do

I do

+ v

erb

Glo

bal M

in0

01

00

10

031

017

00

06

5 G

loba

l Max

911

123

812

720

590

1167

68

541

28 E

arly

Min

to 1

600

0to

160

01

0to

160

20

Ear

ly M

ax2

269

2 M

iddl

e M

in16

00-1

604

126

Mid

dle

Max

4920

Lat

e M

infr

om 1

600

1fr

om 1

605

696

from

160

23

Lat

e M

ax11

127

2011

Blo

cks

Tes

ted

414

Cla

iman

t Pla

ys v

ersu

s Sh

akes

pear

e B

asel

ine:

Rou

nd T

wo

Tes

ts

Pla

ywri

ght:

Pla

yD

ate

Lat

eT

otal

1on

'tne

'er

e'en

i'fai

thth

'i't

h'ha

' 'l

ld

| 'ld

'tis

if th

atth

e w

hich

Oth

er

Fill

ers

I do

I do

+ v

erb

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

ns

Dis

cret

e C

ompo

site

P

roba

bilit

y (1

.3%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Pro

babi

lity

Bea

umon

t: PE

SL16

0716

00

018

00

010

20

381

00

249

168

9.56

7E-1

210

.808

1<

1.00

00E

-15

Cha

pman

U

SHR

1602

00

10

10

00

00

00

00

157

166

3.45

6E-0

87.

5246

1.92

19E

-06

B

USS

1607

00

00

00

00

00

00

00

21

169

1.12

1E-1

38.

9708

1.36

60E

-10

Dan

iel:

CL

EO

1593

01

00

085

00

40

246

14

77

163

1.08

4E-0

315

.800

1<

1.00

00E

-15

Dek

ker

W

BA

B16

0757

05

00

178

1170

239

00

05

216

76.

477E

-10

15.5

854

<1.

0000

E-1

5

HN

WR

1608

412

71

624

1541

139

551

00

06

416

51.

439E

-06

30.1

334

<1.

0000

E-1

5F

letc

her

W

PRZ

1604

670

00

03

00

140

146

00

030

1416

76.

477E

-10

16.6

728

<1.

0000

E-1

5

VA

LN

1610

590

00

00

00

832

400

00

208

166

3.45

6E-0

814

.856

0<

1.00

00E

-15

M

TO

M16

1630

87

01

286

011

40

640

00

2917

164

4.58

7E-0

59.

0992

5.19

58E

-11

C

HN

C16

1760

00

00

00

010

90

490

00

2317

167

6.47

7E-1

015

.376

6<

1.00

00E

-15

L

OY

L16

1810

40

00

00

01

103

054

00

015

416

89.

567E

-12

24.7

698

<1.

0000

E-1

5

DE

MT

1619

677

90

046

201

940

610

00

120

165

1.43

9E-0

615

.987

7<

1.00

00E

-15

B

AR

N16

1941

10

00

50

055

060

10

04

016

76.

477E

-10

11.8

648

<1.

0000

E-1

5

ISL

N16

1970

00

01

50

168

07

00

010

316

89.

567E

-12

17.3

603

<1.

0000

E-1

5G

reen

e

AL

PH15

870

013

03

00

041

011

2519

3227

1616

63.

456E

-08

29.7

590

<1.

0000

E-1

5

FBFB

1591

20

90

02

00

850

269

04

22

163

1.08

4E-0

38.

0417

8.39

84E

-08

lA

M4

1591

00

00

03

00

401

163

02

31

164

4.58

7E-0

55.

3102

2.99

38E

-02

Hey

woo

d: H

EY

W16

032

00

00

00

094

134

00

07

516

63.

456E

-08

6.81

148.

4566

E-0

5Jo

hnso

n

SEJA

1603

20

20

05

00

243

240

10

74

165

1.43

9E-0

66.

2245

1.18

39E

-03

V

OL

P16

0624

00

15

120

2359

541

00

027

1916

63.

456E

-08

17.4

720

<1.

0000

E-1

5

AL

CH

1610

4510

64

619

2559

122

733

00

016

1016

63.

456E

-08

42.2

479

<1.

0000

E-1

5

BA

RT

1614

6014

39

202

170

944

230

00

177

168

9.56

7E-1

251

.558

5<

1.00

00E

-15

N

INN

1629

366

32

220

5540

477

300

00

2113

164

4.58

7E-0

529

.544

8<

1.00

00E

-15

T

TU

B16

3353

1010

52

2253

111

9419

140

00

2619

168

9.56

7E-1

278

.732

8<

1.00

00E

-15

Kyd

: SPT

R15

890

02

00

70

058

019

00

01

116

21.

797E

-02

4.62

511.

6395

E-0

1L

yly:

LY

WM

1597

20

00

20

00

106

013

00

05

216

63.

456E

-08

5.35

082.

6546

E-0

2M

arlo

we

T

AM

115

880

00

02

100

09

012

00

10

016

51.

439E

-06

5.78

996.

2959

E-0

3

TA

M2

1588

00

10

011

00

230

100

00

20

164

4.58

7E-0

55.

3819

2.41

74E

-02

D

F16

1588

51

100

07

00

955

171

40

1511

162

1.79

7E-0

210

.260

73.

5087

E-1

5

JEW

M15

8946

213

00

11

787

042

10

04

016

51.

439E

-06

12.5

319

<1.

0000

E-1

5

ED

W2

1592

20

00

02

00

762

412

01

93

162

1.79

7E-0

25.

4556

1.92

67E

-02

M

APA

1593

20

100

06

00

580

318

00

86

161

1.88

9E-0

16.

8153

8.29

75E

-05

D

IDO

1586

00

70

04

00

630

79

00

74

163

1.08

4E-0

37.

5014

2.19

51E

-06

Mid

dlet

on

PHO

E16

0410

511

2511

023

016

917

690

00

95

167

6.47

7E-1

029

.185

7<

1.00

00E

-15

M

ICL

1606

108

1826

842

3828

3011

69

720

00

155

1610

<1.

000E

-15

38.8

099

<1.

0000

E-1

5

CH

ST16

1144

2027

1334

511

1714

18

850

00

20

1611

<1.

000E

-15

26.3

443

<1.

0000

E-1

5

NW

lT16

1311

621

267

1348

190

9721

640

00

52

1610

<1.

000E

-15

30.5

349

<1.

0000

E-1

5

MD

IS16

1599

1924

64

4415

092

2960

00

04

116

91.

121E

-13

27.3

168

<1.

0000

E-1

5

WIT

C16

1662

3023

41

371

412

99

102

00

014

816

89.

567E

-12

20.1

227

<1.

0000

E-1

5

HE

NG

1618

258

150

18

10

997

680

00

62

167

6.47

7E-1

010

.171

97.

7286

E-1

5

WB

WM

1621

106

2620

161

431

090

1192

00

02

116

91.

121E

-13

28.9

607

<1.

0000

E-1

5

GA

ME

1624

9522

92

024

81

6023

510

00

52

166

3.45

6E-0

824

.008

5<

1.00

00E

-15

Mun

day:

JK

JC15

940

015

00

10

086

012

00

37

016

31.

084E

-03

6.99

023.

4733

E-0

5N

ashe

: WIL

L15

920

113

00

40

036

08

10

46

216

31.

084E

-03

6.25

451.

0455

E-0

3P

eele

A

RPA

1584

20

20

816

04

00

40

00

66

162

1.79

7E-0

26.

4076

5.44

56E

-04

D

BE

T15

940

00

00

50

015

03

00

00

016

51.

439E

-06

6.00

202.

8713

E-0

3Pi

cker

ing:

HO

RE

1567

00

72

00

00

392

014

1717

2422

165

1.43

9E-0

617

.619

9<

1.00

00E

-15

Port

er: A

NW

O15

983

224

021

30

213

30

629

02

2617

164

4.58

7E-0

515

.050

9<

1.00

00E

-15

Sidn

ey: A

NT

O15

900

00

00

90

00

00

00

90

016

63.

456E

-08

8.12

444.

9417

E-0

8Sm

ith: H

EC

T16

155

316

00

97

046

447

51

19

316

31.

084E

-03

8.43

256.

3803

E-0

9W

ilson

: 3L

DY

1581

10

50

10

00

900

310

00

92

162

1.79

7E-0

24.

1843

3.53

46E

-01

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ectio

ns26

2031

106

3421

1130

1822

62

317

2928

650

46 P

erce

ntag

e51

%39

%61

%20

%12

%67

%41

%22

%59

%35

%43

%12

%4%

6%33

%57

%35

%98

%90

% B

lock

s T

este

d51

5151

5151

5151

5151

5151

5151

5151

5181

651

51

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

Tot

al 1

on't

ne'e

re'

eni'f

aith

th'

i'th'

ha'

'll

d | '

ld'ti

sif

that

the

whi

chO

ther

F

iller

sI

doI

do +

ver

bD

iscr

ete

Con

tinuo

us

Glo

bal M

in0

01

00

10

031

017

00

06

51.

889E

-01

2.03

8E-0

2 G

loba

l Max

911

123

812

720

590

1167

68

541

28 E

arly

Min

to 1

600

0to

160

01

0to

160

20

Ear

ly M

ax2

269

2 M

iddl

e M

in16

00-1

604

126

Mid

dle

Max

4920

Sh D

iscr

ete

Rej

ectio

n Pr

ofile

(Se

e no

te in

key

) L

ate

Min

from

160

01

from

160

569

6fr

om 1

602

3 M

inim

um0

Lat

e M

ax11

127

2011

Max

imum

1

415

Play

Dat

e L

ate

Tot

al 1

on't

ne'e

re'

eni'f

aith

th'

i'th'

ha'

'll

d | '

ld'ti

sif

that

the

whi

chO

ther

Fi

ller

sI

doI

do +

ver

bN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ility

(1

.3%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Dub

itan

da a

nd S

et-a

side

s

1H6

1590

20

170

05

00

760

173

20

66

161

1.88

9E-0

17.

0738

2.26

02E

-05

2H6

1591

10

50

18

20

551

362

11

119

160

1.00

0E+

002.

7749

9.57

30E

-01

3H6

1591

00

101

03

01

731

300

03

63

161

1.88

9E-0

15.

0398

6.30

84E

-02

H5

1599

10

30

226

31

330

282

51

1610

160

1.00

0E+

004.

3456

2.74

71E

-01

H8,

(Fl

)16

1312

66

80

064

80

533

640

00

83

163

1.08

4E-0

329

.065

6<

1.00

00E

-15

H8,

(Jt

)16

1380

010

00

4510

060

070

00

015

516

51.

439E

-06

18.9

878

<1.

0000

E-1

5H

8, (

Sh)

1613

197

30

011

020

232

247

05

029

2016

21.

797E

-02

5.65

171.

0176

E-0

2Pe

r, 1

-216

083

50

00

130

011

55

335

200

105

165

1.43

9E-0

612

.458

6<

1.00

00E

-15

Per,

3-5

1608

82

122

224

44

616

384

00

1210

162

1.79

7E-0

26.

9539

4.17

44E

-05

Tim

1608

257

1610

150

35

569

580

00

1211

165

1.43

9E-0

611

.307

6<

1.00

00E

-15

TN

K (

Fl)

1613

375

30

031

160

821

370

00

80

164

4.58

7E-0

59.

4826

2.58

86E

-12

TN

K (

Sh)

1613

337

10

091

150

743

291

60

86

161

1.88

9E-0

18.

0836

6.42

61E

-08

Tit.

1594

00

40

02

00

420

290

00

139

160

1.00

0E+

003.

3334

8.02

55E

-01

Tit

earl

y15

940

00

00

20

051

023

00

215

1316

11.

889E

-01

3.67

856.

3358

E-0

1T

it la

te15

940

08

00

30

031

039

00

313

516

01.

000E

+00

4.19

963.

4560

E-0

1ST

M (

Sh)

1595

00

00

072

00

101

029

140

00

016

63.

456E

-08

16.6

676

<1.

0000

E-1

5

Apo

cryp

ha

HO

RE

1567

70

72

00

00

392

014

1717

2422

165

1.43

9E-0

617

.631

3<

1.00

00E

-15

FVH

515

880

07

20

00

853

024

00

15

216

44.

587E

-05

6.59

152.

3918

E-0

4T

OA

S15

8913

013

23

30

813

92

315

00

2521

164

4.58

7E-0

59.

6404

7.17

77E

-13

IRO

N15

900

07

04

120

057

035

30

09

816

01.

000E

+00

3.65

756.

4500

E-0

1A

RD

N15

920

021

02

00

014

60

361

22

146

163

1.08

4E-0

39.

5229

1.87

02E

-12

YR

KI

1592

00

50

14

00

661

224

11

54

162

1.79

7E-0

24.

3714

2.63

04E

-01

YR

K2

1592

10

90

00

00

781

282

20

76

161

1.88

9E-0

14.

4447

2.31

44E

-01

GU

YW

1593

06

140

80

02

134

313

30

29

616

63.

456E

-08

12.1

359

<1.

0000

E-1

5L

EIR

1593

10

250

04

00

817

214

03

2517

162

1.79

7E-0

217

.000

7<

1.00

00E

-15

RC

D3

1594

00

60

10

00

140

140

01

119

163

1.08

4E-0

34.

3817

2.58

46E

-01

STM

O15

956

315

04

50

494

522

10

210

616

44.

587E

-05

11.9

340

<1.

0000

E-1

5E

DW

315

950

01

00

10

033

016

11

010

816

11.

889E

-01

3.95

544.

7799

E-0

1K

JN1

1595

00

30

00

00

330

171

30

118

161

1.88

9E-0

13.

6112

6.69

77E

-01

KJN

215

950

04

00

60

021

021

24

00

016

31.

084E

-03

4.90

708.

7801

E-0

2L

OC

R15

950

00

00

00

013

01

01

63

116

76.

477E

-10

6.61

962.

1013

E-0

4W

OO

D15

9528

67

29

223

2213

611

390

00

62

167

6.47

7E-1

028

.310

9<

1.00

00E

-15

MU

CE

1598

00

00

20

00

111

031

20

37

316

44.

587E

-05

5.09

265.

4956

E-0

2O

LD

C16

001

53

017

00

033

021

03

010

216

44.

587E

-05

8.83

073.

8375

E-1

0C

RO

M16

021

00

04

04

076

325

43

427

916

44.

587E

-05

5.82

255.

6014

E-0

3D

EV

L16

045

314

012

72

013

33

525

00

93

166

3.45

6E-0

89.

4950

2.34

27E

-12

PRO

D16

058

010

022

00

015

58

381

10

2010

165

1.43

9E-0

612

.333

9<

1.00

00E

-15

PUR

N16

0735

2426

1723

342

212

112

380

10

61

1610

<1.

000E

-15

21.3

018

<1.

0000

E-1

5Y

KSH

1608

313

170

210

107

567

490

00

107

165

1.43

9E-0

614

.116

4<

1.00

00E

-15

MA

ID16

1198

2325

247

112

1395

2386

00

05

116

12<

1.00

0E-1

532

.427

3<

1.00

00E

-15

FAL

S16

1384

410

40

213

052

349

10

017

1216

44.

587E

-05

19.8

733

<1.

0000

E-1

5FA

IR16

310

00

016

00

024

02

23

37

716

89.

567E

-12

10.1

555

<1.

0000

E-1

5M

ER

L16

314

58

18

112

085

741

10

112

716

21.

797E

-02

6.53

733.

0617

E-0

4R

VG

R16

0610

018

2129

2239

144

101

2094

00

04

316

11<

1.00

0E-1

533

.402

4<

1.00

00E

-15

Apo

cryp

ha D

iscr

ete

Dis

crim

inat

ion

Stat

isti

csA

ggre

gate

sA

pocr

ypha

Com

p D

iscr

imin

atio

n St

ats

Rej

ectio

ns7

914

48

209

515

88

11

26

1112

824

20 P

erce

ntag

e25

%32

%50

%14

%29

%71

%32

%18

%54

%29

%29

%4%

4%7%

21%

39%

29%

86%

71%

Blo

cks

Tes

ted

2828

2828

2828

2828

2828

2828

2828

2828

448

2828

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

Tot

al 1

on't

ne'e

re'

eni'f

aith

th'

i'th'

ha'

'll

d | '

ld'ti

sif

that

the

whi

chO

ther

Fi

ller

sI

doI

do +

ver

bD

iscr

ete

Con

tinuo

us

Glo

bal M

in0

01

00

10

031

017

00

06

51.

889E

-01

2.03

76E

-02

Glo

bal M

ax9

1112

38

127

205

9011

676

85

4128

Ear

ly M

into

160

00

to 1

600

10

to 1

602

0 E

arly

Max

226

92

Mid

dle

Min

1600

-160

412

6 M

iddl

e M

ax49

20Sh

Dis

cret

e R

ejec

tion

Prof

ile (

See

note

in k

ey)

Lat

e M

infr

om 1

600

1fr

om 1

605

696

from

160

23

Min

imum

0 L

ate

Max

1112

720

11 M

axim

um1

416

APPENDIX FIVE: KEY TO TABLES FOR ROUND THREE PLAY TESTS PREFIXES, SUFFIXES, INTENSIFIERS, ADVERSIONS, PER 20,000 WORDS



where- | there- Occurrence rate for where- or there- words, exclud-ing wherefore and therefore (per 20,000 words).

3 to 19

dis- Occurrence rate for dis- words. (Examples: distress but not dish.)

19 to 55

whereas | whenas: E Occurrence rate for whereas or whenas. 0

un- Occurrence rate for un- words. (Examples: unable but not union.)

28 to 65

ex- Occurrence rate for ex- words. (Examples: excul-pate, extra.)

17 to 55

fore- Occurrence rate for fore- words. (Examples: fore-warn, but not foreign.)

0 to 8

-able Occurrence rate for -able or -ible words. (Examples: comfortable, but not table.)

10 to 35

-less Occurrence rate for -less words. (Examples: use-less, but not bless or unless.)

2 to 19

-ish Occurrence rate for -ish words. (Examples: British, but not dish.)

1 to 22

-ly Occurrence rate for -ly words. (Examples: heavenly, but not fly.)

98 to 161

-ment Occurrence rate for -ment words. 11 to 36

Very: T Occurrence rate for very. before 1600: 6 to 31 from 1600: 16 to 42

most + mod: T Occurrence rate for most with modifier. Example: most noble but not most do.)

before 1600: 8 to 32 from 1600: 14 to 50

See Occurrence rate for adversions with see. (Includes you see or we shall see, but not I see or you see not.)

0 to 5

hark | listen Occurrence rate for adversions with hark, heark, list, or listen, excluding first person or negatives.

1 to 15

Round Three Rejec-tions

Total number of rejections, this round. 0 to 1

Highlighting Aqua = individual test rejection. Composite scores or ranges in yellow, except: for Shakespeare, gray or red = composite Shakespeare rejection; for oth-ers, gray or red = comp. non-rejection.

SUMMARY OF RESULTS

No Shakespeare play had more than one rejection in these 15 tests, but 98% of the Claimant plays and 96% of the Apocrypha plays had two or more rejections. Test Sensitivities: G = genre; T = time of com-position; E = editing; P = prosody.

417

Shak

espe

are

Pla

y B

asel

ine

Dat

a, R

ound

Thr

ee T

ests

Play

Dat

e L

ate

whe

re-

|

ther

e-di

s-w

here

as |

whe

nas*

un-

ex-

fore

--a

ble

-les

s-i

sh-l

y-m

ent

very

mos

t +

mod

see

hark

| lis

ten

Num

ber

of

Tes

tsD

iscr

ete

Rej

ecti

ons

Dis

cret

e C

ompo

site

Pr

obab

ility

(2

.5%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

R3

1593

1837

060

321

1613

1215

524

818

43

150

1.00

0E+

003.

4964

6.61

94E

-01

Shr

(c)

1594

719

042

200

266

1014

120

214

310

151

3.16

0E-0

13.

9176

4.26

68E

-01

TG

V (

c)15

9419

470

6455

118

1711

151

1219

80

615

01.

000E

+00

4.64

711.

1886

E-0

1E

rr (

c)15

9411

550

5324

17

1811

9811

1418

14

151

3.16

0E-0

14.

4429

1.82

16E

-01

R2

1595

1847

061

274

1416

1613

626

1016

31

150

1.00

0E+

003.

6498

5.77

52E

-01

LL

L (

c)15

959

290

3639

419

66

109

2615

412

215

13.

160E

-01

3.75

225.

1955

E-0

1Jn

1596

643

064

338

2816

2214

722

1413

03

150

1.00

0E+

004.

4201

1.90

41E

-01

MN

D (

c)15

969

460

2820

116

1110

9814

1420

14

150

1.00

0E+

003.

4544

6.84

10E

-01

Rom

15

964

360

4829

319

710

127

2021

80

315

01.

000E

+00

2.43

539.

8089

E-0

11H

415

9715

240

4831

317

1017

128

1814

100

615

01.

000E

+00

2.83

589.

2210

E-0

1W

iv (

c)15

974

540

3522

122

112

131

1523

81

515

13.

160E

-01

3.56

306.

2585

E-0

1M

oV15

9711

210

4335

219

716

130

1929

160

215

01.

000E

+00

3.05

828.

5837

E-0

12H

415

9812

370

2836

715

513

125

2133

271

215

13.

160E

-01

3.77

265.

0797

E-0

1JC

1599

1124

032

171

286

610

219

1329

111

150

1.00

0E+

004.

2099

2.77

49E

-01

Ado

(c)

1599

343

028

483

352

117

725

2921

24

151

3.16

0E-0

15.

2488

2.45

62E

-02

AY

L (

c)15

9910

310

4232

326

722

152

1831

281

015

13.

160E

-01

3.60

696.

0155

E-0

1H

am16

0117

360

6536

421

713

121

3642

641

315

13.

160E

-01

4.66

721.

1363

E-0

1T

N (

c)16

028

400

4949

231

510

114

2441

332

015

13.

160E

-01

3.59

736.

0689

E-0

1T

ro16

0211

500

5141

820

1220

119

2316

273

1215

01.

000E

+00

4.25

022.

5931

E-0

1M

FM16

0310

460

5737

323

97

144

2734

432

415

01.

000E

+00

2.25

569.

9136

E-0

1A

WW

(c)

16

039

470

5837

430

411

157

2234

320

215

01.

000E

+00

3.13

698.

2967

E-0

1O

th

1604

1145

056

352

314

1910

220

3240

16

150

1.00

0E+

003.

0237

8.69

94E

-01

Lr

1605

1053

057

202

1911

1310

317

1733

54

150

1.00

0E+

003.

4344

6.94

47E

-01

Mac

1606

1143

064

190

2219

1411

014

1425

59

151

3.16

0E-0

14.

5750

1.39

06E

-01

Ant

1607

943

039

264

192

812

623

1643

36

150

1.00

0E+

002.

4646

9.78

47E

-01

Cor

1608

1952

046

315

207

215

014

1830

38

150

1.00

0E+

003.

6573

5.73

29E

-01

Cym

1610

1739

147

364

197

1214

035

1634

46

151

3.16

0E-0

13.

5957

6.07

78E

-01

Tm

p16

116

420

3222

432

95

146

1222

500

1515

01.

000E

+00

4.72

339.

9940

E-0

2W

T16

1119

330

5828

222

66

129

1925

355

415

01.

000E

+00

3.15

348.

2325

E-0

1

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ectio

ns0

01

00

01

10

10

23

02

110

1 P

erce

ntag

e0%

0%3%

0%0%

0%3%

3%0%

3%0%

7%10

%0%

7%3%

0%3%

435

2929

Pla

ys B

asel

ine:

Sta

ndar

d St

atis

tica

l Pro

file

Com

posi

te T

hres

hold

s B

lock

s T

este

d29

2929

2929

2929

2929

2929

2929

2929

3.16

0E-0

19.

9940

E-0

2 M

ean

11.1

740

.07

0.20

47.9

731

.62

3.00

21.8

68.

6211

.55

129.

9320

.55

21.9

026

.69

1.86

5.00

Std

Dev

4.68

9.71

0.62

11.8

09.

282.

076.

204.

835.

2819

.99

5.99

9.11

13.8

81.

613.

54Sh

Dis

cret

e R

ejec

tion

Prof

ile (

See

note

in k

ey)

Min

imum

3.00

19.0

00.

0028

.00

17.0

00.

007.

001.

001.

0098

.00

11.0

08.

004.

000.

000.

00 M

inim

um0

Max

imum

19.0

055

.00

1.00

65.0

055

.00

8.00

35.0

019

.00

22.0

017

7.00

36.0

042

.00

64.0

05.

0015

.00

Max

imum

1

Pla

ys B

asel

ine:

Sel

ecte

d St

atis

tics

by

Per

iod

to

1600

Blo

cks

1616

Mea

n19

.25

17.8

1St

d D

ev7.

579.

48 f

rom

160

0B

lock

s13

13M

ean

25.1

537

.62

Std

Dev

9.77

10.1

4

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

lew

here

- |

th

ere-

dis-

whe

reas

| w

hena

s*un

-ex

-fo

re-

-abl

e-l

ess

-ish

-ly

-men

tve

ry m

ost +

m

odse

eha

rk |

liste

n

Glo

bal M

in3

190

2817

010

21

9811

68

01

Glo

bal M

ax19

550

6555

835

1922

161

3642

505

15 M

in to

160

03

190

2817

010

21

9811

68

0.00

1 M

ax to

160

019

550

6555

835

1922

161

3631

325.

0015

Min

fro

m 1

600

319

028

170

102

198

1116

140.

001

Max

fro

m 1

600

1955

065

558

3519

2216

136

4250

5.00

15

*The

mea

n an

d st

anda

rd d

evia

tion

for

this

test

are

com

pute

d us

ing

also

the

foll

owin

g se

t-as

ides

: 2H

6, 3

H6,

H5,

TN

K (

Sh),

Tit

. (Se

e ke

y.)

Blo

cks

Tes

ted

418

Cla

iman

ts v

ersu

s Sh

akes

pear

e B

asel

ine,

Rou

nd T

hree

Tes

ts

Play

Dat

e L

ate

whe

re-

|

ther

e-di

s-w

here

as |

whe

nas

un-

ex-

fore

--a

ble

-les

s-i

sh-l

y-m

ent

very

mos

t +

mod

see

hark

| lis

ten

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

ns

Dis

cret

e C

ompo

site

Pr

obab

ilit

y (2

.5%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abil

ity

Bea

umon

t: PE

SL16

072

280

3410

113

512

133

1114

131

1015

44.

274E

-04

5.28

402.

2071

E-0

2C

hapm

an

USH

R16

0212

450

4051

1017

1319

166

1725

3131

615

35.

674E

-03

18.7

230

<1.

0000

E-1

5

BU

SS16

072

510

3140

716

1314

132

138

222

215

25.

289E

-02

4.64

241.

2010

E-0

1D

anie

l: C

LE

O15

9340

119

371

486

107

197

163

224

015

73.

