The Winter's Tale Appendix

Irregular, Doubtful, and Emended Accidentals in F1

For an explanation of the contents of this list, see here. In the notes, the lemma is the reading of this edition’s text. For emendations, the lemma is followed by the siglum of the edition from which the emendation is drawn and then by the rejected F1 reading and the sigla of the 17th-c. editions reading differently from the lemma. If no source is given for the emendation, the reading adopted is to be found in none of the folios. Doubtful and irregular readings are merely listed. ( | ) indicates that the reading is found in a full line; (?) indicates dubiety or an alternative to the reading adopted, although not a correct one in the judgment of the editor. In notes pertaining to variants in punctuation, a swung dash ( ~ ) shows that a word in the lemma is replaced in substantially the same form, and an inferior caret (‸) calls attention to a lack of punctuation.

Unadopted Conjectures

The Text

Authenticity

That Sh. was not the author of the entirety of WT was casually suggested by Pope (ed. 1725, 1:xx): I make no doubt to declare that those wretched plays [such as Pericles, added to the canon in F3], cannot be admitted as his. And I should conjecture of some of the others, (particularly Love’s Labour Lost, The Winter’s Tale, and Titus Andronicus) that only some characters, single scenes, or perhaps a few particular passages, were of his hand. Robertson (1930, pp. 133–4), a more dedicated disintegrator, also finds WT unworthy. Time’s prologue (lines 1579–1611) has quite an un-Shakespearean aspect. The I mentioned [1601], further, implies a previous prologue, which has been dropped. Perhaps the clearest ground for suspecting a non-Shakespearean hand is the rhyming of after with daughter, a thing unexampled in Shakespeare’s serious work, but emphatically of a kind of perverse rhyming much affected by Chapman. (For the rhyme, see n. 1606–7.) The chief æsthetic difficulty of the play, Leontes’s sudden jealousy, is not like [134] Shakespeare. . . . It would be a more satisfying solution if . . . the unnaturally rapid action had been imposed by a previous constructor, of whom we seem to find plain traces.

The critical problems Robertson discovers have concerned others (see n. 181–92 for Leontes’s jealousy, for example), but no one else attributes the apparent disparities to a previous constructor. WT is generally regarded as authentically Sh.’s. Some critics who accept this opinion believe, nevertheless, that the play as it stands is a revision of a version in which Hermione really dies. See, for example, Craig (Revisions, 1931, pp. 347–8) and, for a more recent expression of the idea, Mueller (1971).

The 1623 Version of The Winter’s Tale

The printing of the First Folio (1623), in which WT was originally published, has been analyzed by Willoughby (1932), by Shroeder (1956), and, most thoroughly and expertly, by Hinman (1963), on whose work much of this account is based. On 8 November 1623, close to the publication date of Mr. William Shakespeares Comedies, Histories, & Tragedies, sixteen of the plays were entered in the Stationers’ Register to Edward Blount and Isaac Jaggard (Arber, 1875–94, 4:69). Blount, a publisher and bookseller, was the leading member of the syndicate sponsoring publication of the First Folio. Isaac Jaggard, a printer, publisher, and bookseller, had replaced his late father, William, as a principal member of the syndicate. Wilson (1925) believes that because of William’s blindness and failing health in 1622–3, Isaac was the chief overseer of the printing of the collection. The entry covers soe manie of the said Copies as are not formerly entred to other men and lists, according to their order in the Folio, eight comedies, two histories, and six tragedies.

WT is the last of the comedies named in the entry; it is also the last play in the first section of F1 (the Comedies), where it occupies sigs. Aa1–Cc2 (pages 277–303); gathering Cc is a single sheet. The play is preceded by AWW (sigs. V1v–Y1v) and TN (Y2–Z6). The text of WT, which concludes in the top third of Cc2, is followed by The Names of the Actors. Sixteen characters are identified by name, and Other Lords, and Gentlemen, and Seruants. Shepheards, and Shephearddesses covers the rest (3370–88). Among the Comedies, the first, second, and fourth plays—Tmp., TGV, and MM—are equipped with similar dramatis personae, and all were probably typeset from manuscripts in the hand of the scribe Ralph Crane, of whom more below. He may have compiled these lists, although Eccles (ed. MM, 1980, p. 3, n. 2940) points out that in the case of a work by Webster, Crane is more likely to have copied than to have originated the list printed in [The Duchess of Malfi] in 1623 [from another Crane transcript], since it names . . . actors . . . who had died in 1614, and . . . in 1619. Wiv., the third play in the Comedies section and probably another Crane copy, lacks space for such a list.

Beneath The Names of the Actors in WT is the satyr tailpiece, in which a small defect acquired during the course of the F1 printing shows that the last page of Jn. (b5v) was printed before the last page of TN (Z6) and the last page of WT (Hinman, 1:179–80). Sig. Cc2v is blank. Since the play following WT in the Folio is Jn., the first of the history plays, the blank may seem to have been left so that the Histories section of the book could begin on a recto. This nicety, however, was actually compelled by the fact that Jn. and part of R2, which follows Jn., had been printed before the typesetting of WT began (Hinman, 1:37). It is more extraordinary that the first page of WT is preceded by blank Z6v. Hunter (1845, 1:417) suggests that there was some danger of losing this play. In the folio collection there is a blank page following Twelfth Night, as if there the collection of comedies ended, and the histories were about to begin: and my copy of the first folio actually wants the Winter’s Tale. Pollard (1909, p. 135) thought there had been simply a miscalculation of the space required for TN, but the anomaly seems to arise from another and more complicated cause.

As Hinman (2:521) explains—repeating to some extent the conclusions of Willoughby (1932, pp. 34–43)—when AWW was nearly completed, Compositor B, who had been working alone on that play, skipped to the Histories. With Compositor C he set all of Jn., which begins at sig. a1, and then two pages of R2, b6–6v. B at that point returned to the Comedies. By himself he finished AWW and TN, which concluded on Z6. He then set two more pages of R2 and, after an interruption for work on another book, with Compositor A set ten more pages of R2 and proceeded to WT. The blank Z6v is thus a legacy of the excursions from the Comedies into the Histories and back again, sheet Z having been printed before copy for WT was available. (Shroeder, p. 42, notices another minor consequence: In the first sheet of Jn. to be printed, sig. a3 is designated Aa3, in the style of the signature alphabet to be used later for WT.) Hinman continues: For some reason the copy for Twelfth Night was not readily available when quire X was finished (though it evidently became so soon afterward), and . . . the copy for The Winter’s Tale was in like manner unavailable when quire Z was finished (though on this occasion the want was made good even more quickly than before). . . . No difficulty over copyright can be supposed—only some short-lived trouble over the copy itself. He thus puts to rest several earlier speculations, such as that of White (ed. 1857, 5:275): It is possible that in gathering the plays together Heminge and Condell forgot this one [WT] until the folio was nearly in type; but it is more probable that, finding it no more tragical [i.e., less so] in its course or its catastrophe than Cymbeline, they first intended to class it with the Tragedies [as Cym. is], and after it was ready to be struck off restored it to its proper place among the Comedies. Equally groundless is the explanation of Furness (ed. 1898, p. vii) that inasmuch as the sheets were printed off . . . at different presses [he seems to be referring, incorrectly, to printing houses], it was undoubtedly easier to leave a whole page blank at the end of a signature than to transfer a single page of The Winter’s Tale to the press that was striking off Twelfth Night.

According to Hinman (2:496–503), the formes of quires Aa through Cc were set by Compositors A and B in the following sequence:

Howard-Hill explains his differentiation between the two Compositors A (pp. 86–7): Whereas the comedies compositor would quite often space a fair number of internal commas, occasionally more than were left without spaces, the practice of the [87] histories A is much more pronounced and never, in the plays . . . examined, is there a greater number of spaced commas to unspaced commas. . . . Also, before WT, the compositor A [of the Comedies] was indifferent to whether he set the first word of the speech together with the speech-prefix in a catchword, or the speech-prefix alone, but in the histories his invariable practice was to supply the first word of the dialogue with which the next page started. In WT, catchwords consisting of speech prefix and a word of dialogue are found on Aa3, Aa5v, Bb5v, and Cc1, all attributed to A, whereas abbreviated speech prefixes only are found on Aa4, Bb1v, Bb2, and Bb2v, all attributed to B. The Compositor A of WT thus appears to be the Compositor A of the Histories, not the Compositor A of the earlier Comedies. The latter was designated F by Howard-Hill (1973, p. 87). But Werstine (1984, p. 92), noticing that in the text assigned to Compositor F, portions of prose speeches that mark a change of address or of topic may be given a new line, a characteristic of Compositor D, wonders whether any distinction can be made between the two workmen. (The subject awaits further investigation.) Moreover, the validity of spacing as a compositorial discriminant has been questioned by McKenzie (1984). He found that its apparent testimony in early books printed at the Cambridge University Press does not accord with the work records kept.

The signatures of WT translate into line numbers as follows:

• Compositor A: (Aa3v) 623–754, (Aa3) 497–622, (Aa2v) 365–496, (Aa5) 1005–1136, (Aa2) 233–364, (Aa5v) 1137–1255, (Aa1v) 101–232, (Aa6) 1256–1387, (Aa1) 1–100, (Bb4v–Cc1v) 2411–3319
• Compositor B: (Aa4–4v) 755–1004, (Aa6v) 1388–1513, (Bb3v–4) 2147–410, (Bb3) 2016–146, (Bb2v) 1884–2015, (Bb2) 1759–1883, (Bb1v) 1635–1758, (Bb1) 1514–1634, (Cc2) 3320–69.

