Picture of Gabriel Egan G a b r i e l   E g a n  .  com

Foucault's epistemic shift and verbatim repetition in Shakespeare

    Since the mid-1980s the ideas of Michel Foucault have had a powerful effect upon Shakespeare studies. First appearing in French in 1969, Foucault's essay "What is an author?" was published in English in 1975 (Foucault 1975), two years before Roland Barthes's "The death of the author" (Barthes 1977, 142-48) which, from different premisses, argued essentially the same point: the 'author' is a phenomenon frequently misunderstood in literary studies. For Foucault, an author was not the simple concept it might seem: some kinds of writing have authors and others do not. A contract, for example, is usually not the writing of any of the persons who sign it. With the rise of empiricism in the seventeenth and eighteenth centuries, science, which had hitherto valued the author, became impersonal (relying on verifiable, reproducible, tests) while literary works, which hitherto had been relatively indifferent to authors, became personal (Foucault 1979, 148-49). Authors are not so much intrinsic originators of texts but rather extrinsic products of the consumption of texts; author are assigned to texts which need them and not others. It is better, Foucault argued, to think not of authors but of the author-function, an exegetic principle applicable not only to written texts but whole fields of study. Above all, the author-function is "the principle of thrift in the proliferation of meaning" (Foucault 1979, 159), used coercively to exclude outlandish interpretations. Appropriately enough for an argument about textual multiplicity, this famous summation of Foucault's idea does not appear in all versions of his essay and is absent from the first English translation. Foucault located the important epistemic break around 1800 when the needs of private intellectual property generated our modern sense of an author. In another area of Shakespeare studies, Foucault's work on sexual identity achieved a wide audience via Alan Bray, whose Homosexuality in Renaissance England (Bray 1982) argued that in the early modern period there was no subject-position from which one could identify oneself as a homosexual. For the early modern, sodomy was a matter of doing, not being: sinful, to be sure, but not a matter of self-definition. This claim is more contentious than the matter of the author-function, and Joseph Cady has made a convincing case that in the work of Francis Bacon and Thomas Heywood is evidence that a homosexual identity was in existence, contrary to Foucault and Bray's claim (Cady 1992). In early modern printing and theatre practice there is evidence to counter the Foucauldian claim that their sense of verbatim repetition was quite unlike ours.

    Near the end of a ground-breaking study of the notion of authenticity in relation to the Shakespearian text, Margreta de Grazia observed a textual phenomenon which is disturbing for us, but apparently was not for the early moderns. Edmond Malone's 1790 edition of The Plays and Poems of William Shakespeare marked a sudden shift in Shakespeare studies in which a new rigorous objectivity, based on factual records, was required, and by reference to the earliest available printings Malone attempted to reproduce Shakespeare with as little interference as possible, ideally 'verbatim'. But Malone noticed that Shakespeare did not share his concern with verbatim reproduction: in a play a single document can be read aloud by different people using different words. Specifically, De Grazia cited 2 Henry 6 in which the articles of peace are twice read aloud with differences in wording. Malone put this down to carelessness on Shakespeare's part, but De Grazia sought an explanation using Michel Foucault's notion of the author-function:

For Malone, these deviations within the text were symptomatic not of the medium's instability, but rather of Shakespeare's 'negligence'. Indeed, he found them so characteristic in the contested works that he considered them conclusive proof of Shakespeare's authorship. At this point, we can see how the notion of a single authorial consciousness (with its occasional lapses into unconsciousness) serves a regulatory function, converting what we have called the 'copiousness' of both mechanical and rhetorical 'copy' into personal idiosyncrasy. Verbatim repetition requires a language in standardized stasis, put under the mastery of precisely the historicized, individuated, and entitled subject Malone both presumed and projected in his 1790s Shakespeare. (De Grazia 1991, 223)

Although De Grazia does not mention Foucault's notion of the author-function for another two pages, and then only in a footnote concerning the anti-Stratfordian conspiracy theorists, the above paragraph is clearly informed by Foucault's assertion that authors are a thrifty brake upon copious interpretative proliferation. Elsewhere De Grazia had already explicitly invoked Foucault's author-function in relation to twentieth-century editing theory (De Grazia 1988, 80-82), especially New Bibliography, and in the book-length study she traced the search for authenticity to its source: Malone's Shakespeare edition (revised in 1821), the first to display the concern for rigorous authenticity which was to characterize subsequent work in the field. With the late-eighteenth century "invention of man", a grand reorganization of knowledge into subject-oriented disciplines according to Foucault, literary scholarship created 'Shakespeare' as "an exemplary instance of the autonomous self" (De Grazia 1991, 10) comprising the biography and the works, and, most enduringly, a set of rules and practices for relating the constructed author to the artistic output.