291E

-08

12.3

658

<1.

0000

E-1

5D

ekke

r

WB

AB

1607

131

054

149

185

611

520

1541

142

155

2.37

7E-0

58.

8969

9.96

63E

-11

H

NW

R16

082

210

2813

37

116

759

2313

41

157

3.29

1E-0

86.

6256

1.14

06E

-04

Flet

cher

W

PR

Z16

040

280

5041

310

422

136

1527

182

215

13.

160E

-01

4.72

809.

8853

E-0

2

VA

LN

1610

040

021

242

62

417

29

315

04

156

1.00

6E-0

66.

4880

2.17

38E

-04

M

TO

M16

163

420

3540

319

47

234

126

360

815

25.

289E

-02

6.49

282.

1263

E-0

4

CH

NC

1617

038

028

110

234

415

411

1220

117

154

4.27

4E-0

46.

2723

5.69

65E

-04

L

OY

L16

180

260

4237

220

25

226

815

250

215

44.

274E

-04

6.66

479.

4551

E-0

5

DE

MT

1619

035

029

381

275

1218

86

2422

06

153

5.67

4E-0

35.

4294

1.39

48E

-02

B

AR

N16

190

480

5944

222

322

163

263

392

115

35.

674E

-03

4.96

385.

5002

E-0

2

ISL

N16

191

380

4419

223

19

231

169

282

415

44.

274E

-04

6.30

075.

0347

E-0

4G

reen

e

AL

PH15

8713

2911

329

40

2132

147

1120

163

715

52.

377E

-05

18.6

488

0.00

00E

+00

FB

FB15

9110

461

2517

47

1232

166

47

22

215

73.

291E

-08

6.84

443.

8973

E-0

5

lAM

415

9118

513

2925

23

2942

9215

36

11

157

3.29

1E-0

89.

9686

1.71

69E

-14

Hey

woo

d: H

EY

W16

031

430

3932

114

91

111

157

141

215

25.

289E

-02

4.86

617.

0750

E-0

2Jo

hnso

n

SEJA

1603

1331

541

455

145

411

917

549

82

153

5.67

4E-0

39.

4478

1.36

22E

-12

V

OL

P16

0610

631

3646

320

69

135

2016

4919

615

35.

674E

-03

11.2

481

<1.

0000

E-1

5

AL

CH

1610

1038

125

2512

131

2511

712

2213

54

156

1.00

6E-0

66.

9520

2.24

92E

-05

B

AR

T16

146

380

2432

411

116

103

1437

47

415

44.

274E

-04

6.13

981.

0001

E-0

3

NIN

N16

295

580

3654

416

122

115

2510

83

015

52.

377E

-05

5.84

523.

2268

E-0

3

TT

UB

1633

214

07

142

20

1287

229

52

515

91.

666E

-11

8.07

723.

1000

E-0

8K

yd: S

PTR

1589

2760

058

464

1446

1414

022

14

120

157

3.29

1E-0

811

.426

7<

1.00

00E

-15

Lyl

y: L

YW

M15

9713

460

3920

83

329

130

183

23

015

52.

377E

-05

6.34

174.

2049

E-0

4M

arlo

we

T

AM

115

8815

510

2643

27

4224

166

133

610

015

91.

666E

-11

10.2

345

1.62

83E

-15

T

AM

215

8819

523

2828

011

2536

138

171

08

215

61.

006E

-06

9.52

137.

4895

E-1

3

DF1

615

8812

411

1732

1214

915

155

144

717

515

61.

006E

-06

11.2

148

<1.

0000

E-1

5

JEW

M15

8922

391

3231

19

1211

152

99

67

115

61.

006E

-06

5.86

103.

0389

E-0

3

ED

W2

1592

947

451

273

1513

1011

916

46

32

153

5.67

4E-0

36.

9311

2.50

54E

-05

M

APA

1593

1952

223

2514

215

1516

214

64

84

157

3.29

1E-0

89.

0498

3.13

48E

-11

D

IDO

1586

2339

1644

191

719

1713

415

13

96

156

1.00

6E-0

626

.306

2<

1.00

00E

-15

Mid

dlet

on0.

0000

0.00

00E

+00

PH

OE

1604

432

04

546

385

1219

822

3129

44

153

5.67

4E-0

36.

8466

3.85

43E

-05

M

ICL

1606

649

148

3610

183

819

525

2123

20

154

4.27

4E-0

45.

7446

4.69

32E

-03

C

HST

1611

126

140

174

247

1914

412

1113

14

154

4.27

4E-0

44.

9763

5.32

10E

-02

N

WlT

1613

548

236

242

242

1314

017

714

02

152

5.28

9E-0

25.

1269

3.51

31E

-02

M

DIS

1615

150

037

3514

222

718

825

1024

43

154

4.27

4E-0

47.

2850

3.75

22E

-06

W

ITC

1616

130

137

304

305

318

915

1013

121

156

1.00

6E-0

67.

2153

5.51

98E

-06

H

EN

G16

182

280

4336

724

55

151

2311

258

1015

35.

674E

-03

5.76

654.

3305

E-0

3

WB

WM

1621

132

034

261

102

616

925

821

48

153

5.67

4E-0

35.

2429

2.50

03E

-02

G

AM

E16

242

680

6031

1129

107

183

3115

572

315

61.

006E

-06

6.74

866.

2822

E-0

5M

unda

y: J

KJC

1594

2564

119

214

1216

1817

915

189

47

155

2.37

7E-0

56.

3440

4.16

24E

-04

Nas

he: W

ILL

1592

1137

141

544

176

1411

321

1014

00

152

5.28

9E-0

23.

8703

4.52

91E

-01

419

Cla

iman

ts v

ersu

s Sh

akes

pear

e B

asel

ine,

Rou

nd T

hree

Tes

ts

Play

Dat

e L

ate

whe

re-

|

ther

e-di

s-w

here

as |

whe

nas

un-

ex-

fore

--a

ble

-les

s-i

sh-l

y-m

ent

very

mos

t +

mod

see

hark

| lis

ten

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

ns

Dis

cret

e C

ompo

site

Pr

obab

ility

(2

.5%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Peel

e

AR

PA15

8425

610

4914

22

456

176

246

202

415

61.

006E

-06

9.70

971.

5694

E-1

3

DB

ET

1594

953

141

287

730

413

69

01

53

156

1.00

6E-0

67.

2659

4.17

33E

-06

Pick

erin

g: H

OR

E15

6724

2712

2017

02

27

105

3710

343

815

61.

006E

-06

20.1

372

<1.

0000

E-1

5Po

rter

: AN

WO

1598

225

317

112

64

371

916

50

1815

91.

666E

-11

8.82

101.

7521

E-1

0Si

dney

: AN

TO

1590

036

072

1818

1845

915

29

09

00

157

3.29

1E-0

811

.740

4<

1.00

00E

-15

Smith

: HE

CT

1615

562

028

434

2514

2610

113

424

01

153

5.67

4E-0

35.

7713

4.25

44E

-03

Wils

on: 3

LD

Y15

8117

700

2218

014

113

144

2430

121

115

35.

674E

-03

5.40

511.

5089

E-0

2

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ectio

ns28

1022

158

916

158

2311

3123

1311

243

5049

Per

cent

age

55%

20%

43%

29%

16%

18%

31%

29%

16%

45%

22%

61%

45%

25%

22%

32%

98%

96%

Blo

cks

Tes

ted

5151

5151

5151

5151

5151

5151

5151

5176

551

51

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

whe

re-

|

ther

e-di

s-w

here

as |

whe

nas

un-

ex-

fore

--a

ble

-les

s-i

sh-l

y-m

ent

very

mos

t +

mod

see

hark

| lis

ten

Dis

cret

eC

ontin

uous

Glo

bal M

in3

190

2817

010

21

9811

68

01

3.16

0E-0

19.

9940

E-0

2 G

loba

l Max

1955

065

558

3519

2216

136

4250

515

Min

to 1

600

319

028

170

102

198

116

80.

001

Max

to 1

600

1955

065

558

3519

2216

136

3132

5.00

15Sh

Dis

cret

e R

ejec

tion

Prof

ile (

See

note

in k

ey)

Min

fro

m 1

600

319

028

170

102

198

1116

140.

001

Min

imum

0 M

ax f

rom

160

019

550

6555

835

1922

161

3642

505.

0015

Max

imum

1

420

Dub

itan

da a

nd S

et-a

side

s, A

pocr

ypha

Pla

ys v

ersu

s Sh

akes

pear

e B

asel

ine:

Rou

nd T

hree

Tes

ts

Play

Dat

e L

ate

whe

re-

|

ther

e-di

s-w

here

as |

whe

nas

un-

ex-

fore

--a

ble

-les

s-i

sh-l

y-m

ent

very

mos

t +

mod

see

hark

| lis

ten

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

ns

Dis

cret

e C

ompo

site

Pr

obab

ility

(2

.5%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Dub

itan

da a

nd S

et-a

side

s

1H6

1590

1765

340

521

1717

3412

323

79

69

154

4.27

4E-0

48.

2373

1.07

44E

-08

2H6

1591

838

238

321

2016

810

817

610

32

151

3.16

0E-0

14.

4591

1.76

44E

-01

3H6

1591

1426

348

237

1325

1311

120

83

43

153

5.67

4E-0

36.

9639

2.11

47E

-05

H5

1599

1046

034

474

235

5214

733

1632

13

151

3.16

0E-0

18.

4993

1.77

84E

-09

H8,

(Fl

)16

130

310

3422

017

622

162

811

310

315

44.

274E

-04

5.21

852.

6896

E-0

2H

8, (

Jt)

1613

020

035

200

515

1018

630

535

00

155

2.37

7E-0

56.

4965

2.09

03E

-04

H8,

(Sh

)16

1318

370

4535

1013

35

147

258

4210

015

44.

274E

-04

7.13

618.

4975

E-0

6Pe

r, 1

-216

085

385

4838

013

85

8723

85

30

155

2.37

7E-0

59.

3784

2.38

44E

-12

Per,

3-5

1608

1228

040

262

268

1014

716

836

48

151

3.16

0E-0

13.

1894

8.08

77E

-01

Tim

1608

935

047

432

419

216

024

1620

30

152

5.28

9E-0

24.

9123

6.29

07E

-02

TN

K (

Fl)

1613

016

019

140

45

311

75

237

00

158

8.39

1E-1

07.

4055

1.89

88E

-06

TN

K (

Sh)

1613

1439

045

321

1015

016

121

1429

08

152

5.28

9E-0

24.

2623

2.53

98E

-01

Tit.

1594

946

146

200

2117

713

613

98

119

152

5.28

9E-0

26.

9042

2.87

63E

-05

Tit

earl

y15

940

620

3613

09

238

113

156

615

915

73.

291E

-08

10.2

718

1.16

69E

-15

Tit

late

1594

2126

356

280

3613

516

913

1310

810

155

2.37

7E-0

57.

7605

2.30

67E

-07

STM

(Sh

)15

9514

140

430

1414

014

101

00

014

015

91.

666E

-11

11.6

313

2.16

84E

-19

Apo

cryp

ha

HO

RE

1567

2427

1220

170

22

710

537

1034

38

156

1.00

6E-0

620

.137

2<

1.00

00E

-15

FVH

515

8816

218

3011

1121

013

104

1129

133

015

52.

377E

-05

13.9

871

<1.

0000

E-1

5T

OA

S15

893

213

138

28

516

188

58

33

1015

73.

291E

-08

8.43

762.

7361

E-0

9IR

ON

1590

459

1959

489

2520

2412

216

85

77

157

3.29

1E-0

830

.899

0<

1.00

00E

-15

AR

DN

1592

1436

230

143

1115

711

08

152

99

155

2.37

7E-0

57.

0509

1.34

02E

-05

YR

KI

1592

1115

425

169

1410

2187

142

50

215

88.

391E

-10

8.86

821.

2346

E-1

0Y

RK

215

9212

315

4427

29

2012

108

185

59

215

61.

006E

-06

9.88

513.

5774

E-1

4G

UY

W15

936

192

386

06

1620

165

69

93

215

52.

377E

-05

7.00

451.

7112

E-0

5L

EIR

1593

953

1258

606

1220

1314

117

1010

24

153

5.67

4E-0

319

.585

5<

1.00

00E

-15

RC

D3

1594

1438

545

249

1316

615

318

86

20

154

4.27

4E-0

49.

0345

3.52

37E

-11

STM

O15

9513

370

2333

713

1121

130

2118

1725

415

25.

289E

-02

14.8

317

<1.

0000

E-1

5E

DW

315

9515

762

4542

422

2531

132

213

92

315

52.

377E

-05

7.42

421.

7056

E-0

6K

JN1

1595

2452

454

374

418

2512

68

14

34

157

3.29

1E-0

88.

8622

1.29

10E

-10

KJN

215

9517

640

5229

617

2331

145

64

60

415

61.

006E

-06

6.82

394.

3206

E-0

5L

OC

R15

958

228

5227

325

3925

152

1312

105

115

35.

674E

-03

14.7

177

<1.

0000

E-1

5W

OO

D15

959

410

6130

710

1323

194

326

347

515

44.

274E

-04

6.67

578.

9661

E-0

5M

UC

E15

9817

535

4127

29

199

112

2919

155

1415

25.

289E

-02

9.25

926.

1577

E-1

2O

LD

C16

0012

380

2625

316

232

132

2112

122

015

52.

377E

-05

5.66

736.

2036

E-0

3C

RO

M16

024

364

3219

19

1013

929

74

13

156

1.00

6E-0

68.

4344

2.79

75E

-09

DE

VL

1604

729

035

177

55

714

319

2616

214

151

3.16

0E-0

15.

4633

1.24

83E

-02

PRO

D16

050

260

4014

010

1017

785

549

53

156

1.00

6E-0

67.

2916

3.61

65E

-06

PUR

N16

075

181

3440

243

110

186

1126

172

615

52.

377E

-05

6.31

344.

7626

E-0

4Y

KSH

1608

335

098

667

287

1712

67

1414

07

154

4.27

4E-0

47.

3880

2.09

85E

-06

MA

ID16

111

371

3723

226

35

155

2812

180

815

35.

674E

-03

4.74

369.

5307

E-0

2FA

LS

1613

664

056

213

1213

1212

331

1616

126

152

5.28

9E-0

27.

7916

1.90

41E

-07

FAIR

1631

5473

376

317

3314

1623

628

1410

50

158

8.39

1E-1

013

.203

5-2

.168

4E-1

9M

ER

L16

317

451

2533

1111

812

127

201

2315

1115

52.

377E

-05

10.0

996

5.42

68E

-15

RV

GR

1606

656

061

3515

2513

917

925

920

84

155

2.37

7E-0

58.

2005

1.37

44E

-08

Apo

cryp

ha D

iscr

ete

Dis

crim

inat

ion

Stat

isti

csA

ggre

gate

sA

pocr

ypha

Com

p D

iscr

imin

atio

n St

ats

Rej

ectio

ns5

819

88

69

87

99

1314

84

135

2728

Per

cent

age

18%

29%

68%

29%

29%

21%

32%

29%

25%

32%

32%

46%

50%

29%

14%

32%

96%

100%

Blo

cks

Tes

ted

2828

2828

2828

2828

2828

2828

2828

2842

028

28

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

whe

re-

|

ther

e-di

s-w

here

as |

whe

nas

un-

ex-

fore

--a

ble

-les

s-i

sh-l

y-m

ent

very

mos

t +

mod

see

hark

| lis

ten

Dis

cret

eC

ontin

uous

Glo

bal M

in3

190

2817

010

21

9811

68

01

3.16

0E-0

19.

994E

-02

Glo

bal M

ax19

550

6555

835

1922

161

3642

505

15 M

in to

160

03

190

2817

010

21

9811

68

0.00

1 M

ax to

160

019

550

6555

835

1922

161

3631

325.

0015

Sh D

iscr

ete

Rej

ectio

n Pr

ofile

(Se

e no

te in

key

) M

in f

rom

160

03

190

2817

010

21

9811

1614

0.00

1 M

inim

um0

Max

fro

m 1

600

1955

065

558

3519

2216

136

4250

5.00

15 M

axim

um1

421

APPENDIX SIX, PLAY VERSE: KEY TO TABLES FOR 3,000-WORD PLAY VERSE TESTS





HC / 20K: E Hyphenated compound words per 20,000 words. 31 to 153

Fem Endings: P Percentage of feminine endings of verse lines. Mostly machine counts.

7 to 25

Open Lines: T, E, P Percentage of open or run-on verse lines. Early, to 1600: 9 to 33 Late, fr. 1600: 15 to 57

Enclitics: P Enclitic microphrases per 1000 lines. 27 to 89

Proclitics: P Proclitic microphrases per 1000 lines. 265 to 476


4 to 36


167 to 586

BoB5 Bundles of badges 5. See text for components. 116 to 556


BoB8 Bundles of badges 8. See text for components. -867 to -265

T-E Slope Test Thisted-Efron Slope test. -0.22 to 0.15

T-E New Word Test Thisted-Efron New Word Test. -32 to 21

T-E Rare Word Test Thisted-Efron Rare Word Test. -33 to 218

Bucket Block: G Modal Bucket Score per Block. -72 to 79

Discrete Rejections Total number of rejections from tests above. 0 to 1


See Table 2. Probability that the observed rejec-tions would occur by chance at Shakespeare’s aver-age rejection rate. Profile minimum probability is Shakespeare’s lowest.

2.900E-01


See Table 2. Probability that the observed compos-ite probability score would occur by chance. Profile minimum probability is Shakespeare’s lowest.

1.78300E-01

Highlighting Aqua = individual test rejection. Composite scores or ranges in yellow, except: for Shakespeare, gray or red = composite rejection; for others, gray or red = comp. non-rejection.

SUMMARY OF RESULTS FOR 3,000-WORD PLAY VERSE BLOCKS

Four of Shakespeare’s 82 baseline blocks (5%) have over one individual rejection in 13-15 tests. All of 38 blocks by others have two or more individual rejections in 13-15 tests. Discrete and continuous com-posite scoring, respectively, pass 95% and 84% of Shakespeare’s blocks and reject 100% of others’ blocks. Perhaps 2 of 38 others’ blocks (5%) are close calls. Test Sensitivities: G = genre; T = time of composition; E = editing; P= prosody.

422

Shak

espe

are

Poe

ms

Bas

elin

e D

ata,

Blo

cksi

ze =

3,0

00

Poem

and

Blo

ckG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Pro

clit

ics

/100

0 lin

esw

ith

(2

lws)

no

/(no

+no

t)B

oB5

BoB

7B

oB8

T-E

Slo

pe

Tes

tT

-E N

ew

Wor

dsM

odal

Blo

ckN

umbe

r of

T

ests

Dis

cret

e R

ejec

tion

s

Dis

cret

e C

ompo

site

Pr

obab

ilit

y

Con

tinu

ous

Com

posi

te

Err

or

Con

tinu

ous

Com

posi

te

Prob

abil

ity

Ven

us1

1015

315

742

270

623

321

182

6-5

93-0

.12

-32

467

141

3.08

4E-0

14.

4067

1.49

55E

-01

210

148

259

4031

76

326

325

579

-536

-0.1

5-2

628

114

01.

000E

+00

4.63

468.

9960

E-0

23

1211

814

1147

317

3441

724

454

2-4

19-0

.13

-711

4914

01.

000E

+00

4.00

863.

0919

E-0

1L

ucre

ce1

1113

39

1734

278

1153

655

694

1-4

05-0

.11

-20

358

140

1.00

0E+

004.

0782

2.76

34E

-01

211

153

1513

4226

56

324

270

867

-580

-0.1

6-8

535

140

1.00

0E+

002.

8607

8.79

54E

-01

312

152

1318

3130

119

351

538

714

-541

-0.2

1-2

096

914

01.

000E

+00

3.55

365.

5600

E-0

14

1213

314

1059

350

1939

119

871

4-6

09-0

.10

-299

414

01.

000E

+00

2.60

749.

4220

E-0

15

1111

413

2165

338

1750

026

252

9-6

25-0

.08

-671

214

01.

000E

+00

3.16

377.

6153

E-0

1So

nnet

s1

1498

1219

7733

415

184

429

818

-300

0.00

-184

214

01.

000E

+00

4.30

331.

8418

E-0

12

1368

815

6136

722

333

282

840

-560

-0.0

87

740

140

1.00

0E+

002.

7706

9.05

50E

-01

313

883

1843

316

750

047

894

4-5

100.

019

543

141

3.08

4E-0

13.

9512

3.37

64E

-01

412

508

1548

321

733

328

861

1-5

20-0

.14

660

114

01.

000E

+00

2.78

019.

0296

E-0

15

1256

1219

8736

07

313

285

818

-560

0.07

-333

114

01.

000E

+00

3.78

924.

2339

E-0

16

1210

47

1781

476

1229

011

682

6-4

120.

060

892

140

1.00

0E+

004.

5704

1.04

52E

-01

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ecti

ons

00

11

00

00

00

00

00

20

0 P

erce

ntag

e0%

0%7%

7%0%

0%0%

0%0%

0%0%

0%0%

0%1%

0%0%

196

1414

Poe

ms

Bas

elin

e: S

tand

ard

Stat

isti

cal P

rofi

leC

ompo

site

Thr

esho

lds

Blo

cks

Tes

ted

1414

1414

1414

1414

1414

1414

1414

3.08

4E-0

18.

9960

E-0

2 M

ean

11.7

911

2.09

12.0

014

.93

54.0

732

9.29

13.4

335

9.36

320.

1475

4.93

-512

.14

-0.0

8-7

.33

672.

60 S

td D

ev1.

0834

.75

4.96

4.13

17.2

450

.74

7.89

97.3

212

6.44

136.

1890

.43

0.08

12.2

326

1.02

Sh D

iscr

ete

Rej

ectio

n Pr

ofile

(Se

e no

te in

key

) M

inim

um10

.00

49.8

83.

007.

0031

.00

265.

006.

0018

4.00

116.

0052

9.00

-625

.00

-0.2

1-3

2.28

281.

21 M

inim

um0

Max

imum

14.0

015

3.11

25.0

021

.00

87.0

047

6.00

34.0

053

6.00

556.

0094

4.00

-300

.00

0.07

9.16

1149

.25

Max

imum

1

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

le

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

litic

s/1

000

lines

Pro

clit

ics

/100

0 lin

esw

ith

(2

lws)

no

/(no

+no

t)B

oB5

BoB

7B

oB8

T-E

Slo

pe

Tes

tT

-E N

ew

Wor

dsM

odal

Blo

ck

Glo

bal M

in10

317

927

265

416

711

613

6-8

67-0

.22

-32

281

Glo

bal M

ax14

153

2557

8947

636

536

556

944

-265

0.15

2111

49 M

in to

160

010

317

927

265

416

711

613

6-8

67-0

.22

-32

281

Max

to 1

600

1415

325

3389

476

3653

655

694

4-2

650.

1521

1149

Min

fro

m 1

600

1031

715

2726

54

167

116

136

-867

-0.2

2-3

228

1 M

ax f

rom

160

014

153

2557

8947

636

536

556

944

-265

0.15

2111

49

Blo

cks

Tes

ted

423

Oth

er P

oets

ver

sus

Shak

espe

are

Bas

elin

e, B

lock

size

= 3

,000

Poe

m a

nd B

lock

Gra

de L

evel

HC

W

/2

0K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

line

sPr

oclit

ics

/100

0 li

nes

with

(2lw

s)no

/(

no+

not)

BoB

5B

oB7

BoB

8T

-E S

lope

T

est

T-E

NW

Mod

alB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ecti

ons

Dis

cret

e C

ompo

site

P

roba

bili

ty

(2.6

%)

Con

tinu

ous

Com

posi

te

Err

or

Con

tinu

ous

Com

posi

te

Prob

abil

ity

Ale

xand

er, A

uror

a

113

453

11bl

ank

blan

k11

380

4668

0-3

50-0

.07

-74

1123

124

1.91

3E-0

47.

0781

1.34

20E

-06

219

704

9bl

ank

blan

k8

385

-142

667

-280

-0.1

2-9

910

0912

58.

075E

-06

11.4

092

<1.

0000

E-1

53

1356

48

blan

kbl

ank

1243

810

776

5-4

16-0

.07

-79

1697

125

8.07

5E-0

67.

9651

5.28

26E

-09

Bac

on P

oem

s12

212

618

149

1128

661

776

0-3

94-0

.13

-11

3159

147

2.34

7E-0

810

.711

3<

1.00

00E

-15

Bar

nefe

ld1

1431

513

20bl

ank

blan

k0

200

179

1000

-415

-0.2

6-6

968

412

58.

075E

-06

9.03

532.

0095

E-1

22

1620

712

13bl

ank

blan

k11

308

473

875

-561

-0.0

2-3

252

312

23.

752E

-02

5.54

082.

1880

E-0

33

1319

713

14bl

ank

blan

k11

462

355

1000

-438

-0.1

3-4

679

012

33.

242E

-03

4.78

372.

8729

E-0

24

1280

127

521

014

241

261

1000

-553

-0.1

3-4

047

614

51.

954E

-05

5.59

505.

0249

E-0

35

1611

413

1118

176

435

687

1000

-150

-0.1

1-3

569

514

72.

347E

-08

7.67

021.

8782

E-0

7B

arne

s1

688

1918

blan

kbl

ank

433

312

071

4-1

60-0

.24

-44

1037

124

1.91

3E-0

48.

0893

2.26

35E

-09

25

161

2216

blan

kbl

ank

152

045

1000

133

-0.3

0-6

984

712

89.

414E

-11

11.9

386

<1.

0000

E-1

53

482

1319

blan

kbl

ank

1240

018

810

0023

-0.1

5-4

780

612

41.

913E

-04

10.2

305

<1.

0000

E-1

54

514

011

10bl

ank

blan

k8

429

139

889

192

-0.2

2-3

319

8612

41.

913E

-04

11.7

822

<1.

0000

E-1

55

560

2618

blan

kbl

ank

1537

5-7

1000

182

-0.1

9-5

311

1512

62.

494E

-07

11.7

339

<1.

0000

E-1

56

253

2513

blan

kbl

ank

716

7-1

6738

5-3

98-0

.17

-24

4650

123

3.24

2E-0

318

.856

6<

1.00

00E

-15

75

4719

18bl

ank

blan

k3

167

139

1000

11-0

.23

-85

857

126

2.49

4E-0

711

.575

6<

1.00

00E

-15

85

4017

16bl

ank

blan

k10

286

118

1000

10-0

.11

-65

854

124

1.91

3E-0

410

.376

0<

1.00

00E

-15

95

6010

16bl

ank

blan

k13

240

155

1000

0-0

.15

-24

848

123

3.24

2E-0

39.

1388

8.76

63E

-13

103

6722

14bl

ank

blan

k3

396

249

1000

-85

-0.2

2-4

812

3812

62.

