"Shakespeare: Editions and Textual Studies in 2022" Not the Year's Work in English Studies

The only major scholarly edition to appear this year was Catherine Richardson's edition of Arden of Faversham. Although Richardson accepts that Shakespeare wrote part of the play, her edition names no writers on its title page and it appears as part of the Arden Early Modern Drama series rather than the Arden Shakespeare. The General Editors, Suzanne Gossett, John Jowett, and Gordon McMullan, provide a preface that says of this series "The title-page of each volume displays the name of the company for which the play was written and the theatre at which it was first staged" and that "in the Introduction the play is discussed as part of a company repertory" (pp. xi-xii). In fact this volume has neither of those features.

    The play was published anonymously in quarto in 1592. Richardson's Introduction has little to say on the evidence that has led multiple scholars to the conclusion that Shakespeare wrote the central part of the play, or the debate about whether it is mainly (or even wholly) the work of Thomas Kyd. In her opening remarks Richardson distances herself from this aspect of debate about the play: "it has . . . become known as the first of William Shakespeare’s collaborations and his first drama in print" (p. 1).

    The real-life murder depicted in the play, and the punishment for it, happened in 1551, and were immediately news-worthy. Richardson contextualizes this within the "Regnal and religious changes" (p. 4) in the 1550s, which made the state's centres of authority particularly keen "to advertise the quelling of rebellions domestic and urban" (p. 4). Richardson details the publication of one of the chronicles, called Breviats by John Mychel in Canterbury, which was the first to tell the story in print. Raphael Holinshed's Chronicles carried a much fuller account of the murder. Richardson traces how the murder was retold in a virtually continuous line of publications up to the present day.

    Before Arden of Faversham, tragedies were about important people and common and middling people were the subject only of comedy. But political theory linked the household and the state authority, so a rupture in the former was implicitly a threat to the latter. Theatre audiences could tell the nuances of characters' places in the social hierarchy by what they wore, and could spot even small disruptions of this hierarchy when a character's speech was not sufficiently deferential. Black Will, Richardson notes, was already a notorious murderer in narratives outside of Arden of Faversham.

    The play is unusual in giving the anxieties and rhetoric of classical tragedy to characters in a middle stratum of society, and for being a tragedy in which much of the action takes place inside an ordinary Elizabethan home. Richardson thinks that Arden himself is a wittol, someone who "turns a blind eye to their wife's adultery" (p. 33). I understand this term to mean someone who does not mind his wife's adultery, or even encourages it for reward, and that is not Arden's case: he minds terribly. Richardson suggests that Arden's insecure position as head of his household, insecure on account of his wife's adultery, is connected to his insecure social status: he has climbed socially and feels vulnerable. Arden is distinctly mercantile rather than aristocratic in his mindset and unmanly in his uxoriousness. 

    The Introduction's section on the play's authorship (pp. 37-61) starts with noting that the earliest claim that the play was co-authored was made in a pamphlet of 1739 by Henry Collyer. Richardson uses G. E. Bentley's guess that about half of all plays from 1590 to 1642 were collaboratively written. (The true number is more like a quarter.) Richardson rightly points out that the work of attributing the play is hampered by there being few plays from around the time of Arden of Faversham that are sole authored and securely attributed: she quotes Brett Greatley-Hirsch's reckoning that there are just 29 from 1580 to 1592.

    Moreover, the shared English language of London was changing quickly in this period, and indeed any one author's style may well have changed rapidly too. Richardson thinks that Arden of Faversham's being a domestic tragedy makes attribution harder than for plays in other genres, where we find "a poetic language far removed from everyday registers of speech" (p. 42). That is, she assumes that everyday speech is inherently less distinctive of authorship than elevated speech.

    In fact, there is no reason to suppose that the language has to be noticeably distinctive and unlike everyday speech in order for attribution methods to have something to work on: the rates of usage of everyday function words turn out to be particularly good for attribution purposes. Richardson thinks it clear that the play has "two distinct patterns of writing" (p. 42) so she is implicitly rejecting Brian Vickers's claim that it is all Kyd's.

    Edward Jacob in his 1770 edition was the first to attribute Arden of Faversham to Shakespeare. Richardson sketches some of the recent debates about the authorship, including Gary Taylor's claim that Thomas Watson was Shakespeare's collaborator. The central scenes 4-8 have few characters while those before and after have a large number of characters, and Richardson points to Stanley Wells's observation about The Two Gentlemen of Verona that at the start of his career the inexperienced Shakespeare seems uncertain how to write scenes with more than four characters.

    Richardson notes that the entrances and exits of the character Michael are uniquely faulty in the early printed text of the play. She observes Rory Loughnane's point that if Arden of Faversham is early Shakespeare then, added to his known early comedies and histories, this shows him working in all three genres right from the beginning of his career. MacDonald P. Jackson found a series of likenesses across the early Shakespeare plays: adulterous liaisons, women who murder, men who reveal their plans in soliloquy, and kings as seducers (p. 51).

    The role of Alice is an extraordinarily long one for a woman: a quarter of all the play's words are hers. Shakespeare gave unusually large parts to his main characters, and was unusual in giving large parts to women, as Holger Schott Syme showed (p. 52). Most of Alice's lines are in the two long scenes at either end of the play, not the Shakespearian middle. But there is a peak for Alice in Shakespeare's Scene 8 (the Quarrel Scene) so perhaps at the start of his stint Shakespeare intentionally gave the actor of Alice a rest before the showpiece Scene 8.

    Terri Bourus observed that Alice has an easy pattern of cues, taking every alternate speech in Scene 8 and speaking with one other character for most of Scene 1. Martin Wiggins suggested that this fits a pattern of 'graduation' roles for senior apprentices in the Lord Chamberlain's Men's repertory: every 18 months to two years one of them got a big role like this. Richardson finds the central, Shakespearian section to be focussed on developing emotional relationships and depths, and remarks that in the plot document this might have looked like an easy bit for a junior contributing writer to fill in (pp. 59-60).

    The play is relatively undemanding of its playing space: no upper playing area is required but it does need a discovery space and probably a trap. The play might have been toured in Shakespeare's time but we don't have any direct evidence for that before the eighteenth century when it was regularly toured around Kent. A puppet-play version was popular too. Richardson considers some eighteenth-century manuscript versions that depart wildly from the early quartos, possibly arising from amateur rewriting. There was at least one professional rewriting of the story too in the eighteenth century. William Poel resurrected the original text in the early twentieth century and then Joan Littlewood did in 1955.

    Richardson briefly sketches the play's limited stage history outside the United Kingdom before turning to its life in printed form. The 1592 first quarto (Q1) was reprinted in 1599 (Q2) and 1633 (Q3). It was printed just once again, in 1770, before becoming more widely known by its inclusion in Henry Tyrrell's collection of Doubtful Plays of Shakespeare in 1851. Q1 was printed by Edward Allde for Edward White, who specialized in crime literature.

    Richardson explores the facts that Q1 and Q2 were printed in black letter and that the play makes unusually frequent use of the phrase "Here enters" at the start of stage directions, and some other particularities, but she draws no firm conclusions from them. Q3 was printed and published by Edward Allde's widow Elizabeth and she probably printed (for "C. Wright") the ballad called "The Complaint and Lamentation of Mistress Arden of Faversham in Kent" which has a woodcut that fits perfectly across the top of the ballad but had to be rotated 90 degrees to fit on the reverse of the title page of Q3. Including woodcuts with play editions was something of a fashion in the 1630s. Some of the clothing in the Q3 woodcut suits the 1620s but other pieces are much older, perhaps gesturing at "the type of costumes originally staged" (p. 103).

    In the section "This edition: How it differs from the quarto, what it adds and why that matters" (pp. 103-109), Richardson explains that she has added to the stage directions "to make it easier for actors to prepare for and readers to imagine more fully the visual impact" (p. 105) of a performance. This includes detailing how a character looks and when people are carrying properties needed in the ensuing action. Emendations of dialogue "have been made sparingly" (p. 106) in this edition, and lineation largely left as found in Q1.

    Richardson makes little attempt to defend this conservatism, remarking vaguely that "The intention here is to understand the text of Arden in the widest possible context" (p. 107). Richardson explains that in her commentary notes she refers more than is usual to non-literary sources, such as catalogues and collections of museums that help us understand the objects referred to in the play (such as "table" mentioned at 14.103).

    Turning to Richardson's text, we will as usual consider some of the interesting editorial decisions rather than seek to discuss them all. Noticeably, most of the textual notes are about lineation and the addition of stage directions that clarify the action. Richardson sensibly divides the play into scenes based on the clearings of the stage, but does not explain that this choice arises from early performance practice.

    At 1.47, and similarly on subsequent occasions, Richardson emends Q1's Ielyouse to Q3's jealous. The OED gives the trisyllable form (spelt multiple ways) as a normal spelling in the period and EEBO-TCP has hundreds of instances, so this is properly a modernization of spelling not an emendation.

    Richardson retains Q1's reading in which Alice complains that "these, my marrow-prying neighbours, blab" (1.134) about her affairs. The difficulty is marrow, for which Q2 has the much easier reading of narrow. Richardson defends marrow as meaning the substance of a matter and also meaning a husband or wife or lover.

    At 1.172, Richardson has Michael asseverate that "the farm of Boughton is mine own", meaning he will get it. In Q1, the place name is Bolton and since there is actually a place called Boughton it is reasonable to suppose that Q1's reading is a misprint for Bocton. Thus emending Q1's Bolton to Boughton is reasonable. But this is an emendation, and yet Richardson's note reads "172 Boughton] (Bolton)", which styling wrongly asserts that Boughton is merely a modernization of Bolton. (The General Editors' Preface page xiii makes clear that italicized brackets have this meaning about modernization.)

    At 1.597, Richardson again records with an emendation note what is merely modernization: "haply] Delius; happely Q1; happily Q2–3". The OED is quite clear that happely is a possible spelling of haply. This recurs at 2.36. Richardson has Arden say to Michael "Sirrah, let me hear no more of this!| Now, for thy life, once write to her a word --" (3.28-29) and explains that in his frustration Arden breaks off his thought, so she provides the long dash. The easier reading she resists is to adopt the emendation of Now to Nor. Richardson does not explain where Arden's mind was going with his broken off thought that begins "Now, for thy life, once write to her a word", which seems the opposite of Arden's obvious wish that Michael not write to Susan.

    Richardson retains Q1's reading to have Black Will say "Say thou seest Mosby kneeling at my knees" (3.96) rather than adopting the emendation ". . . at my feet" or changing kneeling to something else. I do not understand her defence that "The phrase as it stands is sensitive to the mental image of the relative proximity of the kneeler, whose head would be level with Will’s knees". Mosby would have to be extraordinarily short for this to be true.

    K. Warnke and L. Proescholdt provide the emendation plenisht for Q1's perisht to give "Hydra's head that plenished by decay" (4.13). In context, the sense requires that Hydra is indefatigably revived by what should diminish her, just as Alice's vice is indefatigably encouraged by correction. An alternative is Nicolaus Delius's florisht, which has the merit that it is often contrastively paired with decay, but is less easily misread as perisht than plenisht is.

    Richardson follows Q1 to have Shakebag say of his and Black Will's frustration at failing to kill Arden that "’Twould make a peasant swear amongst his boys" (5.57). Keith Sturgess proposed emending Q1's pesant to pedant (meaning schoolteacher), which has the merit of making sense of "his boys" as the pedant's pupils. Richardson has a parallel quotation of peasants being said to be too cowardly to swear, but this does not help explain the idea of the peasant having boys.

    At 6.14-15, Richardson follows Q1 to have Arden say that he dreamt that a forester "rounded me with that beguiling home| Which late, methought, was pitched to cast the deer". The problem is home as something that is pitched to catch a deer. Richardson acknowledges attractive emendations such as T. W. Craik's snare but finds emendation "unnecessary in the context of the play’s repeated questioning of the protective qualities of domestic space".

    Richardson has Michael say "Why, I’ll agree to anything you’ll have me, so you will accept of my company!" (7.30-31) using Q3's accept rather than Q1's except. Q1's reading would mean that Michael wants to be excused from dining with Shakebag and Black Will, which they just proposed doing, even though he is trying to get them to accept his excuse for his part in the failed attempt at murder. Of course, we could just treat except as an allowed form of accept, which the OED entry for except warrants.

    At 8.17, Richardson retains Q1's reading to have Mosby reflect that since he has become socially elevated, having "climbed the top bough of the tree", he finds that "Each gentle starry gale" shakes him and makes him fear a fall. Editors have wanted to emend Q1's stary to something more obviously connected to the weather such as Sturgess's "gentlest airy gale", which supposes that st got attached to the start of airy rather than the end of gentle. But Richardson defends starry as invoking "notions of astrological influence and elevation of the sightline". Richardson also sees an ironic play on gentle meaning not base-born -- that is, the wind is in fact like a ruffian -- that using gentlest would lose.

    Richardson has Mosby decide about Greene that "for his pains I’ll hive him up a while,| And after smother him to have his wax;| Such bees as Greene must never live to sting" (8.26-28 ). Q1 has heave for hive, which is an emendation Richardson gets from Delius, and as she points out there is no intended sense of raising Greene. Rather, the image is of bees being smoked to calm them, and in this context hive means to shelter.

    At 8.30, Richardson prints without comment the line "Chief actors to Arden’s overthrow", which is suspiciously irregular and could be fixed by inserting another monosyllable after actors. Yet she is particular about meter in emending to read "But mads me that I ever thought thee fair" (8.103) where Q1 has ever I for her I ever.

    Alice says that she will meditate on Mosby's letters to her, and "hold no other sect but such devotion" (8.122), which is essentially Q1's reading. In a long note, Richardson rejects emending sect, for instance to faith, because sect has "potent and divisive meanings". She rightly reports that the New Oxford Shakespeare Complete Works Critical Reference Edition claimed that the phrase "no other sect" is not found in books in EEBO-TCP from 1552 to 1592. In fact, though, Richard Bristow's A Reply to Fulke (1580, STC 3802) has it on page Uu4r. EEBO-TCP now has more books in it than it did when the New Oxford Shakespeare Complete Works Critical Reference Edition was being created ten years ago, and either this or limitations in ProQuest's search engine for EEBO-TCP probably explains Bristow's use being overlooked.

    At 8.133, Q1 has "A fence of trouble is not thickned still", which makes no sense. Richardson adopts W. Headlam's suggestion to read "A fount once troubled is not thickened still". Headlam conjectured that first fonte and once got combined by eyeskip to make fonce, and then someone trying to find sense in fonce troubled created Q1's reading; this makes sense if Q1 were printed from a scribal transcript.

    Richardson accepts the essence of Q1's reading to print "Flowers do sometimes spring in fallow lands,| Weeds in gardens; roses grow on thorns" (8.142-143). The second line is suspiciously irregular in meter and sense, but Richardson argues that if we "carry spring over from flowers to weeds" the sense at least is clear. Richardson is more willing to emend for sense than meter. For instance, two lines later she emends Q1's Himself to Himself is to print "Himself is valued gentle by his worth" because without is "the line does not make sense" (8.145).

    Likewise, Richardson retains Q1's reading to print "Well were his date complete and expired!" (8.162) and in order to find metrical regularity she marks the last word as having a stressed final syllable -ed. That is, she makes expired into a cretic, which I cannot recall seeing anywhere else. Other editors have instead suggested the simpler solutions of flipping complete and expired to smooth the meter, or keeping the order but emending complete to completed.

    At 9.20, Richardson has Shakebag say that he hates those who "carry a muscado in their tongue", and the problem is what muscado means. On the basis of this line, OED suggests it might be some kind of gun like a musket, and, on the basis of an occurrence 20 years later, the OED further muses that it (or a homonym of it) might be some kind of herb connected to the word musk. Richardson writes that here muscado is "probably a confusion of 'moschetto', the Italian for 'musket'", but in that case it is unclear why she does not emend to some form of moschetto.

    Q1 has Greene say of Arden's escape from death "The Lord of heauen hath preserued him" and Black Will contemptuously reply "Preserued, a figge, the L. Cheiny hath preserued him" (9.142-143). The reply seems to echo the wrong part of Greene's line: "preserved" instead of "The Lord of Heaven". Richardson adopts Jackson's emendation to make Black Will's reply "'The Lord of heaven', a fig! The Lord Cheyne hath preserved him". It is plausible that the profanity of this line caused someone to censor it to produce Q1's reading.

    At 10.1, Richardson follows Q1 to print "See how the Hours, the guardant of heaven's gate" without a gloss on guardant or an explanation of why it is singular despite Hours being plural. Q1 has Michael say of Shakebag and Black Will that "Theyle be your [meaning Arden's] ferrymen to long home" (10.46), which does not make sense. Richardson adopts the emendation to long to to your long, so the line likens them to Charon bringing Arden to the afterlife.

    In Q1 Mosby says that during his game with Arden the phrase (the watchword) he will use to signal to Black Will and Shakebag that they should emerge from their hiding to kill Arden is "Now I take you" (14.121). In the event, the watchword Mosby uses later in the scene is "Now I can take you", but Richardson thinks that the two forms need not be identical: "The point is . . . that they should strike when Mosby defeats Arden, rather than the exact words spoken in doing so . . .". So no emendation is needed to make these lines agree. At 14.172 Richardson again records as an emendation something that is merely a modernization: Q1's or rejected in favour of Q3s ere. OED shows that or was a way to spell ere in this period.

    Richardson's Appendix One gives the story of Arden's murder as it appeared in Holinshed's Chronicles. Appendix Two ("Doubling of Parts") is a real doubling chart, telling us which actors of major roles can also take minor roles. Richardson finds that the play needs no more than 8 men and 2 boys. Appendix Three lists the press variants in Q1.

    From the same imprint as Richardson's edition, that is the Arden imprint of the publisher Bloomsbury, is Suzanne Gossett's Shakespeare and Textual Theory in the Arden Shakespeare and Theory series generally edited by Evelyn Gajowski. Gossett's book is organized as twelve chapters taking at first a broadly chronological view, from early Shakespeare editions to later ones, and from the New Bibliography of the early twentieth century to reactions against it in the late twentieth century. Later chapters then expand upon particular themes that are current in Shakespeare textual scholarship, such as the nature of authorial collaborations and the materiality of writing.

    Chapter One (pp. 9-35) is concerned with "Shakespeare's Texts from the Sixteenth to the Nineteenth Century", and it has some unfortunate inaccuracies. Gossett reports that "Most quartos were printed by 'formes', that is, by the sheet holding four non-consecutive pages" (p. 13). This is not quite right, in two senses. Some of the pages on a quarto forme are consecutive: pages 2v and 3r appear together on the outer forme and page 1v and 2r and 3v and 4r appear together on the inner forme.

    Moreover, the phrase 'by formes' refers not to the order of printing but the order of typesetting, and the alternative was setting seriatim, meaning typesetting the pages in reading order. But the result is still that the formes contain non-consecutive pages, since it is the folding of the paper not the setting or printing that necessitates this. Gossett also muddies the water somewhat in asserting that the Shakespeare Folio was set by formes (p. 16). In fact once the first six pages (1r to 3v) of a Folio gathering had been cast off, setting could begin from the middle of the gathering (forme 3v:4r) and the second half of the gathering could be set seriatim, and usually was.

    Gossett repeats the fallacy that after being printed on one side ". . . the sheets were hung up until the ink was dry . . . (p. 16) before being returned to the press to have the other side printed. In fact, the sheets were stacked flat between printing of the first side and perfecting. Gossett repeats the fallacy that because of stop-press correction ". . . no two copies of the Folio have exactly the same mixture of corrected and uncorrected pages" (p. 17). Strictly speaking we cannot know this since not all the surviving exemplars have been fully collated. But in any case most corrections do not affect the text (in the sense of giving a viable alternative reading) so many of the exemplars are verbally identical even if they contain corrections to fix such things as spacing errors and pieces of type being inserted into the press upside down.

    Gossett describes the Chamberlain's Men putting on Richard 2 for Essex's supporters "On 7 February 1601, the night before an uprising" (p. 22). This rather gives the impression that the performance was at night, but it was of course given during the afternoon at the Globe. The ccompany demanded compensation for the fact that, the play being an old one, "they should have small or no company at it", and hence this was a public performance by daylight (E. K. Chambers Study of Facts and Problems 2.325).

    Gossett acknowledges that the New Textualists -- she cites Paul Werstine in particular -- showed that manuscripts used to run performances could be inconsistent and "irregular in naming characters . . . [and giving] stage directions with indefinite numbers [of characters]" (p. 19). But she seems to contradict this in asserting that: "A bookholder or prompter needs to know precisely which actors are to enter at any time".

    Gossett defines an issue of a book as "a republication with a few changes rather than a full re-editing" (p. 27). This is unhelpfully indefinite: how many changes are a few? A better definition is that a new issue represents a new publishing intention reusing standing type from an existing edition rather than merely reprinting it unchanged, which would be a new impression. If there are differences between the type in exemplars from the same edition or impression but not a wholesale resetting, that is a new state. G. Thomas Tanselle's classifications in his 1975 essay "The bibliographical concepts of 'issue' and 'state'" are widely accepted as the most useful on this point.

    Gossett's Chapter Two is "The New Bibliography" (pp. 39-46). She remarks that "In all likelihood the company archive [that is, the play manuscripts of the King's Men] went up in the fire that burnt down the first Globe in 1613" (p. 39). This is almost certainly untrue. Multiple contemporary accounts tell us that the fire started during an afternoon performance and although it was fierce there was time enough for the entire audience to evacuate and only the building was lost. It is implausible that in this orderly evacuation no one from the theatre company thought to save the manuscript playbooks on which its future depended.

     Gossett gives an account of the Hinman Collator, a device for detecting press variants within an edition by continuously flashing at the user first an image from one exemplar and then an image from another exemplar, and then back to the first. Gossett confuses this device with the Lindstrand Comparator which achieves the same end by instead using stereoscopic superimposition to find differences and in which it is indeed true that "the resulting single image appears smudged . . . where, for example, a letter varies" (p. 41). On a Hinman machine, by contrast, the variant type will appear to 'jump' from one reading to another as first one then the other is presented to the reader's eyes.