    Certainly in support of Foucault's assertion is De Grazia's observation that around 1800 the use of what we call quotation marks in printed works changed. Previously the symbol marked sententiae, sayings which because of their importance were worth remembering and repeating and thus were worth circulating freely (a form of public ownership), but around the time Foucault identifies as the birth of the modern author-function, quotation marks began to be used to acknowledge borrowing of another's words and thus showed a respect for private intellectual property (De Grazia 1991, 214-19). At this time novels began to distinguish one character's words from another's by use of quotation marks. We can easily accept that when repeating another's words in a play a character might garble, compress, or paraphrase what is heard, but surely a letter read twice should not change its contents. De Grazia commented:

With predictable indignation, Malone observed that when even a written document--a letter or proclamation, for example--was read at one point in a play and then again at some later point, 'inaccuracies' were introduced: 'When [Shakespeare] had occasion to quote the same paper twice (not from memory, but verbatim) from negligence he does not always attend to the words of the paper which he has occasion to quote, but makes one of the persons of the drama recite them with variations, though he holds the very papers before his eyes.'1 In a scene from Henry VI Part II, for example, the same articles of peace between France and England are twice read aloud, first by the Bishop who reads 'dutchy of Anjou and the county of Maine' and then by the Duke who read 'the dutchies of Anjou and of Maine'.

1 Malone, PPWS, vi. 416. (De Grazia 1991, 222-23)

If proven, this claim would take the principle of Shakespearian textual instability to a new level, since indifference to textual variation might then exist not only between authorial versions, as E. A. J. Honigmann showed (Honigmann 1965), but also within a single document, and simply because of a pre-1800 mindset which did not care for logical regularity in the way we do. No longer could such variation be taken as a sign of the imperfect work of memory or unusually careless transcription, but would actually be the normal habit of the time. Fortunately for those who value the Enlightenment, and despite De Grazia's reference to "a letter or a proclamation" and her apparent choice of 2 Henry 6 "for example", there is only the example of Folio 2 Henry 6 and it is not as De Grazia describes it.

    De Grazia's works are concerned with the presence of non-authorial writing, with the "wide array of collective and extended contributions and transformations" (De Grazia 1988, 82) which find their way into textualizations, and she refuses to demote these in relation to authorial writing. Anyone who misquotes De Grazia has the ready excuse that her own work validates such transformations, whether conscious or unconscious. By the same token, one is entitled to imagine a smirk on Jacques Derrida's face as he corrects printers' errors in the proofs of his published work, entirely putting others' contributions and transformations under erasure. Equally playful is Randall McLeod's deliberately inaccurate quotations from and allusions to Shakespeare criticism, such as calling the love of punning Shakespeare's "fertile Cleopatra" (Cloud 1997, 136) instead of Samuel Johnson's "fatal Cleopatra" (Shakespeare 1765, xxiii). Amongst such textual jouissance it might appear boorish to insist that one's quotation of De Grazia renders accurately her quotation of Malone, but mine does. However, her quotation does not render accurately her source, the "Dissertation on the three parts of King Henry VI", in which Malone actually wrote:

When he has occasion to quote the same paper twice, (not from memory, but verbatim,) from negligence he does not always attend to the words of the paper which he has occasion to quote, but makes one of the persons of the drama recite them with variations, though he holds the very paper_ quoted before his eyes. (Shakespeare 1790, 416)

My emphasis marks points where De Grazia's quotation of Malone deleted or inserted material, which error--failure to repeat verbatim--Malone is in the act of lamenting and which troubles the post-Enlightenment sensibility. However, the sense of the passage is not really altered, so why not let this sort of thing pass? Because De Grazia's concern for the materiality of the text will not permit such sloppy idealism: there is, she insists in this book and elsewhere, nothing but the textualization (De Grazia & Stallybrass 1993).