And they translate into sequentially ordered signatures and line numbers as follows:

• Compositor A: (Aa1–3v) 1–754, (Aa5–6) 1005–1387, (Bb4v–Cc1v) 2411–3319
• Compositor B: (Aa4–4v) 755–1004, (Aa6v–Bb4) 1388–2410, (Cc2) 3320–69.

About the order of composition there is little different from the regular procedure for typesetting F1. The manuscript text would be cast off—that is, before typesetting began, the words of the manuscript would be allotted to specific type pages—so that the two compositors could work more or less simultaneously. Both started with the innermost forme; and then one of them usually worked backward through the page sequence of each quire and the other forward. Since much of the text is regular verse, its casting off amounted to little more than counting the number of verse lines needed to fill the two columns of each page. Even on pages largely of prose (e.g., Bb1–2), white lines left above and below stage directions and scene heads permitted expansion or contraction, so that the typeset words could be fitted into the estimated space. Nothing shows what B did while A set Aa1v–2v, 5v–6 by himself.

The few irregularities are minor. On Aa1, the first page of the play but the last of the Aa gathering set by Compositor A, the white space preceding and following Scœna Secunda, near the foot of column a, is reduced, to allow plenty of room for the ornament, head title, and initial entrance direction at the top of the page and to avoid having the scene head or the initial SD stand alone at the foot of the column. At 467 (Aa2vb), perhaps to gain a couple of lines, an entrance direction, rather than being centered, is set on the same line as the conclusion of a speech—or, if the direction had been overlooked earlier, it could have been inserted there after the column was in type. SDs similarly placed are found at 2005 (Compositor B) and at 3029 (Compositor A). On Bb3a there are about two lines of white space above and below Autolycus’s song (2044–55), but on Bb3b his second song in the scene (2139–44) is crowded into the text: A one-line speech has been run into the concluding line of another speech (2138); the logotype yt is used (2146, its only appearance in WT, although ye is found at 1672); and probably line 2080 has been moved from the head of column b to the foot of column a, where it now occupies the direction line, normally blank in column a. Lines 2145–6 may have been reset. Pafford (1961, p. 173) suspects some mistake in the casting off or else some later insertion, perhaps the song. The latter seems likely, and if the song was overlooked, it may have been written on a piece of paper not a part of the main MS. Pafford suggests, too, that the typesetting of Bb3 was done by Compositor E, supposed to be an apprentice, but Hinman and Howard-Hill disagree. Pafford’s opinion, however, is shared by Cairncross (1972, p. 382).

Harrison (1948, p. 242) argues that in his later plays not only did Shakespeare abandon blank verse in conversation, but in the longer speeches he often substituted for the normal pattern of five feet . . . a much freer short line verse. This opinion is reiterated by Bertram (1981), who urges modern editors to forgo relining as iambic pentameter the apparently irregular verse of the early eds. In doing so, he says, they obscure Sh.’s rhetorical instructions to the actors, instructions that are embodied in the lineation of the early texts. Werstine (1984), investigating whether the line division in the Folio is Sh.’s or the F compositors’, finds that in the late plays, including WT, departures from iambic pentameter are usually caused by the compositors’ need to create or waste space to make cast-off copy fit its typographical allocation. The irregular lineation has nothing to do with rhetorical instructions.

The F1 Copy

According to Greg (1957, 3:1111), There is external evidence of trouble over the copy for this play [WT], for on 19 Aug. 1623 the Master of the Revels relicensed the piece since the allowed booke was missing[e] [see here], though by that time the play must have been already printed. And it probably was; Hinman (1:357) believes that the work on quires c and Aa–Cc may well have taken place in December 1622. (Earlier scholarship, summarized by Greg, 1955, p. 461, arrived at nearly the same date, late November.) Even though the manuscript of WT bearing the original license had been lost, however, at least one other copy existed to provide a basis both for the production with which the Master of the Revels, Sir Henry Herbert, was concerned and for F1 printer’s copy. Willoughby (in J. D. Wilson, ed. 1931, p. 113 n.) suggests that what Herbert actually saw was the printed sheets of WT. This is possible, of course; as Greg (1955, p. 417 n.) points out, though, they would have needed a lot of editing [presumably the addition of bookkeeper’s notes for control of the performance and perhaps the cutting of dialogue] and have provided very little room for it. Knowles (privately) suggests that the company may have retrieved from Jaggard the Crane transcript. That is also a possibility. Manuscript returned by modern letterpress printers may be too marked up and inky for theatrical use, but the copy for the 1591 English translation of Orlando Furioso, as described by Greg (1924), is quite neat. In that case, however, it may have made a difference that the manuscript was being received by Sir John Harington, the translator. Because the F1 text has no theatrical stigmata, there is no support for the opinion, voiced by Lee (1902, p. xxvi), that after being represented on the stage, the MS version of the play licensed by Herbert was sent to press.

In his TGV, Wilson (ed. 1921, p. 78) had argued that the copy for The Two Gentlemen was made up by stringing together players’ parts and arranging them in acts and scenes by the aid of a plot. A plot in this sense was an outline of the action of a play, recording entrances and exits, properties, noises, and other details of performance of which the prompter wanted to be reminded (for further information, see Greg, 1922). Wilson’s idea was quickly adapted to WT. Because, in his opinion, WT has no stage-directions, or very few, Rhodes (1922, pp. 59–60) asserts that the text cannot have been set up by the printer from the prompt-book. It was, instead, (p. 60) assembled from the players’ parts, complete with cue, dialogue, and certain directions, and . . . a detailed extract from the stage-directions to serve as his [the prompter’s] remembrancer—that is, the plot. The plot would furnish a guide for sorting the players’ parts and keeping them in order whilst pasting them together into one continuous text, which would, however, inevitably lack many directions. Rhodes finds that WT bears all the stigmata of an assembled text. The entrances of the players are not, as usual, distributed in the places where they are due to appear, but each scene is headed by a list of characters. . . . In all the five acts there are not more than a dozen incidental entrances and exits, and those are mostly of minor characters.

The same idea is expressed more fully by Rhodes (1923, pp. 98–100), and, after it receives favorable mention by Pollard (1923, p. 8), is reiterated by J. D. Wilson (1924, pp. 72–6). This view, reflecting Johnson’s (1756; 1968, 7:52) opinion that Sh.’s plays were printed . . . from compilations made by chance or by stealth out of the separate parts written for the theatre, was obviously influenced by TGV’s and WT’s massed entries (that is, the lists at the head of a scene not only of the characters who appear at its beginning but also of those who enter later). WT has massed entries at 2.1, 2.2, 2.3, 3.2, 3.3, 4.4, 5.1, and 5.3. The characters are listed according to social rank or importance in the drama in 2.1 (in two groups) and 5.3 but in the approximate order of their speaking in 2.2, 2.3, 3.2, 3.3, and 4.4. In 5.1, those present at the beginning are in order of rank; Florizel and Perdita, named last, enter later. Plays in the Ben Jonson folio (1616) also have massed entries, as do the quartos from which the folio texts derive. In Webster’s Duchess of Malfi (1623), according to Bald (1931–2, pp. 244–6), occasionally a character named in a massed entry fails to appear in the scene; and Greg (1931–2) reports that in Massinger’s autograph MS of Believe as you List, a massed entry includes characters who enter later. In these instances, players’ parts cannot have been involved unless, as in the case of BAYL, the point of entry was specified in another direction or could be deduced from another feature of the text. In any event, parts do not seem to have been involved in WT.

Although it was endorsed by J. D. Wilson in his 1931 ed. of WT (p. 122), the theory of assembled texts was to collapse under its impracticalities. (For a history of its career, see Greg, 1955, pp. 156–8.) Speculation about the fate of WT’s allowed book continued, however. Adams (1917, p. 25, n. 2) guesses that the MS was misplaced by the printer of F1. Such an eventuality would be possible if the allowed book was printer’s copy, for WT probably was printed in December 1622, and the entry in Herbert’s office book was dated 19 Aug. 1623 (see here). In the decade following Adams’s study, though, WT’s textual history began to be understood differently. Greg (1926, p. 154) found that the manuscripts of Fletcher and Massinger’s Barnavelt and Middleton’s The Witch (see here) were in the same handwriting, and F. P. Wilson (1926–7) proved that the hand was that of Ralph Crane, a scrivener associated with the King’s Men, who had signed his name to his transcript of Fletcher’s Demetrius and Enanthe; or, The Humorous Lieutenant (see here). To this signed document, Crane’s other surviving work, some also signed, is linked by distinctive and recurrent stylistic features. Wilson’s attribution of the Folio copy of WT to Crane was attacked unsuccessfuly by Tannenbaum (1933, pp. 75–86), who mistakenly thought the MS was quite possibly Shakespeare’s own manuscript.

F. P. Wilson (1926–7, pp. 211–14) describes some general characteristics of [Crane’s] work. 1. A publisher who came by one of Crane’s transcripts might reasonably expect from the printer an accurate text. . . . 2. All Crane’s transcripts are carefully divided into acts and scenes. . . . [212] 3. Those stage-directions in Barnavelt which are in Crane’s handwriting give little information apart from mere statements of exits and entrances [Fletcher and Massinger’s Barnavelt is preserved in British Library MS Add. 18653; see Fredson Bowers, ed., in The Dramatic Works in the Beaumont and Fletcher Canon, 8:485–501]. . . . In Crane’s other transcripts the directions never smack of the theatre. . . . [214] Statements of entrances are massed together at the head of each scene. Chambers (1930, 1:488–9) finds that the SDs in WT consist of little more than entries and exits, and the latter are often omitted. The entries are normally given in [4.3] and [5.2], but for the other scenes all the characters taking part, whether they are present from the beginning or not, are grouped in the initial entry. This follows the order of their appearance, and in [2.1, 3.2, 5.1, and 5.3], but not elsewhere, the successively appearing characters or [489] groups of characters are marked off by colons in the stage-directions. Howard-Hill (1972, p. 130) points out that the massed directions with colons were typeset by Compositor A, those without by Compositor B, who was apparently unwilling to print the colons from copy.

