Gabriel Egan
Department of English and Drama
Loughborough University
Loughborough
LE11 3TU

21 December 2004

    Year's Work in English Studies 2003: Shakespeare: Editions and Textual Studies

    In a review of this kind it can be difficult to know where to stop; after all, 'editions and textual studies' could encompass almost anything written about Shakespeare. In the past I have, at the risk of stepping on other contributors' toes, noticed items that discuss such matters as how the wording of stage directions affects performance. This year, because there are a great many (over 70) items to review, I shall discuss only work showing what an editor has done, or exhorting what an editor should do, or announcing a discovery about evidence that might shape what an editor will do. The burgeoning subject of how the different early versions of what we used to think of as a single play might differ in their performance potentialities bears only indirectly on what an editor should do and it has grown too large to be contained within a section of this book that is properly reserved for matters textual. Of the fine essays in Hardin L. Aasand's collection Stage Directions in Hamlet all but two fall into this category and are ignored here. Even reviewers paid by the word (as we are) must limit the fields of their enquiries.

    It was a busy year for Shakespearian textual studies with three landmark monographs appearing, but only one new substantial edition: Michael Taylor's Henry VI, Part One for the Oxford Shakespeare. The only authoritative early printing of this play is the 1623 Folio, so Taylor's relatively short introduction (77 pages) naturally has much more to say about the meanings and reception of the play than about the textual situation. Taylor first attends to the year 1592, when Thomas Nashe famously alluded to the play in performance, contrasting the heroic English past it presented with his own Puritan-ridden, usurious present (p. 2). Treating that year's Greene's Groatsworth of Wit as though it were simply Greene's work, Taylor finds it betraying a "rancid . . . anxiety of influence" that in a footnote he glosses as "mutual stimulation" and by which he clearly means to imply mental masturbation (p. 5). This seems somewhat unfair to Harold Bloom, whose notion of 'anxiety of influence' can often refer to a positive and indeed productive relation between present and preceding writers. Also somewhat misrepresented is one of Andrew Gurr's admittedly difficult essays ('The Chimera of Amalgamation', Theatre Research International 18 [1993].85-93), in which Taylor thinks Gurr opposed the whole idea that "in the early 1590s companies amalgamated" (p. 6) whereas in fact Gurr was specifically referring to the oft-alleged Strange's/Admiral's men's amalgamation.

    In a section about ' The 'Henry VI' Plays in 1592' (pp. 10-14) that discusses the sequentiality of 1, 2, and 3 Henry 6, Taylor surprisingly neglects to mention the problem that the allusion to 3 Henry 6 in Greene's Groatsworth suggests that 3 Henry 6 was at least written--and presumably, in fact, in performance--before Groatsworth's Stationers' Register entry on 20 September 1592, and yet 1 Henry 6 was apparently 'new' (according to most people's reading of Henslowe's Diary) on 3 March 1592. There seems, on this evidence, insufficient time for parts 1, 2, and 3 being written in that order. Taylor's compressed account of the debate about the order is written from the point of view of someone who knows the data and does not need them repeated. Taylor fails to mention that B. J. Sokol has presented strong evidence that 1 Henry 6 cannot have been written before 25 April 1591 because its 'garden' scene alludes to the refurbishment of the Inner Temple Garden completed by that date ('Manuscript Evidence for an Earliest Date of Henry VI Part One', N&Q 245 [2000].58-63). A much more serious error, however, is Taylor's claim that because the first printings of 2 Henry 6 and 3 Henry 6 do not name Shakespeare as the dramatist, there is nothing to link these plays to Shakespeare until the 1623 Folio firmly located them in his oeuvre (p. 11). In fact, there appeared in 1619 a printing of both plays in one volume called The Whole Contention between the Two Famous Houses Lancaster and York, and its title-page named "William Shakespeare, Gent" as the dramatist.

    Taylor treats the multiple authorship of the play as a matter of "problems and anomalies", not just the way drama was usually made and observes that if, as Gary Taylor argued, Nashe wrote act 1 of 1 Henry 6 then it was surprisingly reticent of him to praise the Talbot scenes so lavishly in his Pierce Pennilesse without mentioning his own contribution. If 1 Henry 6 was written before 2 Henry 6 and 3 Henry 6, as some have recently claimed, it is odd, remarks Taylor, that 2 Henry 6 does not follow up many of its leads and that 1 Henry 6 is more like 3 Henry 6 than like 2 Henry 6. On the other hand, 1 Henry 6 coming first would explain its weaknesses: they are beginner's flaws (p. 12). If 1 Henry 6 is a prequel, why did Shakespeare collaborate on it and so produce something inferior (p. 13)? Here again, Taylor perhaps unconsciously deems collaboration as a sign of weakness, as an anomaly. In a footnote (p. 14n1), Taylor characterizes Marlowe's Tamburlaine as a "daring venture" in being a two-parter. In fact there were many two-parters and Marlowe's play might well have not been conceived as one: the second part simply continues the story (extending an unexpected dramatic success?) and negates the first part's radical innovation by having the anti-hero eventually get his come-uppance.

    Shakespeare inherited rather than invented the history play genre, which collection of plays (mostly not by Shakespeare) Taylor sees as constituting "a critical response to Marlowe's glorification of ruthless foreign individualism." (p. 19). 1 Henry 6  is a piece of patriotic propaganda in a time of national crisis--post-Babington plot, post-Armada--but despite its patriotic elements it is not quite the kind of story one would tell to stiffen the national sinews (p. 21-24). Taylor thinks the second tetralogy (the Henry 4 and 5 plays) "infinitely superior in every way" to the first--it is refreshing to have an editor not extol his play as an undervalued masterpiece--yet the first is also vastly (if not infinitely) superior to the other history plays around at the time (p. 25). At this point, Taylor starts to write using sentence fragments that make sense only in relation to the antecedent subject of the previous sentence, as in "A pleasure we now take for granted" (p. 29n1) referring to Elizabethan audiences' pleasure in being made familiar with great ones. In this case the antecedent subject is in the body text and the fragment in a footnote, which is quite a stretch. This habit gets severely irritating in Taylor's section on the plays in recent performance: "[It was] A clear shape, we may recall, [formed] principally at the expense of Henry VI, Part One" and "[It was] Epic for theatregoers too as the three Henries were performed on the same day on eight occasions, outlasting the Terry Hands 1977 marathon by over an hour" (p. 36). In parts the introduction is repetitive: on page 41 is repeated from page 21 note 21 the claim that the encounter of Talbot and the countess of Auvergne looks forward to the "sharp-edged dialogue" between men and women in Shakespeare's romantic comedies; true, but presumably it also looks back to the same in The Taming of the Shrew. Taylor cites two critics (David Riggs and E. Pearlman) who refer to Tamburlaine playing at the Fortune, without mentioning that that theatre was built at least ten years after the play opened at the Rose.

    Taylor's sections on 'The Text' (pp. 75-77) and his 'Editorial Procedures' are remarkably brief, acknowledging no scholarship since the Textual Companion to the Oxford Complete Works and offering no discussion of what effect it might have upon editing certain parts of the play to think that Nashe, not Shakespeare, wrote them. There is, indeed, no acknowledgement of the New Textualism: "Loose ends, inconsistencies, contradictions, misplaced stage directions, changeable speech prefixes combine, with other irregularities in the Folio text, to make it very unlikely that the manuscript came from the playhouse" (p. 76). Taylor finds some theatrically unnecessary details that would suggest foul paper copy lying behind the Folio text, but too many irregularities for the foul papers to be those of one author, so he goes along with the Textual Companion's view that F's copy was "collaborative foul papers". (The date of the Oxford Complete Works itself, to which the Textual Companion is companion, is wrongly given here as 1988 rather than 1986.) Taylor confesses himself "somewhat eclectic" in choosing names for his characters, preferring historically-correct names for minor figures but Folio names for major ones (p. 80).

    In the text of the play itself there are more signs of incomplete proofing: the note to Talbot's entry in the dramatis personae has a cross-reference to "Introduction, pp. oo-oo" (p. 92). Thankfully, Joan of Arc gets her right name, Joan la Pucelle, rather than the name Joan Puzel used by Edward Burns for the rival Arden 3 edition of the play published three years ago. Likewise Taylor has Bedford say "Our isle be made a marish of salt tears" (1.1.50), sensibly following Alexander Pope, C. J. Sisson and the Oxford Complete Works and rejecting Burns's retention of "nourish of salt tears" from the Folio. At 1.2.21.1 Taylor comments that "Oxford [Complete Works] begins a new scene here, but this adherence to the letter of the dramatic law--an unwritten one--that a new scene begins each time the stage is cleared seems pedantic in this case". Actually, we know that Shakespeare did not mark scene breaks and that theatrical scribes did not add them, so they are only editorial markers anyway; that being the case, it does make sense to consistently follow the rule. Throughout the third scene (1.3.19, 36, 42, 49, 56, 79, and 84) Taylor retains F's repeated insistence that Winchester is a cardinal even though subsequent scenes show him to be only a bishop and that he gets made up to a cardinal in 5.1. This generates an anomaly but Taylor thinks that ". . . there is simply too much dramatic fall-out here from the Cardinal appearing as a Cardinal--his scarlet hat and his robes for instance--to allow us to correct F's contradictory chronology at the expense of the scene's colour and flair". This decision risks being tyrannized by the copy text, which presumably had to be corrected before performance. And yet, having accepted that F's inconsistency should stand, Taylor nonetheless emends F's "Here Glosters men beat out the Cardinalls men" to "Here Gloucester's men beat out the Bishop's men" (1.3.56.1), which suggests that he thinks that Winchester really is only a bishop at this point. If the colourfulness of the error in F is worth preserving when it made real trouble (by being in contradiction with what is said about him elsewhere) it is surely then worth retaining when it does not create trouble within a scene in which the editor has decided that Winchester is a cardinal.

    At 1.5.11 Taylor has Talbot say "Rather than I would be so vile-esteemed", which is essentially Pope's emendation of F's "pil'd esteem'd" for his edition of 1723-25, although the Textual Companion and Taylor himself credit Lewis Theobald in 1733. In support of "vile-esteemed" is Sonnet 121's 'Tis better to be vile than vile esteemed", although this does not mean quite the same thing. The Folio has Mortimer absurdly likening death or locks of hair to Nestor: "And these gray Locks, the Pursuiuants of death, | Nestor-like aged, in an Age of Care, | Argue the end of Edmund Mortimer", which Taylor (like the Oxford Complete Works) sensibly reorders to make it clear that Mortimer is comparing himself to Nestor: "And these grey locks, the pursuivants of death, | Argue the end of Edmund Mortimer, | Nestor-like agèd in an age of care"  (2.5.5-7). Likewise Taylor follows the Oxford Complete Works's assignment of one of Warwick's lines to Gloucester: "WINCHESTER Rome shall remedy this. [GLOUCESTER] Roam thither then. | [WARWICK] (to Winchester) My lord, it were your duty to forbear. | SOMERSET Ay, so the bishop be not overborne" (3.1.51-53), whereas F has Warwick say "Roame thither then. | My Lord it were your dutie to forbeare", which Sisson (New Readings in Shakespeare, pp. 69-70) defended as being said in two tones: first he taunts ('off you go then') and then he softens to remind the bishop of his duty. Following a suggestion made but not enacted by Michael Hattaway in his New Cambridge Shakespeare edition of the play, Taylor reorders F's "Looke on thy Country, look on fertile France, | And see the Cities and the Townes defac't, | By wasting Ruine of the cruell Foe, | As lookes the Mother on her lowly Babe, | When Death doth close his tender-dying Eyes". F makes an (albeit awkward) sense as it is, but is better rearranged as Taylor has it: "JOAN Look on thy country, look on fertile France, | As looks the mother on her lowly babe | When death doth close his tender-dying eyes, | And see the cities and the towns defaced | By wasting ruin of the cruel foe" (3.3.44-48).

    A footnote gloss ought to be something one could substitute for the tricky word or phrase being glossed, but for "Alarum. Enter the Earl of Suffolk with Margaret in his hand" Taylor offers the footnote gloss "0.1 in led by the". To be grammatical this should of course read "in his led by the", else the substituted phrase would read 'led by the his hand', but more importantly still Taylor's gloss switches whose hand it is: in F it is Suffolk's hand, in Taylor's edition it is Margaret's hand. Like many editors since Edward Capell, Taylor finds disorder in Suffolk's Folio lines "For I will touch thee but with reuerend hands, | I kisse these fingers for eternall peace, | And lay them gently on thy tender side" (5.4.3-5). Like Sisson, I cannot see what the problem is. In the famous scene of Suffolk and Margaret making lengthy asides (5.4.17-63), Taylor repeatedly has one or other speak "(To himself)" or "(To herself)". It has been a while since anyone concerned with staging wanted to be quite so specific that an aside is self-communion rather than making the less restrictive assertion that it merely is not to be clearly heard, or directly addressed to, the other character. Finally, at 5.4.148 Taylor deletes "Mad" from Suffolk's "Mad naturall Graces that extinguish Art" on the grounds that it is superfluous and that Burns suggested it was merely a false start in the underlying manuscript that was imperfectly deleted and hence set by the printer. Yet Burns did not just delete the word as Taylor had done. As with Hattaway's suggestion about France as a dead baby (3.3.44-48), Taylor follows another editor's suggestion even though that editor was insufficiently convinced on the point to enact it for himself in his edition.