    Chapter Three, "The Advent of Poststructuralism" (pp. 47-55), gives a brief summary of the implications for Shakespearian textual studies of decentering the author. Chapter Four continues in a similar vein with "Textual and Other Theories" (pp. 56-68). Here Gossett considers the new schools from New Historicism onwards (feminist, postcolonial, queer). Of the last she cites Jeffrey Masten's suggestion that editors of Othello should use the form 't*pping' instead of choosing between 'tupping' and 'topping' in the first scene -- because we can make a case for either -- and letting the reader decide for herself what letter the asterisk stands for (pp. 58-59).

    Less silly examples of suggestions from feminist editors include retaining "join his hand with his" in As You Like It where, because the first pronoun refers to Rosalind, editors usually emend to "join her hand to his". As Laurie Maguire observed, Rosalind is still in male clothes so a male-male stage image might have a complement in this line about male-male handjoining (p. 60). True, but Rosalind is referred to with the pronoun "her" in the immediately preceding line,"Yea brought her hether", so if Shakespeare was intentionally flipping genders it is odd that he did not flip that one too.

    Gossett describes a few efforts at rethinking certain editorial matters in the name of postcolonialism, including Leah Marcus's contrasting of the endeavours of the New Bibliographers in perfecting Shakespeare's tests with "our present goal of reducing the colonial accretions that remains in those texts" (Marcus quoted on Gossett's p. 64). Gossett is somewhat sceptical of Marcus's claims, which when laid out one after another appear to be contradictory.

    For instance, Gossett quotes Marcus claiming that the men but not the women being given asides in The Taming of the Shrew sets up a hierarchy in which only the former have "direct access to the audience" (Marcus quoted on Gossett's p. 65). But Marcus goes on to claim that Shylock is allowed to express his hatred of Christians in an aside, "more venomous because of its secrecy", while ". . . the Christians are allowed to openly articulate their anti-Semitism" (Marcus quoted on Gossett's p. 65). Which is it: are asides a privilege or a constraint?

    Gossett quotes (pp. 66-67) from Marcus's dispute with John Drakakis over whether Shylock should be given the speech-prefix "JEW" throughout The Merchant of Venice, as Drakakis does in his Arden3 edition of the play (reviewed in YWES for 2010). Marcus's objection is not the one made by the present reviewer -- that Drakakis simply misunderstands the type-shortage phenomenon on which he bases his choice of the speech prefix "JEW" -- but rather that doing this has unwanted critical consequences.

    Gossett's Chapter Five on "Authorship, Agency, and Intentionality" (pp. 71-82) is a history of disagreement between poststructuralists of various stripes and everyone else about the nature of authorship. Gossett makes the point that in choosing the longer text in every case where there is a choice of early editions to base a modern one on, the New Oxford Shakespeare (of which the present reviewer is a general editor) necessarily privileges texts close to the moment of authorial creation over those close to the moment of first theatrical realization. This is because, as Jowett acknowledged in the New Oxford Shakespeare, plays tend to get shorter as they move further from their moments of creation, by censorship and other forms of deliberate cutting, and by accidental loss of lines.

    Chapter Six ("Attribution and Collaboration", pp. 83-106) is a summary of the history of authorship attribution and of the currently unresolved disputes. Gossett's account is reliable except in crediting the Text Encoding Initiative (TEI) with creating "searchable texts" (p. 98). The TEI has not done that. Rather, it developed and promoted the standards for encoding digitally not just the raw letters and punctuation of texts -- as for instance in Project Gutenberg's digital transcriptions -- but also encoding the meanings of the various parts, for instance that one part of the text is a speech and another part a stage direction. TEI itself created no texts and was never meant to. (Gossett might be confusing the TEI with the TCP, the Text Creation Partnership.)

    Gossett ends with an uncharacteristic overstatement that "Assuming that an individual writer tends to repeat himself across his oeuvre is a familiar method of identifying authorship, but it is discredited as a form of scholarship on collaboration because it ignores every capable writer's ability to imitate the style of others" (p. 105). Gossett seems here to have forgotten that she earlier wrote that when counting most-frequent words (typically, function words) we are tracking a feature of style "that a writer uses without noticing" (p. 92). If she thinks that writers do not notice this feature then it is hard to see why she also thinks a capable writer could imitate another's habits regarding this feature.

    I suspect that Gossett's remark about imitation was meant to cover only the uncommon words and the noticeable longer phrases that some investigators track to attribute authorship. If so, her remark is indeed apposite: the possibility of imitation of another's noticeable features of writing is a limiting factor in studies that track uncommon words and phrases.

    Chapter Seven on "The (In)Stability of the Text" (pp. 107-125) contains another technical slip. Gossett writes of each gathering in a folio-in-sixes format book, such as the 1623 Shakespeare Folio, that ". . . to print so much text seriatim a printer would need to arrive at the seventh page before he could redistribute his type . . ." (p. 109). But there is no such action as "to print . . . seriatim". Gossett means the action of setting the type seriatim, since in a folio-in-sixes each of the first five pages cannot be printed until its forme-mate (the twelfth, eleventh, tenth, ninth, and eighth page respectively) has been set. And the distribution of type cannot begin when the printer has "arrive[d] at" the seventh page but rather when he has finished printing it along with its forme-mate the sixth page.

    In Chapter Eight ("Editing and Unediting", pp. 126-146) Gossett explains the origins of the unediting movement and why an absolutist position about it cannot be taken in an edition for modern readers. That is, she gives cases where an editor simply has to intervene in the text for the text to make sense. Chapter Nine on "Book History and the Text" (pp. 147-167) pays particular attention to whether Shakespeare was primarily a man of the theatre or whether Lukas Erne is right that he was a literary dramatist, and then turns to recent work on the Pavier Quartos and the treatment in print of the sonnets. Her main point is that as a physical object ". . . the book, rather than the hypothesized author, creates the significance of its contents" (p. 167). It seems to me odd to refer to books having hypothesized authors, as if these persons' existence is uncertain.

    After a brief discussion of adaptation theory, Gossett's main topic in Chapter Ten, "Performance and the Text" (pp. 169-184), is how different performance occasions and venues might be reflected in the differing early editions we have of certain plays. Gossett sketches the well-known cases of Mistress Quickly's speech praising the Order of the Garter in the Folio edition of The Merry Wives of Windsor, Viola-as-Cesario not singing in Twelfth Night (though she said she could), the Willow Song being absent from the 1622 quarto of Othello, Arviragus and Guiderius not singing Fidele's funeral dirge in Cymbeline, and the adaptations by Thomas Middleton of Macbeth, Measure for Measure and All's Well that Ends Well.

    Gossett makes the valid point that the New Oxford Shakespeare's principle of preferring the longer text when there is a choice -- in order to give the reader the most Shakespeare -- might in fact be reducing the proportion of Shakespeare in the edition if the longer version has non-Shakespearian material in it. This is true in principle but it would have been helpful for Gossett to provide an example where this applies in practice; I cannot think of one.

    In Chapter Eleven, "Textual Theories and Difficult Cases: Hamlet and Pericles" (pp. 185-212), Gossett sketches all the difficulties with these two plays and applies to their cases, in turn, the principles outlined in each of the preceding chapters. Gossett mistakenly writes that the New Oxford Shakespeare is one of those editions that "take advantage of their electronic platforms to offer separate editions of all three texts" of Hamlet (p. 205). In fact the New Oxford Shakespeare puts on its electronic platform nothing that it has not also published in printed form.

    Gossett points out that Hamlet and Pericles were each subject to textual theories that "were somewhat upended" by the discovery in the nineteenth century of a new version: Q1 Hamlet and George Wilkins's The Painful Adventures of Pericles. Gossett's Chapter Two, "The Immaterial Text" (pp. 213-227),  is a survey of the advantages and dangers (such as lack of longevity) of the transformation of text to being now largely a digital rather than a print medium.

    Parts of Darren Freebury-Jones's Shakespeare's Tutor: The Influence of Thomas Kyd bear on our topic and we will focus on those and omit the rest. In particular, Freebury-Jones has relevant things to say about the authorship of the Shakespeare plays Arden of Faversham, 1 Henry 6 and Edward 3, and he reckons that Kyd also wrote a lost Hamlet play. Freebury-Jones argues against Bourus's claim that Shakespeare wrote the early Hamlet that Thomas Nashe referred to in 1589, in order to present the case that Kyd did.

    Thomas Dekker in Satiromastix (1601) mockingly uses the phrase "Hamlet revenge" in a passage that refers to Ben Jonson playing Soliman (from Kyd's Soliman and Perseda) and Hieronimo (from Kyd's The Spanish Tragedy, and Freebury-Jones thinks it likely that Dekker was thereby alluding to three Kyd plays, not two Kyd plays and a Shakespeare one. Also, as George Ian Duthie noted, ". . . Shakespeare's work would not be ridiculed by his own company in his own theatre" (p. 25). I think this supposes a greater sense of decorum than we are entitled to assume.

    To help dispose of Bourus's argument, and so leave the way clear for Kyd to compose the early Hamlet, Freebury-Jones cites the evidence -- including an essay by Jackson reviewed in NYWES for 2018 -- that Q1 Hamlet is not an early version of the play but a memorial reconstruction of the Q2/F Hamlet. Freebury-Jones gives an account of his method: he uses the plagiarism-detection software WCopyfind to search with texts downloaded from One Literature (formerly Literature Online) that were first performed in 1580-1600. Freebury-Jones also uses Pervez Rizvi's dataset and website called Collocations and N-Grams.

    Freebury-Jones gives an example of the kind of collocation he thinks indicative of authorship: "'distressful' and 'wound' cluster in The Spanish Tragedy and Arden of Faversham , but in no other plays of the period" (p. 39). The phrases he is referring to are "the wounds of my distresfull sonne" in The Spanish Tragedy and "To forge distressefull looks to wound a breast" in Arden of Faversham. This is certainly a matching collocation but does not really meet Freebury-Jones's self-imposed constraint that a parallel in phrasing has to be matched by a parallel in sense. In The Spanish Tragedy 'wounds' is a noun and in Arden of Faversham 'wound' is a verb, and the noun qualified by 'distressful' is 'son' in The Spanish Tragedy and 'looks' in Arden of Faversham.

    Since the same phrases might be found in different plays either because of "Common authorship or deliberate imitation" (p. 40), Freebury-Jones's technique is to "combine traditional reading-based approaches with modern methods of collecting parallels" (p. 41). The trouble is, of course, that this allows Freebury-Jones to make subjective judgements wherever the raw data do not point towards the attributions he wants to make, and nowhere does he discuss how he avoids this kind of investigator bias. Other investigators avoid it by not allowing themselves to apply "traditional reading-based approaches" and instead they just use the numbers.

    When Freebury-Jones turns to unusual phrasing in stage directions new matter emerges. Arden of Faversham has five stage directions that start "Then they . . .", and Soliman and Perseda  has six, and Freebury-Jones claims that "No other publicly performed play of the Elizabethan period matches" this (p. 52). This is true: the phrase comes up in other plays' stage directions but not so frequently: once each in Captain Thomas Stukely, A Warning for Fair Women, and Mucedorus, and George Peele's The Battle of Alcazar and Nicholas Udall's Ralph Roister Doister.

    Freebury-Jones also claims that ". . . Arden of Faversham and Fair Em share sixty-eight examples of stage directions featuring the relatively rare two-word unit 'Here enters' (there are sixty-four instances in Arden of Faversham and four in Fair Em by my count)" (p. 53). This also is true. But the anonymous play Common Conditions has 34 instances of "Here enter" or "Here entereth" stage directions, and Appius and Virginia, anonymously published in 1575 and perhaps by Richard Bower, has 16. It is not clear what Freebury-Jones thinks the presence of this relatively rare phrasing actually proves.

    Freebury-Jones summarizes the work done by Thomas Merriam and Albert Yang on counting word frequencies, which he says corroborates his new claims for Kyd's authorship. Freebury-Jones writes that he has not enough space to provide the details by which these approaches may be judged. Freebury-Jones is associate editor to Vickers on a new edition of Kyd's works, and Vickers has repeatedly insisted that simply counting word frequencies cannot shed light on authorship. It is easy to see that this would make Freebury-Jones circumspect about Merriam's and Yang's methods.

    Freebury-Jones turns to the topic of Shakespeare's plays echoing phrases from Kyd's plays using a spreadsheet created by Martin Mueller called "'SHCSharedTetragramsPlus". Freebury-Jones says this is available on his (Freebury-Jones's) website at <https://darrenfj.wordpress.com/2017111/> but at the time of writing (21 June 2026) this URL returns a "404. Page Not Found" error. This is a recurrent problem with Freebury-Jones's published works: it is impossible to follow up his references because of broken URLs. In the present case there is little excuse as the URL points to Freebury-Jones's own website, which ought to be stable and under his control.

    A related problem happens when, as here on page 61, Freebury-Jones supports his claim with references to documents on Mueller's own website called Scalable Reading at <https://scalablereading.northwestern.edu/?p= 312>. At the time of writing, this website requires log-in credentials so the general public cannot see what is in it. But even without access to Mueller's data it is clear that Freebury-Jones's claims are untenable.

    Freebury-Jones writes that when he has Queen Margaret say "I will not hence" in 3 Henry 6 Shakespeare was recalling Amurath saying "I would not hence" in Soliman and Perseda (p. 63). Maybe he was, but Kyd was not the only writer to use this phrase: EEBO-TCP has 26 occurrences, including in William Warner's Albion's England (written in the late 1580s), which we know Shakespeare read. Mueller's dataset comprises only plays, but we cannot assume that Shakespeare got phrases only from others' plays.

    Other phrases Freebury-Jones finds in Kyd and early Shakespeare are genuinely rare or unique to those two, but Freebury-Jones seems not to notice that this undermines his claims elsewhere in the book that unique shared phrasing proves common authorship. How are we to distinguish the effect of common authorship from instances of one writer merely echoing another? This question forms the title of one of Freebury-Jones's chapters ("Authorship versus influence") but Freebury-Jones has no discernible answer for it.

    Where Freebury-Jones wants to deny Shakespeare's contribution in Arden of Faversham, its shared phrasing with Shakespeare plays is dismissed as Kyd's influence on Shakespeare, but where Freebury-Jones wants to assert Kyd's sole-authorship of Arden of Faversham and other plays, Freebury-Jones claims that shared phrasing with known Kyd plays establishes common authorship.

    Throughout his argument, Freebury-Jones indiscriminantly mixes phrase matches that are truly rare, such as "the shadow of myself", found only in King Leir and Shakespeare's King John, with commonplaces such as "entire affection to", found in Fair Em and Shakespeare's The Taming of the Shrew but also found in dozens of other works. In his Chapter Four ("Revision", pp. 109-141), Freebury-Jones declares his aim of showing that 2 Henry 6 and 3 Henry 6 were written solely by Shakespeare and that the "Harey the vj" recorded by Philip Henslowe as performed on 3 March 1592 was written by someone else as a prequel and then adapted by Shakespeare (at the behest of the Chamberlain's Men) to make a three-part Henry 6 cycle.

    The crux of disagreements about 3 Henry 6 is just who wrote the play that Shakespeare later adapted. That Nashe contributed to it is widely agreed upon, but whether Christopher Marlowe or Kyd did is not. Freebury-Jones starts with what he calls the "strongest" (p. 115) qualitative evidence for Kyd's hand in the play:

To be enrolled in the brass leaved book
Of never wasting perpetuitie
(Soliman and Perseda)

Deserves an everlasting memory,
To be inrol'd in Chronicles of fame,
By never-dying perpetuity
(King Leir)

Anon from thy insulting tyranny,
Coupled in bonds of perpetuity,
Two Talbots winged through the lither sky
In thy despite shall scape mortality
(1 Henry 6)

I can see what the first two quotations have in common with each other -- the idea of being part of a permanent record and the use of the words enrolled" and "perpetuity" and the likness of "never wasting" and "everlasting" and "never dying" -- but I cannot see what Freebury-Jones thinks links these two to the third, the one he is trying to attribute. Freebury-Jones apparently thinks the likeness is self-evident, since he does not identify the words or phrases in common. I can see only 'perpetuity' being common to all three, and it is far from rare, having over 6800 occurrences in EEBO-TCP.

    Freebury-Jones considers that the "evidential pièce de resistance" that seals his case "is the degree to which the verbal fabric of the non-Nashe or -Shakespeare portions is woven out of Kyd's lexical individuality" (p. 118). This evidence is a set of calculations made by Mueller and provided in the form of a spreadsheet that Freebury-Jones says can be "found . . . on my website" (p. 141n3). This is the website that, as we noted earlier, returns a "404. Page Not Found" error for the URL that Freebury-Jones gives. It is therefore impossible to explore this evidence that Freebury-Jones describes as thpe strongest for Kyd's hand in 1 Henry 6.

    When he tries to refute a recent study by Taylor and John V. Nance that found Marlovian word choices predominating in the non-Nashe parts of 1 Henry 6, Freebury-Jones falls back on the biographical fact that "Kyd and Marlowe shared a room in 1591, and it is therefore probable that they also shared a reading knowledge of each other's works" (p. 120). Of course, consistent application of this logic would undermine Freebury-Jones's attributions to Kyd based on shared phraseology -- we could say that these are cases of Marlowe using Kyd's phrasings -- but Freebury-Jones admits this principle only when needed to explain away others' findings.

    Where Freebury-Jones does present the verbal evidence for his belief that Kyd wrote the non-Nashe parts of 1 Henry 6 he has to quote carefully to conceal how tenuous it is. In King Leir the Captain tells his men they have "To watch in this place, near about the beacon,| And vigilantly have regard" for passing ships. Freebury-Jones finds the same collocation of "watch . . . near . . . vigilant" in 1 Henry 6 in the order "vigilant . . . near . . . watch", which is fine when dealing with collocations. But by quoting the 1 Henry 6 occurrence in two chunks he obscures the fact that the first ellipsis (from "vigilant" to "near") covers seven words while the second (from "near" to "watch") covers 32 words. That is, the three words "watch", "near", and "vigilant" appear within a 10-words chunk of King Leir but are spread across 42 words (and two speakers) in 1 Henry 6.

    If we allow the notion of collocation to cover words as widely dispersed as this, an extraordinary number of verbal matches can be found between almost any two sizeable texts and such evidence proves nothing. Freebury-Jones presents as evidence for their common authorship the fact that Fair Em and the non-Nashe part of 1 Henry 6 contain the phrase "support this". This evidence is weak because the phrase makes hundreds of appearances in books of the period, including many known to have been read by Shakespeare including William Painter's The Palace of Pleasure, George Gascoigne's poetry, and Raphael Holished's and John Stowe's Chronicles, and more.

    Likewise the phrase "to none but to": again found in Fair Em and the non-Nashe part of 1 Henry 6, but also found over 1200 times in books of the period, including works by Nashe, Marlowe, Edmund Spenser, and Thomas Lodge. These examples could be multiplied many times -- "in the hour of death", "dare not speak" -- as examples of shared phrasing that Freebury-Jones thinks unusual enough to indicate shared authorship but which are in fact just commonplace expressions found in hundreds of books of the period, or else are genuinely rare phrases but are found in other Shakespeare plays. Instances of the latter include "make this marriage" in The Spanish Tragedy and the non-Nashe part of 1 Henry 6 but also (unmentioned by Freebury-Jones) in Antony and Cleopatra.

     Freebury-Jones turns to Shakespeare's contribution to 1 Henry 6 and speculates about just which bits, and why, were added by him. In particular Freebury-Jones gives Shakespeare 2.4, the Temple Garden rose-plucking scene, and 4.5 in which Talbot senior and junior each try to get the other to flee from a losing battle. Freebury-Jones reckons 4.5 was written to replace what is now 4.6 (by Kyd) which conveys essentially the same dramatic action but less successfully on account of clumsy enjambement destroying what should be the self-contained nature of each line.

    In Chapter Five, "Collaboration" (pp. 142-165), Freebury-Jones sets out the known facts about Edward 3 and claims that Shakespeare's co-author was Kyd. From the fact that the Countess of Salisbury love scenes are well integrated and alluded to in other parts of the play, Freebury-Jones concludes that Shakespeare did not just take over an existing play and add the Countess scenes to it. Rather, those scenes were there from the beginning and Shakespeare actively collaborated on the creation of this as an original play.

    The first empirical evidence for Kyd's hand in Edward 3 that Freebury-Jones adduces is that its proportions of various rhyme patterns are like those of The Spanish Tragedy, Soliman and Perseda, Cornelia, King Leir, Arden of Faversham, Fair Em, and parts of 1 Henry 6 (p. 150). Freebury-Jones does not give the corresponding data for these rhyme patterns' proportions in other plays so it is impossible to tell how alike are the data for the allegedly Kydian plays. (And the numbers for the Kyd plays do not seem particularly consistent to me: they look like a random scattering from 0 to 30 occurrences across the plays.)

    Regarding the feminine-ending data, Freebury-Jones builds upon his previous attributions to make this new one, so that ". . . we might note that the range for long scenes in these portions of Edward III is strikingly close to that obtained for Kyd's scenes in Henry VI Part One" (p. 151). As observed above, tying new attributions together like this is perilous, since it encourages readers to either accept both new attributions or, just as easily, reject both.

    Freebury-Jones does the same with the pause-pattern data, relying on his attribution of part of 1 Henry 6 to Kyd to support his attribution of part of Edward 3 to Kyd. Freebury-Jones is right that ". . . the bigram 'no issue' appears in Kyd's (by my argument) unhistorical dramatisation of Mortimer's death in Henry VI Part One . . ." and in Edward 3 and in King Leir (p. 155) but wrong to find this significant since there are over 2000 occurrences in other books of the period; the phrase is just a commonplace.

    It would be pointless to continue drawing attention to these logical slips in Freebury-Jones's work: they apply everywhere in the book. He certainly has found some rare verbal parallels between the known works of Kyd and the plays he wants to add to the Kyd canon. The problem is that he has diluted this strong evidence with copious weak evidence and nowhere undertaken a systematic quantifying of what he has discovered.

