Shakespeare and the Oxford Imprint

Times Literary Supplement, 1992 

(The present text is from a copy of the typescript received 1992 from E.S.)



    Everyone reveres Shakespeare. But how do we discover what he really wrote, and when? Nobody has ever compiled an objective Shakespeare work-list derived solely from historical documents, as distinct from literary preconceptions. This article advocates and adopts that approach, for the first time in 400 years.

    My credentials include the Schubert, Schumann and Wolf work-lists in the New Grove Dictionary of Music. Their many thousand of definite facts have of course mostly been known and verified for years. But all the basic data and all their occasional adjustments depend solely and directly on documentary evidence, of exactly the kind which Shakespeare scholarship habitually ignores or rejects. The difference is one of discipline, in every sense. No single attested item of any competent musical work-list derives merely from any musicological theory, however widely accepted, and still less from any personal conviction, however passionately held.

    All the Shakespeare work-lists ever published differ profoundly from that model; and from each other. The present position can be epitomised by reference to recent Oxfordpublications which dominate academic communities and markets world-wide: the revisedOxford English Dictionary, 1989 (hereafter D) and the Complete Works and Textual Companion of the Oxford Shakespeare, 1986-8 (S). These two standard authorities between them cite 50 substantive works, whether plays (in various versions) or poems. Their total tally of complete agreement about authorship and date is just three items. The dating disparities alone add up to some 90 years, and apply to over 100,000 lines, or almost1,000,000 words.

    Even these massive discrepancies are confined to works which both D and S attribute to Shakespeare. In addition, D includes, as Shakespeare’s, at least another 10,000 lines expressly rejected by S, for example in so-called “bad quartos” and “collaborations”. Conversely, the S editors treat as authentically Shakespearean another 4,000 lines not thus classified by D.

    This strange news has never before been reported, because almost no one – not ordinary readers, nor reviewers or critics or commentators, not professional academics, not even the syndics of the Oxford or any other University Press – takes any active interest in such topics. All are content to accept expert opinion about dating, whatever its own date. Thus D relies on the literary theories of the 1880s and S on those of the 1980s, although they have only their anti-historical bias in common. This situation is surely scandalous per se, quite apart from the incalculable damage done to any serious scholarship that may still survive.

    Both D and S, after all, purport to offer a unique service to the world of learning. The latest editors of D specifically claim that it has illuminated every aspect of English linguistic history. Each separable sense of every word is illustrated by a dated quotation, complete with exact and detailed reference. This avowedly enables specialists to identify the origin of each usage and “even, by negative evidence, its non-existence at the given date”. For example, the epithet “choleric” in the sense of “enraged, angry, wrathful” (definition 4a) is first cited from Shakespeare’s Comedy of Errors, dated 1590, and next from his 2 Henry VI, dated 1593. Similarly, the substantive “no”, meaning an utterance of that word, or an instance of its use, or a denial (definition B2) is first cited from Love’s Labour’s Lost, dated 1588. The self-evident inference is that these are, so far as can now be ascertained, Shakespeare’s own innovations in the written language. Such information, if at all reliable, would form a firm foundation for authorship attribution, among other essential studies. But the D dates and texts are just as freely contradicted by S as the S are by D, although all are offered as facts, printed and sold by the same university press at the same time. Each of these mutual rejections is entirely understandable in itself; among their countless disagreements, only one of these Oxford schools of thought could be right. But which? And what if both are wrong?

    The D date of 1588 for Love’s Labour’s Lost (S 1594-5) no doubt derives from the textual fact of Don Armado, and the historical fact of the 1588 Armada, linked by an absurdly illogical assumption; as if character called Dunkirk in a modern play would clearly confirm composition in 1940. In fact, “1588” has no objective validity whatever. It would be agreeable to suppose that the very different S dates derive from new and definite information established by the latest and most reliable researches. In fact however the modern approach of the 1980s habitually proceeds from the same old starting-point of academic assumption. Thus the S date of 1590-1 for The Taming of the Shrew is just as imaginary as the D date of 1588 for Love’s Labour’s Lost. It too derives solely form literary theory, in defiance of the documentary evidence. The present preconceptions are often so strong that they take total precedence, even over the S editors’ own declared intentions and criteria. What is the date of a play? Stanley Wells has publicly answered this vital question (TLS 28 September 1984) thus: “the date of completion of the full text”. But the 1623 Folio first edition of The Shrew certainly contains what the Oxford editors themselves admit may well be an allusion to a play by John Fletcher, who was born in 1579. So The Shrew cannot be definitely dated “1590-1”. Yet that, without any qualification, is the S date, announced as a known fact.

    What powerful magnets of a priori prejudice have pulled the S editorial faculties so curiously askew? One is “style”, which means personal opinion supposedly supported by statistical “tests”. But the S statistics have been rejected as wrong and misleading (see forexample Dr. M.W.A. Smith in the March 1991 Notes and Queries). Pending the evolution of agreed mathematical methods, and the construction of an objective chronology, all dating or attribution by “style” is bound to be question-begging and worthless. An egregious example is the S allocation of The Two Gentlemen of Verona to “probably in the late 1580s” on one page and definitely “1590-1” on another. Both datings lean over-heavily on the thirty-year-old opinion of Stanley Wells that Shakespeare’s dramatic technique in this play is exceptionally naïve. But so perhaps was the early Wellsian critical technique. In fact, there is no objective evidence whatever for any such early date.