    Easily the most important contribution to the field this year--indeed the most important contribution for a long time--is Lukas Erne's Shakespeare as Literary Dramatist, which aims to overturn the oft-repeated orthodoxy that Shakespeare took no interest in the publication of his plays. That Shakespeare might in fact have worked with a readership in mind was the subject of an essay by Richard Dutton seven years ago ('The Birth of the Author', in R. B. Parker and S. P. Zitner (ed.) Elizabethan Theater: Essays in Honor of S. Schoenbaum (Newark: University of Delaware Press [1996])) and of an article by Erne in Shakespeare Quarterly reviewed last year. In the book-length study Erne sets out his case in full. In recognition of the importance of the arguments, the University of Lancaster convened a two-day conference on the subject in July 2004, nominally titled 'New Shakespeare: A writer and his readers: The return of the author in Shakespeare studies' but effectively 'the Erne debate'. Erne begins by getting right the fine detail of Thomas Heywood's letter appended to his Apology for Actors that explains that it was William Jaggard the printer's fault that a couple of Heywood's poems (from Heywood's Troia Britannica) were described as Shakespeare's in the 1612 edition of the miscellany The Passionate Pilgrim. Heywood thought that people might mistakenly assume that Shakespeare put his name to them because Heywood had tried to steal them, which he had not: they were not Shakespeare's in the first place. Since the title-page of The Passionate Pilgrim was reset, we may reasonably assume that Shakespeare got it changed and this presents us with a Shakespeare concerned about his literary property and about publication (p. 2). (Since Heywood clearly objected to the misattribution too, I cannot see why he could not be the agent of change.) Erne thinks that the Greene's Groatsworth attack on Shakespeare, alluding to 3 Henry 6, is about his temerity in outdoing Marlowe and Kyd (writers of two-parters) with a three-parter. Erne twice calls the Henry 6 series an "ambitious project" (pp. 2, 5) but of course the likeliest scenario is that it was a two-parter that grew into a three-parter, which shows not ambition but mere opportunism.

    Erne reminds us that 28 of the Sonnets are about "poetry as immortalization" (p. 5), although it is worth recalling (as Robert Wilcher does in a forthcoming book) that Sonnet 17 refers to "my papers" not pages, suggesting manuscript survival. The view about writing yourself into posterity that we get from the Sonnets is incompatible with the publication-indifferent Shakespeare we get elsewhere, and Erne asserts that in late-sixteenth century England it became a real possibility for a poet to enter posterity (p. 6). An increasing amount of literary work was printed and although it is true that printers, not authors, held legal copyright, it was nonetheless widely recognized that writers had a moral claim to their own works. Erne gives examples of authors complaining that the Stationers' Company monopoly of printing gave these men unjust power over authors' work, and indeed there were royal patents granted that allowed particular authors "to a derive a profit from the sale of their books" (p. 9). Of course, that is not quite the issue: a share in the profits is not the same as control over the material. Thomas Bodley, it is true, kept printed plays out of his library, but Erne insists that he was not typical in that, witness Sir John Harington, and Francis Bacon's grandson Sir Roger Townshend, and others, who took printed plays seriously. Certainly in the first years of the theatrical period (say, 1567 to 1589) plays were primarily for performance only, but thereafter they took on a dual life, just as Shakespeare got going in the early 1590s. The peak of play publication in the 20 years of 1594-1613 exceeded (indeed is double) the combined total for the 10 years before and the 10 years after this 'spike'. That is, it was not simply a matter of ongoing increase: those years were special (p. 15). What we would call literature was only about a quarter of all publishing in these years, but within that literary segment drama was fully a seventh. Drama's share today is much less. Erne notices that particular men--Thomas Creede, Edward Allde, and Valentine Simmes--were especially active in play printing (p. 16n50), which point might usefully be cross-referenced with Gary Taylor's plenary paper at the Lancaster conference (''Shakespeare's First Readers': Plenary Lecture Delivered at the Conference 'The New Shakespeare: A Writer and His Readers: The Return of the Author in Shakespeare Studies' at University of Lancaster, 3 July' [2004]), which made the point that publishers might over-compensate for their lowlife reputation by publishing uncommercial material. This provides a useful counterbalance to Peter W. M. Blayney's empirical demonstration that, in general, play printing was not lucrative ('The Publication of Playbooks', in John D. Cox and David Scott Kastan (ed.) A New History of Early English Drama (New York: Columbia University Press [1997])).

    Material form showed drama's place: printed plays adopted Roman typefaces before other genres did, and hence were "catering to an educated and progressive readership" (p. 17). The anthology England's Parnassus (1600) puts an extract illustrating 'care' from Edmund Spenser's The Faerie Queen (a clear piece of high culture) next to Friar Laurence on 'care', so Shakespeare was not thought of as low-culture in his own time. As an avid reader, Shakespeare must have seen this. Whereas freelance dramatists got their income primarily  from selling the play to the company, Shakespeare's income was primarily from box office (p. 19). And whereas the freelancers (we infer from Henslowe's Diary) wrote just enough to fulfil their commissions, Shakespeare habitually wrote more than was needed, because he could afford to write for the page as well (p. 20). This argument could usefully be nuanced with consideration of Tiffany Stern's work on dramatists' 'benefit days' that were an additional source of income beyond the flat fee for delivery of a playbook. Erne surveys the rise of performance-centered thinking in Shakespeare studies, which he dates as starting in the late 1970s, and dismisses the idea that we only 'overhear' Shakespeare via his scripts, the detritus of performance, as nonsense (pp. 20-25). The publishers, he points out, thought them coherent and whole enough to publish. Erne is explicitly against returning us to the New Critical position that drama is poetry, but he wants to push the pendulum back a little from the current obsession with performance only. The Romantics were not entirely wrong to say that Shakespeare is more suitable to be read than performed, but our view has been skewed by the fact that Shakespeare has been performed from the overlong reading texts that he produced rather than the short theatrical ones. Erne's view of Shakespeare's literariness has amongst its many merits the fact that it solves the long-standing mystery that his dramatic poetry whizzes by one in performance, seeming to demand close attention that it cannot, in the moment, be given.

    Usually Jonson is named as the first man to insist that plays were literature, either in his 1616 Workes, or in more tentative ways in the printings of Every Man Out of His Humour (1600) and Sejanus (1605) that mark a gap between the book and the performance text. For Erne, the gap between performance and print opened up at least as early as the 1590 printing of Tamburlaine and plays became literary artifacts part-way through Shakespeare's career and did so at the behest of publishers not playwrights (pp. 31-33). Pushing back by 10 years (1590 instead of 1600) the date when printed playbooks began to legitimize themselves by stressing their non-theatrical features and their authorial (as opposed to theatrical) origins is important for the case of Shakespeare, because it puts him inside this trend. Erne rightly objects to Jeffrey Masten's erroneous assertion that play title-pages in the late-sixteenth and early-seventeenth centuries generally did not mention authors, and insist rather there was a palpable tension between performance and reading marked in the title-pages. In the 1590s, naming the author became a way to detach printed plays from the taint of the playhouse, but of course collaborative dramatic composition was a hindrance to this detachment. Nearly 60% of the plays written for the Admiral's men from autumn 1597 to summer 1600 were collaborative, yet not a single one of the plays printed in that period mentions multiple authorship on its title-page: either a single author is named or no author is named. (Erne should be careful here of taking the Admiral's men as the norm of collaborative writing: perhaps unattached dramatists (from whom this company got its drama) were more likely to collaborate than the attached dramatists who supplied other companies.) Overall, from 1584 to 1623, fewer than 12% of title-pages that mentioned playwrights mentioned multiple playwrights. So, it seems that collaboratively-written drama is under-represented in publication, either by being not selected or by being printed without mention of the dramatists. Playbills, it seems from John Dryden's evidence, did not name the author until the end of the seventeenth century, while title-pages started to do so 100 years earlier, so the theatre and the printing industry were not treating authors the same way (pp. 43-44). In England's Parnassus, John Allot attributed collaboratively written material to one playwright only, as though writers were supposed, in their very natures, to be loners.

    Non-commercial plays (translations and Latin plays for example) were in the mid-sixteenth century published with their authors'/editors'/translators' names on the title-pages and even appeared in collections of 'Works', but the commercial theatre play printings remained mostly anonymous until the 1590s. However, by the 1610s it was the norm to name the dramatist, so during the span of Shakespeare's career drama went from almost always anonymous in print to almost always attributed (pp. 45-47). The Jonsonian distinction between great dramatic matter to be studied in print by the learned, and the dross that the actors threw in to please the multitude, is already apparent in the preface to the reader in the 1590s printing of Tamburlaine that "left out some fond and friuolous Iestures". It is not clear whether the material excluded was Marlowe's work or actors' interpolations, but the point is that already, by 1590, there was a sense of what suits the stage and what suits the page, and this sense is coming from a publisher not an author (pp. 48-49). Did publishers perhaps carve out the sense of 'dramatic author' that the authors then took up? As well as creating the 'author', publishing might well have helped stabilize genre indeterminacy: Shakespeare's Troilus and Cressida was variably called a comedy (Q-1609 address), a tragedy (F1), and a history (Q-1609 title-page). Likewise, the 1590 printing of Tamburlaine refers to the removal of comic bits and its Stationers' Register entry calls it "commical". Histories and tragedies seem to have been more respectable reading matter than comedies (certainly, Shakespeare's comedies were not reprinted half so often as his histories and tragedies), and thus our modern genre expectations (for example, that Tamburlaine is a tragedy) might themselves have been shaped by the publishers (p. 50-51). When Richard Jones published Tamburlaine there was not yet an established market for printings of plays from the commercial theatre, and he treaded carefully in how he addressed his readers. This care was emulated by others. When The Spanish Tragedy was printed, its multi-lingual inset play was turned into English for the benefit of readers (hence it is not a record of performance), apparently by Kyd who thus wrote the same thing twice, once for the stage and once for the page. Around the same time the same stationer (Edward White) published Kyd's full version of the inset play, Soliman and Perseda, and it is conceivable that the miniature version in The Spanish Tragedy was a taster for the full one (pp. 53-55).

    Turning to Shakespeare's printed output in particular, Erne counts that by 1600 he had written 20 plays of which 15 had been printed. (When the facts are stated baldly like that, it is easy to share Erne's impatience with the orthodoxy that Shakespeare was print-indifferent.) Erne's impatience is heightened by Douglas Brooks's repeated assertion (in From Playhouse to Printing House) that the 1600 quarto of 2 Henry 4 was the first printed play to name Shakespeare on its title-page, made as part of an argument that this fact is extraordinary since 1 Henry 4 caused such a fuss. For Brooks, Shakespeare's authorship was constructed (as Michel Foucault would say) in an act of transgression and as a response to the Oldcastle controversy. Of course, this is nonsense since earlier printings of Love's Labour's Lost, 1 Henry 4, Richard 3, and Richard 2 had all named Shakespeare. Brooks's mistake was to look only at first printings and ignore reprints. What is truly odd (almost unique, in fact) about the Shakespearian cases is that an anonymous edition gets replaced by one that names the author (p. 57-58). For Erne, 1598 is the year that Shakespeare gets invented as an author and around the turn of the century a lot of authors were invented in this way. That is, they cease to get anonymously published and start to get named on title-pages. Shakespeare leads the way. At this point in an extraordinarily detailed argument, Erne makes the first slip that I can find: "the Pavier quarto of 1 Contention was said to be 'Written by W. Shakespere, Gent" (p. 65). In fact it was called The Whole Contention and its title-page reads "William Shakespeare, Gent" (a different spelling and an unabbreviated first name).

