"Shakespeare: Editions and Textual Studies in 2021" Not the Year's Work in English Studies
No major scholarly critical editions appeared this year. The Arden Shakespeare Third Series was completed last year with A. R. Braunmuller and Robert N. Watson's Measure for Measure and the Arden's Fourth Series has yet to produce its first fruit. The Oxford Shakespeare series started with Gary Taylor's Henry 5 in 1982 and ended with Anthony B. Dawson and Paul Yachnin's Richard 2 in 2011.
There were also no monographs or special issues of journals on our topic this year. Darren Freebury-Jones edited what sounds like a volume of the journal Early Modern Digital Review on "Computational Approaches to Examining Early Modern Texts" but this turns out to be just a series of short reviews of particular digital tools and datasets. Similarly, Lukas Erne edited The Arden Research Handbook of Shakespeare and Textual Studies but this is "a reference work aimed at advanced undergraduate and graduate students, and scholars" (p. 1). The book claims to contain no fresh research discoveries, so it is not reviewed here.
We turn then to journal articles, of which the most consequential is Heejin Kim's demonstration that memorial reconstruction cannot explain the fluctuating similarities between the 1594 quarto (Q) of The Contention of York and Lancaster and Folio 2 Henry 6 and between the 1595 octavo (O) of Richard Duke of York and Folio 3 Henry 6, but co-authorship can ("Visualizing Textual Similarity of Shakespearean Suspect Texts: An Examination of the Henry VI Plays", DSH 36[2021] 919-33). The decisive fact that seems to show that memorial reconstruction underlies the 1602 quarto of The Merry Wives of Windsor is that its likeness to the Folio fluctuates with the Host's being on or off stage. The hypothesis is that the actor playing the Host was the 'reporter' who made the memorial reconstruction.
To explore such cases, it would be most useful to have a quantifiable and a visualizable measure of how similar each bit of one edition (say, an early quarto) is to the corresponding bit of another edition (say, the Folio version of the same play), with perhaps breakdowns for likeness by role so that we see how particular characters stand out. The memorial reconstruction theory has been applied to other play editions besides The Merry Wives of Windsor -- such as the early quartos of Hamlet, Henry 5, and Romeo and Juliet -- and Paul Werstine in particular has shown that it does not convincingly explain their features. But what about The Contention of York and Lancaster/2 Henry 6 and Richard Duke of York/3 Henry 6?
Kim sets out to show that in these two cases, Peter Alexander, Alfred Hart, and most recently Kathleen O. Irace were wrong: the statistical evidence does not support the idea that the 1590s editions were memorially reconstructed. Kim surveys the ways that Alexander tried to explain away the bits of quarto The Contention of York and Lancaster that are by his supposed reporters but are unlike Folio 2 Henry 6 and the bits that are not by his supposed reporters but are like Folio 2 Henry 6. If we count how many lines in the plays that Alexander actually explained, it is under one-third for both plays.
To test if the memorial reconstruction hypothesis is viable for the two-thirds of the plays' lines not explained by Alexander, Kim constructed his own parallel texts using "manual assignment of corresponding F[olio] lines" (p. 933n9). He found that he had to manually move a few words to a preceding or following line to get the alignment exactly right and manually tweak "Unusual inflectional endings and proper names" (p. 933n9), but he thinks that the subjective judgements he used to make these tweaks did not affect his results substantially.
Once he established which lines in Q or O pass a minimum threshold (subjectively judged) for being the same as which lines in the Folio, Kim ran a similarity measure on the two lines in each pair to give a score between 0 (for entirely unalike) and 1 (for entirely alike). He incorporated lemmatization and spelling regularization functions from the software called MorphAdorner. This last step means that quarto The Contention of York and Lancaster's "I would false murtherous coward on thy knees" and Folio 2 Henry 6's "I would, false murd'rous Coward, on thy Knee" are identified as being identical except for the difference between singular and plural in 'knee'/'knees'.
To compare text strings, Kim uses what is called Dice Similarity, which counts in any two sets of words how many the sets have in common and then doubles this and divides the result by the total number of words in the two sets. In sets, there can be no duplicates, so the set of words for the line "Tomorrow, and tomorrow, and tomorrow" is just two words, namely 'tomorrow' and 'and', and the set of words for the line "powder me, and eat me too, tomorrow" is just the six words 'powder', 'me', 'and', 'eat', 'too', and 'tomorrow'. The Dice Similarity for these two lines is the overlap between the two sets, which is two words ('tomorrow' and 'and') doubled to make four and then divided by the total number of words in the two sets (which is eight), to make a Dice Similarity of 0.5. Kim usefulyl provides a table of example pairs of lines and their Dice Similarities and talks the reader through each one.
One slight complication is that in the explanation Kim refers to the measure using the average of the two sets size and it is not obvious where that average comes in. It emerges because dividing double the number of words in common by the sum of the number of words in the two sets is the same as dividing the number of words in common by the average of the number of words in the two sets. There are 4 words in common, namely 'set', 'on', 'my', and 'head' between "stand ready to set it on my princely head" (9 words) and "and on my head do set the diadem" (8 words). Twice 4 is 8, and 8 divided by 17 is about 0.47. Equally, 4 divided by the average of 9 and 8 (which average is 8.5) is about 0.47. Kim defends his use of sets of words in which order is not recorded on the grounds that transposition of words is common in textual transmission.
Kim's Table Two is a visualization of the Dice Similarity for each of the sections of The Contention of York and Lancaster and Richard Duke of York upon which Alexander gave his subjective verdict regarding the likeness of Q or O and the Folio at that point. He also provides visualizations of Dice Similarity across the entirety of quarto The Contention of York and Lancaster and Folio 2 Henry 6 and octavo Richard Duke of York and Folio 3 Henry 6, bearing in mind which character's actor Alexander thought was doing the reporting in the 1590s edition. For the bits of the plays that Alexander paid special attention to, the Dice Similarity is "in general agreement with his [Alexander's] observation: with regard to reporters' lines, an average similarity score of 0.85, or 85%, persists for both texts in contrast to 0.63 and 0.66 for nonreporters' parts in The First Part of Contention and [Richard] Duke of York, respectively" (p. 922).
But Alexander was not looking at the whole text of each play. Hart did scene-wise comparison, but he still focussed on the particular characters proposed by Alexander and found that their parts are better reported -- in the sense of the being closer in wording to the Folio -- than the play is in general. But Kim points out that this is misleading. Some roles must be better reported than others, just by random variation, and once Alexander selected out those roles (based on his subjective judgement of their better reporting) Hart was bound to find that those roles were better reported.
In Kim's analysis, the best reported large roles in The Contention of York and Lancaster/2 Henry 6 are Warwick and Suffolk and in Richard Duke of York/3 Henry 6 they are Warwick and Clifford, but in both plays some smaller roles are even better reported. Kim looked at the variability within the alleged reporters' roles, and the problem is that while the average accuracy of reporting is high the results are notably variable: ". . . a large number of the alleged reporters' lines are scored lower than the grand average score while that of nonreporters' lines are given higher scores than the grand mean" (pp. 924-925).
Kim confirms that sometimes the quality of reporting goes up when the alleged reporter enters the stage and down when he leaves, but there are many occasions when it goes up or down without those entrances or exits happening. Alexander got over this obstacle to the memorial reconstruction hypothesis by supposing that the reporters also had access to bits of the written scripts of the plays. But as R. B. McKerrow pointed out, manuscript playbooks usually had writing on both sides of each leaf, so we would expect each stint of good reporting to be accompanied by another good bit nearby (the bit of the play written on the other side of the leaf) and we do not see this happening.
Kim then comes to what is really going on. If we look at the similarity score across the whole play, it goes up for some scenes and down in others. It is not a perfect correlation, but variation-by-scene seems to be one potential explanatory factor alongside the potential explanatory factor of particular speaking roles being collectively well reported and the explanatory factor of all roles being well reported when particular reporters are present on stage.
In an analysis of variance, Kim finds that in The Contention of York and Lancaster/2 Henry 6 the reporters' presence accounts for 3% of the variance, the role being reported accounts for 16%, and scene divisions account for 14% of the variance. In Richard Duke of York/3 Henry 6 the reporters' presence is "marginally associated" (p. 930) with the similarity, and the role being reported accounts for only 12% of the variance, but 22% of the variation in similarity scores is accounted for by the scene divisions. Thus ". . . for both The First Part of Contention and [Richard] Duke of York, neither reporters nor roles can explain the fluctuating similarity between suspect texts and the Folio texts with certainty. Instead, scene division is a considerable factor in explaining fluctuations in [Richard] Duke of York" (p. 930). This is consistent with the plays being collaboratively written in units of acts and scenes and then revised by Shakespeare alone, which is what recent editors of the plays propose.
Also arguing against the theory of memorial reconstruction is Laurie Maguire and Emma Smith's claim that Folio The Merry Wives of Windsor reflects an expansion upon the play as originally written and published in the 1602 quarto ("Theater, Revision, and The Merry Wives of Windsor", SQ 72[2021] 177-202). Maguire and Smith survey the history of explanations for the differences between the 1602 quarto (Q) and 1623 Folio versions (F), including the idea, which they pursue, that the version underlying Q was revised to make the version underlying F. This, they point out, was argued in Peter Grav's essay "Money Changes Everything" (reviewed in YWES for 2006) and Richard Dutton's book Shakespeare, Court Dramatist (reviewed in YWES for 2016)
Smith and Maguire also approve of Alan Farmer's work (reviewed in YWES for 2015) showing that on play title pages Shakespeare was much more often than other dramatists referred to as an author who revised his plays and that this claim was probably true. Smith and Maguire's approach is literary critical, assessing the Folio's material that is not present in the 1602 quarto from the perspective that ". . . comedy is a genre of amplification, repetition, and the unnecessary" (p. 181). We have tended, they observe, to treat play revision as largely a verbal matter and not attended to structural revisions in plays.
Maguire and Smith think that W. W. Greg had as much evidence for Folio The Merry Wives of Windsor being an expansion upon the 1602 quarto version as he had for the quarto being a reduction of the Folio version, but he nonetheless went for the latter explanation. They find evidence against the Host being the maker of a memorial reconstruction of the play in the fact that the quarto and Folio do not always align when he is on stage and the quarto sometimes gives to others lines that are the Host's in the Folio. They remark that ". . . an actor can be expected to remember which lines belong to him" (p. 185n40). True, but I would have liked to hear them explain the remarkable fact that on multiple occasions the quarto and Folio suddenly become identical when the Host enters the stage.
Like memorial reconstruction, Maguire and Smith observe, revision also would most likely affect some parts and not others, and in any case revising the actors' parts rather than the whole play would have a "practical advantage" (p. 186). I am not seeing why they think that. Surely an author would find revising a single document of the play more convenient because he can see the flow of the scene, which you cannot so easily do when reading across multiple documents at once. The only point in writing revisions on the parts would be to preserve the cues, which would be worth doing if the current company had just learnt them all. Maguire and Smith cite Tiffany Stern's book Documents of Performance (Stern's pp. 244-245) on this point about the practical advantage of revising using the parts but Stern has nothing to say about this on those pages of her book.
Maguire and Smith quote Scene 16 of The Merry Wives of Windsor in which Simple (Slender's servant) asks Falstaff about the chain of Slender's that Nim stole. Only in the Folio does Simple also ask about whether Slender will achieve marrying Anne Page. Critics have seen the absence in the quarto of the 14 lines about Anne Page as an omission in the quarto, but Smith and Maguire argue that the Folio simply expands the scene for more fun at Simple's expense.
Maguire and Smith next turn to 'transposition' in two senses: lines given to a different character in the quarto and the Folio, and lines given to the same character but in different parts of the play in the quarto and the Folio. They detail some examples that are rather too complex to summarize and in which there are no verbal problems in the quarto, as Greg acknowledged while still calling the quarto's version a mutilation of the Folio's.
The key bit of evidence of memorial reconstruction, according to Greg, happens in Scene 2 when Doctor Caius finds Simple in his closet. Simple came on an errand from Hugh Evans to get Mistress Quickly (Caius's servant) to help Slender marry Anne Page. But only in the Folio does Simple mention Evans's role in this and only in the Folio does Caius acknowledge that Evans sent Simple. Yet in both versions Caius immediately pens a challenge to Hugh Evans. The problem is that in Q's version Caius has no reason to challenge Hugh Evans, since he has not been told by Simple that Hugh Evans sent him on this errand.
Maguire and Smith dismiss this detail as not enough to hang a whole textual theory on, since the plot is complex and we all lose track of it, and in any case it is just "one missing phrase" in the quarto "for which a compositor might be responsible" (p. 191). But in fact, the Folio version of the scene twice mentions Hugh Evans -- once when Mistress Quickly says that Hugh sent Simple and once when Caius asks Simple if Hugh Evans sent him -- so it is not one missing phrase but two. The Folio takes trouble to make sure that the audience understands that Hugh Evans sent Simple, which is why Caius is angry at Hugh Evans.
Maguire and Smith turn to Scene 6 (Folio 2.2) in which the Folio-only material is a long speech (overheard by Pistol) in which Mistress Quickly tells Falstaff about Mistress Ford as a young woman being wooed by courtiers, and Pistol makes an aside indicating his sexual interest in Mistress Quickly. Maguire and Smith approve of Giorgio Melchiori's claim that this Folio-only material for Pistol is there to keep him in the play so he can reappear in Act 5 and to give motive for his later being married to Mistress Quickly in Henry 5. (This of course supposes that Henry 5 followed The Merry Wives of Windsor when in fact it is more likely that it preceded it, as shown by B. J. Sokol in an article reviewed in YWES 2009.)
For Maguire and Smith this aside by Pistol is not a case of the quarto reporter omitting something but the "F[olio] adapter" (p. 192) tightening the connexion to the history plays. Irace rejected the revision hypothesis because she thought that it would not focus on particular characters, but Maguire and Smith think that character-focussed revision is exactly what we see in other plays revised for revival: The Spanish Tragedy, Doctor Faustus, Mucedorus, A Fair Quarrel, The Malcontent, and Titus Andronicus, all of which got expanded in revision. The Folio-only Latin Lesson scene in The Merry Wives of Windsor is often complained of as superfluous to the plot, but Smith and Maguire defend superfluity itself as a feature of comic theatre.
Maguire and Smith detail the differences between the A and B texts of Christopher Marlowe's Doctor Faustus, focussing on the latter's expansion of comic material. They see this as analogous to how Folio The Merry Wives of Windsor expands on the comic material in the quarto. Maguire and Smith sum up that the revisions of The Merry Wives of Windsor from the quarto to the Folio version: i) amplify by repeating fun bits, ii) would have needed a fresh manuscript since it would be impossible to fit the new material onto an exemplar of the quarto, iii) could have been written by Shakespeare or someone else, iv) give more work to boy actors, and v) tighten the connexion to court culture.
On the same topic of revision, Duncan Salkeld argues that it does not explain the differences between the 1608 quarto (Q1) and 1623 Folio (F) version of King Lear ("Q/F: The Texts of King Lear", The Library 22[2021] 3-32). According to Salkeld, the copy for the quarto was a manuscript virtually identical to the one used as copy for the Folio (perhaps even the same manuscript), and the poor quality of the 1608 quarto arose because the typesetting was done by dictation.
Salkeld believes that no revision of King Lear took place after 1608 "save for a few corrections made in Q2" (p. 4), that is, the quarto of 1619. Thus the Folio-only passages were present in the manuscript used in Nicholas Okes's workshop to print the 1608 quarto but were omitted in that printing, and the quarto-only passages occur by their being marked for theatrical omission in the manuscript used to print the Folio.
"Of the twelve extant quartos, no two are identical" (p. 5), writes Salkeld, on account of stop-press correction. In fact this is untrue. As is clear in Greg's table that Salkeld cites at this point, the exemplars with the sigla BM1 and BODL2 are identical and so are the exemplars TCC and HARV. Salkeld asserts without argument that it is unlikely that Shakespeare would have made the thousand or so small verbal differences between the quarto and Folio version since they are almost pointless. Of course the argument he needs to refute, but ignores, is E. A. J. Honigmann's in The Stability of Shakespeare's Text (1965) that this is exactly what we should expect Shakespeare to do in copying his own work out fairly. Salkeld surveys the publications rejecting the revision hypothesis and endorses their recurrent claim that quarto and Folio King Lear are too much alike despite the many small differences: had Shakespeare revised the play he would have made larger changes.
Salkeld lists some variants where in his view the quarto clearly has a garbling of what we find in the Folio, such as the stratagem of shoeing "a troope of horse with fell" in the quarto where the Folio has "A Troope of Horse with Felt". In fact the quarto's "fell" is acceptable since it means animal skin and occurs elsewhere in the play in that sense; according to Salkeld it is one of the quarto's "absurdities" (p. 9). Regarding the indifferent variants, Salkeld again simply asserts that Shakespeare would not have bothered to make them.
The twentieth-century scholarly debate about the quarto/Folio relationship, prior to the emergence of the revision hypothesis, is summarized by Salkeld. First is E. K. Chambers's thought that the manuscript for the quarto might have been -- which Salkeld hardens into a certainty that it was -- created by surreptitious shorthand in the theatre during a performance. Salkeld sketches G. I. Duthie's and Alice Walker's support of the notion of dictation in the line of transmission leading to the quarto, and P. W. K. Stone's conclusion that underlying the quarto is a memorial reconstruction made by a reporter in the theatre.
