oxford by the numbers: what are the odds …€¦ · · 2009-10-20odds that the earl of oxford...
TRANSCRIPT
* 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
326 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 327
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.
328 TENNESSEE LAW REVIEW [Vol. 72:323
(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
2004] OXFORD BY THE NUMBERS 329
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.
330 TENNESSEE LAW REVIEW [Vol. 72:323
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)
2004] OXFORD BY THE NUMBERS 331
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
332 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 333
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
334 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 335
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
336 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 337
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,
338 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 339
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
340 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 341
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
342 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 343
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
344 TENNESSEE LAW REVIEW [Vol. 72:323
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;
2004] OXFORD BY THE NUMBERS 345
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
346 TENNESSEE LAW REVIEW [Vol. 72:323
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.
2004] OXFORD BY THE NUMBERS 347
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
348 TENNESSEE LAW REVIEW [Vol. 72:323
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-
2004] OXFORD BY THE NUMBERS 349
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
350 TENNESSEE LAW REVIEW [Vol. 72:323
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-
2004] OXFORD BY THE NUMBERS 351
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
352 TENNESSEE LAW REVIEW [Vol. 72:323
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.
2004] OXFORD BY THE NUMBERS 353
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).
354 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 355
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
356 TENNESSEE LAW REVIEW [Vol. 72:323
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.
2004] OXFORD BY THE NUMBERS 357
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
358 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 359
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.
360 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 361
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
362 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 363
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
364 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 365
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
366 TENNESSEE LAW REVIEW [Vol. 72:323
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-
2004] OXFORD BY THE NUMBERS 367
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.
368 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 369
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.
370 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 371
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.
372 TENNESSEE LAW REVIEW [Vol. 72:323
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.
2004] OXFORD BY THE NUMBERS 373
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
374 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 375
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.
376 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 377
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
378 TENNESSEE LAW REVIEW [Vol. 72:323
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;
2004] OXFORD BY THE NUMBERS 379
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
380 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 381
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
382 TENNESSEE LAW REVIEW [Vol. 72:323
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.
2004] OXFORD BY THE NUMBERS 383
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.
384 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 385
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.
386 TENNESSEE LAW REVIEW [Vol. 72:323
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).
2004] OXFORD BY THE NUMBERS 387
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
388 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 389
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
390 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 391
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
392 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 393
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
394 TENNESSEE LAW REVIEW [Vol. 72:323
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
2004] OXFORD BY THE NUMBERS 395
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
396 TENNESSEE LAW REVIEW [Vol. 72:323
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.
2004] OXFORD BY THE NUMBERS 397
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
Discrete Composite Probability
Continuous Composite Probability
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
Discrete Composite Probability
Continuous Composite Probability
Shakespeare Core Profile Bounds 2 2.316E-01 3.6895E-03
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
Discrete Composite Probability
Continuous Composite Probability
Shakespeare Core Profile Bounds 2 2.316E-01 3.6895E-03
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
Play Short TitleLatest
Supposed Date
Number of Words
Discrete Rejections
Discrete Composite Probability
Continuous Composite Probability
Shakespeare Core Profile Bounds 2 2.316E-01 3.6895E-03
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
Play Short TitleLatest
Supposed Date
Number of Words
Discrete Rejections
Discrete Composite Probability
Continuous Composite Probability
Shakespeare Core Profile Bounds 2 2.316E-01 3.6895E-03
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.
Date Latest supposed date of composition.
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%
Discrete Composite Probability
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.
Continuous Composite Probability
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
Column Heading and Test Sensitivities
Meaning Shakespeare Profile
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
Column Heading and Test Sensitivities
Meaning Shakespeare Profile
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
Column Heading and Test Sensitivities
Meaning Shakespeare Profile
Date Latest supposed date of composition.
Grade Level: G, E Reading grade level score. 3 to 8
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
with (2lws) Rate of occurrence for with as the penultimate word of a sentence (per 1,000 sentences).
4 to 36
no / (no + not) Ratio of the number of occurrences of no to that of no plus not combined, times 1000.
167 to 586
BoB5 Bundles of badges 5. See text for components. 116 to 556
BoB7 Bundles of badges 7. See text for components. 136 to 944
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
Discrete Composite Probability
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
Continuous Composite Probability
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
Column Heading and Test Sensitivities
Meaning Shakespeare Profile
Date Latest supposed date of composition.
Grade Level: G, E Reading grade level score. 10 to 14
HC / 20K: E Hyphenated compound words per 20,000 words. 31 to 153
Relative Clauses Total relative clauses per 1,000 words. 7 to 17 (not shown)
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
with (2lws) Rate of occurrence for with as the penultimate word of a sentence (per 1,000 sentences).
4 to 36
no / (no + not) Ratio of the number of occurrences of no to that of no plus not combined, times 1000.
167 to 536
BoB5 Bundles of badges 5. See text for components. 116 to 556
BoB7 Bundles of badges 7. See text for components. 136 to 944
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
Modal Block: G Modal Score per Block. 281 to 1149
Discrete Rejections Total number of rejections from tests above. 0 to 1
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
Column Heading and Test Sensitivities
Meaning Shakespeare Profile
Date Latest supposed date of composition.
Grade Level: G, E Reading grade level score. 4 to 9
HC / 20K: E Hyphenated compound words per 20,000 words. 24 to 243
Fem Endings: P Percentage of feminine endings of verse lines. Mostly machine counts.