    Ben Higgins's Shakespeare's Syndicate: The First Folio, its Publishers, and the Early Modern Book Trade adds to what we know about the stationers -- Edward Blount, William and Isaac Jaggard, William Aspley, and John Smethwick -- who published the First Folio of Shakespeare. This is obviously a matter of interest to book historians, but how much it should interest students in neighbouring fields such as textual criticism and literary criticism is up for debate. Higgins approves of the fact that ". . . recent work is testing the conventional boundaries between literary interpretation and book history" and that a certain kind of critic "grants dramatic character a fresh equivalency with stationer" (p. 14).

    The sensible objection to this claim is obvious. Any one of several stationers might have published a given Shakespeare play and it would be fundamentally the same play. But change the characters in the play and it becomes fundamentally a different play. I suppose Higgins's claim could just about stand if we confine ourselves to the most minor characters in the play. It does not matter greatly if, say, news of the Delphic Oracle's verdict comes from Cleomenes or Dion in The Winter's Tale.

    Higgins offers some useful summaries of his field, including the facts that in the 1590s there were about 250 books published a year and about 25 printshops active (p. 14). He speculates that Othello and the reprints of Richard 3 and 1 Henry 4 in 1622 were produced because their publishers were about to enter into agreements with the Folio syndicate not to publish for a certain period of time the plays they had rights to, and that John Smethwick's undated editions of Romeo and Juliet (1623) and Hamlet (1625) were undated because he wanted to evade such an agreement he had made (pp. 24-25).

    Higgins unintentionally offers an example of why a recent turn in book history is self-defeating. It has become fashionable to treat publishers rather than authors as the loci of meaning for early modern books, and hence to explore together the set of books published by each stationer. What emerges are claims for something like a recognizable style characterizing the output of each publisher. But if we try that with the Folio we have to imagine that the four piles that the final 750 exemplars were divided into -- one for the Jaggards, one for Edward Blount, one for John Smethwick, and one for William Aspley -- are four separate editions (pp. 29-32).

    Chapter One (pp. 41-78) is on Edward Blount as a literary publisher of works by Marlowe, Michel de Montaigne, and Miguel de Cervantes. Blount tended to publish longer and hence more expensive books than other publishers did, and at least at the start of his career he did not form partnerships with other publishers on big books to spread the risk. Blount earned a reputation for being more concerned with intellectual value -- because big books tended to be culturally important ones -- than with mere profit.

     Publishing the First Folio seems to have virtually ruined Blount, and ". . . his business never fully recovered" (p. 57). But Higgins points out that a large part of Blount's decline might also have been due to the considerable sum of 150 pounds he spend over several years around this time trying to recover a two-thousand pounds debt owed to him. We cannot be sure that Blount was simply bankrupted by lack of strong sales for the Shakespeare First Folio.

    Blount more often than other publishers added prefatory material to his books that he signed, usually making literary judgements about the book's contents. Higgins traces how Blount and his bookshop sat at the centre of an international network of literary culture, which high status was then transferred to Shakespeare when Blount published the Folio.

    Chapter Two (pp. 79-121) is about the Jaggard publishing house. William Jaggard was the only Folio publisher who was also a printer and of course the book was made in his, and his son Isaac's, printshop. Jaggard published The Passionate Pilgrim (1599) and he printed (and perhaps co-published) the Pavier Quartos in 1619, both of which editions are somewhat misleading about their own provenance, so an air of mistrust has hung over Jaggard's involvment in the Folio. Perhaps the Jaggards just printed it?

    Higgins surveys the evidence that in fact Jaggard was every bit as much a publisher as a printer for other men's publications. Moreover, Jaggard tended to diminish his visibility on the title pages of books he printed for other men and to emphasize it for books he printed for himself as publisher. That is, he marketed himself as a printer-publisher rather than as a printer for other men's books.

    Higgins sketches what we know of the publication of The Passionate Pilgrim, including Thomas Heywood's complaint after the third edition in 1612 about the misuse of Shakespeare's name, and the Pavier/Jaggard quartos. Higgins's key point is that Jaggard did not emphasize his involvement in the latter -- it took twentieth-century sleuthing to uncover it -- so there was no publicly obvious contradiction in his thereafter printing and publishing the Shakespeare Folio.

    The idea for the Folio having a table of contents, its 'Catalogue', was probably Jaggard's: his other books did this and other publishers' books generally did not. From 1615, Jaggard had the monopoly for printing playbills so his name was commonly associated with matters theatrical, which is what the Folio preliminaries, including the commendatory poems, are much concerned with.

    Ideas about deceptive printing practices that were held by the New Bibliographers were based on assumptions about the sole-authorship of the Folio that we now know to be false. That is, the Folio has a lot in it that is not by Shakespeare. Thus, argues Higgins, the allegedly deceptive practices of Jaggard in The Passionate Pilgrim and the Pavier/Jaggard quartos do not look so bad as they used to. Or at least, Jaggard's deceptive practices in the Folio -- and the deceptive practices of the others involved in the edition -- are more of a piece with Jaggard's earlier deceptive practices than we formerly realized (pp. 117-119). Of course, as Higgins notes, we cannot tell just what the compilers and publishers of the Folio actually knew about how much of the writing in it is not by Shakespeare.

    Chapter Three (pp. 125-169) is about William Aspley, who had the rights to, and had earlier published, Much Ado About Nothing and 2 Henry 4. It is odd that Aspley did not reprint these plays, since to judge from contemporary allusions they seem to have been popular on the stage. The lack of a reprint of 2 Henry 4 has given rise to the idea that it was something of a failure when compared to its wildly popular Part One. Perhaps, suggests Higgins, Aspley was simply not much interested in these plays for some reason. The main publishing activity of Aspley's career was to put out a great many editions of the writings of the conformist preacher John Boys, which Higgins details.

    Chapter Four (pp. 170-210) is about John Smethwick. That he was not closely connected to the actual printing of the Folio is suggested by his name being given, unusually for him, as "Smithweeke" in the colophon. Presumably he was not around to correct this. Smethwick had the rights to Romeo and Juliet, Hamlet, Love's Labour's Lost, and The Taming of the Shrew. There is really not much to say about Smethwick, so Higgins spins it out with material about the parish of Saint Dunstan in the West where Smethwick's bookshop was located, and about understanding the whole book trade through the lens of geography, attending to where in London the printshops and bookshops were located.

    Higgins's "Epilogue" is about the women in the book trade and naturally has little to say since the profession was almost exclusively male. In three appendices he presents the texts of the letters of Edward Blount, a list of the publications of William and Isaac Jaggard, and the locations in London where Shakespeare's books could be bought wholesale.

    Molly Yarn's monograph Shakespeare's 'Lady Editors': A New History of the Shakespearean Text builds upon and expands the research presented in her essay "Katharine Lee Bates and women's editions of Shakespeare for students" reviewed in NYWES for 2020. Unfortunately, Yarn is implacably opposed to the idea that ". . . the traditional textual work of collation and emendation is the most serious and important element of editing . . ." (p. 17). For Yarn, ". . . assigning value to editions based on the originality of the textual work behind them privileges the single (usually white, male) editor as a surrogate author-figure" (p. 125).

    Yarn comments that

The primary goal of editing, for [Fredson] Bowers, is establishing a text. This is clearly an extremely narrow and limiting view of editorial work. 'Practical' editors may not produce original textual work, but surely the person who guides the reader and directs her attention is at least as important as the person who decides whether the word should be 'Indian' or 'Judean'. (pp. 130-1)

Yarn here confuses the labour with the labourer, since the issue is not the importance of the person but the importance of the work. The present review is founded on a principle that Yarn rejects -- that deciding what the author wrote (that is, establishing the text) is indeed the prime editorial activity -- so her book has nothing of interest for us.

    There are no collections of essays to review this year. Terri Bourus's Shakespeare and the First Hamlet is just a repackaging of the special issue of the journal Critical Survey on the topic of "Canonizing Q1 Hamlet" that was reviewed in NYWES for 2019. The most technically complex of this year's articles is Douglas Bruster and Anna Christoffersen's argument that the chronology of Shakespeare's works can be refined by use of Eliot Slater's rare-word linkage method applied across his canon ("Shakespeare's Rare Words and Chronology", SQ 72[2022] 279-324).

    Bruster and Christoffersen explain that 'rare' here means that a word is seldom used by Shakespeare, and their central idea is that works Shakespeare wrote around the same time share rare words because these words were disproportionately uppermost in Shakespeare's mind at that time. Their study necessarily excludes the extremely seldom-used words since these by their nature will not appear in multiple works and hence cannot serve as 'link-words' between different works.

    Hugh Craig has charted how certain words became more popular and others less popular across the 1590s and 1600s. Despite this, rare words demonstrably cluster at certain times in the Shakespeae canon and hence can be used to date the works. Naturally, the Shakespeare plays thought to be collaboratively written are split by Bruster and Christoffersen into the bits by and not by Shakespeare.

    Aside from some custom-made transcriptions for works not covered by it, Bruster and Christoffersen's base text is The Riverside Shakespeare Second Edition "Because it is well supported through the Spevack Concordance" (p. 282). This seems a poor justification for choosing a base text, since for any complex computational analysis of text you might as well make your own concordance while you are at it.

    By moving The Two Gentlemen of Verona to 1594, Bruster and Christoffersen find that all early Shakespeare is collaborative. They also find Shakespeare writing the Fly Scene in Titus Andronicus, and the Additions to The Spanish Tragedy. They find that The Taming of the Shrew, 3 Henry 6, Romeo and Juliet, Hamlet, Troilus and Cressida, Measure for Measure, All's Well that Ends Well, and Macbeth "were drafted in one year, then revised and/ or augmented in still others" (p. 282). Shakespeare wrote non-collaboratively only for 10 years in the middle of his career but these were by far his most productive years.

    It emerges from the description of their method that Bruster and Christoffersen do not study lemmatized forms of words but simply strings (including punctuation). They define how they count links so that the "5-word" link word 'orphans' counts as five because it appears once in Sonnet 97, once in Macbeth, once in the Shakespearian part of 3 Henry 6, and twice in Richard 3 (p. 283). Like Slater they categorize rare words into groups by how many links they form, and they use 2-to-6-link words and 7-to-11 link words, noting as Slater did that the former set are the most discriminating of chronology.

    Bruster and Christoffersen provide an impressively clear explanation of their use of Jackson's modification of Eliot Slater's way of counting and comparing rare-word usage and of the chi-squared method for measuring how far observations of rare-word usage differ from what our expectations would be if the rare-words were simply randomly scattered across the Shakespeare canon. (Jackson's modification of Slater's work was reviewed in NYWES for 2015.)

    But Bruster and Chistoffersen do not use exactly Jackson's formula for the expected number of links between two plays that we can call P and Q. The expected number gives the baseline for how many links we should find if closeness of dating is not causing P and Q to have additional links. Jackson took the number of rare-word links in P and divided this by the result of subtracting the number of rare-word links in P from the number of rare-word links in the whole Shakespeare canon, and then he multiplied this by the number of rare-word links in Q.

    Bruster and Christoffersen instead simply divide the number of rare-word links in P by the number of rare-word links in the whole canon and multiply the result by the number of rare-word links in Q. That is, they omit the step where Jackson's denominator first subtracts the number of rare-word links in P from the number in the whole canon. Bruster and Christoffersen report that they do this "to achieve symmetrical results" (p. 284), but they do not explain what symmetry means in this context.

    Usefully, Bruster and Christoffersen provide a worked example of how they use the rare-work links in The Two Gentlemen of Verona to date that play as being closest to Lucrece, which is the work with which the chi-squared calculation above produces the largest value. That is, the number of observed links between The Two Gentlemen of Verona and Lucrece far outstrips the number of expected links, and by more than is found between The Two Gentlemen of Verona and any other Shakespeare work. Because it squares the result of a subtraction -- specifically, the number of observed links minus the number of expected -- the chi-squared test gives a positive result even if the number of observed links is less than the expected, but Bruster and Christoffersen leave such outlier results out of their tables.

    Bruster and Christoffersen explain their charts' nomenclature for labelling texts, so that "TNKsh" is the Shakespearian part of The Two Noble Kinsmen, and "TROrev?" is "a hypothesized revision". I am not sure I understand the latter point about the revision of Troilus and Cressida.  I think (but may be mistaken) that the '-rev' suffix is used when two works not close in time give roughly equal numbers of matches of rare-words for a given text being tested. In such a case, Bruster and Christoffersen invent a revised version in their chart and place it near the second of the two matches in this "bimodal distribution" as they call it (p. 286)].

    Bruster and Christoffersen acknowledge that genre tends to be a confounding variable in attempts to date works by internal evidence, but they have no way to mitigate against that. To explore how the language preferences of all authors drift collectively over time, Bruster and Christoffersen chart how each Shakespeare play differs from Kyd's parts of The Spanish Tragedy and as expected Shakespeare's early plays are much more like Kyd's play than his late plays are.

    Bruster and Christoffersen go on to use this as a general yardstick, expecting a Shakespeare's play's difference from The Spanish Tragedy to serve as a proxy for its date of composition, although their own Figure Two shows that this measure correlates with date only loosely. For instance, their method suggests that Venus and Adonis and Lucrece are alike (sharing lots more rare words than we would expect by chance) and so according to Bruster and Christoffersen they were written at almost the same time, yet Venus and Adonis has one of the lowest likenesses to The Spanish Tragedy shown in Figure Two and Lucrece has nearly the highest. If being unlike The Spanish Tragedy were a good proxy for a play's date we should not expect to see this. This problem vitiates their use of comparisons wih The Spanish Tragedy throughout the essay.

    Before they started applying their method to the testing plays of uncertain dating, I would have expected Bruster and Christoffersen to take some plays with fairly certain datings derived in part from external evidence and test those by their method to see if their predicted dates from this method match their known dates. Bruster and Christoffersen mention no such validation of their method.

    Bruster and Christoffersen test the 684-word Fly Scene in Titus Andronicus and find that it dates to "a general location in the mid-1590s" (p. 289). This scene is of course of disputed authorship. If Bruster and Christoffersen compared it only to Shakespeare's works and it really is by Middleton, then finding this likenesss to mid-1590s Shakespeare means only that this bit of Middleton writing is, of all the Shakespeare works, most like the mid-1590s Shakespeare works; this tells us nothing about authorship or dating.

    Discussing possible revision of 2 Henry 6, Bruster and Christoffersen make no mention of there being two substantive versions of the play (the 1594 quarto and the 1623 Folio) and their being so different from each other, which is the basis of the argument for revision. Unless you have two texts, how can you have evidence for revision? And yet it seems that from just one text, Bruster and Christoffersen feel they can produce two datings: "Owing to its chi-square figures, we locate 3H6sh in two positions: an original date of composition and then a hypothesized revision" (p. 290).

    I would have thought that when you get two matches for the dating of one text, that is perhaps a sign that your experiment is not producing conclusive results, and the text having differently aged strata due to revision might be the cause of that. It does not seem right to accept the doubled results as conclusive of revision. Rather, the next step is to try to separate the strata and date each independently.

    Considering the Shakespearian and non-Shakespearian parts of Arden of Faversham Bruster and Christoffersen again seem to assume that authorship is not a confounding variable for dating, so that ". . . even if written by different hands, both parts of Arden of Faversham were composed closer to its publication in 1592" (p. 290) than to the period 1588-90 found by other investigators, to judge from its likeness to Shakespeare's works around 1592.

    Bruster and Christoffersen bring in genre as a confounding variable for certain cases in order to explain likenesses that their method would seem otherwise to be saying is due to closeness of dating. Thus with A Lover's Complaint they write that its appearance "in the 2-6 associations [with Venus and Adonis especially] may come from its shared representational mode (narrative verse) and complaint theme" (p. 292). If sharing a genre is an explanation for this case, why not for other cases? (By "2-6 associations" they mean they are concentrating on words appearing between twice and six times across the canon.)

    The confounding variable of genre can be wheeled in whenever Bruster and Christoffersen's method produces results they do not like:

Note that the chart for 3H6sh (see figure 6) also flags a strong association with R3. If this is not merely the product of shared subject matter, it may indicate either (a) an earlier phase of composition for R3 or (b) a later revision of 3H6. (p. 15)

These three hypotheses -- actual closeness of dating, shared subject matter, and revision -- all fit the evidence and the method is not capable of distinguishing between them. This looks like the familiar error of over-working a classifier that has insufficient power to make the discriminations the investigators are trying to make.

    This exploitation of the confounding variable is recurrent:

Though Titus feels like a remarkably early association, perhaps challenging our placement of SPTAD [the Additions to The Spanish Tragedy], it is possible to argue that the genre of Richard III, like that of Titus, is close enough to revenge tragedy and Senecan practice that its verbal overlap with SPTAD comes from tradition and form. (p. 17)

The remainder of the article picks through the results for each period of Shakespeare's career, but it is not worth going through them if you are not convinced -- as I am not, on the evidence presented -- that Bruster and Christoffersen have created a working classifier.

    Bruster and Christoffersen's narrative is full of special pleading and speculations, such as "The appearance of four tragedies . . . in the three The Merchant of Venice chi-square tests is harder to explain, but fits with the play's reputation as a generically mixed drama" (p. 301), and "the association with . . . may come from Shakespeare's revision to that text prior to its 1599 publication" (p. 303), and  "The pronounced association with . . . may derive from shared subject matter" (p. 303).

    In their conclusion, Bruster and Christoffersen repeat the claim that if two Shakespeare works that we are confident are widely separated in time are the best matches for a given work under examination, then this is evidence of "phased composition" (p. 324). In fact it may just be noise.

    The only other substantively relevant article from Shakespeare Quarterly this year was Gary Taylor's argument that Folio Henry 5 (hereafter F) reflects a revision of the play that lies behind the 1600 quarto (Q1), and most likely this was a revision that Shakespeare undertook between first reading the play to the company and its licensing ("Shakespearean Magnitudes", SQ 73[2022] 246-79). Q1 Henry 5 is remarkably short at just 12,272 dialogue words. Taylor takes the lengths of the 10 Folio history plays and treats these as a normal (Gaussian) frequency distribution and asks how often we would expect to sample a value as low as 12,272 dialogue words from this distribution. The answer is that this low value is far over on the low side of the distribution and getting a value this low or lower will happen in 1 out of 500,000 random samples from the distribution.

    This Taylor reports as there being "only one chance in 500,000 that The Chronicle History [= Q1 Henry 5] belongs to the same set as Shakespeare's ten Folio histories" (p. 246), but that is not what such as test can tell us. The outcome of such a test does not tell us the probability of a sample coming from a population but how frequently or infrequently we would get such an extreme sample from that population. (These sound like the same thing but are not.)

    Moreover, it is not clear that the premises of this test are coherent. What would it even mean for this edition to be a sample from this population, since we know that Q1 Henry 5 is not a sample from the population of Folio editions of the history plays? Taylor builds a larger population and repeats the test, finding that there is a "one chance in 250 that it [Q1 Henry 5] belongs to the same set as the thirty-six canonical playtexts" (p. 246). Again, the premise of the test is not coherent and Taylor's conclusion is not what such a test is capable of telling us. He is of course right that the length of Q1 Henry 5 is extremely anomalously low, but these statistical tests are not the way to express that.

    F is just over twice as long as Q1, with a difference more than that of any other play's quarto/Folio comparison. But is the difference due to the version underlying F being shortened to make the version underlying Q1, or the version underlying Q1 being expanded to make the version underlying F? Q1 famously lacks the Chorus, who identifies himself as an actor in the troupe, and it has much less for Pistol to say although we know that he was a popular character. Q1 also lacks Nim's allusions to the 'humours' play craze of the late 1590s and the big speeches for Henry that everyone remembers.

    These are all things it is hard to see being removed from the play underlying F. Andrew Gurr and Erne indpendently argued that Q1 Henry 5 reflected what the performers did to Shakespeare's over-long script, which long version is represented in the Folio. Taylor finds it inherently implausible that Shakespeare's fellow actors would omit to perform more than half of what he wrote, and points out that in the Folio preliminaries John Heminges and Henry Condell claim that what is in the book has stood the test of performance.

    Taylor studies the Q1/F variants that arise from Shakespeare's use of his sources, and he finds that most often Q1 is nearer than F to the source, suggesting a Q1-to-F transmission. That is, it is hard to think of a means by which Q1 was derived from F during which it got closer to the play's sources. Richard Dutton pointed out that none of the material that Folio Henry 5 has that Q1 lacks comes from the source play The Famous Victories of Henry the Fifth, which is odd if Q1 is a cutdown version of F.

    But the F-only passages are dependent on the prose chronicles, so perhaps to expand the play Shakespeare returned to those because, unlike The Famous Victories of Henry the Fifth, they had much material he had not mined. In Q1 and in all the sources the Dauphin is not at Agincourt but Bourbon is, whereas F gives Dauphin a prominent role at Agincourt and gives him Bourbon's lines. This makes sense if Q1 is simply following the sources and F reflects subsequent revision to make a place for the Dauphin, which certainly improves the play. Regarding the English nobles, F departs from historical reality considerably more than Q1 does. The motivation seems to be to mention characters from the Henry 4 and Henry 6 plays.

    If we accept that 1623 Folio Henry 5 represents Shakespeare's own revision of the play represented in the 1600 quarto, we are bound to ask when he revised it. Taylor explores various possibilities, including that the revision followed hard upon the first performances (which used something like the 1600 quarto version as its script) and happened because the play was unsuccessful at first.

    Taylor's preferred hypothesis is that revision took place before the play was sent to the Master of the Revels for licensing, perhaps because the company did not like it at first reading. After the revision, the manuscript of the first version would not be needed for anything. One possible objection to this reasoning is that if Shakespeare's revision made the play so much better, would the Chamberlain's Men not have objected to the vastly inferior version being published? In an appendix, Taylor lists all the variants that underpin his argument.