    In 1598 Falstaff had just become a popular character and Shakespeare had become a gentleman, but Erne does not think these things account for his emergence as an author. Rather, what mattered was his being canonized by Francis Meres and put amongst the acknowledged great writers (pp. 66-67). This seems to be why reprints of Shakespeare suddenly, thereafter, name him where the first editions had not: now, post-Meres, Shakespeare's name sells. This might also be why The Passionate Pilgrim (1599), a collection with a few bits of Shakespeare in it, is attributed to Shakespeare on its title-page, and might also explain by 4 non-Shakespearian plays (Cromwell, London Prodigal, Puritan Widow, and Yorkshire Tragedy) were printed with his name on their title-pages. Thus ". . . the social cachet of printed playbooks increased well before the advent of Ben Jonson and the publication of his Workes in folio in 1616" (p. 71) Allot's selection for his collection England's Parnassus was strongly influenced by non-anonymous publication: most excerpts were from printed plays and of those most were from printings with named authors. Identifiable playwrights, then, already qualified for inclusion amongst "the choicest flowers of our modern poets" as the collection called them. Another compilation is A. M.'s [probably Anthony Munday's] Belvedere, or the Garden of the Muses (1600), which gives much shorter snippets (a couple of lines) and does not attribute them, although Charles Crawford was able to identify more than half of them by hand in the early twentieth century. The situation with Belvedere is like that with England's Parnassus: plays are given place amongst the literary, and Shakespeare most of all. So, we must review our standard history and stop saying that plays were, before the big dramatic folios, sub-literary, and we must stop thinking of the publishers as the enemies of the dramatists. To a considerable extent, the publishers made the dramatic authors. By 1600, Shakespeare had a substantial body of published work and must have expected that what he wrote next would also appear in print. The remarkable fact is that mostly it did not and only 5 more of his plays--Merry Wives of Windsor, Hamlet, King Lear, Troilus and Cressida, and Pericles--were printed in his lifetime.

    This fact structures Erne's next two chapters, which examine in detail Shakespeare's publication career up to (pp. 78-100) and after (pp. 101-14) the fulcrum year 1600. The gist of these chapters appeared as Erne's Shakespeare Quarterly article reviewed last year and need not detain us too long. Erne examines the first 12 plays that Shakespeare wrote for the Chamberlain's men. Romeo and Juliet, The Merry Wives of Windsor, Henry 5, and perhaps Love's Labour's Lost first appeared in 'bad' quartos. Perhaps because subsequent printings were 'good', we can say that players did care about the quality of their plays in print and intervened. For Q1 Romeo and Juliet John Danter got licence but not Stationers' Register entry before printing it, but by 1599 Cuthbert Burby seems to have acquire the rights to Romeo and Juliet and owned a 'good' manuscript that he went on to print. It is hard to be certain what happened, but if the company sold a good manuscript to Burby, they might well have done this before Q1 appeared. (True, but equally they might not have.) That the good manuscript underlying Q2 Hamlet changed hands before Q1 Hamlet appeared is evidenced by the fact that James Roberts who went on to print Q2 entered his copy in the Stationers' Register before Q1 was printed for Nicholas Ling and John Trundle. Presumably, Roberts, pointing out the unintentional breach of the publishers of Q1, resolved the potential dispute by selling Ling and Trundle his good manuscript and having them pay him to print it. Q1 Love's Labour's Lost calls itself "Newly corrected and augmented", implying the existence of an earlier, lost 'bad' quarto; but there are other cases where such claims about correcting and augmenting are demonstrably false and Paul Werstine showed that Q1 Love's Labour's Lost was probably set up from printed copy, in which case the lost Q0 was good too.

    So, it is likely that rather than responding to 'bad' publication, the company sale of manuscripts to publishers preceded the bad publication. The 'bad' Henry 5 and Merry Wives of Windsor were not superceded by good ones in Shakespeare's lifetime, but the latter did not sell well (no reprint until 1619) so the players would not have been able to get a good text printed if they wanted to, for the publisher would have wanted to shift his existing copies first. Here Erne repeats the jumping from list to list that marred his Shakespeare Quarterly article: "Of eight other plays Shakespeare is likely to have written for his company from 1594 until close to the turn of the century . . ." (p. 82) Love's Labour's Won might be the same as Much Ado About Nothing or might be another play since lost, and King John could not be printed because Troublesome Reign's publication blocked it. (It would be doing the reader a service if every time a list of plays were referred to, the items in the list were spelled out. A little wasted space would save a lot of readerly guesswork.) Of the remaining 6 plays--Richard 2, A Midsummer Night's Dream, The Merchant of Venice, 1 Henry 4, 2 Henry 4, and Much Ado About Nothing--we are now more sceptical than ever about our ability to tell the nature of the underlying copy from the printed text. Still, it remains quite possible that these plays were printed from good manuscripts supplied by the company. As a rule, 2 years seems to elapse between composition and Stationers' Register entry, although Erne has to dodge around a little amongst the critics for his dates of composition in order to make this all fit; there is a faint sense of shoehorning the evidence to fit the theory. To his credit, Erne admits the danger of circularity: some of the datings of plays are based on the idea that they would not get printed soon after composition because that would hurt performance income.

    The people involved in publishing the good texts of Shakespeare were a tightly-knit group of primarily just three men: the publisher of playbills James Roberts, plus Andrew Wise (with whom Roberts worked) and Cuthbert Burby. Why wait 2 years before selling the manuscript to the publisher? Perhaps to keep up the income from scribal dissemination (which was certainly more costly per book than print) to discerning aristocrats, or perhaps to get publicity for a revival (pp. 87-91). A few performances (including revivals) that we can date--Titus Andronicus June 1594, The Taming of the Shrew June 1594, A Knack to Know an Honest Man October 1594 to November 1596, and Massacre at Paris June-September 1594) coincide with printed texts being published. Of Shakespeare's pre-Chamberlain's men's plays the evidence is generally hazy and it is no surprise that a clear pattern cannot be determined. Although Erne does not explicitly make the connection, it follows from his argument that once the two main companies have what Andrew Gurr called 'settled practices' resulting from their state-enforced duopoly of London playing (The Shakespearian Playing Companies, pp. 78-104), settled practices in publication also emerged.

    Then, suddenly, Shakespeare playbook publishing fell off after 1600, with just 5 plays appearing from 1601 to 1616. The decline was in new editions and in reprints, and only 2 new plays, Hamlet and Pericles, went into a second edition. This is harder to explain than the fairly straightforward pattern of the pre-1600 years, which was clearly one of intended company publication (p. 100). One possible explanation is that there was a glut of playbooks around 1600-1 and supply outstripped demand; this would make publishers reluctant to take on new books. Another explanation would be competition from the newly re-formed Paul's and Blackfriar's boys: their plays got into print and hurt the adult players' publications. The Stationers' Register 'staying' order of Much Ado About Nothing and 2 Henry 4 (not Henry 5 as usually thought), As You Like It, and Every Man In His Humour, was not an attempt to block publication but rather an acknowledgement that the licence lacked ecclesiastical authorization; otherwise, how come 3 of the 4 plays were regularly entered within 20 days and published the same or the next year? Looked at this way, As You Like It failed to get published, rather than was blocked (p. 103). Dutton suggested that the prefatory address to the Troilus and Cressida quarto about the play never being performed refers to the readerly, long text of that edition, just as Q2 Hamlet boasts of being longer and new in the sense of being newly available in this long version. James Roberts entered Troilus and Cressida in the Stationers' Register on 7 February 1603, the entry recording that he still needed "sufficient aucthority" (that is, ecclesiastical approval). This does not mean that there was anything wrong: Roberts entered several plays that had not been allowed, but as he seems to have been trading in manuscripts (entering them and then selling his rights to let someone else publish them) he did not need to pay the 10 pence to have the plays allowed, since he was not going to publish them. So, we do not have to imagine two different Troilus and Cressida manuscripts, an acting version that was not allowed and a reading version that was.

    Tracing Stationers' Register entry of Shakespeare plays, there is a big gap from Troilus and Cressida on 7 February 1603 to King Lear on 26 November 1607 (after a steady flow before then), and King Lear is the most badly printed of the 'good' quartos. This Erne admits presents a problem for his argument that Shakespeare cared about its publication (p. 107). The same is also true of the next-but-one Stationers' Register entry in respect of Shakespeare: Pericles on 20 May 1608 by Edward Blount, which was followed in 1609 by Henry Gosson's publication of the 'bad' quarto without a transfer of rights from Blount to Gosson. That Gosson was not contravening Blount's rights seems indicated by the fact that he printed a second edition in 1609. Erne rather cryptically here asserts that if Gosson got his manuscript and his rights from Blount, it is unlikely that the company were involved. (Why? Because Pericles is textually so bad? If so, that stands as an objection no matter what Gosson or Blount did.) If the King Lear, Antony and Cleopatra and Pericles entries in the Stationers' Register did originate in the playing company passing their manuscripts to publishers, then there was at least a trickle of the old flood after 1600. But since we cannot assume that, we must see what else might account for the Stationers' Register entry gaps from 1603 to 1607. Plague closure of the theatre would have hurt everyone, and it is only Shakespeare's printed output that dwindles. Erne has no certain answer, but wonders if there were a collected works of Shakespeare being planned, or perhaps the company chose to limit their print output in order to favour patrons, especially William Herbert, who probably got them royal patronage, probably is the dedicatee of Sonnets, and certainly is, with his brother, joint dedicatee of the 1623 Folio and almost certainly stopped Pavier's collected works in 1619. (In an article to be reviewed here next year, Gary Taylor argues that Shakespeare's dramatic powers failed him and he simply did not have a hit until Pericles; that would explain the gap)

    The history of Shakespeare printing duly (and largely convincingly) rewritten, Erne moves to the consequences of his ideas. Regarding the players' alleged opposition to print (pp. 115-28) Erne can build on existing knowledge with which his narrative is compatible. That printing did not lead to other companies playing a company's play is indicated by companies paying for new plays on subjects about which printed plays already existed: the Master of the Revels's licence did not allow just anybody to perform a play, only the company that licensed it. There is no evidence--just the tradition started by A. W. Pollard--that printing a play reduced its box office. Contrary to the usual explanation that players sometimes sold their plays to publishers when they were hard up, the amount given by publishers (Blayney reckons 30 shillings) was trivial when compared with the cost of costumes; it were better to sell those to raise cash. Richard Brome's deal with Queen Henrietta's men, come to court in 1640, prevented him from getting his plays published. G. E. Bentley extrapolated from this that Shakespeare was bound by the same rule, but in fact Brome's contract prevented him only from printing his plays without company agreement, which suggests that in fact companies did support publication. Heywood's address to the reader in The English Traveller seems to imply that actors were against publication and that he has no such ambitions, but needs to be read in the context of his frustration that his Age plays did not come out as a collection, as he had been promised, so he was making a virtue of non-publication necessity. Likewise, the Articles of Agreement amongst the Whitefriars sharers seem to prohibit publication, but only by individuals: collectively the company may print its plays. Moreover, the articles most particularly protect the 'playbook' (the copy with the all-important performance licence) currently being used, rather than any copy of any of their plays. All the mistakes about publication of plays in Shakespeare's time stem, as Blayney pointed out, from the failure to look at it from the publisher's point of view. Once we get that straight, we can deal with such questions as why Shakespeare wrote overlong plays, how performance relates to published text, what is a 'socialized' text in this context, and what lies behind the 'bad' quartos. These matters occupy the rest of Erne's book.

    Erne detects a neat irony in performance-centered study of early modern drama: the very fact that we have the plays relies on someone not thinking that the scripts existed only to be performed (p. 131). Since we have only the printings, its delusional to think we can achieve performance-centered criticism in any meaningful sense. (Here Erne makes another of his rare errors, referring to "Elizabethan groundlings standing in the pit of the Globe" p. 136; he means of course standing in the yard.). We can, however, detect a wide gap between performance and printing in the excessive length of many Shakespeare printed plays; even a company specializing in speedily delivery such as the Shenandoah Shakespeare Express cannot get through Hamlet or Richard 3 in under 3 hours. (This reviewer's telephone call to Barrie Rutter of the Northern Broadsides company revealed that with the cutting of about one-sixth of the Arden 2 Richard 3, about 600 lines, the remaining text of 3,400 lines was routinely performed in under 2½ hours. There may well be truth in Rutter's claim that the southern regional accent draws out Shakespeare's lines and makes them tedious.) It is indeed hard to see why playwrights would have written hundreds of lines that could not be performed, and Alfred Hart's counts in the 1930s showed that Shakespeare and Jonson were unusual in writing such long plays as they did. Erne twice (pp. 139, 144) refers to the belief that abridgement was done for provincial touring while full texts were performed in London, without commenting on whether he accepts this idea about provincial touring and without noting Scott McMillin's demonstration that abridgement often increases rather than reduces the number of actors needed by reducing the opportunities for doubling. Shakespeare's comedies average around 2,500 lines, which would be performable in just a few minutes over 2 hours, while the histories and tragedies are markedly longer, averaging around 3,000 lines. The most heavyweight plays (in subject matter and sources) are also the longest: Hamlet, Richard 3, Troilus and Cressida, and Coriolanus. Contemporary accounts also seem to suggest that comedies were thought lightweight and not much worth reading, while tragedies had readerly gravity.