    So apt for my argument are De Grazia's slips that one is tempted to resort to Freud to explain them: surely something other than inadvertence must be at work. Whatever witty spirit commanded De Grazia's hand, it did not leave for at least another sentence, making her write that the articles of peace are read "first by the Bishop . . . and then by the Duke". Rather, in all early versions of the play the order of speaking in the scene (1.1) is the other way around: the Duke reads, breaks off, then the Bishop takes over. De Grazia quoted from the Folio text and modernized the spelling, as the rules of quotation allow, but she added an extra word of her own, which they do not. The second reading is not "the dutchies of Anjou and of Maine" (as she has it) but "the dutchies of Anjou and Maine" (a modernization of F's "the Dutchesse of Aniou and Maine"). De Grazia invoked Foucault and 2 Henry 6 to claim that the early moderns would not have perceived a significant discrepancy in a document being read aloud twice with different wordings. Whether there was such an epistemic shift is too large a question to address here, but it is demonstrable that the evidence from Shakespeare does not support the argument that there was. To see why, we must attend to the processes of the early modern theatre and the printing-house that, quite impersonally and for no epistemological reason, tended to produce the textual coherence that De Grazia and Foucault claim was not typical until after 1800.

        The First Part of the Contention/2 Henry 6 begins with Suffolk's return from France bringing Margaret, King Henry 6's bride-to-be, and the articles of peace which specify a kind of negative dowry of English possessions in France to be given to her father. In the quarto, Humphrey Duke of Gloucester begins to read the articles:

    Humphrey. Imprimis, It is agreed betweene the French King
    Charles, and William de la Poule, Marquesse of Suffolke, Embas-
    sador for Henry King of England, that the said Henry shal wed
    and espouse the Ladie Margaret, daughter to Raynard King of
    Naples, Cylles, and Ierusalem, and crowne her Queene of Eng-
    land, ere the 30. of the next month.
Item. It is further agreed betwene them, that the Dutches of An-
    ioy and of Maine, shalbe released and deliuered ouer to the
    King her fa.
                       Duke Humphrey lets it fall.

Gloucester breaks off reading, saying he is overcome by a "sodain qualm", and Cardinal Beaufort, Bishop of Winchester, takes over:

Cardinall. Item, It is further agreed betweene them, that the
Dutches of Anioy and of Mayne, shal be released and deliue-
red ouer to the King her father, & she sent ouer of the King
of Englands owne proper cost and charges without dowry. (Shakespeare 1594, A2v-A3r)

Where the two recitations overlap ("Item . . . fa") the words in the quarto version are identical, although the punctuation and spelling vary. The same events in the Folio version are thus:

    Glo. Reads. Inprimis, It is agreed betweene the French K.
, and William de la Pole Marquesse of Suffolke, Am-
bassador for Henry King of England
, That the said Henry shal
espouse the Lady Margaret
, daughter vnto Reignier King of
, Sicillia, and Ierusalem, and Crowne her Queene of
, ere the thirtieth of May next ensuing.
    Item, That the Dutchy of Aniou, and the County of Main,
shall be released and deliuered to the King her father
    King. Vnkle, how now?
    Glo. Pardon me gracious Lord,
Some sodaine qualme hath struck me at the heart,

As in the quarto, Cardinal Beaufort, Bishop of Winchester, takes over reading the articles:

    Win. Item, It is further agreed betweene them, That the
Dutchesse of Aniou and Maine
, shall be released and deliuered
ouer to the King her Father
, and shee sent ouer of the King of
Englands owne proper Cost and Charges
, without hauing any
. (Shakespeare 1623, m2v)

As De Grazia noted, in the Folio there are substantial differences between the first and second readings of the articles. The second reading inserts "It is further agreed betweene them" after the listing tag "Item", changes "Dutchy" to "Dutchesse", cuts "the County of" and inserts "ouer" before "to the King her Father". If the Folio represents what was performed, the theatre practitioners must have thought the audience would not notice or would not care if two men read the same piece of paper differently, and this suggests habits of mind which we, as modern readers and writers, do not share. Even if the Folio does not represent what was performed, it might well represent what was in the underlying manuscript (whose writer did not notice or noticed and did not care) and in any case it shows that those involved in printing the play were not sufficiently concerned, or not sufficiently attentive, to regularize the two readings.