494E

-07

10.8

761

<1.

0000

E-1

5C

hapm

an, H

ero

118

9010

3174

384

2425

013

976

0-5

83-0

.10

-45

952

142

5.00

0E-0

28.

2274

5.04

17E

-09

216

859

2557

305

1224

0-1

1169

2-6

00-0

.19

-40

622

143

5.15

9E-0

36.

7655

3.05

47E

-05

316

367

33bl

ank

blan

k11

400

308

571

-460

-0.1

0-2

393

312

12.

710E

-01

6.73

209.

0825

E-0

64

1479

431

blan

kbl

ank

036

035

781

8-5

21-0

.11

-35

1044

123

3.24

2E-0

35.

7589

9.12

27E

-04

Dan

iel,

Del

ia1

1368

1113

917

318

500

-14

810

-521

-0.0

9-1

010

3514

35.

159E

-03

5.57

065.

4873

E-0

32

1217

1117

1813

311

462

147

1000

-579

0.01

281

014

43.

710E

-04

6.01

629.

7565

E-0

4D

onne

, Poe

ms

110

134

822

blan

kbl

ank

1134

123

486

2-1

32-0

.04

-95

907

122

3.75

2E-0

28.

8381

9.44

50E

-12

212

315

22bl

ank

blan

k7

333

277

556

-211

-0.1

5-5

617

9312

41.

913E

-04

7.70

502.

9435

E-0

83

1455

526

blan

kbl

ank

034

723

944

0-2

99-0

.02

-35

1508

124

1.91

3E-0

46.

7963

6.43

52E

-06

413

489

31bl

ank

blan

k0

436

234

667

-173

-0.2

2-5

242

112

33.

242E

-03

7.48

681.

1714

E-0

75

1345

1431

blan

kbl

ank

1238

923

157

7-2

22-0

.14

-61

1216

123

3.24

2E-0

37.

5768

6.67

07E

-08

613

918

25bl

ank

blan

k5

176

268

528

-447

-0.0

6-5

374

012

12.

710E

-01

5.51

732.

3959

E-0

37

1353

729

blan

kbl

ank

037

716

058

8-2

83-0

.12

-76

2044

123

3.24

2E-0

39.

4100

9.42

59E

-14

Dra

yton

, Ide

a1

755

122

blan

kbl

ank

848

836

681

8-4

020.

05-2

281

312

23.

752E

-02

6.36

675.

8592

E-0

52

1386

84

blan

kbl

ank

033

322

777

8-3

66-0

.08

-59

1069

123

3.24

2E-0

35.

9962

3.29

50E

-04

Dye

r Po

ems

114

81

9bl

ank

blan

k8

395

387

1000

-525

0.10

1265

512

33.

242E

-03

5.63

001.

5408

E-0

32

110

22

blan

kbl

ank

2150

027

310

00-4

460.

050

2092

125

8.07

5E-0

68.

0145

3.77

89E

-09

E.C

., E

sq.

1922

516

13bl

ank

blan

k5

216

234

1000

-794

0.16

-57

634

125

8.07

5E-0

69.

9119

1.22

72E

-15

Ele

gy22

101

1246

2421

18

8960

976

9-8

8-0

.10

-24

789

146

7.74

9E-0

713

.855

6<

1.00

00E

-15

Ess

ex P

oem

s12

6012

2bl

ank

blan

k6

444

137

938

-562

0.06

-30

1672

122

3.75

2E-0

26.

2435

1.05

98E

-04

Ford

, Fam

e's

Mem

oria

l 1

1918

818

1712

190

020

065

410

0016

7-0

.25

-112

495

1410

1.28

7E-1

314

.629

5<

1.00

00E

-15

218

123

1932

514

30

200

518

1000

265

-0.1

9-6

794

314

72.

347E

-08

13.6

335

<1.

0000

E-1

53

1312

720

2415

182

752

462

590

525

0-0

.19

-67

2045

146

7.74

9E-0

712

.527

0<

1.00

00E

-15

Ford

, Chr

ist's

Blo

ody

Sw

eat

116

5810

2228

279

050

064

975

0-2

29-0

.17

-568

214

43.

710E

-04

6.83

932.

0922

E-0

52

1611

520

2518

280

816

771

875

0-3

26-0

.14

-26

1109

143

5.15

9E-0

37.

2993

1.70

48E

-06

311

3117

2721

241

534

369

593

5-7

1-0

.09

-113

8414

51.

954E

-05

8.08

511.

3146

E-0

84

1476

1723

1924

77

366

382

667

-364

-0.1

2-1

665

914

25.

000E

-02

4.65

478.

5738

E-0

25

1668

1122

3023

20

303

603

1000

-200

-0.1

1-1

090

514

67.

749E

-07

7.05

136.

7997

E-0

6G

reen

e Po

ems

112

713

110

139

952

529

278

9-6

71-0

.18

-35

473

144

3.71

0E-0

46.

5789

7.72

15E

-05

213

5412

17bl

ank

blan

k21

455

336

854

-639

-0.2

1-6

442

712

12.

710E

-01

5.80

697.

4659

E-0

43

1211

67

11bl

ank

blan

k14

541

296

882

-784

-0.2

5-5

354

212

33.

242E

-03

5.81

687.

1611

E-0

4

424

Oth

er P

oets

ver

sus

Shak

espe

are

Bas

elin

e, B

lock

size

= 3

,000

Poe

m a

nd B

lock

Gra

de L

evel

HC

W

/2

0K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

line

sPr

oclit

ics

/100

0 li

nes

with

(2lw

s)no

/(

no+

not)

BoB

5B

oB7

BoB

8T

-E S

lope

T

est

T-E

NW

Mod

alB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ecti

ons

Dis

cret

e C

ompo

site

P

roba

bili

ty

(2.6

%)

Con

tinu

ous

Com

posi

te

Err

or

Con

tinu

ous

Com

posi

te

Pro

babi

lity

Gri

ffin

, Fid

essa

14

6215

7bl

ank

blan

k6

325

165

955

-660

-0.0

1-2

520

3512

41.

913E

-04

9.77

304.

1960

E-1

52

563

1312

blan

kbl

ank

1947

711

494

9-6

74-0

.13

-44

803

124

1.91

3E-0

47.

8418

1.20

42E

-08

Hey

woo

d, O

enon

e 19

223

234

2019

40

111

258

1000

-439

-0.3

2-5

066

714

101.

287E

-13

8.59

593.

7803

E-1

02

919

226

412

154

1737

515

290

9-6

20-0

.26

-102

594

148

5.45

1E-1

010

.692

3<

1.00

00E

-15

Hey

woo

d, T

roia

116

156

1816

blan

kbl

ank

722

953

681

8-2

07-0

.21

-45

840

124

1.91

3E-0

46.

9117

3.42

51E

-06

221

5321

8bl

ank

blan

k9

478

569

667

-396

-0.1

0-1

737

612

33.

242E

-03

9.55

852.

6856

E-1

43

2273

188

blan

kbl

ank

066

761

810

00-3

26-0

.10

-85

757

127

2.34

7E-0

812

.713

9<

1.00

00E

-15

420

102

1119

blan

kbl

ank

2435

037

944

4-5

27-0

.15

-51

806

122

3.75

2E-0

28.

9333

4.49

87E

-12

Jons

on, s

elec

ted

1039

622

blan

kbl

ank

530

847

881

8-3

630.

02-8

589

412

23.

752E

-02

7.76

392.

0085

E-0

8L

odge

, Ros

alyn

de10

9210

73

188

737

916

010

00-3

61-0

.06

-16

1259

145

1.95

4E-0

56.

1124

6.50

25E

-04

Lov

er's

Com

plai

nt14

109

1120

1226

70

120

335

500

-452

-0.2

2-3

357

214

43.

710E

-04

5.79

172.

4035

E-0

3M

arlo

we,

Her

o an

d L

eand

er1

942

818

5129

816

296

292

1000

-238

-0.1

4-1

697

514

35.

159E

-03

5.26

431.

5529

E-0

22

913

919

3020

67

240

448

625

-383

-0.2

92

1072

144

3.71

0E-0

46.

2782

3.14

49E

-04

Mer

itum

90

112

2020

418

400

203

1000

-263

0.17

-39

1717

1410

1.28

7E-1

38.

7715

1.04

13E

-10

Mid

dlet

on, G

host

1424

39

11bl

ank

blan

k8

444

297

515

-471

-0.2

5-9

920

7612

41.

913E

-04

10.6

522

<1.

0000

E-1

5N

ashe

Poe

ms

818

55

10bl

ank

blan

k0

519

322

882

-675

-0.2

4-7

215

9512

72.

347E

-08

8.61

095.

3230

E-1

1O

xfor

d P

oem

s*7

320

713

115

550

029

010

00-3

01-0

.01

-928

9214

67.

749E

-07

11.9

588

<1.

0000

E-1

5P

eele

, Tal

e of

Tro

y10

632

1014

225

1844

447

910

00-7

67-0

.20

-769

814

43.

710E

-04

5.98

561.

1074

E-0

3Q

E1

Tra

nsla

tions 1

1027

933

blan

kbl

ank

747

649

810

00-4

510.

12-1

0661

812

33.

242E

-03

10.3

752

<1.

0000

E-1

52

87

1137

blan

kbl

ank

715

267

110

00-2

670.

13-1

0413

9612

89.

414E

-11

12.2

455

<1.

0000

E-1

53

1235

1124

blan

kbl

ank

1551

759

510

00-5

450.

15-1

1012

8512

41.

913E

-04

10.2

561

<1.

0000

E-1

54

1423

1634

blan

kbl

ank

035

163

310

00-4

350.

16-7

315

8712

89.

414E

-11

9.76

774.

3952

E-1

5R

alei

gh P

oem

s1

1258

1510

3019

00

347

338

905

-302

-0.1

5-2

378

414

25.

000E

-02

5.07

662.

7669

E-0

22

2071

1516

1620

10

333

200

1000

-56

-0.1

3-1

217

1314

72.

347E

-08

10.9

459

<1.

0000

E-1

5S

ackv

ille

, Mir

ror

2318

324

blan

kbl

ank

1166

721

210

00-6

57-0

.18

-87

1090

126

2.49

4E-0

713

.647

8<

1.00

00E

-15

P. S

idne

y, A

stro

112

803

21bl

ank

blan

k6

318

204

1000

-315

-0.0

7-4

210

1112

33.

242E

-03

5.08

211.

1352

E-0

22

1116

52

26bl

ank

blan

k13

341

191

1000

-473

-0.0

1-2

913

7712

41.

913E

-04

5.47

022.

8678

E-0

3M

. Sid

ney,

Psa

lms

1081

1224

blan

kbl

ank

644

444

310

00-2

71-0

.18

-67

2418

123

3.24

2E-0

39.

5454

3.00

31E

-14

Wm

. Sm

ith, C

hlor

is1

714

66

19bl

ank

blan

k9

304

157

1000

342

-0.2

6-4

764

912

62.

494E

-07

11.5

259

<1.

0000

E-1

52

845

1617

blan

kbl

ank

1251

225

810

0032

-0.0

8-2

779

212

33.

242E

-03

7.86

421.

0381

E-0

8Sp

ense

r, A

mor

etti 1

1361

710

1026

79

333

179

1000

-301

-0.1

1-3

464

314

35.

159E

-03

5.40

069.

9215

E-0

32

1046

311

blan

kbl

ank

1040

016

710

00-3

09-0

.09

-10

894

122

3.75

2E-0

24.

6256

4.48

71E

-02

312

414

14bl

ank

blan

k8

160

205

1000

-293

0.00

-15

885

123

3.24

2E-0

34.

8480

2.37

46E

-02

War

ner,

Alb

ion'

s13

142

1724

blan

kbl

ank

034

530

310

00-3

860.

18-8

870

512

41.

913E

-04

8.35

493.

4846

E-1

0W

ebst

er, P

oem

s14

127

1039

blan

kbl

ank

025

043

925

0-3

77-0

.04

-113

1439

123

3.24

2E-0

311

.940

0<

1.00

00E

-15

Wil

lobi

e, A

visa

112

482

4bl

ank

blan

k0

261

123

750

-742

0.16

-60

1529

126

2.49

4E-0

78.

0936

2.19

74E

-09

211

222

4bl

ank

blan

k0

328

329

448

-754

0.19

-34

1303

127

2.34

7E-0

87.

4243

1.72

23E

-07

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ecti

ons

4223

2519

2222

248

2342

2718

5728

380

8787

Per

cent

age

48%

26%

29%

22%

76%

76%

28%

9%26

%48

%31

%21

%66

%32

%34

%10

0%10

0% B

lock

s T

este

d87

8787

8729

2987

8787

8787

8787

8711

0287

87

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

litic

s/1

000

line

sPr

oclit

ics

/100

0 li

nes

with

(2lw

s)no

/(

no+

not)

BoB

5B

oB7

BoB

8T

-E S

lope

T

est

T-E

New

W

ords

Mod

alB

lock

Dis

cret

eC

onti

nuou

s

Glo

bal M

in10

317

927

265

416

711

613

6-8

67-0

.22

-32

281

3.08

4E-0

18.

9960

E-0

2 G

loba

l Max

1415

325

5789

476

3653

655

694

4-2

650.

1521

1149

Min

to 1

600

1031

79

2726

54

167

116

136

-867

-0.2

2-3

2.28

281

Max

to 1

600

1415

3.11

0048

2533

8947

636

536

556

944

-265

0.15

2111

49.2

5Sh

Dis

cret

e R

ejec

tion

Prof

ile (

See

note

in k

ey)

Min

fro

m 1

600

1031

715

2726

54

167

116

136

-867

-0.2

2-3

2.28

281

Min

imum

0 M

ax f

rom

160

014

153.

1100

4825

5789

476

3653

655

694

4-2

650.

1521

1149

.25

Max

imum

1

*OL

, FE

, enc

liti

cs, a

nd p

rocl

itic

s ar

e I

-5-

test

s, c

ompa

ring

Oxf

ord'

s I-

5 po

ems

(130

4 w

ords

) w

ith

Shak

espe

are'

s 15

00-w

ord

bloc

k p

oem

pro

file

. On

thre

e of

thes

e O

xfor

d sc

ores

bel

ow S

hake

spea

re's

150

0-w

ord

prof

ile, b

ut O

pen

Lin

es b

arel

y fi

ts it

.

425

Oth

er P

oets

ver

sus

Shak

espe

are

Bas

elin

e, B

lock

size

= 3

,000

, Agg

rega

te A

naly

sis

for

Nar

row

Rej

ecti

ons

Poem

and

Blo

ckG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

with

(2lw

s)no

/(

no+

not)

BoB

5B

oB7

BoB

8T

-E S

lope

T

est

T-E

NW

Mod

alB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

ns

Dis

cret

e C

ompo

site

Pr

obab

ility

(2

.6%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Ford

, Chr

ist's

Blo

ody

Swea

t1

1658

1022

2827

90

500

649

750

-229

-0.1

7-5

682

144

3.71

0E-0

46.

8393

2.09

22E

-05

216

115

2025

1828

08

167

718

750

-326

-0.1

4-2

611

0914

35.

159E

-03

7.29

931.

7048

E-0

63

1131

1727

2124

15

343

695

935

-71

-0.0

9-1

1384

145

1.95

4E-0

58.

0851

1.31

46E

-08

414

7617

2319

247

736

638

266

7-3

64-0

.12

-16

659

142

5.00

0E-0

24.

6547

8.57

38E

-02

516

6811

2230

232

030

360

310

00-2

00-0

.11

-10

905

146

7.74

9E-0

77.

0513

6.79

97E

-06

Ave

rage

s14

.60

69.6

915

.00

23.8

023

.20

255.

804.

0033

5.80

609.

4082

0.40

-238

.00

-0.1

3-1

1.72

947.

79A

ggre

gate

Val

ues

Std

. Err

ors

5.82

-2.7

31.

354.

80-4

.00

-3.2

4-2

.67

-0.5

45.

121.

086.

78-1

.22

-0.8

02.

3614

201.

887E

-15

13.4

588

<1.

0000

E-1

5

Ral

eigh

Poe

ms

112

5815

1030

190

034

733

890

5-3

02-0

.15

-23

784

142

5.00

0E-0

25.

0766

2.76

69E

-02

220

7115

1616

201

033

320

010

00-5

6-0

.13

-12

1713

147

2.34

7E-0

810

.945

9<

1.00

00E

-15

Ave

rage

s16

.00

64.7

515

.00

13.0

023

.00

195.

500.

0034

0.00

269.

0095

2.50

-179

.00

-0.1

4-1

7.68

1248

.50

Agg

rega

te V

alue

s S

td. E

rror

s5.

51-1

.93

0.86

-0.6

6-2

.55

-3.7

3-2

.41

-0.2

8-0

.57

2.05

5.21

-1.0

1-1

.20

3.12

149

2.39

4E-0

810

.267

7<

1.00

00E

-15

Spen

ser,

Am

oret

ti 113

617

1010

267

933

317

910

00-3

01-0

.11

-34

643

143

5.15

9E-0

35.

4006

9.92

15E

-03

210

463

11bl

ank

blan

k10

400

167

1000

-309

-0.0

9-1

089

412

23.

752E

-02

4.62

564.

4871

E-0

23

1241

414

blan

kbl

ank

816

020

510

00-2

930.

00-1

588

512

33.

242E

-03

4.84

802.

3746

E-0

2

Ave

rage

s11

.67

49.4

84.

6711

.67

10.0

026

7.00

9.00

297.

6718

3.67

1000

.00

-301

.00

-0.0

7-1

9.54

807.

30A

ggre

gate

Val

ues

Std

. Err

ors

-0.1

9-3

.12

-2.5

6-1

.37

-4.4

3-2

.13

-0.9

7-1

.10

-1.8

73.

124.

040.

31-1

.73

0.89

148

5.08

0E-0

68.

8284

6.80

53E

-11

Poe

ms

Bas

elin

e: S

tand

ard

Stat

isti

cal P

rofi

le B

lock

s T

este

d14

1414

1414

1414

1414

1414

1414

14 M

ean

11.7

911

2.09

12.0

014

.93

54.0

732

9.29

13.4

335

9.36

320.

1475

4.93

-512

.14

-0.0

8-7

.33

672.

60 S

td D

ev1.

0834

.75

4.96

4.13

17.2

450

.74

7.89

97.3

212

6.44

136.

1890

.43

0.08

12.2

326

1.02

Min

imum

10.0

049

.88

3.00

7.00

31.0

026

5.00

6.00

184.

0011

6.00

529.

00-6

25.0

0-0

.21

-32.

2828

1.21

Max

imu m

14.0

015

3.11

25.0

021

.00

87.0

047

6.00

34.0

053

6.00

556.

0094

4.00

-300

.00

0.07

9.16

1149

.25

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

le

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

with

(2lw

s)no

/(

no+

not)

BoB

5B

oB7

BoB

8T

-E S

lope

T

est

T-E

New

W

ords

Mod

alB

lock

Glo

bal M

in10

317

927

265

416

711

613

6-8

67-0

.22

-32

281

Glo

bal M

ax14

153

2557

8947

636

536

556

944

-265

0.15

2111

49 M

in to

160

010

317

927

265

416

711

613

6-8

67-0

.22

-32.

2828

1 M

ax to

160

014

153.

1100

4825

3389

476

3653

655

694

4-2

650.

1521

1149

.25

Min

fro

m 1

600

1031

715

2726

54

167

116

136

-867

-0.2

2-3

2.28

281

Max

fro

m 1

600

1415

3.11

0048

2557

8947

636

536

556

944

-265

0.15

2111

49.2

5

426

APPENDIX SIX, POEMS: KEY TO TABLES FOR 3,000-WORD POEM TESTS






Relative Clauses Total relative clauses per 1,000 words. 7 to 17 (not shown)


7 to 25





4 to 36


167 to 536



BoB8 Bundles of badges 8. See text for components. -867 to -265

T-E Slope Test Thisted-Efron Slope test. -0.22 to 0.15


Modal Block: G Modal Score per Block. 281 to 1149


Discrete Comp. Prob. See Table 2. Probability that observed rejections would occur by chance at Sh’s avg. rejection rate.

3.084E-01

Continuous Comp. Prob.

See Table 2. Prob. that observed comp. probability score would occur by chance, Sh. lowest

8.9660E-02

Highlighting Aqua = individual test rejection. Composite scores or ranges in yellow, except: for Shakespeare, gray or red = composite Shakespeare rejection; for oth-ers, gray or red = comp. non-rejection.

SUMMARY OF RESULTS FOR 3,000-WORD POEM BLOCKS

Only 2 of 14 Shakespeare baseline poem blocks (14%) have even one individual rejection in 14 tests. Only 3 of 87 poem blocks by others (3%) have fewer than two individual rejections in 12 tests. Both dis-crete and continuous composite scoring pass 100% of Shakespeare’s blocks and reject 100% of others’ blocks. 3% are close calls, but most composite rejection scores are not close at all. A Lover’s Complaint and Marlowe’s most Shakespeare-distant block have four rejections each and are hundreds of times less likely to have come from Shakespeare by chance than Shakespeare’s least typical baseline block. Fu-neral Elegy by W.S. and the poems of Bacon and the Earl of Oxford have 6-7 rejections each and are tril-lions of times less likely than Shakespeare’s own outlier to be his. Test Sensitivities: G = genre; T = time of composition; E = editing; P = Prosody.

427

Shak

espe

are

Pla

y V

erse

Bas

elin

e D

ata,

Blo

cksi

ze =

3,0

00

Play

and

Blo

ckD

ate

Lat

eG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

with

(2lw

s)no

/(no

+no

t)B

oB5

BoB

7B

oB8

T-E

Slo

pe

Tes

tT

-E R

are

Wor

dsT

-E N

ew

Wor

dsB

ucke

tsB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

ns

Dis

cret

e C

ompo

site

Pr

obab

ility

(2

.6%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Ric

hard

III

1593

16

107

1418

2932

413

333

366

571

-778

0.06

679

1715

01.

000E

+00

2.30

829.

8897

E-0

12

586

1617

4032

012

262

273

548

-698

0.02

815

1015

01.

000E

+00

1.78

179.

9943

E-0

13

540

1222

4334

30

262

483

737

-537

0.01

120

191

151

3.26

4E-0

13.

6044

6.02

94E

-01

45

8016

1538

359

1934

835

076

5-4

670.

0518

83

5515

01.

000E

+00

3.39

467.

1469

E-0

15

734

2219

2734

222

344

411

517

-636

-0.0

214

615

-13

151

3.26

4E-0

13.

0944

8.45

57E

-01

67

7917

1447

364

538

249

971

4-4

400.

0071

1129

150

1.00

0E+

003.

6065

6.01

77E

-01

75

120

1717

2737

415

188

188

939

-634

-0.1

135

8-5

415

01.

000E

+00

3.65

015.

7735

E-0

18

611

117

1658

403

1628

929

787

2-6

340.

0111

7-5

-86

151

3.26

4E-0

13.

7753

5.06

44E

-01

95

7815

1365

370

732

137

181

8-5

57-0

.03

488

2515

01.

000E

+00

3.05

088.

6090

E-0

1R

icha

rd I

I15

951

878

1120

4829

26

333

466

852

-432

-0.0

546

-11

-63

150

1.00

0E+

004.

0388

3.61

63E

-01

28

979

1936

304

2240

041

782

4-6

78-0

.04

-24

-5-9

915

13.

264E

-01

4.69

321.

0712

E-0

13

777

1223

4627

45

236

514

671

-265

-0.1

010

60

-215

01.

000E

+00

4.05

843.

5148

E-0

14

778

823

4728

41

242

458

941

-618

0.04

276

-115

13.

264E

-01

4.11

633.

2223

E-0

15

785

928

4231

711

357

466

829

-738

-0.0

284

81

150

1.00

0E+

003.

1786

8.13

18E

-01

65

847

2163

341

1342

334

072

2-6

67-0

.07

899

1915

01.

000E

+00

3.60

826.

0082

E-0

17

573

719

6635

30

250

321

692

-781

-0.0

615

6-1

2915

13.

264E

-01

4.02

803.

6727

E-0

1R

omeo

1596

15

145

614

3442

812

282

197

524

-610

0.04

55-1

0-2

915

13.

264E

-01

3.91

394.

2872

E-0

12

414

27

1939

396

1518

412

840

5-5

960.

0544

4133

151

3.26

4E-0

15.

2633

2.35

08E

-02

35

102

816

8842

412

233

192

667

-779

0.03

109

4-1

0715

13.

264E

-01

5.16

413.

1561

E-0

24

411

97

1459

450

1835

724

180

4-6

590.

0210

4-5

-180

151

3.26

4E-0

16.

3960

3.30

19E

-04

54

6712

1250

459

2219

719

980

0-6

430.

1325

74

2615

13.

264E

-01

5.69

545.

6103

E-0

36

410

511

16bl

ank

blan

k13

325

236

500

-857

-0.0

410

4-1

664

130

1.00

0E+

003.

4575

5.31

39E

-01

75

6310

23bl

ank

blan

k15

300

186

787

-748

0.09

100

83

130

1.00

0E+

003.

2987

6.20

75E

-01

1 H

enry

IV

1597

110

140

1233

blan

kbl

ank

2938

244

189

5-5

11-0

.03

13-1

87

131

2.90

0E-0

15.

0975

1.70

82E

-02

27

126

1226

blan

kbl

ank

1130

228

365

7-5

94-0

.03

53-3

218

130

1.00

0E+

002.

7784

8.61

38E

-01

37

9910

30bl

ank

blan

k6

238

466

892

-671

0.01

268

1313

01.

000E

+00

3.47

475.

2162

E-0

14

750

1128

blan

kbl

ank

2625

533

370

4-7

210.

0413

511

3113

01.

000E

+00

3.39

475.

6701

E-0

1H

amle

t16

011

754

1929

blan

kbl

ank

2225

644

723

8-6

40-0

.03

81-2

673

130

1.00

0E+

003.

2857

6.27

92E

-01

25

6016

27bl

ank

blan

k9

254

443

588

-867

-0.0

337

-44

7913

12.

900E

-01

3.95

262.

7009

E-0

13

740

1625

blan

kbl

ank

1137

046

135

1-7

78-0

.16

81-1

313

130

1.00

0E+

003.

3385

5.98

62E

-01

47

6013

32bl

ank

blan

k22

325

424

208

-659

0.06

6-1

63

130

1.00

0E+

002.

9690

7.86

77E

-01

55

2712

26bl

ank

blan

k9

280

453

396

-705

-0.0

223

-21

1413

12.

900E

-01

2.75

348.

6977

E-0

16

686

1430

blan

kbl

ank

1028

935

074

3-7

23-0

.09

33-3

25

130

1.00

0E+

002.

4494

9.46

17E

-01

74

4614

21bl

ank

blan

k0

385

300

784

-810

0.02

37-3

4913

12.