    Laurence Publicover's essay in Shakespeare Quarterly is largely a literary-ecocritical argument about Pericles, but it touches on some textual matters ("Pericles's Humming Waters: Nonhuman Agency, Textual Criticism, and the Practice of Material Ecocriticism", SQ 73[2022] 280-302). In the 1609 quarto, on sig. E2v, the First Mariner tells Pericles that the dead Thaisa must be thrown overboard and the following exchange ensues:

    Per. That's your superstition.

    1. Pardon vs, sir; with vs at Sea it hath bin still obserued.
And we are strong in easterne, therefore briefly yeeld'er,

    Per. As you thinke meet; for she must ouer board straight:
Most wretched Queene

Most editors think that "for she must ouer board straight" is really part of the First Mariner's speech and meant to follow "therefore briefly yeeld'er".

    Publicover argues instead that at this point "the prince begins to think and speak as a sailor" (p. 292) and to accept the tradition of throwing the dead off a ship. As Publicover acknowledges, the first word of the disputed phrase, "for", is "difficult to comprehend" in Pericles's mouth, but perhaps "the prince speaks what is, in part, the sailor’s thought" (p. 291) as he becomes one. This seems a long way to go to avoid acknowledging a simple error in the quarto.

    In the threnody that Pericles gives Thaisa ("A terrible Child-bed . . . farewell to her") there are some odd phrasings that Publicover thinks editors have not been able to emend into coherence. But that is okay because "Shakespeare may be aiming here for a speech whose internal incoherence evokes its speaker's mental turmoil" (p. 298).

    In a second essay on the provenance of the 1600 quarto of Henry 5 (Q1), Taylor shows that it shares errors with the Folio version (F) that must be transcriptional, so Q1 cannot be based on a memorial reconstruction or a script taken down by a reporter of the early performances ("Play Manuscripts, Vectors of Transmission, and Shakespeare's Henry the Fifth", PBSA 116[2022] 343-78). Q1 was one of the five 'bad' quartos identified by R. B. McKerrow in 1909, the others being the 1597 Romeo and Juliet, the 1602 The Merry Wives of Windsor, the 1603 Hamlet, and the 1609 Pericles. W. W. Greg's The Editorial Problem in Shakespeare found only four Folio plays to be printed from Shakespeare's foul papers: Henry 5,  The Comedy of Errors, King John, and Timon of Athens. That makes Henry 5 the only play where we can -- according to the New Bibliography -- compare the good edition printed from foul papers with a 'bad' quarto.

    Virtually everybody, including the younger Taylor, accepted this binary opposition between the two versions: F being essentially authorial and Q1 essentially not authorial but rather based on performance. This went along with a chronology: the manuscript behind the F ("The Life") being created before the manuscript behind Q1 ("The Chronicle History"). The binary opposition has survived the collapse of Pollard's argument about the 'bad' quartos being surreptitiously printed, and of the theory of memorial reconstruction, and of the foul-paper/promptbook binary.

    For Taylor, what now finally undermines the standard binary-opposition model of Henry 5 is the names of two women in the play, Prince Katherine and her woman Alice. Princess Katherine appears in the source play The Famous Victories of Henry the Fifth. In Q1 and F, Katherine's woman gets given the name Alice in speech prefixes in the English-lesson scene and is then given the name Lady in speech prefixes in the final scene. The name Alice cannot have come from The Famous Victories of Henry the Fifth since there Katherine's attendants never speak.

    Could F have got the name Alice from consultation of Q1 during typesetting of the Folio? This seems unlikely, since Q1 gives the Princess the speech prefix "Kate." and F uses "Kath." or "Kathe.". Also, Q1 has only 21 speeches in the language-lesson scene where F has 35. We would expect consultation of Q1 during the making of F to create greater consistency than this. Similar Q1/F differences appear in the final scene, involving the Princess and her woman.

    Since Alice in the language-lesson scene is the Lady of the final scene, we have here Q1 and F agreeing on giving a character one form of her name (a personal form) in one scene and a different form (a social-rank form) in a later scene, without F getting this directly from the quarto by consultation. Also, whereas Q1 has the name in her entrance stage direction in the language-lesson scene match her speech prefix (both being Alice), F has her entrance call her "an old Gentlewoman" while her speech prefixes are "Alice". In the final scene, F does not name her in the stage directions but Q1 calls her "gentlewoman" (disagreeing with her speech prefixes of Lady). Thus both versions at some point call her "gentlewoman" in a stage direction that disagrees with her speech prefixes.

    Thus by New Bibliographical lights, Q1 and F look like they were printed from foul papers that had authorial variations of a name in them. If we now accept (from Werstine) that playhouse manuscripts could have such variations of name in them, then Q1 and F could both have been printed from playhouse manuscripts. Their likeness means we cannot say that Q1 is bad compared to F.

    The variation in names that Taylor traces -- particularly between the language-lesson scene and the final scene -- cannot have derived from a notetaking spectator nor from a junior actor making a memorial reconstruction: neither would have seen a document with this character's speech prefixes in it. Rather, there must be a textual transmission connection between the manuscript underlying Q1 and the manuscript underlying F. In short, "Either (a) one of the underlying manuscripts must have been the ancestor of the other underlying manuscript, or (b) both underlying manuscripts derive from a shared manuscript ancestor that contained the inconsistency" (p. 356).

    The forms of the speech prefixes in Q1 are nearer than F's to those in the source play, The Famous Victories of Henry of the Fifth, in that both use "Kate." where F uses "Kath." and "Kathe." In F's version of the language lesson there is a run of speech-prefix errors -- involving unwanted switches of speaker -- that are hard to explain by the usual lapses in transmission but which make sense if we think of the quarto's version as the original version of the scene and the Folio's text as a revision in which new speeches were added.

    There is a run of stage-direction errors that link Q1 and F. Both omit to bring attendants on with Henry and the three traitors in the scene where he exposes them, yet attendants are needed to take the traitors away, so this is a shared error. The other cases are where Q1 has a stage direction that makes sense in relation to its dialogue or action and F has the same stage direction but it does not make sense in relation to the dialogue or action.

    Taylor thinks this is hard to explain unless F inherited the stage directions unchanged from Q1 even though F had revised dialogue or action. We might object that perhaps Q1 is the more theatricalized version in which stage directions conflicting with dialogue/action were fixed in rehearsal, but this would require that the fix was to change the dialogue/action when of course the simpler fix is to rewrite the stage directions. Moreover, in these stage directions, Q1 is mostly closer to the sources than F is, which implies Q1-to-F progression.

    Q1 contains only two sound cues while F has 19 sound cues that Q1 lacks (mainly in scenes the quarto also has), which itself suggests that Q1 is based on authorial papers and F on a script that has been through the processing of readying for performance. Taylor picks through the Q1/F variants that Duthie thought were conclusive of Q1 being a memorial reconstruction and finds that in fact they might reasonably be attributed to intentional revision.

    For instance, Q1 has "To his vnfurnish nest the weazel Scot| Would suck her egs" where F has "To her vnguarded Nest, the Weazell (Scot)| Comes sneaking". In Q2 and Q3, set from Q1, "vnfurnish" got its obvious correction to "vnfurnisht". Duthie thought that "vnfurnish[t]" was a memorial reconstructor's anticipation of "vnfurnisht Kingdom" that (in all editions) occurs later in the play, but Taylor defends it as a perfectly reasonable adjective here: an unfurnished nest is one that is not armed against attack. Duthie thought that Q1 having the unusual word "vnfurnisht" twice was inherently suspicious, but Taylor shows that F also has many such repetitions that have not aroused suspicion.

    In Q1, Flewellen recounts the bravery of a man in Exeter's group who defended a bridge but says he cannot remember the man's name ("I do not know how you call him") but when Gower asks "How do you call him?" Flewellen gives his name as Pistol. F does not have this contradiction because there Flewellen never says he cannot remember the man's name. Supporters of memorial reconstruction see Q1 here as the result of mangling by the reconstructor, but Taylor suggests that it is a case of Shakespeare's second stab at an idea existing alongside his first one, as with other well-known false starts in his plays. Taylor traces other moments where anomalies in Q1 might be authorial false starts.

    Q1 and F have Pistol fondly and sadly give bad news about Doll Tearsheet -- that she is sick in one, dead in the other -- and reflects that he is thereby financially ruined. However, in both versions he says near the start of the play that he had married Mistress Quickly (not Doll) and in both in the middle of the play he is disparaging about Doll as disease-ridden. This baffling reference to Doll in Q1 and F, then, is a shared error. From all these shared errors, Taylor argues, we have to accept that the link between Q1 and F Henry 5 is transcriptional.

    An article by Pervez Rizvi published in 2021 was omitted from last year's round-up because it was the subject of a rebuttal being written by this reviewer that was under peer review ("The Use of the T-test in Shakespeare Scholarship", DSH 36[2021] 712-8). The journal Digital Scholarship in the Humanities has now, in July 2026, published this reviewer's rebuttal (DOI 10.1093/llc/fqag106) so Rizvi's article about what is called a t-test will be briefly noticed here.

    The t-test is an important and commonly used statistical tool that produces a probability known as a p-value. In computational stylistics, a t-test is often used to make sense of the data arising from counting the frequencies of certain words in different texts. These frequencies may vary across different texts from different eras, in different genres, and by different authors.

    Rizvi argues that Hugh Craig's essay "Authorial Attribution and Shakespearean Variety: Genre, Form and Chronology" (reviewed in NYWES for 2017)  misused the t-test in an attempt to show that in early modern plays word-choice frequencies vary more significantly across different authors than they vary across writings in different genres by one author. The rebuttal by the present reviewer shows that Craig in fact used t-tests responsibly, that Rizvi’s essay misrepresents this branch of statistics, and that Rizvi recurrently commits a well-known statistical 'howler' -- his claim that a p-value tells us how likely it is that our null hypothesis is true -- that vitiates his entire critique.

    The field of Shakespeare studies is currently divided on certain recent authorship attributions. This year there were substantive journal articles on only one of these cases -- the authorship of Arden of Faversham -- and quite a few on the validity of various technical methods used to attribute authorship on the basis of a text's internal features.

    Rizvi's first article this year is on both topics: he thinks that Taylor made a methodological slip in his claim that Watson wrote part of Arden of Faversham ("The Unsoundness of the Stylometric Case for Thomas Watson's Authorship of Arden of Faversham", ANQ: A Quarterly Journal of Short Articles, Notes and Reviews 35[2022] 261-70). Rizvi's aim is to show flaws in the method used in Taylor's essay "Finding 'Anonymous' in the Digital Archives: The Problem of Arden of Faversham" (reviewed in NYWES for 2019). The problem, according to Rizvi, is that Taylor scaled the number of Watson's verbal matches with Arden of Faversham to compensate for the small size of Watson's dramatic canon and that the resulting large multiplier makes it certain that Watson will be found to be the author of any text with which his canon has a match.

    Rizvi's point about scaling is correct, but it is not clear why Rizvi thinks this makes Watson bound to win in Taylor's tests. It was entirely possible for Watson's tiny canon to have no matches with Arden of Faversham, and indeed that is what we would expect with any small canon. Moreover, the principle of scaling hit-counts by size of the authorial canon is one that Rizvi has explicitly advocated and on his Collocations and N-Grams website he provides the formula for doing this that Taylor uses.

    Rizvi's current formula for doing it is not the one that Rizvi first developed. In response to a discussion in 2018 with the present reviewer about how best to write such a formula, Rizvi changed the one he uses. The present reviewer does not think anyone has come up with a satisfactory formula that fairly compensates for different sizes of authorial canons across our authorship candidates.

    Other objections to Taylor's essay made by Rizvi include that his notion of a word or phrase being unique was poorly defined: "Does he mean unique within the works written by the candidate authors in the date range 1585 to 1594 inclusive; or unique within the canons of all authors in all periods; or something between those extremes?" (p. 264). It is unclear why Rizvi needs to raise this question since he finds the answer for himself in Taylor's essay: ". . . he is taking 'unique' to mean unique only within the works of those authors" in his candidate list "and only within his date range" (p. 264).

    Instead of Taylor's approach, Rizvi thinks that "True uniqueness or rarity should be judged by checking how common a collocation was in the literary culture generally" (p. 264). This is fair enough, and someone could indeed do a different experiment from Taylor's using such a different experimental method. But that is not itself a critique of Taylor's experiment.

    Rizvi faults as too convoluted the rules Taylor applied for sifting his evidence, such as Taylor's "In the absence of an exact match, I counted unique matches of variant forms . . ." (Taylor's p. 860). Rizvi also objects to the inconsistency of Taylor's ". . . where possible, the corpus is limited to works written in the decade 1585-94 . . . . For some candidates, however, it has been necessary to extend the date range" (Taylor's p. 857; Rizvi's p. 265). Ultimately Rizvi thinks that none of this matters since "the mathematical near impossibility of anyone beating Watson" (p. 265), arising from the scaling needed to compensate for Watson's small canon size, trumps all other considerations.

    Taylor counted sub-matches within matches such that a match of three words X Y Z between two texts also counts as a pair of matches of words X Y and Y Z between them (so three matches rather than one). As Rizvi admits, by this means "Taylor inflates everyone's totals, not just those of his preferred candidate" (p. 265) and Rizvi offers no reason to suppose that this inflation favours Watson's authorship. Indeed, it is not really inflation at all: it is just a particular way of counting matches in a situation where there are multiple possible ways we could count them.

    Rizvi claims that Taylor chose for testing the particular 274-word passage of Arden of Faversham that he did, the start of Scene 10, because he knew that it would produce matches with Watson's work. That is, according to Rizvi, Taylor cheated: "Clearly, the passage was picked because its first three lines happen to contain some matches with [Watson's] Elvetham" (p. 266). Rizvi offers no evidence for this impugning of Taylor's scholarly probity.

    Taylor gave his reasons for selecting this particular passage from Arden of Faversham: i) it is not influenced by the sources, ii) no one has claimed it is by Shakespeare, and iii) it has no entrances and exits. Each of these in turn Rizvi rejects, replying regarding (i) that in that case the matches in other writers' works should also not be influenced by sources, and regarding (ii) and (iii) that there are plenty of other passages in Arden of Faversham that meet each of these criteria. Rizvi here misses the point, since what Taylor needed was a passage that meets all three criteria at once, and these are not abundant in Arden of Faversham.

    Moreover, Rizvi's response to (i) that the same criterion (of not being influenced by sources) would have to apply to the works in which the match is found is simply a misunderstanding. Taylor was observing that since the true story of Arden's murder appears in multiple books of the period we might get spurious matches where Arden of Faversham reuses phrasing from its sources and it were better to choose a part of the play wholly invented by the author of that part.

    Rizvi next objects to Taylor's treating Watson's Elvetham as a play and his bringing into consideration Watson's non-dramatic works too. As Rizvi rightly claims, no one knows for sure that using non-dramatic writings to attribute plays actually produces reliable attributions, but Rizvi offers no evidence against this approach.

    On discrepancies in Taylor's arithmetic Rizvi cites an essay by Freebury-Jones called "Unsound Deductions in Early Modern Attribution: The Case of Thomas Watson", reviewed in NYWES for 2020. In that review it was noted that Freebury-Jones's essay plagiarizes an earlier review of Freebury-Jones's work written by the present reviewer. It is gratifying to be able to report that in response to legal action brought by the present reviewer, the publisher Taylor & Francis has now inserted into the online version of Freebury-Jones's essay a reference to my research that Freebury-Jones formerly presented as if it were his own.

    Rizvi ends with the observation that since a single unique match between an extract from Arden of Faversham and Watson's writing is all it takes for Taylor's method to attribute that extract to Watson, we are entitled to look for other plays that have a single match to Watson's writing and treat them as anonymous plays to be attributed. Rizvi points out that Shakespeare and others' 2 Henry 6, Marlowe's 2 Tamburlaine, and Kyd's Cornelia all have a single match to Watson's writing and hence ". . . Taylor's algorithm attributes them all to Watson" (p. 269).

    According to Rizvi this procedure "is a standard validation process in attribution studies; that is, to treat a play of known authorship as anonymous and check who it is attributed to by the method" (p. 269). But that is not at all what Rizvi does here, since he sets out looking for plays that have one match with Elvetham. Rizvi does not start with an extract from Arden of Faversham, as Taylor did, and see whose writing is most like it, but rather he starts from Taylor's endpoint -- that Elvetham is the writing most like the Arden of Faversham extract -- and works his way back to others' writings, looking for matches.

    Working this way and with many plays to choose from, it is unsurprising that Rizvi finds three that have such a match to Elvetham; it would be surprising if he did not. Rizvi's procedure is rather like a familiar mistake of police work. Instead of starting with all the suspects and by a process of elimination finding the one with most evidential links to the crime, incompetent detectives start with a single suspect and see which pieces of evidence are consistent with her guilt. That this approach is self-deluding is now widely understood.

    In NYWES for 2021 this reviewer remarked that ". . . the journal EMLS [Early Modern Literary Studies] seems defunct, having published no issues since 2019" and hence that an article by Freebury-Jones that Vickers described in 2021 as forthcoming in EMLS had not appeared. In fact, the problem seems to be that the website of the journal EMLS is broken and nothing that the journal published after 2019 can be accessed on it.

    The Internet Archive's miraculous Wayback Machine captured a snapshot of the EMLS website from before it broke and this reveals Freebury-Jones's article ("Elevating Thomas Watson: An Investigation Into New Authorship Claims", EMLS 22.i[2021] n. pag.). Readers wanting to read Freebury-Jones's article, reviewed now, should get a copy from the Wayback Machine.

    Like Rizvi, Freebury-Jones disputes Taylor's recent claim that Watson wrote most of Arden of Faversham and maintains his view that Kyd did. Freebury-Jones starts with a survey of the authorship attribution debate around the play, claiming that "a consensus of scholars recognized him [Thomas Kyd] as sole author in the twentieth century" (p. 1). But in fact not all the scholars named by Freebury-Jones as members of this consensus were as certain about it as Freebury-Jones is.

    On the very page of Philip Timberlake's book that Freebury-Jones cites in support of his claim that Timberlake "recognized him [Kyd] as sole author" of Arden of Faversham -- that is, page 53 of his The Feminine Ending in English Blank Verse -- Timberlake wrote that Kyd certainly "copies Marlowe's verses in Soliman and in Arden (always granting that these are Kyd's)". That sounds distinctly like Timberlake sitting on the fence.

    Freebury-Jones complains that in his essay "Shakespeare, Arden of Faversham, and Four Forgotten Playwrights" (reviewed in NYWES for 2020), Taylor neglected to mention that ". . . every academic publication he [Taylor] cites claiming to eliminate Kyd’s authorship of the play has been challenged or refuted" (p. 3). This cuts both ways, since Freebury-Jones here neglects to mention that every recent publication he cites supporting Kyd's authorship has been challenged or refuted.

    As he does with Timberlake's book, Freebury-Jones assigns to people positions that they simply do not hold. For instance, Freebury-Jones reports that "The methodological issues in Brett Greatley-Hirsch and Jack Elliott’s chapter in the New Oxford Shakespeare: Authorship Companion [on the authorship of Arden of Faversham] have been scrutinized by . . . Joseph F. Stephenson . . ." among others (pp. 3-4). They have not.

    Stephenson's review of the New Oxford Shakespeare: Authorship Companion that Freebury-Jones cites makes no mention of "The methodological issues" in Greatley-Hirsch and Elliott's chapter. Indeed, Stephenson found Greatley-Hirsch and Elliott impressively tentative about the authorship of the play, in contrast to Jackson, and found them so broadminded that ". . . evidence [is] included in the chapter suggesting Kyd or even Marlowe as the author" of parts of it (Sixteenth Century Journal 14 (2018) page 1316).

    Freebury-Jones summarizes what we know about Watson and his works and then turns to Taylor's claims about Watson's life experiences suiting him to write Arden of Faversham. Freebury-Jones finds that Kyd's life contained experiences that likewise suited him to write the play, for instance his experience translating Torquato Tasso's Padre di Famiglia as The Householder's Philosophy. Freebury-Jones has a lot of these 'and so did Kyd' cases and of course they do not settle the authorship debate. Instances include Taylor observing that Watson was celebrated for his skill in comic and tragic modes, to which Freebury-Jones replies that Jonson's reference to "sporting Kyd" in the Shakespeare Folio preliminaries indicates that Kyd was known for comedy too.

    Halfway through his essay, Freebury-Jones makes his first specific language-use claim. He finds that Kyd's character Bel-Imperia in The Spanish Tragedy uses the phrase "Is this the" twice to start successive lines ("Is this the love thou bear'st Horatio?| Is this the kindness that thou counterfeits?) and so does Arden of Faversham's character Alice ("Is this the end of all thy solemn oaths?| Is this the fruit thy reconcilement buds?"). The trouble, of course, is that "is this the" is a commonplace of the language with over 12,000 occurrences in EEBO-TCP, and over 30 in Shakespeare's works.

    What is more, in Edward 2 Marlowe has King Edward use the same phrase the same way that Bel-Imperia and Alice do, to start successive lines: "Is this the love you bear your sovereign?| Is this the fruit your reconcilement bears?" And indeed, the second of King Edward's lines is identical to Alice's line except for the final word: her "buds" for his "bears". The evidence that Freebury-Jones here adduces points more strongly to Marlowe's hand in Arden of Faversham than to Kyd's.

    Freebury-Jones next looks at Taylor's 'microattribution' technique of counting rare phrases found in the text to be attributed and in the works of each candidate author, and seeing which candidate has the most matches. Freebury-Jones finds this an "arbitrary and unrepresentative method" (p. 11) but in fact it is essentially the same method that Freebury-Jones uses, not least in his claim, just made, that Bel-Imperia and Alice both using the phrase "is this the" points to one author, Kyd, writing The Spanish Tragedy and Arden of Faversham. The only substantive difference is that Taylor systematically counts the occurrences of shared phrases and attributes authorship to the candidate with the most of them, while Freebury-Jones finds and quotes verbal matches to his favoured candidate without mentioning or counting other writers using the same phrasing.