    Erne notes that the printings of Webster's The Duchess of Malfi (1623) and of Brome's The Antipodes (1640) explicitly refer to the printing being longer than what was performed. (True, but the latter says that the printing included extra material from "the allowed original" meaning the licensed playbook, so that is still a theatrical rather than a readerly origin.) Of course, with the introduction of musical act intervals performance took longer for the same total number of lines, so line counts might go down if performance duration were to be kept the same. Erne makes much of the strongest piece of evidence of cutting-for-performance: Humphrey Moseley's publisher's address to the reader in the 1647 Beaumont and Fletcher folio. Moseley claims that whole scenes were cut from the plays for performance and that the private transcripts that circulated were of the cut texts, so the folio is the first opportunity to see the uncut versions. The plays in the folio are not particularly long, averaging about 15% fewer lines than those in the Shakespeare folio; so, if these shortish Beaumont and Fletcher plays had to be cut for performance, much more cutting would have been needed to perform Shakespeare plays. Moseley's claim about cutting was rejected by W. W. Greg but is borne out by the fact that the plays Moseley did not initially enter in the Stationers' Register (presumably because he did not have manuscript copies) are indeed much shorter than the others, perhaps because printed from those private, post-cutting transcripts (p. 153). In particular, the private transcript of The Woman's Prize seems to give the cut version of what the folio has in full, albeit minus what the censor removed in 1633. Likewise the manuscript from which the folio's Beggar's Bush is printed seems to have had restored to it some things that were originally cut for performance, to judge from repetitions and metrical lines split in two. Erne's own counts of lines in manuscript playbooks from 1576 to 1642 confirm the general picture that plays of under 2,500 lines are left virtually uncut and those with substantially more than that number are cut down to about that number (pp. 158-64). The only-indirectly applicable evidence of the post-Restoration stage points the same way: the long Shakespeare plays were not performed in their entirety.

    What are the implications of all this for editorial policy? At the height of New Bibliography, Erne's conclusion that plays were cut for performance would not have mattered much for editors because their aim was to recover the play as it would have stood in the author's manuscript as he handed it over to the players. However, for the stage-centered new New Bibliography (most obviously manifested in the 1986 Oxford Complete Works) that tries to recover the play as it was first performed, routine cutting for performance is devastating (p. 175). The underlying rationale for new New Bibliography that Shakespeare's intentions extended only so far as performance (and hence that the performed text is primary) has been buttressed by the idea that he was indifferent to print; this latter has fallen in the first half of this book, so the former is vulnerable. In other words, perhaps Shakespeare's ideals for his works were fully realized by him in the print versions. From this point of view, Greg's seeking after the authorial manuscript before the players got hold of it is a less distorting ideal than the Oxford Complete Works's seeking after the first performance. Gurr's New Cambridge Shakespeare edition of Henry 5 argued that Q1 shows us that the choruses were omitted in performance, so an edition done along the principles of new New Bibliography would have to omit them.

    Folio Hamlet might well reflect some of the cuts made to make the play performable: it has absences in common with Q1. On a couple of occasions, F omits some things that are in Q2 and at that point where the extra material is in Q2 there is a lost half line: presumably there was a mark for a theatrical cut in the manuscript underlying Q2 that the compositor took for a half-line deletion. Because printers printed everything in their copy, ". . . no printed text allows us to recover how much would have been marked for omission in its copy-text and, consequently, would have been cut in performance" (pp. 180-81). How come Folio Hamlet contains some of the cuts (from Q2) that Q1 shows, but not all? Because it is based on a preliminary abridgement. The history of The Honest Man's Fortune by Fletcher, Massinger, and Field provides a parallel: the play was 'lost' in 1625 so the censor Henry Herbert relicensed a transcript of the author's foul papers for performance. The play was printed in the Beaumont and Fletcher folio of 1647 from the author's foul papers. Thus, the folio shows the play-as-written before cuts and whatever the folio has that the manuscript has not (primarily one scene, 5.3, and a different version of another, 5.4) are things that the players cut or changed before sending the play for relicensing. However, the manuscript also has marginal bars showing further cutting, so is itself another example of a "preliminary abridgement" because already reduced from its copy and marking more reduction to be made, just like Folio Hamlet. It follows from this that, since there was cutting before licensing, Gurr's Maximal/Minimal textual model--that they licensed the most text they could ('Maximal and Minimal Texts: Shakespeare V. the Globe', ShSurv 52 [1999].68-87)--is not corroborated (p. 182n31). One might well respond to this that the circumstances of Honest Man's Fortune, with the licensed playbook being lost, are scarcely typical.

    In this view, neither Q2 nor F Hamlet, nor editions based on them, show us the play as performed. Equally, Q1 Othello's being 160 lines shorter than F might be explained by its deriving from a preliminary abridgement, since 160 lines is not enough to bring it down to the performable length. The argument that the revision of King Lear caused the loss of the mock-trial scene simply because it was artistically ineffective is strained, and we would have to factor in the certainty that the play as represented in Q1 is unperformably long. The idea that Q1 Richard 3 was produced by collective, company memorial reconstruction while on tour (because they lost their licensed playbook) is inherently implausible: if you know your parts (which is what a memorial reconstruction presupposes) there is no point recreating the playbook when what you really need is the licence. Also, at 3,400 lines, Q1 Richard 3 is too long to represent performance. Here Erne might useful have acknowledged John Jowett's brilliant demonstration, by entirely independent means, that Q1 Richard 3 cannot be a memorial reconstruction (''Derby', 'Stanley', and Memorial Reconstruction in Quarto Richard III', N&Q 245 [2000].75-9). Macbeth is short (about 2,000 lines) and if it does indeed, as many suspect, derive from a posthumous theatrical adaptation, we should have to accept that in performance the other long tragedies might have lost more than about a third of their lines too. Concluding this section, Erne asserts that once we admit revision as an agent that separates versions that have competing authority, we have accepted an artistically self-conscious Shakespeare and we should go the extra mile to accept that not everything he did was driven by the exigencies of live performance (p. 189).

    There follows a chapter on the origins of the so-called 'bad' quartos for which there are also overlong 'good' companion texts (Romeo and Juliet, Henry 5, and Hamlet), which chapter adds little that is new to the debate and largely echoes the best of the work of the New Textualists (pp. 192-219). Erne's intention is not to show that the 'bad' quartos are really 'good' (p. 194) but that they do nonetheless yield important evidence about performance. McMillin's point about the absurdity of cutting for a smaller cast (because in fact cutting often removes opportunities for doubling) finally appears here (p. 207). Kathleen O. Irace pointed out that certain alleged memorial reconstruction texts retain characters' parts disproportionately--as in Q1 Henry 5 retaining 84% of Exeter's part as represented in F, although Q1 is only half the length of F overall, and Q1 Hamlet retaining 92% of Marcellus's part, although it is only 60% as long as F--and she thought this incompatible with the claim that the actors were remembering a text already abridged for provincial performance. After all, would not Exeter's and Marcellus's parts have been cut with something approaching the severity that the entire play was cut? Erne thinks not: Marcellus's lines are largely unscathed in Q1 Hamlet because they are necessary to the plot (p. 208). Taking F Henry 5's part for Exeter and cutting it by 16% (or rather, taking the combined reporter's parts for Exeter, Pistol, and Gower, which overall are cut by 20%) is a process we can imagine being done to the whole play, and if the choruses were entirely removed too (as Gurr thinks happened in performance) this would put the play that the reporters were trying to reconstruct at around 2,500 lines. The point that Erne is making (and it is not easy to keep sight of it in all this detail) is that the memorial reconstruction was indeed aiming to reproduce a text that had indeed been cut (relative to F in each case), but not drastically for regional performance but by the usual amount necessary to make a play performable. The short quartos are performance texts, and their long companions are essentially literary works not for performance.

    Pursuing this claim about literariness further, Erne attempts to show that the short quartos are speakerly and rely on performance showing things to the audience while the long texts are readerly and have things described. Thus Q2 Romeo and Juliet has Juliet say that she is kneeling to beseech her father while Q1 simply has a stage direction for it. Rather illogically Erne runs the same kind of evidence the other way too: Q2 Hamlet (but not Q1) gives a stage direction for the cock crowing that startles the ghost in the first scene (pp. 222-24). One might well respond to this by pointing out that stage directions can be readerly and theatrical at once: after all, someone in the theatre needs to know that a cock crowing noise is to be made. Erne surveys other passages in the long texts (absent in the short ones) that seem gauged to please a reader and, although performable, would probably be the first to go when cutting for length. For Erne, Friar Laurence's 'grey-eyed morn' speech in Q2 Romeo and Juliet is "purple"--surely that is too strong a word--and appears in two places because Shakespeare was planning where to drop this detachable, literary bit of anthologizable writing. Risking a charge of ethnocentricity, Erne quotes Walter Ong using E. M. Forster's notion of 'rounded' character to argue that oral culture cannot produce characters that challenge our expectations, only ones that fulfil them. (By this logic, we would have to accept the implausible notion that the complexity of Odysseus's character in The Odyssey was something absent in all the oral retellings and that emerged only once it got written down.) Erne is quite serious on this concluding point, and suggests that criticism based on a better understanding of "the cultural contingency of characterization" might take its proper place in Shakespeare studies (p. 243).

    Erne provides three appendices. The first ("The plays of Shakespeare and his contemporaries in print, 1584-1623") provides useful tables of all the first editions of all the plays (abbreviated by using Greg's numbering in the Bibliography of English Printed Drama). These are marred by attempts at typographical distinctions--such as solid or dashed underlining to show how firmly Shakespeare's authorship is advertized on the title-page--that in the review copy were almost impossible to see. In the second appendix Erne ponders whether the "stolen and surreptitious" copies of Shakespeare plays that the 1623 Folio preliminaries refer to could have been Thomas Pavier's quartos of 1619; they might but Erne has no new evidence to offer. In the third appendix Erne gives reasons to believe in the circulation of dramatic manuscripts of successful plays well before the phenomenon is referred to in the Beaumont and Fletcher folio of 1647. Greg observed that we have no surviving examples before 1624, but Erne notes that the publisher's epistle to the 1619 printing of Beaumont and Fletcher's A King and No King refers to Sir Henry Nevill having one, and the printer's address to the 1620 printing of The Two Merry Milk-Maids refers to "false Copies" travelling abroad. Also, Gabriel Harvey's reference to Hamlet, written into his copy of Chaucer, sounds like a reference to a reading copy, yet is pre-Essex's execution and hence pre-Q1 Hamlet.

    2003 certainly was a year for landmark monographs, and Andrew Murphy's extraordinary Shakespeare in Print: A History and Chronology of Shakespeare Publishing will remain the standard single-volume reference work for a long time. Murphy covers the entire timespan from first printings to the present, and geographically he covers English, Scottish, Irish, and American editions, but thankfully for our purposes he gives equal space to each period covered so that the material relevant to this review is only a small part of the whole. Murphy's introduction attends to the matter of balance and observes that some editions that are not distinguished as textual scholarship are nonetheless important for their place in the struggle over Shakespeare as intellectual property, or for sheer numbers sold (pp. 6-7). To complement his scrupulous scholarship Murphy has a fine feeling for a good anecdote, as shown in his account of the publishing house Macmillan's employment of an imperfect mechanical process to copy their outgoing correspondence. The images turned out to be impermanent, and Murphy has us picture him in the British Library receiving to his desk bound volume after bound volume of carefully stored and respectfully handled material, the pages of which were, upon opening, entirely blank (p. 11).

    Murphy's first chapter ("The early quartos") is a fine work of print history but adds nothing to our knowledge of matters textual. As with Erne's book reviewed above, Murphy seldom makes a slip, but I am sure that when he twice uses the formula "X published Y for Z" he really means "X printed Y for Z" (pp. 24-25). If not, I do not understand what 'publishing for' someone might mean. In his second chapter ("Early collected editions") Murphy gives a potted history of the Pavier quartos, with a useful summary table, and likewise for F1, F2, F3, F4, and the putative F5. Chapters 3 and 4 cover the Tonson era from Nicholas Rowe to Edmond Malone and are helpful to have to hand when examining one of the many multi-volume Shakespeare editions produced in the eighteenth century (often republished with alterations) and trying to work out what it is. Murphy credits Samuel Johnson as the first to articulate the principle that derivative texts must be given lower authority than the texts they derive from, from which they "only deviate . . . by the printer's negligence" (p. 82), and he, correctly in my view, demurs from Margreta De Grazia's Foucauldian view that Malone's edition was a whole new way of doing things. Following Simon Jarvis, Murphy prefers to see it within the developing tradition (p. 97). In chapters 5 and 6 Murphy breaks off the historical narrative to consider the matter of copyright disputes amongst English, Scottish, and Irish publishers that emerged because of the legal differences across the (only fictionally 'united') kingdom. Murphy describes the eighteenth-century price war between Tonson and the publisher Robert Walker, who between them flooded the market with cheap editions (4 pence per play, and under) and drove a boom in Shakespeare appreciation generally (pp. 107-10). There was some confusion about whether English law applied in Scotland after the union of 1707, and it certainly did not apply in Ireland, so the copyright situation in those places was unclear in a way that favoured printing and exporting to England books previously printed in England. These editions' original English publishers believed them to be subject to perpetual, rather than time-limited, copyright, and after a test case of 1774 rejected this interpretation, Shakespeare printing doubled in rate.