900E

-01

3.76

713.

6053

E-0

1T

roilu

s16

021

986

1532

3834

711

250

415

586

-586

-0.2

4-1

54-3

9-7

115

44.

955E

-04

6.06

521.

3595

E-0

32

966

1738

4138

129

294

448

489

-692

-0.0

6-2

5-2

87

151

3.26

4E-0

14.

0465

3.57

63E

-01

36

7318

29bl

ank

blan

k27

261

459

512

-634

-0.1

117

-814

130

1.00

0E+

002.

5484

9.26

30E

-01

45

4815

2049

395

2235

335

047

2-6

78-0

.06

45-1

67

150

1.00

0E+

002.

3641

9.85

86E

-01

55

6614

2129

393

1821

618

640

4-7

95-0

.08

66-3

21

150

1.00

0E+

002.

7991

9.30

23E

-01

Mea

sure

1603

16

4020

33bl

ank

blan

k24

189

420

378

-544

0.06

105

-4-4

213

01.

000E

+00

3.24

976.

4758

E-0

12

641

2237

blan

kbl

ank

2330

438

420

9-6

53-0

.04

121

-125

130

1.00

0E+

002.

8293

8.43

28E

-01

35

4017

30bl

ank

blan

k7

333

381

563

-765

0.04

73-1

521

130

1.00

0E+

002.

4022

9.54

11E

-01

46

6820

27bl

ank

blan

k21

378

356

423

-646

-0.0

415

4-5

3713

01.

000E

+00

2.58

189.

1855

E-0

1O

thel

lo16

041

675

1833

blan

kbl

ank

2019

527

041

2-5

840.

0065

-27

-913

01.

000E

+00

2.28

449.

7006

E-0

12

674

1726

blan

kbl

ank

1516

723

742

9-7

69-0

.03

52-1

62

130

1.00

0E+

002.

5625

9.23

09E

-01

35

3717

27bl

ank

blan

k11

234

246

136

-691

0.13

142

-515

130

1.00

0E+

003.

5985

4.51

74E

-01

44

3117

25bl

ank

blan

k18

232

117

51-8

200.

0112

1-1

1-1

713

12.

900E

-01

3.99

612.

5082

E-0

15

331

1718

blan

kbl

ank

2026

021

223

1-7

470.

1015

9-2

313

01.

000E

+00

4.08

482.

1408

E-0

16

225

1517

blan

kbl

ank

1835

015

023

3-7

98-0

.01

215

2014

132

4.36

0E-0

25.

2159

1.16

59E

-02

428

Shak

espe

are

Pla

y V

erse

Bas

elin

e D

ata,

Blo

cksi

ze =

3,0

00

Play

and

Blo

ckD

ate

Lat

eG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

with

(2lw

s)no

/(no

+no

t)B

oB5

BoB

7B

oB8

T-E

Slo

pe

Tes

tT

-E R

are

Wor

dsT

-E N

ew

Wor

dsB

ucke

tsB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

ns

Dis

cret

e C

ompo

site

Pr

obab

ility

(2

.6%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Lea

r16

051

673

2133

3030

918

313

224

538

-474

-0.1

227

1-2

815

01.

000E

+00

2.04

509.

9708

E-0

12

580

1733

blan

kbl

ank

734

029

723

5-5

11-0

.01

27-1

69

130

1.00

0E+

002.

0404

9.89

33E

-01

34

159

1930

blan

kbl

ank

3232

830

726

8-5

26-0

.06

86-2

97

131

2.90

0E-0

13.

9762

2.59

53E

-01

44

100

1930

blan

kbl

ank

925

032

048

4-5

86-0

.06

46-1

846

130

1.00

0E+

002.

1995

9.78

59E

-01

54

100

1632

2932

310

245

305

500

-539

0.07

90-5

115

01.

000E

+00

1.50

539.

9993

E-0

1M

acbe

th16

061

679

2239

blan

kbl

ank

431

464

572

2-3

530.

0210

-28

1213

12.

900E

-01

4.40

631.

1077

E-0

12

491

1933

blan

kbl

ank

3025

045

418

4-3

71-0

.12

41-8

4913

01.

000E

+00

3.81

903.

3399

E-0

13

487

1334

blan

kbl

ank

1320

645

033

3-3

40-0

.03

21-1

722

130

1.00

0E+

003.

0901

7.30

35E

-01

45

117

2037

blan

kbl

ank

1129

135

150

0-6

31-0

.07

27-9

2213

01.

000E

+00

1.94

949.

9314

E-0

1A

nton

y16

071

547

1940

6339

336

255

474

231

-325

-0.0

633

-20

-615

01.

000E

+00

4.45

381.

7830

E-0

12

613

022

4149

299

2328

331

426

3-3

16-0

.10

41-2

314

150

1.00

0E+

003.

7126

5.42

02E

-01

34

6715

3277

410

326

526

228

0-5

24-0

.04

101

-14

3215

13.

264E

-01

3.33

067.

4598

E-0

14

573

1945

8226

911

208

467

371

-487

-0.0

455

-38

815

13.

264E

-01

4.05

433.

5360

E-0

15

460

1542

blan

kbl

ank

2032

337

037

8-7

69-0

.06

82-3

011

130

1.00

0E+

002.

6912

8.89

22E

-01

64

4716

43bl

ank

blan

k24

385

173

474

-580

0.06

106

-46

130

1.00

0E+

003.

2432

6.51

10E

-01

74

3612

45bl

ank

blan

k13

375

250

693

-600

-0.0

920

-919

130

1.00

0E+

003.

0254

7.61

29E

-01

Cym

belin

e16

101

563

2047

4828

614

289

336

720

-478

-0.0

882

-2-3

215

01.

000E

+00

2.67

249.

5360

E-0

12

554

2153

3731

914

346

271

309

-495

0.00

36-1

20

150

1.00

0E+

002.

6430

9.58

05E

-01

35

102

2351

4027

814

347

300

440

-525

-0.0

618

-13

115

01.

000E

+00

2.95

138.

9218

E-0

14

490

1742

4526

916

250

206

155

-604

0.01

50-7

1715

01.

000E

+00

2.95

558.

9096

E-0

15

461

2543

3831

122

357

381

289

-506

-0.1

059

-18

1715

01.

000E

+00

3.17

118.

1621

E-0

16

541

1943

5132

47

357

296

217

-236

-0.1

021

-31

151

3.26

4E-0

13.

7764

5.05

82E

-01

75

6217

4638

301

816

725

270

0-4

290.

0361

23-2

315

13.

264E

-01

3.66

285.

7019

E-0

1T

empe

st16

111

611

422

4989

358

1332

712

60

-295

-0.1

3-1

0-1

1-3

215

13.

264E

-01

5.21

572.

7121

E-0

22

467

2243

8336

510

583

166

409

-525

-0.0

673

-419

150

1.00

0E+

005.

4590

1.26

61E

-02

36

199

2249

5032

24

351

183

412

-488

-0.2

6-2

4-4

016

153

6.32

5E-0

36.

2279

6.89

58E

-04

46

8423

4682

332

427

318

962

5-5

47-0

.07

37-4

1815

01.

000E

+00

3.54

706.

3461

E-0

1W

inte

r's16

111

515

023

42bl

ank

blan

k7

313

347

348

-634

-0.2

327

-33

-40

132

4.36

0E-0

24.

5660

7.59

83E

-02

26

9421

50bl

ank

blan

k21

288

262

170

-267

0.05

92-2

613

01.

000E

+00

3.73

853.

7551

E-0

13

644

2348

blan

kbl

ank

1425

028

223

5-4

48-0

.07

90-9

-14

130

1.00

0E+

003.

0447

7.52

24E

-01

46

8723

53bl

ank

blan

k4

286

233

514

-556

-0.1

511

2-2

3013

01.

000E

+00

3.54

204.

8348

E-0

15

579

1750

blan

kbl

ank

1525

031

960

0-6

42-0

.01

161

1-2

613

01.

000E

+00

2.68

728.

9040

E-0

16

556

1945

blan

kbl

ank

3027

527

245

5-4

25-0

.10

984

3213

01.

000E

+00

3.19

486.

7695

E-0

1

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ectio

ns4

42

00

05

01

21

32

74

354

13 P

erce

ntag

e5%

5%2%

0%0%

0%6%

0%1%

2%1%

4%2%

9%5%

3%5%

16%

1,15

082

82

Poe

ms

Bas

elin

e: S

tand

ard

Stat

isti

cal P

rofi

leC

ompo

site

Thr

esho

lds

Blo

cks

Tes

ted

8282

8282

4242

8282

8282

8282

8282

822.

900E

-01

1.78

30E

-01

Mea

n5.

4478

.36

16.0

930

.82

49.5

234

5.34

14.5

629

4.39

325.

7950

6.52

-592

.26

-0.0

368

.36

-8.2

02.

47 S

td D

ev1.

3633

.28

4.63

11.4

217

.11

49.3

08.

0067

.11

109.

2122

9.08

146.

660.

0757

.97

15.5

738

.77

Sh D

iscr

ete

Rej

ectio

n Pr

ofile

(Se

e no

te in

key

) M

inim

um2.

0024

.59

6.00

12.0

026

.53

268.

710.

0016

7.00

117.

000.

00-8

67.0

0-0

.26

-154

.16

-44.

00-1

80.4

3 M

inim

um0

Max

imum

10.0

019

8.79

25.0

053

.00

89.3

945

8.75

36.0

058

3.00

645.

0094

1.00

-236

.00

0.13

257.

0041

.00

78.8

4 M

axim

um1

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

le

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

with

(2lw

s)no

/(no

+no

t)B

oB5

BoB

7B

oB8

T-E

Slo

pe

Tes

tT

-E R

are

Wor

dsT

-E N

ew

Wor

dsB

ucke

tsB

lock

Glo

bal M

in3

317

927

265

416

711

613

6-8

67-0

.22

-33

-32

-72

Glo

bal M

ax8

153

2557

8947

636

583

556

944

-265

0.15

218

2179

Min

to 1

600

331

79

2726

54

167

116

136

-867

-0.2

236

-33

-32

-72

Max

to 1

600

815

317

3389

.385

4749

476

3658

655

694

4-2

650.

1521

821

79 M

in f

rom

160

03

3112

1527

265

416

711

613

6-8

67-0

.223

6-3

3-3

2-7

2 M

ax f

rom

160

08

153

2557

89.3

8547

4947

636

586

556

944

-265

0.15

218

2179

Blo

cks

Tes

ted

429

Oth

ers

Pla

y V

erse

ver

sus

Shak

espe

are

Bas

elin

e, B

lock

size

= 3

,000

Play

and

Blo

ckD

ate

Lat

eG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

wit

h

(2lw

s)no

/(no

+no

t)B

oB5

BoB

7B

oB8

T-E

Slo

pe

Tes

tT

-E R

are

Wor

dsT

-E N

ew

Wor

dsB

ucke

tsB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

ns

Dis

cret

e C

ompo

site

Pr

obab

ility

(2

.6%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Tam

215

881

1135

619

blan

kbl

ank

2416

736

140

0-5

080

-63

-75

-50

134

2.70

6E-0

47.

5988

1.32

64E

-07

212

189

24bl

ank

blan

k7

333

473

667

-773

0-9

5-9

5-9

713

51.

284E

-05

9.04

235.

1346

E-1

23

923

720

blan

kbl

ank

028

048

785

2-6

550

-81

-68

1213

51.

284E

-05

6.69

372.

2549

E-0

54

1012

919

blan

kbl

ank

621

946

510

00-6

410

-71

-27

-58

134

2.70

6E-0

46.

2701

1.78

11E

-04

58

127

215

164

1032

444

487

1-4

940

-55

-46

1015

52.

868E

-05

7.14

298.

1909

E-0

6Ja

m4

1591

17

143

1817

167

931

627

660

0-7

330

154

-19

-97

155

2.86

8E-0

56.

5000

2.05

68E

-04

25

76

2022

150

1928

635

875

0-6

420

45-9

-915

44.

955E

-04

5.57

438.

5898

E-0

33

40

314

312

37

250

314

700

-506

079

-20

3415

44.

955E

-04

7.02

751.

5165

E-0

54

514

418

1712

28

161

338

892

-797

048

-10

-16

155

2.86

8E-0

56.

8027

4.80

42E

-05

55

136

1019

122

1131

633

390

5-6

320

76-2

1315

44.

955E

-04

6.29

335.

1998

E-0

4C

leo

1593

110

274

36bl

ank

blan

k9

400

291

667

-321

081

-19

-13

134

2.70

6E-0

45.

4175

5.83

83E

-03

211

2818

20bl

ank

blan

k20

388

164

250

-584

013

-13

-111

134

2.70

6E-0

45.

9806

6.44

67E

-04

310

712

22bl

ank

blan

k9

188

261

360

-333

011

07

-177

133

4.13

3E-0

36.

8848

8.26

78E

-06

410

3914

2816

206

034

212

481

8-6

040

105

-12

-53

154

4.95

5E-0

46.

1581

9.26

49E

-04

Dbe

t15

941

1214

09

1616

238

2833

357

077

8-8

440

-121

-64

-220

157

4.30

0E-0

810

.618

94.

8030

E-1

72

1071

1017

1615

715

250

623

920

-815

011

0-7

9-3

115

52.

868E

-05

8.27

618.

2737

E-0

93

1234

1115

1615

141

381

666

1000

-889

038

-49

-127

159

2.35

8E-1

19.

9014

3.10

96E

-14

412

2810

21bl

ank

blan

k21

200

607

1000

-600

034

-41

-713

51.

284E

-05

6.85

399.

7529

E-0

65

1133

1015

blan

kbl

ank

1219

055

392

0-6

800

-27

-23

-272

132

4.36

0E-0

29.

3931

3.04

54E

-13

JKJC

1594

16

3012

921

237

1627

042

492

3-7

630

68-2

510

153

6.32

5E-0

34.

6941

1.06

90E

-01

25

020

845

288

1046

232

970

7-7

110

124

-932

153

6.32

5E-0

34.

6593

1.15

66E

-01

36

017

715

255

031

426

747

4-6

720

150

-10

1515

52.

868E

-05

5.26

072.

3695

E-0

24

628

194

2127

00

340

351

742

-868

013

80

2915

61.

263E

-06

5.02

524.

6652

E-0

2Sa

ja16

031

632

1230

2021

110

358

505

438

-326

025

-61

4515

36.

325E

-03

5.85

613.

0961

E-0

32

639

1236

1021

515

289

497

800

-465

0-3

-19

1315

25.

672E

-02

4.70

451.

0438

E-0

13

738

1033

1726

718

362

482

619

-447

0-2

3-4

6-2

815

36.

325E

-03

5.01

784.

7599

E-0

24

625

1728

2417

710

261

458

1000

-523

010

-62

3115

52.

868E

-05

6.17

768.

5365

E-0

45

70

1237

522

411

273

578

1000

-433

017

-38

-115

61.

263E

-06

6.06

781.

3452

E-0

36

532

1228

1327

512

345

509

652

-480

0-3

9-3

119

152

5.67

2E-0

24.

6170

1.27

01E

-01

75

1911

2615

265

1433

345

760

0-4

240

34-4

937

154

4.95

5E-0

44.

7723

8.90

39E

-02

85

3713

2717

260

1836

751

891

7-5

310

-4-4

8-8

153

6.32

5E-0

35.

2798

2.23

56E

-02

Wpr

z16

041

526

3324

105

285

1034

714

113

5-8

130

-26

-29

2715

44.

955E

-04

6.75

426.

1112

E-0

52

674

3331

8828

815

429

101

-167

-842

013

6-1

221

153

6.32

5E-0

36.

4410

2.69

49E

-04

36

3336

24bl

ank

blan

k4

591

-101

-314

-735

012

7-3

040

134

2.70

6E-0

48.

6763

8.44

88E

-11

45

2736

31bl

ank

blan

k17

381

-100

-136

-903

037

-49

3113

64.

528E

-07

7.67

388.

2996

E-0

85

853

3421

blan

kbl

ank

410

3-2

713

3-7

010

157

-28

6213

42.

706E

-04

7.04

903.

3717

E-0

66

626

3229

8221

313

321

-92

-24

-719

010

6-3

245

155

2.86

8E-0

57.

0756

1.17

54E

-05

76

934

17bl

ank

blan

k13

208

61-2

41-8

670

177

-21

1813

42.

706E

-04

6.86

049.

4207

E-0

6

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ectio

ns13

2419

422

185

311

123

07

157

163

3838

Per

cent

age

34%

63%

50%

11%

88%

72%

13%

8%29

%32

%8%

0%18

%39

%18

%30

%10

0%10

0% B

lock

s T

este

d38

3838

3825

2538

3838

3838

3838

3838

544

3838

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

wit

h

(2lw

s)no

/(no

+no

t)B

oB5

BoB

7B

oB8

T-E

Slo

pe

Tes

tT

-E R

are

Wor

dsT

-E N

ew

Wor

dsB

ucke

tsB

lock

Dis

cret

eC

ontin

uous

Glo

bal M

in3

317

927

265

416

711

613

6-8

67-0

.22

-33

-32

-72

2.90

0E-0

11.

7830

E-0

1 G

loba

l Max

815

325

5789

476

3658

355

694

4-2

650.

1521

821

79 M

in to

160

03

317

927

265

416

711

613

6-8

67-0

.223

6-3

3-3

2-7

2 M

ax to

160

08

153

1733

89.3

8547

4947

636

586

556

944

-265

0.15

218

2179

Sh D

iscr

ete

Rej

ectio

n Pr

ofile

(Se

e no

te in

key

) M

in f

rom

160

03

3112

1527

265

416

711

613

6-8

67-0

.223

6-3

3-3

2-7

2 M

inim

um0

Max

fro

m 1

600

815

325

5789

.385

4749

476

3658

655

694

4-2

650.

1521

821

79 M

axim

um1

430

APPENDIX SEVEN, PLAY VERSE: KEY TO TABLES FOR 1,500-WORD PLAY VERSE TESTS







3 to 29




Bob5 Bundles of badges 5. See text for components. 93 to 761

BoB7 Bundle of badges 7. See text for components. 0 to 1000

BoB8 Bundle of badges 8. See text for components. -889 to -209

T-E Slope Test Thisted-Efron Slope Test. -0.22 to 0.15


T-E Rare Word Test Thisted-Efron Rare Word Test. -40 to 116

Bucket Block: G Modal Bucket Score per Block. -77 to 100


Discrete Comp. Probability

See Table 2. Probability that observed rejections would occur by chance at Sh.’s avg. rejection rate.

2.710E-01

Continuous Comp. Probability

See Table 2. Prob. that observed comp. probability score would occur by chance, Sh. lowest.

1.2386E-01

Highlighting Aqua = individual test rejection. Composite scores or ranges in yellow, except: for Shakespeare, gray or red = composite rejection; for others, gray or red = composite non-rejection.

SUMMARY OF RESULTS FOR 1500-WORD PLAY VERSE BLOCKS

Only 5 of 140 Shakespeare baseline poem blocks (4%) have more than one individual rejection in 11-13 tests. Thirty-eight of forty-three non-Shakespeare blocks (88%) have more than one rejection. As sam-ple size gets smaller, more overlap and close calls appear, including 4% false negatives and 12% false positives for discrete rejections. Stated differently, of 183 blocks tested, the computer correctly discrimi-nated between Shakespeare and non-Shakespeare (95%) of the time (173 blocks). But roughly half of the rejections for this block size could be considered close calls. Test Sensitivities: G = genre; T = time of composition; E = editing; P = prosody.

431

Shak

espe

are

Poe

ms

Bas

elin

e D

ata,

Blo

cksi

ze =

1,5

00

Poem

and

Blo

ckG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

liti

cs/1

000

line

sP

rocl

itic

s/1

000

line

sB

oB5

BoB

7B

oB8

T-E

Slo

pe

Tes

tT

-E N

ew

Wor

dsM

odal

Blo

ckN

umbe

r of

T

ests

Dis

cret

e R

ejec

tion

s

Dis

cret

e C

ompo

site

P

roba

bili

ty

(2.6

%)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Pro

babi

lity

Ven

us1

1019

511

822

199

255

857

-478

-0.1

9-4

7912

12.

710E

-01

4.21

241.

2368

E-0

12

1010

017

763

333

174

778

-678

-0.1

2-2

818

812

12.

710E

-01

4.19

501.

2845

E-0

13

1011

523

722

349

306

583

-530

-0.1

0-7

135

120

1.00

0E+

003.

7221

3.10

13E

-01

410

168

1810

5229

234

657

1-5

43-0

.15

-19

8212

01.

000E

+00

3.64

673.

4773

E-0

15

1215

115

1046

301

149

652

-333

-0.1

9-1

017

412

01.

000E

+00

2.97

257.

1689

E-0

16

1376

1013

4127

532

744

0-5

00-0

.13

357

512

12.

710E

-01

3.61

783.

6264

E-0

1L

ucre

ce1

1512

07

2231

276

595

846

-508

-0.2

2-1

720

812

01.

000E

+00

3.97

622.

0009

E-0

12

1013

29

1237

298

525

1000

-286

-0.1

1-3

194

120

1.00

0E+

003.

1703

6.11

50E

-01

312

145

1713

3625

027

780

0-4

67-0

.24

-380

121

2.71

0E-0

13.

1470

6.24

42E

-01

410

160

1219

4828

226

010

00-7

22-0

.16

-532

212

01.

000E

+00

3.07

376.

6430

E-0

15

1119

913

1026

235

671

777

-625

-0.1

9-1

919

812

01.

000E

+00

4.53

295.

7418

E-0

26

1210

613

1237

372

412

647

-476

-0.2

1-1

264

120

1.00

0E+

002.

1625

9.67

92E

-01

710

125

1319

2632

712

173

3-6

00-0

.05

-424

312

01.

000E

+00

2.51

278.

9945

E-0

18

1512

811

1782

372

271

666

-619

-0.1

03

560

121

2.71

0E-0

13.

6277

3.57

50E

-01

913

8113

2458

381

226

750

-800

-0.0

6-7

311

120

1.00

0E+

003.

3112

5.32

00E

-01

1010

148

518

6530

529

833

3-5

00-0

.08

116

012

01.

000E

+00

3.38

204.

9181

E-0

1So

nnet

s1

1314

28

1892

372

500

777

-321

-0.0

1-2

310

120

1.00

0E+

003.

0759

6.63

12E

-01

216

5314

2082

396

371

846

-277

0.00

132

512

01.

000E

+00

4.15

481.

3999

E-0

13

1475

717

9232

114

180

0-4

550.

029

319

120

1.00

0E+

003.

3666

5.00

53E

-01

413

628

1456

342

403

867

-689

-0.0

8-2

238

120

1.00

0E+

002.

1432

9.70

20E

-01

513

102

317

3130

647

089

5-5

810.

000

373

120

1.00

0E+

002.

8286

7.85

05E

-01

613

872

1846

286

486

1000

-455

0.01

940

712

12.

710E

-01

3.65

823.

4187

E-0

17

1110

112

1356

286

284

579

-572

-0.0

85

207

120

1.00

0E+

001.

8892

9.90

02E

-01

812

123

1736

347

293

647

-455

-0.1

41

234

121

2.71

0E-0

13.

0796

6.61

13E

-01

913

6316

2331

372

385

556

-369

-0.0

8-5

152

120

1.00

0E+

003.

2472

5.68

32E

-01

1012

494

1512

230

618

010

00-7

690.

072

329

120

1.00

0E+

004.

7490

3.17

70E

-02

1112

104

717

8748

011

682

6-4

120.

06-2

202

120

1.00

0E+

004.

1331

1.46

52E

-01

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ecti

ons

01

10

01

00

01

12

70

2 P

erce

ntag

e0%

4%4%

0%0%

4%0%

0%0%

4%4%

7%2%

0%7%

324

2727

Poe

ms

Bas

elin

e: S

tand

ard

Stat

isti

cal P

rofi

leC

ompo

site

Thr

esho

lds

Blo

cks

Tes

ted

2727

2727

2727

2727

2727

2727

2.71

0E-0

11.

2368

E-0

1 M

ean

12.0

411

1.18

10.7

515

.19

52.5

632

0.75

327.

4874

9.11

-519

.26

-0.0

9-3

.87

254.

45 S

td D

ev1.

7144

.18

5.06

4.64

25.3

256

.13

142.

9516

8.06

137.

700.

088.

4112

3.24

Sh D

iscr

ete

Rej

ecti

on P

rofi

le (

See

note

in k

ey)

Min

imum

10.0

012

.44

2.04

7.00

21.5

119

8.92

116.

0033

3.00

-800

.00

-0.2

4-2

8.44

78.6

0 M

inim

um0

Max

imum

16.0

019

9.47

22.5

824

.00

122.

4547

9.59

671.

0010

00.0

0-2

77.0

00.

078.

7757

4.81

Max

imum

1

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

le

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

liti

cs/1

000

line

sP

rocl

itic

s/1

000

line

sB

oB5

BoB

7B

oB8

T-E

Slo

pe

Tes

tT

-E N

ew

Wor

dsM

odal

Blo

ck

Glo

bal M

in10

243

718

235

930

-889

-0.2

2-2

479

Glo

bal M

ax16

243

2924

123

561

671

1000

-209

0.15

1240

7 M

in to

160

010

243

718

235

9318

8-8

89-0

.22

-24

79 M

ax to

160

016

243

2924

123

561

761

1000

-209

0.15

1240

7 M

in f

rom

160

010

243

1318

235

930

-889

-0.2

2-2

479

Max

fro

m 1

600

1624

329

2312

356

176

110

00-2

090.

1512

407

Blo

cks

Tes

ted

432

Oth

er P

oets

ver

sus

Shak

espe

are

Bas

elin

e, B

lock

size

= 1

,500

Poe

m a

nd B

lock

Gra

de L

evel

HC

W

/2

0K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

liti

cs/1

000

line

sP

rocl

itic

s/1

000

line

sB

oB5

BoB

7B

oB8

T-E

Slo

pe

Tes

tT

-E N

ew

Wor

dsM

odal

Blo

ckN

umbe

r of

T

ests

Dis

cret

e R

ejec

tion

s

Dis

cret

e C

ompo

site

P

roba

bili

ty

(2.6

%)

Con

tinu

ous

Com

posi

te

Err

or

Con

tinu

ous

Com

posi

te

Pro

babi

lity

Bac

15k-

112

06

710

230

628

1000

-409

-0.0

13

1039

124

1.91

3E-0

48.

0744

2.50

81E

-09

Gre

15k-

19

935

130

118

315

625

-533

-0.1

8-1

723

712

33.

242E

-03

5.11

191.

0283

E-0

2G

re15

k-2

1696

29

014

227

890

9-7

55-0

.18

-19

314

123

3.24

2E-0

35.

7573

9.18

34E

-04

Dde

l15k

-112

9414

2616

109

960

0-4

00-0

.13

-822

112

41.