The entrance directions are of four types. The first, reflecting English tradition, marks entrances as the characters appear (4.3 and 5.2). The second, reflecting neoclassical tradition, includes some of the massed entries discussed earlier (here). In this type, all the characters appearing in the scene are named in the initial SD, although some actually enter later; their entrances are unmarked (2.1, 2.2, and 3.2; the SD for 3.2 omits Paulina, unless she is included among the Ladies, and the Servant who enters at 1323). The third is neutral: all characters who appear in the scene are present from the beginning (1.1, 1.2, 3.1, 4.1, 4.2, and 5.3). The fourth is hybrid: all who appear in the scene are named in the initial entry but some actually enter later, and their entrance is marked (2.3, 3.3 [except for the Shepherd who enters at 1501], 4.4 [except for the Servant who enters at 2145], and 5.1 [except for the Lord who enters at 2940]). Since the stage directions in all the Shn. texts thought to be printed from foul papers are in the English style and since there is no reason to think that Jaggard would have introduced the neoclassical, Crane must be its source and the source of the inconsistencies as well.

Only a few SDs add anything to the bare entrances and exits: as to her Triall (1174–5), pursued by a Beare (1500), the Chorus (1579), singing (1668 and 2043), Heere a Da(u)nce of (1988 and 2164), and like a Statue (3185). Of these, singing is common, of course. As to is unusual; it occurs in only two other Shn. SDs (Trumpets sound as to a charge [TNK 5.3.55.1]; Enter as to the Parliament [R2 4.1.0.1 (1921)] and there in F only, not in Q). WT’s pursued by is unique in Sh. Like is sometimes preceded by a participle—for example, habited like Shepheards (H8 1.4.63 [753]) or drest like Vincentio (Shr. 4.4.0 [2180])—but more often in Sh. it is not—Enter Ariel like a water-Nymph (Tmp. 1.2.316 [453–4]) and like a Harpey (3.3.52 [1583]), like Gentlemen (MM, DP [2946]), and like himselfe (Tim. 1.2.0 [341]). Similar SDs appear in AYL—like Forresters (2.1.0 [606]) and like Out-lawes (2.7.0 [972]); in Tit.—like a Cooke (5.3.25 [2525]); and in Cym.—like a poore Souldier (5.2.0.4 [2894]). These texts were printed from a variety of sources, according to Wells & Taylor (pp. 145–7), who represent a recent, although not universally shared, opinion: Tmp. from a Crane transcript possibly of foul papers, MM from a Crane transcript of what we now call a promptbook, Tim. from foul papers (Middleton’s and Sh.’s), AYL from a transcript or a promptbook, Tit. from foul papers, and Cym. from a transcript probably by Crane, of a manuscript in two hands. The SDs in WT obviously contribute nothing to our knowledge of Crane’s copy.

Punctuation marks, especially parentheses and apostrophes, provide another approach to the study of the F1 version of WT. Thorndike (1934), having assiduously counted the parentheses in the F1 texts thought not to have been printed from quartos, finds the greatest number in WT (369, 2 short of the number given by a later critic—see below, here), 2H4 (259), Wiv. (219), Cym. (158), and TGV (129). The printer’s copy for all but 2H4 is now believed, with more or less conviction, to have been Crane transcripts. Thorndike also notices that TGV, Wiv., and WT have collective entries of the type found in Crane’s (MS Malone 25) transcript of Thomas Middleton’s A Game at Chess. Pafford (1961, pp. 175–7) notes that, in WT, apostrophes mark—in addition to omitted vowels (as in ’d preterits)—parts of words (cam’t for came it, le’t for let it) and entire words before exclamations or appeals (’Beseech you) or in the middle of a phrase (Who taught ’this). The apostrophe is also used where apparently nothing has been dropped. . . . E.g., has—the verb, not a contraction of he has—is usually printed ha’s. . . . [176] The apostrophe was sometimes perhaps used vaguely, to indicate that writer or compositor realized that a word was abbreviated but did not know how to show it; e.g. Gillyvors, presumably for gillyflowers . . . is first spelt Gilly-vors [line 1890] and then Gilly’vors [1910]. In contrast there are cases where something is dropped but no apostrophe used. Instances include Ile (73, 95, 97, 121, etc.), th for the (1408, 1450, 2375), tane (1730), ha (1743), and fore (2221). Howard-Hill (1972, p. 129) also finds many possessive singulars of nouns with apostrophes after vowels [e.g., Camillo’s (792), Cytherea’s (1936)], it’s with the apostrophe [for its—e.g., 231, 236, 357, 1488], it for the possessive [e.g., 1110, 1279]. To the suggestion by Pafford (p. 176) that the colons punctuating many SDs might be traceable to Edward Knight, at one time bookkeeper to the King’s Men, Howard-Hill (1966) objects that Crane’s MS Malone 25 uses colons in massed entries to separate characters who enter later from those with whom the scene begins, whereas Knight did not use colons for this purpose.

The fundamental study of Crane’s work is Howard-Hill (1972), which superseded such earlier studies as that of Somer (1962). A summary of Howard-Hill’s and other research on Crane is provided by Haas (1989). Eight of Crane’s dramatic manuscripts survive, three being of Middleton’s A Game at Chess. Three additional MS copies of this play, one in Middleton’s hand, also exist, and comparison of Crane’s versions with Middleton’s affords an unusual opportunity to isolate many details attributable to the scribe rather than to the copy from which he worked. Eight nondramatic transcripts in Crane’s hand have also survived, and comparison of these and the dramatic transcripts permits more characteristics of Crane’s work to be identified. Those evident in WT include The Names of the Actors (3370–88) and the massed entrances mentioned above; division of the text into acts and scenes; descriptive stage directions (see here); some characteristic spellings (the very common ’em; see below for many more); the Jonsonian elision at 109 (Verely’is); numerous hyphens linking, for example, prepositions and objects, prefixes and stems, adverbs and adjectives, adjectives and substantives, and stems and suffixes (6, 17, 24, 30, 50, 73, 79, etc.); and, according to Howard-Hill (p. 82), huge quantities of colons and parentheses, 839 of the former and 371 of the latter. He points out (p. 87) that of the possible situations where parentheses [to enclose vocatives] could have been used, 39% of the vocatives in WT have them. Wells & Taylor (p. 601) add parentheses to mark passages spoken aside at 2516–17 and 2520. Crane may not be entirely responsible for the text’s numerous hyphens, however; McKenzie (1959, p. 81) finds that in setting MV Q2 (1619), Compositor B hyphenated such words as bed-fellow and me-thinks 25 times and deleted the hyphen in his copy only 9 times. (While John F. Andrews found that me-thinks was hyphenated only by Compositor B’s fellow-workman on MV [1619] and the other Pavier quartos [The Pavier Quartos of 1619—Evidence for Two Compositors (Vanderbilt diss., 1970)], nonetheless it remains true that Compositor B did introduce hyphens into many words he set in these quartos. There is reason to doubt, however, that Compositor B worked on the Pavier quartos: see R. Knowles, SB 35 [1982], 202.)

A great many of Crane’s characteristic spellings appear in WT, although, as one might expect, there are some exceptions. Howard-Hill’s summary (pp. 64–8) includes, but is not limited to, the following:

• -ei-, never -ie- (all such spellings in WT accord with modern convention—e.g., perceiue [263] and Heire [1315])
• -lly, never -ly (the same—e.g., naturally [2595])
• -ing, never -eing (the same—e.g., mouing [431])
• -nck (e.g., prancks [2583] and wrinckled [3217], although these are far outnumbered by such -nk forms as ranke [369] and Winke [414])
• -ll, never -l (Royall [28], wooll [1703], but wil [453] and shal [1304], both full lines)
• -s, never -es, for plurals of nouns with short vowels (e.g., Gifts [31], things [2939])
• -es, never -s, for plurals of nouns ending in -th/sh/ch (Oathes [85], blushes [1872], Wenches [2136])
• -ings, never inges, for words ending in -ing (Fadings [2020] but singes [2008], a full line; the spelling is not found elsewhere)
• -s or -sse, never -ss (e.g., killes [1750], Presse [74], kisse [3284], but also Princess [2832])
• b is not doubled before -’d or -ed (e.g., rob’d [past tense of rob; 1729], but crabbed [173]); it is doubled before -ing after short vowels (e.g., stabbing [2607])
• d is not doubled before -s or -es (but toddes [1701], Goddes [1826], addes [1902], oddes [2976]), before -ing (but bidding [733, 1100, and 1143]); or before -er (e.g., hinder [79])
• g is not doubled before -ing, -er, -es, but is probably doubled after an unvoiced short vowel (e.g., Egges [206], Dagger [235], pugging [1675])
• Doubled consonants are retained before -ed, but one -l may be dropped before -d (e.g., crabbed [173], spotted [427], Added [793], muzzel’d [235], and Il’d [462]). An exception to the doubling of l before -ed is (vn)setled (this word only) (224, 424, 1767, 2385, and 3272)
• m is doubled medially only after short vowels (e.g., (be)comming [8 and 1464], Commission [97], and command [1665]) but not before -es or -s (e.g., lames [3066] and redeemes [3311], plurals in which the vowels are long)
• n is doubled medially after short vowels (e.g., Sonne(s) [46 and 3136], manner [3093], and winners [3344]) but not after long vowels or before -s (e.g., finer [312] and begins [469])
• p is generally not doubled medially, but there are exceptions (e.g., Coppy [197], slipperie [365], Lippe [479], Appollo’s [801])
• r is doubled after short vowels (e.g., Starre [50], morrow [61], and Iarre [100])
• s is doubled after short vowels (e.g., Hostesse [121], Kisse [165 and 3284], and Glasse [402])
• t is doubled after short vowels and before -ing (e.g., pitty [670] and committing [1189])
• w is always single
• z is always single (reversed in WT: muzzel’d [235] and Chizzell [3279]).