    Where there exist verbal parallels that point away from Kyd authoring Arden of Faversham, Freebury-Jones argues that Kyd is parodying or imitating Watson's work. Because they were "close associates" (p. 13) they are bound to sound like each other. Freebury-Jones seems not to notice that this explanation makes vulnerable much of the evidence he adduces for Kyd's authorship, since any work's shared phrasings with Kyd might as easily be explained as someone parodying or imitating Kyd.

    Turning to the evidence from rates of feminine ending use, Freebury-Jones finds that the 62 lines of "The Poet's Speech to Her Majesty" in Watson's 1591 dramatic work Entertainment at Elvetham are "not a reliable sample for the feminine ending test" (p. 16), since it is "not a play" (p. 16). Freebury-Jones thinks that Watson's Eclogue upon the Death of Francis Walsingham is a much better source for judging Watson's feminine-ending rate, and it has far too few to be a match for Arden of Faversham. But of course that eclogue is even less like a play than the Entertainment at Elvetham is.

    Quite a bit of Freebury-Jones's argument is recycled from his earlier publications, which he treats as if they had settled the matters they discuss. For instance, because he has elsewhere attributed the play Fair Em to Kyd, he here treats likenesses between Fair Em and Arden of Faversham as evidence that Kyd also wrote Arden of Faversham. It is unwise to build new attributions upon recent ones in this way. Freebury-Jones goes on to make the same construction using the likeness of phrasing in Arden of Faversham and the anonymous King Leir, which he finds significant because he has recently attributed King Leir to Kyd too.

    A particular likeness that interests Freebury-Jones is Fair Em having four stage directions beginning "Here enter[s]" and Arden of Faversham having 64 of them. The obvious question to ask is why, if Kyd wrote both, he put so many more instances of this form into Arden of Faversham than he did Fair Em. In any case, as we saw above in relation to the same evidence being used elsewhere by Freebury-Jones, the phrasing is not as rare as he thinks.

    Brian Vickers also believes that Watson did not write any part of Arden of Faversham, nor did Shakespeare, and Kyd wrote all of it ("Arden of Faversham, the Authorship Problem: Shakespeare, Watson, or Kyd?", DSH 37[2022] 580-93). Vickers too aims to refute Taylor's stylometric claims for Watson's authorship of part of Arden of Faversham, reviewed in NYWES for 2019 and NYWES for 2020. Vickers starts with a brief biography of Watson and a description of the scholarly publications and ends with Watson's contribution to the Entertainment for Queen Elizabeth at Elvetham that took place on 20 September 1591.

    Taylor quoted the likeness between the Entertainment's "And lightfoote Howrs, the guardians of heav’ns gate,| With joyned forces doe remove those blocks" and Arden of Faversham's "See how the howrs the gardeant of heavens gate| Have by their toyle removed the darksome cloudes". Taylor claimed this as a unique parallel between Watson and Arden of Faversham but Vickers disputes this, quoting three passages from Kyd:

Ere Sol had slept three nights in Thetis lap,
And slakte his smoaking Charriot in her floud
(The Spanish Tragedy)

the golden Sunne where ere he drive
His glittring Chariot, finds our Ensignes spred
(Cornelia)

But say my deare, when shall the gates of heaven
Stand all wide ope[n] for celestiall Gods
With gladsome looks to gaze at Hymen's robes?
(Soliman and Perseda)

The obvious problem is that these are not parallels of the kind Taylor found. In Taylor's parallel the following words are in common: 'hours', 'heaven's gate' and 'remove'.

    Vickers's first two Kyd quotations have none of these words and the third has only a form of 'heaven's gate', which is not a rare phrase. Vickers  points out the 'heaven's gate' is not rare, as if this undermined Taylor's claim, but that is not the point of Taylor's claim: the parallel is the coincidence of 'heaven's gate' with 'hours' and 'remove', used by Watson and the author of Arden of Faversham but not Kyd.

    Next Vickers turns to Taylor's means of performing stylometry using digital databases. Vickers echoes Rizvi's objection, made in the essay reviewed above, to Taylor's counting as separate hits the words and phrases that form part of a larger match between two texts. Taylor's experiment involved counting the phrasal matches between a short extract of 274 words from Arden of Faversham and the canons of several candidate authors. Taylor showed that whether we count matches that are part of larger matches or just those larger matches, the counts of Watson matches to this 274-word sample from Arden of Faversham outnumber anybody else's, including Kyd's.

    In response, Vickers offers Freebury-Jones's essay (reviewed in NYWES for 2020) that if we choose an even shorter extract, just 173 words, the tables are turned and Kyd comes out on top. Curiously, Vickers writes that ". . . Taylor’s self-restriction to the first 173 words privileges Watson at Kyd’s expense" (p. 584). But Vickers has got himself confused here: it was Freebury-Jones who confined himself to 173 words, while Taylor used 274.

    As an alternative explanation for the stylometric results, Vickers offers the fact that Kyd and Watson were friends and they borrowed phrases from one another. Since most of the early modern dramatists seem to have known one another and had ample opportunity to sample one another's plays in performance, this explanation can of course be used to dismiss any stylometric finding.

    Because Watson's canon is small, Taylor scaled up the number of matches found with this canon to put him on a level playing field with candidates who have much larger canons. Vickers repeats Rizvi's objection that this scaling factor is so large -- because the Watson canon in so small -- that finding any Watson matches at all cannot fail to make Watson look like the author of Arden of Faversham.

    In explaining this objection, Vickers misunderstands Rizvi's point and seems to think that Rizvi accuses Taylor of getting his numerator and denominator around the wrong way, calculating canon-size divided by hit-count instead of (as he should) hit-count divided by canon-size. But of course Rizvi knows that it does not matter which way around we put the fraction so long as we do the same for each author, as Taylor does.

    Getting the fraction wrong is not what Rizvi accuses Taylor of doing. Rizvi's critique is simply that with so small a canon-size as Watson's finding even one hit for Watson would make Watson's ratio of hits-to-canon-size larger than any rival. This is true but it is hard to see why it undermines Taylor's argument. After all, the smaller is Watson's canon the more surprised we should be by any match with that tiny canon.

    If anything, Rizvi's is a critique of the whole principle of scaling rare-word matches by the size of an author's canon, which is a technique that Rizvi himself uses in his authorship attribution studies. There are indeed good reasons to think that such scaling is not a sound solution to the problem of differing canon sizes, So far nobody has found a better solution to the problem that large canons are more likely than small ones to contain a match with any word or phrase we are looking for.

    Vickers reruns some of Taylor's searches to check the claimed parallels and finds that the collocation of 'my deserts' and 'desires' is not unique to Arden of Faversham and Watson's work but is also present in John Lyly's play Galathea. Vickers writes as if Taylor failed to find this match, but in fact Taylor included it in the appendix to his article that gives all the matches (Taylor's p. 587). Vickers also finds 'desert' and 'desire' in Kyd's plays, but without discussing the contrastive pronouns that according to Taylor make the uses in Arden of Faversham and Watson's work so unusually alike.

    This last omission is ironic, since Vickers then goes on to explain for the next parallel that such decontextualization "can be misleading" and that "It is preferable to cite complete verse lines since they often give additional information, such as the actual meaning of an utterance . . ." (p. 588). Thus Vickers discounts one of Taylor's claimed matches using a principle that, if applied to Kyd's use of 'desire'/'desire' would remove one of Vickers's.

    Regarding the collocation in Watson of the nouns 'desire'/'love' (as in Arden of Faversham), Vickers writes that the two words "come from two different sources, and each is a self-contained sententia" and hence "This cannot count as a 'hit'" (pp. 588-589). I cannot see why Vickers thinks so. Likewise, of the next match Vickers's writes ". . . to state the obvious, 'Wherein he wount' and 'where he was wonte' are not matches" (p. 589). In quoting this match Vickers has shortened it and thereby reduced its significance: the actual match is 'Wherein he wount to' in Arden of Faversham and 'where he was wont to'; notice the additionally matching 'to'. And in context this is a match since in both 'wherein' has the sense of 'where'.

    For the match of "my deserts or your desires" in Arden of Faversham with Watson, Vickers claims that when he (Vickers) tracked it down he found that Taylor's claim was based on "misquotation" (Vickers p. 589). But Vickers seems to be simply looking in the wrong place among Taylor's list of hits, for on Taylor's page 867 we find "Hekatompathia 34: 'To follow her desert, and my desire'". The match claimed by Taylor is the combination of the two "nouns, contrasting pronouns immediately preceding each noun, and association of 'desert' with a female and 'desire' with a male". Vickers reports that he found other mistakes in Taylor's list of claimed matches between Arden of Faversham and Watson's work but does not identify them. "I stop at this point" Vickers writes, "for it is clear that the textual evidence supporting Taylor's argument is inadequate" (p. 589).

    Next Vickers attempts to show that verbal matches between Kyd's plays and Arden of Faversham are more significant than Taylor made them seem. "Given as complete verse lines", Vickers writes, "the similarities of phrasing and rhythm are much more apparent" (p. 589). But even with this wider context I cannot see why Vickers thinks that Arden of Faversham's "The season fits, come Francklin, let's away" is unusually like Kyd's "Come, lets away to seek my Lord the Duke". This seems like a simple match of "come' followed by 'let's away', which of course occurs many times in the drama. I cannot see what extra value the context adds to the match.

     Likewise Arden of Faversham's "See how the howrs the gardeant of heauens gate" and Kyd's "See how the Rocks do heave their heads at thee" seems no more than a match of 'see how the' in both texts. Vickers lists a few more like this and concludes his objection to Taylor's essentially quantitative approach of match-counting with the remark that "the relevant criterion is the quality of those hits" (p. 590).

    Vickers turns to Rizvi's online set of data and experimental results called Collocations and N-Grams. The key element that Vickers describes is called 'Experiment 4 -- Arden of Faversham and the Extended Kyd Canon' for which Vickers gives the URL <https://www.shakespearestext.com/can/experiments/results/4-Arden-of-Faversham-and-the-Extended-Kyd-Canon>. At the time of writing (22 June 2026) this URL leads to a type-404 error message: "That page has moved or you may have used an incorrect URL". That Vickers's citations of authorities for his claims are broken URLs has been a recurrent complaint of this review for the past 10 years. According to Vickers, Rizvi's experiment confirms the attribution of Arden of Faversham to Kyd.

    David L. Hoover responds to Rizvi's essay "The Interpretation of Zeta Test Results" reviewed in NYWES for 2019, showing that it is mainly wrong ("Zeta Revisited", DSH 37[2022] 1002-21). When presenting Zeta results, investigators typically offer a two-dimensional (x/y) scatterplot in which a datapoint's x-axis value represents the degree to which a text (being represented by the datapoint) contains many of the words that are favoured in one candidate's author's canon and many of the words that are disfavoured in a second candidate author's canon. Conversely, the datapoint's y-axis value represents the degree to which a text contains many of the words that are favoured in the second candidate's author's canon and many of the words that are disfavoured in the first candidate author's canon.

    When texts from these two candidate authors' canons are plotted in this way, the first candidate's works tend to score highly on the x-axis because they contain many of the words the first candiate favours and many of the words the second candidate disfavours. And they tend to score lowly on the y-axis because they contain few of the words the second candidate favours and few of the words the first candidate disfavours. This puts the datapoints for the first candidate's works in the bottom-right (or south-east) quadrant of the scatterplot.

    The opposite is true of the datapoints for the second candidate's works, since these tend to score lowly on the x-axis because they contain few of the words the first candidate favours and few of the words the second candidate disfavours, and they tend to score highly on the y-axis because they contain many of the words the second candidate favours and many of the words the first candidate disfavours. This puts the datapoints for the second candidate's works in the top-left (or north-west) quadrant of the scatterplot. It is typical for Zeta scatterplots to include a line running from south-west to north-east that divides the whole into two 'zones' so that the one author's points are above and to the left of this line and the other author's points are below and to the right of it.

    Rizvi objected that simply getting two such clusters, one in the south-east and one in the north-west, is not itself an experimental achievement: the investigator chose the words being counted precisely to produce this effect. But Hoover points out that a sharp clustering is not in fact guaranteed and arises only if the authors really do have contrasting habits with certain words, and that is by no means always the case.

    To test whether the method really can detect authorship, Hoover makes a dataset of about 400,000 words by Henry James and another of about the same size by William Dean Howells. A standard Zeta test puts them into clearly distinct clusters (bottom-right and top-left in the scatterplot). Then Hoover puts a couple of James's works into the Howells dataset (so the method would treat them as if they were by Howells) and likewise a couple of Howells's works into the James dataset.

    This affects the lists of favoured-by-James and favoured-by-Howells words and affects the positions of the segments on the scatterplot, but strikingly the dots for the deliberately misattributed works hardly change positions: in the new scatterplot they still cluster (as they did in the original) close to their true authors and away from the cluster for the 'wrong' author into whose dataset we put them. Thus the method is indeed capable of putting an authorial datapoint far away from that author's cluster if the method 'thinks' that we have got the attribution wrong.

    The amount of separation between the clusters, then, is a marker of how unalike the authors' word choices are. Hoover points out that in his critique of the use of a bisector line to divide a scatterplot into two authorial zones, Rizvi offered a made-up (and admittedly problematic) example, Rizvi's Figure 1, that looks nothing like the actual scatterplots that scholars using Zeta usually generate from their data. Hoover agrees with Rizvi that we do not really need the bisector lines as the important facts are how tightly compacted is the cluster of dots for each author and how far each dot representinga a suspect text is from the two authorial clusters.

    Hoover remarks, as did the present reviewer, that Rizvi seemed to reject the idea that some kinds of separability emerge only when we consider two or more dimensions at once. Rizvi demanded that in an x/y scatterplot the values for each author must be fully distinct along each of the two dimensions. As Hoover puts it ". . . Rizvi’s analysis is a one-dimensional critique of a two-dimensional method, and for that reason alone is hard to take seriously" (p. 1006).

    Where Elliott and Greatley-Hirsch, in the New Oxford Shakespeare: Authorship Companion, use hold-out segments from the same plays as those used to make a scatterplot to evaluate the accuracy of the scatterplot, Rizvi objected that this merely showed that a segment is like the play it was taken from. But Hoover points out that in fact segments from a single authorial work can be widely dispersed across -- even on opposite edges of -- the author's cluster. It is not a given that segments from a single work will be alike just because they are all from that one work.

    Rizvi objected that because Zeta derives the word-types that it counts using not the overall frequency but the number of segments each word-type appears in, it favours consistent use of a word across the text over the frequency of the word. Hoover points out that that there is "no a priori reason to believe that frequency is more important than consistency" (p. 1007) when judging authorship. Hoover describes Craig's refinements of Burrows's original Zeta -- dividing the authors' texts into equally sized segments and "calculating Zeta scores as the proportion of the types in each segment that are each author's marker words" (p. 1008) -- and points out that this form of Zeta is a useful complement to methods that measure only words' frequencies.

    Hoover runs a series of experiments in which he tries out certain variations in Zeta. He tests whether words (or n-grams) with high Zeta scores are more effective for authorship attribution than those with low ones, and whether Zeta itself is an effective authorship classifier. Hoover tries "words, word 2-grams, 3-grams, and 4-grams, and character 3-grams, 4-grams, and 5-grams" (p. 1009) using datasets of nineteenth-century American fiction (6.75 million words), twentieth-century American poetry (1.47 million words), and Victorian drama (2.1 million words). 

    Rizvi objected to how the centroid of each authorial cluster has, in previous Zeta experiments, been calculated from the means of the x and y values of the points in the cluster, noting that this allows the occasional outlier to shift the centroid substantially. Accepting this critique, Hoover instead uses the median values in place of the mean values. What should we do about points that fall close to the bisector line, for which Rizvi thinks the Zeta method is rather crude in its simple binary classification of all points based on which side of the line they fall?

    Hoover derives a calculation for quantifying how far a point is from the bisector line. He creates a right-angled triangle in which the point forms the corner of the right angle, the bisector forms the hypotenuse, and vertical and horizontal lines drawn from the point to the bisector form the other two legs. The length of the horizontal leg is expressed as a percentage of the x value of the bisector where this leg meets it, and the length of the vertical leg is expressed as a percentage of the y value of the bisector where this leg meets it, and these two percentages are summed.

    This gives an overall percentage that allows comparison between scatterplots whose scales are different. For these experiments, Hoover sets his threshold at 7%: if a point's overall 'distance' (as calculated by the above method) from the bisector is below this then it would be classified as being only marginally on that side of the line, but if the 'distance' is 7% or more then it would be classified as definitely on that side of the line. Hoover ignores all marginal points, whether they are correctly or incorrectly assigned, in his experiments.

    Hoover calculates the standard machine-learning metrics of Accuracy, Precision, Recall, F-score and also something called Matthews Correlation Coefficient, and discusses why no measure of a classifier's performance is perfect. Hoover presents his results as a comparison of Zeta's performance against that of the function called 'Classify' in the R Stylo software library. Overall both methods perform very well, and both show that word n-grams where n is 4 are less useful from classification than shorter ones.

    But for character n-grams the results are mixed and overall, across all the variables, correct attribution is not "strongly correlated with either type or length of n-gram" (p. 1017). Rizvi claimed that the Zeta method inherently favours short n-grams but this is not borne out by Hoover's results. As Hoover remarks, ". . . no method, variable, or measure is consistently preferable for all corpora" and "Nothing works (best) all the time under all circumstances" (p. 1018).

    Hoover concludes that Rizvi is wrong that finding two distinct clusters in the bottom-right and top-left of a scatterplot is inevitable and meaningless: rather, it tells us how distinct the datasets (and hence the authors) really are and hence how likely we are to get reliable attribution results. Hoover concludes that Rizvi is right to be concerned about how we construct a bisector, and right that the median rather than mean values should be used to find the centroids from which the bisector is derived, and right to object that points falling near the bisector line should be treated as less significant than those far from it. Whether long n-grams (as Rizvi holds) or short n-grams are better for author attribution is a matter that Hoover thinks remains open: his experiments do not settle it.

    These days, journal articles are often made available in standalone 'Advance Access' form a year or two before they appear in their assigned position within a volume of the journal. By this means, Rizvi was able to read Hoover's article we just reviewed, and respond to it is an article that also appeared in 2022 ("The Interpretation of Zeta Results: A Supplement", DSH 37[2022] 1172-8). Rizvi finds Hoover wrong about the significance of Zeta's method for discriminating word preferences and about the meaning of the bisector line used in the interpretation of Zeta scatterplots.

    To assess Rizvi's article it is necessary to recapitulate key aspects of the Zeta method using the terminology that he employs. In the Zeta method, the set of texts known to be by one author (or a group of authors considered collectively) is called the 'base' set and the set of texts known to be by another author (or group of authors) is called the 'counter' set. Typically, the texts are not whole works but equally sized chunks, called 'segments', into which whole works are first divided.

    Rizvi starts with a false contrast in a question that is central to the first half of his article: "When a Zeta graph shows a good separation between the base and counter segments, as it often does, is that a finding about the authorship of those segments, or is it an inevitable consequence of the method and therefore of no significance?" (p. 1172). Rizvi tries to show that the latter of his two suggested answers is correct: the separation is inevitable given the method and hence is of no significance.

    I would say that the correct answer depends, as we shall see, on where the base and counter segments came from and that if the segments that the method separates are from different authors then their separation is a finding about the authorship (namely that these author's habits can be distinguished), and that this is an inevitable consequence of the method if the segments are distinguishable and the method is being applied correctly. Rizvi's misframing of the question is the reason that his investigation of it leads to the false conclusion that Zeta is able to tell us nothing.

    I do not think any investigator using Zeta is under the delusion that Rizvi claims they are all under, which is that finding base/counter separation in the scatterplot is itself a validation of the method's ability to reliably attribute authorship. This is, rather, a finding that the segments being tested can objectively be distinguished. To try to show that authorship is not necessarily at play when segments are being distinguished, Rizvi randomly assigns each word from Jane Austen's Pride and Prejudice to a base set and a counter set and then divides the base and counter into 2000-word segments and applied the Zeta method to produce a scatterplot.

    Rizvi finds a sharp separation of the two sets, and because this cannot be the result of an authorship difference -- all the segments came from one writer's novel --  Rizvi concludes that this separation will be found for any sole-authored text and hence the separation tells us nothing about who wrote what. Then he does essentially the same experiment for an undifferentiated set of Shakespeare plays and Fletcher plays (treated as one pool of writing), and finds that the scatterplot again divides into two regions, but not along authorial lines.

    Rizvi misses the point that Zeta is a supervised method not an unsupervised one, meaning that we start with labelled data -- we assert that certain blocks are definitely by author X, and others definitely by author Y -- and we ask whether there are features of the writing that demonstrably differ between the two sets and hence may be markers of the two authorial styles. That any single-authored body of writing also displays internal variation that Zeta will identify, because it is a method for finding any such variation, is not evidence that authors do not differ along the dimensions that Zeta can measure.

    No one who understands Zeta would expect that if we treat Shakespeare's plays and Fletcher's plays as one pool of writing, Zeta will separate them out into two sets by author. Rather, being a supervised method, we use it to find which words do distinguish the two writers and we then see whether a text of unknown authorship reflects the preferences of one or other writer. Rizvi's procedure is of the kind we would use to interrogate the validity of an unsupervised method, meaning one in which we do not label the blocks but ask the computer to find whatever internal variations it can.

    The proper test for whether Zeta's supervised method can distinguish authorship is blind validation in cases where the authorship is known in advance. That is, if Zeta correctly 'predicts' authorship in cases where the true authorship is known -- 'predicting' in the scientific sense of asserting that a particular text belongs to a particular writer -- then it is a useful tool. Zeta has repeatedly been shown to have this predictive ability.

    In the second half of his article, Rizvi returns to the use of bisector lines in the interpretation of Zeta scatterplot, which he first brought up in "The Interpretation of Zeta Results" reviewed in NYWES for 2019. He is quite right that for a block of unknown authorship an attribution to the author represented by the datapoints for the base set on a scatterplot (the ones in the south-east quadrant with high x values and low y values) should be made only when the datapoint for the unknown block likewise has a high x value (meaning that it is like the blocks in the base set) and a low y value (meaning that it is unlike the blocks in the counter set).