913E

-04

5.41

593.

5163

E-0

3D

del1

5k-2

1353

1528

530

8-7

610

00-6

86-0

.09

-259

812

41.

913E

-04

5.78

518.

1803

E-0

4B

arne

s15k

-16

5521

1719

211

295

0-1

36-0

.16

-14

410

125

8.07

5E-0

67.

4430

1.53

54E

-07

Bar

nes1

5k-2

712

216

19bl

ank

blan

k-1

3310

00-1

82-0

.24

-30

266

105

2.68

4E-0

66.

4594

8.39

63E

-06

Mhe

15k-

19

389

2259

307

308

1000

-253

-0.2

3-1

532

012

23.

752E

-02

4.35

728.

8886

E-0

2M

he15

k-2

834

1127

4328

627

910

00-2

40-0

.14

-132

512

23.

752E

-02

4.77

912.

9117

E-0

2O

xI5s

t8

310

713

115

257

1000

111

-0.0

1-2

1209

126

2.49

4E-0

710

.808

1<

1.00

00E

-15

Idea

15k-

17

8110

933

207

285

733

-321

-0.1

0-9

324

122

3.75

2E-0

24.

3057

1.00

28E

-01

Idea

15k-

27

2711

7bl

ank

blan

k45

483

3-5

690.

05-1

446

010

22.

648E

-02

4.85

918.

7031

E-0

3N

ash1

5k-1

769

514

3635

034

284

6-6

84-0

.23

-34

359

123

3.24

2E-0

35.

4343

3.28

30E

-03

Spam

15k-

112

107

1014

1035

629

110

00-1

84-0

.09

-18

241

122

3.75

2E-0

23.

7628

2.90

69E

-01

Spam

15k-

214

279

610

168

6710

00-4

18-0

.11

-17

282

124

1.91

3E-0

45.

2668

6.03

90E

-03

Fid1

5k-1

562

178

2527

829

410

00-5

570.

00-8

868

122

3.75

2E-0

27.

2442

5.09

94E

-07

Fid1

5k-2

362

135

blan

kbl

ank

4593

5-7

55-0

.01

-17

816

104

8.46

2E-0

58.

1446

2.25

00E

-10

Che

15k-

120

6713

2845

392

346

1000

-364

-0.2

4-2

315

412

33.

242E

-03

6.69

051.

1315

E-0

5C

he15

k-2

1611

716

3211

744

7-8

364

7-8

26-0

.10

-22

434

123

3.24

2E-0

37.

1604

8.34

22E

-07

Mid

g15k

-111

411

138

1017

736

525

0-6

32-0

.23

-59

344

125

8.07

5E-0

610

.708

3<

1.00

00E

-15

Mid

g15k

-214

134

811

1827

930

687

5-4

86-0

.25

-39

281

122

3.75

2E-0

25.

2063

7.46

13E

-03

Che

ttle

1313

70

6bl

ank

blan

k32

162

5-6

75-0

.26

-31

205

104

8.46

2E-0

55.

0281

4.83

62E

-03

Ffam

15k-

120

192

2517

1520

064

410

00-2

00-0

.30

-70

126

126

2.49

4E-0

710

.979

0<

1.00

00E

-15

Ffam

15k-

219

309

1117

blan

kbl

ank

665

1000

680

-0.2

5-4

211

710

52.

684E

-06

12.0

711

<1.

0000

E-1

5D

on15

k-1

1120

68

2345

290

319

765

26-0

.17

-70

294

122

3.75

2E-0

29.

3608

1.42

12E

-13

Don

15k-

29

688

20bl

ank

blan

k12

610

00-2

97-0

.04

-25

491

103

1.83

9E-0

34.

7441

1.27

22E

-02

Bur

ton

Ana

t8

3121

7bl

ank

blan

k20

010

00-7

50bl

ank

blan

k42

28

38.

922E

-04

4.87

472.

5114

E-0

3

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ecti

ons

163

39

1010

61

79

910

9327

26 P

erce

ntag

e59

%11

%11

%33

%50

%50

%22

%4%

26%

35%

35%

37%

30%

100%

96%

Blo

cks

Tes

ted

2727

2727

2020

2727

2726

2627

308

2727

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

liti

cs/1

000

line

sP

rocl

itic

s/1

000

line

sB

oB5

BoB

7B

oB8

T-E

Slo

pe

Tes

tT

-E N

ew

Wor

dsM

odal

Blo

ckD

iscr

ete

Con

tinu

ous

Glo

bal M

in10

243

718

234.

6938

7893

0-8

89-0

.22

-24

792.

710E

-01

1.23

68E

-01

Glo

bal M

ax16

243

2924

123

561

671

1000

-209

0.15

1240

7 M

in to

160

010

243

718

234.

6938

7893

188

-889

-0.2

2-2

479

Max

to 1

600

1624

329

2412

356

176

110

00-2

090.

1512

407

Sh

Dis

cret

e R

ejec

tion

Pro

file

(S

ee n

ote

in k

ey)

Min

fro

m 1

600

1024

313

1823

4.69

3878

930

-889

-0.2

2-2

479

Min

imum

0 M

ax f

rom

160

016

243

2923

123

561

761

1000

-209

0.15

1240

7 M

axim

um1

433

APPENDIX SEVEN, POEMS: KEY TO TABLES FOR 1,500-WORD POEM TESTS







3 to 29






BoB8 Bundle of badges 8. See text for components -889 to -209







2.710E-01


See Table 2. Prob. that observed comp. probability score would occur by chance, Sh.’s lowest block.

1.2386E-01


SUMMARY OF RESULTS FOR 1500-WORD POEM BLOCKS

Only 7 of 27 Shakespeare baseline poem blocks (26%) have even one individual rejection in 12 tests. None of 27 blocks of poems by others have fewer than two individual rejections in 12 tests. Both dis-crete and continuous composite scoring pass 100% of Shakespeare’s blocks and reject 96-100% of oth-ers’ blocks. Perhaps a quarter could be considered close calls, but most, including Oxford, Bacon and Marlowe, could not. Test Sensitivities: G = genre; T = time of composition; E = editing; P = prosody.

434

Shak

espe

are

Pla

y V

erse

Bas

elin

e D

ata,

Blo

cksi

ze =

1,5

00

Play

and

Blo

ckD

ate

Lat

eW

ords

Lin

esG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

BoB

5B

oB7

BoB

8T

-E S

lope

T

est

T-E

Rar

e W

ords

T-E

New

W

ords

Buc

kets

Blo

ckN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ility

(2

.6%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Ric

hard

III

1593

114

9419

88

134

927

1031

842

650

0-8

41-0

.17

250

2313

12.

900E

-01

3.33

685.

9957

E-0

12

1496

192

580

1624

4728

127

666

7-7

040.

0642

88

130

1.00

0E+

002.

4638

9.43

56E

-01

315

0919

06

5312

1937

300

243

500

-520

-0.0

648

130

130

1.00

0E+

001.

4088

9.99

78E

-01

415

0818

35

106

2014

4432

231

060

0-9

450.

0233

412

131

2.90

0E-0

13.

1232

7.13

89E

-01

514

9718

76

4013

2437

332

579

571

-647

-0.1

034

93

130

1.00

0E+

002.

6031

9.13

32E

-01

614

8618

45

4011

2049

337

388

833

-439

0.01

8610

5913

01.

000E

+00

2.89

338.

1861

E-0

17

1496

182

510

718

1566

418

315

700

-500

0.01

78-2

5813

01.

000E

+00

2.87

358.

2646

E-0

18

1521

187

426

1416

1133

737

985

7-4

380.

0511

15

142

132

4.36

0E-0

24.

9918

2.36

63E

-02

914

6317

75

2723

1834

407

430

278

-814

0.04

9812

-913

01.

000E

+00

4.26

321.

5099

E-0

110

1517

190

740

2119

2128

439

266

7-4

67-0

.02

473

1313

01.

000E

+00

3.02

267.

6259

E-0

111

1493

193

954

1916

4738

357

369

2-3

85-0

.15

233

5213

01.

000E

+00

4.03

922.

3253

E-0

112

1555

193

690

1412

4734

246

075

0-5

000.

0048

856

130

1.00

0E+

002.

5469

9.26

63E

-01

1315

0819

15

146

2020

1634

627

490

9-4

73-0

.01

21-2

2013

12.

900E

-01

3.28

256.

2968

E-0

114

1497

187

612

014

1437

401

9310

00-7

73-0

.11

1410

-50

130

1.00

0E+

004.

0353

2.34

15E

-01

1514

9518

46

8020

1854

380

100

867

-565

0.01

59-3

-64

130

1.00

0E+

003.

4780

5.19

75E

-01

1615

6619

26

153

1515

6845

849

187

5-7

220.

0159

-39

130

1.00

0E+

003.

5148

4.98

87E

-01

1715

1919

24

7914

1157

370

253

867

-500

0.10

295

-213

01.

000E

+00

3.18

086.

8430

E-0

118

1556

193

577

1613

7336

846

077

7-6

00-0

.03

183

4113

01.

000E

+00

2.52

079.

3232

E-0

1R

icha

rd I

I15

951

1549

197

910

311

1846

315

370

833

-542

-0.0

822

-833

130

1.00

0E+

002.

5110

9.34

34E

-01

215

1119

68

4011

2251

270

574

867

-273

-0.0

524

-4-1

0313

12.

900E

-01

4.32

161.

3349

E-0

13

1497

189

812

07

2148

222

342

1000

-576

-0.0

80

-4-5

013

12.

900E

-01

3.19

646.

7610

E-0

14

1596

201

975

1114

2538

349

372

7-7

41-0

.04

-24

-1-4

513

01.

000E

+00

3.74

163.

7387

E-0

15

1473

186

795

922

2728

557

765

5-4

48-0

.06

35-1

-313

01.

000E

+00

2.31

699.

6617

E-0

16

1633

204

661

1523

6426

546

568

2-1

28-0

.10

711

5213

12.

900E

-01

3.65

534.

2031

E-0

17

1548

193

990

727

4124

436

980

0-7

570.

0629

217

130

1.00

0E+

002.

8020

8.53

16E

-01

815

3119

06

788

2653

326

558

1000

-486

0.04

-14

-58

130

1.00

0E+

003.

1256

7.12

68E

-01

915

8319

67

639

2746

316

471

867

-769

-0.0

632

315

130

1.00

0E+

001.

7551

9.97

63E

-01

1014

6618

17

8210

2939

320

461

800

-724

-0.0

252

510

130

1.00

0E+

001.

6853

9.98

45E

-01

1114

9318

46

548

2360

315

266

500

-878

-0.0

560

218

130

1.00

0E+

001.

4996

9.99

56E

-01

1215

9619

44

125

618

6736

641

082

3-3

72-0

.07

297

1413

01.

000E

+00

2.19

059.

7937

E-0

113

1367

166

488

613

6633

137

346

7-7

650.

0096

-144

130

1.00

0E+

002.

3124

9.66

73E

-01

1413

6816

56

589

2667

376

263

833

-792

-0.0

660

047

130

1.00

0E+

001.

6793

9.98

51E

-01

Rom

eo15

961

1553

191

411

64

1327

415

368

680

-510

-0.0

15

0-3

130

1.00

0E+

001.

8232

9.96

49E

-01

214

7417

85

122

815

5544

116

294

-758

0.04

50-1

019

131

2.90

0E-0

13.

2392

6.53

26E

-01

315

5519

05

167

522

3856

110

218

8-5

10-0

.07

-27

3091

131

2.90

0E-0

15.

7127

1.93

26E

-03

414

0116

94

114

915

4436

515

857

1-6

890.

0571

1114

913

12.

900E

-01

4.05

532.

2591

E-0

15

1493

181

610

78

1557

472

9858

3-8

05-0

.03

4012

-26

130

1.00

0E+

003.

2282

6.59

18E

-01

614

5317

64

839

1712

337

026

575

0-7

500.

0369

-810

013

01.

000E

+00

4.34

221.

2768

E-0

17

1547

196

411

69

1943

395

310

727

-600

0.01

412

371

131

2.90

0E-0

17.

9407

1.48

40E

-08

812

5416

04

128

69

7751

615

786

2-7

580.

0263

-736

130

1.00

0E+

003.

8799

3.04

02E

-01

915

9018

84

6314

1356

441

236

714

-600

0.00

147

415

131

2.90

0E-0

14.

2890

1.43

07E

-01

1014

1716

84

719

1140

484

166

909

-682

0.13

110

028

130

1.00

0E+

004.

4569

9.86

55E

-02

1113

2616

26

6017

19bl

ank

blan

k32

176

5-7

62-0

.10

59-6

3011

01.

000E

+00

2.17

639.

4326

E-0

112

1328

163

412

06

12bl

ank

blan

k15

635

5-1

000

-0.0

445

-10

1511

12.

516E

-01

2.92

076.

6514

E-0

113

1552

193

564

1020

blan

kbl

ank

139

1000

-733

-0.0

750

816

110

1.00

0E+

002.

6331

8.04

47E

-01

1416

6320

34

6010

25bl

ank

blan

k22

267

7-7

580.

0950

021

110

1.00

0E+

002.

2523

9.27

59E

-01

1 H

enry

IV

1597

115

4320

213

143

1133

blan

kbl

ank

406

867

-375

-0.1

3-2

3-2

14

111

2.51

6E-0

15.

7240

5.74

50E

-04

214

6518

68

150

1331

blan

kbl

ank

475

1000

-652

-0.0

336

3-2

110

1.00

0E+

003.

0241

6.08

49E

-01

316

3320

16

8614

27bl

ank

blan

k36

677

8-6

47-0

.08

50-5

911

01.

000E

+00

1.04

579.

9992

E-0

14

1510

194

615

910

25bl

ank

blan

k20

025

9-5

33-0

.03

3-2

7-1

111

2.51

6E-0

13.

1201

5.54

37E

-01

514

1617

97

999

32bl

ank

blan

k41

087

5-6

75-0

.06

62

2711

01.

000E

+00

1.37

799.

9882

E-0

16

1580

203

789

1128

blan

kbl

ank

510

905

-667

0.01

206

1911

01.

000E

+00

2.18

799.

4105

E-0

17

1577

200

976

1430

blan

kbl

ank

490

467

-744

-0.0

544

8-2

211

01.

000E

+00

2.88

386.

8472

E-0

18

1591

194

525

926

blan

kbl

ank

167

1000

-702

0.04

913

4611

01.

000E

+00

3.17

955.

2059

E-0

1

435

Shak

espe

are

Pla

y V

erse

Bas

elin

e D

ata,

Blo

cksi

ze =

1,5

00

Play

and

Blo

ckD

ate

Lat

eW

ords

Lin

esG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

BoB

5B

oB7

BoB

8T

-E S

lope

T

est

T-E

Rar

e W

ords

T-E

New

W

ords

Buc

kets

Blo

ckN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ility

(2

.6%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Ham

let

1601

115

5219

67

5217

31bl

ank

blan

k56

933

3-5

69-0

.17

13-2

314

110

1.00

0E+

003.

4942

3.48

11E

-01

214

1917

66

5622

27bl

ank

blan

k33

314

3-7

72-0

.03

68-3

150

111

2.51

6E-0

13.

9058

1.71

11E

-01

315

7619

56

6318

26bl

ank

blan

k53

868

4-7

96-0

.16

4-2

20

110

1.00

0E+

003.

4874

3.51

59E

-01

414

4317

85

5515

28bl

ank

blan

k31

846

7-9

41-0

.03

32-2

261

111

2.51

6E-0

12.

8793

6.87

08E

-01

514

4517

57

1417

27bl

ank

blan

k60

153

8-8

42-0

.03

78-8

-911

12.

516E

-01

3.51

833.

3588

E-0

16

1570

192

764

1623

blan

kbl

ank

333

250

-731

-0.1

63

-419

110

1.00

0E+

002.

5790

8.26

58E

-01

715

0018

57

9317

38bl

ank

blan

k46

911

1-7

39-0

.13

13-8

-10

110

1.00

0E+

002.

8304

7.12

30E

-01

815

1618

86

2610

27bl

ank

blan

k37

026

7-5

560.

06-7

-88

110

1.00

0E+

002.

0173

9.67

84E

-01

Tro

ilus

1602

115

1019

17

4016

3334

310

318

667

-778

-0.1

5-6

4-2

4-1

813

12.

900E

-01

4.45

429.

9275

E-0

22

1514

201

1111

913

3142

380

519

500

-451

-0.2

4-9

0-1

5-1

013

34.

133E

-03

6.07

344.

3167

E-0

43

1534

191

926

1535

4252

843

633

3-7

02-0

.06

4-1

0-1

013

01.

000E

+00

3.94

802.

7217

E-0

14

1481

190

912

220

4247

325

460

700

-678

-0.0

6-2

9-1

914

130

1.00

0E+

003.

9626

2.65

59E

-01

515

0918

55

9320

30bl

ank

blan

k43

145

5-8

89-0

.06

-1-9

1811

01.

000E

+00

2.58

428.

2452

E-0

16

1506

186

853

1628

blan

kbl

ank

489

579

-372

-0.1

118

2-2

111

01.

000E

+00

2.80

947.

2288

E-0

17

1427

176

456

1620

7131

840

433

3-7

040.

0314

-718

130

1.00

0E+

002.

3828

9.57

10E

-01

815

1818

65

4013

1938

437

296

586

-657

-0.0

631

-10

1313

01.

000E

+00

1.87

669.

9530

E-0

19

1485

186

554

1624

1943

821

928

0-8

24-0

.02

26-7

3113

01.

000E

+00

2.86

788.

2869

E-0

110

1534

189

478

1218

3735

415

454

5-7

73-0

.08

39-2

42

130

1.00

0E+

002.

9255

8.05

41E

-01

1120

6025

94

117

1212

5844

035

481

2-7

24-0

.04

38-2

91

132

4.36

0E-0

23.

5526

4.77

50E

-01

Mea

sure

1603

114

8018

48

1417

41bl

ank

blan

k37

762

5-6

10-0

.05

241

411

12.

516E

-01

2.66

107.

9249

E-0

12

1497

180

467

2224

blan

kbl

ank

455

241

-474

0.06

81-5

-37

110

1.00

0E+

003.

3906

4.02

68E

-01

314

7017

85

2723

31bl

ank

blan

k34

013

3-5

830.

0652

634

110

1.00

0E+

003.

1553

5.34

36E

-01

414

2217

17

5620

31bl

ank

blan

k43

038

5-7

20-0

.04

70-8

2111

01.

000E

+00

2.29

659.

1718

E-0

15

1474

185

554

1729

blan

kbl

ank

273

379

-758

-0.0

411

-951

110

1.00

0E+

001.

8861

9.81

13E

-01

615

2919

15

2616

31bl

ank

blan

k47

271

4-7

690.

0463

-58

110

1.00

0E+

002.

5974

8.19

23E

-01

714

9318

46

6720

32bl

ank

blan

k33

348

4-6

720.

0279

-754

110

1.00

0E+

002.

3073

9.14

49E

-01

814

3517

75

7020

21bl

ank

blan

k38

033

3-6

09-0

.04

751

1311

01.

000E

+00

2.20

109.

3847

E-0

1O

thel

lo16

041

1661

195

596

1332

blan

kbl

ank

260

517

-460

0.06

34-1

91

110

1.00

0E+

002.

0724

9.60

49E

-01

215

8419

68

6323

33bl

ank

blan

k28

427

3-7

000.

0031

-814

110

1.00

0E+

002.

6207

8.09

66E

-01

316

1719

76

8717

27bl

ank

blan

k26

766

7-7

62-0

.04

30-5

2311

01.

000E

+00

1.42

719.

9836

E-0

14

1598

195

550

1724

blan

kbl

ank

205

200

-774

-0.0

322

-10

211

01.

000E

+00

2.30

399.

1534

E-0

15

1666

203

524

1827

blan

kbl

ank

271

-47

-714

0.05

471

4311

12.

516E

-01

3.17

805.

2144

E-0

16

1613

188

550

1626

blan

kbl

ank

221

304

-659

0.13

95-6

1411

01.

000E

+00

3.33

164.

3496

E-0

17

1645

199

549

1427

blan

kbl

ank

119

0-8

49-0

.07

61-7

911

01.

000E

+00

3.06

235.

8707

E-0

18

1634

196

412

2022

blan

kbl

ank

116

86-7

780.

0160

-4-5

311

12.

516E

-01

3.97

921.

4741

E-0

1L

ear

1605

114

7817

88

6819

3347

296

242

500

-440

-0.1

022

1-2

113

01.

000E

+00

2.66

628.

9647

E-0

12

1532

188

578

2234

1730

720

757

1-5

00-0

.12

40

4613

12.

900E

-01

3.02

517.

6143

E-0

13

1493

181

540

1929

4820

629

677

-440

-0.0

129

-854

131

2.90

0E-0

13.

2570

6.43

62E

-01

415

1618

85

119

1537

blan

kbl

ank

299

333

-591

-0.0

1-2

-813

110

1.00

0E+

001.

7252

9.91

04E

-01

514

7917

75

243

1837

blan

kbl

ank

305

263

-391

-0.0

140

-23

1111

01.

000E

+00

4.77

301.

8972

E-0

26

1537

182

452

2124

blan

kbl

ank

309

273

-647

-0.0

646

-716

110

1.00

0E+

002.

2218

9.34

21E

-01

715

2318

04

9221

26bl

ank

blan

k38

038

5-6

00-0

.02

39-3

4311

01.

000E

+00

1.98

879.

7122

E-0

18

1485

180

494

1735

blan

kbl

ank

264

556

-568

-0.0

67

-15

2111

01.

000E

+00

2.00

609.

6921

E-0

19

1507

177

410

615

3440

302

203

529

-682

-0.0

420

-341

130

1.00

0E+

001.

8748

9.95

34E

-01

1015

1018

35

146

1830

1834

140

347

8-4

040.

0770

-213

130

1.00

0E+

003.

2567

6.43

78E

-01

1115

7318

44

8922

2552

295

300

241

-600

0.03

5614

1713

12.

900E

-01

3.15

566.

9738

E-0

1M

acbe

th16

061

1524

192

652

1736

blan

kbl

ank

641

692

-404

-0.1

5-3

-810

110

1.00

0E+

003.

4517

3.70

12E

-01

214

9918

16

9326

42bl

ank

blan

k64

973

9-3

090.

0214

-20

1111

01.

000E

+00

4.20

198.

9914

E-0

23

1528

185

492

2029

blan

kbl

ank

407

273

-436

-0.1

023

-14

3911

01.

000E

+00

2.54

828.

3851

E-0

14

1532

186

591

1837

blan

kbl

ank

504

111

-320

-0.1

217

631

110

1.00

0E+

003.

2012

5.08

26E

-01

515

1418

25

5312

29bl

ank

blan

k47

629

7-3

64-0

.06

5-9

1311

01.

000E

+00

1.90

689.

7938

E-0

16

1477

177

410

813

39bl

ank

blan

k42

240

0-3

22-0

.03

15-9

2311

01.

000E

+00

2.14

009.

4981

E-0

17

1593

195

688

1839

blan

kbl

ank

369

789

-793

-0.0

925

-124

110

1.00

0E+

002.

1669

9.45

01E

-01

813

1016

14

153

2234

blan

kbl

ank

328

238

-500

-0.0

72

-720

110

1.00

0E+

002.

9933

6.25

60E

-01

436

Oth

ers

Pla

y V

erse

ver

sus

Shak

espe

are

Bas

elin

e, B

lock

size

= 1

,500

Play

and

Blo

ckD

ate

Lat

eW

ords

Lin

esG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

BoB

5B

oB7

BoB

8T

-E S

lope

T

est

T-E

Rar

e W

ords

T-E

New

W

ords

Buc

kets

Blo

ckN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ility

(2

.6%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Tam

215

881

1722

236

1112

519

blan

kbl

ank

390

600

-550

-0.2

1-5

6-5

0-3

114

1.30

2E-0

47.

3751

1.00

10E

-07

217

0822

110

597

18bl

ank

blan

k33

320

0-4

83-0

.15

-6-2

5-2

611

23.

181E

-02

4.76

391.

9511

E-0

23

1687

235

1212

727

blan

kbl

ank

572

692

-727

-0.1

7-4

7-5

9-3

111

41.

302E

-04

8.18

424.

5608

E-1

04

1710

230

1123

1020

blan

kbl

ank

371

636

-818

-0.1

4-4

7-3

6-4

711

41.

302E

-04

6.14

238.

6955

E-0

55

1668

215

836

915

blan

kbl

ank

451

733

-682

-0.2

3-6

3-4

242

113

2.47

9E-0

36.

2636

4.83

61E

-05

617

6523

39

235

25bl

ank

blan

k52

010

0-6

28-0

.10

-17

-26

-811

32.

479E

-03

4.85

831.

4514

E-0

2Ja

m4

1591

114

9818

07

131

1822

167

232

524

-764

-0.0

383

-247

133

4.13

3E-0

35.

0249

2.14

00E

-02

214

5317

67

285

1811

165

316

684

-707

-0.0

372

-17

-34

132

4.36

0E-0

24.

8125

3.97

94E

-02

315

0618

35

138

2627

175

208

857

-600

-0.0

633

-910

132

4.36

0E-0

23.

9231

2.83

60E

-01

414

4818

05

04

136

133

488

692

-703

-0.0

813

027

133

4.13

3E-0

35.

3223

8.14

56E

-03

514

8918

45

02

175

136

341

636

-500

-0.0

240

-15

1713

42.

706E

-04

5.26

309.

9652

E-0

36

1457

177

40

411

010

728

977

8-5

120.

0440

-532

133

4.13

3E-0

35.

7355

1.76

63E

-03

Map

a15

9315

2819

59

137

1515

103

364

500

-824

-0.0

3-1

511

32.

479E

-03

5.53

933.

7494

E-0

3C

leo

1593

115

6219

010

133

3626

268

456

714

-359

-0.2

1-9

9-3

714

135

1.28

4E-0

57.

4269

3.78

96E

-07

213

4317

311

605

36bl

ank

blan

k12

660

0-2

83-0

.19

-21

-27

311

32.

479E

-03

5.88

242.

8809

E-0

43

1446

182

1211

114

26bl

ank

blan

k20

325

0-5

93-0

.07

33-3

0-3

112

3.18

1E-0

25.

0909

6.67

64E

-03

414

4518

49

2821

14bl

ank

blan

k11

825

0-5

71-0

.09

78-4

8-1

3311

23.

181E

-02

6.76

463.

5651

E-0

65

1452

185

80

1221

blan

kbl

ank

207

571

-429

-0.0

940

-35

-128

113

2.47

9E-0

35.

4704

1.62

80E

-03

613

8517

412

1411

23bl

ank

blan

k31

591

-231

-0.1

1-1

-14

-64

113

2.47

9E-0

35.

7356

5.46

74E

-04

Dbe

t15

941

1561

204

1515

49

1925

270

616

750

-949

-0.2

1-9

9-3

7-1

2913

51.