Other Crane preferences are for w in such words as sowr’d (173), Lowt (397), scowre (631), lowd’st (865), Perswades (868), powre (927 and 1586), and howre (1587); for internal a in roab’d (1830), Coarse (1944, 1946), poaking (2052), and coap’st (2267); for -que in Basilisque (496), publique (815), and Heretique (1042); for internal -y- in prayse (18), trayn’d (25), and tyre (66); for blood (20 times; no other spelling occurs). He was elsewhere indifferent to the termination ance/aunce, but in WT aunce is found only in daunce(s) (183 and 1988), whereas dance occurs eight times, once at 1991—two lines from Daunce. Among spellings preferred by Crane and not likely to have been altered to justify lines of type, Howard-Hill (pp. 100–2) lists, for WT, Councels (326; plural noun), Physick (282), powrefull (284, 622), and wayting (163).

Those that also may signify Crane are councell (singular nouns: 2777, 2786), extreames (1804), flowre(s) (1879, 1886, 1889, 1916, 1919, 1927, 1931, 1941), graunt (187, 2994; but also grant [1616]), howre (1587), moneth(s) (98, 173, 2089, 2871, 2998), powr’d (2172), and publique (815). And to these Howard-Hill adds (pp. 171–2) Crane spellings adopted by Compositor B against his preference: Angell (2034), answere (2023), approach (1465, 1857, 2036), choice (2137, 2257), daylie (1434 as dayly), deed (797, 1571), extreames (1804), forth (837, 979, 1606), happie (2178), howre (1587), mightie (921, 1725), moneth (2089), need (1722, 2256; neede, 6 times), old (1860, 2089, 2179, 2263, 3328, 3345; olde, twice), powre (927, 1586, 1839), son (1638, 3363; sonne, 39 times), sun (1717, 1918), wee’ll (2135; wee’l 5 times), yeere(s) (1585, 1617, 1671; yeare(s) [1399, 1887]).

Spellings in Crane’s transcripts that Eccles (ed. MM, 1980) discovered in MM also appear in WT: beleeue (12 times), Coyne (2607), deere (10 times; deare 11 times), (vn)easie (118, 1661, 2662, 3298), medler (2143), meere (901, 1322, 1326), mistris (6 times; mistresse, 7 times), neere (8 times), peece (11 times; pieces once), practise/practis’d (189, 1352), sence (8 times), thether (1662, 2125; thither, 3 times), vertue (7 times), and yong (5 times; young, 9 times). Of all these spellings, Kable (1968, pp. 157–9) found extreames (1804), howre (1587), and moneth (2089) to be contrary to Compositor B’s preference, hence his copy spellings. Kable’s opinion, however, is compromised by R. Knowles’s finding that the work was not Compositor B’s (see here).

The two plays also display some similarities in punctuation. In WT as in MM, colons are preferred to semicolons, although more strongly (1 per 29 words, as opposed to 1 per 140 in MM). The exclamation point does not appear. Sometimes the question mark is exclamatory (at 265, for example), but in many instances its exclamatory function cannot be distinguished clearly from its interrogative one. One difference from MM and the transcripts is that the suffix is nearly always -nesse in WT; only one -nes is found, in 1663, a full line.

Howard-Hill (pp. 66–8) separately categorizes Crane’s spellings of words that have been used to identify the stints of Jaggard’s compositors. His tabulation may be compared with the occurrence of the spellings in WT (see table). Ignoring counts

Crane’s Copy

Because Crane exerted such a strong influence on the form of the texts he transcribed, it is not surprising that critics find it difficult to penetrate his version of WT as it is represented in F1. Greg (1955, pp. 416–17) believes that WT was late reaching the printer only because of the time required to make for Jaggard a transcript of the foul papers, an opinion with which Evans (ed. 1974, p. 1604) agrees. Howard-Hill (1965, p. 340), having protested against too facile an acceptance of numerous parentheses as indicative of Crane’s transcription, suggests, nevertheless, that the number in WT may indicate that the play was printed from a Crane transcript of his own earlier transcript, an idea developed further by Howard-Hill a year later (1966, p. 140). When, according to Howard-Hill, copy was being gathered for the Folio, Crane was given foul papers for Winter’s Tale with instructions to prepare also the promptbook for playhouse use. Crane prepared the promptbook first before making the transcript for the printer for by so doing he would release the foul papers in the shortest possible time to return to the security of the players’ collection of manuscripts. Had he prepared the Folio copy before the promptbook, we should not have the clean and sophisticated text, showing an unusually large number of parentheses, that we find in the Folio. On the same evidence, Crane apparently retained the promptbook, but probably returned the foul papers, so that he would avoid having to make both transcripts from foul papers. . . . He chose, therefore, to delay delivery of the Folio copy, perhaps thinking, or indeed knowing, that the printers had enough to go on with. [Howard-Hill (1992, p. 128) conjectures that the delay arose from Crane’s occupation with his transcript of Webster’s Duchess of Malfi.] From this order of events, we get a good, clean, literary text, with an unusually high number of parentheses consistent with other Crane transcripts from his own earlier papers, and, of course, a delay which affected the printing of Winter’s Tale in the First Folio. In his major work on Crane, Howard-Hill (1972, p. 131) adds, The massed entries [appearing in F1 and deriving from the printer’s copy] would have been out of place in a copy made for a promptbook but [they] easily could have been compiled from a transcript with conventional entries. Why he [Crane] should choose massed entries when he had a clean manuscript available is not readily explained, but the variety of the directions argues that the copy was prepared in some haste, under conditions which did not allow the scribe to adopt the massed convention completely.

The idea that the WT copy was a transcript of a transcript had appeared earlier in Howard-Hill (1965, pp. 337–8). The number of WT’s parentheses (371, 268), the article indicates, is substantially greater than that in the other Folio Comedies believed to have been set from Crane manuscripts—Tmp. 98 (115), TGV 128 (150), Wiv. 218 (241), and MM 75 (78); the first figure is the actual incidence and the second, the actual adjusted for the length of the plays. Since the Folio compositors did not have an identical fondness for parentheses, the figures adjusted for length become, when further adjusted to allow for Compositor A’s preference, Tmp. 116, TGV 131, Wiv. 308, and MM 62, in contrast to WT’s 415. Moreover, Crane’s several transcripts of A Game at Chess indicate that every time Crane recopied a text, he tended to reproduce the parentheses of his copy, but added to them, and . . . added more parentheses when he was transcribing from his own copy (p. 336). The WT transcript, Howard-Hill (1972, p. 70) observes, must have been made after any other transcripts he made for the Folio. Therefore the scribe’s habits would have been influenced both by the character of the copy before him, and by his increasing familiarity with the kind of material he had to work from.

A question about this theory arises from the massed entries. If they were undesirable in a promptbook, as Howard-Hill says, and if the WT promptbook was created before the printer’s copy, would Crane, having copied in the first transcript the entrances where the play’s action required them, have then taken the trouble to extract and compile these entries for each of the play’s eight scenes they head? The compilation is incomplete, moreover (4.3 and 5.2, which might have been massed, are not), and the style of the massed entries differs (see here). Crane’s motive would have been professional pride or a classical preference (massed stage directions serve no essential literary or dramatic purpose), which he indulged even though, as noted above, the copy was prepared in some haste, under conditions which did not allow the scribe to adopt the massed convention completely (Howard-Hill, 1972, p. 131). In response, Wells & Taylor (p. 601) state that as the Folio had been printed [WT in Dec. 1622] long before Herbert saw the new prompt-book [on 19 Aug. 1623], this [Howard-Hill’s explanation] seems unlikely. The original prompt-book might even have been lost as a result of being copied by Crane for the Folio. They also allude to Thorndike’s idea that the dance of satyrs at 2164 was taken from Jonson’s Oberon (see n. 2164): The passage introducing this dance could be omitted without disturbing the dialogue; no one comments upon the dance afterwards; moreover, the Clown’s comment that My Father, and the Gent. are in sad talke [2134–5] would be naturally followed, after the exit of Autolycus and his clients, by Polixenes’ O Father, you’l know more of that heereafter [2165], which indicates that they have been carrying on a conversation which we have not heard. Polixenes’ comment is not nearly so natural after the satyr dance, since it suggests that he had been talking to the Old Shepherd rather than attending to the dance he had himself insisted upon witnessing [2156–7]. There is no reason to doubt Shakespeare’s authorship of the passage introducing the dance, but it could be a late addition; if so, Crane was copying a prompt-book, and the original composition was earlier than January 1611. (see here.) Regarding the supposed unnaturalness of 2165, however, not one of the acting editions collated for this edition cuts it (see here). Since there are no other traces of Crane’s having copied WT’s promptbook, the apparent anomaly probably arises from an aberration in foul papers, toward which the clothing crux may also point (see n. 2557–8 and here).