Salkeld gives an account of the dispute about whether the first quarto or the 1619 second quarto (Q2) was the copy for the Folio. Trevor Howard-Hill argued in 1982 that if a quarto was used then it was the second quarto, not the first as Greg had argued on the basis of stop-press corrections in the first quarto. Howard-Hill thought that this second quarto was used merely to help in the making of a manuscript from which the Folio was printed. This argument Salkeld seems to misunderstand, for he reports Howard-Hill claiming that the Folio was printed directly from the second quarto, when in fact Howard-Hill's 1982 article explicitly rejected that idea.
Next Salkeld reports Richard Knowles's 2008 rejection of the idea that the Folio was set from either quarto as if this were an attack on Howard-Hill's 1982 position when in fact it is consistent with it. (To be fair, this arises in part because Knowles himself misrepresented Howard-Hill's 1982 paper.) Salkeld reports Knowles's claim that "F rejects some 1,300 substantive readings found in Q2, and seems to accept a mere 90" (p. 14) as if this has been established ("Knowles pointed out") when in fact Knowles made this claim in 2008 without citing any evidence for it.
Salkeld's summary of all this is that "In sum, Howard-Hill doubted Q's influence on F, and Knowles doubted Q2's . . ." and hence "The case for Q as a first shot and F as a later revision was weakening under scrutiny" (p. 15). This misrepresents Howard-Hill's view, which in fact changed between his 1982 paper and a 1986 paper ("Q1 and the copy for Folio Lear") in which Howard-Hill accepted that the Folio was printed from an exemplar of the second quarto, and is also a non-sequitur since the case for revision is not dependent on the nature of the copy for the Folio.
Reporting René Weis's rejection of the revision hypothesis, Salkeld takes the example of the word 'pottage' in the first quarto being 'porridge' in the Folio and "since one [of these readings] is an attempt at the other" Salkeld concludes that "only one of these readings can be correct" (p. 15). But potage and porridge were the same thing -- a kind of thick soup -- so not only might both readings be correct but also this is not even a substantive variant. Here as with pairs such as 'metal'/'mettle' and 'travel'/'travail' a distinction we now make was not available in Shakespeare's time.
Salkeld begins to address the genetic line of the early editions and reports that "Almost all studies of the problem have taken the evolution of the Lear texts to have been Qa->Qb->Q2->F" (p. 16). It is hard to understand what the arrows mean in Salkeld's line of descent. No one disputes that the copy for the second quarto was an exemplar of the first quarto, so there Salkeld's arrow must mean "was printed from", but that cannot be the meaning of the arrow in "Qa->Qb" since the stop-press corrected Q1 (called Qb by Salkeld) was not printed from the uncorrected Q1 (called Qa by Salkeld). Moreover, Salkeld is mistaken in thinking that the second quarto was printed from an exemplar of the first quarto in which all the readings altered by stop-press correction were in the corrected state, as is asserted by his "Qb->Q2". Rather, there are readings in the second quarto that undoubtedly come from the uncorrected state of the first quarto and since these readings are crucial to the arguments he is trying to survey it is clear that Salkeld does not understand those arguments.
Again Salkeld misunderstands the combined effect of Howard-Hill's 1982 article on the second and not the first quarto being the quarto that influenced the Folio and Knowles's 2008 rejection of Howard-Hill's argument as being a rejection of the Folio having "any ancestry in Q at all" (p. 17). Salkeld works through the examples of agreements-in-error between the first quarto and the Folio that P. A. Daniel listed in his 1885 facsimile of the first quarto, which showed that the Folio was printed from quarto copy. Salkeld's aim is to show that the readings are explicable by the first quarto being printed from a manuscript much like the one underlying the Folio.
A key example that Salkeld tackles is this:
Uncorrected First Quarto
Bast. Sir I thought it fit,
To saue the old and miserable King to some retention,
Whose age has charmes in it, whose title moreCorrected First Quarto
Bast. Sir I thought it fit,
To send the old and miserable King to some retention, and ap-
Whose age has charmes in it, whose title more, (pointed guardSecond Quarto
Bast. Sir I thought it fit,
To send the olde and miserable King
To some retention, and appointed guard,
Whose age has charmes in it, whose Title more,Folio
Bast. Sir I thought it fit,
To send the old and miserable King to some retention,
Whose age had Charmes in it, whose Title more,
Salkeld argues that the words "and appointed guard" found in the corrected state of the first quarto were "probably supplied by Okes's proof-corrector" (p. 20) but he does not give a coherent reason why someone would go to this trouble since it does not, as he acknowledges, solve anything. Moreover, it was a particularly difficult addition to make in stop-press correction since it required altering and rejustifying two lines because of the turn-under; it is hard to see the inexperienced Okes's workmen going to this trouble on the basis of a reading they themselves had invented rather than one they found in the manuscript.
The Folio has Lear swear by the "miseries of Heccat" (where the quartos have "mistresse of Heccat") and the Second Folio supplies what most people think is the correct reading of "mysteries of Hecat". Salkeld makes the surprising claim that the Folio's "miseries" is correct, on the analogy of the "terrors" and the "railing" of Hecate written about elsewhere and because ". . . Shakespeare never uses the word 'mystery' in a pagan, occult sense" (p. 21).
In fact, Shakespeare has Lear use 'mystery' in the required sense later in the play when he consoles Cordelia that together in prison they may reflect upon "the mystery of things" like "Gods spies" watching earthly events "ebbe and flow by th'Moone". Lear's oath near the start that Salkeld is discussing shares with the consolation near the end the religious invocation of heavenly bodies as shapers of human destiny: "the sacred radience of the Sunne | . . . the operation of the Orbes,| From whom we do exist, and cease to be" in the oath, and the ebb and flow of human affairs under the moon in the consolation. Repeating the pagan, occult sense of 'mysteries' as part of this mirroring is distinctly Shakespearian patterning, whereas Salkeld's suggestion of "miseries of Heccat" is poetically weak.
Where the quartos and Folio have "Historica passio" instead of "Hysterica passio" (the correct reading supplied by the Fourth Folio), Salkeld supposes that the mistake was in the manuscript underlying the first quarto (from where the second quarto inherited it) and in the manuscript underlying the Folio (which might have been the same manuscript as that used to make the first quarto) rather than the Folio inheriting this error from the first quarto via the second. Since everyone agrees that the first quarto was poorly printed, it is easy to imagine this error occurring in that process and hard to imagine why it would stand in a manuscript of the play, much less two of them.
Madeleine Doran listed the quarto/Folio variants that seem to be mishearings. Salkeld quotes Joseph Moxon referring to copy being read aloud in the printshop for the purpose of proofing and suggests that "It is not implausible to think that copy might have sometimes been initially set from dictation" (p. 22). Salkeld cites the scant evidence for this occasionally happening, including the infamous ". . . my head is broken, | Within a parenthesis, in every corner" from the first edition of John Fletcher's play The Elder Brother, where instead of the words "Within a parenthesis" appearing as dialogue the typesetter should have understood this as an instruction and set "(my head is broken)".
Salkeld gets this example from McKerrow's 1927 primer on bibliography and omits to mention that McKerrow conceded that this mistake might have arisen not by dictated typesetting but by a proof corrector writing "within a parenthesis" as a marginal instruction for alteration of the type. Salkeld thinks that typesetting-by-dictation helps explain the first quarto's mistaken setting of verse as prose, but it is hard to see why someone reading aloud a manuscript would not say a phrase such as "new line" as needed, since a similar phrase would be needed whenever a new speech was to be started.
There is plenty of verse set correctly as verse in the first quarto. Stanley Wells's first-quarto-based edition for The Oxford Shakespeare records his relining of 1092 lines (Wells's pp. 293-302), which is less than a third of the play. Some of the other two-thirds is prose correctly set as prose, but much of it is verse correctly set as verse, for which a dictator would have to have verbally indicated where he wanted a new line to start. Moreover there are 64 examples of turn-ups and turn-downs where Q1's compositor for some reason took considerable trouble to make the line end on a particular word.
Salkeld claims that the evidence for dictation is greater than that acknowledged by Doran, but he is including such things as added or omitted 's' at the ends of words. It is hard to see why he thinks this is something that dictation is prone to, especially where the difference amounts to a whole syllable, as in his example 'reuenge'/'reuenges'. He includes "the definite article 'the' for the pronoun 'thee'" (p. 24), but of course this is not even an error: either spelling was acceptable for both words.
Salkeld lists the quarto/Folio variants where he thinks the quarto reading is either a misreading or a mishearing of the Folio reading. Salkeld refers to "a Gentleman (whom Q, in a moment of confusion, forgets is a Doctor)" (p. 27). In this case, the scene of Lear's recovery, it is impossible to see the first quarto's readings as garblings of the Folio's readings, as Salkeld insists they are. The first quarto has the stage direction "Enter Cordelia, Kent and Doctor" (K1v) and then seven speeches assigned to "Doct." The Folio instead has "Enter Cordelia, Kent and Gentleman" and assigns the Doctor's speeches to "Gent." It is impossible to see this happening by the makers of the quarto mishearing or misreading a manuscript containing the Folio's words. Also, Salkeld does not explain how the speech prefix for the play's final speech got misheard/misread so that the first quarto assigns it to "Duke." where the Folio has it as "Edg.".
Salkeld considers alternative explanations for the origins of the first quarto and finds stenography and scribal copying unable to account for the aural errors. He does acknowledge a few quarto/Folio differences that cannot be explained as mishearings or misreadings but dismisses them as "A handful of puzzling instances" (p. 30). Salkeld does not accept the abundant evidence that the Folio was printed from an exemplar of the second quarto: he thinks the Folio's copy was a manuscript.
To explain the instances of stage directions being fuller in the first quarto than in the Folio, Salkeld suggests that the person doing the dictation in Okes's printshop "had an eye and an ear for the dramatic" (p. 31). There was no revision of the play: behind the first quarto and the Folio there was either one manuscript or two that were "almost identical" (p. 31) apart from the cuts marked on the one used to make the Folio.
Salkeld supposes that dictation was resorted to because it speeds things up and because the light was not good enough to read by. But he never explains just why he thinks dictation speeds things up: how does someone reading the text aloud to a compositor save time over the compositor reading it for himself? And if the printshop was too dark to read by how did the dictator manage to do the reading aloud? Certainly, compositing can be done with less light than is needed to read by, but distributing cannot be because the compositor has to read the type to distribute it. (Indeed, it takes more light to distribute type than to read a manuscript, since the type are harder to see.) And distributing must have taken place during the printing of King Lear since Okes had nowhere near enough type to do the job without it.
Still on the subject of revision, Richard Stacey argues that Folio 3 Henry 6 reflects a revised version of the play that was printed as the 1595 octavo of Richard Duke of York and that the revisions toned down the play's Yorkist sympathies ("The True Tragedy as a Yorkist Play? Problems in Textual Transmission", ShSurv 74[2021] 271-82). As an example of the octavo's pro-Yorkist sympathies, Stacey offers that "Richard of York’s assertion that the throne is 'mine inheritance as the kingdome is' (sig. A3v) is modified in the folio to 'as the earldom was' (1.1.78), placing his claim as the next step on the ladder of ambition rather than the assertion of a de jure fact" (p. 271). Stacey quotes here a few such variants where he thinks the octavo is obviously more pro-Yorkist, but I cannot really see what he thinks makes them so.
Stacey mentions the recent scholarship (including by the present reviewer) that suggests that Richard Duke of York and 3 Henry 6 derive from a lost play by Marlowe, Shakespeare, and others. This means that where we see revision making the play "more bipartisan" and less "pro-Yorkist" (p. 272) as we move from octavo to Folio, we now have to consider that ". . . this would be a revision of the original collaborative playtext, rather than the text which comprises the octavo version" (p. 273). That is, we now have to factor in revision of the lost original before it got into print and factor in disrupted transmission in its getting into print; the case is even more complex than we thought.
Stacey offers an appendix of 13 examples of Richard Duke of York being more pro-Yorkist than 3 Henry 6. It would have been helpful for Stacey to talk his reader through why he reads these variants as he does, since looking over them I cannot clearly see "Either a Softening of a Pro-Yorkist Position or a Strengthening of Henry's Claim in F" as the appendix's caption puts it. Stacey finds that some phrasings in the octavo and Folio that support his argument are ones that the Early English Books Online Text Creation Partnership (EEBO-TCP) dataset tells him are unique to those two editions, in the sense of being found in no other play from 1588 to 1600.
Stacey also finds phrases that are in other plays, for instance the octavo's "lawful heir" becoming "lawful chosen", in which "lawful heir" is also found in other history plays. For Stacey this suggests that the phrase "lawful heir" was "inserted into the play by actors who have recalled the phrase" (p. 274). I do not see the logic of this: if it was a fairly common phrase that suited a particular context that comes up in history plays (about legitimate inheritance) why do we need to suppose it was recalled by actors rather than simply being written by a playwright?
Likewise because the phrase "yield thy crown" in the octavo (where the Folio has "kneel for grace") is found in other plays including The Contention of York and Lancaster, Stacey argues that "The number of possible sources for O, in tandem with the unstable copy of the text of The Contention, dilutes the chances that the variant phrase in F, rather than O, could be the result of interpolation" (p. 275). Why does it? I am not seeing what light any of this throws on the relative likelihood of actors' interpolation versus authorial revision.
Stacey repeatedly asserts distinctions between phrases in the octavo and Folio that to his satisfaction establish different kinds of causes. For instance, he offers that ". . . O's 'Englands royall king' (sig. B6r) is diluted to the less assertive 'England’s royal throne'" in F, which is part of "a pattern of revision at work in F" (p. 275). I cannot see why he thinks the word 'throne' to be less assertive than 'king'. Indeed, Stacey's kind of literary criticism makes little sense to me; I cannot make anything out of the assertion that "Metaphoric spaces are opened up which add texture to the discourse of Lancastrian loss . . ." (p. 276).
Stacey writes that "The phrase 'Henries heires and your succession' (sig. B5r), spoken by Warwick to Edward in reference to the Yorkists as adopted successors, is changed in F to 'henry's oath and your succession' (2.1.119); an expectation of royal status is turned to a lawful pact, introducing a degree of fulfilled obligation -- and its obverse, jeopardy -- into the transference of rule" (pp. 276-277). This seems to me a clear example of over-reading. Warwick is here referring to King Henry's formal agreement, his oath, that upon his (King Henry's) death the crown will pass to the Yorkists, and since this oath is about Henry's heirs (because they will not inherit the throne) the octavo and Folio readings mean substantively the same thing.
Repeatedly Stacey describes textual variants in terms that prejudge how they came about, referring to lines in the Folio but not the octavo as "additions" and "interpolations" (p. 277) before establishing that this is what they are rather than being omissions from the octavo. He writes of certain examples that "As these interpolations simply do not occur in O, there is strong evidence of a sustained revisionary impulse in F" (p. 277). But this is illogical, since if the lines were present in the octavo they would not constitute variants let alone be interpolations in the Folio. That is, we cannot meaningfully refer to certain lines being interpolations that are absent from the octavo if their being absent from the octavo is what makes them interpolations in the Folio. Thus Stacey is begging the very questions he seeks to answer.
To account for the changes made to the play behind the octavo and Folio, Stacey reads the pro-Yorkist sympathies of the former as being shaped by anxieties in the 1590s over who would succeed Queen Elizabeth: the play advances the claim for a York-descended successor such a James 6 of Scotland "over a Spanish or Portuguese rival" (p. 280). Stacey suggests that the play's repeated references to molehills -- one where Richard Duke of York is mocked and killed and another where Henry 6 sits to reflect on his situation -- are allusions to the Mouldwarp (or Mole) Prophecy as applied to Henry 4 and later Henry 8.
Ros Barber is unconvinced that the Zeta method of authorship attribution has satisfactorily shown that each of the three Henry 6 plays has multiple authors, insisting that they each might in fact be sole-authored ("Big Data or Not Enough? Zeta Test Reliability and the Attribution of Henry VI", DSH 36[2021] 542-64). The Zeta method was invented by John Burrows and refined by him and Hugh Craig and their investigations are one of the several independent lines of research that led to the New Oxford Shakespeare putting Marlowe's name alongside Shakespeare's on all three Henry 6 plays' title pages
Barber starts to repeat Pervez Rizvi's critique of the method of interpreting Zeta results used by Burrows and Craig (which critique was reviewed in NYWES for 2019) using a 'bisector line' method, but because she has not explained that this refers to a line drawn on a scatterplot, her account will make no sense to readers who have not also read Rizvi's essay. The Zeta method uses two sets of training data: i) texts all by one candidate author, and ii) texts by another author or set of authors. Zeta determines which words are favoured in one set and disfavoured in the other. Burrows and Craig's method for representing their results is to put a dot on a scatterplot for each unit of text tested (typically a play, or a chunk of a play) and to label each dot by the name of its author. The x value of the dot represents how much it is like the style of the first set of texts (typically the candidate author) and unlike the second, and the y value of the dot represents how much it is like the style of the second set of texts (by one or more rival candidates) and unlike the first.