    He is also right that conversely an attribution to the author represented by the datapoints for the counter set on a scatterplot (the ones in the north-west quadrant with low x values and high y values) should be made only when the unknown block has a low x value (meaning that it is like the blocks in the counter set) and a high y value (meaning that it is unlike the blocks in the base set).

    To conveniently make this distinction, Hugh Craig and Arthur Kinney simply drew a bisector separating the south-east quadrant from the north-west quadrant, specifically by finding the centre of the cluster for the base set (in the south-east) and the center of the cluster for the counter set (in the north-west), joining these centres with a straight line, and then dividing this line in half and extending a line perpendicular to this mid-point.

    Rizvi is right that this bisector does not precisely isolate only the points that are high-x-low-y (the points near the 'cloud' of base set points) from those that are low-x-high-y (the ones near the 'cloud' of counter set points), since the bisector line extends to infinity in both directions. Because the bisector runs to infinity, points that are high-x-high-y (way off in the distant north-east) might fall either side of it without significance, and so might points that are low-x-low-y (way off in the distant south-west). But so long as the points under consideration fall reasonably close to the line between the two 'clouds' (that is, they fall in the space between the base 'cloud' and the counter 'cloud') then this bisector is a good way to distinguish south-easterlyness from north-westerlyness, which is the feature we are trying to isolate.   

    Rizvi's objection to the use of bisector lines comes down to the simple point that it is a reasonable tool only when the datapoints being classified are close to the line that joins the centres of the two 'clouds' and hence the datapoints fall between them. Rizvi is right about this, although of course to apply his rule we would need some way of quantifying this closeness/betweenness, and he suggests none. As we have seen, Hoover offered a practical suggestion about this.

    Two more of this years articles used computational methods and both come to startling conclusions that we confidently dismiss as false. In the first, Hartmut Ilsemann attempts to confirm what he has claimed in previous articles, that most of the plays that we have attributed to Marlowe were in fact written by someone else ("A Brief Supplement to 'The Marlowe Corpus Revisited' and Phantom Marlowe", DSH 37[2022] 462-8). Ilsemann starts by claiming that 1 Tamburlaine and 2 Tamburlaine are more like each other than either is like the other plays traditionally attributed to Marlowe: Dido Queen of Carthage, The Jew of Malta, The Massacre at Paris, Edward 2, and Doctor Faustus. The obvious retort is that this is to be expected since they are two parts of one play.

    Ilsemann uses Maciej Eder's 'General Imposters' method on the problem. Ilsemann explains that the input to the method is that ". . . all the texts to be analysed are already pre-processed and represented in a form of a matrix with frequencies of features (usually words)" (p. 462). Unfortunately, Ilsemann neglects to report what the rows and the columns of the matrix actually represent (that is, what are the column and row labels), which makes it hard to see what the experimental method actually involves. What is worse, Ilsemann then simply lists the commands that he gave to the computer software called R, running Eder's Stylo software package, rather than explaining the experimental steps.

    Crucially, it is impossible to discern from Ilsemann's account just what got compared with what in the experiment. From his file names it looks like Shakespeare was represented by Twelfth Night, Hamlet, King Lear, Romeo and Juliet, and The Winter's Tale, and Ilsemann never explains why he used these five plays. Why use three tragedies and two comedies (one arguably a tragicomedy) and none of his histories to represent Shakespeare's style?

    It is also unstated how Ilsemann's set of non-Shakespearian plays was chosen. Aside from the allegedly Marlovian plays Ilsemann is trying to attribute, the experiment seems to use the five Shakespeare plays, plus two by Peele, one by Nashe, one by Kyd, and one anonymous. The one by Kyd is his translation Cornelia instead of the plays widely agreed to be Kyd's own inventions, The Spanish Tragedy and Soliman and Perseda. With Shakespeare plays forming half of the dataset of non-Marlovian candidates (5 out of 10 plays), it is not exactly a surprise that Shakespeare gets more than his share of the new attributions.

    Ilsemann seems to have misunderstood the fundamental point of Eder's 'General Imposters' method, since he has included in his dataset no imposters. The point of the method is to introduce into your dataset works by candidates who could not possibly have written the work to be attributed, the 'imposters' of the method's name. This gives the investigator a sense of how reliable the attribution method is at detecting real stylistic differences between writings rather than picking up on the mere random 'noise' (that is, variation in stylistic habits) present in any samples of writing. None of the authors chosen by Ilsemann for his experiment -- Shakespeare, Kyd, Nashe, and Anonymous -- is an impossible candidate for writing the Marlowe plays.

    Ilsemann refers to some distance measures used in the experiment, such as Wurzburg (a form of cosine distance) and "minmax" (pp. 464-465) distance, and reports that Table 2 uses the former and Table 3 uses the latter. He does not explain why he uses different measures in different tables nor what significance we should attribute to his getting different results from the two cases. The impression given is that Ilsemann is just randomly 'twiddling the knobs' on his experimental software to see what happens.

    Ilsemann presents tables arising from a final twiddle of the knobs to make his software use word bigrams and character trigrams. The last of these shows that Shakespeare wrote Doctor Faustus, Dido Queen of Carthage, Edward 2, The Jew of Malta, and The Massacre at Paris. To account for this surprising result Ilsemann concludes that "These plays must be seen as collaborations, and one of the contributors in varying degrees must have been William Shakespeare" (p. 468). It does not seem to occur to him that maybe his experiment is not doing what he thinks it is doing.

    The other article with a highly implausible conclusion is Anna Faktorovich's attempt to show that  90% of the plays identified in Henslowe's Diary never existed and were invented as a book-keeping fraud ("Manipulation of Theatrical Audience Size: Nonexistent Plays and Murderous Lenders", Critical Survey 34.iii[2022] 82-99). According to Faktorovich there was not even a person called Philip Henslowe and nor was there, in all likelihood, a man called William Shakespeare. Instead there was what Faktorovich calls 'The Workshop', a collective of writers. Shakespeare's plays were written by "Percy, Jonson, Harvey, Byrd and Sylvester" (p. 83).

    The source for Faktorovich's attributions is a "computational-linguistic author-attribution method" that she invented and that informed the reshaping of the field that is described in the 20-volume British Renaissance Re-Attribution and Modernization Series of books that Faktorovich has self-published via her company Anaphora Literary Press. There is no way to separate her claims that are worth exploring from the utterly bizarre ones in references to the ". . . Verstegan-ghostwritten Pierce Penniless His Supplication to the Devil (1592) and her claim that ". . . it is very likely that Verstegan ghostwrote the other 'Dekker'-bylined theatrical pufferies, which include Dead Term (1608), Seven Deadly Sins of London (1606) and Ravens' Almanac (1609)" (p. 84) and her references to "Jonson's Winter's Tale" and "Jonson's Coriolanus" (p. 85).

    The article is full of many such startlingly implausible assertions about play writing and theatre history. Indeed, according to Faktorovich, the entire body of English Renaissance drama between the 1580s and the 1640s cannot have flourished in the way that standard historical narratives tells us, since if it did we cannot explain "the absence of theatrical entertainments in Wales across the same decades" (p. 89). The editor and publisher of the journal Critical Survey ought to be embarrassed that they accepted and disseminated this illogical and historically ill-informed essay.

    Three articles published this year deal with matters of dramatic convention that are of interest to editors so they are noticed here. In the first of these, Margaret Jane Kidnie addresses some awkward cases of mid-scene entrance of one or more characters where it is unclear whether those on stage are supposed to notice the entrance and unclear whether the person(s) entering hear(s) the dialogue of those already on the stage ("Warning the Stage: Shakespeare's Mid-scene Entrance Conventions", ShSurv 75[2022] 208-28). For Kidnie, the problems presented by these tricky cases are not hard enough for us to have to conclude that rehearsals were used to iron them out. Kidnie is able to construct rules of dramaturgy that solve virtually all the problems.

    Kidnie looks at the moment just before Osric first enters in Hamlet. In the second quarto of 1604-5 (hereafter Q2) he seems to interrupt Hamlet's speech, since Hamlet's thought (about what he is justified in doing to Claudius) is not completed, whereas in the 1623 Folio (hereafter F) Hamlet first completes the thought ("To quit him with this arm") and then has another one (that Laertes's case as a grieving son is like his own). Kidnie wonders whether we are right to suppose that in Q2 Osric overhears Hamlet's incomplete speech, which implies, without completing the thought, that Hamlet plans to kill the king.

    Without resolving this, Kidnie switches to Hamlet's entrance behind Claudius at prayer and asks "How does the actor playing the King know not to look back (or over, or up) when Hamlet enters?" (p. 211). The obvious answer that they work this out in rehearsal is not available to her because she accepts Tiffany Stern's argument that early modern acting companies seldom rehearsed. So Kidnie seeks to derive conventions governing mid-scene entrances that explain Claudius's non-reaction.

    To find the conventions, Kidnie looks at mid-scene entrances in 26 plays in print and in playhouse manuscripts. They form eight kinds of mid-scene entrance, falling into two main groups: 'warned' and 'unwarned'. The four 'warned' types are for characters Hailed from Stage, Hailing on Entry, Summoned, and having a Pre-Arranged Meeting, and the four 'unwarned' are for characters Sizing Up [the] Stage, where dramatists use Simultaneous Staging, or where characters Pass Over [the] Stage, or Wait Attendance.

    Kidnie finds evidence that "upon entry a character could, or at least might, hear any lines delivered during her approach" (p. 215). In The Honest Man's Fortune Kidnie finds the bookkeeper Edward Knight taking special pains to place the entrance for a couple of characters so they have no chance of overhearing the onstage conversation they are interrupting. This example of the exactness off placement contradicts Kidnie's arguments about the indeterminacy of stage-direction placement that she made in her essays "Text, Performance, and the Editors: Staging Shakespeare's Drama" (reviewed in YWES for 2000) and "The Staging of Shakespeare's Drama in Print Editions" (reviewed in YWES for 2004).

    Kidnie finds that in general entrances of which those on stage are unwarned are ignored by them. For an entering character to be noticed, he or she must be given a line of dialogue to draw the attention of those on stage, or else one of the onstage characters must be given a line to indicate that he or she notices the entrance. Thus the onstage characters are under "no pressure to keep one eye on the tiring-house doors" (p. 218).

    Kidnie gives another example, from John a Kent and John a Cumber, of an entrance that is precisely timed so that John a Cumber enters at exactly the right moment to hear a reference to himself ("so sayd a deuill or boy, sent to vs from Iohn a Cumber") immediately after he enters. In general, then, an entering character can be assumed to hear what those on stage are saying, and in general those on stage do not notice the entrance unless and until the entering character speaks to them, or one of the characters on stage has dialogue (including a cue from the entering character) indicating that he notices the entrance.

    Henry 5 is alone on stage after his interaction with Pistol when Fluellen and Gower enter and discuss the need for them to keep quiet so the enemy does not hear them. Henry never acknowledges them and they do not notice him, but it is made clear that Henry hears what Fluellen and Gower said because he remarks on it after they leave. For Kidnie, it is not a given that because Fluellen and Gower are unaware of Henry's presence he will necessarily overhear what they say. She writes: "That he [Henry] was listening is confirmed only after the others exit" (p. 220).

    Edward Alleyn's part for the title character of Robert Greene's Orlando Furioso requires him to enter into the middle of a scene but does not tell him exactly when to do so. David Bradley and Stern decided that Alleyn listened in to the stage dialogue from off stage and figured out when he needed to enter from that, but Werstine instead supposed that it was the bookkeeper's job to tell Alleyn when to enter. Entrances and exits from and to the 'arras' somewhat blur the distinctions being made here because it is a liminal place and indeed we do not know for sure where it was.

    Kidnie talks the reader through a few mid-scene entrances that seem not to conform to the conventions she proposes, but none of them presents serious problems of staging if we suppose that the players rehearsed a little before performing. Kidnie resorts to this solution only when all else has failed: "I can only think that this mid-scene entrance was rehearsed . . ." (p. 224). The idea that actors did not generally rehearse is maintained by theatre historians despite its being decidedly counter-intuitive. Did the players not notice that their theatre was entirely unused from sunrise to 2pm? Did it never occurred to them that a bit of practice in that empty space might be useful?

    Kidnie ends by returning to Osric's intrusive, speech-interrupting, entry to Hamlet and Horatio in Q2 Hamlet that she began with. If this really is an interruption of Hamlet's speech then it is essential that Osric enters at exactly the right moment: too soon and he will not be interrupting Hamlet, too late and Hamlet will run out of things to say. For that reason, Kidnie thinks that this entrance is unique.

    It is noticeable that the variant text in F resolves the problem: Osric's entrance is noticed by Horatio and remarked upon. Kidnie toys with but does not commit herself to the explanation that in the Q2 version Shakespeare introduced a problem (the unannounced entrance of Osric) that the Folio version solved by giving Horatio a line to notice Osric.

    Despite one case almost dissuading her, Kidnie returns to the dogma: ". . . it is not necessary to invoke group rehearsal to explain how early modern actors working from parts managed to shape mid-scene entrances during live performance" (pp. 227-228). That is because in general "shared theatrical conventions" (p. 228) were enough.

    Mariko Ichikawa marshalls evidence from across early modern plays to show that the expression 'at the arras' meant 'behind the arras', this being the cloth that hung in front of the central opening in the back wall of the stage ("'At the Arras': What is the Location Implied in This Phrase?", ShakB 40[2022] 215-37). One reason why theatre historians have been reluctant to accept that early modern theatres had a middle entrance in the back wall, between the two stage doors, is that just three plays (quoted by Ichikawa) explicitly call for use of this third entrance. Ichikawa argues that we have overlooked other implicit references to this middle entrance that take the form of references to the 'arras' or 'hangings' or 'curtain'. She focusses on six instances of 'at the arras' in five King's Men's plays from the 1620s and 1630s.

    John Fletcher and William Rowley's The Maid in the Mill has the stage direction "Six Chaires placed at the Arras" fully 275 lines before these chairs are required by the stage action. During those 275 lines the two stage doors are needed, so these chairs must not obstruct the doors and hence the arras must be between the two doors. But are the chairs in front of or behind the arras? We cannot tell.

    William Davenant's The Cruel Brother has the stage directions "Chaire at the Arras" and "A Chayre at the Arras". Again the need to use the stage doors makes it likely that this meant somewhere between the stage doors. Again, we cannot tell if the chair is in front of or behind the arras.

    The manuscript playbook of Nathan Field, Fletcher, Philip Massinger, and Robert Daborne's The Honest Man's Fortune has "Lamyra showes hir Self at the Arras". This seems to mean that she pokes her head out to let the audience know that she is behind the arras (for some overhearing) since later there is the stage direction "Ent: Lamyra: from the Arras". Before poking her head out, Lamyra had exitted the stage, so there must have been a discovery space behind the arras, accessible from back stage. But in this case we cannot be sure it was a central discovery space: they could have used one of the stage doors.

    Massinger's Believe as You List has the stage direction "Harry: Wilson: & Boy ready far the song at y^e Arras". The dialogue makes clear that these performers are invisible. They cannot be in a curtained music room in the stage balcony, which we know the theatres had, because this space was being used by actors, so the musical performers are on the stage level behind an arras. But we cannot tell from the use of the stage doors that this is a centrally located arras.

    The last bit of evidence is from Massinger's The City Madam, which has the directions "Whil'st the Act Plays, the Foot-step, little Table, and Arras hung up for the Musicians" and "Musicians come down to make ready for the song at Aras". So, an arras is hung up in the interval between Act 4 and Act 5. From the dramatic context and action, Ichikawa infers that "at Aras" must here mean behind it and out of sight. This is because, for the performance of a show about Orpheus, Luke's table is set up with him at its upper end (the important person's seat) nearest the back wall of the stage, so the musicians cannot plausibly be in sight behind him: they must be out of sight behind him.

    Since the stage doors are needed for the ensuing action, the arras must be in the centre of the back wall of the stage. And since the musicians from the stage balcony music room need to come down and get into it, the central space concealed by this arras can be accessed from the tiring house: it is a central opening.

    Putting the six stage directions together, Ichikawa points out that all seem to be bookkeepers' notes and all are written from the bookkeeper's perspective from within the tiring house. For him, getting properties into a central discovery space concealed by an arras is not a matter of putting them behind the arras but in front of it, within the discovery space. Thus 'at' the arras is the bookkeeper's idiom for 'behind' it from the audience's perspective. Bookkeepers and authors did also use the phrase 'behind the arras'.

    James Hirsh explores the dramatic conventions of 'aside' and 'soliloquy' using his own unusual definitions ("The Convention of Self-address Speech in Shakespeare's Plays: New Empirical Data", BJJ 29[2022] 192-223). He defines a soliloquy as "a passage [of speech] not directed at the hearing of any other character" (p. 192). This definition entails that what we normally call 'asides' to the audience -- such as Polonius's "Though this be madness, yet there is method in 't" -- are called soliloquies by Hirsh.

    The usual definition of 'soliloquy' comes from its Latin etymology: 'solus' (alone) and 'loqui' (to speak). In an endnote, Hirsh objects that using this derivation to decide what the word 'soliloquy' means is "a flawed procedure" since the word "did not become a theatrical term until after his death" and hence there is no chance that Shakespeare "tailored his dramatic technique to make it fit the etymology" (p. 219n1). But that of course is why we should respect the etymology: the word was coined by later theorists of the stage to fit what Shakespeare actually did in having a character speak while alone.

    For Hirsh, what other critics call soliloquies are in fact spoken when the character is not alone, such as Juliet on her balcony unaware of Romeo below. Indeed, but the important fact is that such characters think themselves alone in all these cases. (Hirsh gets around to acknowledging this crucial distinction near the end of his essay.) Because he understates the importance of whether a character believes herself to be alone, Hirsch is forced to construct a rather complex set of conventions such as asides that some characters cannot hear and others can. Speeches that no one on stage hears are by Hirsh's definition, asides and soliloquies too, or as he puts it "soliloquies guarded in asides" (p. 193).

    Hirsh reckons that part of the thrill of soliloquies guarded in asides is that "Playgoers were aware that, if the character became preoccupied with what she was saying to herself, she would lower her guard, and the other characters would begin to overhear" (p. 193). Later Hirsh repeats this claim: playgoers "knew that guarding a soliloquy was a skill and entailed a risk" (p. 209). Hirsh offers no examples of characters ever making the mistake of lowering the guard that makes their speech inaudible and I doubt it is possible. Rather, a speaker can, using the aside convention, deafen the ears of anyone she knows is present and for as long as she likes.

    Hirsh finds that soliloquies are essentially self-addressed speech and that the speaker frequently names herself in the speech as its addressee, as in Titus's "Titus, when wert thou wont to walk alone" in Titus Andronicus and Petruccio's ". . . and now, Petruccio, speak" in The Taming of the Shrew. Hirsh finds no evidence that soliloquies were ever addressed to the audience. Hamlet's "Am I a coward?| Who calls me villain, breaks my pate across . . . ?" might seem like a candidate, but Hirsh insists that since no audience member has done these things or could do these things Hamlet is still just talking to himself. In fact, Mark Rylance as Hamlet at the London replica Globe theatre in 2000 delivered these and other lines as direct questions to the audience, and waited for responses, which he frequently got.

    Hirsh makes the surprising claim that "some female characters were played by adult males" (p. 208) without giving any evidence or indicating which characters he is thinking of. The conventional wisdom is that only boys played women. Towards the end of the essay Hirsh finally gets to the vital point about the convention of audibility: ". . . a character cannot guard a soliloquy from a character of whose presence the speaker is unaware" (p. 215).

    The remaining articles this year all deal with particular moments of interest in Shakespeare's texts. Joshua R. Held argues that the double rejection of Pandarus found in the Folio edition of Troilus and Cressida (but not the 1609 quarto) should be retained by editors as it fits a wider pattern of repetition in the play ("Editing Shakespeare's Troilus and Cressida: The Folio's Double Rejection of Pandarus", Shakespeare Studies 50[2022] 101-25).

    In the Folio, at the end of 5.3, Troilus and Pandarus have an exchange:

    Pand. Why, but heare you?

    Troy. Hence brother lackie; ignomie and shame
Pursue thy life, and liue aye with thy name.

(TLN 3329-3330; sig. ¶¶6r) 

In the Folio, almost the same exchange happens again just before the final speech of the play:

    Pand. But heare you? heare you?

    Troy. Hence broker, lackie, ignomy, and shame
Pursue thy life, and liue aye with thy name.

(TLN 3570-3571; ¶¶¶1r)

Held calls the Folio's final speech by Pandarus that follows this an epilogue. It is generally agreed that one or other version of the exchange quoted above should be treated as a first attempt that was meant to be deleted.

    The 1609 quarto does not have this repetition since the first of the above exchanges is absent and just before Pandarus gives the play's final speech there is this exchange:

    Pan. But here you, here you.

    Troy. Hence broker, lacky, ignomyny, shame,
Pursue thy life, and liue aye with thy name.

(sig. M1v)

    Editions vary in how they deal with the Folio's duplicated exchange, but typically they delete one or other as an error. Held's argument is that the doubleness of this repetition fits a wider pattern of doubled things in the play and we should allow it to stand as a repetition as it is in the Folio. The Second Folio used "brother" for both rejections (instead of "brother" in the first then "broker" in the second), and the Third and Fourth Folios used "brother" for the first rejection but "brothel" for the second.

    Some of the great eighteenth-century editors seem not to have been bothered by the repetition of the rejection, but Edward Capell was and he cut the first rejection and moved a chunk of what preceded it to the end to be part of the final rejection. George Steevens could see the problem too, and Edmond Malone relied on the 1609 quarto to simply cut the three lines that form the first rejection.

    Held thinks that Malone was motivated not so much by the problem here in Troilus and Cressida as by his belief that Shakespeare's quartos in general are more authoritative than the Folio. Held remarks that "Yet this thinking is problematic since scholars have shown that a first printed version may reflect an early modern theatrical author’s manuscript less precisely than a later printing" (p. 110). This point about early quartos Held supports with an endnote (p. 124n39) about the differing scholarly opinions of the origins of Q1 Hamlet -- early version versus memorial reconstruction versus script recorded by notes made during performance -- that appears to indicate that Held thinks that if we discovered the origins of Q1 Hamlet we would thereby have discovered the origins of Q1 Troilus and Cressida too.