    There are many other S shibboleths unknown to D, such as the muddle-headed and baseless notions of “memorial reconstruction”. These also ignore all such explicit historical evidence as title-page dating and attribution. They have been disproved in detail for decades; the latest round of refutation began in these pages (TLS 2 September 1983). Not a syllable of rational defence has ever appeared in print anywhere, over the last seventy years; yet the same old wrong assumptions continue to be presented as proven, especially by the S editors Wells, Taylor and Schoenbaum. But the rest of the profession, especially inAmerica, is again beginning to see and say, clearly and distinctly, that the Augean stables must now be cleansed, even without help from the senior hands.

    Then at last a rational chronology and attribution will become possible. It will begin with the earliest facts instead of the latest fads. The starting-point must therefore be 1592, with Greene’s implicit attribution to Shakespeare of the line “O tiger’s heart wrapped in a woman’s hide”. Those words occur in 3 Henry VI and also in its forgotten counterpart version The True Tragedy of Richard, Duke of York. But only the latter text was printed in Tudor times (1595); the former remained unpublished until the 1623 First Folio. A new historico-documentary viewpoint would readily perceive that the 1592 reference is prima facie to the 1595 not the very different 1623 version; that Greene (who ought to know) was accordingly attributing the 1595 version to Shakespeare; that the title-page attribution of its undated third edition to “William Shakespeare, Gent.” is accordingly authentic; and that the S date of 1591 for the 1623 Folio version is entirely imaginary and irrational.

    So is the S denial to Shakespeare of other texts, such as the early quartos accepted by D as first versions, not “memorial reconstructions”. So is the attribution to Shakespeare of the poem “Shall I die?” immediately after its so-called discovery in Bodleian first-line poetry index under the heading “Shakespeare, William”. So is the S view of certain other acts and plays as “collaborative” (1 Henry VI assisted by “Nashe”, Timon of Athens and Macbeth by “Middleton”, Pericles by “Wilkins”, and so forth). So is the S dismissal of yet another work as “dubious” (such as The Taming of the Shrew2 Henry VI, 3 Henry VI and Titus Andronicus). On all these eight plays, the D attributions are in entire accord with common sense and the 1623 Folio editors.

    In some respects however the S attributions may be preferred, as genuinely deriving from the later objective research. Thus the S editors have now belatedly acknowledged thatEdward III in its entirety is indeed a Shakespeare play; and they have expressed their regret at omitting it from their so-called Complete Works. Similarly Sir Thomas More admittedly has Shakespearean components, the full extent of which has yet to be determined.

    The next step is for the D and S authorities to convene a joint conference, with Oxford as the relevant and convenient venue, to get their acts and their plays together, and reach agreement about the questions of authorship and chronology on which the world looks to them for guidance. If posterity should enquire how the present horrendous discrepancies and errors ever arose, the answer is implicit in the following typical dictum (Oxford Shakespeare Textual Companion, 1988, p. 109): “The study of authorship and chronology begins and ends in editing”. Of course only an editor could believe that. Everyone else will instantly see that such studies should begin and end in reasoning, which is a very different matter. Further, such reasoning in turn should begin with the documentary facts, which S very often excludes or decries.

    One major problem is how to date the work of a revising artist, as S rightly classifies Shakespeare. The Brahms piano trio Op. 8, for example, exists in two published versions; the 1853 original, and its 1889 revision. What is meant by “Op. 8”, and how should it be dated in the work-list? The practice varies. But the blunt description “1889” is surely wrong for a revised version that still consists mainly of 1853 inspiration, just as “1853” would be equally misleading for a work incorporating stylistically later material. On any assessment, the original should surely be separately available for study and performance. Finally, even without extant documents, the discipline of musicology could never have countenanced, let alone canonised, the bizarre conjecture that the 1853 score was a “memorial reconstruction” of the allegedly pre-existing 1889 version, prepared by a disaffected and confused cellist for a provincial tour with new partners. By analogy, The True Tragedy of Richard, Duke of York, the first version of 3 Henry VI, should also be available in print as an early Shakespeare play, not suppressed as a mere “memorial reconstruction”. Similarly the 1623 Folio text should not be assigned to either “1593” (D) or “1591” (S). The c. 1000 lines it shares verbatim with True Tragedy must of course be dated before 1592. But all that can safely be said about the completed revision is that it pre-dated Shakespeare’s death in 1616.

    This permits the further inference that he spent his Stratford retirement in revising his plays for Folio publication, like the actual sensible person that so many scholars forbid him to be. It also suggests that those style-critics who have disconcertingly discerned late strata among those supposedly apprentice works (such as the speech of the young Clifford over his dead father in 2 Henry VI) were right after all. No doubt D is also right in implying that Shakespeare coined the telling expression “chair-days”; but in his own, not in 1593.

    Dating and attributing Shakespeare by 16th or 17th century evidence alone, without any regard to 19th or 20th century theory (that is, by an exact reversal of present procedures) imposes an entirely new pattern; for example:


by 1588 (not S “1607”): the first version of Pericles.

by 1588 (not S “1592”): the first version of Titus Andronicus.

by 1589: the first versions of Hamlet and The Taming of the Shrew.

by 1592; the first versions of 2-3 Henry VI.

by 1598 (not D “1590”): A Midsummer Night’s Dream.

by 1598 (not D “1588”): Love’s Labour’s Lost.

by 1598 (neither S “late 1580s” nor S “1590-1”); The Two Gentlemen of Verona

              performed (no text published until 1623)

by 1609 (not S “1590-1”): The Taming of the Shrew, performed

              (no text published until 1623)

by 1616 (not S “1591”): the Folio 2-3 Henry VI.


    It may prove possible to adjust these dates by modern stylometric analysis, if such studies ever become demonstrably objective and reliable. Meanwhile the urgent need is to correct the culpably irrational dating methods and results of the Oxford Dictionary and the Oxford Shakespeare, which are not only mutually incompatible but demonstrably wrong.