It is part of the set up of the test that we expect to produce a series of dots for one author having high x and low y values, and another series of dots for the other author(s) that have low x and high y values. That is, we get two distinct clouds of dots, one for each set of texts. One cloud is in the bottom right of the scatterplot (high x, low y) and the other in the top left of the scatterplot (low x, high y). We can make the same calculations for a text to be attributed and make a dot for it on the scatterplot and see which authorial cloud of dots it falls nearer to. Burrows and Craig used a dividing line between the two clouds that is simply a line drawn half-way along and perpendicular to a line that joins the centres of the clouds.
What Rizvi (and now Barber) rightly object is that this bisector line method takes no account of how tightly or loosely clustered the cloud of dots is, and that the calculation of the centre of the cloud can be pulled off in one or other direction by an outlying, anomalously large or small, single data point, or by a few of them. While true in principle, it is not clear that this is a problem in practice: Barber would need to show that outlying datapoints do distort the calculation of the centres of the clouds in Burrows and Craig's work. The solution, proposed by Barber (and proposed to her by Rizvi she says) is to calculate the standard deviation of the points in the cloud and require that the new point to be attributed falls within so many standard deviations from the cloud centre.
Barber says that her aim is to "examine the accuracy of Burrows and Craig’s claim that Zeta can tell Shakespeare from Marlowe with 99.9% accuracy" (p. 543), which is a claim they made for one particular validation experiment in an essay in the New Oxford Shakespeare Authorship Companion co-edited by the present reviewer Necessarily, requiring that each text (or chunk thereof) to be tested has to fall within so many standard deviations of the candidate author or the rival author(s) creates a "no man's land" (p. 545) between these zones into which a suspect text might fall. By contrast, Burrows and Craig's bisector method divides the entire space on the scatterplot into two zones so of necessity the test must give us a verdict in favour of one or other even if the difference being judged is small.
As Barber shows, declaring that some chunks of a work fall into no man's land can actually be the better result. In validation runs, Burrows and Craig's method wrongly attributed many chunks of The Jew of Malta to Shakespeare (simply because they do not fall on Marlowe's side of the bisecting line), whereas Barber's method has them fall into no man's land, meaning that they are like neither Marlowe's nor Shakespeare's style.
Barber's main aim is to critique Burrows and Craig's work attributing parts of the three Henry 6 plays to Marlowe. She notes that they were forced to use sample writings of each candidate that were not of the same size. What seems particularly objectionable to Barber is denial of Kyd's hand in 1 Henry 6. The problem of course is that we have so little writing that is securely attributed to Kyd: at best the agreed canon is The Spanish Tragedy, Solimon and Perseda, and his translation of someone else's play, Cornelia. Barber is right to say that large discrepancies in canon size affect authorship attribution reliability, but this has long been known and of course Craig and Burrows acknowledge this.
A related problem arises when the candidate authors' canons do not reflect all the genres. Shakespeare wrote comedies, histories, and tragedies, and according to one's classification preferences perhaps also tragicomedies and/or Romances. But Marlowe left us "a canon consisting almost entirely of histories and tragedies" (p. 549) as Barber rightly points out. To test whether a "small genre-canon like Marlowe's" (p. 550) really would affect Zeta's reliability, Barber set up a test using just four Shakespeare plays (Coriolanus, King Lear, Antony and Cleopatra, and Romeo and Juliet) versus 89 plays by others (not including Marlowe) and then checked whether the setup would correctly attribute Hamlet to Shakespeare. It did not, even when Barber then repeatedly cut down the size of the non-Shakespeare set.
Next Barber tried the same thing with Julius Caesar, and here it becomes apparent what part of the problem might be. In Barber's experiment, "'Mark' is a Shakespeare marker, being not only a verb and common noun but a major character's name, mentioned in both speech and stage directions, in Anthony and Cleopatra" (p. 550). But in fact, Craig removes proper nouns, speech prefixes, and stage directions from his plays before doing the testing: he tests dialogue only. Thus Barber's experiment is fundamentally unlike the method it is attempting to replicate, and this alone is enough to explain its failing.
When Barber repeats the experiments that Burrows and Craig used to attribute 3 Henry 6 partly to Marlowe (or rather, runs her version of the experiments), with the refinement of using a standard-deviation measurement so that text segments are allowed to fall into no man's land where the method gives no verdict on authorship, ". . . the vast majority of 3 Henry VI's segments fall in no man's land . . ." (p. 554). Looking at the particular marker words for Marlowe in Craig's experiments on 1 Henry 6, including 'arms', 'realm', and 'death', Barber reflects that these are simply bound to come up in the Joan of Arc scenes "given that Joan is at war with the English" (p. 555).
Attempting to rework Burrows and Craig's approach to 1 Henry 6 but using sets of plays that "have a high level of reliability for distinguishing the two authors for most (though not all) of their sole-attributed plays" (p. 557) and using the standard-deviation approach instead of the bisecting-line method, Barber found that Marlowe is attributed "a single segment early in Act 1 . . . the second half of Act 3, and early Act 4 . . . [and] almost all of Act 5" (p. 557).
This last is a remarkable result given that Barber and the rest of the team led by Brian Vickers (including Rizvi and Freebury-Jones) have been arguing for years that the New Oxford Shakespeare's claim of Marlowe's hand in 1 Henry 6 is mistaken. In describing this result Barber seems torn between celebrating the power of her refinement of Burrows and Craig's method and dismissing it in order to dismiss their findings too. Barber is silent on the fact that her results are similar to those produced by the Word Adjacency Network method and published by a team containing the present reviewer, mentioned briefly in NYWES for 2016.
Barber does the same test for 2 Henry 6 and again by the standard-deviation method a lot of segments fall in no man's land, but again she finds Marlowe's hand in the play and in the same places that the Word Adjacency Network method did, about which she says nothing. Considering why The Jew of Malta does not seem Marlovian by Burrows and Craig's tests and her own, Barber wonders if revision between its being written around 1589 and its being published in 1633 explains the problem.
In conclusion, Barber retreats from her findings about the three Henry 6 plays and remarks that ". . . any one of the plays might be written wholly by the same author" (p. 561) and ". . . any one of the Henry VI plays may or may not be co-authored" (p. 562). It is impossible to reconcile that conclusion with the results in the body of her essay detailed above.
On the same theme as Barber is Pervez Rizvi's attempt to show that all the authorship attribution scholarship in the New Oxford Shakespeare Authorship Companion should be disregarded because its Zeta and its Principal Component Analysis texts (PCA) are flawed ("Shakespeare and Principal Components Analysis", DSH 36[2021] 1030-41). Rizvi correctly explains what PCA is but objects to its being called a data-reduction technique since it "does not reduce our data. It manufactures new sets of data" (p. 1030-1031). Oddly, though, he cites as an authority on PCA an article by Ian T. Jolliffe and Jorge Cadima which repeatedly refers to PCA reducing dimensionality, which is the same thing as data reduction.
Rizvi explains that his chief complaint is that in their book Shakespeare, Computers, and the Mystery of Authorship (reviewed in YWES for 2009) Hugh Craig and Arthur F. Kinney did not specify in each of the PCA plots how much of the variance was captured by each principal component. Rizvi gives an account of Zeta that is slightly askew, claiming that "The number of occurrences of each chosen function word in each segment is counted . . ." (p. 1031). In fact what is counted is the proportion of the set of chosen function words that each segment contains at least one occurrence of, not the number of occurrences of the word in each segment. Rizvi here repeats the complaint made in his "The Interpretation of Zeta Test Results" (reviewed in NYWES for 2019) that using a bisector line on Zeta results scatterplots does not really add anything to them.
Because in their explanation of PCA in their book Craig and Kinney do not tell us how much of the variance is retained by the first and second components they use, Rizvi attempts to replicate their experiments on Henry 8 and The Two Noble Kinsmen but with quite a few variations. Rizvi uses different Fletcher plays, fewer function words, and employing his own versions of the play transcriptions, and looking at just the sole-authored plays of each man. Because in his resulting scatterplot "The dots are clearly in the same places relative to each other as in the two graphs in C[raig] & K[inney] . . ." (p. 1032), Rizvi is content that he has replicated their work.
In Rizvi's scatterplot, each dot stands for the data about one play, either by Shakespeare or Fletcher. In Craig and Kinney's scatterplots, this is also true but they overlay dots for certain scenes from Henry 8 and The Two Noble Kinsmen to show that the dots for these scenes fall within the clouds of dots for the man who wrote each of those scenes. To show who is the author of the text that each dot represents, Craig and Kinney use a different shape of dot for each author.
It is most peculiar, then, that Rizvi writes that "It is important to understand that they", meaning the dots in the scatterplot, "are so represented because that was a choice made by me, and by C&K, in drawing the graphs" whereas in a Zeta plot ". . . the presentation of the dots by different shapes is not at the researcher's discretion: it was determined at the start by his choice of the base and counter sets" (p. 1032). This is untrue. In both cases the investigator decides how to style the dots based on the same attribute of the text the dot represents: its author. That is, both methods are supervised in the sense that we decide who wrote each text that each dot represents and we instruct the software to label the dots accordingly.
To explain what he means by the labelling of the dots being deceptive, Rizvi proposes a bizarre thought experiment in which instead of labelling the dots to show the author of each text that each dot represents, we label them according to where they fall on the scatterplot. He rightly says that this would make for a meaningless scatterplot, but he gives no reason why anybody would produce such a thing in the first place. There are many illogical choices that one might make within any investigative technique, and just naming one of these illogical things does not itself reveal a weakness in that investigative technique.
Rizvi rightly observes that his scatterplot for Henry 8 and The Two Noble Kinsmen (and it is true of Craig and Kinney's too) separates Shakespeare and Fletcher along the first principal component but not along the second. Rizvi claims that this is "common" in the PCA tests in Craig and Kinney's book (in fact it is rare) and that ". . . at no point do they draw the reader's attention to this" (p. 1033).
Craig and Kinney do draw the reader's attention to this happening, as in their explanation of the PCA analysis of Edward 3:
The Shakespeare texts fall to the left of the plot, and occupy a smaller span along the vertical axis than the non-Shakespearean segments, which range higher in this dimension. . . . On the horizontal axis the Countess scenes (the black triangles) are well to the Shakespeare side, though they are beyond the northern edge of the Shakespeare cluster on the vertical axis. (Craig and Kinney p. 124)
As well as describing here the greater separation along one axis than the other, Craig and Kinney draw attention to additional information that is relevant even when complete separation along one axis is not observed: we can see the two authors' different ranges across that axis even though they overlap. This is true of the particular case Rizvi points to, which is the PCA analysis of Shakespeare and George Peele's Titus Andronicus (Craig and Kinney's p. 32).
Because in his 'replication' of Craig and Kinney's illustration of PCA using Henry 8 and The Two Noble Kinsmen the first principal component captured only 17.2% of the variation in the data, Rizvi assumes that is about how much of it was captured by the first principal component in Craig and Kinney's experiment. He considers 17.2% to be "far below an acceptable level" (p. 1033) but this is really just a complaint about stubborn reality. By its very nature the first principal component is the one that, of all the components calculated, captured the most variation in the data. Rizvi does not offer what he thinks an acceptably high level of variation for the first principal component to capture.
Rizvi believes that by accepting a first principal component that captured only 17.2% of the variation, Craig and Kinney accepted that "a significant amount of information was indeed discarded" (p. 1034) and hence their results are unreliable. But this is just a consequence of starting with as many dimensions as 59 -- that is, by counting the frequencies of 59 function words -- and reducing those dimensions to two. What is striking is that in such a drastic reduction in dimensionality as much as almost a fifth of the variation can be captured along just one principal component.
To try to show that an authorial attribution cannot be reliably made by Craig and Kinney's methods, Rizvi put segments from John Webster's The Duchess of Malfi into the scatterplot he made for Henry 8 and The Two Noble Kinsmen and finds that they fall into the Shakespeare part of the scatterplot, so that if The Duchess of Malfi were an anonymous play and we were trying to attribute it, this test would make it look like Shakespeare's work. But of course we do not do that with anonymous plays since it would be an obviously unsound way to attribute one unless we already had reason to suppose that the author of the anonymous The Duchess of Malfi was either Shakespeare or Fletcher and not someone else. An experiment specifically set up to tell apart the styles of Shakespeare and Fletcher is naturally unsuited for trying to distinguish either from the style of Webster.
Rizvi attempts a replication of the analysis of words spoken by characters written by Shakespeare and those written by Fletcher that Craig presented in his essay "Style, Statistics, and New Models of Authorship" (reviewed in YWES for 2010). Rizvi's PCA plot comes out as quite different from Craig's and he remarks "It is inconceivable that minor differences between his plays and mine could cause such a fundamental difference in outcome . . ." so he concludes that ". . . Craig inadvertently swapped the two axes on his graphs, or perhaps something went wrong with the software he used" (p. 1035).
Rizvi's main objection is that in this test (his version), the first principal component (abbreviated to PC1) does not separate the authors: only the second (PC2) does. (In Craig's version, PC1 separates the authors but PC2 does not). Again, Rizvi misses the point that even when one dimension does not entirely separate two authors (as Craig's PC2 does not) it contributes information about the range and spread of the data. For instance, in Craig's PCA, 12 Shakespeare characters have PC2 scores lower than the lowest Fletcher character's score, and five Fletcher characters have PC2 scores higher than the highest Shakespeare characters. Even with the overlapping PC2 scores for many of their characters, the Shakespeare and Fletcher characters have a meaningful difference along PC2 as well as a clear differentiation along PC1.
That is why the line bisecting the Shakespeare and Fletcher dot-clouds is not parallel to one of the axes: it runs at a diagonal taking in both axes and making a sharp distinction with just one Fletcher dot in the Shakespeare zone, and that is for a character from Fletcher's famously anomalous The Faithful Shepherdess. Students of machine learning will readily recognize Craig's procedure: he has shown that the Shakespeare and Fletcher sets are linearly separable in two dimensions. None of this is true of Rizvi's attempted replication, so we should not simply accept his explanation that Craig must have made an error that he avoided.
Next Rizvi adds dots to his PCA plot for the characters of Hamlet, Antony from Antony and Cleopatra, and Petruchio from Fletcher's The Tamer Tamed. Hamlet is way off to the left of the plot, having a far lower PC1 and a higher PC2 than any other character. Shakespeare's Antony and Fletcher's Petruchio have almost identical PC1 scores but highly unalike PC2 scores. Rizvi thinks that perhaps the test is distorted by size of a role, since Hamlet is an especially long one and some other long roles are nearby.
Rizvi could easily have measured this for all the characters and tested if there is a correlation between PC1 on this test and role length. Rizvi reiterates his objections to Craig's (and hence the New Oxford Shakespeare's) use of PCA, claiming again that both the two PCs need to show clear separation between authors if the test is to be trusted (so, again denying the value of linear separability in two dimensions).
Rizvi closes with a brief critique of the essay by Jack Elliott and Brett Greatley-Hirsch the New Oxford Shakespeare Authorship Companion and his key objection is right. Elliott and Greatley-Hirsch thought that their PCA analysis showed that the true author of Arden of Faversham must be amongst the nine authors they were considering because its segments appeared within the large cloud formed by the segments for works by those nine authors.
As Rizvi correctly objects, this is a logical error: any number of additional new authors might be added to the experiment and their dots might also fall within the cloud that already exists and hence form the space that encloses the dots for Arden of Faversham. Why might not one of these new authors be the author of Arden of Faversham? According to Rizvi, the way to avoid the problems he has identified in this essay is to go back to reliable old-fashioned methods of attribution using "rare n-grams and collocations" (p. 1039), manually reading the context of each occurrence and doing literary criticism to "check for similarity of thought" (p. 1039).
Mark C. Hulse attempts to show that Peele and not Shakespeare (as claimed by the New Oxford Shakespeare) wrote scene 4.1 of Titus Andronicus ("The Case for Peele's Authorship of Titus Andronicus 4.1: Cross-examining Attribution Methods in a Disputed Scene", RES 72[2021] 860-81). His approach is largely to critique the recent work by which William W. Weber and Anna Pruitt reassigned the scene from Peele to Shakespeare. The important question Hulse starts with is what to do about differing canon sizes when comparing counts of rare words in Shakespeare's works and in Peele's in order to attribute scenes.
In his pioneering work that inaugurated the searching for rare phrases in the two authors' canons, MacDonald P. Jackson arbitrarily cut Shakespeare's down to seven works to match Peele's, but admitted that even this artificially reduced canon gave Shakespeare an advantage since it was still bigger than Peele's actual canon. In replicating Jackson's work, Weber and Pruitt used an eight-work Shakespeare canon, making the problem of bias towards Shakespeare even worse.
By Hulse's calculations, Peele's canon contains about 89,000 words while Jackson's seven-work Shakespeare canon contains about 143,000 words (about 60% more), and Weber and Pruitt's eight-work Shakespeare canon contains about 165,000 words (about 85% more than Peele's). If we accept Hulse's counts, then this is indeed a problem. If we are looking for matches for any phrase in a Shakespeare dataset that is significantly bigger than our Peele dataset, we will find more matches simply because the body of Shakespeare writing we are looking in is bigger than the body of Peele writing we are looking in.