    Murphy's chapter 7 on American editions is especially interesting on the genesis and development of the Furness Variorum, and his surveys of nineteenth-century popular (chapter 8) and scholarly (chapter 9) editions are exemplary condensations of his extensive work in archives, but fall outside our scope here. In chapter 10 ("The New Bibliography") Murphy covers the faltering Oxford Shakespeare (under R. B. McKerrow and then Alice Walker) and the rival ('New') Cambridge Shakespeare (under Arthur Quiller-Couch and John Dover Wilson). Murphy has not much to say about the principles of New Bibliography itself, although he makes the important observation that Wilson's series was the first to be completed (albeit very late, in 1966) along New Bibliographical lines. Discussing the later twentieth century, Murphy too cautiously (as we shall shortly see) remarks that Pericles "may have been co-authored" (p. 250). In his conclusion ("Twenty-first century Shakespeares") Murphy repeats the familiar line about hypertext as a new configuration that decentres the author because of the capacity for movement from one place to another. Actually, a print library itself can be seen as a hypertext with the individual works' footnotes as the links; electronic media just make the jumps from text to text that we all do anyway a little less time consuming. To my mind, focussing on the links misses the truly important differences between paper and etext, which have to do with alterability (in postmodern terminology, 'textual stability') and copyability.

    Murphy ends with an anecdote about printed Shakespeare crossing class boundaries and with an implied prediction that print will survive a long time for certain kinds of readers. Murphy contrasts such a printed book with the electronic text that is "dispersed" and not so much for reading as "surfing" (p. 275). Such terminology taints the medium, etext, by association with allegedly mindless recreations (surfing television channels, surfing the Internet) and sounds rather like the pre-web pseudo-theoretical mistakings of the nature of etext, which could see no further than the 'magic' of hypertext. Quite possibly the relative advantages of the printed book will soon disappear with the mass production of e-paper: illuminated or reflective electronic displays capable of showing as much detail as paper and consuming power only when changing the image and flexible enough to be folded (or rolled into a tube) when not in use. About a third of Murphy's book is an appendix--a substantial work of scholarship in itself, that lists all the major Shakespeare editions to date and provides multiple indices sorted on such fields as 'publisher name' and 'editor name'. In a revealing footnote (p. 412n4) Murphy cites Erne's demurral from the general consensus that Shakespeare was indifferent to publication, which Murphy announces is forthcoming in a book called Lines to Time: Shakespeare and Literary Drama. This must have been an earlier title for Erne's book Shakespeare as Literary Dramatist reviewed above, and it is worth knowing that, like Wilcher, Erne alighted on Sonnet 18 ("Shall I compare thee to a summer's day?") for its exploration of the potential for immortality through art.

    MacDonald P. Jackson's Defining Shakespeare: Pericles as Test Case is rather like Brian Vickers's Shakespeare, Co-author reviewed last year in its subject matter and its deprecation of certain aspects of modern theory's application to dramatic authorship. More focussed than Vickers's book, however, Jackson's aims for one target and hits its squarely: to show that George Wilkins wrote Pericles with Shakespeare. Indeed, Jackson hits his mark many times over from multiple new angles, taking the demonstration of his correctness far beyond the limits of anyone's reasonable scepticism. Unusually for our subject, we may truly say that the case is now closed. Jackson begins by quoting Jeffrey Masten's Textual Intercourse to disagree with the view that when collaborating, dramatists submerged their personal distinctiveness (p. 6), and instead cites Vickers's book on collaboration being done "by acts or scenes or other large units dominated by differing plot lines or sets of characters". As noted last year, that was one of the weakest points of Vickers's book since there is not as much evidence to support the claim as Vickers's rhetoric suggests. Happily, however, Jackson adds to the evidence that collaboration was not done at the level of the line but something larger by showing Middleton's preference for the shortening I've and Rowley's absolute avoidance of it, across many sole-authored plays "with their diverse textual histories". When they collaborate on a play, each section (be it Middleton's or Rowley's, determined on other grounds) betrays its author's particular preference for or against I've (p. 7). Jackson makes the important point that every act of 'disintegration' (say, splitting off a part of a play from Shakespeare) is simultaneously an act of 'integration' (say, giving that part to the rest of the Middleton canon). This is worth saying because there is a current mistaken orthodoxy (articulated by Masten amongst others) that an author-based approach to drama is anachronistically post-Enlightenment. In fact, Jackson (like Vickers) points out, the ancient Greek dramatic authorship contests, and the pre-Christian canonizing of Aeschylus, Sophocles, and Euripides, show otherwise (p. 12).

    The ground cleared of theoretical objections, Jackson turns to the necessary preliminary matter of statistics, stylishly addressed via the opening scene of Tom Stoppard's Rosencrantz and Guildenstern are Dead concerning the strangeness of a tossed coin coming up heads 92 times in a row (p. 23). Everyone uses statistics, asserts Jackson, even those who attribute something on the basis of verbal parallels with a work of known authorship, because there the implied reasoning is that chance could not produce those parallels with work by a different known author. In truth, chance often can produce the unexpected and Jackson promises that when verbal parallels are used in his argument there will be also "a search for similarities with other playwrights" and that statistics will be used to test what chance can do (p. 22). Jackson thinks the clincher for Wilkins's prose narrative Painful Adventures being based on the play is that it has Lychorida bring the newborn Marina up on deck to show Pericles: this utterly implausible event (Marina is just a few minutes old) happens in the play because Pericles cannot leave the deck without leaving the Globe stage bare. Had Wilkins not been copying the performance, he would have written the scene differently. (I would not have thought plausibility a reasonable criterion in relation to this story and would also object that Marina's association with the sea cannot be formed in a scene set below decks. Indeed, it is not dramatic practice that prevents Pericles going below deck: Shakespeare could have written a new scene or written the existing scene differently; he did not have to have Lychorida come on deck.)

    Jackson traces Ernest Schanzer's and then Gary Taylor's demolition of Philip Edwards's claim that the whole of the quarto is by Shakespeare, but that acts 1-2 had a different (and worse) reporter than acts 3-5. Were this so, Wilkins's Painful Adventures and the Pericles quarto would be independent witnesses to the play as performed, and hence where they agree they must be right since chance could hardly make them agree in error. And yes, they agree on 5 lines in 2.3 that are too bad to be Shakespeare and are typical of the first half of the play. The obvious inference is that the offending lines are by Wilkins, which is why his novel has them right (pp. 27-28). Also, if acts 3-5 of the Pericles quarto are simply a more-accurately reported account of the play as performed than acts 1-2 were, these last three acts should be more similar to the corresponding parts of Wilkins's novel than the first two acts were, since in Edwards's theory Wilkins's novel is just a uniformly good/bad account of the performance. In fact, acts 3-5 of the Pericles quarto diverge from rather than converge upon the novel (p. 29). More speculatively, Jackson insists that if Pericles had been Shakespeare's solo work, Heminges and Condell would not have left it out of F1; the only other play they left out was the undisputedly collaborative The Two Noble Kinsmen. Pericles was in the company repertory around 1623, so they had a playbook of it (hence availability of copy was not the problem), and Blount (one of the F1 publishers) had a Stationers' Register entry proving his copyright priority over Gosson; thus Blount could have printed the play if he had wanted. Indeed, claims Jackson, printing the play would have established a text "which would have differed sufficiently from Gosson's quarto to be exempt from any copyright claim Gosson's successors might have made" (p. 31). This is not right: stationers were protected from another stationer publishing the same story even if it were from a different text, as Blayney pointed out in relation to King Lear and King Leir ('The Publication of Playbooks', in John D. Cox and David Scott Kastan (ed.) A New History of Early English Drama (New York: Columbia University Press [1997])). From Blayney's view it seems to follow that Blount had effectively conceded the right to print Pericles, or even privately agreed to it, by not objecting to Gosson's 1609 quarto, for as Blayney showed it was printing that firmly established ownership of copy. Jackson claims that the 'problem' with Troilus and Cressida in F1 seems to have been overcome "in this way" (that is, by printing a good text) and cites the Oxford Textual Companion page 425 (p. 31), but in fact the Companion at that point makes no mention of how "difficulties over copyright" in respect of Folio Troilus and Cressida were overcome, only that we can presume they existed because F1's printing of the play was interrupted. Jackson is assuming that getting a manuscript to supplement the Troilus and Cressida quarto--to substantially alter its readings--was how the F1 publishers got around the copyright problem, but that is not what Taylor is arguing at the place cited.

    Finally, the idea that Pericles acts 1-2 are early Shakespeare material that he reused at the end of his career can be dismissed because whereas everything else he wrote can, by certain independent tests, be shown to belong to a particular phase in his career, these two acts have some kinds of stylistic links with early Shakespeare, some other kinds of links with mid-career Shakespeare, and other kinds of links still with late Shakespeare. (Actually, that is what I would expect to find if it were reworked juvenilia--a mix of old and new characteristics--but Jackson clearly means to imply that it belongs to no definite stage of Shakespeare's career because it does not belong to Shakespeare at all.) To refute the late Eric Sams's claim that Pericles is, in part at least, a Shakespeare play from the 1580s, Jackson shows that the alleged early allusions to it are weak or simply mistaken, and useful lists all the clear allusions to it that cluster after 1609, appearing in the plays Pimlico, or Run Red-Cap, Robert Tailor's The Hog Hath Lost His Pearl, and (via a plot echo) in Robert Armin's The Two Maids of Moreclacke (pp. 34-39). Perhaps seeing Gower's tomb in St Saviour's church during the burial of his brother Edmund on 31 December 1607 gave Shakespeare the idea for the play, Jackson wonders. Having established that the sole-authorship claim cannot stand, Jackson turns to the particular evidence for dual authorship, which is where the hard matter of this book begins. 

    The key to Eliot Slater's work in this field is rare-word usage, which is also what interested the most famous stylometrician Donald Foster and for which he was soundly excoriated in Vickers's other book reviewed last year, 'Counterfeiting' Shakespeare. The principle followed by Slater and Jackson is that ". . . works written by the same author at about the same time are apt to have more of their low-frequency words in common than works whose dates of composition are separated by many years" (p. 40). Low-frequency means that the writer concerned does not use them often, and hence they are for him or her (but not necessarily anyone else) 'rare' words. Slater's precise criterion was words used in at least 2 plays and a total of less than 10 times overall in the canon, and he counted how many such words should be common between play x and other plays in the canon on the basis of expectation derived simply from play x's total vocabulary. Jackson reckons that Slater should have derived his expectation from the proportion of rare words in x compared to the total number of rare words in the canon. This notion of 'expectation' is the ground upon which stylometricians start to leave the rest of literary scholarship behind, primarily because the stylometricians can put a number on it. The numbers generally refer to how unlikely a given event is, a matter about which non-specialists are apt to be wildly mistaken in their assumptions. Unless they are extremely careful with phrasing, stylometricians tend to make claims that non-specialists find either wholly persuasive or absurd. Here (p. 41) Jackson writes of the link between The Tempest and acts 3-5 of Pericles in the form of rare words that "the possibility of this discrepancy [between expected links and found links] being due to chance is infinitesimal". In fact he means that, were random chance all that connected rare word choice in The Tempest and Pericles acts 3-5--the whole of Shakespeare's career considered as a single word-pool with no chronological forces shaping his selection of words--then the likelihood of this high linkage (way above 'expectation') occurring by chance alone were infinitesimal. A little logic, however, shows that this does not necessarily link The Tempest and Pericles 3-5 directly by shared authorship. Imagine a world in which all dramatists were choosing their rare words according to the seasons (words beginning with the letters A-B in January, C-D in February, and so on) then the same results might occur if The Tempest and Pericles 3-5 were written at the same time of the year. The high-sounding mathematics ("infinitesimal") does not tell us that we are onto the right connection between The Tempest and Pericles acts 3-5, only that the chance of there being no causal connection is small.