The foul papers of one of Sh.’s compositions—147 lines of Sir Thomas More, a play of uncertain date originally by Anthony Munday and Henry Chettle—may have survived. Having been heavily censored by the Master of the Revels for political reasons, the play was parceled out to several playwrights for salvage. The 147 lines are indeed foul—punctuation lacking, SDs absent or incomplete, SPs mistaken or vague, words omitted, and verse mislined—but they have perhaps created too strong an impression of Sh.’s scribal carelessness. It may be true that Jaggard’s initial editorial policy called for scribal transcripts to be given to the printer if quartos were not available (Eccles, ed. MM, 1980, p. 293). Nevertheless, recent critical opinion, as represented by Wells & Taylor (pp. 145–7), is that foul papers more or less certainly were the initial printer’s copy for as many as 17 of the 35 plays included in F1. If this opinion is correct, the foul papers of WT could have served as Crane’s copy as well.

Crane’s Reliability

In about 1625, Crane transcribed Thomas Middleton’s The Witch for a presentation by the author (see Greg, 1941–2). Three of the songs in this transcript are preserved in other versions. In 1625, Crane made a private transcript of Fletcher’s Demetrius and Enanthe, or the Humorous Lieutenant, which may be compared with the independent text of the play published in the Beaumont and Fletcher folio of 1647, a version representing the play as cut for production (see Cyrus Hoy, ed., in The Dramatic Works in the Beaumont and Fletcher Canon, 5:293). Nosworthy (1965, p. 221), having examined these sources, concludes: It is clear that Crane, though an elegant scribe, was at times an extraordinarily perfunctory one. Carelessness, combined with strange orthography, occasionally results in misreadings which would have baffled any compositor. . . . That he was, like many another scribe, occasionally guilty of sins of omission is a ready inference. . . . [Sh.] Folio texts based on Crane transcripts must obviously be viewed with suspicion. He was doubtless responsible for the sprinkling of apparent nonce-words which defy emendation, and there are often grounds for suspecting small omissions. Howard-Hill (1972, p. 133) differs, however: Even when he [Crane] may be suspected of error, the reading of his transcript is at the least plausible. If his sophistication of the texts he transcribed had been less, more could be discovered of the nature of his copy. The goodness of . . . WT means little more than that the printer’s copy was free from obvious error. The general level of Crane’s accuracy was high, but he was not reluctant to interfere with his text, consciously or unconsciously, when its meaning was obscure to him. That the F1 text of WT is not considered word-perfect is evident from the long history of its emendation recorded in this edition’s textual notes; for a contrast of the repairs made by two recent eds., see here below.

The Printer’s Reliability

There is almost no evidence of proofreading. As Pafford (1961, p. 178 n.) notes, Some copies of F1 may have a Bears. [at 1500], but apparently in most it is a Beare. This possible press variant is not mentioned in his ed. 1963, however. Hinman (1:264) reports that page number 281 (sig. Aa3) is 285 in one copy (Folger 24) and that in about a third of the more than fifty copies he collated, a mark appears between the first e and the long s of these in 1880. One cannot be at all exact in judging the accuracy of the compositors, for their errors may have been corrected in an earlier stage of proof than the one that has survived, or they may have faithfully transmitted incorrect readings introduced by Crane. Nevertheless, it is interesting to observe that in a conservatively edited version of the play and a more radically edited version, the rate of error is similar. Evans (ed. 1974) makes twenty substantive emendations:

• Compositor A: 175, 237, 290, 368, 1185, 1207, 2617, 2739–40, 2798, 2818–19, 3206
• Compositor B: 945, 1559, 1678, 1810, 1910, 2262, 2272, 2316, 3363

The Oxford editors (1986) emend in nineteen of these instances (2818–19 excepted) and nineteen more:

• Compositor A: 11, 213, 1353, 2464, 2488, 2530–1, 2588, 2751, 2795, 2821, 3191, 3302
• Compositor B: 896, 1617, 1811, 1965, 1980, 2184, 2266

Subsequent Early Editions

The later history of WT in the 17th c. is told primarily by the entries in the Stationers’ Register, quoted and annotated by Greg (1957, 3:1113–21). After the publication of F1, Isaac Jaggard’s widow transferred her parte in Shackspheere playes to Thomas and Richard Cotes on 19(?) June 1627, and on 16 Nov. 1630 Edward Blount transferred his right to sixteen plays, including WT, to Robert Allot. The publication of F2 (1632) ensued; the work was printed by Thomas Cotes for Allot and four other stationers. This edition was twice reissued. Because Allot’s widow was about to marry Philip Chetwind—who was a clothworker rather than a stationer—she was forced to give up her copies (on 7 November 1636; the Stationers’ Register entry is dated 1 July 1637). Chetwind recovered the copyrights, however, and became the publisher of the two issues of F3 (1663–4), in the manufacture of which three printers participated. Yet the Stationers’ Company continued to regard the copyrights as the property of Richard Cotes, and on 6 August 1674 transferred them (including the right to WT) to John Martin and Henry Herringman; these rights were again transferred by Martin’s widow to Robert Scott, on 21 August 1683. F4 followed in 1685, its three sections (Comedies, Histories, and Tragedies) evidently having been printed simultaneously by three printers. Textual changes made in WT in the three derivative folios and in the principal later versions may be found in the notes of this edition. None of these changes has independent textual authority.

The Date of Composition

External Evidence

Early attempts to date WT were influenced by the opinion that so irregular a play must have been the work of an inexperienced playwright and by the entry dated 22 May 1594 in the Stationers’ Register of a booke entituled a Wynters nightes pastime (Arber, 1875–94, 2:650), which was taken to be WT. Ulrici (1839; trans. 1846, p. 269) wrote, for example, that this is probably the same drama as we now have, which, upon its revision, received a name more suited to its altered form. The equation was not unreasonable, for the Accounts of the Revels at Court record a performance by the King’s Men on 5 November 1611 of A play called ye winters nightes Tayle (see Cunningham, 1842, p. 210, and Streitberger, 1986, p. 48); Chambers (1923, 4:125) identifies this later wintry amusement as WT.

Overly imaginative discoveries of allusions also interfered. Walpole (1768, 2:114–16) thought that WT was certainly intended (in compliment to queen Elizabeth) as an indirect apology for her mother Anne Boleyn. . . . [115] The unreasonable jealousy of Leontes, and his violent conduct in consequence, form a true portrait of Henry the Eighth, who generally made the law the engine of his boisterous passions. . . . Several passages . . . touch the real history nearer than the fable. Hermione on her trial says, [For Honor, ’Tis a deriuatiue from me to mine, And onely that I stand for (1217–19)]. This seems to be taken from the very letter of Anne Boleyn to the king before her execution, where she pleads for the infant princess his daughter. Adding several similar details, Walpole concludes (116): The Winter Evening’s Tale was therefore in reality a second part of Henry the Eighth. For the compliment to have point, Queen Elizabeth must have been alive to receive it; hence, in Walpole’s view, WT preceded her demise, in 1603. The letter to which Walpole alludes is found in The Harleian Miscellany (1808, 1:201–2). According to DNB (s.v. Anne), the letter is a manifest fabrication of the time of Queen Elizabeth; nevertheless, it speaks with pathetic dignity of the foul blots on the most dutiful wife and on the infant princess. A modernized text is in Black (1933, pp. 46–7).

Far-fetched notions abound. Capell (1783, 2.4:176) finds several absurd reasons to think that at the time he wrote WT, Sh. had his mind on a country matter, his retirement (e.g., the mention of th’Grange, or Mill at 2126). The play is a writing for Stratford, or a writing at it. Capell dates the play 1613, after H8 and before Tmp. (1614). Ironically, Capell is nearer current opinion than the more judicious Malone (in Steevens, ed. 1778, 1:285). Influenced by the entry of A Winter Night’s Pastime, which might have been the same play, Malone assigns WT to 1594, although his respect for Walpole, the silence of Francis Meres’s Palladis Tamia: Wit’s Treasury (1598), which mentions a dozen other Shn. titles, and the paucity of couplets in WT (characteristic of Sh.’s early style) make him doubt whether it ought not to be ascribed to the year 1601, or 1602. By the time of his own edition of WT (Malone, ed. 1790), he was convinced that A Winter Night’s Pastime was not WT, and its removal from consideration made Walpole’s conjecture extremely plausible (1.1:350). Meanwhile, Blackstone (in Steevens, ed. 1785) discovered in the lines If I could find example Of thousand’s that had struck anoynted Kings, And flourish’d after, Il’d not do’t: But since Nor Brasse, nor Stone, nor Parchment beares not one, Let Villanie it selfe forswear’t (460–4) an allusion to the death of the queen of Scots. The play therefore was written in king James’s time. His argument is that an allusion to Mary, Queen of Scots, King James’s mother, could never have been made before Queen Elizabeth’s death, for Queen Elizabeth, however reluctantly, had consented to Queen Mary’s execution. Trapped, Malone (1.1:351) attempted to have it both ways: Sh. lay’d the scheme of the play in the very year in which the queen died [1603], and finished it in the next. He discovers, however, in the Stationers’ Register, 2 April 1604, the entry of The Strange Report of a Monstruous Fishe (see n. 2097–103), to which, he believes, Autolycus alludes, and he also finds the Puritan who sings psalms to hornpipes (1712–13) a corroborating detail, because (1.1:352) the precise manners of the puritans was at this time much ridiculed by protestants. As for style, the meter is less easy and flowing than is usual in Sh.’s plays and the phraseology more involved and parenthetical. . . . In this harshness of diction and involution of sentences it [WT] strongly resembles Tro. and H8. The latter play is now dated 1612–13, not long after WT, but Tro. some ten years earlier. Nevertheless, Malone knew that Jonson had alluded to WT, as well as to Tmp., in the Induction to Bartholomew Fair, first produced in 1614 (see n. 0), and so again changed his mind (1.2:286): Jonson joined these plays in the same censure, in consequence of their having been produced at no great distance of time from each other; and . . . The Winter’s Tale ought to have been ascribed to the year 1613.