But is this as bad a problem as Hulse thinks? He writes:
The majority of lexical combinations are mundane and occur in both canons (n-grams like 'I know not', 'see how', and so on), meaning they cancel out and benefit neither candidate. Imagine that an unattributed passage produced 60 n-gram correspondences with Peele's oeuvre, but 80 with a longer subset of works granted to Shakespeare. The majority would likely cancel out when we adopted the uniqueness requirement. If 50 were common to both authors, there would remain 30 unique hits for Shakespeare against only 10 for Peele--a perceived three to one advantage. Thus, even a modest majority in total collocations or n-grams (caused by improperly balanced canons) can snowball into a supermajority when foregrounding unique hits, and this amplification effect is never identified by Weber or Pruitt as an alternative explanation for their data. (p. 864)
This sounds convincing but is illogical since the snowballing effect is imaginary. In this hypothetical situation there is no baseline of "50" matches . . . "common to both authors". Hulse just made this number up and even if it were the result of an actual experiment it would have no significance.
It is true that commonplace (what Hulse calls "mundane") phrases in the Titus Andronicus 4.1 would most likely also be found in the artificially constrained Shakespeare canon and the real Peele canon, but we do not know how many there are until we count them. Because these phrases are commonplace, they tell us nothing about authorship since anyone might use them. There is no significance to the ratios of the counts of these commonplace phrases to the counts of rare phrases -- 50:30 for Shakespeare and 50:10 in Peele, in Hulse's example -- no matter whether Hulse's "50" is a reasonable or a wildly inaccurate guess for the commonplace phrases.
Comparing the meaningful data about rare matches -- 30 rare hits for Shakespeare, 10 rare hits for Peele -- to the meaningless data about common matches (50 for both authors) reveals nothing about the significance of the meaningful data. Objectively, if there are three times as many rare-phrase matches between Titus Andronicus 4.1 and Shakespeare as there are rare-phrase matches to Peele, that is information about authorship. It matters whether the investigators are using an artificial Shakespeare canon that is bigger than the Peele canon, but it would have to be three times bigger before we would say that the discovered difference is only to be expected.
Thus Hulse is right to object to an "advantage stemming from canon size disparity" but wrong to refer to "the distorting effect of emphasizing only unique correspondences" (p. 864). The "snowball" effect emerges only from his starting point of considering commonplace phrases, which are not meaningful. Hulse also objects that the Shakespeare canon was inflated by Weber and Pruitt allowing matches to either of the alternative early editions of some plays, such as the 1597 and 1599 quartos of Romeo and Juliet: because these two editions have variant readings, the 'surface area' for Shakespeare is again larger than it should be. Because of these distortions Hulse declares that "fresh analyses are needed" (p. 866).
Hulse complains that Weber and Pruitt did not include in Peele's canon the play The Troublesome Reign of King John, which Charles R. Forker's 2011 Revels Plays edition attributed to him in 2011, before they published their findings. But Martin Wiggins's Catalogue published in 2012 wrote of the play that it "has been ascribed to George Peele" (whereas for cases he accepts he simply identifies an author), and Alan Farmer and Zachary Lesser's Database of Early English Playbooks (which keeps up with new scholarship) still, as of April 2025, lists the play as "Anonymous (?)".
So it was reasonable in 2014 and 2017 for Weber and Pruitt to leave this play out of the Peele canon. Inclusion of The Troublesome Reign of King John as Peele's would likely bias the results of this particular authorship investigation (a Peele-versus-Shakespeare case) since there is an as yet undetermined relationship between this play and Shakespeare's King John, possibly arising from Shakespeare adapting Peele's play to make his own. This is just the sort of unresolved textual relationship that authorship-attribution scholars seek to avoid when choosing the contents of their datasets.
Hulse revisits Weber's claims about epizeuxis in Titus Andronicus 4.1 in the light of a new Peele canon including The Troublesome Reign of King John and a new restricted Shakespeare canon of just The Comedy of Errors, Richard 2, Richard 3, The Taming of the Shrew, The Two Gentlemen of Verona, and Venus and Adonis, which is now about the same size as the Peele canon. Hulse reports that with these adjusted canons Shakespeare's use of epizeuxis is much less frequent but he does not offer totals for either candidate so the point has to be taken on trust.
Hulse includes Richard 2 in his restricted Shakespeare canon, and it famously has dozens of examples of epizeuxis including "tut, tut", "why, why", "well, well", "arm, arm", "see, see", "alack, alack", "little, little", "down, down", "thou, thou", "repealed, repealed", "that, that", "rest, rest", "pardon, pardon", "help, help", and "mount, mount", and many more, and so does Richard 3, including "see, see", "blush, blush", "well, well", "come, come", "you, you", "alas, alas", "it, it", "to, to", "do, do", "pardon, pardon", "boy, boy", "both, both", "monstrous, monstrous", "Pomfret, Pomfret", "after, after", "tut, tut", "perhaps, perhaps", "thus, thus", "delay, delay", "myself, myself", "this, this", "awake, awake", "perjury, perjury", "bustle, bustle", and "rescue, rescue". It would have been interesting to see the examples from Peele that Hulse thinks outnumber Shakespeare's uses of this device, but he confines himself to asserting that "the epizeuxis device is scarce in the early [Shakespeare] canon and in other suspected Shakespearean writings from this phase" (p. 868).
Hulse turns to some rare words that are in Titus Andronicus 4.1 but not the restricted Shakespeare canon, including 'fere' (found only in Titus Andronicus 4.1, Wilkins's portion of Pericles and Shakespeare's part of The Two Noble Kinsmen) and 'sware' (found only in Titus Andronicus 4.1 and 2 Henry 4), but he does not mention how often nor where he thinks they appear in Peele's works, so we cannot evaluate this finding. Hulse provides a table in which he summarizes the results of repeating Weber's investigation but using his newly restricted Shakespeare canon and adding The Troublesome Reign of King John to Peele's canon, and he says that is shows that "Far from finding twice as many Shakespeare-only words [in Titus Andronicus 4.1], Peele wins comfortably . . ." (p. 869).
Hulse's precise numbers are 18 words from Titus Andronicus 4.1 found in Peele's canon but not Shakespeare's and 9 words from Titus Andronicus 4.1 found only in Shakespeare's canon but not Peele's. This is largely the consequence of adding The Troublesome Reign of King John to Peele's canon, since this turns many Shakespeare-only words into Shakespeare-and-Peele words, so they drop out of consideration. Six of his 18 Peele-only words also occur in the Shakespeare sections of Titus Andronicus but they do not count as Shakespeare words because the Shakespeare sections of Titus Andronicus are not in Hulse's restricted Shakespeare canon.
Hulse does not accept Weber's argument that Titus Andronicus 4.1 shows a fluency with classical-myth allusion that we find elsewhere in Shakespeare but not in Peele: he thinks we find it in Peele too. Hulse claims that "Although Shakespeare does refer to the Trojan queen [Hecuba] several times in the canon starting around 1600, only one late play (Cymbeline) speaks of her madness" (p. 873) as is mentioned in Titus Andronicus 4.1 ("ran mad for sorrow").
But surely the following lines in Hamlet refer to the madness of Hecuba: "[Who had seen her] Run barefoot up and down, threat'ning the flames | With bisson rheum, a clout upon that head | Where late the diadem stood, and, for a robe, | About her lank and all o’erteemèd loins | A blanket, in th'alarm of fear caught up | . . . | The instant burst of clamour that she made | . . . | Would have made milch the burning eyes of heaven". This sounds to me like a woman running mad.
Hulse wonders why in Titus Andronicus 4.1 Marcus calls Young Lucius "Hector's hope" since Hector and his son end badly in the story of Troy, and he explains it via a line in The Troublesome Reign of King John in which the French king invokes the myth that Hector's son got away from Troy and founded a French dynasty. (But this would still not make the line yield good sense in Titus Andronicus, which does not invoke this myth.)
Lastly, Hulse turns to the allusion to Sybil's leaves bearing prophecies and being blown apart by the wind in Titus Andronicus 4.1 and an allusion to the same in Peele's The Arraignment of Paris, which also has a wronged woman (Oenone, abandoned wife of Paris) and her written complaint. Overall, Hulse considers the abundance of classical mythological allusion in Titus Andronicus 4.1 to be more typical of Peele than Shakespeare.
Hulse turns to rates of feminine endings, which are found in 1.5% of the lines in Titus Andronicus 4.1. Hulse does not actually state the rate in Peele's works, but Philip W. Timberlake in The Feminine Ending in English Blank Verse has it as "between 0.0 and 5.4 per cent in all the work which can be definitely assigned to him" (Timberlake's p. 24), which overlaps with Shakespeare's range of 4.3% to 16.8% (Timberlake's p. 117). But these are averages for whole plays, and when he counted only in long scenes Timberlake found Peele's range to be "0.0 to 2.4" and Shakespeare's "0.0 to 22.1" (Timberlake's p. 117). Since both writers wrote long scenes with no feminine endings, this feature cannot be used to attribute Titus Andronicus 4.1. Hulse concludes that rare-phrase matching is not an especially good method for authorship attribution and promotes the desirability of "integrating our quantitative and qualitative processes" (p. 880).
Thomas Merriam has new light to throw on the relationship between the anonymously published The Troublesome Reign of King John and King John in the 1623 Shakespeare Folio ("For John or for Arthur in King John", ANQ: A Quarterly Journal of Short Articles, Notes and Reviews 34[2021] 274-81). His conclusion is that the latter was written by Shakespeare and someone else, and that this someone else also wrote the former. Merriam describes some of the contradictions in the way that King John is presented in Folio King John, including his rejection of Rome (which would gain the sympathy of Shakespeare's audience) and his suborning of Hubert to murder Prince Arthur (which would lose it).
Merriam breaks Folio King John's dialogue into 204 one-hundred-word segments (the last being a little short of 100 words) and identifies four runs of segments that are pro-John (segments 180-92 depicting the defying of Pandolf) or pro-Arthur (99-103 the suborning of Hubert, 119-129 Hubert threatening Arthur, and 197-200 showing John's death). Merriam used the software package R Stylo to find occurrences of the top 1000 most frequent character 5-grams in each of the 100-word sections and then used PCA to reduce the dimensions for these counts from 1000 to two (PC1 and PC2).
This analysis gives a very strong division along PC1: all the pro-Arthur segments are to the left and all the pro-John segments are to the right. Merriam reckons this difference is due to these two parts of the play having different authors. Merriam also puts into the plot three points representing the same PCs of counts for the same character 5-grams in three chunks of the play The Troublesome Reign of King John, being i) the bit where John puts Hubert in charge of Prince Arthur, ii) the bit where Arthur confronts Hubert, and iii) the bit where John dies.
If Merriam's analysis were only really distinguishing themes of King John then these three bits of The Troublesome Reign of King John ought to be with the pro-Arthur segments on the scatterplot (since they are three pro-Arthur bits of The Troublesome Reign), but instead they sit with the pro-John segments of King John. Why? Because The Troublesome Reign was written by someone other than Shakespeare and so were the pro-John bits of King John.
In recent years, debates about the extent of the Kyd canon and its connexions with the Shakespeare canon have taken up much of this review. This year, just two articles need be noticed. In the first, Brian Vickers argues that Gary Taylor is wrong to think that Thomas Watson wrote Arden of Faversham since in fact Kyd did ("Authorship Candidates for Arden of Faversham: Kyd, Shakespeare, and Thomas Watson", SP 118[2021] 308-41). Vickers summarizes Taylor's two publications on Thomas Watson being the author of Arden of Faversham outside of Scenes 4-8 (Shakespeare's scenes), and then gives a brief history of claims that Kyd wrote the play.
Vickers gives a potted biography of Watson and his writings, pointing to the uncertainty about whether he was ever a playwright in addition to being a poet and scholar. He then describes the one quasi-dramatic writing we know that Watson did: a lavish entertainment for Queen Elizabeth when she visited the Earl of Hertford's home at Elvetham. Vickers indicates how unlike a commercial stage play this was, and how the apparent verbal parallels and collocations with Arden of Faversham -- and in particular the 'Hours' (hora) being guardians of heaven's gate -- are weaker than they seem because these parallels and collocations were commonplace.
Vickers thinks it merely coincidence that Franklin's epilogue to Arden of Faversham has 18 lines and ends with a couplet and that Watson wrote 18-line poems ending with couplets, whereas Taylor made much of this fact. Vickers quotes an argument from Freebury-Jones about the authorship of Arden of Faversham -- noting that Franklin's epilogue provides a list of the dead as do Kyd's epilogues to The Spanish Tragedy and Solimon and Perseda -- that is cited as forthcoming in Early Modern Literary Studies (EMLS). But today, four years after the publication of the article being reviewed, the journal EMLS seems defunct, having published no issues since 2019.
Vickers points out that other poets also wrote 18-line poems: it was not peculiar to Watson and the epilogue of Arden of Faversham. Taylor claimed that Arden of Faversham is unique among plays of the 1580s and 1590s in having stage directions beginning with "Here" (which also occurs in Watson's play-like writings) but Vickers rightly says that this occurs in other plays too, including Fair Em, The Pedlar's Prophecy, and The True Tragedy of Richard III. Taylor described Arden of Faversham as the first city comedy and Vickers remarks that "there are hardly any opportunities for laughter in this domestic tragedy" (p. 336) This is not my experience of seeing the play performed: it can be riotously funny.
Vickers does not accept that Arden of Faversham is like Watson's poetry in its special attention given to women. Vickers ends with a literary-critical case for Kyd in fact being the author of Arden of Faversham, pointing out that in The Spanish Tragedy, Solimon and Perseda, and King Leir, Kyd had created three vengeful women (Bel-Imperia, Perseda, and Ragan respectively) and that Alice Arden was the fourth.
In the second article on Kyd this year, MacDonald P. Jackson argues that the so-called 'expanded' Kyd canon promoted by Vickers most likely contains plays that are not Kyd's ("The Use of N-grams to Determine the Dramatic Canon of Thomas Kyd", Medieval and Renaissance Drama in England 34[2021] 126-54). The widely accepted Kyd canon is The Spanish Tragedy, Solimon and Perseda, and Cornelia (a translation of a French play). Jackson starts by summarizing Vickers's claims for expanding Kyd's canon to include Arden of Faversham, King Leir, Fair Em, part of 1 Henry 6, and part of Edward 3. Jackson reports that he has the data for Vickers's original n-gram claims that began this expansion, as announced by Vickers in the Times Literary Supplement (in an article reviewed in YWES for 2008).
Jackson finds that the unique n-grams shared between any pair of the three plays that we all agree are Kyd's -- The Spanish Tragedy, Solimon and Perseda, and Cornelia -- are more numerous than the shared matches for any pair that includes one of the plays that Vickers newly attributes to Kyd. This is most unlikely -- it would happen by chance 1 time in 500 -- if all the plays are equally by Kyd. Jackson does not show his working out for this, but since he refers to samples from "the same population" (p. 127) I suspect that he is referring to a test leading to a p-value of <0.002. It must be remembered that this is not the probability that the same-author hypothesis is true -- that is, that Kyd authored one of the 'expanded' canon plays -- but rather the probability of getting the result Jackson did when the same-author hypothesis is true.
Vickers has claimed that where we find his allegedly Kydian rare phrases popping up in Shakespeare, the explanation is that Shakespeare echoed Kyd, but Jackson points out that Kyd's influence would have to be so strong for this to explain all the occurrences in Shakespeare that this "proposal sabotages the very notion of using unique matches as evidence of authorship" (p. 128). Jackson reviews Vickers's subsequent publications on Kyd's authorship, observing that Vickers no longer confines the matches he cites to those phrases that are rare: he accepts commonplace expressions as having as much evidentiary value as rare ones.
Then Jackson turns to postings on Martin Mueller's blog about the enlarged Kyd canon, tested using discriminant analysis, critiquing Mueller's methods and noting that his approach finds that Solimon and Perseda, Cornelia, and The Spanish Tragedy are less similar to the new extended Kyd canon than the new plays themselves are. That is, we could just as easily carve out a new authorial canon for the new plays as their own set, by someone other than Kyd. Jackson points out that Mueller did not do the work of turning his investigation into a peer-reviewed article so there are many unanswered questions about exactly what his method was.
Jackson describes Rizvi's Collocations and N-Grams dataset and is largely approving of his methods, finding just a couple of flaws, including Rizvi's retention of stage directions and paratextual materials that can lead to false matches between plays that do not have dialogue phrases in common. Jackson details some of Rizvi's experiments, including ones that attribute Kyd as the author of Marlowe's The Jew of Malta and of Shakespeare's The Two Gentlemen of Verona and Marlowe as the author of Shakespeare's The Taming of the Shrew. Jackson speculates that what Rizvi's method measures is as much chronology as authorship, which would explain its misattributing early Shakespeare, since Marlowe wrote nothing after 1592 and Kyd nothing after 1593.