    Held claims that the ". . . F[olio] -- and the tradition that develops from it -- extends, and retroactively highlights, an aesthetically rich layer of the text: double plot, double cross, doublespeak, double entendre" into which the double rejection of Pandarus fits properly (p. 113). No one would deny the pattern of repetition that Held's literary-critical study of the play's dialogue and action traces. But whether this justifies making the actors do what the Folio requires in repeating themselves is a matter of subjective judgement. Held strains to avoid admitting that the Folio simply has an error, most likely caused by authorial second thoughts.

    Emma Smith argues that the publication of the Folio itself imposed what we now perceive as the seriality of the Shakespeare history plays ("Shakespeare's Serial Histories?", Memoria di Shakespeare: A Journal of Shakespearean Studies 9[2022] 1-20). The order of the Comedies and Tragedies in the Folio seems entirely random, but the Histories are in regnal order. The Folio's seriality of the History plays is not so much an inherent aspect of the plays themselves as a "specific intervention" (p. 4) made for that collection.

    Smith considers the pre-Folio editions of the plays and finds them rather more disparate and discontinuous than the Folio makes them. Calling them tragedies, as the Richard 2 and Richard 3 quarto title pages do, makes them stand alone, since by generic tradition there cannot be a sequel to a tragedy, and these are among the most reprinted of the history plays.

    The publication of the Henry 6 plays in the 1590s really disrupts the idea of seriality: Part One was not published at all, Part Two "was proposed as a first part to an incomplete story" (p. 7) by being called "The first part of ...", and Part Three's title page claimed that it contained "the whole contention between the two houses Lancaster and York".

    The two Henry 4 plays were not published in a way that stressed their interdependence. Part One was published as "The history of . . ." with no mention of a second part, which makes Part Two seem more like an unanticipated sequel rather than something planned from the start. And Part Two apparently sold poorly and was not reprinted in Shakespeare's lifetime. (But see the review of Ben Higgins's book below for an alternative explanation.)

    That there was going to be a follow-up to 1 Henry 4 seems clear from the way it sets up the promised reformation of Hal and the banishment of Falstaff, so a plausible explanation is that Shakespeare intended the next play to cover his succession and exploits in France but instead chose to spread that material (perhaps rather too thinly) across two plays: 2 Henry 4 and Henry 5.

    It could be argued that an "alternative narrative arc" (p. 9) is not Hal's story but Falstaff's, which culminates in his own play: "The most pleasant and excellent conceited comedy of Sir John Falstaff and the Merry Wives of Windsor". This title the Folio heavily emends to The Marry Wives of Windsor.

    Noting that non-Shakespearian history plays sometimes dealt with the same events and historical figures as Shakespeare's history plays -- such as Thomas of Woodstock and The Famous Victories of Henry the Fifth -- Smith argues that playgoers and readers may well have read the genre with less regard to particular authors than we have and with greater regard to shared topics and stories.

    Perhaps Shakespeare's Richard 2, with its strange refusal to apportion blame for the death of Thomas of Woodstock, was meant to be understood in relation to the earlier play Thomas of Woodstock. The quarto of Shakespeare's Richard 2 calls him Woodstock but the Folio calls him Gloucester, so "It's as if the Folio, implicitly designating Richard II as the start of its own historical sequence, has to erase a reference to something that came before, even though the play is deeply dependent on versions of its own lost past" (p. 13). Smith concludes that the seriality imposed on the History plays in the 1623 Folio "was not inevitable or necessary then, and should not be now" (p. 18).

    Andy Reilly shows that the 1710 Copyright Act stimulated two publishers to create radically new editions of Hamlet that have been unjustly overlooked and that anticipated some emendations we wrongly attribute to Lewis Theobald ("Thomas Johnson's and Mary Wellington's Single-text Editions of Hamlet and the Copyright Act of 1710", MP 120[2022] 212-36). Nicholas Rowe's 1709 complete-works edition created the modern canon with his decision to conflate Q2 and Folio Hamlet.

    The important stimulant behind all this was the 1710 Copyright Act that ended perpetual copyright in 1731. Rowe, Alexander Pope, and Theobald established much of what we now think of as editions of Shakespeare, for the publishers Jacob Tonson Senior and then Jacob Tonson Junior. The Copyright Act made the Tonsons look to the editorial apparatus as the place to stake their claim for copyrightable innovation in an edition despite its reusing the text of an earlier edition. Publishers still do this today.

    Reilly focusses on four under-studied editions of Hamlet: "the first single-text octavo (1710) [hereafter O1], the first single-text duodecimo (1717, but dated '1718' in the title page imprint) [hereafter D1], the second single-text octavo (1720) [O2], and the second single-text duodecimo (1723) [D2] editions" (p. 214). The octavos were published by Thomas Johnson (who had no rights) in The Hague and the duodecimos were published by Mary Wellington (who owned the right to Hamlet, and other plays) in London. These single-play editions were cheap (8 pence and one shilling respectively) and had "extensive new textual conflation, novel metric emendations, and conjectural emendations" (p. 215) that modern editors have overlooked.

    Reilly has fully collated these four editions against the quartos and Folio and against Rowe's, Pope's, and Theobald's editions. The four draw heavily on Rowe's 1709 edition, and include its act and scene divisions and scene locators. Reilly reckons that O1 and D1 used Q11 as their base texts and O2 and D2 are based on O1 and D1 respectively. The Tonsons knew the copyright situation and engaged Rowe in order to retain control of the publishing of complete Shakespeare editions.

    Thomas Johnson was a pirate working in the Netherlands while Wellington was a legitimate rights owner working in London. Why did they independently commission "an anonymous editor or editors to prepare textually innovative editions of the play that, like the more famous collected editions, introduced new textual conflation, metric emendations, and conjectural emendations" (p. 218)? Reilly thinks that again the answer is the 1710 Copyright Act. Both thought that what they were doing would enable them to navigate the new act: Johnson to avoid being accused of piracy and Wellington to stake a claim for innovation that would give her editions protection under the new act.

    Reilly traces just which eighteenth-century editors conflated whole passages to combine the quartos and the Folio and which adopted particular words and phrases from outside their copy text to correct the script. The standard narrative, invented by Theobald, is that Rowe did no systematic work to improve the text (that is, no collating or conflating), Pope did a little (but pretended to do more than he actually did), and that he, Theobald, was the first to systematically collate the quartos and Folio and conflate to give us the best possible text (p. 221)

     Reilly thinks that this narrative is essentially true, but he wants to nuance it. Reilly's account distinguishes between the importing into one text's whole passages from a non-copy-text edition, which Reilly calls section-by-section conflation, and the adoption of individual readings from a non-copy-text edition, which he calls word-by-word conflation, and he retains the word 'emendation' only for editorial conjectural emendation. This is unfortunate, since what Reilly calls word-by-word conflation is really emendation too. After all, it is not the case that when adopting a non-copy-text reading because the copy-text reading is manifestly deficient the editor prints both readings, which is what the word 'conflation' implies.

    Theobald's claims for originality in his edition were encouraged by the publisher Tonson as they fitted his desire to claim another fourteen years of copyright protection for the Tonson Shakespeare. But Thomas Johnson's and Wellington's single-play editions of Hamlet show more textual conflation than any pre-Theobald edition and O2 "is almost as fully conflated as Theobald's edition" (p. 224). Theobald's preface to his edition does not mention Thomas Johnson's or Wellington's editions. As a pirate, Thomas Johnson would not have cared too much about challenges to his rights but Wellington might well have wanted her edition to be easily shown to be unlike existing editions so that she did not get challenged about her right to publish the play.

    Reilly reports that EEBO-TCP shows the phrase "sweet prince" being "first recorded in print in Lodge's A Margarite of America (1596)" (p. 233) and that it was a popular phrase between 1603 and 1606. In fact, John Partridge used it in his The Notable History of . . . Astianax and Polixena in 1566, and Anthony Munday used it in Palmerin d`Oliva in 1588, and others also used it before Lodge did. Reilly attributes to its use in Hamlet the fact that the phrase "explodes in popularity between 1603 and 1606 with nine separate publications (excluding editions of Hamlet) containing the phrase" (p. 233).

    Reilly uses ProQuest's version of EEBO-TCP, but EarlyPrint confirms this explosion but with different numbers and starting a bit earlier than Reilly has it, with 10 occurrences by 1598. The phrase became widely known as a quotation from Hamlet, and when Wellington's edition emended the line it occurs in to give "Now cracks the Cordage of a noble Heart; good night, sweet Prince", people would have noticed.

    Reilly counts that three conjectural emendations to Hamlet we routinely credit to Theobald in fact originated in Thomas Johnson's editions. The first of these is in place of Q2's "the dram of eale" the emendation "the dram of base" (present in Johnson's 1710 O1). The second is in place of Q2's "Coleagued with this dreame of his aduantage" the emendation "collogued with this Dream of his Advantage" (present in Johnson's O1 and O2 of 1720). The third is in place of Q2's "Of habits deuill, is angell yet in this" the emendation "Of habits evil . . ." (present in Johnson's O1 and O2).

    Margaret Tudeau-Clayton thinks that the Folio The Merry Wives of Windsor reflects a revision of the play after the accession of James 1, while the 1602 quarto (Q) shows the play before the revision ("The Kings English' and the Language of the King: Shakespeare and the Linguistic Strategies of James I", Memoria di Shakespeare: A Journal of Shakespearean Studies 9[2022] 74-101). The phrase "the King's English" occurs in Shakespeare only in Folio The Merry Wives of Windsor; it is absent from Q. Tudeau-Clayton's argument is that the new king James 1 had an explicit policy of promoting plain speaking and Shakespeare reflected this project while remaining sceptical about its being possible.

    The Merry Wives of Windsor and Henry 5 appeared in quarto versions, before James's succession, that lack references to the plainness of the King's English and then appeared in Folio versions that include it. Tudeau-Clayton concludes that they were revised by Shakespeare to include this detail. Shakespeare smuggled in the idea that what passes as plainness might in fact be a cover for malice, as Cornwall suspects of the plain-speaking Kent-as-Caius in King Lear.

    Historically, it was in Henry 5's reign that the English spoken by Londoners became officially privileged over the French formerly spoken at court. Ironically, in telling his new court and parliament in Basilicon Doron that he would use plain English and that this kind of speaking and writing would unite Scotland and England in his person, James 1 use the word 'leide' (meaning a style of speaking or writing) that was current in Scotland but not England.

    Richard Dutton noted that in Falstaff's first speech, quarto The Merry Wives of Windsor has him say that Shallow wants to complain of him "to the Council" whereas in the Folio version he says that Shallow wants to complain of him "to the King", and that in general the quarto makes no mention of the king or queen. The plays of the time reflect changing attitudes in that, as Hugh Craig has shown, references to England declined and references to Britain increased.

    Tudeau-Clayton sees Macbeth as putting a Scot at the centre of a tragedy where earlier plays had Scots as comic alien figures. In the person of Paroles in All's Well that Ends Well and especially in the exploration of plain speaking in King Lear, Shakespeare went along with James 1's project of promoting the use of plain English.

    Tudeau-Clayton reads The Merry Wives of Windsor as continuing the process begun in 1 Henry 4 and 2 Henry 4 of contrasting the plain-speaking characters (Prince Harry, the London citizens) with the linguistic and sartorial extravagance of Falstaff who must be banished both the court (in the Henry 4 plays) and the city (in The Merry Wives of Windsor). Her conclusion might usefully have engaged with the fact that in the Folio but not the quarto of The Merry Wives of Windsor Falstaff is explicitly included ("Sir Iohn and all") in the happy reintegration of the community in the play's closing speeches.

    David M. Bergeron thinks that the character of the Mercer named in the opening stage direction of Timon of Athens should not be editorially effaced ("Timon of Athens, the Absent Mercer, and Nothing Poet: 'To Th' Dumbness of the Gesture / One Might Interpret' (1.1.33-34)", Medieval and Renaissance Drama in England 35[2022] 161-77). The Folio's opening stage direction is "Enter Poet, Painter, Ieweller, Merchant, and Mercer, at seuerall doores". The Mercer is technically a 'ghost' in that he has no lines to speak. Most editors omit him. It all boils down to what one thinks of his being present but silent; is he worth keeping? The Folio list of characters omits the Mercer, but then it also omits essential characters such as the Steward/Flavius and all the women, so it is not to be trusted.

    Bergeron thinks that maybe the silent Mercer could be "moving about, showing the garments that he brings, engaging in unrecorded conversation with other characters, smiling, frowning" (p. 166). Bergeron finds it inconsistent to delete the Mercer but not delete the senators who also enter in Scene 1 and say nothing. I would respond that the Mercer is just one person in a scene-opening stage direction for multiple people, about whom the writer might not have fully finalized his intentions, while by contrast the senators have their own mid-scene entrance direction and that is more obviously intentional.

    Bergeron explores what the actual Mercers of London were like, in other plays and in reality. They dealt in expensive textiles and so their presence on stage could stand for self-indulgent excess. But they were also philanthropists and benefactors of good works. Bergeron argues that Timon becomes like the Mercer in that he descends into nothingness. Middleton was heavily involved with the City of London and its livery companies, so perhaps he added the Mercer to the play's opening stage direction. Perhaps the opening scene's speech prefix "Mer." is really for the Mercer not the Merchant. But I would say that the dialogue involving him makes it clear that "Mer." is the Merchant: "I know the Merchant", as the Poet says.

    Rory Loughnane revisits some standard narratives about Shakespearian authorship and printing in order to present new perspectives on them ("Shakespeare, Marlowe, and Traces of Authorship", Memoria di Shakespeare: A Journal of Shakespearean Studies 9[2022] 21-54). An opening footnote reports that this essay first appeared in the collection The Birth and Death of the Author: A Multi-Authored History of Authorship in Print edited by Andrew J. Power and published by Routledge in 2020. But Routledge's own website says the book was published in 2022.

    Loughnane challenges the oft-repeated claim that about half the plays in the Folio had not been printed before. He points out that if we take into account substantive differences between quartos and Folio editions of the same play then "75% rather than 50% of the works included" are "substantively new or variant" (p. 23). Loughnane is, for instance, including the Fly Scene posthumously added to Titus Andronicus as making a new variant.

    Nine of the plays in the Folio (one quarter of them) contain writing by someone other than Shakespeare: Measure for Measure, All's Well that Ends Well, 1 Henry 6, 2 Henry 6, 3 Henry 6, Henry 8, Titus Andronicus, Timon of Athens, and Macbeth. And then there are the six plays with Shakespeare's writing in that are not in the Folio: Arden of Faversham, Edward 3, Sir Thomas More, The Two Noble Kinsmen, The Spanish Tragedy, and Cardenio.

    Loughnane summarizes what we know about the performances and publication of 2 Henry 6, 3 Henry 6, and Richard 3. He gives the details of which plays were published anonymously and which were attributed, and explains that before the late 1590s plays generally were published anonymously. But even after it became usual to put Shakespeare's name on the title pages of his plays, seven plays were reprinted in Shakespeare's lifetime without his name on them: The Contention of York and Lancaster, Richard Duke of York, Arden of Faversham, Titus Andronicus, Edward 3, Romeo and Juliet, and Henry 5.

    Why these? Loughnane finds that for the first five of these the answer is that they were early collaborative plays from before the 1594 duopoly's settled practices. Thus "Shakespeare and the Lord Chamberlain's Men/King's Men were, overwhelmingly in print, a package deal" (p. 35). By contrast, we do not know precisely what Shakespeare was doing before 1594, who he worked for or with, and the datings and attributions of his plays are less secure.

    An outlier is the sixth of the seven plays, Romeo and Juliet, which was first printed in Q1 in 1597 in a version quite markedly different from Q2 in 1599 from which subsequent editions derive. Another outliner is the seventh of those seven, Henry 5, first printed in a version, Q1 of 1600, markedly different from the Folio version, and reprinted as Q2 in 1602. Loughnane summarizes the claim for the three Henry 6 plays being co-authored.

    The offence given to Shakespeare by Greene's Groatsworth of Wit was, Loughnane notes, a hard allusion to spot at the time. One would have to recognize "Tygers hart wrapt in a Players hyde" as being close to a line from the unpublished 3 Henry 6 and spot that "Shake-scene" means Shakespeare. William Montgomery pointed out that the printers of Folio 2 Henry 6 consulted the Pavier Quarto Whole Contention of 1619 for certain passages, so they at least could see that The Contention of York and Lancaster was the same play as 2 Henry 6, and presumably could see that Richard Duke of York was the same play as 3 Henry 6.

    Aleida Auld argues that Malone was not the first to read Shakespeare's Sonnets through a biographical lens, as earlier editors also did this ("Biographical Reconfigurations of Shakespeare's Sonnets: John Benson, Charles Gildon, and the Catullan Epigram", Shakespeare 18[2022] 197-222). Eighteenth-century editors mostly reproduced John Benson's edition of Shakespeare's Sonnets until Malone returned to the 1609 edition for his 1780 and 1790 editions. It is generally thought that going back to the 1609 edition made it possible to link Sonnets to Shakespeare's biography because Benson's reworkings occluded that link, depersonalizing the poems. Auld shows that the link was there in the 1640 edition. After all, the Petrarchan tradition that Shakespeare was writing within was an intensely personal one.

    Auld accepts that Benson toned down the biographical implications of the poems, but insists that his rearrangements -- such as putting multiple sonnets together to make long poem -- did not entirely eliminate the biographical. Auld details the changes and her argument seems to be that while Benson did indeed tone down the homoeroticism in places, and gave the poems names that made them seem to embody general philosophical abstractions such as "'belief', 'temptation', 'vow', and 'exchange'" (p. 200), he also made groupings such as "The Benefit of Friendship" that are necessarily personal, and invented titles that foregrounded personal relationships.

    Charles Gildon's 1709 edition, based on Benson's, stressed the biographical aspect by making clear that the sonnets are addressed to Shakespeare's mistress. Indeed, by providing a biographical essay, Rowe's 1709 edition met a desire to read the works via the life. Auld traces the increasing eighteenth-century interest in the intersection of Shakespeare's life and his works, and the history of the publication of the Sonnets in that century. Gildon referred to what we call Shakespeare's sonnets as epigrams and likened them to Catullus's epigrams and the love poems of Abraham Cowley.

    Also on the topic of Sonnets, Stanley Wells wants to dispel some stubborn myths ("Shakespeare and Sonnet Form", Cahiers Élisabéthains 109[2022] 32-43). It was Henry Tottel's Miscellany of 1557, including Thomas Wyatt's poems among others, that introduced the Italian sonnet form to England. Each of the six-line stanzas forming Shakespeare's Venus and Adonis is a sestet: the final quatrain and couplet from a sonnet. Lucrece is similar but uses seven-line rhyme royal units: ababbcc instead of ababcc.

    Wells considers it a myth that the first 125 sonnets, plus the 126th which is composed in couplets and has only 12 lines, are addressed to a young man and the rest (apart from the last two) are about a Dark Lady. The last two sonnets are two versions of a translation of a Greek poem and one (the first) revises the other; Wells thinks them schoolboy exercises.

    The reason scholars have read the first 126 sonnets as written to one person and the next 26 as written to another is that other poets' sonnets sequences were usually addressed to a person, usually a desired woman who is named (often by a pseudonym). But the only person Shakespeare's sonnets name is himself, and it really is a miscellany not a sequence. The first 17 sonnets do indeed form a sequence, but it is the longest one in the book.

    Wells sketches the homoerotic poetry of Richard Barnfield, an early admirer of Shakespeare's poetry. He explores the line "Within thine own bud buriest thy content" in Sonnet 1, with "bud" meaning "the glans of the penis" and the reference being to his wasting his semen. Wells remarks that Shakespeare must have known the young man well in order "to address him in such sexually intimate terms" (p. 42).

    Ann Thompson and Neil Taylor describe what various editors have done with certain metrical problems in Shakespeare, in particular in Q1 Hamlet ("Metrical Variation and Metrical Emendation in Q1 Hamlet", Shakespeare 18[2022] 305-18). Much of their essay is long quotations from various editions and summaries of their differences, with little analysis. On the last page they give their advice to editors of Q1 Hamlet: "surely its lineation should be left alone" (p. 317). Lastly among the articles, the present reviewer and his collaborators published a layperson's account of the Word Adjacency Network method of authorship attribution by computational analysis ("How the Word Adjacency Network (WAN) Algorithm Works", DSH 37[2022] 321-35). Their aim was to fully document the method in language understandable by readers who do not know the mathematics or Computer Science that is taught beyond the age of 16.

    So to the round-up from Notes and Queries. In our field, the most substantive of these take us back to questions of authorship attribution. Thomas Merriam reckons that not only the first act but also the last 2000 words of Titus Andronicus are by Peele not Shakespeare ("Titus Andronicus: Peele, Shakespeare, Peele", N&Q 267[2022] 97-102). The turning point is the entrance of Titus and Lavinia with a knife and a basin to desanguinate Chiron and Demetrius. Merriam's evidence for authorship is the relative frequency of the word 'and'. In Act 1 it is 4.8% of all words, which is within Peele's range of 3.9% to 5.3% but outside Shakespeare's range of 2.39 to 3.84% of all words. Across Acts 2 to 5 it is 3.57%, within Shakespeare's range.

    Merriam finds that if he breaks the play into four 5000-word chunks (the last being 146 words short of 5000) then although Acts 2-5 averages 3.57%  for 'and' there is a sharp upturn at the end. In the last 5000-words the rate is 4.57%, which is within Peele's range but not Shakespeare's. Might Peele have written the end of the play? The same picture emerges when Merriam counts the word 'but': Shakespeare's range is 0.53% to 0.92% but the first 5000 words of Titus Andronicus come out at 0.52% and the last 5000 words at 0.49%, more like Peele's typical values.