    In the quarto version of this scene, the second reading is identical to the first, as we have seen. More importantly, however, the two quarto readings are almost identical to the Folio's second reading. The New Bibliographical consensus is that the quarto text was probably based on a memorial reconstruction (perhaps made for the purpose of touring the play) and the Folio printing which was based on foul papers, perhaps reworked for a revival, supplemented by Q3, itself a Q1 reprint (Wells et al. 1987, 175-78; Montgomery 1989, 22). Without accepting this view, we can make some initial inferences about the performances represented by these texts. Whether or not the papers that were passed to the actors contained a discrepancy between the first and second readings of the articles of peace, one additional copy of those words would have been made as a property document to be used on the stage. In a study of such documents, Tiffany Stern argued that the words to be read on stage would be specially marked in authorial papers: they would be headed 'letter' (or some such) so that the theatrical scribe would know to leave them out of the actor's 'part' and instead to write them on a separate piece of paper to be handed to the actor during the performance (Stern 1999). In the playbook used backstage to run the performance, Stern argued, it would be essential that speakers' names were clearly marked but the words themselves could be omitted since these existed on the property itself, so only the tag 'The letter' need be recorded. However, it would be wise for the playbook to repeat the speech prefix after the letter to make absolutely clear that the same person is still speaking, and thus in early printed texts the use of a particular speech prefix after the reading of a letter does not indicate that someone else read the letter, as has often been thought in the case of Antonio's letter to Bassanio in The Merchant of Venice (Stern 1999, 232). Not all of this is convincing, since it is difficult to believe that a playbook would be licensed if it were manifestly an incomplete record of the words to be spoken on stage, and we have no evidence of the licensing of supplementary textual properties. Stern quoted E. K. Chambers's belief that early modern actors did not memorize their 'letters' since the property documents would be available to them during the performance.1 Whether or not actors memorized their 'letters', the creation of a property document enforces singularity: it can be only one thing or another. A professional dramatist who in his authorial papers created a discrepancy between two readings of the same letter--a discrepancy of the kind seen in Folio 2 Henry 6 and pointed out by Malone and De Grazia--could hardly be unaware that the making of the property document would in any case eliminate one of the variants. Strong evidence that during performance actors read aloud the lines in their property documents (rather than memorizing them beforehand) is supplied by Edward Alleyn's 'part' for the title role in Orlando Furioso lacking the verses he reads aloud, as Stern pointed out (Stern 1999, 231), so any authorial discrepancy between two readings of a document necessarily disappeared in performance.

    Whether or not the papers passed to the theatrical scribe were consistent in the two readings of the articles of peace in 2 Henry 6, the theatre scribe must have chosen one of the readings to form the copy for the property document. A moment's reflection would make him choose Cardinal Beaufort's reading, because it is complete, rather than Duke Humphrey's which is not. If the actors relied on the property document for their lines, both men would speak what was written down for Cardinal Beaufort in the papers passed to the theatrical scribe. This conclusion is consistent with (but does not prove) the New Bibliographical view that the Folio represents the discrepant authorial papers and the quarto represents the non-discrepant performance, since the two quarto readings are virtually identical with Cardinal Beaufort's reading from the Folio. There is a small difference between Cardinal Beaufort's reading in the Folio and the two readings in the quarto, and it too is consistent with F representing the authorial papers and Q representing the performance. If in the authorial manuscript Cardinal Beaufort referred to "the Dutches of Anioy and Maine" to be "released and delivered" to Margaret's father, there might be an unwanted suggestion that a person (a duchess of Anjou and Maine) rather than lands were to be freed and handed over. Indeed this is what Beaufort appears to say in the Folio: "the Dutchesse of Aniou and Maine". There is no such duchess in the play, only lands are to be handed over. To prevent the potential misunderstanding that the spelling "Dutches" creates, the theatre scribe making the property document could have repeated the word "of" before "Maine" so that the phrasing becomes unambiguous, since no-one could be the 'duchess of Anjou and of Maine'. Thus a long 'e' pronunciation (in the phonetic alphabet, 'i:') of "Dutches", indicating two duchies, would be ensured. This is what Q has, and again this is consistent with that text representing what was performed and, in this case, including an alteration which removed an unwanted ambiguity.