Jackson devises some new tests using Rizvi's dataset to find and count unique trigrams and these show that the three accepted Kyd plays -- The Spanish Tragedy, Solimon and Perseda, and Cornelia -- are like one another but the plays that Vickers wants to add to the Kyd canon are not like those three, except that the part of Arden of Faversham outside of Scenes 4-8 tests like Solimon and Perseda. Jackson picks through his findings and explains that some matches are due to conscious parody of or allusion to one play in another and others are due to closeness in content, such as King Leir with King Lear and within the multiple plays set in the Turkish world.
Jackson creates a further experiment in which he invents some "factitious canons" (p. 139) out of plays for which we do not know the author and compares whether they are less like the agreed Kyd canon than are the Putative Kyd Canon plays (that is, the ones that Vickers wants to attribute to Kyd). The answer is that the Putative Kyd Canon is no more like the agreed Kyd canon than the factitious canons are, and indeed the actual Marlowe canon is more like the agreed canon Kyd than either.
Jackson then slices the data another way and shows how like the rest of the Putative Kyd set each play in that set is, and how like it is to the agreed Kyd canon, the Peele canon, the Greene canon, and the Marlowe canon (the last divided in two to be about the same size as the others) and the two factitious canons. The upshot is that no clear pattern of likeness joins any of Vickers's Putative Kyd plays with the agreed Kyd canon. The only exception is the part of Arden of Faversham outside of Scenes 4-8, which part does test somewhat like agreed Kyd. Jackson weighs this result against the substantial work by other investigators, which he summarizes, that point away from Kyd having a hand in Arden of Faversham.
In our last article this year on authorship attribution, Hartmut Ilsemann continues his argument (noticed in recent NYWES reviews) that most of what we think Marlowe wrote was written by somebody else ("The Marlowe Corpus Revisited", DSH 36[2021] 333-60). Ilsemann describes some of his exploratory investigations using the tests provided in the R Stylo software package such as Rolling Delta, and reports on findings that cannot be right, such as ". . . there were only just a few plays up to 1600 in which he [Shakespeare] had not participated" (pp. 334-335). This ought to ring alarm bells that perhaps Ilsemann's experiments were not working the ways he thinks they are.
When Ilsemann starts to describe his experiment on Marlowe's 1 Tamburlaine (p. 335), not enough detail is given and his account of the data in his Table 1 defeated this reviewer's attempt to make sense of it. Even such simple matters as how to read numbers in the table are not straightforward. There are dozens of numbers of the kind "19,7" and "18,1". Is that a total of four numbers (19 and 7 and 18 and 1) or is Ilsemann using the European convention, unfamiliar in the Anglophone world, of representing a decimal point with a comma, so that these are simply the two numbers 19.7 and 18.1?
Ilsemann writes that "column A contains the plays" (p. 335) but in fact column A of Table 1 contains a heading of "words" in cell A1 and then a list of numbers from 0 in cell A2 to 17000 cell A70. When Ilsemann writes "In each column (except A of course), the three-lowest deltas are then highlighted and those plays and windows that do not have a single-lowest delta are eliminated" (p. 335). I do not understand this. It is possible that Ilsemann is here describing not his Table 1 but the spreadsheet from which he made Table 1. If so, it is unclear why he bothers to talk his reader through a spreadsheet she cannot see and does not talk his reader through the actual data she can see.
Ilsemann's conclusion is that by his experiment 1 Tamburlaine is most like 2 Tamburlaine and like Kyd's Cornelia, Peele's The Battle of Alcazar, and the anonymous Locrine, but not like the plays we traditionally attribute to Marlowe. Among the troubling aspects of this experiment is that Ilsemann does not list the Marlowe plays he tested nor what features of them he was testing. Remarks earlier in the paper imply that he looked for most-frequent character trigrams, but this is not stated definitively. As Ilsemann reports, his previous investigations (reviewed in NYWES for 2018) found that Locrine was written by Marlowe.
Ilsemann repeats his experiment but this time using 2 Tamburlaine, and it tests as most like 1 Tamburlaine and Greene's Friar Bacon and Friar Bungay and Peele's The Battle of Alcazar. To explain this, Ilsemann tests The Battle of Alcazar itself and finds that it was written by Marlowe not Peele. Likewise Peele's David and Bethsabe is in fact, according to Ilsemann, a Marlowe play. Ilsemann continues with experiments using R Stylo but the descriptions are scant, with mere hints such as ". . . a large number of single-authored and well-attributed reference files were used . . ." (p. 339).
The captions to tables are equally opaque: Table 5's is just "Principal component analysis of corpus" with no mention of what the experiment even counted. The body text says "The classifiers nsc (nearest shrunken centroid), svm (support vector machine), and delta in its classic Burrowsian flavour then tested the reference corpus with words, character bigrams (mf2c) and character trigrams (mf3c) following the pattern of Table 5" (p. 340). I do not know what "following the pattern of" means: is Table 5 the result of one experiment or some kind of model for another?
Of what I can gather about Ilsemann's experimental method -- and I am not sure I have properly understood it -- it appears that he starts from the assumption that Tamburlaine Parts One and Two are Marlowe's and then compares the other Marlowe plays to these two and, finding that they are not alike, treats this as evidence that plays such as Edward 2, The Jew of Malta, and Doctor Faustus are not Marlowe's.
External evidence such as Marlowe's name on play title pages does not seem to influence Ilsemann, and in this regard his approach to Marlowe's canon is like the anti-Stratfordian approach to Shakespeare's. That is, unreliable internal evidence is allowed to trump reliable external evidence. It is worth remembering that false attributions on title pages are actually rather rare, as Luke Erne showed in Chapter Two of his book Shakespeare and the Book Trade (reviewed in YWES for 2013).
Jennifer Young argues that the second quarto (Q2) of Othello, published in 1630, was made by conflating the texts of the first quarto (Q1), published in 1622, and the 1623 Folio (F), in a conscious attempt to appeal to a readership of senior lawyers that its publisher, Richard Hawkins, specifically catered for ("Shakespeare for the 'Triers': Richard Hawkins and Q2 Othello at the Serjeants' Inn", Journal for Early Modern Cultural Studies 21.i[2021] 94-119). The Stationers' Register records transfer of the right to publish Othello from Thomas Walkley to Richard Hawkins on 1 March 1628. Q2 was Hawkins's first Shakespeare play; later he was a co-publisher of the Second Folio.
Young explains where Hawkins's shop was, adjoining the gate to the Serjeants' Inn where resided the most senior lawyers. Hawkins published books that were particularly of interest to this high-class readership, with paratexts that made this clear. The copy for Q2 was an exemplar of Q1 marked up with readings from F supplemented by slips of paper holding Folio-only passages (p. 95).
Young considers some readings where Q2 follows F rather than its main copy of Q1, and finds that the choice is for the "harder or more imaginative readings" (p. 108). Young tries to find reasons for groups of choices regarding variants, such as "interest in the rhetorical balance of parallels and various forms of repetition" (p. 109) but the examples do not really bear this out. Rather, it seems to me that in general the Folio was understood to be the better text so its readings were copied into an exemplar of Q1 because that could more conveniently be used in the printshop than the larger, expensive Folio.
At one point in Young's argument, the variant under consideration is this:
Is not to leave undone, but keepe unknowne (Q1)
Is not to leave’t undone, but kept unknowne (F)
Is not to leave’t undone, but keepe’t unknowne (Q2)
Young writes that ". . . combining the 'leave't' of F1 with Q1's 'keepe' creates previously non-existent assonance within the adjoining phrases" (p. 110). But there might not be any combining going on here. Since 'keep' could be spelt 'kep' in Shakespeare's time, F's 'kept' and Q2's 'keepe't' are substantively the same: they differ merely in spelling. (If instead we read 'kept' as the past participle of 'to keep' then the line has an awkward shift in tense.)
A similar case is this:
it is not very well, by this hand, I say tis very scurvy (Q1)
nor tis not very well. Nay I think it is scurvy (F)
nor t'is not very well; I say t'is very scurvy (Q2)
Young remarks: "Combining the single use of 'tis' from both Q1 and F1 creates a new parallel phrasing of repetition" (p. 110). But Q2's reading can be explained by its Q1 exemplar being annotated this way to reflect F's readings:
nor it is not very well,
by this hand,I say tis very scurvy
That is, the annotator introduced 'nor' and deleted 'by this hand' in order to make his Q1 copy reflect F's readings. He might also have rewritten Q1's 'it is' as 't'is' but this could just have easily been Q2's compositor's doing.
Young seems to attribute more artistic agency to Q2's readings than we have warrant for. Discussing Q2's "approach to the vocative case" (p. 110), Young again overstates the complexity of the matter: the evidence she adduces can be accounted for simply by an exemplar of Q1 being annotated to make it have F's readings. And rather than account for this as Young does via a complex set of reasons regarding the kind of people who bought Hawkins's book -- using speculations that may be correct -- we need no explanation more complicated than Hawkins considering the Folio to be the better text and therefore wanting to bring his Q1 copy in line with it.
Because her topic is one of means by which we distinguish Shakespeare's style from Fletcher's in their collaborative plays, an article by Marina Tarlinskaja falls within the scope of this review ("Fletcher's Versification", Medieval and Renaissance Drama in England 32[2021] 102-25). In reviewing Tarlinskaja's work we have complained before (for instance in YWES for 2020) that what she means by the typographical variations she employs is not obvious and is not explained by her. In this article she again combines the use of capital letters, boldface, and underlining in examples such as ". . . my boy, my SWEET boy" without explaining what each bit of styling means. For this reason alone her examples are largely incomprehensible to this reader.
We have also complained before that Tarlinskaja does not fully explain her rules for whether monosyllabic words are to be stressed or not. Here Tarlinskaja begins to give the detail of her system, where W means the odd-numbered syllables of an iambic pentameter line (the first, third, fifth, seventh, and ninth syllables) and "S" means the even-numbered syllables (the second, fourth, sixth, eighth, and tenth). Tarlinskaja explains:
I divide monosyllables into three stressing categories: those that are always unstressed, both on S and W (articles, prepositions, conjunctions), always stressed (nouns, lexical verbs, adjectives, adverbs), and ambivalent: usually stressed on S and always unstressed on W (personal, demonstrative, possessive and relative pronouns; relative adverbs). (p. 103)
This is helpful, as far as it goes. Tarlinskaja gives a few illustrative examples, but not nearly enough to exhaust the above categories, nor to show how she decides the "ambivalent" cases. Where Tarlinskaja cites authorities for her rules, the key ones are in Russian: the entries for M. L. Gasparov, E. V. Kazartsev, A. N. Kolmogorov, and V. M. Zhirmunsky in her bibliography.
Also hard on the non-specialist is a baffling first footnote in which Tarlinskaja explains that "Iambic pentameter is a string of ten syllabic positions: weak, W, usually unstressed, and strong, S, usually stressed syllabic positions of the metrical line: W S W S W S W S W S W S W S W S W S W S (W W)" (p. 124n1). I count this as 20 syllables (10 "W" and 10 "S"), not ten, plus the last two optional ones. The editors of the journal ought at least to have spotted the absurdity of this explanation and asked Tarlinskaja to fix it. This review will proceed no further because the above uncertainties about the significance of typographic conventions and the rules for stressing monosyllabic words mean that I cannot make sense of what Tarlinskaja is claiming.
Pervez Rizvi proposes an emendation to fix a puzzling line in the 1623 Folio ("An Emendation of 'By Louing Likelyhood' in Henry V", ANQ: A Quarterly Journal of Short Articles, Notes, and Reviews 34[2021] 199-201). In Folio Henry 5, the Chorus compares the welcome Henry gets in London after his victory in France to the welcome that the Earl of Essex would get coming back from Ireland:
As by a lower, but by louing likelyhood,
Were now the Generall of our gracious Empresse,
As in good time he may, from Ireland comming,
. . .
How many would the peacefull Citie quit,
To welcome him?
The problem is "by louing likelyhood". Some editors try to justify retaining this reading and others have attempted emendation; neither produces a convincing result. Rizvi proposes the emendation to "billowing likelyhood". But how plausible is it that Shakespeare wrote only one '-l-' in 'billowing' where we usually write '-ll-', as this explanation of the error requires?
Rizvi searched Shakespearian good quartos for this habit in lines that are not full -- since in full lines a compositor might drop a letter to save space -- and found three: 'cals' for 'calls' in the first quarto of 1 Henry 4, 'reueling' for 'revelling' in the first quarto of A Midsummer Night's Dream, and 'tels' for 'tells' in the quarto of Much Ado About Nothing. Then he looked in the same corpus for an unusual space within what we think of as one word, and he found three: 'a shamed' in the quarto of Love's Labour's Lost, 'any where' in the second quarto of Romeo and Juliet, and 'In sisture' in the quarto of Troilus and Cressida.
So, according to Rizvi, it is not inherently unlikely that the manuscript underlying Folio Henry 5 had 'bi lowing' for 'billowing'. We know that minim misreadings caused certain errors in the Shakespearian good quartos, including 'w' being misread as 'u'. This could cause 'lowing' to be misread as 'louing' and from there the compositor simply regularized 'bi' to 'by' to give us the Folio reading 'by louing'.
The word 'billowing' occurs elsewhere in Henry 5 but not in any other Shakespeare play, but as Rizvi rightly points out the same is true for a lot of words that Shakespeare uses. He points to the weakness of other emendations, but does not quite offer a compelling explanation of what a billowing likelihood actually is. That is, what is the adjective telling us about the noun?
An article by Petr Plecháč that is relevant to this review is in fact just the same as the one with the same title that was reviewed in NYWES for 2019 when it appeared in the pre-print repository called arXiv ("Relative Contributions of Shakespeare and Fletcher in Henry VIII: An Analysis Based on Most Frequent Words and Most Frequent Rhythmic Patterns", DSH 36[2021] 430-8). The remaining articles this year are concerned with the history of Shakespeare editions and their wider cultural effects. We will take them in chronological order of the editions or, where this is the point, their effects.
Mathieu D. S. Bouchard argues that the differences between the 1714 Jacob Tonson complete works of Shakespeare and its copy text (the second of the 1709 Nicholas Rowe editions) are greater than we have appreciated and they are the work of John Hughes, who adopted readings from an exemplar of the Second Folio ("A Revised Account of the 1714 Works of Mr. William Shakespear", PBSA 115[2021] 419-61). Bouchard starts with a brief history of the publishing of Rowe's 1709 octavo edition of Shakespeare, the first to be in multiple volumes and the first to be edited. A second edition, a line-by-line reprint of the first, appeared in late 1709 or early 1710, and in 1714 a third edition appeared, in duodecimo.
The textual differences between this 1714 edition and the 1709 editions have long been dismissed as insignificant, but Bouchard shows that in fact an exemplar of the second edition of 1709 was copy for the 1714 edition but it was first collated with a Folio, probably the Second Folio, and that the person doing this collation had different ideas about meter and punctuation from Rowe. A manuscript recording payment by Tonson to John Hughes suggests that Hughes was the one who did the editing needed for the 1714 edition (although Rowe's was still the only name on the title page), but it seems to be for rather too much money: 28 pounds 7 shillings, compared to the 36 pounds 10 shillings that Rowe got for creating the whole thing.
Bouchard shows that more work than we used to think went into the 1714 edition, which might explain the large payment to Hughes. He provides a full collation of how the 1714 edition departs from its 1709 second edition copy regarding the first play, The Tempest, and talks the reader through the various choices in the 1714 edition that make collation of the copy against the Second Folio more likely than collation against any other prior edition. Adding to the likelihood of this is the demonstration by Peter Holland (in an article reviewed in YWES for 2000) that a trial sheet of the edition, set and impressed in 1708, had the Second Folio as its copy text. Most likely Tonson made the same exemplar of the Second Folio available to Hughes for the 1714 edition.
Overall, "It is difficult to know what precise editorial strategy guided the various reversions to the earlier folio readings in The Tempest. There does not appear to be any single, cohesive logic to the changes" (p. 433). But smoothing the meter does seem a recurrent concern. Bouchard is somewhat shaky on meter, though. He considers the line "But that the Sea, mounting to the Welkins Cheek" and remarks that ". . . the word 'cheek' falls on the eleventh beat, violating a strict adherence to iambic pentameter" (p. 434). Of course, in this line 'cheek' is the eleventh syllable not the eleventh beat, and the line is a regular iambic hexameter with the permissible omission of an unstressed syllable at the caesura where the comma is.
Bouchard detects in changes of punctuation introduced in the 1714 edition the hand of someone "who recognized the value of Rowe’s pointing but saw the potential for improvement" (p. 438). Bouchard ends with biographical information about how the work on the 1714 edition would have fitted into Hughes's life and career. Ghost writing and ghost editing seem to have been Hughes's specialisms. In his 1709 edition Rowe somewhat slighted Shakespeare's non-dramatic works, and these remarks being reprinted for the 1714 edition stand somewhat at odds with that edition's inclusion of the narrative poems and sonnets as a seventh volume. This is further reason to suppose that Rowe did not edit the 1714 edition, along with the fact that the poetry volume was edited and contained essays by Charles Gildon, whom Rowe detested.