284E

-05

9.78

891.

0790

E-1

42

1304

171

1012

39

13bl

ank

blan

k52

080

0-6

80-0

.19

-21

-27

-78

113

2.47

9E-0

35.

7408

5.34

71E

-04

314

0818

49

4311

1727

120

463

714

-667

-0.0

733

-30

-45

132

4.36

0E-0

25.

4681

4.86

83E

-03

414

2318

111

988

17bl

ank

blan

k77

010

00-9

03-0

.09

78-4

844

114

1.30

2E-0

47.

3910

9.06

87E

-08

514

4519

19

2810

15bl

ank

blan

k76

710

00-9

51-0

.09

40-3

5-1

811

32.

479E

-03

6.18

697.

0225

E-0

56

1559

193

1638

1115

blan

kbl

ank

561

1000

-807

-0.1

1-1

-14

-138

112

3.18

1E-0

28.

2982

2.01

09E

-10

JKJC

1594

115

0619

06

2711

411

237

463

857

-758

-0.0

330

-17

7013

24.

360E

-02

4.51

968.

5068

E-0

22

1540

184

526

1613

1618

533

688

9-7

420.

0339

-8-6

913

24.

360E

-02

4.43

671.

0337

E-0

13

1356

164

50

189

blan

kbl

ank

361

680

-742

0.07

82-8

7311

12.

516E

-01

3.85

241.

8989

E-0

14

1371

168

60

228

blan

kbl

ank

295

750

-689

-0.0

843

-137

111

2.51

6E-0

13.

2794

4.64

06E

-01

513

7616

66

017

7bl

ank

blan

k18

961

9-6

970.

0688

-10

2711

23.

181E

-02

3.82

382.

0049

E-0

16

1370

166

70

167

blan

kbl

ank

344

294

-647

0.12

630

1111

23.

181E

-02

3.87

871.

8048

E-0

1Sa

ja16

031

1535

192

726

1129

blan

kbl

ank

664

435

-350

-0.0

7-9

-27

5111

12.

516E

-01

4.38

465.

7177

E-0

22

1641

204

537

1332

1021

138

042

9-3

06-0

.09

35-3

442

133

4.13

3E-0

34.

6631

5.94

49E

-02

315

1219

56

2613

3310

215

537

765

-576

-0.0

4-1

8-2

1613

24.

360E

-02

3.89

222.

9813

E-0

14

1594

201

650

1138

1024

446

484

6-3

680.

0116

-17

1013

12.

900E

-01

3.80

673.

4020

E-0

15

1595

204

813

1038

blan

kbl

ank

459

700

-212

-0.0

6-2

1-2

6-3

211

23.

181E

-02

4.67

242.

5713

E-0

26

1594

206

663

1028

1022

350

454

6-6

28-0

.11

-1-2

023

132

4.36

0E-0

23.

8334

3.26

78E

-01

Wpr

z16

041

1491

168

427

3320

155

250

111

500

-933

0.03

84-7

1313

34.

133E

-03

7.06

933.

0111

E-0

62

1562

185

638

3327

blan

kbl

ank

170

-143

-733

-0.0

453

-527

112

3.18

1E-0

24.

4761

4.48

57E

-02

314

8617

04

108

2822

blan

kbl

ank

106

-272

-938

0.01

86-2

310

112

3.18

1E-0

25.

4034

2.11

57E

-03

414

9017

27

5438

4199

343

970

-773

-0.0

642

-621

131

2.90

0E-0

15.

7330

1.78

39E

-03

515

0917

67

2740

22bl

ank

blan

k-1

76-3

85-6

97-0

.12

14-2

228

113

2.47

9E-0

37.

3734

1.01

16E

-07

615

3317

75

2631

25bl

ank

blan

k-8

-273

-771

-0.1

024

-27

2111

41.

302E

-04

5.69

446.

5139

E-0

4

Dis

crim

inat

ion

Stat

isti

csA

ggre

gate

sC

ompo

site

Dis

crim

inat

ion

Stat

isti

cs R

ejec

tions

1218

75

1212

46

51

620

511

338

42 P

erce

ntag

e28

%42

%16

%12

%67

%67

%9%

14%

12%

2%14

%47

%12

%22

%88

%98

% B

lock

s T

este

d43

4343

4318

1843

4343

4342

4342

507

4343

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

BoB

5B

oB7

BoB

8T

-E S

lope

T

est

T-E

Rar

e W

ords

T-E

New

W

ords

Buc

kets

Blo

ckD

iscr

ete

Con

tinuo

us

Glo

bal M

in4

243

818

235

930

-889

-0.2

2-4

0-2

4-7

72.

516E

-01

3.48

11E

-01

Glo

bal M

ax9

243

2955

123

561

761

1000

-209

0.15

116

1210

0 M

in to

160

04

243

818

235

9318

8-8

89-0

.223

6-4

0-2

4.37

35-7

7 M

ax to

160

09

243.

4077

0823

3312

3.37

6623

561

761

1000

-209

0.15

116

12.3

612

100

Sh D

iscr

ete

Rej

ectio

n Pr

ofile

(Se

e no

te in

key

) M

in f

rom

160

04

243

1318

235

930

-889

-0.2

236

-40

-24.

3735

-77

Min

imum

0 M

ax f

rom

160

09

243.

4077

0829

5512

3.37

6623

561

761

1000

-209

0.15

116

12.3

612

100

Max

imum

1

437

Shak

espe

are

Pla

y V

erse

Bas

elin

e D

ata,

Blo

cksi

ze =

1,5

00

Play

and

Blo

ckD

ate

Lat

eW

ords

Lin

esG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

BoB

5B

oB7

BoB

8T

-E S

lope

T

est

T-E

Rar

e W

ords

T-E

New

W

ords

Buc

kets

Blo

ckN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ility

(2

.6%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Ant

ony

1607

115

1218

34

4019

3461

473

469

308

-257

-0.0

943

-8-1

313

01.

000E

+00

3.62

784.

3546

E-0

12

1454

179

655

1945

6532

448

115

4-3

78-0

.06

-9-1

218

130

1.00

0E+

003.

2900

6.25

55E

-01

315

0718

67

1321

4650

313

355

48-3

620.

0639

-63

131

2.90

0E-0

13.

8363

3.25

34E

-01

414

1917

45

113

2436

4728

127

352

9-2

41-0

.10

2-1

81

130

1.00

0E+

003.

5964

4.52

92E

-01

514

9217

84

5412

3275

353

176

83-4

71-0

.05

55-6

2213

01.

000E

+00

2.50

359.

3587

E-0

16

1491

180

480

1733

7946

434

246

2-5

86-0

.04

46-8

1613

01.

000E

+00

2.49

759.

3708

E-0

17

1509

193

566

1842

5520

557

429

4-4

15-0

.22

17-3

72

132

4.36

0E-0

25.

3792

6.68

46E

-03

814

9418

45

8020

4810

833

137

844

4-5

68-0

.04

38-2

2013

01.

000E

+00

3.47

025.

2418

E-0

1C

ymbe

line

1610

115

4018

44

5221

4460

255

390

680

-630

0.13

52-1

-19

130

1.00

0E+

003.

8330

3.26

98E

-01

216

0919

45

7518

5036

314

281

760

-263

-0.0

830

-211

130

1.00

0E+

003.

1406

7.05

07E

-01

317

0220

96

4720

5014

325

355

142

-564

-0.0

31

-341

131

2.90

0E-0

13.

4270

5.48

71E

-01

416

6019

65

7221

5561

311

190

412

-417

0.00

35-9

-24

130

1.00

0E+

003.

3716

5.80

04E

-01

516

2119

76

111

2452

4127

920

433

3-4

10-0

.12

28-5

2113

01.

000E

+00

3.51

914.

9643

E-0

16

1713

204

693

2229

3927

538

951

7-7

00-0

.06

-10

-8-5

130

1.00

0E+

002.

5788

9.19

27E

-01

716

6720

34

108

1847

3924

119

412

8-6

740.

0720

-14

3613

01.

000E

+00

3.73

733.

7614

E-0

18

1666

196

360

1637

5129

621

818

8-5

240.

0130

715

131

2.90

0E-0

12.

8854

8.21

76E

-01

Tem

pest

1611

115

0018

37

4026

5498

333

188

0-3

22-0

.11

62

1413

01.

000E

+00

4.94

602.

7129

E-0

22

1485

177

520

217

4479

384

640

-265

-0.1

3-1

5-1

34

131

2.90

0E-0

15.

3276

7.99

84E

-03

315

0017

64

5323

4085

318

147

500

-464

-0.0

144

844

130

1.00

0E+

003.

5551

4.76

09E

-01

415

1117

75

7920

4573

390

185

333

-567

-0.0

629

-13

713

01.

000E

+00

2.75

438.

6948

E-0

15

1505

179

513

322

5145

307

232

200

-415

-0.0

227

-417

130

1.00

0E+

003.

3199

6.09

00E

-01

612

9616

57

231

2245

5534

511

871

4-5

61-0

.26

-51

-36

1913

34.

133E

-03

6.88

758.

1489

E-0

67

1442

175

612

522

4657

337

202

429

-546

-0.0

83

-620

130

1.00

0E+

002.

8227

8.45

70E

-01

814

1017

05

4323

4610

633

517

768

0-5

48-0

.07

342

4113

01.

000E

+00

3.91

012.

8967

E-0

1W

inte

r's16

111

1422

167

411

322

41bl

ank

blan

k44

538

5-8

79-0

.11

26-1

40

110

1.00

0E+

003.

2916

4.57

22E

-01

213

6516

96

176

2444

blan

kbl

ank

241

333

-469

-0.2

31

-19

-39

111

2.51

6E-0

14.

8022

1.73

28E

-02

313

9416

75

7222

55bl

ank

blan

k29

044

0-2

09-0

.06

51-4

1511

01.

000E

+00

3.59

222.

9965

E-0

14

1369

166

688

2045

blan

kbl

ank

234

-72

-319

0.05

412

411

12.

516E

-01

3.64

402.

7550

E-0

15

1336

157

545

2641

blan

kbl

ank

263

364

-580

-0.0

548

-81

110

1.00

0E+

002.

6214

8.09

37E

-01

613

9217

05

4319

54bl

ank

blan

k30

40

-347

-0.0

743

-137

110

1.00

0E+

003.

6038

2.94

15E

-01

713

1616

06

7629

54bl

ank

blan

k24

147

4-5

00-0

.04

386

1011

01.

000E

+00

3.40

123.

9696

E-0

18

1461

179

796

1851

blan

kbl

ank

226

556

-592

-0.1

574

-855

110

1.00

0E+

003.

2675

4.70

74E

-01

Dis

crim

inat

ion

Stat

isti

csA

ggre

gate

sC

ompo

site

Dis

crim

inat

ion

Stat

isti

cs R

ejec

tions

34

01

53

22

43

46

542

533

Per

cent

age

2%3%

0%1%

6%4%

1%1%

3%2%

3%4%

4%2%

4%24

%1,

702

140

140

Poe

ms

Bas

elin

e: S

tand

ard

Stat

isti

cal P

rofi

leC

ompo

site

Thr

esho

lds

Blo

cks

Tes

ted

140

140

140

140

8181

140

140

140

140

140

140

140

2.51

6E-0

13.

4811

E-0

1 M

ean

5.66

81.6

316

.02

29.4

550

.66

348.

7533

1.54

510.

21-5

90.9

8-0

.04

33.0

6-4

.38

18.7

5 S

td D

ev1.

5941

.02

5.25

11.3

621

.22

71.1

813

3.35

269.

7917

4.73

0.07

33.0

89.

8545

.17

Sh D

iscr

ete

Rej

ectio

n Pr

ofile

(Se

e no

te in

key

) M

inim

um3.

0012

.24

4.00

9.00

10.1

020

5.48

16.0

0-7

2.00

-100

0.00

-0.2

6-9

0.04

-36.

57-1

03.0

0 M

inim

um0

Max

imum

13.0

024

3.41

29.0

055

.00

123.

3856

0.61

649.

0010

00.0

0-1

28.0

00.

1314

7.34

29.7

437

1.36

Max

imum

1

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

le

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

BoB

5B

oB7

BoB

8T

-E S

lope

T

est

T-E

Rar

e W

ords

T-E

New

W

ords

Buc

kets

Blo

ck G

loba

l Min

424

38

1823

593

0-8

89-0

.22

-40

-24

-77

Glo

bal M

ax9

243

2955

123

561

761

1000

-209

0.15

116

1210

0 M

in to

160

04

243

818

235

9318

8-8

89-0

.223

6-4

0-2

4-7

7 M

ax to

160

09

243.

4077

0823

3312

3.37

6623

561

761

1000

-209

0.15

116

1210

0 M

in f

rom

160

04

243

1318

235

930

-889

-0.2

236

-40

-24

-77

Max

fro

m 1

600

924

3.40

7708

2955

123.

3766

2356

176

110

00-2

090.

1511

612

100

Blo

cks

Tes

ted

438

APPENDIX SEVEN, POEMS: KEY TO TABLES FOR 1,500-WORD POEM TESTS







3 to 29






BoB8 Bundle of badges 8. See text for components -889 to -209







2.710E-01


See Table 2. Prob. that observed comp. probability score would occur by chance, Sh.’s lowest block.

1.2386E-01



Only 7 of 27 Shakespeare baseline poem blocks (26%) have even one individual rejection in 12 tests. None of 27 blocks of poems by others have fewer than two individual rejections in 12 tests. Both dis-crete and continuous composite scoring pass 100% of Shakespeare’s blocks and reject 96-100% of oth-ers’ blocks. Perhaps a quarter could be considered close calls, but most, including Oxford, Bacon and Marlowe, could not. Test Sensitivities: G = genre; T = time of composition; E = editing; P = prosody.

439

Shak

espe

are

Poe

ms

Bas

elin

e D

ata,

Blo

cksi

ze =

750

Poem

and

Blo

ckG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

liti

cs/1

000

line

sP

rocl

itic

s/1

000

line

sno

/(no

+no

t)B

oB5

BoB

7B

oB8

Mod

alB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ecti

ons

Dis

cret

e C

ompo

site

P

roba

bili

ty

(4.0

%)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Pro

babi

lity

Ven

us1

1023

18

1131

188

074

714

-333

5111

13.

618E

-01

4.17

389.

6035

E-0

22

916

011

411

211

200

570

1000

-636

2011

13.

618E

-01

4.05

271.

2609

E-0

13

876

126

2129

237

529

600

-684

108

111

3.61

8E-0

13.

2667

4.71

19E

-01

412

150

188

104

375

200

297

1000

-667

5811

01.

000E

+00

3.10

135.

6503

E-0

15

1075

195

1026

033

338

257

1-4

2968

111

3.61

8E-0

13.

1273

5.50

28E

-01

610

160

209

3344

442

922

560

0-6

9284

110

1.00

0E+

003.

1906

5.14

28E

-01

78

126

235

4235

426

326

360

0-5

2450

111

3.61

8E-0

13.

4927

3.48

88E

-01

812

216

1716

6322

940

045

650

0-5

7271

110

1.00

0E+

002.

7619

7.46

18E

-01

911

142

1011

4628

736

434

862

5-3

6652

110

1.00

0E+

001.

6739

9.93

09E

-01

1013

138

218

4631

560

0-1

971

4-3

0475

111

3.61

8E-0

13.

3295

4.36

12E

-01

1111

4313

1361

272

471

171

538

-385

260

111

3.61

8E-0

13.

8829

1.79

00E

-01

1215

111

413

1927

820

047

733

3-6

3643

110

1.00

0E+

003.

2349

4.89

14E

-01

Luc

rece

114

8311

2431

265

500

612

1000

-421

1211

01.

000E

+00

3.34

554.

2729

E-0

12

1518

412

2031

286

889

581

778

-630

166

111

3.61

8E-0

14.

2575

7.86

96E

-02

38

128

1211

6129

627

326

210

00-2

8651

110

1.00

0E+

002.

6948

7.77

49E

-01

411

136

613

1129

750

075

010

00-2

8611

511

01.

000E

+00

3.32

194.

4033

E-0

15

1116

017

1341

245

500

273

500

-333

7411

01.

000E

+00

2.48

608.

6107

E-0

16

1315

822

1331

255

500

262

1000

-619

7511

01.

000E

+00

2.78

927.

3290

E-0

17

1111

518

1377

242

500

256

333

-273

9511

01.

000E

+00

3.03

806.

0072

E-0

18

1418

810

1520

316

333

167

1000

-750

146

110

1.00

0E+

002.

7270

7.62

71E

-01

911

268

1312

1020

433

366

010

00-4

7473

110

1.00

0E+

003.

9605

1.53

21E

-01

1011

134

127

4126

566

768

055

6-8

4699

110

1.00

0E+

003.

6992

2.50

97E

-01

1115

129

1313

1031

636

449

710

00-5

8373

110

1.00

0E+

002.

3065

9.14

69E

-01

1210

8215

1166

429

273

294

455

-333

189

110

1.00

0E+

003.

3155

4.43

89E

-01

139

152

2012

4137

837

513

166

7-5

2011

311

01.

000E

+00

2.58

908.

2261

E-0

114

1610

016

2510

276

375

113

778

-680

131

110

1.00

0E+

003.

3260

4.38

06E

-01

1516

131

1116

8232

70

415

600

-250

186

111

3.61

8E-0

13.

7164

2.43

59E

-01

1615

153

1117

8241

842

911

510

00-8

4616

511

01.

000E

+00

3.48

013.

5535

E-0

117

1189

1419

4842

950

00

1000

-714

5211

13.

618E

-01

3.16

535.

2867

E-0

118

1571

1130

7132

940

046

760

0-8

9514

311

01.

000E

+00

3.98

741.

4491

E-0

119

1120

04

2152

312

556

7750

0-6

5511

911

01.

000E

+00

3.15

705.

3339

E-0

120

999

116

7829

950

050

520

0-3

33-1

911

26.

923E

-02

4.25

457.

9272

E-0

2So

nnet

s1

1324

16

1815

342

912

557

960

0-4

6783

110

1.00

0E+

004.

6212

2.98

70E

-02

211

5013

1831

286

364

432

1000

-263

8511

01.

000E

+00

2.36

288.

9970

E-0

13

1799

1618

8243

910

037

710

00-5

6513

711

13.

618E

-01

3.60

352.

9429

E-0

14

150

821

8334

511

136

375

00

129

112

6.92

3E-0

24.

1580

9.96

14E

-02

520

149

1018

9230

625

022

550

0-4

2711

711

13.

618E

-01

3.76

092.

2511

E-0

16

100

515

9233

725

025

1000

-471

4211

26.

923E

-02

3.41

393.

9015

E-0

17

1324

1015

7139

825

059

800

-750

9511

01.

000E

+00

2.78

677.

3413

E-0

18

1210

25

1341

286

556

697

1000

-655

7511

01.

000E

+00

2.88

856.

8225

E-0

19

1115

27

1931

306

636

543

833

-478

6011

01.

000E

+00

2.57

098.

2977

E-0

110

1552

214

3130

666

737

910

00-7

0011

011

13.

618E

-01

3.26

754.

7074

E-0

111

1473

419

8230

633

354

510

00-5

2686

110

1.00

0E+

002.

4906

8.59

48E

-01

1213

101

117

1026

533

341

910

00-2

9422

311

26.

923E

-02

3.81

812.

0265

E-0

113

1115

121

1341

306

286

249

636

-467

107

110

1.00

0E+

002.

1100

9.54

78E

-01

1410

505

1371

265

300

341

500

-692

9311

01.

000E

+00

2.50

388.

5483

E-0

115

1424

418

5132

742

939

110

00-4

12-1

111

01.

000E

+00

3.16

605.

2827

E-0

116

140

517

2036

730

019

350

0-4

8247

111

3.61

8E-0

13.

0150

6.13

56E

-01

440

Shak

espe

are

Poe

ms

Bas

elin

e D

ata,

Blo

cksi

ze =

750

Poe

m a

nd B

lock

Gra

de L

evel

HC

W

/2

0K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

liti

cs/1

000

line

sP

rocl

itic

s/1

000

line

sno

/(no

+no

t)B

oB5

BoB

7B

oB8

Mod

alB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ecti

ons

Dis

cret

e C

ompo

site

P

roba

bili

ty

(4.0

%)

Con

tinu

ous

Com

posi

te

Err

or

Con

tinu

ous

Com

posi

te

Pro

babi

lity

1714

101

1423

4143

935

748

650

0-4

0811

111

01.

000E

+00

2.90

426.

7395

E-0

118

1126

2023

2030

637

523

466

7-3

3377

110

1.00

0E+

003.

0107

6.15

96E

-01

1914

04

2210

430

237

513

710

00-6

0057

111

3.61

8E-0

13.

5011

3.44

59E

-01

2010

954

814

331

616

721

710

00-9

2691

111

3.61

8E-0

14.

3562

6.15

17E

-02

2114

100

712

6151

023

117

963

6-3

1010

411

01.

000E

+00

3.41

513.

8951

E-0

122

1012

26

1911

244

933

310

710

00-4

3870

110

1.00

0E+

003.

4106

3.91

92E

-01

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ecti

ons

24

34

00

34

01

324

41

Per

cent

age

4%7%

6%7%

0%0%

6%7%

0%2%

6%4%

7%2%

594

5454

Poe

ms

Bas

elin

e: S

tand

ard

Stat

isti

cal P

rofi

leC

ompo

site

Thr

esho

lds

Blo

cks

Tes

ted

5454

5454

5454

5454

5454

543.

618E

-01

6.15

17E

-02

Mea

n12

.24

112.

9311

.15

14.8

752

.65

320.

0536

5.37

325.

3175

3.41

-511

.46

92.3

2 S

td D

ev2.

5061

.60

5.92

5.42

33.2

868

.82

166.

5719

6.18

230.

4518

9.20

51.9

2S

h D

iscr

ete

Rej

ecti

on P

rofi

le (

See

not

e in

key

) M

inim

um8.

000.

001.

004.

0010

.20

187.

500.

00-1

9.00

200.

00-9

26.0

0-1

8.63

Min

imum

0 M

axim

um20

.00

267.

7423

.00

30.0

015

3.06

510.

2088

9.00

750.

0010

00.0

00.

0026

0.11

Max

imum

1

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

le

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

liti

cs/1

000

line

sP

rocl

itic

s/1

000

line

sno

/(no

+no

t)B

oB5

BoB

7B

oB8

Mod

alB

lock

Glo

bal M

in8

243

610

152

100

59-1

46-9

29-1

1 G

loba

l Max

1626

828

5115

751

066

775

010

00-1

4318

9 M

in to

160

08

243

610

152

1059

200

-929

-11

Max

to 1

600

1626

828

3215

751

0.21

667

750

1000

-143

189

Min

fro

m 1

600

824

312

1015

210

5914

6-9

29-1

1 M

ax f

rom

160

016

268

2851

157

510.

2166

775

010

00-1

4318

9

Blo

cks

Tes

ted

441

Oth

er P

oets

ver

sus

Shak

espe

are

Bas

elin

e, B

lock

size

= 7

50

Poem

and

Blo

ckG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

liti

cs/1

000

line

sPr

ocli

tics

/100

0 li

nes

no /(

no+

not)

BoB

5B

oB7

BoB

8M

odal

Blo

ckN

umbe

r of

T

ests

Dis

cret

e R

ejec

tion

s

Dis

cret

e C

ompo

site

Pr

obab

ilit

y (4

.0%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abil

ity

Bac

75-1

110

34

2233

025

055

010

00-3

3369

711

38.

291E

-03

12.2

410

<1.

0000

E-1

5B

ac75

-214

08

90

140

400

695

1000

-530

517

114

6.73

4E-0

49.

2909

8.81

76E

-14

Gre

75-1

1210

76

90

118

444

435

1000

-412

105

112

6.92

3E-0

23.

8809

1.79

70E

-01

Gre

75-2

879

318

011

776

919

325

0-6

9212

611

38.

291E

-03

5.38

972.

2307

E-0

3D

del7

5-1

1112

520

2010

121

615

-91

500

074

113

8.29

1E-0

35.

3653

2.44

97E

-03

Dde

l75-

213

587

3224

9442

910

066

7-8

0023

811

26.

923E

-02

5.89

182.

7627

E-0

4B

arne

s75-

17

2623

1113

187

333

511

0-1

5422

711

38.

291E

-03

6.16

267.

8919

E-0

5B

arne

s75-

25

8618

2511

227

400

115

0-1

1114

211

38.

291E

-03

5.83

423.

5657

E-0

4M

he75

-19

266

2083

240

250

191

1000

-250

189

110

1.00

0E+

003.

8703

1.83

45E

-01

Mhe

75-2

950

1224

3836

833

342

210

00-2

2212

411

01.

000E

+00

3.22

844.

9282

E-0

1O

x75-

19

510

822

108

400

386

1000

-304

356

113

8.29

1E-0

36.

8189

2.64

03E

-06

Ox7

5-2

70

07

011

775

0-1

210

0060

057

511

91.

340E

-11

12.4

544

<1.

0000

E-1

5Id

ea75

-16

7612

910

242

842

215

1000

-131

215

114

6.73

4E-0

45.

4706

1.62

67E

-03

Idea

75-2

887

78

7115

30

369

692

-467

104

111

3.61

8E-0

14.

0468

1.27

72E

-01

Nas

h75-

16

973

1249

272

333

273

1000

-667

161

111

3.61

8E-0

13.

5589

3.15

73E

-01

Nas

h75-

27

657

1626

410

636

400

714

-714

264

112

6.92

3E-0

24.

7768

1.87

50E

-02

Spam

75-1

1115

17

1610

356

200

387

1000

-154

5111

13.

618E

-01

3.08

715.

7307

E-0

1Sp

am75

-212

5813

1111

356

333

206

1000

-217

161

110

1.00

0E+

002.

9964

6.23

89E

-01

Fid7

5-1

710

118

630

303

1000

323

1000

-667

116

112

6.92

3E-0

25.

0505

7.67

76E

-03

Fid7

5-2

424

1510

2025

342

926

110

00-4

8249

211

26.

923E

-02

8.76

166.

2070

E-1

2C

he75

-119

127

1025

4847

633

353

810

00-5

4815

011

13.

618E

-01

4.43

495.

0102

E-0

2C

he75

-221

014

3143

298

015

810

0077

6311

46.

734E

-04

6.44

921.

9052

E-0

5M

idg7

5-1

1450

66

120

232

333

400

600

-667

190

113

8.29

1E-0

37.

1573

3.76

21E

-07

Mid

g75-

29

317

194

2012

180

032

6-3

34-5

7199

115

3.86

2E-0

57.

5455

3.42

09E

-08

Ffam

75-1

1916

318

149

264

500

517

1000

-200

9011

26.

923E

-02

4.13

871.