Hurdis (1792, pp. 22–3), is unconvinced: The faults of its [WT’s] metre and its language are so numerous, that it must be ranked with [Ant., H8, Cor., and Cym.], . . . the earliest efforts of our poet’s muse. The anointed kings passage, Hurdis believes (p. 23), was inserted later. The compliment he has paid to the Queen in the fable of the play . . . affords a strong proof that it was written during her life-time. For it is not likely that he [Sh.] would endeavour to exculpate Anne Bullen in the reign of James. Chalmers (1799, pp. 396–401)—whom F. W. Clarke (in Furnivall, ed. 1908, p. ix) calls the Sir Politick-Would-Be of Shakespearean criticism—observes several historical allusions in WT but not Blackstone’s, for, Chalmers says, Blackstone’s mind was not very amply stored with historical knowledge of the Elizabethan period. He believes that the lines Blackstone cited—If I could find example Of thousand’s that had struck anoynted Kings (460–1)—reflect the public prayers offered for the queen after the failure of Essex’s rebellion; an allusion such as Heire-lesse it hath made my Kingdome (2737) would have been inappropriate in King James’s time, for he had heirs. Thus, Chalmers concludes, WT was written in 1601. Also rejecting Blackstone and his use of the execution of the Queen of Scots as evidence, Douce (1807, 1:347) notes: The perpetrator of that atrocious murder did flourish many years afterwards. Douce therefore thinks the allusion in lines 460–4 is to King James’s escape from the Gowrie conspiracy (1582), an event often brought toi the people’s recollection during his reign.

Opposition to such fanciful guessing was bound to arise. Boswell (ed. 1821, 14:234–5), for example: I confess I am very sceptical as to these supposed allusions by [235] Shakspeare to the history of his own time. If the plots of his plays had been of his own invention, he might possibly have framed them with a view of that kind; but this was unquestionably not the case with the play before us; and if any one had intended a courtly defence of Queen Elizabeth’s mother, it must have been Greene, and not Shakspeare. Garinter, the Mamillius of our poet, dies under the same circumstances, in the novel [see here]; nor is it, as Mr. Walpole seemed to suppose, an unnecessary incident, because it fulfils the declaration of the oracle, that if the child which was lost could not be found, the king would die without an heir [1315–16]. To say that a child resembles her father is surely not so uncommon a remark as to make it evident that it had reference to a particular individual; nor is there any thing very courtly or complimentary in Paulina’s angry allusion to the old proverb.

Moreover, Malone had already found the clue needed to begin working through the early speculations, although he did not immediately recognize its value. In his Historical Account of the Rise and Progress of the English Stage (ed. 1790, 1.2:1–284), Malone first makes use of a document that has since disappeared, the office book of Sir Henry Herbert, Master of the Revels to King James. As Chambers (1930, 2:347) explains, the Office Book of Sir Henry Herbert, Master of the Revels, . . . is now lost, but in 1790 was in the house of Francis Ingram at Ribbesford, Worcestershire, which had belonged to Herbert. Malone (1.2:226) quotes Herbert’s memorandum: For the king’s players. An olde playe called Winters Tale, formerly allowed of by Sir George Bucke, and likewyse by mee on Mr. Hemmings his worde that there was nothing prophane added or reformed, thogh the allowed booke was missinge; and therefore I returned itt without a fee, this 19 of August, 1623. Book in this context is probably the technical term for the prompt copy of the play on which the license to act was inscribed (see Greg, 1931, 1:192–3), although Pollard (1920, p. 67) had thought that the term possibly [referred to] the original manuscript. Bucke, or Buc, was Master of the Revels from 1610 to 1622; Hemmings, or Heminge, was an actor with, and the business manager of, the King’s Men.

George Buc had served as deputy Master from 1603, however, and Chalmers (1799, pp. 200 ff.) prints extracts from the Stationers’ Registers showing Buc as licenser for the printing of plays as early as 1606. Malone may or may not be right that Buc did not license plays for performance before he became Master (see Eccles, 1933, p. 458). Thus critics like Rolfe (ed. 1879, p. 10 n.), who asserts that the Stationers’ Registers show . . . that he [Buc] had practically the control of the office from the year 1607, are saying more than the entries in the Registers prove. The same applies to critics like Wickham (1973, p. 96 n.); they maintain that until his official installation as Master, Buc licensed for printing only. Albright (1927, p. 246), incidentally, asserts that WT was relicensed to prevent its being taken over by another company because it was in print, but Greg (1928, p. 96) points out the obvious—that Buc acted not for Albright’s reason but because the authorised copy had been lost. That copy was apparently on hand when the play was acted at court on 7 April 1618 and perhaps in 1619 as well (see here); as Baldwin (Division, 1965, p. 51) notes, its absence may have been discovered when the company began to prepare for the court performance of 18 January 1624 (see here). Greg (1954, p. 150): The Winter’s Tale is peculiar in that it may have been a late addition to the folio, and that a new prompt-book was licensed by Herbert in August 1623. . . . These facts are probably related, but it is not clear that they have any bearing on the nature of the text.

Drake (1817, 1:504–5), convinced that Tmp. was written towards the close of 1611, argues (1:497) that WT was written towards the close of 1610 and was licensed and performed during the succeeding year. The order of the two plays is assumed to be that in which they were named by Jonson. Without mentioning Drake, Malone (in Boswell, ed. 1821, 2:463) states: I . . . suppose The Winter’s Tale to have been originally licensed by him [Buc] in the latter part of that year [1610] or the beginning of the next. He therefore assigns the play to 1611. Although Dutton (1991, p. 151) finds no cogent evidence that Buc was involved . . . in the censoring of plays for performance, prior to Tilney’s death, critics by and large have accepted Malone’s terminus a quo. According to Halliwell (ed. 1859, 8:40), for example, In the absence of any direct evidence to the contrary, it seems . . . unnecessary to suggest that the Winter’s Tale may have been one of the dramas that passed under Buck’s review during the tenancy of Tylney in the office; and it may fairly, at present, be taken for granted that the comedy was not produced until after the month of August, 1610. Hunter (1845, 1:416) believes that prior to 1610, Buc licensed both for performance and for printing; on the basis of a supposed affinity of WT with TN, he dates WT not later than 1601 or 1602—or, if licensing by Buc must be taken into account, 1606, the year after the Gunpowder Plot, because of the anointed kings passage (see n. 460–4). As for the absence of the allowed copy, Malone suggests that it was destroyed in the Globe fire of 1613, but Chambers (1930, 1:488) disagrees, pointing out that there was a performance in 1618 and probably another about 1619. The former is attested by Cunningham (1842, p. xlv): To John Heminges &c upon a warrant dated 20 April 1618 for presenting two severall Playes before his Maty. on Easter Monday Twelfte night the play soe called and on Easter Tuesday the Winter’s Tale xxli. The latter is inferred from the appearance of The Winters Tale on a piece of waste paper from the Revels Office (see Marcham, 1925, pp. 7, 13).

Important new evidence of WT’s date came to light with the discovery of a passage in Simon Forman’s Bocke of Plaies (Bodleian Library MS Ashmole 208, fols. 201v–202). As Pafford (Forman, 1959, pp. 289–90) remarks, W. H. Black began to catalogue the Ashmole manuscripts in 1830 or 1831, and there is a note by Black on a proof-sheet of the catalogue against [290] the entry of the Bocke of Plaies which reads I made a transcript of this curious article, in 1832, for my friend J. P. Collier. The so-called Book of Plays consists of memoranda made in 1611 by Forman after seeing four plays at the Globe—Macbeth, Cymbeline, Richard II (not Sh.’s), and WT. Of WT he wrote: IN the Winters Talle at the glob 1611 the 15 of maye ☿ [Wednesday]

Obserue ther howe Lyontes the kinge of Cicillia was overcom wth Ielosy of his wife with the kinge of Bohemia his frind that came to see him. and howe he Contriued his death and wold haue had his cup berer to haue poisoned. who gaue the king of bohemia warning therof & fled with him to bohemia/Remēber also howe he sent to the Orakell of appollo & the Aunswer of apollo. that she was giltles. and that the king was Ielouse &c and howe Except the child was found Again that was loste the kinge should die wthout yssue. for the child was caried into bohemia & ther laid in a forrest & brought vp by a sheppard And the kinge of bohemiā his sonn maried that wentch & howe they fled into Cicillia to Leontes. and the sheppard hauing showed the letter of the noble man by whom Leontes sent a was [away? it was?] that child and the Iewells found about her. she was knowen to be leontes daughter and was then 16 yers old

Remember also the Rog that cam in all tottered like coll pixci /. and howe he feyned him sicke & to haue bin Robbed of all that he had and howe he cosoned the por man of all his money. and after cam to the shep sher with a pedlers packe & ther cosoned them Again of all ther money And howe he changed apparrell wth the kinge of bomia his sonn. and then howe he turned Courtiar &c / beware of trustinge feined beggars or fawninge fellouse.