3 to 29
Open Lines: T, E, P Percentage of open or run-on verse lines. Early, to 1600: 8 to 33 Late, fr. 1600: 13 to 55
Enclitics: P Enclitic microphrases per 1000 lines. 18 to 123
Proclitics: P Proclitic microphrases per 1000 lines. 235 to 561
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 New Word Test Thisted-Efron New Word Test. -24 to 12
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 Rejections Total number of rejections from tests above. 0 to 1
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
Column Heading and Test Sensitivities
Meaning Shakespeare Profile
Date Latest supposed date of composition.
Grade Level: G, E Reading grade level score. 10 to 16
HC / 20K: E Hyphenated compound words per 20,000 words. 24 to 243
Fem Endings: P Percentage of feminine endings of verse lines. Mostly machine counts.
3 to 29
Open Lines: T, E, P Percentage of open or run-on verse lines. Early, to 1600: 7 to 24 Late, fr. 1600: 13 to 23
Enclitics: P Enclitic microphrases per 1000 lines. 18 to 123
Proclitics: P Proclitic microphrases per 1000 lines. 235 to 561
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 New Word Test Thisted-Efron New Word Test. -24 to 12
Modal Block: G Modal Score per Block. 79 to 407
Discrete Rejections Total number of rejections from tests above. 0 to 1
Discrete Composite Probability
See Table 2. Probability that observed rejections would occur by chance at Sh.’s avg. rejection rate.
2.710E-01
Continuous Composite Probability
See Table 2. Prob. that observed comp. probability score would occur by chance, Sh.’s lowest block.
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 = comp. non-rejection.
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
Column Heading and Test Sensitivities
Meaning Shakespeare Profile
Date Latest supposed date of composition.
Grade Level: G, E Reading grade level score. 10 to 16
HC / 20K: E Hyphenated compound words per 20,000 words. 24 to 243
Fem Endings: P Percentage of feminine endings of verse lines. Mostly machine counts.
3 to 29
Open Lines: T, E, P Percentage of open or run-on verse lines. Early, to 1600: 7 to 24 Late, fr. 1600: 13 to 23
Enclitics: P Enclitic microphrases per 1000 lines. 18 to 123
Proclitics: P Proclitic microphrases per 1000 lines. 235 to 561
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 New Word Test Thisted-Efron New Word Test. -24 to 12
Modal Block: G Modal Score per Block. 79 to 407
Discrete Rejections Total number of rejections from tests above. 0 to 1
Discrete Composite Probability
See Table 2. Probability that observed rejections would occur by chance at Sh.’s avg. rejection rate.
2.710E-01
Continuous Composite Probability
See Table 2. Prob. that observed comp. probability score would occur by chance, Sh.’s lowest block.
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 = comp. non-rejection.
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.
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
Column Heading and Test Sensitivities
Meaning Shakespeare Profile
Date late Latest supposed date of composition.
Grade Level: G, E Reading grade level score. 8 to 16
HC / 20K: E Hyphenated compound words per 20,000 words. 24 to 268
Fem Endings: P Percentage of feminine endings of verse lines. Mostly machine counts.
3 to 28
Open Lines: T, E, P Percentage of open or run-on verse lines. Early, to 1600: 6 to 32 Late, fr. 1600: 12 to 51
Enclitics: P Enclitic microphrases per 1000 lines. 10 to 137
Proclitics: P Proclitic microphrases per 1000 lines. 152 to 510
No/no+not Ratio of the number of occurrences of no to that of no plus not combined, times 1000.
100 to 667
Bob5 Bundles of badges 5. See text for components. 59 to 750
BoB7 Bundle of badges 7. See text for components. -146 to 1000
BoB8 Bundle of badges 8. See text for components. -929 to -142
Modal Block: G Modal Score per Block. -11 to 189
Discrete Rejections Total number of rejections from tests above. 0 to 1
Discrete Composite Probability
See Table 2. Probability that observed rejections would occur by chance at Sh.’s avg. rejection rate.
3.618E-01
Continuous Composite Probability
See Table 2. Probability that observed composite probability score would occur by chance, here measured for Sh.’s lowest-probability block.
6.1517E-02
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 750-WORD POEM BLOCKS
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
Column Heading and Test Sensitivities
Meaning Shakespeare Profile
Date late Latest supposed date of composition.
Grade Level: G, E Reading grade level score. 8 to 18
HC / 20K: E Hyphenated compound words per 20,000 words. 0 to 240
Fem Endings: P Percentage of feminine endings of verse lines. Mostly machine counts.
3 to 40
Open Lines: E, P Percentage of open or run-on verse lines. 7 to 28
Enclitics: P Enclitic microphrases per 1000 lines. 17 to 196
Proclitics: P Proclitic microphrases per 1000 lines. 183 to 589
No/no+not Ratio of the number of occurrences of no to that of no plus not combined, times 1000.
0 to 800
Bob5 Bundles of badges 5. See text for components. 55 to 805
BoB7 Bundle of badges 7. See text for components. 0 to 1000
BoB8 Bundle of badges 8. See text for components. -1000 to -167
Modal Block: G Modal Score per Block. -35 to 154
Discrete Rejections Total number of rejections from tests above. 0 to 1
Discrete Composite Probability
See Table 2. Probability that observed rejections would occur by chance at Sh.’s avg. rejection rate.
4.378E-01
Continuous Composite Probability
See Table 2. Probability that observed composite probability score would occur by chance, here measured for Sh.’s lowest-probability block.
3.163E-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 470-WORD POEM BLOCKS
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