    In a footnote (p. 41n2) Jackson explains his method for linking The Tempest and Pericles 3-5. The Tempest contains about 2.4% of the rare words in all Shakespeare and Pericles acts 3-5 contains 1228 Shakespearian rare words itself, so we would expect about 30 of these words (2.4% of 1228) to be in common, that is to be the same words, were a writer's changing habits over time not a factor. But a writer's changing habits over time are a factor--plays written about the same time tend to share rare-words because the writer is favouring those words at that moment--and indeed The Tempest and Pericles acts 3-5 have 62 rare words in common, more than twice what we would otherwise expect. Jackson uses a procedure called chi-square to give a sense of how big a difference from expectation this is (and explains it well) and cautions that it is not really an index of probability (a comparison with random chance) but of the alleged association (of rare words with phases in a writer's career) being tested. That is, a high chi-square indicates that two variables are unlikely to be randomly associated and likely to be causally linked somehow, but it does not (despite some popular misuse) tell you how likely something is. Rather, it tells you how unlikely the result you got would be if the variables were linked only by random chance, which is to say it tells you how infrequently chance alone will produce the result obtained. In Great Britain a one-in-14-million unlikelihood happens to someone about every other week in the National Lottery: the chance of a particular person winning is tiny, but the chance of someone winning is about 1 in 2. The likelihood that the phenomenon 'winning' will emerge from these events needs to be closely defined with qualifiers before we can put numbers on it. This objection to terminological slippage applies throughout Jackson's book: he repeatedly makes claims of the kind 'the odds of this being by chance are 1 in 100' (which sounds like 'it almost certainly is not chance') which strictly means the same as, but has different rhetorical force from, 'by chance this will happen 1 time in a 100 anyway' (so, eventually chance alone does it).

    Returning to Jackson's main argument, it emerges that if the same rare word tests are repeated with Pericles acts 1-2, the association with the last phase of Shakespeare's career disappears. One can also do the same tests for the absence of rare-word linkages between plays since chance operates equally on the non-selection of the same rare words in two plays. I would have thought this no more illuminating than the test for present links, since non-selection is selection's mathematical complement, but Jackson seems to think it highly significant. At this point, I suspect most non-mathematical readers will become lost. Jackson admits that Pericles 1-2 has above-expectation links with Antony and Cleopatra and Coriolanus but rejects the obvious conclusion that his makes Pericles 1-2 likely to be Shakespeare's too with the assertion that this can happen by verbal osmosis from one's collaborator. That is, when working on Antony and Cleopatra and Coriolanus the rare words of the other man's work in Pericles were in Shakespeare's mind. I cannot see how this can be accepted without it diminishing the significance of the claimed links of Pericles 3-5 with other plays: might not they too be explained this way? Jackson anticipates this objection by pointing out that Pericles acts 3-5's departures from random chance are way above those of Pericles acts 1-2, and indeed taking the Shakespeare canon together, the links of Pericles acts 1-2 are within what random chance would be expected to produce (p. 42). Better still, taking the canon in 4 sections, Pericles acts 3-5 has a strong association with the plays from King Lear to All is True and against the 3 earlier sections, while Pericles acts 1-2 has no significantly strong associations with any period. The pattern looks much the same if one considers the even-more-rare words, those occurring 2 to 6 times rather than 2 to 10 times (p. 43), although there emerges a hint of connection between Pericles acts 1-2 and Titus Andronicus and 1 Henry 6. Jackson does not think that this supports the idea of Pericles acts 1-2 being early Shakespeare because the evidential base becomes so small--in that case (I would say) he should draw no conclusions from it either way--and in any case Titus Andronicus and 1 Henry 6 were probably collaborative and in any case early Shakespeare is less idiosyncratic (uses more common words) so will have more links with what other men do than later Shakespeare will (p. 44). (Here Jackson sails closest to special pleading for his case.)

    Jackson 'proves' the insignificance of Pericles acts 1-2 having links with early Shakespeare by showing that a sole-authored Wilkins play, The Miseries of Enforced Marriage, or at least a random sample of its rare words, has links with early Shakespeare too. Jackson here brings in the fact that Pericles acts 3-5 has fewer than 'expected' rare word links with 1 Henry 6 and Titus Andronicus, but by this point the exhausted reader has surely forgotten that 'expected' means 'were only chance operating' and that in fact Jackson believes that changing authorial preference is what he is tracking. That is, we would expect old Shakespeare to have given up rare words he favoured as a young man and hence falling below 'expectation' also confirms the 'changing-tastes' hypothesis. Yet if that is so--if dropping words over time is as important as acquiring new ones--we should expect Jackson to be presenting all the evidence of non-linking between chronologically separated plays too (p. 44). At this point Jackson introduces a fundamental principle that so many attributors have neglected: the importance of negative testing. It is one thing to show that writing X is like writing Y in certain respects, but for authorship attribution you have also to show that lots of other writings are not like writing Y in those respects (pp. 45-46). This will become important later in Jackson's argument.

    Karolina Steinhäuser's work on Shakespearian rare words in each scene of Pericles (including the choruses) can be used to show that the choruses in acts 1-2 are like the scenes in acts 1-2 (and like early Shakespeare) and those in acts 3-5 are like the scenes in acts 3-5 (and like late Shakespeare); so again the play is internally divided and hence probably not all by Shakespeare. Even if he were, in the choruses, imitating an archaic style, then Shakespeare would have to have dropped this imitation part-way through. Steinhäuser's more finely reticulated counting of rare words (by scene) allows Jackson to rank the scenes in terms of rare words per thousand words even though the scenes differ in length. When he does this, not all the 17 scenes of acts 3-5 come ahead (in terms of Shakespearian rare word richness) of all the 11 scenes of acts 1-2, but most of them do and that is not likely by chance distribution of rare words (pp. 47-49). Gregor Sarrazin's work on very rare words confirms the foregoing: there are far fewer of these (expressed in lines per rare word) in Pericles acts 1-2 and Middleton's part of Timon of Athens and Fletcher's part of All is True than is normal for all the other Shakespeare plays (pp. 51-53). While it is admittedly much more subjective, poetic parallels (such as calling eyelashes the "fringes" of the eye) between Pericles acts 1-2 and the rest of the Shakespeare canon and between Pericles acts 3-5 and the rest of the canon confirm what has already been found: the latter is much more Shakespearian (parallels occur twice as often) than the former. This also makes unlikely the possibility that Pericles acts 1-2 are early Shakespeare later reworked (pp. 56-59).

    Metrical features such as extra syllables over the usual 10, various degrees of enjambment, and absence of a caesura, can all be measured (pp. 59-68), and Karl Wentersdorf produced a table showing how a single indexical figure derived from these features rises steadily over Shakespeare's career. The figure for Pericles acts 3-5 takes its expected place amongst the late plays but the figure for Pericles acts 1-2 is more like the figures for Shakespeare's late-sixteenth century plays. Jackson describes Ants Oras's work on where the caesura falls, which shows that increasingly over his career Shakespeare put the pause in the second half of a verse line, and the figure for Pericles acts 3-5 matches the late plays and the figure for Pericles acts 1-2 again matches the late-sixteenth century plays. Charles A. Langworthy's work on rates at which sentences start at a verse-line beginning and/or end at a verse-line end--in early plays they usually do, in late plays usually not--confirms the preceding work. Marina Tarlinskaja's works on rates at which 'slots' for unstressed and stressed syllables in iambic pentameter verse (1, 3, 5, 7, and 9, and 2, 4, 6, 8, and 10, respectively) are actually filled by stressed or unstressed syllables shows that over his career Shakespeare decreasingly put stressed syllables in slots 1 and 4 and increasingly put stressed syllables in slots 3, 6, and 9. Into this trend of changing habits Pericles acts 3-5 fits well as late Shakespeare and Pericles acts 1-2 looks more like late-sixteenth century Shakespeare. Also, in the second half of his career, Shakespeare increasingly allowed polysyllabic words to be the cause of loss of stress in a slot where we would expect it, compared to how often he allowed monosyllabic words to do this work of taking away stress. 

    As well as summarizing these mutually-supportive (and under-recognized) works, Jackson has done his own fresh work on elision. In early Shakespeare we/you/they are and I/we/you/they/to have are rarely best spoken as monosyllables (such as we're and they've) and frequently are best spoken as disyllables if one wants to follow strict iambic pentameter. Jackson does not make it clear whether he means this claim regardless of how the words are spelt, but in fact he is confining himself only to cases where they are fully-spelt out, since he uses only Marvin Spevack's concordance entries for are and have and these do not include the elisions, which get listed under the first word (so under they for they're). Of course, one would want to check whether the edition Spevack used, the Riverside edited by Gwynne Blakemore Evans, had ever changed non-elided to elided spelling for the sake of metre. Evans's introduction (The Riverside Shakespeare, p. 40) refers to his regularization of -ed and -'d endings (all made -'d) in prose passages on the assumption that mere compositorial convenience might be the determinant; Evans might also, on that basis, have wondered whether convenience shaped practice in full-width verse lines where the compositor faced the same pressure in justifying the line of type. Evans's saying nothing of are and -'re and have and -'ve implies that he left them as he found them, but it would be reassuring to know and Jackson's argument is not complete without this information. Pericles acts 1-2, it turns out, has a much higher proportion of fully-spelt out we/you/they are and I/we/you/they/they have being pronounced monosyllabically (if we want to preserve iambic pentameter) than ought to be the case were it early Shakespeare (p. 71). Shakespeare's use of rhyme fell (albeit unevenly) over his career and was low by 1607; rhyme use is high in Pericles acts 1-2 (one in 4 lines) while low in Pericles 3-5 (one in 33 lines), and the particular kinds of rhyme in Pericles acts 1-2 (especially the pattern aabcc) are unlike what Shakespeare does elsewhere (pp. 72-73).

    To conclude this section on why Pericles simply cannot be the work of one writer, Jackson touches briefly on Barron Brainerd's tests, which relied mostly on usage of certain words across the career (unto, because, and with decreased over time while might, more, and most increased over time) and the tests of the Claremont McKenna College Shakespeare Clinic that pass Pericles acts 3-5 as Shakespearian but Pericles acts 1-2 as not  (pp. 75-79). Jackson does not give much detail about the Claremont McKenna tests, but criticizes their method of 'badges and flukes'. This method relies on Shakespeare's most-commonly-used and least-commonly-used words, relative to the dramatists Marlowe, Greene, Kyd, and Munday. The problem with this is that Shakespeare started writing in a style that was like everyone else and got distinctive over time, and the Shakespeare plays chosen to form a baseline of his style were about 10 or more years later than the plays from Marlowe, Greene, Kyd, and Munday. Thus by this test the early plays of Shakespeare tend to look non-Shakespearian because they sound too much like Marlowe, Greene, Kyd, and Munday.

    Having established that someone other than Shakespeare wrote Pericles acts 1-2, Jackson turns to identifying the writer, paying most attention to the likeliest candidate, George Wilkins. The only known sole-authored play by Wilkins is The Miseries of Enforced Marriage, so that is not much to go on for establishing if the Wilkins oeuvre shares features with Pericles acts 1-2. We can also see if things unique to Miseries and Pericles acts 1-2 crop up in Wilkins's share of Wilkins, Day, and Rowley's The Travels of the Three English Brothers. Of course, most of the stylometricians so far discussed never thought to look at Wilkins's work, but F. G. Fleay and H. Dugdale Sykes found that Miseries has metrical affinities with Pericles acts 1-2 (pp. 83-86). Jackson has not done Langworthy's or Tarlinskaja's kind of analysis on Wilkins' play, but he has repeated Oras's work on where the caesura falls (how often after the first syllable, how often after the second syllable, and so on) and finds that Miseries is much like Pericles acts 1-2 and unlike Pericles acts 3-5 and Coriolanus. This result Jackson submits to a series of comparisons between Pericles acts 1-3, Pericles acts 3-5, Miseries and all the Shakespeare plays, with statistical computation to show how likely it is that the result achieved could happen by random chance (pp. 88-94). Consistently Pericles acts 1-2 is like Miseries and Pericles acts 3-5 is like the rest of Shakespeare around 1607. Likewise, repeating the we/you/they are and I/we/you/they have as monosyllable or disyllable test for Miseries shows it to be like Pericles acts 1-2, and so on for the Wilkins share of Travels too. Jackson reports David J. Lake's work on rhymes, especially assonantal near-rhymes (such as ship/split, sung/come) and finds Pericles acts 1-2's high frequency of them to be like Wilkins's work and not like Shakespeare's or anyone else's (pp. 95-96). Jackson's fresh work on rhymes (of the direct, law/awe kind) shows that where Miseries and Pericles have a rhyme in common, it is almost always in Pericles acts 1-2 and not in Pericles acts 3-5, and that three of the rhymes that Miseries and Pericles acts 1-2 share (consist/resist, him/sin, impudence/offence) occur nowhere in the Shakespeare canon (pp. 97-99). Comparison of all the rhymes in Miseries with the rhymes in all the Shakespeare works shows that the Shakespeare works have far fewer rhymes in common with Miseries than Pericles acts 1-2 has rhymes in common with Miseries, and indeed that after Pericles acts 1-2 the text with the next greatest number of rhymes in common with Miseries is Wilkins's share of Travels. But what if the rhyming links between Pericles acts 1-2 and Miseries are due to them both being full of commonplace rhymes? To exclude this possibility Jackson ran a couple of unShakespearian rhymes that appear in Pericles acts 1-2 and in Miseries (consist/resist and impudence/offence) through Chadwyck-Healey's Literature Online (LION) database and found that they are extremely rare (pp. 100-104).