Forman’s notes were first published by Collier (1836, p. 20), in a somewhat modernized version. As Collier’s forgeries and impostures were revealed, Forman’s Book of Plays also fell under suspicion, but W. H. Black’s note proves it genuine. Further authentication became available with the publication by Cunningham (1842) of extracts from the Revels Accounts, in which the performance of WT on 5 November 1611, mentioned above, is recorded. Collier (ed. 1842, 3:425–6), however: The Tempest and The Winter’s Tale were both acted at Whitehall, and included in Sir George Buc’s account of the expenses of the Revels from October, 1611, to October, 1612. How much older The Tempest might be than The Winter’s Tale, we have no means of determining; but there is a circumstance which shows that the composition of The Tempest was anterior to that of The Winter’s Tale. . . . [426] There is . . . one remarkable variation [between Pandosto and WT]; in the former the infant Fawnia is put into a boat [to be carried into the midst of the sea, and there left to the wind and wave]. Shakespeare . . . describes the way in which the infant [Perdita] was exposed very differently, and probably for this reason:—that in The Tempest he had previously (perhaps not long before) represented Prospero and Miranda turned adrift at sea in the same manner [as Fawnia]. When, therefore, Shakespeare came to write The Winter’s Tale . . . he varied from the original narrative, in order to avoid an objectionable similarity of incident in his two dramas. Although Collier may be mistaken that Tmp. is the earlier play, his opinion carried some weight; it is quoted with approval later, by Hudson (ed. 1852, 4:6), for example, and it reappears in Muir (1957, p. 243): Possibly the first version of The Tempest had been written before The Winter’s Tale, so that Shakespeare could not easily repeat the incident of the babe adrift in a boat. That there was more than one version of Tmp. is a supposition, however, that disappears from Muir (1977). Nevertheless, it is possible that Tmp. antedates WT. According to Orgel (ed. Tmp., 1987, pp. 63–4), There is . . . not . . . any way of determining chronological priority between The Tempest and The Winter’s Tale. . . . [64] The most we can say is that the evidence supports a date of late 1610 to mid-1611 [for Tmp.], and that Shakespeare was writing the play just after, or just before, or at the same time as The Winter’s Tale. Halliwell (ed. 1859, 8:44): With equal probability, it might be conjectured that Shakespeare, having omitted the incident in the construction of the Winter’s Tale, introduced it in the Tempest as one especially suited to a romantic drama of that description.

Another approach is represented by Procter (1844, 1:x–xi), who finds WT too badly written to be a late play: As a general principle, . . . I would say, that the plays in which signs of imitation (particularly imitation of style) are manifest, should be accounted the earliest; and that those wherein the poetry is redundant and far exceeds the necessities and purposes of the story, should be held to have preceded, in point of time, the great and substantial dramas, in which the business of the play is skilfully wrought out, and where the poetry springs out of the passion or humour of the characters, and serves to illustrate and not to oppress them. In conformity with this view, I think that the [xi] Winter’s Tale, although perhaps not actually performed until the year 1611, can never have been the last work of Shakspere. It is far more like the labour of his youth, an idea to be revived seventy-five years later on supposedly scientific as well as impressionistic evidence (see here). Nevertheless, by the time of Hudson (ed. 1852, 4:6), the composition of WT was assigned to the winter of 1610–11. The later limit is 15 May 1611, when Forman saw the play. The earlier limit, Thorndike (1900, pp. 116–17) argues, was shortly after 1 January 1611, the date of the performance at court of Ben Jonson’s Oberon, the action of which opens upon Silenus, a shaggy old forest god, and at least five of the satyrs over whom he presides (the number who speak). For the probable reappearance of three of these satyrs in WT, see n. 2164 and here. The shortly arises from the assumption that the three saltiers who danced before the king (2158–9) must have done so recently in order for the audience, or at least the members of it who knew about courtly amusements, to grasp the allusion and perhaps also in order for the men still to be available to perform in WT. Yet the line is spoken in modern performances, its significance having become that the three will unknowingly dance before the king once more—that is, King Polixenes. It possibly never meant more than this. Moreover, as Nicoll suggests (see n. 2164), the satyr dance could be an interpolation. Thus WT may have no precise anterior limit, but, as Wells & Taylor (1987, p. 601) point out, if the Crane transcript from which the F text was typeset derived from the promptbook, the original composition predates January 1611 (see here).

Two attempts to prove WT an occasional play were made by Wickham (The Winter’s Tale, 1969, and Investiture, 1969), in the second of which the first is summarized: Critical discussion of The Winter’s Tale has centred on the sixteen-year gap dividing the Sicilian court scenes from the Bohemian pastoral scenes bridged only by Time as Chorus. . . . This odd structural pattern may have stemmed from a deliberate emblematic purpose. Since the skeleton of the plot is the fusing of seemingly irreconcilable opposites, and as nowhere in contemporary politics were such opposites more glaringly apparent than in Anglo-Scottish relations . . . Shakespeare deliberately reworked (and altered) . . . Pandosto at a narrative level in order to reflect emblematically the reunion of the British Isles under his master, James I. The datum point was the legend of the division of Britain by King Brutus among his three sons and Merlin’s prophecy of eventual reunion under a descendant of King Arthur. Henry VII had seen himself as that descendant and James viewed his own claim to the crowns of Scotland and England through Henry’s daughter Margaret (James’s great-great- grandmother) as the fulfillment of the prophecy. Wickham continues: The Winter’s Tale may be regarded not only as a figurative compliment to James I, but also Shakespeare’s contribution to the investiture of Henry, Prince of Wales and Heir Apparent, in 1610. The evidence is drawn from two pageants by Anthony Munday, two masques by Ben Jonson, a poem and a masque by Samuel Daniel, three of James I’s own speeches and two statues, the memorial effigies of Queen Elizabeth and Mary, Queen of Scots, commissioned by King James and placed on the queens’ tombs in Westminster Abbey. Examination of this evidence combines to designate The Winter’s Tale as written for performance in the autumn of 1610 before the King and the Heir Apparent . . . ; to reveal it as [figuring] the mystical marriage of Prince Henry (Florizel) to the three kingdoms whose original unity was lost but has been found (Perdita) thanks to Time and King James’ own piaculous action; and finally to show that, by substituting for the dead Queen of Greene’s novel the living statue of Hermione . . . , Shakespeare created a work of art which was as effective an emblem for his court audience as it was enjoyable dramatic romance for his wider public in the city. Everett (1970) supports Wickham’s parallel of Hermione and Mary, Queen of Scots: Hermione’s blend of royal dignity and pathos . . . [is] expressed in The Winter’s Tale by a few phrases so like echoes of Mary’s words [as given by Antonia Fraser in Mary, Queen of Scots, 1969] as to suggest the possibility of a real connexion. Mary told the deputation of lords who came to announce her trial: I am myself a Queen, the daughter of a King. Hermione refers to herself during her trial as a great king’s daughter [1213], and later . . . says, The Emperor of Russia was my father [1299]. Having cited several similar parallels, Everett concludes, There is, in short, enough dignity and pathos in the Jacobean (as distinct from the Elizabethan) image of Mary Stuart, to give an added interest to Shakespeare’s creation of Hermione, perhaps his queenliest heroine.

Apart from Everett, however, Wickham’s topicality seems to have attracted no supporters. Regarding the statues of the two dead queens, Smith (1972, pp. 217–18) observes: If either had anything to do with The Winter’s Tale, it would seem to be Mary’s, not Elizabeth’s; Mary could in some sense be thought to be resurrected by the investiture of her grandson as Prince of Wales. The trouble is that Mary’s statue was not finished until 1612 . . . and by this time The Winter’s Tale had been on the stage for two years. But Wickham, undismayed, says [in Heritage, 1969] the sculptors lived in Southwark . . . and Shakespeare could have dropped in at their studio any time from 1606 to 1610 to see [218] how the statue was coming. On this frail evidence we are asked to believe that Florizel represents Prince Henry. Nevertheless, these ideas are reiterated and further developed in Wickham (1973). Bullough (1975, 8:117) objects, though: It is going too far to suggest that the play figures the mystical marriage of Prince Henry (Florizel) to the three kingdoms whose original unity was lost but had been found (Perdita). . . . If there was such an allegorical intention why did not Shakespeare make Florizel (like Greene’s Dorastus) rule over two kingdoms at the end of the play? To which it might be replied that Florizel, as Perdita’s consort and Mamillius reborn, will reign over two kingdoms, although that fact will hardly save Wickham’s case.

Internal Evidence

As Furness (ed. 1898, p. 316) says, Roderick (1758, p. 225) was apparently the first to observe that the verse lines of H8 more frequently than those of other plays end in a redundant syllable. Malone (in Steevens, ed. 1778, 1:280 n.) adds that a mixture of rhyming lines and blank verse is a circumstance which seems to characterize and distinguish our poet’s earliest performance. From these beginnings developed a close study of Sh.’s style to determine, more exactly than external evidence allows, when he wrote his works.

The earliest critic to divide Sh.’s career into four periods distinguished by stylistic differences was Bathurst (1857). By the third quarter of the 19th c., one of the internal tests of chronology given by Ward (1875, 1:359–63) was versification, subdivided as follows: (a) Rhyme: Progress from more to less rhyme may be held to accompany the general progress of Shakspere as a dramatic writer. (b) Stopped and unstopped lines: A stopped line is one in which the sentence, or clause of the sentence, concludes with the line; but it is not always possible to determine what is to be regarded as the clause of the sentence, whether e.g. and is to be regarded . . . as beginning a new clause. The stopping of the sense is . . . often of more importance than the stopping of the sentence. (c) Feminine endings, or line endings with an eleventh, unaccented, syllable: While it is certain that Shakspere employed the feminine endings sparingly in many of his plays . . . regarded as early, it is certain that in those plays which on other grounds may be regarded as . . . late . . . he employed these endings largely. Ward’s examples, supplied by Fleay, include Shr. (now dated 1593–4), line 260, and WT 639, but also R3 (1592–3), line 570. The tabulations are not adjusted for the length of the plays or the proportions in them of verse and prose. And (d) Other verse tests, such as irregularities in the trimeter couplets of the early plays and, in the late plays, such carelessness as the failure to mark the caesura.