Bouchard sees Hughes working in the kind of anonymous-corrector capacity that Sonia Massai argued (in a book reviewed in YWES for 2007) was commonly employed in the making of seventeenth-century reprints of Shakespeare. Thus although Rowe's 1709 edition inaugurated a new kind of editorship, Hughes's 1714 edition looked back at the old kind described by Massai. Bouchard ends with an appendix giving his full collation of the 1709 second edition and the 1714 edition for the play The Tempest.
Jane Kingsley Smith's article is about the rivalry between two eighteenth-century editors of Shakespeare ("'Distinguished By the Letter C': Edmond Malone and Edward Capell as Rival Editors of Shake-speares Sonnets", SQ 72[2021] 52-78). Edmond Malone was the first modern Shakespeare editor to provide an edition of Sonnets (1609), but Edward Capell anticipated him in various ways. Malone sought to conceal Capell's achievements in this regard. Before Capell, John Benson's 1640 edition of Shakespeare's Poems was often the base text for publishing the sonnets, although the 1609 Sonnets was also sometimes used despite the difficulty of getting an exemplar of it. When Malone got a copy of the 1609 Sonnets he emphasized its obscurity, claiming that Capell did not possess one, although in fact Capell did and Malone most likely knew this.
Capell championed the 1609 Sonnets as the true early text of these poems, but much of his work on this was unpublished so it is widely unknown. Kingsley-Smith debunks some myths that have arisen about Capell's work on Shakespeare's poems. She points to a number of emendations of the sonnets accepted today that Capell was responsible for. Malone adopted many of them without crediting Capell or else attributing them to an unidentified person "C". Kingsley-Smith ends by providing a full transcription of a preface about the sonnets that Capell wrote into his exemplar of Bernard Lintott's 1709 edition based on the 1609 Sonnets.
Cyrus Mulready argues that the literary-historical category of 'old plays' of Shakespeare's time was invented in the eighteenth century to form a contrast with what Shakespeare himself achieved ("Old Plays: Shakespeare, Robert Dodsley, and the Early Modern Dramatic Canon", Renaissance Drama 49[2021] 229-57). Mulready gives a sketch of the publication of Shakespeare's plays across the century after 1623, pointing out in particular the occasions when they, and other plays by his contemporaries, were characterized as old. Then he describes the project of Robert Dodsley to reprint some of the best of the plays of Shakespeare's contemporaries as his Select Collection of Old Plays in 12 volumes (1744).
Dodsley suggested that the chief merit of these old plays was the light they shone on Shakespeare's times, and subsequent republishers of his collection developed this idea of their relative inferiority, and their aid in serving as foils, to the achievements of Shakespeare. The collections of Dodsley and his followers show a greater diversity of genres than the Shakespeare canon shows, particularly because so many of Shakespeare's plays (about a quarter) are histories, whereas outside of Shakespeare this genre is much less common.
From the publication of these collections of non-Shakespearian plays there emerged a sense that 'old plays' were a genre, and this was useful in explaining certain aspects of the plays published under Shakespeare's name. Thus Lewis Theobald characterized 1 Henry 6 as being the result of Shakespeare adding a few "finishing beauties" to someone else's 'old play' that was brought to him. Edmond Malone, by contrast, dismissed 1 Henry 6 as wholly an 'old play' by someone else. But for both editors and others, the point of studying the 'old plays' was to appreciate what it was that Shakespeare learnt from and surpassed; this applied particularly to the 'old play' of King Leir. We can tell that 'old' is being used here in a judgemental not a chronological sense from the fact that the 'old' plays in question did not need to be much earlier than Shakespeare's writing.
In the 1608 quarto of King Lear, the King of France leads the invasion but then is "suddenly gone back", leaving his Marshall to take over. In the Folio, this detail is cut and the impression is given that Cordelia leads the French army. The earlier play King Leir makes it clear that the French king leads the invasion. Thus when editors conflate the editions to make a modern King Lear, their reliance on the quarto is in a way a return to the King Leir version. Mulready takes the view that Folio King Lear "presents a less coherent narrative" (p. 247), and this is why modern editors use Q's version: "because it clarifies the plot" (p. 247). Mulready makes no mention of the theory that Folio King Lear reflects Shakespeare's revision of the play.
Mulready considers the term 'old' (in "so old and so long out of use") used by Augustine Phillips to explain why the Chamberlain's Men were reluctant to perform Richard 2 for the Essex rebels in 1601 and why in King Lear Edmund refers to Edgar as entering "like the catastrophe of the old comedy" and why Biron in Love's Labour's Lost observes that his story does not end happily "like an old play". This section seems somewhat to conflict with Mulready's wider argument that the category 'old play' was invented long after Shakespeare's time in order to contrast with Shakespeare's work. In an appendix, Mulready tabulates the genres of the plays included in a series of 'old plays' collections published between 1744 (Dodsley's) and 1885 (A. H. Bullen's), using the genre assignments from Alfred Harbage's Annals. He draws here no conclusions about genre from his tables.
Scott Shepherd's article is a consideration of how an early edition has affected recent theatrical productions ("The First Quarto of Hamlet and its Performance in London, 2010-11", Shakespeare 17[2021] 472-87). Between 2000 and 2010, Shepherd argues, scholarly views of the first quarto of Hamlet (Q1, published in 1603) changed substantially, particularly because of Lukas Erne's 2003 book Shakespeare as Literary Dramatist (reviewed in YWES for 2003). Shepherd looks at how these changes affected two productions of the play in 2010-2011: the London replica Globe's directed by Dominic Dromgoole and the National Theatre's directed by Nicholas Hytner.
Dromgoole's programme text showed him in touch with scholarly thinking regarding the merits of Q1, while Hytner's programme text parroted the standard New Bibliographical lines about piracy and memorial reconstruction. But Dromgoole also wrote that Q1 was a touring text, which in fact we do not know. Hytner used Harold Jenkins's Arden Second Series edition for his script, while Dromgoole used the Moby digital version of the Victorian Globe edition.
Shepherd has examined the promptbooks for these two production and recorded the directors' variations from their base texts. Hytner occasionally used a Q1 reading, but quite possibly only because it was identified in Jenkins's textual collation. But Dromgoole was "downright profligate in his adoption of the text; he inserts around 100 separate lines from Q1, as well as substituting or adding individual words from the text at various points throughout the play" (pp. 481-482). Hytner's production was much more enthusiastically received by critics than Dromgoole's, although critics repeatedly stated (wrongly) that Hytner had cut almost nothing.
Abigail Rokison-Woodall offers a description of what her Arden Shakespeare Performance Editions are trying to achieve ("Editing Hamlet for Performance", Actes des Congrès de la Société Française Shakespeare 39[2021] n. pag.). She starts with an account of the history of acting editions from the seventeenth to the twenty-first century. Actors, she reports, are often taught that when a line within a verse section is noticeably short we must assume that it is followed by silence that lasts as long as it would take to complete the meter of the line.
Where there are three short lines in succession it might be possible to join them to make two verse lines in either of two ways -- joining the first and second or joining the second and third -- these are sometimes called amphibious lines. Rokison-Woodall points out that it is difficult for an actor trained that shortness implies a pause to make sense of the text if the editor has joined two of the lines from an amphibious triplet. Rokison-Woodall thinks it better to leave all three unjoined. Some practitioners, she reports, think that reading the performance history of a play they are putting into production is likely to harm their creative process.
Rokison-Woodall argues that there is need for a new set of acting editions for practitioners, "providing them with clear, straightforward, easily accessible glossing; visible, performatively viable, textual variants; guidance regarding the pronunciation of unusual or obscure words and the metrical scansion of lines" and dispensing with "the weighty introductory material" other editions have (p. 8). And her Arden Shakespeare Performance Editions series is meeting that need. The texts are from the Arden Third Series, but the notes (glossing the words and identifying textual variants) are placed on the recto of each opening, "immediately opposite the line to which they refer" (p. 8). For Hamlet, though, this edition does not use the three-version Arden Third Series text, instead providing a conflation of Q2 and F.
Regarding amphibious lines, the Arden Shakespeare Performance Editions "differ from most other editions in the representation of three consecutive short lines. Rather than printing them as one shared and one short line, they seek to make the ambiguous metrical connection apparent by indenting both the second and third portions of the line" (p. 9). She gives an illustration of this from the three-line exchange "Saw, who? . . . my father" in Hamlet 1.2, but I cannot see how it differs from her quotation on page 7 of the same exchange as it appears in Philip Edwards's New Cambridge Shakespeare edition, where the second and third lines are both indented from the left margin (the latter more than the former).
In Rokison-Woodall's series, textual notes are not given for departures from the copy text that are mere corrections of errors. Also provided, unusually, is a guide to pronouncing character names and place names. Wherever there is a transition from verse to prose or prose to verse, a facing-page note alerts the reader to this.
Andrew Hadfield traces Shakespeare's role in the evolving meanings of the word 'hint' ("Shakespeare's Hints", MP 119[2021] 235-47). Hadfield mentions that Othello "was first performed around 1600" (p. 235) but does not say where he gets this date from. Editors typically give 1603 as the terminus a quo because that is when Richard Knolles's History of the Turks and King James's Lepanto were published in London, and Othello seems to depend on both. Hadfield points out that Othello's 'hint' is the first Oxford English Dictionary citation and that the word was presumably unfamiliar at the time.
The first time 'hint' is used in the play is "It was my hint to speak" during Othello's account of telling Desdemona about his life of adventure, and although the spelling in the Folio is "hint" it is "hent" in the 1622 quarto and the sense is more of occasion or opportunity than of suggestion. The second use is later in the same speech, where Othello reports that Desdemona made clear signs of being attracted to him and he says "Upon this hint I spake". But this time where the Folio has "hint", the quarto has "heate" and in context that also makes sense: Othello is describing the heat that developed between him and Desdemona as she listened to his life story.
Hadfield surveys some other Shakespearian uses of the word 'hint' and makes a remark about one in Antony and Cleopatra ("When the best hint was given him, he not took’t, | Or did it from his teeth") that I do not understand. He writes "Again, this could mean that Octavius was given cues that he refused to accept, or subtle suggestions that he failed (or refused) to notice. Alternatively -- or, even, simultaneously -- the word might mean that he was given a cue" (p. 238). I cannot see how Hadfield's second sentence is making an alternative suggestion to the first: in both, a hint is a cue.
Overall, Hadfield finds that in Shakespeare the word 'hint' can simultaneously connote a cue or suggestion and an opportunity or occasion. Hadfield traces the origins of 'hint', which arose in theological texts and in part as an offshoot of the word 'hent' meaning (as a verb) to seize to take away and (as a noun) an attack or blow, and first meant an opportunity or occasion. Shakespeare was a pioneer in the use of 'hint' in literature: other playwrights and poets generally did not use it, except Ben Jonson in Epicoene and Nathan Field in Amends for Ladies.
The last article this year concerns the song in Measure for Measure ("Thomas Middleton's Absent Stanza: Musical Transplantation and the Textual Lacunae of William Shakespeare's Measure for Measure", TN 75[2021] 158-72). William David Green notes that the 1623 Folio text, our only authority, includes the first stanza of the song "Take O take those lips away" from Fletcher and Philip Massinger's play Rollo Duke of Normandy, written no earlier than 1617. The second stanza of the song is not used in Measure for Measure, presumably because it reveals that the singer is a man longing for a woman, which does not suit the context in Measure for Measure. Green reckons that the revision of Measure for Measure by Thomas Middleton deliberately draws attention to the fact that the song is not completed.
At the indoor Blackfriars theatre, music played during the act intervals, so maybe the tune of "Take O take those lips away" was played in the interval between Act 3 and Act 4 so that Act 4 started with this song "to smooth over the break in the action" (p. 165) caused by the interval. Mariana's command to the Boy to "Break off thy song" would highlight that the song's second stanza is not to be sung. In Rollo Duke of Normandy and in Measure for Measure the Boy who sings the song has no lines of dialogue and Green wonders if the same boy sang it in the two plays. Having Mariana cut off the Boy's song "can itself be said to play with the idea of theatrical cutting" (p. 171).
The two chapters-in-books relevant to our review this year both came from Hamlet: The State of Play edited by Sonia Massai and Lucy Munro for the Arden Shakespeare State of Play series. In the first, Charles Adam Kelly and Dayna Leigh Plehn argue that the first quarto (Q1) of Hamlet was an early version of the play by Shakespeare ("Q1 Hamlet and the Sequence of Creation of the Texts", in Massai & Munro, eds. Hamlet: The State of Play, pp. 151-74). Kelly and Plehn begin by sketching the textual situation of Hamlet, essentially correctly although their remark that 18 of the 36 plays in the 1623 Folio had been printed before in quarto (p. 151) leaves unstated which ones they are referring to and what it means for two editions to contain the same play.
There are 16 indisputable cases and four disputable cases of a Shakespeare plays being published in quarto before the Folio. Before King John there had been the two-part Troublesome Reign of King John in 1591, before The Taming of the Shrew there had been The Taming of a Shrew in 1594, before 2 Henry 6 there had been The Contention of York and Lancaster in 1594, and before 3 Henry 6 there had been Richard Duke of York in 1595. Presumably Kelly and Plehn get to 18 cases of prior quartos by rejecting the claims of The Taming of a Shrew and The Troublesome Reign of King John. As we shall see, this problem of judging when one piece of writing is essentially the same as another is central to their entire essay.
Kelly and Plehn's methodology owes something to Sherlock Holmes. They list the "Possible Creators of the Text of Q1" Hamlet and "Possible Sources of the Text" (p. 153) and argue that since their lists are exhaustive the process of eliminating all but one for each leads to knowledge: ". . . the remaining possibility . . . becomes a certainty" (pp. 152-153). What they overlook is that the possibilities are not exclusive: the manuscript from which Q1 Hamlet was printed might, in principle, have been created by memorial reconstruction of the performance script of an early authorial version of the play, which are two explanations (memorial reconstruction and early authorial version) that Kelly and Plehn list as if mutually exclusive.
Kelly and Plehn rightly insist that publishing a book was a significant investment for a publisher, who would want to be sure of a return on investment. But they seem to go beyond what we know when they write that Nicholas Ling "must have been absolutely assured that a more desirable text would not be released and printed immediately to eclipse his investment in the 1603 quarto of Hamlet" (p. 155). Ling of course also published the second quarto (Q2) in 1604-5. Perhaps Ling was assured by Shakespeare's playing company that no other publisher would get a better text of the play, but we cannot know. If Shakespeare's company tacitly approved of the publication as Kelly and Plehn suggest ("Perhaps none of them [the bad quartos] were unauthorized", p. 155) then why not give Ling for his first edition the considerably better script of Hamlet that went on to be published as Q2, since it would reflect better on them, being the better play?
Historical details aside, the real methodological problem with Kelly and Plehn's essay comes when they attempt to compare the texts of Q1, Q2, and Folio (F) Hamlet and quantify the differences. For each page of Q2 they tabulate the percentage of "Q1 Lines Same as Q2", "Q1 Lines Similar to Q2", and "Q1 Unique Lines" (Table 6.5). The obvious question this raises is just where Kelly and Plehn draw the boundary between two lines being the same and their being similar, but this is addressed nowhere in their essay nor in the supplementary material they provide online to support it. We can get a sense of their strictness from another table in which they attempt to show that in the second scene of the play there is a run of 16 lines in which the Q2 version is essentially just the rearrangement (not the garbling) of the Q1 version's lines and part lines (Table 6.9).
In this table, Kelly and Plehn assert that Q1's "And wee did thinke it right done, | In our dutie" is the same as Q2's "And we did thinke it writ downe in our dutie". The Oxford English Dictionary does not give 'right' as a possible spelling of the past participle 'writ' nor 'writ' as a possible form of the adverb 'right', and nor is 'done' a form of the preposition 'down', although 'down' is a possible spelling of the past participle 'done'. Kelly and Plehn do not comment on whether they think "right done" and "writ down" are the same phrase spelt two ways, but it seems on balance more likely that these are two similar-sounding but distinct phrases meaning 'properly executed' and 'recorded in writing' respectively.
Elsewhere, Kelly and Plehn's table asserts that Q1's "know it" is the same as Q2's "knowe of it", Q1's "watched" is the same as Q2's "watch", "as he" is the same as "as it", and so on for several more differences that go beyond mere spelling. These are counted by Kelly and Plehn as identical phrasings in lines that they considered merely reordered between Q1 and Q2, so their sense of what counts as similar phrasing (rather than identical) must encompass even greater differences. But how much greater? There is no way to know: Kelly and Plehn never define what it means for two words or phrases to be identical or to be merely similar. This silence vitiates their study of the textual problem.
In their Table 6.7 Kelly and Plehn record the proportion of "words rather than lines" (p. 158) spoken by Marcellus, Hamlet, and Horatio that are the same in Q1 and Q2. They give totals for the numbers of words spoken by each, but it is not clear whether the shared use of, say, 'the' by Hamlet in Q1 and in Q2 counts as an example of what Kelly and Plehn mean by such "concordance" (p. 158). Since in English the top 25 most used words account for about a third of all writing we should expect a lot of this kind of concordance when comparing any two texts -- even when comparing speeches by different characters -- yet Kelly and Plehn again give no indication of what they mean by the concordance they are tabulating.