    To show that the counting of function words (articles, conjunctions, pronouns, prepositions, auxiliary verbs, and so on) can be a stylometric tool, Jackson quotes passages from Marlowe's Tamburlaine (full of of . . . the constructions), Kyd's The Spanish Tragedy (full of ands), and Middleton's Women Beware Women (full of it . . . a constructions), which features, were they typical of the plays at large, would make for good general discriminators of authorship (pp. 105-106). Jackson touches on analysis of the so-called Federalist papers from the mid-eighteenth century and the Pauline New Testament using function-word frequency--in the latter case the analysis is now considered suspect--and Thomas Merriam's application of function-word analysis to Sir Thomas More. The Federalist work was especially good because the researchers produced their discriminators using just half the available material, reserving the other half (of known authorship) for the unbiased testing of the efficacy of their discriminators (p. 107). Jackson reports M. W. A. Smith's function-word analysis of Pericles acts 1-2 and Pericles acts 3-5 compared to the works of Shakespeare, Chapman, Middleton, Jonson, Webster, Tourneur, and Wilkins, in which Pericles acts 3-5 comes out as consistently (over different kinds of test) closest to Shakespeare and Pericles acts 1-2 gets inconsistent results, with some tests favouring Shakespeare and other tests favouring others dramatists (pp. 109-13). Jackson describes his own function-word analysis done by hand in the 1960s and 1970s in which he counted the frequencies (relative to one another) of the occurrences of a, and, but, by, for, from, in, if, of, that, the, to, and with in Shakespeare. Unsurprisingly, amongst a group of contemporary dramatists chosen, the work of Wilkins is closest in function-word frequencies to Pericles acts 1-2 and the work of Shakespeare closest to Pericles act 3-5 (pp. 113-18). Here again Jackson slips into unhappy phrasing about chance ". . . the probability [of Revenger's Tragedy sharing the discovered function-word frequencies with a set of Middleton plays] being considerably less than one in ten thousand that it was due to chance" (pp. 114-15), when in fact he means were chance responsible for the rates at which function-words are chosen in the test texts, we would get this result one time in ten thousand. Even chance can produce these results on rare occasions, and there are lots of other factors to consider in real writing, such as one writer imitating another.

    When one puts the 27 scenes of Pericles in descending order of the frequency with which they use 'to + verb-infinitive' (such as the opening line's "To sing a song"), 7 of the top 8 scenes are from Pericles acts 1-2, the interloper being scene 4.4, which for other reasons people have long suspected is not by Shakespeare (pp. 118-22). This distribution into two distinct populations is most unlikely if all the scenes were by one hand and the use of the 'to + verb infinitive' usage were randomly distributed amongst the scenes. There are about 20 such uses per 1000 words in Pericles acts 1-2 (close to the Wilkins norm and unlike almost all other writers) and about 10 such uses per 1000 words in Pericles acts 3-5, which is close to the Shakespeare norm. Simply counting how often the function word to is used corroborates Wilkins's writing of Pericles acts 1-2, and so do unusual uses of which, such as the which, favoured by Wilkins and no-one else and occurring in Pericles acts 1-2 (pp. 122-29). Jackson reports that Jonathan Hope's sociolinguistic approach says little about Pericles because for most of the things he measures acts 1-2 are like acts 3-5, although in the detail there are some things that indicate Wilkins for Pericles acts 1-2; examples are use of non-personal who and 'non-restrictive zero forms', which means the dropping of relative pronouns, as in 'all the examples [that] I can think of' (pp. 129-34). Early in his career Shakespeare overused the word unto, and late in his career he overused the word most. Pericles acts 3-5 has few untos and many mosts, while Pericles acts 1-2 and Wilkins's other work has more untos than mosts (pp. 136-38). Jackson also surveys certain kinds of analysis begun by Cyrus Hoy and used successfully to determine what Middleton wrote, but which is of limited value in relation to Pericles because acts 1-2 and acts 3-5 are not unalike in this regard (pp. 138-42). (This should, of course, have been counted amongst the evidence against dual authorship rather than listed as non-evidence.)

    A bridge to chapter 5 ("A Literary-Critical Approach to Style in Pericles") appears at the end of chapter 4 where Jackson reports some interesting verbal parallels and collocations between Wilkins's work and Pericles acts 1-2 (pp. 142-48). Chapter 5 itself is old-fashioned literary criticism and is the least successful part of this book simply because it is subjective. If one cannot hear the difference between acts 1-2 and acts 3-5--and this reviewer confesses that he cannot--then literary criticism is not likely to make good the deficiency. Chapter 6 is devoted to summarizing and defending the case for Wilkins as the co-author of Pericles, and refuting the claims for Shakespearian sole-authorship made by James O. Woods (based on imagery that is in fact commonplace), Karen Csengeri (likewise for diction), and A. Q. Morton (bad stylometry, especially for measuring the placing of words at sentence boundaries, which in fact is editorially not authorially made). Jackson points out that the consistent use of the two main sources in the play (Gower's Confessio Amantis and Laurence Twine The Pattern of Painful Adventures) is not of itself evidence for a single shaping hand across acts 1-2 and acts 3-5, because collaborating writer simply agree about these things. J. C. Maxwell (in the Cambridge New Shakespeare) pointed out that in Wilkins's Painful Adventures Marina does not know of her parentage (that Cleon and Dionyza are only foster-parents) until the dying Lychorida tells her, which is how Twine has it, and hence thought that Wilkins did not have anything to do with the play in which, of course, Marina knows her parentage all along. Jackson points out that this shows only that Wilkins did not know the second half of the play well: he had presumably seen it in performance but not having written these scenes it were easier to plagiarize Twine on this point than follow what happened in performance (pp. 180-81). Jackson mocks the New Cambridge Shakespeare editors Doreen DelVecchio and Anthony Hammond's absurd adherence to all the errors in the Pericles quarto and their strained attempts to defend the misreadings, and hence he justifies the use of Wilkins's prose novella to help patch the deficiencies in the quarto (pp. 183-89).

    Jackson's final chapter explains his "New Technique for Attribution Studies", which turns out to be fairly obvious in principle and novel only in exploiting new technology. The idea is to look in LION for words/phrases (using 'near to' proximity searching for phrases) in the passage under investigation amongst the known works of the rival contenders for authorship, and to count the hits. One has to make the corpora of the rival candidates roughly equal in size, otherwise getting a hit amongst, say, Shakespeare plays means little if the rival is Kyd whose entire corpus is just The Spanish Tragedy (pp. 193-203). In the case of Pericles, the corpus of Wilkins is already tiny (just Miseries of Enforced Marriage) and to match this the Shakespeare corpus is reduced to just The Tempest. Unsurprisingly, Pericles acts 1-2 has many more words/phrases in common with Miseries than with The Tempest, and for Pericles acts 3-5 the reverse is true. Turning to the detail, Jackson reports that Pericles scenes 4.2, 4.5, and 4.6, which he expected to be more like The Tempest than like Miseries, are in fact more like Miseries so perhaps Wilkins had a hand in them (p. 206). When, as a test of the methodology, the words/phrases that The Tempest and Miseries share with Antony and Cleopatra, The Winter's Tale, and Wilkins's supposed part of The Travels of the Three English Brothers are counted, Antony and Cleopatra gives "ambiguous results", The Winter's Tale shows up more Tempest connections than Miseries connections, and the Travels bit shows more Miseries connections than Tempest connections (p. 207). Jackson seems to think his test has come out rather well, but he is calling "ambiguous results" the fact that Antony and Cleopatra (which is definitely Shakespeare's) has fewer links with The Tempest (definitely Shakespeare's) than it has with Miseries (definitely Wilkins), in the ratio of 2:3. That is to say, 40% of Antony and Cleopatra's links are to The Tempest and 60% of them are to Wilkins's writing. I would have thought this to be devastating evidence of the insufficiency of this discriminator: it finds a known Shakespeare play to be, if anything, unShakespearian.

    At this point Jackson starts to (rightly) fret over the Antony and Cleopatra results and wonders whether the test is skewed by Miseries being almost half as long again as The Tempest. If we adjust the figures for this (by making more of the Shakespeare connections, proportionally) then Antony and Cleopatra comes out as Shakespearian again, but of course if Jackson really thinks that The Tempest's being short has skewed the test he should go back and recalculate all the figures on the previous page: presumably the things that looked Wilkinsian should now seem a bit less Wilkinsian in the light of the greater weight to be place on the (numerically fewer) links to Shakespeare (p. 208). Clearly still worried, Jackson turns to a subset of his Pericles/Miseries links to focus on just those that are unlike anything elsewhere in the entire Shakespeare canon. He produces a pretty extensive and impressive list of 59 links, of which 47 are in Pericles acts 1-2 and 12 are in Pericles acts 3-5, hence the former is by Wilkins. The ones in acts 3-5 cluster in tiny bits of that part of the play (such as part of a brothel scene) that might also be by Wilkins; this might explain why the brothel scene is contradictory about Lysimachus's intentions in going to the brother: Shakespeare toned down, but did not entirely eliminate, Lysimachus's vice (pp. 208-12).

    Using Sarrazin's rare-word links (from chapter 3) it emerges that the bits of the brothel scenes that seem, on the new evidence, to be Wilkins's have links with the 4 periods of Shakespeare (Two Gentlemen of Verona to A Midsummer Night's Dream, Romeo and Juliet to Julius Caesar, As You Like It to Timon of Athens, and King Lear to All is True) that are distributed much like the links between Pericles acts 1-2 and those 4 periods are distributed. And conversely, the bits of the brothel scenes that do not, on the new evidence, seem to be Wilkins's have links with the 4 periods that are distributed much like the links between Pericles acts 3-5 and those 4 periods are distributed. Thus, although the sample of data is too small to prove much, the bits of Wilkins that we seem to have found in the brothel scenes are like the work of Wilkins in Pericles acts 1-2 (p. 213). If, as seems to be emerging, Wilkins had a hand in the brothel scenes, then the versions of those scenes in his prose novella have added usefulness in supplementing the Pericles quarto versions of those scenes. The important conclusion of the book, of course, is that Wilkins wrote Pericles acts 1-2. There is no reason to suppose he ever wrote more than that (say, a full play) from which Shakespeare just extracted two acts. Renaissance theatre was more economical with revision than that, and the likeliest collaborative scenario is that Shakespeare and Wilkins worked together as collaborators with an agreed division of labour (pp. 215-16).

    His main claim effectively (indeed, multiply) proven--my quibbles notwithstanding--Jackson offers a couple of appendices. In the first (pp. 218-32) Jackson defends the view that Pericles says "till she be married . . . all unscissored shall this hair of mine remain" instead of "vnsisterd . . . heyre" (as the play quarto has it) on the grounds of the unemended text making no sense of "till she be married"; it is not as if Marina would get a sister on her marriage day. (Actually, Pericles could plausibly be saying that he will not produce another heir, to whom Marina would be sister, until Marina is married. It is a wonder Jackson could not see that.) Jackson points out that Painful Adventures, which was unknown when the unscissored emendation was first proposed by George Steevens, independently confirmed the emendation, and Gower and Twine's versions of the story also confirm the hair-growing vow. Although it is never going to be explained with complete satisfaction, Jackson surveys the theories about the manuscript underlying the Pericles quarto and concludes that Gary Taylor's account (in the Oxford Textual Companion) is the best overall. Jackson reconsiders the brothel scenes (especially in their repetitiveness and contradictions) in the Pericles quarto, in Wilkins prose novella, and in the sources, in the light of the 'discovery' that Wilkins had a hand in them. In the second appendix (pp. 232-43) Jackson gives the LION data showing words/phrases that each scene of Pericles shares with The Tempest or Miseries but not both.