    For a fresh analysis, Merriam divides Titus Andronicus into three parts covering the whole play: Titus-alpha = start to 2.1.26, Titus-beta = 2.2.1 to 5.2.165, and Titus-gamma = 5.2.166 to the end. Titus-alpha is the same as his first 5000-word chunk in the previous experiment. Notice that Merriam is calling "2.1" what some editions call 2.2, the 26 lines from "The hunt is up" to "doe to ground".

    The ending of Titus Andronicus-beta at 5.2.165 is made on purely literary grounds, namely Merriam's reading of what starts at that point: a Peelean anti-Catholic parody of Eucharistic blood-drinking. This change in tone distinguishes the anti-Catholic ending from the main body of the play, which main body "engages Shakespeare's lasting theme of the superiority of unjustly treated heroines to the naturalist violence of their tyrants" (p. 102).

    Using this three-way division of Titus Andronicus, Merriam employs the R Stylo software package to run two experiments. In the first he takes these three parts of Titus Andronicus plus the other 35 First Folio plays plus four Peele plays -- The Arraignment of Paris, Edward 1, David and Bethsabe, and The Battle of Alcazar -- and has R Stylo find the most-frequent 688 words that are also found in 80% of these 42 texts. That is, the 42 texts being the three parts of a dlvided Titus Andronicus, 35 from the Folio, and four Peele plays. Merriam has R Stylo make a dendrogram representing the clustering of the results for these 42 texts in the sense of their similarity to one another in their scores.

    Merriam does not report which algorithm R Stylo used to make the clusters, nor what it means numerically for two texts to be on the same branch rather than on different branches. The x-axis to Merriam's Figure One dendrogram is labelled "Classic Delta distance" and runs numerically from 0.0 to 3.0. Merriam's body text says nothing about what this Delta is (presumably John Burrows's Delta metric) nor how these numbers 0 to 3 are derived from the frequencies of these 688 most-frequent words.

    Figure One's dendrogram puts all the Peele play's on one branch together with Titus Andronicus-alpha and Titus Andronicus-gamma (the beginning and ending of the play) and all of Shakespeare's plays on a different branch together with Titus Andronicus-beta (the middle of the play). Essentially the same result is achieved when R Stylo counts the 1200 most-common character 3-grams, with 100% culling instead of 80% so that every text has to have at least one occurrence of each of the 3-grams used, as shown in Figure Two. Merriam reports but does not show us that the same result is gained with character 4-grams, 5-grams, and 6-grams. This result is consistent with his previous results (from counts of 'and' and 'but') showing that Peele wrote the end of Titus Andronicus as well as the beginning

    Presumably because of the Advance Access publishing of journal articles mentioned above, Brian Vickers was able in 2022 to respond to Merriam's essay and dispute its claim that Peele wrote the end of Titus Andronicus ("The Limitations of Stylometry: Idiolect and the Authorship of Titus Andronicus", N&Q 267[2022] 207-11). Vickers begins by sketching Timberlake's work on feminine endings being introduced by Kyd and taken up by Shakespeare. Vickers gives mean-average frequencies of their use across a whole play: "Peele’s usage ranged from 0.8 per cent (David and Bethsabe) to 1.5 (Battle of Alcazar), rising to 1.8 per cent (Edward I); Greene's from 0.1 to 1.1 per cent; Marlowe’s from 1.2 per cent (1 Tamburlaine) to 3.0 (Jew of Malta), rising to 3.7 per cent (Edward II), dated to 1592, where he probably learned from Kyd" (p. 207)

    These mean averages conceal widely differing practices in different scenes of a play. For instance, Timberlake's table shows that the 0.8% for David and Bethsabe is a mean average derived from combining totals for scenes with as many as 10.9% of their lines using feminine endings with scenes using none; indeed the Prologue has 19% of lines using this feature. We see the same problem with Vickers's mean averages for the other plays. Vickers claims that Peele cannot have written the end of Titus Andronicus because its final scenes use feminine endings at the rate of  "18.3, 5.4, and 11.5 per cent", which Vickers considers impossible for Peele (p. 208.)

    But Vickers is here comparing whole-play means with per-scene means, and if we look at Peele's scenes in David and Bethsabe (as tabulated in Timberlake's book, which is Vickers's source) we find that it begins with scenes in which feminine endings appear in 10.9%, 9.0%, and 6.0% of the lines. There is of course variation in how we count feminine endings, and Timberlake has what he calls 'strict' counts as well as 'all' counts; Vickers makes no case for preferring one over the other.

    Vickers summarizes the plot of the opening scene of Titus Andronicus, and surprisingly gets it back to front: "Unexpectedly, Saturninus takes as his wife the captive Tamora, Queen of the Goths, who is accompanied by her sons Alarbus (who is soon sacrificed by the Romans), Chiron, and Demetrius, and the Moor Aaron" (p. 208). In fact, of course, Alarbus is sacrificed before Saturninus takes Tamora as his wife, and indeed we are encouraged to suspect that she agrees to marry Emperor Saturninus only to achieve the power necessary to wreak revenge on those who sacrificed him.

    Continuing his back-to-front reading, Vickers claims that next and "Equally unexpectedly, Bassianus (Saturninus' brother), seizes Lavinia, Titus’ daughter, and claims her as his betrothed" (p. 208). Again, and crucially, this happens before Saturninus takes Tamora as his wife and it is the cause not the consequence of him taking Tamora: he wanted Lavinia but she was already taken.

    Vickers lists some of the lines near the end of the play that have more occurrences of the word 'and' than we would expect from Shakespeare and argues that they are conscious artistic uses of 'and' and hence we should not think them merely symptomatic of unconscious literary style. Vickers method of quotation is rather misleading, as he uses ellipses at the ends of lines to show the omission of whole lines, while the usual scholarly convention is to add whole lines of dots to mark missing lines.

    Thus Vickers's quotation runs:

And set them upright at their deere Friends doore . . .
And on their skinnes, as on the Barke of Trees . . .
And nothing greeues me hartily indeede,

This gives the impression that three successive lines begin with 'and' but in fact the above quotations span eight lines of the play not three: the first ellipsis stands for Vickers's omission of a whole line and the second his omission of four whole lines. Vickers does this for a few quotations as part of his literary-critical argument that 'and' is being deployed for rhetorical effect.

    Quoting Titus's words to his prisoners Chiron and Demetrius in 5.2, Vickers moves the line "And this the Banquet she shall surfet on" from its proper place (where the preceding and following lines do not begin with 'and') and puts it after the two lines "And with this hatefull Liquor temper it,| And in that Paste let their vil’d Heads be bakte", thereby producing a run of three lines beginning with 'and'. This is odd as he is quite explicit that "Quotations are from the Folio text . . ." (p. 2n9).

    Vickers is quite right that 'and' is strategically deployed in the final scenes of Titus Andronicus; what is not clear is why he thinks this rules out Peele's authorship and proves Shakespeare's. Vickers ends by claiming that stylometry based on the frequencies of the most-common words (typically function words) is utterly disabled by the fact that dramatists deploy function words strategically. He believes that in their mature works dramatists use function words differentially for different characters. (This last claim was taken up by the present reviewer in NYWES for 2018 and shown to be false.)

    Darren Freebury-Jones sets out to show that Arden of Faversham Scenes 4 to 8 are by Kyd not Shakespeare ("New Perspectives on Thomas Kyd's Restored Canon", N&Q 267[2022] 15-7). He starts by responding to Jackson's essay "The 'Restored' Kyd Canon: Interpreting the Evidence of Unique Verbal Matches" (reviewed in NYWES for 2021). Jackson's method was to take Scenes 4-8 of Arden of Faversham on the one hand and the rest of the play on the other and look for trigrams found in each of those two text extracts and in the plays of Shakespeare, Kyd, and Marlowe.

    Arden of Faversham Scenes 4-8 is just 16% of the play but it provides 32% of the whole play's links to Shakespeare, and 20% of the whole play's links to Marlowe and 7% of the whole play's links to Kyd. Other independent tests have made scholars think that Arden of Faversham Scenes 4-8 is by Shakespeare, and it is most unlikely that the new evidence of a preponderance of links to Shakespeare falls by chance in exactly the place where we already suspected the presence of Shakespeare's hand.

    To test Jackson's reasoning, Freebury-Jones divides The Spanish Tragedy into chunks the same length as Arden of Faversham Scenes 4-8, which gives him six full chunks, each of which should contain about 16% of The Spanish Tragedy's links to Shakespeare plays and 16% of the play's links to Kyd plays (since each chunk is about 16% of The Spanish Tragedy). But Freebury finds that some chunks have considerably more links to Shakespeare's plays than they should have: Chunk 2 has 27% of the play's links to Shakespeare, and Chunk 6 has 21% of the play's links to Shakespeare.

    Freebury-Jones concludes that this shows the unreliability of Jackson's attribution of part of Arden of Faversham to Shakespeare since "According to Jackson's method, Shakespeare would be much more likely than Kyd to have authored the second segment" of The Spanish Tragedy. This is absurd since we all agree that Kyd wrote the whole play. Freebury-Jones results are not quite as stark as Jackson's: the bit of Arden of Faversham that Jackson tested (Scenes 4-8) was likewise 16% of the play but had 32% of the links to Shakespeare.

    The disabling logical flaw here is that Freebury-Jones went looking for anomalous chunks of The Spanish Tragedy and by testing all the chunks he manages to find one. By contrast, Jackson started with Scenes 4-8 because other independent evidence has made investigators think that Shakespeare wrote this part of Arden of Faversham. As Jackson put it, his results "would be less significant had not a combination of diverse tests previously honed in on scenes 4-8 as those within which Shakespeare’s contribution was most evident" (Jackson's p. 77).

    To dismiss Jackson's results as a random occurrence we would have to suppose a remarkable coincidence: that the anomalous preponderance of links to Shakespeare lies just where other evidence told us to look for links to Shakespeare. By contrast, since Freebury-Jones tests all the chunks he is quite likely to hit one having an anomalous preponderance of links to Shakespeare. If we trawl through any dataset we are bound to find some place where random variation produces anomalies. That is expected. What is unexpected is that when existing evidence leads us straight to one part of a text we find there an anomaly that is consistent with that existing evidence.

    Freebury-Jones ends with the claim that since the Shakespeare plays used in Jackson's test were almost all written after Arden of Faversham the obvious explanation for the unusually high number of links found by Jackson is that Arden of Faversham influenced those Shakespeare plays rather than sharing authorship with them. This explanation rather implausibly requires Shakespeare to have been influenced, time and again and in multiple plays that he wrote across the 1590s, by the same small part (comprising just 16%) of Arden of Faversham that is its Scenes 4-8.

    Where Jackson shows that the other plays that Vickers and Freebury-Jones want to add to the Kyd canon test rather more like Marlowe's than Kyd's, the explanation offered by Freebury-Jones is that Marlowe and Kyd lived together and wrote alike. Indeed, for Freebury-Jones, finding that these plays test like Marlowe actually corroborates his case: "we can expect the putative Kyd plays" to "share dense parallels with plays by his room-mate Marlowe" and finding that they do shows "they really are by Kyd" (pp. 16-17). By such reasoning we could just as easily reattribute all of Marlowe's works to Kyd too, since their testing like Marlowe would likewise tell us they are really by Kyd.

    Ian McAdam too explores Shakespeare's recollections of Marlowe ("Further Allusions to Marlowe in The Comedy of Errors", N&Q 267[2022] 211-4). Harold F. Brooks noted that the attempted exorcism of Antipholus of Ephesus in The Comedy of Errors has him say "What, will you murder me?", which echoes the king's line in Marlowe's Edward 2. The allusion continues: in Edward 2 the king's beard is shaved off with puddle water, while in The Comedy of Errors the situation is reversed as Antipholus of Ephesus and Dromio of Ephesus break free and are said (in a later report) to burn off the Doctor's beard and throw on him "puddled mire" to put the fire out. These Marlovian allusions are now added to by McAdam.

    The lines "Spread o'er the silver waves thy golden hairs,| And as a bed I’ll take them, and there lie" in The Comedy of Errors echo Marlowe's "Sometime a lovely boy in Dian's shape,| With hair that gilds the water as it glides" in Edward 2. McAdam thinks that the name of greasy Nell in The Comedy of Errors recalls that of Helen of Troy in Marlowe's Doctor Faustus. McAdam weaves these links into an argument about Shakespeare's turning the "homoerotic aggression" (p. 213) in Marlowe back into heteroeroticism.

    MacDonald P. Jackson finds evidence that The Taming of A Shrew contains at least some material from Shakespeare's original version of the play ("A New Perspective on the Endings of The Taming of the Shrew and The Taming of a Shrew", N&Q 267[2022] 102-3). Jackson has previously observed that Shakespeare's comedies often end with the anticipation of a feast in which all will learn everything about what just happened in the play. The Taming of the Shrew does not fit this pattern, since the feast is on stage and there is no promise of further revelations after the end of the play. But The Taming of A Shrew does fit the Shakespearian comic norm.

    Most scholars now think A Shrew was written after The Taming of the Shrew and that in part it recalls and mixes Shakespeare's play with other writing. But unlike The Taming of the Shrew, A Shrew finishes off the opening frame by bringing back Sly and the tapster who threw him out of the tavern at the start. The tapster ends A Shrew saying "Nay, tarry Sly, for I’ll go home with thee,| And hear the rest that thou hast dreamed tonight", which fits the Shakespearian comic pattern.

    Sona Sahoo thinks that Thomas Hickock's 1588 English translation of Cesare de Federici's account of his travels in the Middle East, India, and South East Asia, and in particular a storm taking him to the island of Sandwip, may have influenced The Tempest ("Relocating The Tempest in the Bengal Delta", N&Q 267[2022] 219-21). Sandwip was a major exporter of salt and the play mentions "brine pits". The further links claimed by Sahoo are even more tenuous.

    Various "adventurers" (Sahoo's term for "missionaries, renegade soldiers, convicts, and [Portuguese] sailors") took over Sandwip in the early seventeenth century. According to Sahoo, they "courted friendship" with the "native rulers only to exploit them for their own ends and gradually establish their autonomy. This recalls the inherently exploitative nature of Prospero's governance on Caliban's native land and his dependence on the latter's labour and resources" (p. 220). An obvious retort is that Caliban is scarcely indigenous to the island, having been born there to an immigrant mother, Sycorax.

    Portuguese inter-marriage with the natives of Sandwip produced children who were dark-skinned but blue-eyed, just as Sycorax is said to be in the play. Also, Sycorax came from Algeria where the Barbary slave trade operated, and Sandwip too was a centre for piracy. In the hybridity and subjugation of Caliban, Sahoo sees the "lived reality of socioeconomic circulations, geographic relocations, and racial miscegenation that once marked the Bengal delta" (p. 220) that Sandwip sits in.

    T. M. Vozar traces a variant in the three early editions of Hamlet ("Mallecho or Malligo? A Crux in Hamlet Revisited", N&Q 267[2022] 291-3). The three readings are "This is myching Mallico, that meanes my chiefe" in the 1603 quarto (Q1), "Marry this munching Mallico, it meanes mischiefe" in the 1604-5 quarto (Q2) and "Marry this is Miching Malicho, that meanes Mischeefe" in the Folio. In all three editions the 'Mallico'/'Malicho' word is capitalized and in Q2 and in the Folio it is italicized too, so presumably a proper noun was meant.

    Vozar's suggestion is that this is a form of the word 'Malligo' meaning wine from Malaga in Spain. Shakespeare uses Canary to mean Canary wine in The Merry Wives of Windsor and Twelfth Night. (And in 1 Henry 4 too but this is not mentioned by Vozar.) The idea of munching Malaga wine (as in Q2) obviously makes little sense, but what about miching it (as in Q1 and the Folio)? Vozar is unconvinced by any of the explanations of this that he can think of.

    Vozar ends by suggesting that maybe 'Malaga' refers not to wine but to Malaga raisins, said in John Webster's The Devil's Law Case and in John Gerard's Herbal to make one speak at length. Perhaps in Hamlet it is meant as an ironic reference to the speechless mime going on too long. Vozar concludes that he "can hardly claim to have resolved this vexing crux" in Hamlet (p. 293).

    Paul Hammond adds to the previously noted influence of Seneca's play Medea (as translated by John Studley) on Shakespeare and Middleton's Macbeth ("A Further Echo of Studley's Translation of Seneca's Medea in Shakespeare's Macbeth", N&Q 267[2022] 217-9). Medea says "I in the wandering sky,| In nimble-wheeled wagon swift, will ride advanced high", to which Jason replies "Go through the ample spaces wide, infect the poisoned air,| Bear witness grace of God is none in place of thy repair".

    Hammond likens this to Macbeth's remark about the witches: "Infected be the air whereon they ride| And damned all those that trust them". Connecting the moments are the slaughter of children and the idea of God's grace being absent from where Medea is or the witches are. Hammond thinks that Studley's word 'infect' struck Shakespeare as particularly apt for what he was exploring in Macbeth.

    John Henry finds a supernatural significance to rain in Macbeth ("'Let it Come Down': On Banquo's Murder in Macbeth", N&Q 267[2022] 215-7). Just before he is murdered, Banquo says "It will be rain tonight" and the First Murderer responds in an aside "Let it come down". This completes Banquo's thought and meter if understood in a literal sense about the rain, but it also completes the thought in a cosmic sense. Something impersonal ("it") is at work to bring about Banquo's destruction. But also, argues, Henry, there is the rain of stage blood coming down. And since the ancients, the unnatural phenomenon of the raining of blood from the sky has been understood as a portent of cosmic disorder.

     After the catastrophe of Actium in Antony and Cleopatra, what does Antony mean by saying to Cleopatra "See| How I convey my shame out of thine eyes"? It is not, according to Russell M. Hillier, that he simply turns his ashamed face away so that Cleopatra cannot see it ("'See/How I Convey My Shame Out of Thine Eyes': Breathing Dramatic Life Into an Ancient Proverb in William Shakespeare's Antony and Cleopatra, III.xi.51-4", N&Q 267[2022] 293-7). This cannot be it, since ". . . Antony’s words indicate that he is receiving his shame while looking directly into Cleopatra's eyes" (p. 295).

    Unfortunately Hillier does not tell the reader exactly which words in the speech indicate that Antony is looking at Cleopatra, although the whole point of his note rests on this. Hillier claims that Antony means that he "looks into Cleopatra's eyes and from there gathers an acute sense of the depth of his shame" (p. 295). Hillier traces the history of the proverb that 'Shame is in the eyes' in writings by Erasmus and Aristotle.

    Izumi Hombu explores the significance of the word 'latch' to Shakespeare ("Latches in Shakespeare", N&Q 267[2022] 106-8). Shakespeare uses 'latch' as a noun in Lucrece and The Winter's Tale. In the former, the latch on the door to Lucrece's bedchamber is said to be "yielding" to Tarquin's hand. But is that a failure of the latch or is that what a latch is meant to do?

    Hombu explores the history of door construction and concludes that latches could be designed to allow someone to let themselves in from outside. But that is distinctly not what Polixenes means when he warns Perdita she must never: "These rural latches to his [Florizel's] entrance open" in The Winter's Tale. These are "rural" latches because they are not the sophisticated urban type that had a mechanism to enable them to be operated from the outside. The latches in Lucrece are the urban type.

    Ceri Sullivan traces in Coriolanus a distinct line of imagery of good-versus-bad wheat ("Musty and Unbolted Grain in Coriolanus", N&Q 267[2022] 214-5). The citizens in the opening scene want the "superfluity while it were wholesome" but are themselves called by Martius the "musty superfluity". Comenius says that Coriolanus will not bother to separate and save his Roman friends from the "noisome, musty chaff" and reuses the phrase to call the Tribunes that "musty chaff". Menenius's Fable of Belly has the stomach pass on the flour while retaining the bran, and Menenius later says that Coriolanus cannot sift his words (using "bolted language") but rather "Meal and bran together| He throws without distinction".

    Gustav Zamore finds that the "worm of conscience" that Richard 3 refers to, and is plagued by before the Battle of Bosworth, has a distinct history in medieval theological theory and appears twice in the Coventy Mystery plays that we think Shakespeare saw ("The Medieval Origins of the Worm of Conscience in Shakespeare's Richard III ", N&Q 267[2022] 103-6). Siobhan Keenan strengthens the circumstantial evidence linking the joiner and royal performer John Birche of Saint Stephen Coleman Street parish with James Burbage, who also lived there and may have been Birche's apprentice ("New Evidence About Tudor Royal Players, George and John Birche and Their Connection with James Burbage", N&Q 267[2022] 204-7). Specifically, she finds a parish register death record of 1558 for "John Bretche" who is likely the same man.

    Lastly, Brian Vickers shows that a speech in Kyd's The Spanish Tragedy is essentially lifted from Thomas Watson's Hekatompathiai which is itself lifted from Ovid's Ars Amatoria ("Thomas Watson, Thomas Kyd, and the Re-use of Ovid", N&Q 267[2022] 88-9). Since, as we have seen above, Vickers rejects Taylor's attribution of Arden of Faversham to Watson, the implication behind this is presumably that where Arden of Faversham has verbal matches to Watson's poetry we should not take this as evidence of Watson's authorship of Arden of Faversham, since Kyd was apt to borrow from the writing of his friend Watson.

Books reviewed

Anonymous Arden of Faversham, ed. Catherine Richardson, Arden Early Modern Drama. Bloomsbury [2022]. 360 pp., £14.29, ISBN 978-1474289290

Freebury-Jones, Darren Shakespeare's Tutor: The Influence of Thomas Kyd. Manchester University Press [2022]. 248 pp., £20.99, ISBN 978-1526182616

Gossett, Suzanne Shakespeare and Textual Theory, Arden Shakespeare and Theory. Bloomsbury [2022]. 272 pp., £21.15, ISBN 978-1350121249

Higgins, Ben Shakespeare's Syndicate: The First Folio, its Publishers, and the Early Modern Book Trade. Oxford [2022]. 310 pp., £87.55, ISBN 978-0192848840

Yarn, Molly G. Shakespeare's 'Lady Editors': A New History of the Shakespearean Text. Cambridge University Press [2022]. 352 pp., £22.40, ISBN 978-1316518359