    There is an alternative explanation for the differences between Q and F concerning the articles of peace. In the textual introduction to the Oxford Complete Works, William Montgomery recorded that in editing the play for his doctoral thesis he decided that F's second recitation derived not from the single manuscript used for most of the play, but from a quarto, probably Q3 (Wells et al. 1987, 176). In the textual note for Cardinal Beaufort's recitation (numbered 1.1.55-59 in the Oxford edition), Montgomery reported that "F has been contaminated by Q here: the Cardinal's reading should be identical with Gloucester's" (Wells et al. 1987, 179). To understand why Montgomery thought this we need to turn to his doctoral thesis in which he considered the long-recognized phenomenon that certain passages in the play show "extraordinarily close correspondence in Q and F", so close that the memorial reconstruction hypothesis cannot provide the explanation because no-one's memory could be so good (Montgomery 1985, 2:xxxvii). Peter Alexander's answer was that the reconstructors had scraps of manuscript to supplement their memories, but R. B. McKerrow's explanation has won out: F was intermittently set up from a copy of Q. Curbing the excesses of Andrew S. Cairncross's Arden edition of the play (Shakespeare 1957), Montgomery observed that the only way to demonstrate the dependence of one edition on another is to show that the later maintains a clear error which is also in the earlier. It is no good showing that indifferent variants (that is, those which are equally as good as a different word which appears in a third text) agree since these can happen independently of one another and it is equally pointless to show, as Cairncross frequently did, that good readings agree since these can come from a reliable manuscript source and not the earlier printing. An error which Montgomery thought significant was the spelling "Dutchesse" in the Cardinal's reading of the articles of peace in the Folio text and in Q3, as against "Dutches" which appears in Q2 and Q1. Montgomery admitted that a compositor might make such a change in spelling whatever his copy read, and especially if he saw "Dutches" and thought this meant a person more unambiguously spelt "Dutchesse". If a compositor did this, he is responsible for creating in Q3 the absurdity of "the Dutchesse of Anioy and of Maine", which perhaps does not seem wrong until one has to say it aloud. Because the spelling could be compositorial, Montgomery thought this not a strong Q3/F link, but nonetheless suggestive of some Q/F connection or that "very similar manuscripts lay behind this part of both Q and F" (Montgomery 1985, 2:xlii). Considering all the Q/F agreements in error, Montgomery showed that if a quarto was consulted to make F then it was probably Q3 (Montgomery 1985, 2:xlvii). In all Montgomery found 7 moments in the play where F seems dependent on Q, and he decided that because the link is transcriptional--Q3 was consulted to fill gaps in the copy for F--it was now reasonable for him to "extend these seven points of demonstrable transcriptional contact to include that portion of their immediate context in which Q and F, for the most part, verbally agreed" (Montgomery 1985, 2:xlvi). Starting from each moment of agreement in error, Montgomery worked outwards until F and Q3 ceased to agree, and because several of the 7 spots of agreement are close to one another, this 'join the dots' procedure makes them merge, producing 3 substantial chunks of F where Q3 was consulted: Folio TLN 63-79, 858-904, 2598-2639. To these three sections Montgomery added two more where stage directions in F are so like those in Q that a transcriptional link was, he thought, certain: the entrance of the mayor of St Albans (F TLN 795-6) and the entrance of the armourer and his neighbours in 2.3 through to the end of that scene and the opening direction of the next one (F TLN 1115-1170). The first section where Montgomery saw a transcriptional link between Q and F is the Cardinal's reading of the articles of peace and continuing on for a dozen lines until the King, Queen, and Suffolk exit.