Kelly and Plehn apply statistical methods to their counts of lines that are the same. By their reckoning, there are 10 passages totalling 209 lines that are present in Q2 but absent from F. Overall, Kelly and Plehn found that "Q1 contains 1,631 Q2 lines, or 41.8 per cent of the text of Q2 . . ." (p. 160). Notice, this is not a comparison of how long Q1 is compared to Q2: Kelly and Plehn mean that "1,631 is the sum of the 1,054 identical or near identical lines and 577 identifiably concordant lines" shared between Q1 and Q2 (p. 173n23).
To test the hypothesis that Q1 is based on the recollection by one or more persons of the lines in Q2, Kelly and Plehn ask what proportion of the 209 lines present in Q2 but absent in F should we expect to find in Q1 if a random process of forgetting made someone fail to record in Q1 some of Q2's lines. Since Q1 has about 42% of the lines found in Q2 (plus some more that are not found in Q2), Kelly and Plehn reason that we should expect 42% of those 209 lines to make it into Q1, or around 87 lines. But in fact none of those 209 lines is present in Q1. Kelly and Plehn write that ". . . it is a statistical certainty that the number of lines remembered would fall in the range of six standard deviations [from the mean of 87] or 31-143 lines" (p. 161).
To support their claim of statistical certainty, Kelly and Plehn cite (on their page 173n24) Appendix E of their earlier work The Evidence Matrix for the 1st Quarto of Shakespeare's 'Hamlet' (2016), where they made the calculation using the concepts of "expected value", "normal curve" (meaning a Gaussian distribution), and "standard deviation". This indicates that they were using the wrong method for the value they are trying to calculate, which is the likelihood of randomly selecting 1631 lines from Q2's 3900 lines (to make the 1631 lines in Q1 that are found in Q2) and managing to avoid selecting any one of the 209 lines that Q2 has that F lacks.
The correct calculation for this uses the hypergeometric not the Gaussian distribution, since the analogy Kelly and Plehn are drawing with memorial reconstruction is the random selection, one at a time and not putting them back, of 1631 white balls from a bag containing 3900 balls (one for each line in Q2) of which 3691 are white balls (representing the lines in Q2 also found in F) and 209 are black balls (representing the lines present in Q2 and absent in F).
Of course, the act of memorially reconstructing a play is not necessarily analogous to selecting lines at random from that play since characters' trains of thought often run across lines. But Kelly and Plehn are right that it is unlikely that one could randomly select 1631 lines from Q2 and happen not to select any of the 209 lines in Q2 that are absent in F. Recall, however, where that count of 1631 comes from: it is "the sum of the 1,054 identical or near identical lines and 577 identifiably concordant lines" (p. 173n23) shared between Q1 and Q2. As Kelly and Plehn acknowledge, there are hundreds of other lines in Q1 that are not identical to, nearly identical to, or identifiably concordant with lines in Q2.
Kelly and Plehn's hypothesis does not test any possible and complete account of where Q1's text came from, since it leaves out those hundreds of lines that Kelly and Plehn judge as having no likeness to lines in Q2. And their count of 1691 lines in Q1 that are like lines in Q2 is itself dependent on an unstated method for judging the degree of likeness between a line in one edition and a line in another, and nowhere do Kelly and Plehn indicate how they decide this. (As we saw with Heejin Kim's article at the start of this review, there exist well-established methods such as Dice Similarity for doing this.) The numerical aspect of their approach gives an appearance of precision and certainty that the method itself does not warrant.
The same problems of subjectivity apply to Kelly and Plehn's analysis of how the German-language play Der Bestrafte Brudermord is related to Q1, Q2, and Folio Hamlet. To allow for the language difference, Kelly and Plehn counted not words or lines but "plot elements" (p. 167), such as "Hamlet instructs Corambus to reward players", that are shared between, or are unique to one of, the texts
Because Der Bestrafte Brudermord has over 50 unique plot elements not found in Q1/Q2/F, Kelly and Plehn conclude that it must in fact have been written first. And since they have excluded (in the above statistical analysis) the possibility that Q1 Hamlet was derived from the version of the play underlying Q2, they understand this as evidence "to further position BB [Der Bestrafte Brudermord] as the earliest text, with Q1 at an intermediate point between BB and Q2/F1 . . ." (p. 168).
Kelly and Plehn point the reader to an online table that gives "The analysis of BB vs Q1 vs Q2/F1" that "reveals several 'trajectories' of revision passing from BB, through Q1, to Q2/F1" (p. 168). Having looked at that table, which has no explanatory text about how to read it, I cannot see that the plot likenesses it identifies support the BB-to-Q1-to-Q2-to-F trajectory in preference to any other.
None of the above tells us that Kelly and Plehn are wrong to assert that Q1 Hamlet represents Shakespeare's first attempt at the play and that Q2 and F represent subsequent revisions of it. Rather, their essay is a combination of others' arguments, which they recount reasonably accurately, and their own numerical analysis that is fatally flawed by concealed subjectivity, misuse of standard statistical methods, and illogical deductions.
This year's second and final chapter-in-book follows immediately after Kelly and Plehn's in the same book. In it, William Dodd argues that what are called pragmatic markers found in the first quarto (Q1) of Hamlet but not in the second quarto (Q2) or the Folio (F) show us how the first actors of the play departed from the script in their performances ("The Hamlet First Quarto: Traces of Performance?", in Massai & Munro, eds. Hamlet: The State of Play, pp. 175-97).
Can we test if the small interpolations that we find in some early editions, especially bad quartos, really do come from actors? To explore this, Dodd counted how often actors in the BBC Shakespeare on Film videos added or removed pragmatic markers (p. 176) but he does not yet give any examples of these. It turned out that the actors on average added three for every one they removed. And the first quartos of Romeo and Juliet, Richard 3, Henry 5, Hamlet, and The Merry Wives of Windsor do have higher numbers of these pragmatic markers than the later editions of the same plays, and they drop a few that appear in the later editions. From this Dodd determines that these editions "show clear signs of (further) oral adaptation of a prior written text" (p. 178).
Dodd lists and classifies the pragmatic markers in groups such as 'Openers/closers, attention-getters' including 'come', 'go', and 'how now', 'Deictics' including 'here', 'this', and 'now', and 'Vocatives' including 'father', and 'my good lords'. He gives tables for his counts of how often actors in the BBC films add, substitute, displace, and omit these markers, and the absolute counts are far less than happens in the Shakespeare bad quartos, which he puts down to the BBC films being "carefully-rehearsed" (p. 180).
Dodd takes a passage for which the Q1 and Q2 Hamlet versions are highly similar -- Hamlet asking Horatio and Marcellus about the Ghost -- and highlights the pragmatic markers in Q1 not found in Q2. They tend to be the same kinds of markers that Dodd found modern actors adding. The added pragmatic markers in Q1 Hamlet are disproportionately in the mouth of Hamlet: he speaks 38% of the play's lines but 50% of the added pragmatic markers.
Dodd starts to refer to Q1 as reflecting what Richard Burbage did in performance and seems to think the case for their being his additions is made simply by detailing the pragmatic markers found in Q1. Dodd writes "I believe that we have accumulated enough quantitative evidence by now to point confidently towards Shakespeare's players as the primary source of pragmatic marker variants" (p. 187).
For Dodd, Q1's version of "To be or not to be" is "clearly a debased version of a more coherent text" (p. 188). Dodd highlights these lines in particular as meaningless:
And borne before an euerlasting Iudge,
From whence no passenger euer retur'nd,
The vndiscouered country, at whose sight
The happy smile, and the accursed damn'd
Dodd asks "What passenger ever failed to return from an everlasting judge? And what kind of judge can suddenly turn into an undiscovered country that arouses a smile in the happy and (leaves?) the accursed damned?" (p. 188). These questions are over-literal in respect of poetry, since being "borne before" a judge means being taken to a place (the site of judgement), and it is that place, not the judge, that is called an undiscovered country because no one returns to report it to humankind.
Dodd quotes two more passages from Q1, with annotations to convey differences from Q2/F, and describes what he thinks they do to the characterization, for instance: "make Hamlet appear more aggressive" (p. 190), "harping allusively" (p. 192), and "stepping up the aggressiveness" (p. 192). I cannot hear how these added words achieve these things on their own: surely it is at least in part a matter of how the actors say them.
So to the round-up from Notes and Queries. In the first of two articles, MacDonald P. Jackson offers evidence that the play King Leir is not by Kyd ("Is Thomas Kyd the Author of the Anonymous King Leir?", N&Q 266[2021] 70-4). This is Jackson's response to Brian Vickers's "Kyd's authorship of King Leir" reviewed in NYWES for 2018. Jackson lists how often the words/phrases 'for to', 'I do', 'unto', 'I'd', 'ne'er', and 'thine' appear in The Spanish Tragedy, Solimon and Perseda, Cornelia, and King Leir.
For each word, the contrast between its appearance in the first three plays -- which everyone agrees are Kyd's -- and the last one is striking: King Leir uses some of these words/phrase at much higher rates than known Kyd plays do, and uses the last of them ('thine') not at all although they abound in the known Kyd plays. An obvious objection is that perhaps the contractions were made in the process of transmission, but Jackson points out that ". . . the monosyllabic 'I'd' fits the verse, so appears to be original" (p. 72).
Jackson also finds that King Leir uses the phrase 'ere't be long' six times, while the three known Kyd plays never use it. Likewise the phrase 'as much' is used 12 times in King Leir but nowhere in The Spanish Tragedy and Solimon and Perseda and three times in Cornelia. Jackson acknowledges that King Leir is not only unlike known Kyd plays, but also unlike most other plays of its time.
As an independent check, Jackson ran the kind of experiment made possible by Pervez Rizvi's Collocations and N-Grams dataset. For each 'target' play, Jackson used Rizvi's data to count how often just one other play contains a 'weighted formal trigram' from that play. The term 'weighted' means that instead of using raw counts of matches, the count is "divided by the combined number of word-tokens in the pair of plays" (pp. 72-73). The term 'formal' means that trigrams found within larger matching n-grams (4-grams, 5-grams, and so on) are counted once for each time they occur. We could instead choose to count only the larger matching n-gram and ignore the smaller matches within it.
Jackson's experiments show that in a list of plays that is rank-ordered by how many such unique trigrams it shares with a target play, the other plays by the author of the target play tend to come out near the top of the list. This works for each of the known Kyd plays taken as the target play, but when Jackson did it with King Leir as the target play, the known Kyd plays were not at the top of the rank order for shared unique trigrams. The closest matching play to King Leir on this test is Robert Yarington's Two Lamentable Tragedies.
In his second note, Jackson argues that the plays that Darren Freebury-Jones wants to add to the Kyd canon are not by Kyd ("The 'Restored' Kyd Canon: Interpreting the Evidence of Unique Verbal Matches", N&Q 266[2021] 74-8). Jackson is responding to Freebury-Jones's note "Unique Phrases and the Canon of Thomas Kyd" (reviewed in NYWES for 2020), objecting that in it Freebury-Jones seems to think that showing similarities between the plays he wants to add to the Kyd canon -- Arden of Faversham, King Leir, and Fair Em -- constitutes evidence that they are Kyd's. It does not: you have to show that these plays are like the known Kyd plays: The Spanish Tragedy, Solimon and Perseda, and Cornelia.
Jackson describes a test that Rizvi conducted for him using the Collocations and N-Grams dataset. Of 86 plays by Chapman, Fletcher, Jonson, Marlowe, Middleton, and Shakespeare, this experiment in uniquely shared trigrams correctly assigned 84 of the plays to their actual authors. In this test, Fair Em and Arden of Faversham came out matching with The Spanish Tragedy, Solimon and Perseda, and Cornelia (so it looks like Fair Em and Arden of Faversham are Kyd's), but King Leir and the "dominant shares" (p. 76) of Edward 3 and 1 Henry 6 matched with Marlowe not Kyd.
Jackson suspects that chronology is a confounding factor here: after Marlowe's death, playwrights got more distinctive in their voices so late-play attributions are more accurate then early-play ones. Countering some specific claims by Freebury-Jones, Jackson shows by counting in a different way that Arden of Faversham scenes 4-8 have far more matches with Shakespeare's canon than with Marlowe's or Kyd's when using Rizvi's data on 'unique formal trigram matches', as defined in the previous Jackson article. The same tests for Fair Em and King Leir show that they match much more with Marlowe's plays than Kyd's, not because they are Marlowe's but because they are not Kyd's but someone else's.
Thomas Merriam also contributed two notes this year. In the first he argues that perhaps Marlowe did not write The Jew of Malta ("Possible Light on a Kyd Canon--continued", N&Q 266[2021] 78-80). Merriam took 54 1000-word chunks from 1 Tamburlaine, The Spanish Tragedy, and The Jew of Malta and compared them with the 36 plays in the 1623 Shakespeare Folio looking at the 24 most-frequent characters (including the space). Reducing these 24 dimensions to two with Principal Component Analysis shows a clear distinction by author: Marlowe in the top-left corner, Shakespeare in the top-right corner. By this test, The Jew of Malta does not look much like Marlowe.
Using the 247 most-common character trigrams instead of single characters, Merriam does the experiment again using just seventeen 1000-word segments from 1 Tamburlaine and twenty 1000-word segments from The Spanish Tragedy and presents his findings as a "cluster analysis" (that is, a dendrogram). With the exception of one out of the 20 chunks from The Spanish Tragedy, all 17 of the 1 Tamburlaine chunks are on one branch and all 20 of The Spanish Tragedy chunks are on the other branch, so this method is clearly able to separate parts of these two plays. But that does not mean it is separating them because of authorship: it could be separating them on one or more other variables. Merriam remarks that: "The one error in thirty-seven is equivalent to a probability of one in 237 or 137 billion". Merriam does not say what he thinks this is the probability of, so the claim is meaningless.
Merriam repeats that last experiment using this time 1 Tamburlaine and The Jew of Malta and finds that these two likewise appear on two distinct branches in his dendrogram. From this Merriam concludes that Marlowe did not write The Jew of Malta, and, since the dendrogram looks like the one in the previous experiment that pitted a known Marlowe play against a known Kyd play, that Kyd might be the author of The Jew of Malta. These conclusions are not justified by the experimental results as described.
At the very least, Merriam would need to put a number of plays by different authors into his experiment and show that the chunks largely or wholly cluster by authorship and that the chunks from The Jew of Malta largely or wholly cluster with those written by some other author or at least do not cluster with chunks from Marlowe plays. Short of such analysis, all that Merriam has shown is that what he has created can tell apart the plays 1 Tamburlaine and The Jew of Malta. That this dendrogram looks like the previous one tells us nothing about whether Kyd might have authored The Jew of Malta.
In his second note, Merriam argues that Marlowe wrote the play published as the 1595 octavo of Richard Duke of York ("The Authorship of The True Tragedy of Richard, Duke of York", N&Q 266[2021] 92-4). Merriam divided Richard Duke of York into seventeen 1000-word segments and counted in those, and in 24 Shakespeare plays and three Marlowe plays, the frequencies of 315 most-common character trigrams and then reduced the dimensions from 315 to two using Principal Component Analysis. In the resulting scatterplot there is a clear left-right distinction along Principal Component 1 between the Shakespeare texts of the left and the Richard Duke of York segments on the right, with Marlowe in the middle.
Merriam draws a line through the scatterplot such that Richard Duke of York and the three Marlowe plays are on one side of the line and all the Shakespeare plays are on the other side. Merriam reports, but does not show the scatterplot for it, that the same analysis of Folio 3 Henry 6 gives most but not all of the play to Marlowe. Merriam concludes that Shakespeare adapted Marlowe's play Richard Duke of York, rather as he did Peele's The Troublesome Reign of King John.
In the last authorship attribution study this year, Naseem A. Alotaibi offers new reasons for thinking that Fletcher was a major contributor to the play underlying Lewis Theobald's Double Falsehood ("'[T]hat Other Great Poet': Double Falsehood, the Arden Shakespeare, and Further Evidence for Fletcher", N&Q 266[2021] 184-9. Brean Hammond's 2010 Arden Third Series Double Falsehood noted the parallels between names in that play and Shakespeare's works, but missed some with parallels in Fletcher's works, such as Roderigo in Fletcher's The Pilgrim, and Julio in Fletcher's The Maid in the Mill co-written with William Rowley, and Don Martino Cardenes in Fletcher and others' A Very Woman.
Alotaibi also lists some Double Falsehood plot parallels with Fletcher plays that Hammond missed, and more names in common including Violente and others where Fletcher in other plays used the Cervantean names of characters in Double Falsehood. That is, Fletcher elsewhere used the name used by Cervantes for a character who has the same function in the source Don Quixote but has a different name in Double Falsehood. Hammond concentrated on Shakespearian uses of the word 'inward' that appears in Double Falsehood, but in fact, argues Alotaibi, the stronger match is the full phrase 'inward with him' in Double Falsehood and Fletcher's The Island Princess and Fletcher and Rowley's The Maid in the Mill.