0412

E-0

1Ff

am75

-221

223

3120

2013

10

767

1000

-200

2811

53.

862E

-05

7.16

303.

6365

E-0

7D

on75

-110

362

326

5036

637

531

460

023

810

011

26.

923E

-02

6.31

813.

6943

E-0

5D

on75

-213

5113

2039

223

333

323

833

-222

127

110

1.00

0E+

002.

6818

7.83

32E

-01

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ecti

ons

127

32

79

83

36

1070

2019

Per

cent

age

43%

25%

11%

7%25

%32

%29

%11

%11

%21

%36

%23

%71

%68

% B

lock

s T

este

d28

2828

2828

2828

2828

2828

308

2828

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

liti

cs/1

000

line

sPr

ocli

tics

/100

0 li

nes

no /(

no+

not)

BoB

5B

oB7

BoB

8M

odal

Blo

ckD

iscr

ete

Con

tinuo

us

Glo

bal M

in8

243

610

152

100

59-1

46-9

29-1

13.

618E

-01

6.15

17E

-02

Glo

bal M

ax16

268

2851

157

510

667

750

1000

-143

189

Min

to 1

600

824

36

1015

210

5920

0-9

29-1

1 M

ax to

160

016

268

2832

157

510.

2166

775

010

00-1

4318

9Sh

Dis

cret

e R

ejec

tion

Pro

file

(Se

e no

te in

key

) M

in f

rom

160

08

243

1210

152

1059

146

-929

-11

Min

imum

0 M

ax f

rom

160

016

268

2851

157

510.

2166

775

010

00-1

4318

9 M

axim

um1

442

APPENDIX EIGHT, POEMS: KEY TO TABLES FOR 750-WORD POEM TESTS



Date late Latest supposed date of composition.




3 to 28




No/no+not Ratio of the number of occurrences of no to that of no plus not combined, times 1000.

100 to 667


BoB7 Bundle of badges 7. See text for components. -146 to 1000


Modal Block: G Modal Score per Block. -11 to 189




3.618E-01


See Table 2. Probability that observed composite probability score would occur by chance, here measured for Sh.’s lowest-probability block.

6.1517E-02



Of 54 Shakespeare blocks tested, 4 (7%) have more than one rejection in 11 tests. Of 28 non-Shakespeare blocks tested, 20 (71%) have more than one rejection. For blocks this small, accuracy is significantly lower than for larger blocks, but much better than chance: 7% false negatives, 29% false positives, many close calls, 85% of 82 blocks correctly assigned. Some rejection calls, notably Bacon and Oxford, are still not close even at 750 words. Test Sensitivities: G = genre; T = time of composition; E = editing; P = prosody.

443

Shak

espe

are

Pla

y V

erse

Bas

elin

e D

ata,

Blo

cksi

ze =

750

Play

and

Blo

ckD

ate

Lat

eW

ords

Lin

esG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Pro

clit

ics

/100

0 lin

esB

oB5

BoB

7B

oB8

Buc

kets

Blo

ckN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ilit

y (4

.0%

)

Con

tinu

ous

Com

posi

te

Err

or

Con

tinu

ous

Com

posi

te

Prob

abil

ity

Ric

hard

III

1593

174

997

916

013

3021

278

423

500

-943

2910

13.

352E

-01

3.03

465.

1241

E-0

12

746

947

107

1819

038

342

950

0-7

1413

101

3.35

2E-0

12.

4700

8.06

72E

-01

374

994

527

1214

6426

644

171

4-6

4313

100

1.00

0E+

002.

0211

9.43

44E

-01

474

893

613

412

1032

312

132

600

-769

3410

01.

000E

+00

2.36

128.

4960

E-0

15

742

956

2714

1942

253

250

556

-523

910

01.

000E

+00

1.51

809.

9343

E-0

16

768

975

7810

1931

340

237

429

-517

3210

01.

000E

+00

1.40

109.

9662

E-0

17

761

935

131

1816

4335

532

766

7-1

000

-57

101

3.35

2E-0

12.

9187

5.78

30E

-01

874

791

510

722

1244

286

230

556

-875

-18

100

1.00

0E+

002.

5111

7.88

97E

-01

974

796

654

1126

3117

740

050

0-6

6719

100

1.00

0E+

002.

1561

9.13

37E

-01

1075

092

527

1522

4348

971

260

0-6

3018

100

1.00

0E+

003.

3972

3.16

95E

-01

1175

194

680

1320

4336

222

485

7-6

9713

100

1.00

0E+

001.

3621

9.97

33E

-01

1273

590

50

920

5631

158

180

0-8

379

101

3.35

2E-0

13.

6641

2.00

84E

-01

1373

492

519

118

1354

348

241

714

-714

-410

01.

000E

+00

2.67

177.

1236

E-0

114

762

915

2618

1677

484

376

692

-286

4410

01.

000E

+00

3.17

604.

3290

E-0

115

753

934

279

2422

333

351

1000

-167

6910

01.

000E

+00

3.15

474.

4470

E-0

116

768

944

7819

70

340

406

800

-600

8110

13.

352E

-01

3.40

103.

1509

E-0

1R

icha

rd I

I15

951

755

9611

7916

1852

323

381

333

-400

2510

13.

352E

-01

2.76

816.

6178

E-0

12

794

102

815

17

1839

304

359

1000

-624

5510

01.

000E

+00

2.69

776.

9900

E-0

13

740

956

278

2163

305

611

1000

-369

-30

100

1.00

0E+

002.

9250

5.74

73E

-01

477

210

19

2613

2440

238

541

714

-143

-128

101

3.35

2E-0

14.

5582

2.27

02E

-02

574

597

1113

44

2352

289

407

1000

-419

-48

101

3.35

2E-0

13.

6919

1.90

54E

-01

675

292

610

610

2643

152

285

1000

-750

-710

01.

000E

+00

2.72

396.

8532

E-0

17

782

989

5117

180

469

642

400

-572

-43

101

3.35

2E-0

13.

9281

1.17

15E

-01

881

410

39

986

1149

301

321

1000

-800

-16

100

1.00

0E+

002.

8683

6.06

66E

-01

976

095

513

26

2111

284

627

647

-429

-37

100

1.00

0E+

003.

1770

4.32

34E

-01

1071

492

1156

1223

4328

351

466

7-4

6732

101

3.35

2E-0

12.

9446

5.63

62E

-01

1175

995

653

1520

7424

244

358

3-9

117

100

1.00

0E+

002.

6823

7.06

94E

-01

1287

510

96

6915

2555

284

484

800

-177

6610

01.

000E

+00

2.56

457.

6471

E-0

113

786

957

102

1021

4218

934

966

7-6

9212

100

1.00

0E+

002.

0383

9.40

10E

-01

1476

298

979

320

4129

640

210

00-8

00-2

210

01.

000E

+00

2.95

555.

5743

E-0

115

795

986

126

722

5134

758

010

00-2

86-1

0710

13.

352E

-01

3.91

901.

1953

E-0

116

737

928

010

3054

304

538

1000

-619

1610

13.

352E

-01

2.69

137.

0231

E-0

117

797

101

875

932

3021

860

310

00-8

6825

100

1.00

0E+

003.

2270

4.04

99E

-01

1878

896

625

821

6341

730

671

4-6

36-3

101

1.00

0E+

002.

0193

9.43

78E

-01

1975

691

710

610

3022

418

471

818

-706

810

01.

000E

+00

2.25

388.

8580

E-0

120

710

907

5610

2756

222

450

778

-750

410

01.

000E

+00

2.05

249.

3726

E-0

121

713

876

010

1792

345

355

1000

-810

610

13.

352E

-01

2.97

405.

4691

E-0

122

780

996

103

627

3028

318

525

0-9

2955

100

1.00

0E+

002.

6972

6.99

26E

-01

2378

697

715

31

2362

330

483

1000

-385

-38

101

3.35

2E-0

13.

2779

3.77

76E

-01

2481

097

699

1013

7240

235

575

0-3

6424

100

1.00

0E+

002.

1562

9.13

35E

-01

2567

383

1530

1117

7216

949

655

6-7

6559

101

3.35

2E-0

15.

3912

1.21

62E

-03

2669

484

411

51

860

488

236

333

-765

1110

13.

352E

-01

3.63

962.

1020

E-0

127

677

828

895

3049

378

226

857

-680

2310

01.

000E

+00

2.21

778.

9658

E-0

128

691

835

2913

2184

373

301

800

-913

4410

01.

000E

+00

2.69

427.

0082

E-0

1R

omeo

1596

176

898

615

61

1310

265

571

1000

-200

2710

13.

352E

-01

4.00

989.

7407

E-0

22

786

953

102

612

3236

818

661

9-7

24-4

810

01.

000E

+00

2.95

245.

5919

E-0

13

716

885

168

818

6836

4-6

920

0-5

30-6

910

13.

352E

-01

3.84

781.

3931

E-0

14

758

915

158

812

2235

288

429

-100

015

101

3.35

2E-0

13.

3105

3.60

69E

-01

577

096

720

83

2310

385

141

636

-273

010

01.

000E

+00

3.74

861.

7063

E-0

16

785

963

127

721

4238

568

-48

-704

6310

13.

352E

-01

3.38

553.

2270

E-0

17

703

873

114

1011

3446

027

565

2-8

8216

100

1.00

0E+

003.

0697

4.92

48E

-01

869

884

511

58

1848

214

947

4-5

7247

101

3.35

2E-0

12.

8413

6.21

71E

-01

974

991

513

48

922

407

075

0-6

19-1

910

13.

352E

-01

3.23

653.

9986

E-0

110

744

935

134

820

8649

521

750

0-1

000

-158

102

5.81

5E-0

25.

4212

1.07

74E

-03

444

Shak

espe

are

Pla

y V

erse

Bas

elin

e D

ata,

Blo

cksi

ze =

750

Play

and

Blo

ckD

ate

Lat

eW

ords

Lin

esG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Proc

liti

cs/1

000

lines

BoB

5B

oB7

BoB

8B

ucke

tsB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

ns

Dis

cret

e C

ompo

site

Pr

obab

ility

(4

.0%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Tro

ilus

1602

170

189

729

1721

439

778

-778

28

01.

000E

+00

1.86

259.

0159

E-0

12

726

883

5516

1868

307

394

231

-750

-710

01.

000E

+00

2.27

928.

7779

E-0

13

749

935

2714

2522

376

182

1000

-667

810

01.

000E

+00

2.23

988.

9006

E-0

14

769

944

2613

1443

362

389

400

-647

3310

01.

000E

+00

1.98

549.

4993

E-0

15

731

925

2716

2922

424

313

500

-790

4210

01.

000E

+00

2.36

808.

4710

E-0

16

754

955

8015

1811

337

120

77-8

6710

100

1.00

0E+

002.

6949

7.00

45E

-01

776

393

50

1517

3250

520

633

3-5

399

101

3.35

2E-0

13.

0338

5.12

86E

-01

877

197

415

69

2041

206

9269

2-8

711

101

3.35

2E-0

13.

0915

4.80

15E

-01

997

412

65

8213

1748

373

405

600

-636

1110

01.

000E

+00

1.23

609.

9885

E-0

110

1087

134

412

912

752

388

306

1000

-778

2610

13.

352E

-01

2.73

126.

8147

E-0

1L

ear

1605

172

889

755

1934

4528

124

0-1

000

-546

2910

13.

352E

-01

4.32

414.

4273

E-0

22

750

888

5319

3145

284

244

636

-357

-60

100

1.00

0E+

002.

5176

7.86

08E

-01

376

194

50

2334

2130

930

550

0-6

1578

101

3.35

2E-0

13.

0558

5.00

36E

-01

477

193

610

421

3311

269

104

667

-421

1110

01.

000E

+00

2.58

457.

5530

E-0

15

736

884

019

2057

159

270

143

-565

8810

25.

815E

-02

3.65

622.

0383

E-0

16

757

926

7919

3833

217

324

0-3

3418

100

1.00

0E+

002.

6369

7.29

85E

-01

Ant

ony

16

071

782

974

5115

3521

371

467

429

-304

110

01.

000E

+00

2.34

068.

5702

E-0

12

730

864

2722

3381

395

471

167

-167

-28

100

1.00

0E+

003.

4778

2.78

74E

-01

370

387

80

1748

4631

052

225

0-4

7418

101

3.35

2E-0

13.

1157

4.66

51E

-01

475

191

510

721

4277

308

433

0-3

0850

100

1.00

0E+

003.

1579

4.42

92E

-01

574

794

627

1841

6427

725

9-7

7-6

257

100

1.00

0E+

002.

6240

7.36

22E

-01

676

091

80

2450

2226

435

225

0-2

26-1

410

13.

352E

-01

3.87

181.

3240

E-0

17

733

936

136

2640

3217

221

910

00-5

313

101

3.35

2E-0

14.

1250

7.40

19E

-02

868

680

487

2130

3825

033

333

3-6

003

100

1.00

0E+

001.

9581

9.54

52E

-01

977

093

478

1338

4324

728

140

0-4

7312

100

1.00

0E+

001.

8325

9.71

70E

-01

1072

284

328

1225

7128

671

-146

-467

3010

01.

000E

+00

3.17

374.

3417

E-0

1T

empe

st

1611

174

989

50

2449

157

360

186

0-2

8646

101

3.35

2E-0

15.

7296

2.91

08E

-04

275

195

1080

2857

4230

518

90

-357

-16

101

3.35

2E-0

14.

5268

2.49

29E

-02

372

386

513

821

3510

540

712

833

3-1

7712

100

1.00

0E+

003.

7182

1.81

12E

-01

476

292

523

613

5154

359

-20

-667

-313

-13

102

5.81

5E-0

25.

4018

1.16

54E

-03

573

685

354

2840

118

341

630

-600

3710

01.

000E

+00

4.66

531.

6347

E-0

26

765

925

5217

4054

293

223

1000

-250

1910

01.

000E

+00

2.50

007.

9384

E-0

17

750

915

5320

5155

297

250

500

-600

-710

01.

000E

+00

2.65

117.

2277

E-0

18

761

875

105

2139

9248

312

10

-500

510

01.

000E

+00

3.69

821.

8826

E-0

19

759

905

7926

4933

256

153

-111

-625

-610

01.

000E

+00

3.66

851.

9918

E-0

110

747

906

187

1853

5635

634

545

5-2

8019

101

3.35

2E-0

13.

4593

2.87

29E

-01

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ectio

ns4

104

33

04

35

440

310

Per

cent

age

4%11

%4%

3%3%

0%4%

3%6%

4%4%

3%11

%89

890

90

Poe

ms

Bas

elin

e: S

tand

ard

Stat

isti

cal P

rofi

leC

ompo

site

Thr

esho

lds

Blo

cks

Tes

ted

9090

9090

8989

9090

9090

3.35

2E-0

11.

1715

E-0

1 M

ean

5.96

80.7

313

.51

25.2

647

.04

323.

5731

7.51

533.

74-5

59.8

68.

19 S

td D

ev2.

0553

.87

6.23

11.6

125

.73

80.4

716

4.41

382.

1623

6.74

40.8

6Sh

Dis

cret

e R

ejec

tion

Prof

ile (

See

note

in k

ey)

Min

imum

3.00

0.00

1.00

7.00

0.00

152.

17-6

9.00

-100

0.00

-100

0.00

-158

.01

Min

imum

0 M

axim

um15

.00

236.

2228

.00

57.0

015

7.30

505.

3871

2.00

1000

.00

-53.

0087

.68

Max

imum

1

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

le

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

litic

s/1

000

lines

Proc

liti

cs/1

000

lines

BoB

5B

oB7

BoB

8B

ucke

tsB

lock

Glo

bal M

in3

263

610

152

63-1

46-9

29-6

9 G

loba

l Max

1023

628

5115

750

571

210

00-8

381

Min

to 1

600

325

.906

7358

36

1015

263

200

-929

-69

Max

to 1

600

1023

6.22

0472

2332

157.

3033

7150

5.37

6344

712

1000

-83

81 M

in f

rom

160

03

25.9

0673

5812

1210

152

63-1

46-9

29-6

9 M

ax f

rom

160

010

236.

2204

7228

5115

7.30

3371

505.

3763

4471

210

00-8

381

Blo

cks

Tes

ted

445

Oth

ers

Pla

y V

erse

ver

sus

Shak

espe

are

Bas

elin

e, B

lock

size

= 7

50

Play

and

Blo

ckD

ate

Lat

eW

ords

Gra

de L

evel

HC

W

/2

0K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

liti

cs/1

000

lines

Proc

liti

cs/1

000

lines

BoB

5B

oB7

BoB

8B

ucke

tsB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ectio

ns

Dis

cret

e C

ompo

site

Pr

obab

ilit

y (4

.0%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abil

ity

Tam

215

881

874

1223

618

016

354

833

3-6

6714

103

6.21

4E-0

34.

6325

1.81

04E

-02

284

910

04

219

132

221

1000

-375

-47

103

6.21

4E-0

34.

5571

2.27

77E

-02

385

310

236

179

111

293

0-3

332

104

4.42

6E-0

44.

3593

4.02

18E

-02

485

612

937

190

9637

533

3-5

0012

103

6.21

4E-0

34.

6765

1.57

83E

-02

583

813

08

3017

143

612

500

-556

410

36.

214E

-03

4.96

036.

1476

E-0

36

850

1224

724

017

953

577

8-8

46-5

910

36.

214E

-03

4.87

788.

1694

E-0

37

833

90

1119

1812

836

310

00-8

46-1

5610

36.

214E

-03

5.59

285.

2751

E-0

48

878

1546

921

011

537

733

3-8

0728

103

6.21

4E-0

35.

6809

3.60

62E

-04

Jam

415

911

750

90

218

3212

627

560

0-6

6770

103

6.21

4E-0

34.

1511

6.93

93E

-02

274

95

270

1912

209

200

500

-875

110

13.

352E

-01

3.52

702.

5670

E-0

13

734

80

416

013

335

955

6-7

27-4

610

36.

214E

-03

4.18

036.

4494

E-0

24

720

656

620

3519

827

380

0-6

84-1

410

01.

000E

+00

2.38

178.

4199

E-0

15

759

526

526

4314

117

181

8-5

24-6

101

3.35

2E-0

13.

1215

4.63

25E

-01

674

85

010

2611

207

236

1000

-684

1510

13.

352E

-01

2.98

565.

4030

E-0

17

720

428

710

018

233

373

3-6

0013

101

3.35

2E-0

13.

3856

3.22

65E

-01

872

96

02

1511

8765

763

6-8

33-1

310

36.

214E

-03

4.79

561.

0755

E-0

2C

leo

1593

175

915

264

3211

333

615

1000

-304

710

13.

352E

-01

5.56

805.

8614

E-0

42

804

70

239

4120

432

360

0-4

0036

103

6.21

4E-0

33.

2424

3.96

68E

-01

365

710

306

420

167

9433

3-2

50-1

410

25.

815E

-02

4.42

333.

3640

E-0

24

686

1287

330

1110

016

471

4-3

1015

102

5.81

5E-0

24.

8571

8.76

20E

-03

572

99

555

2722

120

333

-200

-613

1510

25.

815E

-02

3.93

451.

1551

E-0

16

718

1828

2224

4414

373

1000

-572

-28

102

5.81

5E-0

26.

8385

1.04

19E

-06

771

77

011

1633

256

39-5

00-8

18-1

7310

44.

426E

-04

5.94

731.

0793

E-0

48

729

1255

3113

1121

123

810

00-4

84-6

710

25.

815E

-02

5.20

522.

5165

E-0

3D

bet

1594

174

116

243

1122

1037

157

710

00-9

17-7

110

36.

214E

-03

6.75

081.

7134

E-0

62

821

1473

717

3718

565

033

3-1

000

-133

103

6.21

4E-0

36.

3075

1.84

91E

-05

364

68

936

1412

226

456

778

-667

610

01.

000E

+00

2.84

406.

2021

E-0

14

659

1112

111

110

159

600

1000

-692

-101

103

6.21

4E-0

35.

2562

2.07

03E

-03

570

513

8512

1833

207

515

1000

-455

-70

102

5.81

5E-0

24.

6269

1.84

20E

-02

670

47

011

1622

118

422

600

-846

-70

103

6.21

4E-0

34.

0405

9.06

82E

-02

771

513

567

240

191

798

1000

-800

910

36.

214E

-03

5.49

437.

9823

E-0

48

709

1014

110

110

194

740

1000

-100

041

103

6.21

4E-0

35.

0110

5.14

06E

-03

JKJC

1594

175

36

534

130

250

491

750

-846

-31

101

3.35

2E-0

13.

4177

3.07

00E

-01

275

46

09

411

242

438

1000

-700

8410

36.

214E

-03

3.87

631.

3113

E-0

13

753

653

129

2229

031

410

00-7

3323

100

1.00

0E+

002.

3636

8.48

72E

-01

478

85

021

170

185

351

833

-750

810

25.

815E

-02

3.46

822.

8316

E-0

15

675

50

1511

4822

627

083

366

711

610

36.

214E

-03

6.32

581.

6837

E-0

56

682

50

207

2524

746

153

8-7

9017

101

3.35

2E-0

13.

0723

4.91

01E

-01

768

55

017

548

410

242

1000

-556

3210

25.

815E

-02

3.00

085.

3165

E-0

18

687

60

2711

5925

933

355

688

910

103

6.21

4E-0

36.

8382

1.04

36E

-06

446

Oth

ers

Pla

y V

erse

ver

sus

Shak

espe

are

Bas

elin

e, B

lock

size

= 7

50

Play

and

Blo

ckD

ate

Lat

eW

ords

Gra

de L

evel

HC

W

/2

0K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

liti

cs/1

000

line

sPr

ocli

tics

/100

0 li

nes

BoB

5B

oB7

BoB

8B

ucke

tsB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ecti

ons

Dis

cret

e C

ompo

site

Pr

obab

ilit

y (4

.0%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abil

ity

Seja

1603

176

48

265

2963

221

750

1000

-333

-11

102

5.81

5E-0

23.

9421

1.13

57E

-01

277

26

2617

2810

286

597

0-3

7534

100

3.35

2E-0

13.

0865

4.82

97E

-01

386

45

4614

300

148

418

857

-259

-710

25.

815E

-02

3.42

703.

0253

E-0

14

777

526

1333

018

832

8-4

29-3

6448

103

6.21

4E-0

33.

9879

1.02

43E

-01

576

212

2615

360

182

525

1000

-667

910

25.

815E

-02

4.50

852.

6311

E-0

26

751

427

1030

2124

754

760

0-4

6711

101

3.35

2E-0

12.

5471

7.72

76E

-01

779

26

257

4710

190

452

500

-579

2210

25.

815E

-02

3.35

963.

3561

E-0

18

804

675

1528

1029

448

210

00-1

58-1

010

13.

352E

-01

2.81

596.

3574

E-0

1W

prz

1604

175

85

5330

2212

642

592

500

-100

049

102

5.81

5E-0

25.

0002

5.34

17E

-03

273

44

035

1712

220

713

550

0-8

67-1

910

25.

815E

-02

5.42

861.

0455

E-0

33

794

70

3734

5328

413

7-3

33-6

3626

103

6.21

4E-0

34.

9198

7.07

59E

-03

476

96

7829

2055

253

203

0-8

2616

101

3.35

2E-0

13.

3148

3.58

46E

-01

573

78

028

2548

226

128

-400

-895

-510

25.

815E

-02

4.41

763.

4186

E-0

26

750

324

028

2080

184

84-1

67-1

000

2210

36.

214E

-03

5.48

298.

3680

E-0

47

748

580

4239

108

325

042

9-5

654

102

5.81

5E-0

25.

6573

3.99

66E

-04

874

314

2734

4312

232

222

6-4

29-1

000

-20

104

4.42

6E-0

46.

9715

4.81

79E

-07

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ecti

ons

1626

154

1815

58

79

123

4237

Per

cent

age

29%

46%

27%

7%32

%27

%9%

14%

13%

16%

22%

75%

66%

Blo

cks

Tes

ted

5656

5656

5656

5656

5656

560

5656

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

liti

cs/1

000

line

sPr

ocli

tics

/100

0 li

nes

BoB

5B

oB7

BoB

8B

ucke

tsB

lock

Dis

cret

eC

ompo

site

Glo

bal M

in3

263

610

152

63-1

46-9

29-6

93.

352E

-01

1.17

15E

-01

Glo

bal M

ax10

236

2851

157

505

712

1000

-83

81 M

in to

160

03

25.9

0673

583

610

152

6320

0-9

29-6

9 M

ax to

160

010

236.

2204

7223

3215

7.30

3371

505.

3763

4471

210

00-8

381

Sh D

iscr

ete

Rej

ecti

on P

rofi

le (

See

note

in k

ey)

Min

fro

m 1

600

325

.906

7358

1212

1015

263

-146

-929

-69

Min

imum

0 M

ax f

rom

160

010

236.

2204

7228

5115

7.30

3371

505.

3763

4471

210

00-8

381

Max

imum

1

STM

SHV

S84

513

2423

4449

284

757

636

-600

-110

44.

426E

-04

447

APPENDIX NINE, POEMS: KEY TO TABLES FOR 470-WORD POEM TESTS



Date late Latest supposed date of composition.




3 to 40

Open Lines: E, P Percentage of open or run-on verse lines. 7 to 28



No/no+not Ratio of the number of occurrences of no to that of no plus not combined, times 1000.

0 to 800




Modal Block: G Modal Score per Block. -35 to 154




4.378E-01


See Table 2. Probability that observed composite probability score would occur by chance, here measured for Sh.’s lowest-probability block.

3.163E-01



Of 89 Shakespeare blocks tested, 7 (8%) have more than one rejection in 11 tests. Of 40 non-Shakespeare blocks tested, 29 (73%) have more than one rejection. For blocks this small, accuracy is significantly lower than for larger blocks, but better than chance: 8% false negatives, 27% false posi-tives, many close calls, 72% of 129 blocks correctly assigned. Some rejection calls, notably Bacon and Oxford, are still not close even at 470 words. Test Sensitivities: G = genre; T = time of composition; E = editing; P = prosody.

448

Shak

espe

are

Poe

ms

Bas

elin

e D

ata,

Blo

cksi

ze =

470

Poem

and

Blo

ckD

ate

Lat

eG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

no /(

no+

not)

BoB

5B

oB7

BoB

8M

odal

Blo

ckN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ility

(5

.1%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Ven

us15

931

1020

013

1233

133

019

210

00-6

0019

111

4.37

8E-0

13.

0604

5.88

14E

-01

29

160

37

1721

70

118

500

-167

4811

01.

000E

+00

3.29

424.

5577

E-0

13

820

023

817

267

200

630

1000

-100

0-3

511

01.

000E

+00

4.24

838.

0472

E-0

24

840

73

1730

050

0-2

900

-100

068

112

1.05

4E-0

14.

6519

2.73

14E

-02

59

200

133

1740

00

354

1000

-400

411

14.