    Having noted that the Dutchesse/Dutches spellings were not strong evidence that F was printed directly from Q3, Montgomery admitted that the other evidence pointing to a transcriptional link between F and one of the quartos is even weaker (Montgomery 1985, 2:xlvii). This other evidence is 5 cases of mislineation of verse which F shares with all the quartos, and a speech prefix problem in F at TLN 2625-28 where Butcher and Dicke (the same person) get different successive speeches (Montgomery 1985, 2:xliii-xliv). Cairncross explained the Butcher/Dicke confusion by imagining that an additional speech was added to the Q used to make F and this should have been accompanied by the deletion of a speech prefix, but by error the deletion was not made. This error in F, then, seems to be at a point where F depends on Q, but of course the error could just as easily be an error in the authorial manuscript underlying F, as Montgomery observed (Montgomery 1985, 2:xliv). The mislineation evidence Montgomery characterized as "not conclusive" of Q influencing F, but he did not speculate how else the agreement in error might have come about (Montgomery 1985, 2:xlvii); coincidence must be one possibility. Montgomery decided that "The 'duches' [sic] evidence is perhaps the strongest . . . but again, by no means conclusive" (Montgomery 1985, 2:xlvii). Having declared the evidence for a transcriptional link between F and Q inconclusive, it is surprising that Montgomery chose to merge the 7 points of "demonstrable transcriptional contact" into 3 larger sections were F can be assumed to have been set from Q3. Montgomery assumed that Q3 was used as copy for F starting from the Cardinal's reading of the articles of peace and continuing on for the next dozen lines as well. But this hypothesis requires that, as Q3 was copied, some of it was left out and invented additions put in, for there are differences. The second "of" was removed from Q3's "of Anioy and of Mayne", Q3's "without dowry" was supplemented by two additional redundant words to make F's "without hauing any Dowry". An additional "the" was added to Q3's "we heere create thee first Duke of Suffolke" to make F's "create thee the first Duke of Suffolke", which extra "the" is semantically redundant but regularizes the metre. In what might be a simple correction, a redundant "and" was removed from Q3's list "Winchester, Gloster, Yorke, and Buckingham, Somerset, Salisbury, and Warwicke". In Montgomery's hypothesis, the person who copied this chunk of Q3 to make F (whether scribe or compositor) made these 4 minor alterations as he went, one of which improved metre and another removed redundancy, and yet he made no attempt to adjust the new material to fit what had already been read out from the same document in Gloucester's broken-off reading of it. It is hard to know how to take the word "should" in Montgomery's observation that "the Cardinal's reading should be identical with Gloucester's" (Wells et al. 1987, 179), since his explanation of the textual situation rests upon the existence of an early modern copyist or compositor even less concerned with coherence and fidelity than the haphazard norm posited by De Grazia.

    The New Bibliographical consensus that Q1 represents a necessarily imperfect memorial reconstruction of a play better represented by F was attacked by Steven Urkowitz who saw Q1 as an equally viable dramatic version (Urkowitz 1988), but Roger Warren's response convincingly countered with a series of moments for which a conjecture of garbling best explains Q1's relation to F (Warren 2000, 195-201). Warren did not, however, explicitly counter Urkowitz's observation that Q1's stage directions contain verbal parallels with F's, which ought not to be the case in a report since these elements of the script are not spoken, nor memorized other than as actions (Urkowitz 1988, 252-53). How can memorial reconstruction explain actors remembering not only their lines but the exact phrasing of a play's stage directions? Montgomery had spotted these parallels in the phrasing of stage directions but pointed out that there is only a limited number of ways to describe an action (Montgomery 1985, 2:xliv-xlv). In any case, all but one of the stage directions that Urkowitz thought "terminally embarrassing" to the theory that the quartos derive from a memorial reconstruction--Folio TLN 784, 795-6, 902-4, 1115-9, 1169-70, and 2633-4 (Urkowitz 1988, 253)--fall in sections of F which Montgomery had decided were directly copied from Q3, this being Montgomery's explanation of the Q/F likenesses. Urkowitz's one example not covered by Montgomery's transcriptional-link-hypothesis is a stage direction that includes dialogue (Q1 C2r: "Enter one crying, A miracle, a miracle"; F TLN 784: "Enter one crying a Miracle"), so even if not discounted by Montgomery's principle that there is only a limited number of ways to describe an action, it can nonetheless be discounted because verbal memory was needed for it, and hence it is not out of place in an alleged memorial reconstruction.

    Montgomery's claim of an F/Q transcriptional relationship for Beaufort's reading of the articles and the following twelve lines is unconvincing because of the small differences listed above. If, however, Montgomery is right then the discrepancy between Gloucester's reading of the articles and Beaufort's that is seized on by De Grazia exists only in F and arises from the printer switching copy between Gloucester's and Beaufort's readings; such a printer's error does not constitute evidence of an epistemological shift. If Montgomery is wrong, then De Grazia is right inasmuch as the Folio shows that a writer could make a manuscript in which there was a discrepancy between the two readings of a single document in a play. But since it would be known to all that the textual processes of the theatre would eliminate one of the variants before performance we can hardly conclude from this that the early moderns did not share our dislike of inconsistency. Only one other Shakespeare play dramatizes the reading and re-reading of a single property document, and it shows no discrepancy between the first and second reading. Indeed, it shows such extraordinary agreement in every particular between the first and second readings that it can only be understood in terms of the economics of printing. In Cymbeline 5.4 Posthumus awakes to find a "tablet" (also called a "label" and a "book") on his breast and he reads it aloud:

Whenas a Lyons whelpe, shall to himselfe vnknown, with-
out seeking finde, and bee embrac'd by a peece of tender
Ayre: And when from a stately Cedar shall be lopt branches,
which being dead many yeares, shall after reuiue, bee ioynted to
the old Stocke, and freshly grow, then shall Posthumus end his
miseries, Britaine be fortunate, and flourish in Peace and Plen-
(Shakespeare 1623, bbb3v)

In the next scene Philharmonus the Soothsayer is called upon to read aloud this prophecy and to declare its meaning (Shakespeare 1623, bbb6r). Although the dramatist presumably wrote the prophecy twice, the theatre scribe would have made only one property document and if this controlled what was said on the stage there could be no variation between the two readings. But the two Folio readings are identical not only in wording but also punctuation and spelling and what is more the line-endings fall in exactly the same places in each reading and the relative positions of the lines within the measure are also preserved with greater accuracy than can be achieved when type is distributed and recomposed. For example, in both readings the bottom of the stem of the "f" in "himselfe" exactly meets the top of the "b" in "by". A close examination of irregularities such as the break at the top of the second "e" in "tender" indicates what happened: the same block of type was used in the first (5.4) and second (5.5) readings. After the printing of forme bbb3v:4 (the end of 5.4 and beginning of 5.5), the block of italic type was set aside to be reused for the printing of forme bbb1v:6 (the end of 4.2 and final printed page of the Folio). In a study of the recurrence of individual types, Charlton Hinman's analysis of the printing of these formes did not draw attention to this unusual movement of a whole block of type (Hinman 1963, 322-24). Because the Folio was set by formes it is likely that this labour-saving opportunity was noticed during casting off when the content and sequence of formes was determined. Peter W. M. Blayney's useful table of compositor attributions confirms that the two readings of the prophecy were printing in the order that they appear in the story, and both by compositor B (Shakespeare 1996, xxxv-xxxvii). If verbal differences existed between the two readings of the "Labell" in the underlying copy, they must have been sufficiently small for this interference in the text to have been acceptable. Warren Smith noticed that the two printings of the prophecy were identical in the tiniest matters, but like Stern he believed that the words of the prophecy did not occur in the promptbook and so had to be set by consulting the property document twice (Smith 1950, 180). However, being set twice from the same copy would not produce identical letter spacing, nor would the same broken "e" be likely to occur in the same word twice.

    There are no other occasions in a Shakespeare play when a document is read and re-read on stage, although in Contention of York and Lancaster/2 Henry 6 there is a conjuring scene (1.4) during which the spirit Asnath's words are written down as he speaks them and this incriminating document is later read aloud. The case of Asnath is inconclusive because the onstage transcriber might be supposed to have failed to record exactly what he heard, so there is a perfectly simple explanation for the reading of this document not matching Asnath's prophecies. Thus, in the only two Shakespearian examples of the onstage re-reading of documents (neither of which is De Grazia's claimed "letter or proclamation") the technology of textual reproduction (scribes making properties and printers making books) and the needs of the repertory system (which impelled actors to rely on properties documents) worked together to promote coherence. Knowledge of these economies of labour should inform a materialist approach to the subject. The fashionable insistence that there are only the surviving printed texts and that we should have nothing to do with Platonic idealizations of the work ignores that the fact that those who spoke and printed the texts did idealize, even if those who wrote and copied the texts had not. Perhaps the underlying manuscript for the printing of Cymbeline had the two readings of the prophecy in identical words, spelling, and punctuation. This would be untypical, but not impossible and it would make a useful example to counter De Grazia's use of non-verbatim repetition in 2 Henry 6. If, as is somewhat more likely, the two manuscript versions of the prophecy differed in small details such as spelling or larger ones such as punctuation and choice of words, a printing-house worker nonetheless felt that the opportunity to save labour was more important than the preservation of these differences, and he effaced them by reusing his first setting of type. The textual materialism advocated by De Grazia treats non-verbatim repetition as symptomatic of an epistemological difference between then and now, but the printing-house worker must have decided that a verbatim, literatim, and indeed punctilious reproduction would be perfectly acceptable. It is the same prophecy, after all. Our textual materialism is not materialist enough if it overlooks the regularizing effects of performance practices and of textual reproduction in the playhouse and the printing-house, and too materialist if it denies the professionals concerned their reasonable idealizations, such as 'it is the same prophecy, after all'. The economies of labour in playhouse and printing-house operated to reduce the proliferation of variation well before Foucault's claimed epistemic shift of 1800.