There is also the parallel of "'Twas but the coyness of a modest bride" in Double Falsehood and "this is but the coyness of a bride" in Francis Beaumont and Fletcher's The Maid's Tragedy. Another parallel is 'seat' being used to mean skill in horsemanship, which appears in Double Falsehood and in a Fletcher scene in Shakespeare and Fletcher's The Two Noble Kinsmen. For that last you have to emend 'feat of horsemanship' in The Two Noble Kinsmen to 'seat of horsemanship', as Lois Potter did in revising her Arden Third Series edition of The Two Noble Kinsmen for a 2015 reprint, under the influence of a note by MacDonald P. Jackson reviewed in YWES for 2009. Alotaibi thinks that Fletcher and Massinger's play The Spanish Curate might be a sequel to Cardenio.
B. J. Sokol argues that Falstaff alludes to a serious mythological trinity in passing in The Merry Wives of Windsor ("The Transformation of Falstaff in The Merry Wives of Windsor and His Allusion in 5.1.4-5 to the Treble Goddess", N&Q 266[2021] 96-9). The Falstaff of The Merry Wives of Windsor is different from the one in the earlier history plays in that he cannot support his followers. Also, Mistress Quickly has fallen from tavern owner engaged to Falstaff and then married to Pistol (in the histories) to unmarried housekeeper for Caius. Most importantly of all, Falstaff is resurrected.
When Falstaff decides to have a third go at seducing Mistress Ford he says "They say there is divinity in odd numbers, either in nativity, chance, or death". Those last three words are the three symbolic meanings of women in Sigmund Freud's essay "The Theme of the Three Caskets": the three fates of Clotho (the tendencies we are born with), Lachesis (accident), and Atropos (death). Freud claimed that in these myths the youngest silent one of three sisters is the good one (as in King Lear). Silence is death, so the right choice is the death choice.
Substitution of something by its opposite is a frequently occurring mental process in Freud's work. When Lear sees Cordelia as the right choice, he is able to die, which for Freud is a reversal of the image of the Death goddess carrying off the dead hero in German mythology. Why does Shakespeare have Falstaff refer to "nativity, chance, or death" at the point where he decides to have a third go at Mistress Ford? Because Shakespeare was anticipating the serious theme he would explore in King Lear and The Tempest, including "every third thought shall be my grave".
Fernando Castanedo reckons that the correct reading in A Midsummer Night's Dream is "Seeking sweet savours for this hateful fool" (from the second quarto and Folio) not "Seeking sweet favours . . ." as the first quarto has it ("The Case for 'Savours' in A Midsummer Night's Dream, IV.i.48", N&Q 266[2021] 402-8). The hateful fool is Bottom, for whom Titania has been foraging, according to Oberon. Editors generally adopt Q1's 'favours' but Castanedo argues for 'savours'. Q1 does have notable 'f'/'long-s' confusions, in both directions.
Castanedo's case for 'savours' is based on euphony and the pattern of alliteration in the wider speech in which this line appears. In this line, "Seeking sweet fauours for this hateful fool" gives a pattern of sounds that runs s s f f f f (if we include the 'ful' of 'hateful') whereas if the reading is 'savours' the pattern is a more balanced (less parodic) s s s f f f. What is a 'favour'? Those defending this reading say it means flowers used as love-tokens. But in A Midsummer Night's Dream and elsewhere Shakespeare uses 'favours' to mean flowers only after having used the word 'flowers' to make it plain (that is, 'favours' refers back to named flowers), otherwise it means gifts such as rings, gloves, and the like. Also, flowers seem to abound in Titania's bower and she has already wreathed Bottom with flowers, so that she should go seeking them elsewhere does not make much sense.
Castaneda thinks it more likely that Titania is seeking "honey, provender, oats, hay, and/or the nuts previously mentioned in the scene" (p. 406), that is savoury things for Bottom to eat. This would also better suit the humiliation of Titania, since foraging for food is a low-class activity while gathering flowers is not. Elsewhere Shakespeare uses 'sweet favour' twice, but both in the sense of visually appreciated beauty, and he uses 'sweet savours' in The Taming of the Shrew and The Comedy of Errors to mean things pleasing to taste/smell. And of course there is "savours sweet" earlier in A Midsummer Night's Dream referring to fragrant flowers.
'Sweet savour' also occurs dozens of times in the Geneva Bible, frequently in relation to offerings made to God. This too enhances the sense in A Midsummer Night's Dream: Titania is being considerably more idolatrous if she is seeking for Bottom the kinds of offering rightly due to God than if she is merely hunting for 'favours'. Possibly the pious nineteenth-century editors, consciously or unconsciously, preferred 'favours' to 'savours' because they detected the impious Biblical echo.
William Baker and Andrew Thompson list some annotations by George Henry Lewes in his copy of Charles Knight's 1842-1844 edition of the works of Shakespeare ("George Henry Lewes's Annotations of the Comedies in Charles Knight's Shakspere (2nd Edition, 1842-1844)", N&Q 266[2021] 304-6). The only one of textual interest is Lewes's defence of Orsino in Twelfth Night saying of a favourite tune that "it came o'er my ear like the sweet south" where 'south' is Alexander Pope's emendation of the Folio's 'sound'. Knight overturned Pope's emendation and reverted to the Folio's 'sound' on the principle that Shakespeare "nowhere else made the south an odour-breathing wind". Lewes in his annotation disagreed: "The sound may breathe but it does not steal nor give odour to violets, hence South is the word. Shakspere does not compare the sound of music to the sound of a breeze; but the affect of music to the effect of the breeze". Baker and Thompson have a second note about Lewes's annotations but none are of textual interest ("G. H. Lewes Reads Cymbeline: His Annotations in Knight's Shakspere", N&Q 266[2021] 306-9).
Andy Reilly shows that the first edition to disambiguate the speech prefixes and stage directions in Hamlet for the Player King and Queen from those for Claudius and Gertrude was John Hughs's in 1718, not Steevens in 1778-1780 as usually claimed ("The Origins of the Player King and Player Queen Speech Prefixes in Hamlet", N&Q 266[2021] 104-6). Reilly traces the origins of the editorial habit of rewriting the early editions' stage directions and speech prefixes to disambiguate the king and queen within the inset play from King Claudius and Queen Gertrude watching it.
Philip Edwards in his New Cambridge Shakespeare Hamlet attributed the stage direction "Enter the PLAYER KING and QUEEN" to Peter Alexander's 1951 complete works edition, about which Reilly comments that this "can only be explained by a potential mistake between the somewhat similar names Peter Alexander and Alexander Pope" (p. 105). But there is no mistake: Edwards thought that Alexander was the first to use exactly this wording and so rightly credited him, while Pope used "Enter King and Queen, Players". Reilly has found that the earliest edition to use "Enter Player King and Queen" and the speech prefixes "Pl. King" and "Pl. Queen" is John Hughs's 1718 edition published by Mary Wellington.
In a second note, Reilly shows that although Hughs's edition of Hamlet is dated 1718 it was actually on sale in December 1717, since there was an advertisement for it in The Post Boy for 5-7 December 1717 ("The Correct Publication Date of Mary Wellington's '1718' Edition of Hamlet", N&Q 266[2021] 217-9). Also, The Daily Courant for 27 December 1717 carried an advertisement showing the book on sale then. Reilly finds other Wellington books that have title page dates later than the date the books actually became available. The Hamlet edition purports to reflect the play "As it is now Acted by his Majesty's Servants", and the new dating of December 1717 rules out this referring to any of the 1718 productions and makes likeliest the one running in the late-1716-to-early-1717 season.
James P. Bednarz argues that, contrary to Roslyn Knutson's claim, Hamlet does not allude to James 1's coronation, nor does it call his government tyrannical ("Does Hamlet Allude to the Coronation of James I?", N&Q 266[2021] 102-4). In the second quarto of Hamlet, Rosencrantz explains that the players of the city are arriving at Elsinore with the remark "I thinke their inhibition, comes by the meanes of the late innouasion". In the Folio he says this and expands on it with speeches about the eyrie of little eyases, meaning the boy players in the city. Knutson suggested that the late innovation was James 1's accession of 1603.
But, as Frederick Boas earlier pointed out, when used of a change of government the word 'innovation' was invariably negative. Knutson supported her interpretation of the word 'innovation' by citing its uses in 'The Triumph of Love' in Fletcher and Field's Four Moral Representations in One and in James Shirley's The Coronation, but reading these instances of 'innovation' Bednarz finds that they clearly "use it in its ordinary political sense: to signify the forced overthrow of legitimate authority. They do not use it to describe an orderly or legal succession such as James 1's assumption of power in 1603/4" (p. 103).
Knutson also claimed that the phrase "they are tyrannically clapped for it" in Hamlet refers to the Children of the Chapel losing their royal patronage in 1608. But as Bednarz points out, Shakespeare would not have dared suggest that James 1's regime was tyrannical. Rather, the word 'tyrannical' merely means 'excessively': the boy players were wildly applauded for their performances that mocked the adult theatre. In all, this passage alludes to contemporary theatre not politics.
Brian Vickers reckons that in Much Ado About Nothing, Dogberry's repeated (and unintentionally self-mocking) injunction that everyone remember that he has been called an ass -- "me . . . an ass", "I am an ass", "I am an ass", and "I have been writ down an ass" -- was inspired by Gabriel Harvey doing something similar in his book Pierce's Supererogation or A New Praise of the Old Ass of 1593 ("Dogberry, The Book of Job, and Gabriel Harvey", N&Q 266[2021] 99-100). Thomas Nashe had called Harvey an ass and in defending himself Harvey sounds rather like Dogberry in repeating the accusation and calling for public sympathy.
When Hamlet says he recognizes Polonius with "You're a fishmonger", editors usually gloss 'fishmonger' as meaning a pimp and related terms, but Quentin Skinner offered textual evidence that it could mean one who blows his knows on his sleeve. Vanessa Lim has more evidence for Skinner's suggestion that fishmonger's were said to have runny noses and that wiping one's nose on one's sleeve was thought disgusting ("Polonius the Nose-blowing Fishmonger", N&Q 266[2021] 100-2). One is from Plutarch's Quaestiones Convivales and another from Erasmus's De Civilitate Morum Puerilium, the latter aimed at children and likely to be known by the schoolboy Shakespeare.
Boris Borukhov has new evidence to date The Phoenix and the Turtle ("A Possible Source for William Shakespeare's 'The Phoenix and the Turtle'", N&Q 266[2021] 106-7). Part of the poem uses trochaic tetrameters in groups of three lines (tercets) rhyming aaa. Shakespeare never did this before or after (although he did use trochaic tetrameters elsewhere), but this form was used in Nicholas Breton's poem "A Report Song in a Dream between a Shepherd and his Nymph" that appeared in 1600 in the anthology England's Helicon compiled by John Flaskett. Since England's Helicon also contained Shakespeare's poem "On a day, alack the day!" it is likely that Shakespeare read it and saw Breton's poem and imitated its meter for part of The Phoenix and the Turtle. If so, The Phoenix and the Turtle was written after England's Helicon was entered into the Stationers' Register on 4 August 1600.
We do not have the music for the song "Let me the cannikin clink" that Iago sings in Othello. Thomas Erik Nielsen ("'The Clinke': A Potential Source for Iago's Cannikin Song in Othello", N&Q 266[2021] 107-11) points out that Cambridge University's Matthew Holmes Manuscript Dd.9.33, a lutebook from around 1600, has a tune called "The Clink" that "fits Iago's lyrics perfectly in meter and length" (p. 108). Rebecca Ann Bach surveys Shakespeare's use of the word 'chuck, said affectionately of a child or spouse and derived from the sound that chickens make and the fact that chickens were thought to be especially kind to their family members ("Chuck in Shakespeare's Plays", N&Q 266[2021] 112-3). But as Bach shows it could also be used contemptuously by one man to another. Bach writes as if the word 'chuck' were now only an archaic term, but in fact it is still common in the sense Bach describes in the north of England and it can be heard in television programmes made and/or set there.
Igor Djordjevic has a new source for one aspect of King Lear ("The Possible Origins of Edmund's Birtherism: Thomas, Lord Cromwell By 'W.S.'", N&Q 266[2021] 113-5). By 'birtherism' Djordjevic means a person's perceived unfitness for an elevated social role because of an accident of birth. Djordjevic thinks that Shakespeare may have got the idea for Edmund's sense of injustice about being kept down because of his illegitimacy from the character of young Cromwell in the play Thomas Lord Cromwell, who soliloquizes about how base men rise and great ones fall, and hence maintains hope that he may rise. Djordjevic acknowledges that the parallel is not terribly close: Cromwell is not a younger bastard son as Edmund is. Cromwell just thinks that the prevailing social hierarchy is unfair and he wants a meritocracy.
Roger Stritmatter has a new source for another aspect of King Lear ("'Something Rich and Strange': Senecan Influences on the Dover Cliff Scenes of King Lear", N&Q 266[2021] 116-9). Stritmatter traces the influence of Seneca's play Phoenician Women on the Dover Cliff scene in King Lear, including "an attempted suicide of the blinded father (Oedipus) guided by a sympathetic child (Antigone)" (p. 116). Also, "In both plays old men wander as outcasts with the support and guidance of one pious child opposed by two warlike and ambitious children . . ." (p. 116). Stritmatter thinks that among the strongest parallels is the ekphrastic description of a high cliff to which a child is asked to lead a parent who wants to commit suicide by leaping off it. He details some more minor parallels of phrasing and action.
Bailey Sincox reckons that the cup that Leontes in The Winter's Tale imagines having a spider in it -- saying that the drinker takes no infection from it if he does not know the spider is there -- comes from a wider romance topos of the cup or horn test of female chastity ("A Missed Shakespeare-Ariosto Connection", N&Q 266[2021] 127-9). In Ludovico Ariosto's epic poem Orlando Furioso, Rinaldo, while gazing on statues of great women, is invited to drink from a cup that is able to tell a man if his wife is faithful. If she is not, the wine from the cup will splash onto the man's chest. Rinaldo wisely elects not the try the test because he is secure about his wife's fidelity. According to Sincox, Ariosto's poem and Shakespeare's play are linked by the theme of metacognition -- thinking about how one's thinking works -- and the problem of the unknowableness of female chastity. Plus of course there is the link that Robert Greene adapted Orlando Furioso to the stage and wrote Pandosto that The Winter's Tale is based on.
Simon Reynolds finds a source for Prospero's island in The Tempest in one described in an essay by Plutarch ("Plutarch's Island of Saturn as a Source for The Tempest", N&Q 266[2021] 133-6). Plutarch's collection of essays called Moralia has a tale called 'The Face of the Moon' about a magic island in the Atlantic with a continent beyond it. Perhaps this gave Shakespeare the idea. Moralia was published in English in 1603 based on a French translation published in 1572. The likenesses with Prospero's island are the sweet airs and climate of the place, its bounteous natural advantages, and the population of helpful spirits. Reynolds details some lesser and more tenuous likenesses. Another one of the tales in the collection features a tempest.
Carl D. Atkins ("Shifting Heads to Solve a Crux in Comedy of Errors", N&Q 266[2021] 400-2) explains that the following speech by Adriana in The Comedy of Errors does not need the emendation it usually gets:
2.1.109 I see the Iewell best enameled
2.1.110 Will loose his beautie: yet the gold bides still
2.1.111 That others touch, and often touching will,
2.1.112 Where gold and no man that hath a name,
2.1.113 By falshood and corruption doth it shame:
2.1.114 Since that my beautie cannot please his eie,
2.1.115 Ile weepe (what’s left away) and weeping die.
Atkins argues that "that others touch" refers back to "Iewell" not to "gold": jewels being touched will lose their beauty, gold being touched will not. This is what is meant by a shifted head in Atkins's title. The "and often touching will" just repeats the point: touching will tarnish the jewel's beauty. That takes care of the first three lines, which complete a thought.
A new thought starts with line 112, in which "Where" means 'whereas' and "no" means 'any', so the idea of 112-113 is that gold and any man of reputation are two things that can be shamed (disgraced) by lies. In 114-115, "Since that" means simply 'since' and the sense is straightforward: because my beauty cannot please him I will cry away the rest of it and die.
Alexander Gourlay, and Patricia Southard Gourlay ("'Widow Dido' and 'Widower Aeneas': Some Undeciphered Ribaldry in The Tempest", N&Q 266[2021] 408-10) aim to illuminate this exchange from The Tempest:
ANTONIO Widow? A pox o' that! How came that 'widow' in? Widow Dido!
SEBASTIAN What if he had said 'widower Aeneas' too? Good Lord, how you take it!
They suggest that "how you take it!" must mean that Antonio is laughing and that "Widow Dido" was perhaps pronounced as if it were "We do Dido", which they call "not a great joke" (p. 409). I am not clear what the joke is, unless simply that Dido was reputed to be promiscuous, as they remark Virgil paints her.
Gourlay and Gourlay think that "widower Aeneas" then builds on the joke by sounding like "we do her in the arse". Strangely, Gourlay and Gourlay think that ". . . it can be objected that 'to do [someone]' was not then and is not now a common phrase in the sense required by our reading . . ." (p. 410). To 'do' someone certainly has that slang sense now in British English and Aaron uses that sense in "I have done thy mother" in Titus Andronicus.
Books reviewed
Sonia and Lucy Munro, eds. Hamlet: The State of Play, Arden Shakespeare State of Plays Series. Bloomsbury [2021]. 272 pp., £20.29, ISBN 978-1350232747