378E

-01

3.30

514.

4968

E-0

16

1412

020

1216

721

725

040

510

00-5

0034

110

1.00

0E+

003.

4583

3.66

67E

-01

78

4027

217

350

200

234

455

-200

-611

14.

378E

-01

3.97

221.

4956

E-0

18

1180

3310

1728

342

930

010

00-7

3355

110

1.00

0E+

003.

2237

4.95

49E

-01

97

160

207

3350

040

040

577

8-6

0060

111

4.37

8E-0

13.

0803

5.76

92E

-01

109

120

405

5038

344

444

429

-667

3111

14.

378E

-01

4.35

776.

1281

E-0

211

716

020

733

350

167

421

500

-530

3411

14.

378E

-01

2.73

127.

6075

E-0

112

1020

017

1283

233

500

506

333

-333

4211

01.

000E

+00

2.59

238.

2128

E-0

113

1326

013

1767

300

044

666

7-7

7855

111

4.37

8E-0

12.

8824

6.85

46E

-01

1410

4013

1067

317

375

317

1000

-455

5511

01.

000E

+00

1.53

819.

9674

E-0

115

1216

010

100

333

800

-12

-200

-250

2411

21.

054E

-01

4.00

451.

3980

E-0

116

1120

023

783

217

500

119

1000

-273

6511

01.

000E

+00

3.07

555.

7963

E-0

117

1240

2010

1735

037

5-5

555

6-3

3492

110

1.00

0E+

002.

6969

7.76

54E

-01

1810

4010

1310

040

050

019

560

0-4

2983

110

1.00

0E+

002.

0815

9.59

16E

-01

1912

120

1012

021

742

957

820

0-6

9237

111

4.37

8E-0

12.

6492

7.97

60E

-01

2013

140

1313

3835

816

740

542

9-5

5625

110

1.00

0E+

001.

5991

9.95

38E

-01

Luc

rece

1594

112

803

3533

233

072

310

00-3

3469

111

4.37

8E-0

14.

3784

5.81

03E

-02

212

8010

717

283

857

410

1000

-556

-26

111

4.37

8E-0

13.

2919

4.57

05E

-01

315

160

1220

3325

010

0063

271

4-6

0056

111

4.37

8E-0

13.

5999

2.96

00E

-01

48

200

217

6738

337

544

010

00-6

8417

511

21.

054E

-01

3.52

833.

3086

E-0

15

980

1712

6726

70

215

1000

125

2611

14.

378E

-01

3.53

893.

2558

E-0

16

1080

1312

031

733

380

510

00-3

0031

111

4.37

8E-0

12.

8431

7.05

83E

-01

79

180

713

1720

010

0034

210

00-3

3357

111

4.37

8E-0

13.

5996

2.96

14E

-01

811

120

2013

6731

70

317

0-3

3384

110

1.00

0E+

002.

7664

7.44

01E

-01

912

120

1013

3315

050

028

010

00-6

679

111

4.37

8E-0

12.

4244

8.81

40E

-01

1010

200

337

5030

038

525

910

00-4

5559

110

1.00

0E+

003.

3493

4.25

20E

-01

117

120

817

5023

312

540

410

00-8

0061

111

4.37

8E-0

12.

6216

8.09

29E

-01

1211

180

1310

8336

733

347

110

00-8

8250

110

1.00

0E+

002.

3109

9.13

58E

-01

1314

140

818

031

70

790

-500

135

111

4.37

8E-0

13.

5208

3.34

62E

-01

1411

300

1212

1720

025

053

310

00-4

2978

111

4.37

8E-0

13.

3314

4.35

07E

-01

159

180

237

3320

060

075

050

0-6

3640

110

1.00

0E+

003.

4768

3.57

05E

-01

1611

180

137

5018

350

070

710

00-1

000

3111

01.

000E

+00

3.48

443.

5314

E-0

117

1112

015

1217

467

082

010

00-7

7893

111

4.37

8E-0

13.

3620

4.18

24E

-01

189

120

1212

018

330

80

1000

-619

7611

14.

378E

-01

2.72

937.

6164

E-0

119

1440

1010

8348

350

021

233

3-2

0040

110

1.00

0E+

002.

9813

6.32

22E

-01

209

4018

817

400

041

566

7-3

6912

411

01.

000E

+00

3.00

416.

1962

E-0

121

1128

017

1850

383

444

8810

00-6

2573

111

4.37

8E-0

12.

9749

6.35

73E

-01

228

4012

2317

233

0-2

510

00-7

1454

110

1.00

0E+

003.

3947

4.00

46E

-01

2310

120

817

3326

766

730

420

0-4

0010

211

01.

000E

+00

2.48

678.

6083

E-0

124

1412

012

2310

038

30

627

0-4

0076

110

1.00

0E+

003.

4451

3.73

59E

-01

2511

403

1011

738

350

0-3

010

00-6

2511

411

01.

000E

+00

3.10

415.

6345

E-0

126

1424

023

2050

367

400

292

1000

-100

015

911

14.

378E

-01

4.05

841.

2454

E-0

127

1012

015

1850

533

667

-100

1000

-625

6311

14.

378E

-01

3.30

624.

4907

E-0

128

1480

1013

8328

340

055

910

00-8

5763

110

1.00

0E+

002.

2883

9.19

18E

-01

2910

180

828

5021

750

012

0-8

2615

011

01.

000E

+00

4.32

066.

7328

E-0

230

912

03

1783

350

571

371

500

-473

4011

01.

000E

+00

2.07

589.

5999

E-0

131

910

013

1374

315

051

533

3-3

3357

110

1.00

0E+

002.

3063

9.14

74E

-01

449

Shak

espe

are

Poe

ms

Bas

elin

e D

ata,

Blo

cksi

ze =

470

Poem

and

Blo

ckD

ate

Lat

eG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs

(%C

)

Ope

n L

ines

(%

C)

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

no /(

no+

not)

BoB

5B

oB7

BoB

8M

odal

Blo

ckN

umbe

r of

T

ests

Dis

cret

e R

ejec

tions

Dis

cret

e C

ompo

site

Pr

obab

ility

(5

.1%

)

Con

tinuo

us

Com

posi

te

Err

or

Con

tinuo

us

Com

posi

te

Prob

abili

ty

Sonn

ets

1603

111

180

710

143

321

046

910

00-5

5612

811

01.

000E

+00

3.32

924.

3629

E-0

12

1728

07

2071

536

142

758

0-1

67-2

111

4.37

8E-0

15.

1852

4.78

04E

-03

311

4013

1354

143

400

400

1000

-333

4411

14.

378E

-01

2.37

328.

9675

E-0

14

1440

328

3628

640

049

210

00-5

306

110

1.00

0E+

003.

2378

4.87

50E

-01

515

803

389

536

200

279

1000

-250

9811

14.

378E

-01

3.71

212.

4542

E-0

16

1340

1025

107

357

053

177

8-2

9431

110

1.00

0E+

003.

1405

5.42

77E

-01

718

4020

2289

250

167

254

1000

0-4

111

4.37

8E-0

14.

3207

6.73

11E

-02

822

120

718

107

339

0-1

610

00-2

8616

811

21.

054E

-01

5.34

892.

6078

E-0

39

1480

310

107

214

375

333

600

-500

141

110

1.00

0E+

002.

8921

6.80

35E

-01

109

07

1810

748

225

038

1000

-546

6111

01.

000E

+00

3.15

305.

3567

E-0

111

110

1315

3642

925

0-1

2350

0-7

7825

111

4.37

8E-0

13.

0398

5.99

71E

-01

1216

403

1571

286

250

228

1000

-556

2111

01.

000E

+00

2.55

608.

3553

E-0

113

1280

37

5423

271

459

410

00-6

6722

110

1.00

0E+

002.

8921

6.80

35E

-01

1414

100

725

1839

30

803

1000

-375

117

110

1.00

0E+

003.

6761

2.61

08E

-01

1511

803

1818

321

667

636

1000

-539

7211

01.

000E

+00

2.54

398.

4013

E-0

116

1314

03

1854

339

833

262

778

-636

2911

14.

378E

-01

2.60

428.

1646

E-0

117

1210

00

536

339

800

333

1000

-667

2211

21.

054E

-01

3.04

655.

9596

E-0

118

170

018

7137

525

060

610

00-9

0939

111

4.37

8E-0

13.

7450

2.31

61E

-01

1911

120

317

5433

950

053

810

00-3

3350

110

1.00

0E+

001.

9311

9.77

18E

-01

2011

400

130

232

500

492

1000

-333

115

112

1.05

4E-0

12.

9837

6.30

89E

-01

2114

140

722

1835

716

749

310

00-2

6343

110

1.00

0E+

002.

5693

8.30

39E

-01

2210

180

2712

5428

650

023

142

9-7

1423

611

14.

378E

-01

4.59

063.

2616

E-0

223

1312

07

736

214

286

-154

333

-539

9611

14.

378E

-01

2.97

936.

3332

E-0

124

110

312

8930

425

522

1000

-600

8011

01.

000E

+00

2.65

427.

9545

E-0

125

1140

017

7130

40

345

-100

0-6

0056

112

1.05

4E-0

15.

0386

7.99

71E

-03

2610

03

1236

339

462

429

1000

-400

-911

01.

000E

+00

2.67

657.

8568

E-0

127

130

317

1853

616

731

120

0-3

3318

110

1.00

0E+

003.

5577

3.16

32E

-01

2813

407

2018

196

500

9671

4-5

0040

110

1.00

0E+

002.

5252

8.47

09E

-01

2913

4017

2071

482

417

584

500

-429

151

110

1.00

0E+

003.

2705

4.69

06E

-01

3013

8022

230

375

333

291

1000

-500

3011

14.

378E

-01

2.70

497.

7291

E-0

131

1340

1020

1825

041

736

033

3-4

0035

110

1.00

0E+

002.

2932

9.17

99E

-01

3215

010

1571

375

417

8110

00-6

75

111

4.37

8E-0

13.

3202

4.41

28E

-01

3312

03

2010

723

237

515

110

00-1

000

4811

01.

000E

+00

3.44

903.

7154

E-0

134

1180

78

196

304

333

343

1000

-100

065

110

1.00

0E+

004.

2660

7.70

82E

-02

359

200

08

161

464

7749

1000

-857

143

111

4.37

8E-0

14.

6838

2.48

58E

-02

3616

4010

1371

554

053

600

-250

3011

01.

000E

+00

3.73

132.

3731

E-0

137

80

011

5442

935

711

181

8-4

8318

111

4.37

8E-0

12.

9726

6.36

99E

-01

3813

180

923

143

589

375

173

1000

-455

3511

01.

000E

+00

4.03

561.

3084

E-0

1

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ectio

ns4

47

77

34

52

34

507

22 P

erce

ntag

e4%

4%8%

8%8%

3%4%

6%2%

3%4%

5%8%

25%

979

8989

Poe

ms

Bas

elin

e: S

tand

ard

Stat

isti

cal P

rofi

leC

ompo

site

Thr

esho

lds

Blo

cks

Tes

ted

8989

8989

8989

8989

8989

894.

378E

-01

3.16

3E-0

1 M

ean

11.5

410

8.09

11.1

313

.93

54.6

732

4.86

329.

8532

7.13

731.

20-5

21.0

159

.71

Std

Dev

2.64

72.8

78.

106.

2740

.57

101.

0224

9.03

241.

7138

4.72

236.

7747

.58

Sh D

iscr

ete

Rej

ectio

n Pr

ofile

(Se

e no

te in

key

) M

inim

um7.

000.

000.

002.

000.

0013

3.33

0.00

-290

.00

-100

0.00

-100

0.00

-35.

17 M

inim

um0

Max

imum

22.0

030

0.00

40.0

035

.00

196.

4358

9.29

1000

.00

820.

0010

00.0

012

5.00

236.

24 M

axim

um1

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

le

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

litic

s/1

000

lines

Proc

litic

s/1

000

lines

no /(

no+

not)

BoB

5B

oB7

BoB

8M

odal

Blo

ck

Glo

bal M

in8

03

717

183

0-5

50

-100

0-3

5 G

loba

l Max

1824

040

2819

658

980

080

510

00-1

6715

4 M

in to

160

08

03

717

183

0-5

50

Max

to 1

600

1824

040

2819

658

980

080

510

00 M

in f

rom

160

08

03

717

183

0-5

50

Max

fro

m 1

600

1824

040

2819

658

980

080

510

00

Blo

cks

Tes

ted

450

Oth

er P

oets

ver

sus

Shak

espe

are

Bas

elin

e, B

lock

size

= 4

70

Poem

and

Blo

ckG

rade

Lev

elH

CW

/20K

Fem

E

ndin

gs(%

C)

Ope

n L

ines

(%

C)

Enc

liti

cs/1

000

line

sP

rocl

itic

s/1

000

line

sno

/(no

+no

t)B

oB5

BoB

7B

oB8

Mod

alB

lock

Num

ber

of

Tes

tsD

iscr

ete

Rej

ecti

ons

Dis

cret

e C

ompo

site

P

roba

bili

ty

(5.1

%)

Con

tinuo

usC

ompo

site

E

rror

Con

tinuo

usC

ompo

site

P

roba

bili

ty

Bac

5h1

120

33

011

525

059

510

00-3

3421

911

41.

671E

-03

5.10

356.

3890

E-0

3B

ac5h

29

03

1033

197

333

437

1000

-429

618

111

4.37

8E-0

112

.049

3<

1.00

00E

-15

Bac

5h3

180

109

015

733

380

410

00-5

0027

711

31.

607E

-02

6.21

666.

0831

E-0

5B

arne

s5h1

738

2112

2416

733

346

00

-250

150

112

1.05

4E-0

14.

1618

9.87

44E

-02

Bar

nes5

h26

023

2016

164

034

20

5913

611

41.

671E

-03

5.20

404.

4665

E-0

3B

arne

s5h3

412

719

2117

293

500

770

-273

132

111

4.37

8E-0

14.

4688

4.57

51E

-02

Che

5h1

2017

214

2831

492

333

571

1000

-556

192

112

1.05

4E-0

15.

3456

2.64

07E

-03

Che

5h2

2237

1123

4246

50

273

1000

-177

110

111

4.37

8E-0

15.

1338

5.74

30E

-03

Che

5h3

180

1232

6224

60

302

1000

-333

2211

14.

378E

-01

4.54

723.

6876

E-0

2D

del5

h111

215

1718

1414

166

749

1000

046

113

1.60

7E-0

24.

0024

1.40

42E

-01

Dde

l5h2

1542

2126

1810

550

012

90

-500

1911

14.

378E

-01

4.33

496.

4943

E-0

2D

del5

h311

444

3318

7050

0-1

1666

7-8

3315

711

41.

671E

-03

5.27

673.

4208

E-0

3D

on5h

19

398

330

4531

350

029

80

143

3111

31.

607E

-02

6.05

711.

2993

E-0

4D

on5h

212

177

521

6941

40

524

1000

9116

311

21.

054E

-01

4.23

588.

2936

E-0

2D

on5h

312

7015

1927

173

455

143

778

-143

4611

21.

054E

-01

2.71

887.

6652

E-0

1Ff

am5h

119

148

915

029

550

041

510

00-2

0026

112

1.05

4E-0

13.

7083

2.47

05E

-01

Ffam

5h2

2220

237

1728

113

073

510

000

2611

31.

607E

-02

6.56

771.

0290

E-0

5Ff

am5h

320

236

3019

1817

50

803

1000

-400

1211

21.

054E

-01

5.46

471.

6650

E-0

3G

re5h

112

124

77

011

537

550

010

00-6

3680

112

1.05

4E-0

13.

0260

6.07

43E

-01

Gre

5h2

1241

815

082

1000

257

1000

-333

8511

31.

607E

-02

4.16

989.

6932

E-0

2G

re5h

36

113

018

015

275

018

425

0-7

1411

811

41.

671E

-03

4.28

917.

2827

E-0

2Id

ea5h

113

131

1214

024

666

744

710

0067

115

112

1.05

4E-0

13.

5965

2.97

61E

-01

Idea

5h2

444

94

1822

882

4-8

3-1

000

-647

9811

51.

229E

-04

6.42

702.

1343

E-0

5Id

ea5h

37

708

885

155

042

983

3-3

3380

112

1.05

4E-0

13.

2359

4.88

58E

-01

Mhe

5h1

100

313

111

206

500

7010

0025

012

311

14.

378E

-01

4.61

883.

0078

E-0

2M

he5h

210

118

1030

1526

90

356

1000

-429

103

112

1.05

4E-0

13.

3843

4.06

09E

-01

Mhe

5h3

80

1323

5342

733

348

410

00-3

3311

111

01.

000E

+00

3.13

915.

4357

E-0

1M

idg5

h116

847

816

032

833

346

90

-636

7711

21.

054E

-01

10.5

936

<1.

0000

E-1

5M

idg5

h29

409

616

940

111

500

-579

7211

31.

607E

-02

3.56

003.

1519

E-0

1M

idg5

h311

408

203

1411

310

0045

90

-750

8611

51.

229E

-04

6.24

805.

2202

E-0

5N

ash5

h17

782

1431

200

333

150

1000

-530

145

112

1.05

4E-0

13.

2408

4.85

81E

-01

Nas

h5h2

680

610

8138

725

038

510

00-8

1810

711

14.

378E

-01

3.06

085.

8792

E-0

1N

ash5

h37

826

1711

426

700

433

667

-800

173

113

1.60

7E-0

23.

9253

1.64

57E

-01

Ox5

h19

390

816

127

308

714

1000

-59

220

115

1.22

9E-0

45.

2460

3.83

17E

-03

Ox5

h29

400

70

115

1000

120

1000

-500

267

115

1.22

9E-0

46.

2132

6.18

42E

-05

Ox5

h36

00

421

128

500

-304

053

834

411

72.

468E

-07

9.06

285.

7000

E-1

3Q

E1p

oem

s9

010

100

250

545

370

1000

-250

7811

14.

378E

-01

2.93

766.

5605

E-0

1Sp

am5h

111

221

318

1731

750

025

410

0023

196

111

4.37

8E-0

14.

0564

1.25

08E

-01

Spam

5h2

1243

1213

036

720

063

210

00-5

3020

111

4.37

8E-0

12.

4315

8.79

16E

-01

Spam

5h3

1270

1311

1336

025

059

1000

-158

7911

21.

054E

-01

2.46

328.

6883

E-0

1

Dis

cret

e D

iscr

imin

atio

n St

atis

tics

Agg

rega

tes

Com

posi

te D

iscr

imin

atio

n St

atis

tics

Rej

ecti

ons

153

59

1919

43

112

1010

029

30 P

erce

ntag

e38

%8%

13%

23%

48%

48%

10%

8%3%

30%

25%

23%

73%

75%

Blo

cks

Tes

ted

4040

4040

4040

4040

4040

4044

040

40

Shak

espe

are

Cor

pus

Bas

elin

e: C

onso

lida

ted

Dis

cret

e P

rofi

leC

ompo

site

Thr

esho

lds

Gra

de L

evel

HC

W

/2

0KFe

m

End

ings

Ope

n L

ines

Enc

liti

cs/1

000

line

sP

rocl

itic

s/1

000

line

sno

/(no

+no

t)B

oB5

BoB

7B

oB8

Mod

alB

lock

Dis

cret

eC

onti

nuou

s

Glo

bal M

in8

03

717

183

0-5

50

-100

0-3

54.

378E

-01

3.16

3E-0

1 G

loba

l Max

1824

040

2819

658

980

080

510

00-1

6715

4 M

in to

160

08

03

717

183

0-5

50

00

Max

to 1

600

1824

040

2819

658

980

080

510

000

0S

h D

iscr

ete

Rej

ecti

on P

rofi

le (

See

not

e in

key

) M

in f

rom

160

08

03

717

183

0-5

50

00

Min

imum

0 M

ax f

rom

160

018

240

4028

196

589

800

805

1000

00

Max

imum

1

451

APPENDIX TEN: KEY TO CHRONOLOGICAL INDICATORS IN SHAKESPEARE PLAYS

Column Heading Meaning

Title Short title of play.

Riv Seq Riverside Shakespeare late dating sequence.

MLE Seq Midline Speech Ending sequence.

Hess Seq W. Ron Hess recent Oxfordian dating sequence.

Rdate Late Late Riverside Shakespeare dates.

Oxfd Clark Late Eva Turner Clark old Oxfordian dates.

Oxfd Ogbn Sr Dorothy and Charlton Ogburn old Oxfordian dates.

Hess W. Ron Hess new Oxfordian dates.

1st clear First clear recorded mention of the play.

1st clear - Riv Years from 1st clear mention to Riverside date.

1st clear – Hess Years from 1st clear mention to Hess date.

F. End Halli Feminine Endings, per Halliday.

OL% TC Open-Line Percentage per Textcruncher computer counts.

ML Sp. End Midline Speech Endings, Percentages, per Halliday.

Light End Light Endings Percentages, per Halliday.

Weak End Weak Endings Percentages, per Halliday.

Most/10K Most’s per 10,000 words.

Colloq/20K Colloquialisms per 20,000 words, adapted from Wells & Taylor.

Arch/20K Archaisms per 20,000 words, adapted from Wells & Taylor.

SUMMARY OF RESULTS

If the Riverside play sequence is roughly correct, eight of Shakespeare’s listed style habits clearly evolved during his writing lifetime, all steadily increasing except archaisms such as “hath,” “-eth,” etc., which decreased. The best stylistic sequencer for plays is Midline Speech Endings. These trends, applied to Oxford’s early poems, could keep his very low rates of feminine endings and open lines from disqualifying him as a Shakespeare claimant. But they badly damage his case by continuing apace for years after his death. New Oxfor-dian redating tries to move all the plays earlier, while keeping vestiges of the Riverside se-quence. But it is mostly pulled from a hat; it improbably assumes an 11-year average delay between a play’s first performance and its first clear recorded mention, and it all but wipes out the trends which otherwise would shelter Oxford from a clear mismatch. Dates in bold are considered more firm than others, either by Hess or by us.

452

Appen

dix

Ten

: C

hro

nolo

gic

al I

ndic

ators

in S

hak

espea

re's

Pla

ys,

Riv

ersi

de

Lat

e D

atin

g

Title

Riv

Seq

ML

E

Seq

Hes

s

Seq

Rdat

e

Lat

e

Oxfd

Cla

rk

Lat

e

Oxfd

Ogbn

Sr

Oxfd

Hes

s

1st

clea

r

1st

Cle

ar

-Riv

1st

Cle

ar

- Hes

s

F.E

nd

Hal

li

OL

% TC

ML

sp.e

n

Lig

ht

end

Wea

k

end.

Most

/10K

Collo

q /20K

Arc

h

/20K

1h6

11

1590

1587

1587

1592

28

14

13

17

31

143

3h6

22

21591

1580

1581

1580

1592

112

14

12

13

02

40

75

2h6

33

11591

1579

1581

1579

1594

315

14

14

12

16

55

83

r34

63

1593

1581

1581

1581

1597

416

20

17

34

011

15

90

tit

55

81594

1577

1577

1584

1594

010

915

35

05

17

108

err

64

14

1594

1577

1577

1587

1594

07

17

12

10

09

38

82

tgv

78

41594

1579

1579

1582

1598

416

18

16

60

05

50

80

shr

87

51594

1579

1579

1582

1607

13

25

18

11

41

13

75

81

r29

96

1595

1582

1582

1583

1597

214

11

23

74

08

19

65

lll

10

10

71595

1579

1579

1583

1598

315

814

10

30

25

43

56

jn11

11

17

1596

1581

1582

1590

1598

28

623

13

70

727

130

rom

12

13

19

1596

1582

1583

1591

1597

16

816

15

61

747

99

mnd

13

14

91596

1581

1583

1584

1598

214

715

17

01

12

31

98

1h4

14

12

21

1597

1584

1584

1592

1598

16

529

14

52

526

115

wiv

15

18

10

1597

1585

1585

1585

1602

517

27

16

21

10

4128

60

mov

16

20

11

1597

1579

1579

1585

1598

113

18

27

22

61

924

99

2h4

17

15

12

1598

1585

1585

1585

1600

215

16

27

17

10

16

13

100

h5

18

16

22

1599

1586

1586

1592

1600

18

21

26

18

20

16

69

117

ayl

19

21

25

1599

1582

1582

1593

1600

17

26

23

22

20

17

59

57

jc20

17

15

1599

1583

1583

1587

1599

012

20

22

20

10

015

50

82

ado

21

19

16

1599

1583

1583

1587

1600

113

23

19

21

11

12

35

84

453

Appen

dix

Ten

: C

hro

nolo

gic

al I

ndic

ators

in S

hak

espea

re's

Pla

ys,

Riv

ersi

de

Lat

e D

atin

g

Title

Riv

Seq

ML

E

Seq

Hes

s

Seq

Rdat

e

Lat

e

Oxfd

Cla

rk

Lat

e

Oxfd

Ogbn

Sr

Oxfd

Hes

s

1st

clea

r

1st

Cle

ar

-Riv

1st

Cle

ar

- Hes

s

F.E

nd

Hal

li

OL

% TC

ML

sp.e

n

Lig

ht

end

Wea

k

end.

Most

/10K

Collo

q /20K

Arc

h

/20K

ham

22

25

27

1601

1585

1585

1594

1602

18

23

27

52

80

27

156

52

tn23

23

32

1602

1580

1580

1600

1602

02

26

23

36

31

17

92

46

tro

24

22

18

1602

1583

1584

1590

1603

113

24

26

31

60

13

84

47

aww

25

31

20

1603

1579

1579

1591

1623

20

32

29

32

74

11

218

205

43

mfm

26

24

23

1604

1581

1581

1592

1604

012

26

30

51

70

25

159

55

oth

27

26

24

1604

1583

1583

1592

1604

012

28

24

54

20

22

183

42

lr28

27

26

1605

1589

1589

1593

1606

113

29

31

61

51

20

164

52

mac

29

32

33

1606

1589

1590

1600

1611

511

26

35

77

21

215

232

29

ant

30

33

28

1607

1579

1580

1595

1608

113

27

41

78

71

28

25

251

21

tim

31

28

1608

1576

1576

1623

15

22

31

63

16

520

222

51

cor

32

34

29

1608

1581

1580

1596

1623

15

27

28

46

79

60

44

19

345

29

per

sh33

29

35

1608

1577

1577

1603

1608

05

22

34

71

15

519

164

65

cym

34

35

30

1610

1578

1578

1598

1611

113

31

47

85

78

52

17

250

18

wt

35

37

31

1611

1586

1586

1599

1611

012

33

48

88

57

43

20

307

17

tmp

36

36

34

1611

1583

1583

1600

1611

011

35

46

85

42

25

26

231

25

h8sh

37

30

1613

1601

1603

1613

032

51

72

45

37

22

254

30

tnksh

38

38

1613

1634

21

30

46

92

50

34

17

335

19