1My informal poll of actors working at Shakespeare's Globe in London found them equally divided between learners and non-learners of their 'letters', but they are only ersatz early moderns.

Works Cited

Barthes, Roland. 1977. Image-Music-Text. Trans. Stephen Heath. London. Fontana.

Bray, Alan. 1982. Homosexuality in Renaissance England. London. Gay Men's Press.

Cady, Joseph. 1992. "'Masculine Love', Renaissance Writing, and the 'New Invention' of Homosexuality." Homosexuality in Renaissance and Enlightenment England: Literary Representations in Historical Context. Edited by Claude J. Summers. New York. Haworth Press. 9-40.

Cloud, Random. 1997. "What's the Bastard's Name?" Shakespeare's Speech-headings: Speaking the Speech in Shakespeare's Plays: The Papers of the Seminar in Textual Studies, Shakespeare Association of America, March 29, 1986, Montreal. Edited by George Walton Williams. Newark. University of Delaware Press. 133-209.

De Grazia, Margreta and Peter Stallybrass. 1993. "The Materiality of Shakespeare's Text." Shakespeare Quarterly 44. 255-83.

De Grazia, Margreta. 1988. "The Essential Shakespeare and the Material Book." Textual Practice 2. 69-86.

De Grazia, Margreta. 1991. Shakespeare Verbatim: The Reproduction of Authenticity and the 1790 Apparatus. Oxford. Clarendon.

Foucault, Michel. 1975. "What is an Author?" Partisan Review 42. 603-14.

Foucault, Michel. 1979. "What is an Author?" Textual Strategies: Perspectives in Post-structuralist Criticism. Edited by Josue V. Harari. Ithaca. Cornell University Press. 141-60.

Hinman, Charlton. 1963. The Printing and the Proof-reading of the First Folio of Shakespeare. Vol. 2. 2 vols. Oxford. Clarendon.

Honigmann, E. A. J. 1965. The Stability of Shakespeare's Text. London. Edward Arnold.

Montgomery, William. 1985. The Contention of York and Lancaster: A Critical Edition. Unpublished D.Phil. thesis, University of Oxford.

Montgomery, William. 1989. "The Original Staging of The First Part of the Contention (1594)." Shakespeare Survey 41. 13-22.

Shakespeare, William. 1594. [The Contention of York and Lancaster] The First Part of the Contention Betwixt the Two Famous Houses of Yorke and Lancaster. STC 26099 BEPD 119a (Q1). London. Thomas Creede for Thomas Millington.

Shakespeare, William. 1623. Comedies, Histories and and Tragedies. STC 22273 (F1). London. Isaac and William Jaggard for Edward Blount, John Smethwick, Isaac Jaggard and William Aspley.

Shakespeare, William. 1765. The Plays. Ed. Samuel Johnson. Vol. 1: Preliminary Matter; The Tempest; A Midsummer-Night's Dream; The Two Gentlemen of Verona; Measure for Measure; The Merchant of Venice. 8 vols. London. J. and R. Tonson [etc.].

Shakespeare, William. 1790. The Plays and Poems. Ed. Edmond Malone. Vol. 6: 1 Henry 6; 2 Henry 6; 3 Henry 6; Dissertation on the Three Parts of King Henry VI; Richard 3. 10 vols. London. Rivington.

Shakespeare, William. 1957. The Second Part of King Henry VI. Ed. Andrew S. Cairncross. The Arden Shakespeare. London. Methuen.

Shakespeare, William. 1996. The Norton Facsimile of the First Folio of Shakespeare. Ed. Charlton Hinman. Second edition with a new introduction by Peter W. M. Blayney. New York. Norton.

Smith, Warren D. 1950. "New Light on Stage Directions in Shakespeare." Studies in Philology 47. 173-81.

Stern, Tiffany. 1999. "Letters, Verses and Double Speech-prefixes in The Merchant of Venice." Notes and Queries 244. 231-33.

Urkowitz, Steven. 1988. "'If I Mistake in Those Foundations Which I Build Upon': Peter Alexander's Textual Analysis of Henry VI Parts 2 and 3." English Literary Renaissance 18. 230-56.

Warren, Roger. 2000. "The Quarto and Folio Texts of 2 Henry VI: A Reconsideration." Review of English Studies. n.s. 51. 193-207.

Wells, Stanley, Gary Taylor, John Jowett and William Montgomery. 1987. William Shakespeare: A Textual Companion. Oxford. Oxford University Press.