    Richard Proudfoot is a general editor of the Arden Shakespeare and at an age when others might be thinking of winding down he launched its third series (the current one) in 1995. In a large collection of fairly short essays called In Arden: Editing Shakespeare: Essays in Honour of Richard Proudfoot edited by Ann Thompson and Gordon McMullan, 21 Arden editors, past and present, pay tribute to the generosity with which Proudfoot gives away ideas that others turn into books of their own. "In all fairness", write Thompson and McMullan, "the name of Richard Proudfoot should be on the cover of every volume, not just as general editor but as, in effect, co-editor" (p. xii). Proudfoot's work has brought stage-centered thinking to prominence in Shakespeare scholarship, via Arden and as textual adviser to the Oxford Complete Works that so dramatically (in both senses) altered the scene in 1986. The second Arden was not stage-centered, but the third is and indeed George Walton Williams argues that editing is itself a kind of performance directing "for the page" (p. xv). For Thompson and McMullan, after years of division between editing and theory, suddenly they have come together (pp. xvi-xvii). The essays in the book are divided into 5 categories: "Part I: Bibliography/Theory of Editing" (pp. 1-61), "Part II: Editing and Feminism" (pp. 63-108), "Part III: Editing and Stage Practice" (pp. 109-74), "Part IV: Annotation and Collation" (pp. 175-218), and "Part V: The Playwright and Others" (pp. 219-68). Not all the essays are relevant to this review. A. L. Braunmuller ("Shakespeares various", pp. 3-16) starts with early-twentieth century comments on editorial notes and collations from E. M. W. Tillyard and Stephen Potter, but is primarily concerned with eighteenth- and nineteenth-century variorum editions.

    In the next chapter, Giorgio Melchiori asserts "The continuing importance of new bibliography" (pp. 17-30) but has nothing substantial with which to support that (entirely reasonable) claim. Melchiori rightly comments that Werstine "could hardly be more unfair" for claiming that New Bibliography did not take account of the multiple, non-authorial inputs that could get registered in early printed texts (p. 19). Something goes awry in Melchiori's quotation of Stanley Wells: "There is no doubt that Stanley Wells is right in stating that 'The primary surviving texts of Shakespeare's plays represent those plays in various states of composition' and that none of these texts 'necessary represent in anything but in a definitive state the words that Shakespeare wished to be spoken or a larger action that he wished to be bodied forth' (in Elam, 340)" (p. 24). This is such an odd thing to claim--surely Wells meant that a definitive state was not available to us--that I followed Melchiori's footnote. "Elam, 340" is supposed to refer to an article by Keir Elam but is in fact an article by Melchiori himself ('What Did Shakespeare Write?', Textus: English Studies in Italy 9 [1996].339-56) and in it Wells is indeed quoted thus, and the quotation cited as a letter to the Times Literary Supplement on 18 January 1986. There was no issue of the TLS on 18 January that year, and a glance at the previous and following years' issues shows that Wells's letter actually appeared on 18 January 1985. And, unsurprisingly, Wells wrote that we should not expect to find the plays in "anything like a definitive state" ('Editing Shakespeare', 18 January. ISSN 0307661, TLS 4268 [1985].63) rather than the opposite as Melchiori's quotation has it. Melchiori concludes his essay by giving his views on the provenance of all the plays that exist in quarto form, without bothering to explain how he came to these opinions (pp. 26-29).

    The next essay is much better, and in it Anthony B. Dawson argues that we do no have to give up entirely our notions of the authorial ideal, only to moderate them ("Correct impressions: Editing and evidence in the wake of post-modernism", pp. 31-47). In Troilus and Cressida at 4.4.47 (Folio TLN 2434) F, but not Q, has "Enter Aeneus" and then both texts have "Aeneas within. My lord, is the lady ready?". Is he within or not? This "Enter" could be the bookkeeper's reminder to himself to have the actor ready, and is not likely to be an authorial revision: it is unlikely to be Shakespeare saying, no, let us have him come in there, since the whole point of the scene is "the pressure exerted by Aeneas's invisible presence" (p. 34). For the Oxford Complete Works, Gary Taylor argued that the "Enter" had to be purposeful, while the "within" might be a failed deletion, so perhaps he just stands in the doorway, making a kind of half-entrance. (In fact, work by the scholar who has spent the longest considering Shakespearian entrances and exits, Mariko Ichikawa, shows that an actor being 'within' did not necessarily mean he could not be seen by the audience (''Acting Spaces' in English Renaissance Drama: An Unpublished Research Report' [2003]).) Later in the scene there is an "Exit" after Cressida's last line, which cannot be her exit as she needs to be silently onstage for what happens next. Taylor took this as an exit for Paris whom he also, like Aeneas, imagines 'at the door' rather than 'within' even though he speaks lines marked 'within' in Q and F. The trouble is that this staging makes awkward Troilus's line "come you hither" to Paris, which is more easily spoken to someone offstage ('come on, get on stage') than someone in the doorway. Dawson thinks that perhaps this "Exit" is an error made by a bookkeeper looking quickly over the text, seeing that Cressida has nothing more to say, and so erroneously giving her an exit (p. 35). The scene, Dawson argues, needs Aeneas and the others to be offstage not loitering in doorways; perhaps the actors were keen to show themselves and so spoilt Shakespeare's design. Or perhaps, and the mistaken entry stage direction "Enter the Greekes" supports this, the manuscript underlying F was imperfectly annotated by a bookkeeper--who did not care that there is actually only one Greek, Diomedes, in the group that enters, only that a group enters--and the "Enter Aeneus" was him warning himself to have the actor ready and the "Exit" after Cressida's last line was him noticing that she had finished her speeches in the scene (p. 36). Sensibly, Dawson advises that just because performance never quite conforms to a given manuscript (there are always the little flourishes not scripted) does not mean we cannot reasonably treat it as though it does (p. 37).

    Surveying how these matters have been theorized, Dawson relates that for Jerome McGann, there is the 'text' (the "literary product . . . as a lexical event"), the 'poem' (the place where this happens in "a specific process of production . . . and consumption"), and the 'work' (the superset containing "all the texts and poems which have merged in the literary production and reproduction processes"). For W. B. Worthen in Shakespeare and the Authority of Performance a performance is like McGann's 'poem', which can never be authentic because the 'poem' in this sense is utterly tied to Shakespeare's own time: there are only performance events in their own specific times, each of which is different because the time is different. McGann argues that the contingency of meaning--its being the result of collaborative forces (and, I should argue, conflictual ones such as censorship)--makes the author's final intention unsuitable as a guiding criterion for selecting one's copy-text (p. 39). In Unediting the Renaissance, Leah Marcus attacked the New Bibliography but as Michael Bristol argued (and Dawson agrees) there is a difference between veridical and circumstantial evidence. The New Bibliographers were not trying to establish a single coherent history of the texts (what veridical evidence helps with), only a set of plausible explanations for the texts (for which there is only circumstantial' evidence; this is what Marcus misunderstood) (p. 40). Dawson points out the problems with Marcus's view--most importantly that a commitment on principle to "discontinuity and rupture" is as ideological as a commitment to continuity and order--and insists that McGann's terminology does not give enough space for the idea, the non-material version, of the created artefact: there has to be a 'work' that is not simply Q or F Troilus and Cressida (p. 41). Where is it? Dawson answers: "The text is born in the brain of its comically talented author, but when it grows up it becomes a book" (p. 42). With ideas like this in circulation, 2003 really was a good year for literary approaches to Shakespeare. The 'work', Dawson goes on, is a dialectic of the immaterial idea and the physical embodiment, and without some element of the immaterial in one's conceptual framework for editing you could not even fix the grossest errors (p. 42). Quite right. Just because Shakespeare is not the single author of his plays (they are indeed collaborative in a number of ways) does not mean that he is not the 'primary' author; he is, and his "initiating authorial act", even though not wholly recoverable, is not wholly lost either: pluralizing authority does not undermine it completely (p. 43). And that, as Dawson observes, is one of the concerns of Troilus and Cressida too.

    H. R. Woudhuysen's essay explores the peculiar phenomenon that printed plays had unnecessary blank pages at the front and the back despite this being an expensive waste ("Early play texts: Forms and formes", pp. 48-61). Generally the versos of title-pages were left blank to prevent show-through, but that was expensive: Blayney reckons that when making 800 copies of a quarto about half the cost was the paper. We do not know who made book-design decisions. Compositors were paid by printers by the page, so the non-labour of setting a blank page was 'fat work' (as it was called), yet printers seem to have liked leaving a blank page, or even a whole leaf of two pages, at the end of a book to protect it when unbound. Such a blank end-leaf could be folded round to protect the title-page at the front, and there developed in the eighteenth century the half-title to protect the title-page proper. It is possible that blank pages and even whole leaves were there to make a short play seem bulkier in print than it really was, for the phenomenon is more common in plays under 2,000 lines long such as the first quartos of Shakespeare's Henry 5, Hamlet, and The Merry Wives of Windsor than it is in longer plays. This would suggest that plays were not quite the ephemera we have thought (the point made also by Erne above). The practice that Greg called 'continuous printing'--not starting a new line for a new speech but printing it on the same line as the end of the preceding speaker's speech--looks like a compositorial trick to save paper (and so correct for casting-off errors) but it appears in Q1 King Lear, which was set seriatim. Here the effort seems to be to leave the final verso unprinted upon, since there is much 'continuous printing' on the last three pages (p. 57). That leaving the final verso unprinted was considered important is witnessed by books such as Q2 Romeo and Juliet, which increases its lines-per-page rate towards the end, but has an unprinted final verso. Ultimately, having described the phenomenon in detail, Woudhuysen admits that he has not solved the puzzle that even though paper was expensive and worth saving, pointless blank pages were considered worth including in books.

    Easily the richest and most intellectually stimulating section of the book is the second one, "Editing and feminism". It begins with a demonstration by Suzanne Gossett (whose Arden 3 edition of Pericles will be reviewed next year) that, in the absence of hard evidence, editing Pericles is a critical, not a scientific, matter. In the play, Cleon asks Dionyza what she will say to Pericles about Mariana's disappearance, and in Q1 Dionyza replies "That shee is dead. Nurses are not the fates to foster it, not euer to preserue, she dide at night, I'll say so" (sig. G2r). H. H. Vaughan suggested emending to "That she is dead. Nurses are not the fates. To foster is not ever to preserve. . . .", which posits minimal error (is to it) and some repointing, and mirrors the repartee in the previous scene about bastards being brought up and down in the brothel (p. 65). Gossett thinks that "to foster is not always to preserve" is a good editorial aphorism. The Oxford Complete Works emended and patched the play wholesale while the New Cambridge Shakespeare edition bent over backwards to trust Q1, finding "To foster it" acceptable because it can be the indefinite object of the verb to foster, as in Lear "I cannot daub it further", and ignoring the problem of "not ever to preserve". The New Cambridge Shakespeare editors thought the Pericles quarto to be printed from foul papers because it is like the King Lear quarto (which probably was). However, Werstine has deconstructed the category 'foul papers', showing it to be metaphysical, and what is worse the New Cambridge Shakespeare's unemended lines of Pericles are often just gibberish. On the other hand, Taylor's solution for the Oxford Complete Works was vastly over-confident in the face of indeterminability and according to Gossett what the text needs is "a post-modern, post-structuralist approach" that takes each decision locally  (p. 67).

    As an example of how feminism affects editing, Gossett notes that editors often have Lychorida exit after giving Pericles his baby as though she were too low-status to stay while Pericles welcomes his daughter to the world, but in fact midwives were well thought of. Equally, Pericles is often made to hand over his baby while wailing his wife's death, as though ". . . a man cannot emote with a baby in his arms" (p. 68). Another example is Marina's startled response to Leonine's commission from Dionyza: "Why would she haue mee kild now? as I can remember by my troth, I neuer did her hurt in all my life" (sig. F3r), which editors since Malone alter the punctuation of to make "Why would she have me killed? Now, as I can remember . . .". but the point is indeed 'why now?' after fostering her all these years. The answer (that Marina does not know) is the recent unfavourable comparisons that people have drawn between her biological and her adopted daughter, and the subtext of their being now pubescent (p. 69). A common substitution of a passage from Wilkins's novella for what seems a faulty bit of the quarto--Cerimon's speech about the cold and apparently dead being revived (sig. E4r)--is unnecessary and apparently motivated by a concern that the Egyptian should be male and hence should be the agent, not the recipient, of a reviving. In fact, there was much interest in cold, apparently (but not really) dead women in the period (p. 70). The peculiar repetition of Pericles's vow about cutting his hair (first said after Thaisa's apparent death, 13.29, and repeated after hearing about Marina's death, 18.28) is the consequence of an emendation of the first one, which reads: "till she [Marina] be maried, | Madame, by bright Diana, whom we honour, | All vns