Fleay himself (1876) divides Sh.’s career into periods characterized by more or less distinct metrical features. In the last period (pp. 70–1), doggerel, alternately rhyming lines, and couplets are absent; alexandrines, with considerable variation in the position of the caesura, increase; and so do feminine endings; and so do lines of less than five measures. Fleay assigns WT (p. 136), along with Cym., Cor., Ant., and Tmp., to Sh.’s fourth (and last) period, but in a postscript (p. 138) creates a fifth period for Tmp. and WT alone. The actual date he assigns to WT is 1610–11 (p. 54) or 1611 (p. 130), 1611? having earlier been proposed by Furnivall (1874, p. xlv). Later, Fleay (1886, p. 65) decides that WT was certainly produced early this year [1610], before Jonson’s Alchemist, which was acted and entered S[tationers’] R[egister] October 3. Fleay’s techniques and conclusions raised questions immediately (see the discussion following the reading of his paper before the New Shakspere Society on 13 Mar. 1874, in Fleay, 1874), and he changed his mind as his investigations continued. In any event, his date for WT was not affected; in the 1886 work (p. 247) as in the 1874 volume (p. 10), it is 1610. By introducing this type of analysis, Fleay hoped to apply a scientific technique to the problems of Shn. chronology, but his followers proved capable of wild variations in method and conclusions, and he himself could not count.

The predilection for Shn. statistics, however, extends to Bather’s Table of Plays, According to Number of Puns (1887, p. 74), which shows a decrease in the number of puns per 100 lines from 3.97 in LLL to 0.26 in WT. As a technique for dating, the scheme is demolished by the fact that Tit. contains only 0.15 and 3H6 only 0.14. Bayfield’s (1920, pp. 402–3) verse analysis reveals that WT must have been written some years before the generally received date 1610. . . . It suggests that the performance at the Globe theatre on May 15, 1611 . . . must refer to a revival. . . . [403] [WT’s] versification is in fact of a quite different period [from that of the late plays. WT was written] near Troilus and Cressida, The Taming of the Shrew, and Timon. On the other hand, impressionistic critics who did not conduct verse tests could also reach strange conclusions. Spens (1922, p. 92), for example, suspects that all the Romances . . . were written originally by Shakespeare at the very beginning of his career, and that they were for the most part one or two act pieces forming part of a series. Mathew (1922, p. 41) agrees in part: These Plays were written first when Shakespeare was young and revised when he was mature.

The testimony of internal evidence is usually linked to the historical and practical facts of Sh.’s career, the latter being, for instance, that Sh. could not write more than a certain amount in a certain period, about two plays a year when his known work is distributed over the period during which he is believed to have been active. The distribution may also be colored by assumptions, not necessarily wrong, about Sh.’s development as a man and an artist. As Dowden (1877, pp. 37–41) puts it, We need no scientific test to make us aware that, in passing from Love’s Labour’s Lost to Hamlet, and from Hamlet to The Tempest, we pass from youth to manhood, and again from a manhood of trial and sorrow to a riper manhood of attainment and of calm. Affected by the transitions are (a) style and diction. In the earliest plays . . . the idea is at times hardly sufficient to fill out the language in which it is put; in the middle plays . . . there seems a perfect balance and equality between the thought and its expression[; in the latest] this balance is disturbed by the preponderance or excess of the ideas over the means of giving them utterance. . . . (b) The growth of Shakspere’s taste and judgment. . . . [38] (c) In the structure of the play and grouping of characters there is, in some of the early plays, a tendency to formal symmetry, an artificial setting of character over against character, and group against group. . . . Afterwards the outline of the play is drawn with a freer because a firmer hand. (d) . . . By degrees the characterisation becomes profound and refined. . . . (e) The entire reflective power deepens. . . . (f) The imagination . . . becomes passionately energetic, of daring and all-comprehensive power, as in King Lear, or lofty and sustained, with noble ideality, as in The Tempest. (g) The sympathy with human passion and the power of conceiving and dramatically rendering it in its most massive and most intense [39] forms increases. (h) . . . The humour of the dramatist . . . becomes full of grave significance, and works in conjunction with his (i) Deepening pathos. . . . (j) Finally, in moral reach, in true justice, in charity, in self-control, in all that indicate fortitude of will, the writings of the mature Shakspere excel, in an extraordinary degree, those of his younger self. These characteristics obviously cannot be quantified, but the alteration in Sh.’s verse can be. At first Shakspere has his breaks and pauses at the end of the line—the verse is end-stopt; gradually he more and more [carries] on the sense from one line to another without a pause at the end of the line [but] in some part of the line other than the end. Light and weak endings increase. The first group, in Dowden’s classification, encompasses words on which (p. 41) the voice can to a small extent dwell (am, are, be, can, could; the auxiliaries do, does, has, had; I, they, thou). The weak endings are more fugitive and evanescent . . . , including such words as and, for, from, if, in, of, or. Now weak endings hardly appear in Shakspere’s early or middle plays. . . . Nor do they come in by slow degrees at a later period. . . . In Macbeth light endings appear for the first time in considerable numbers; weak endings in considerable numbers for the first time in Antony and Cleopatra. This test serves perfectly to pick out the plays which form the group belonging to Shakspere’s last period of dramatic authorship; and within that period it probably serves to indicate nearly the precise order in which the plays were written. Dowden reproduces part of a table devised by Ingram (1874, p. 450) showing percentages of light and weak endings, from Mac. through H8, in a sequence that does not quite bear Dowden out, in that it places Tmp. earlier than Cym. and WT. Similarly, Dowden reproduces a table he attributes to Hertzberg; it shows the percentage of double, or feminine, endings increasing from LLL (4) through Tmp. and puts WT (31.09) before Cym. (32) by a whisker. Actually, the table Hertzberg published (1878, p. 252) reverses the sequence (WT 32.5), but in neither accounting is the difference great enough to establish priority. (Similar data from which similar conclusions were drawn persisted; see, for example, Neilson & Thorndike, c. 1913, pp. 69–75.)

The history of stylistic analysis and the efforts of other contributors to it are described in detail by Chambers (1930, 1:242–74). As a prelude to his own assignment of dates, he provides a thorough discussion of the problem of chronology, an evaluation of the evidence of dates, and (in 2:397–408) his own metrical tables, several of them corrected versions of Fleay’s. He believes (1:489) that the style and metre group Winter’s Tale with Cymbeline and Tempest, and it may reasonably be placed between them. A date early in 1611 is suggested by the probability that the bear of iii.3 and the dance of satyrs at [2164] were both inspired by Jonson’s mask of Oberon on 1 January 1611. For the bear, see n. 1500, and for the satyrs, n. 2164. Gray (1931, p. 148) averages the percentages of double [i.e., feminine] endings, run-on lines, and speeches ending with the line and discovers that the results for WT, Tmp., and Cym., though close to those for Cor. on the earlier side and H8 on the later, are also close to each other. Having related independent clauses to verse lines, Langworthy (1931, p. 748) gets similar results for the same plays, although his data order them as Tmp., Cym., and WT. Law (1936, pp. 50–1) tabulates the dates assigned by Adams (1923), Alden (1925), Chambers (1930), Craig (ed. 1931), Campbell (1932), Parrott (1934), and Brooke (ed. 1935). All give 1610 or 1610–11. So does Reinhold (1942, p. 87); like Oras (see below), Reinhold calculates split lines (pentameters shared by two or more speakers) as a percentage of the total lines in each play to show that their number generally increases with time. In an early statistical study, Yardi (1946) uses multiple measurements of metrical data for the discrimination of groups; the results place WT with Cym. and Tmp. but do not provide actual dates. Brainerd’s (1980) statistical study of Shn. chronology has no bearing, for WT, dated 1610.5, is a member of the test set, plays for the most part unambiguously dated.

A few recent critics dissent. Wentersdorf (1951, p. 178), combining such stylistic features as split lines, extra syllables within and at the end of lines, feminine endings, and alexandrines into a metrical index for each play, finds the three last to have been written in Chambers’s (1930) order, but he moves each work back by a year, WT to 1609–10. As Tmp. was performed at court on 1 November 1611, it would have been acted publicly at least by the summer of 1611, since public success presumably preceded the selection of any play for presentation at court. If the dance of satyrs in WT derives from Jonson’s Oberon, WT may nevertheless have been in existence prior to January 1611, when the masque was performed, and the satyrs subsequently added. It has also been suggested that the bear scene in Act iii was borrowed from . . . Mucedorus, which was revived by the King’s Men shortly before 1610. If this was the case (as the bear episode is not in Shakespeare’s source for WT), it points to an upward [i.e., anterior] time limit late in 1609. . . . Finally, the reference at [461] is sufficiently motivated by the story and in any case too general to warrant interpretation to the murder of Henri IV on 14 May 1610 (see Chambers, 1930, 2:489). Having assigned Per. to 1607–8 and Tmp. to 1610–11, Wentersdorf accordingly dates Cym. and WT 1608–9 and 1609–10, respectively. Also dissenting is Oras (1960), who studies three types of pauses in the verse of Sh. and a number of other early writers, under three heads: all pauses indicated by the internal punctuation of the earliest eds. (A-patterns); strong pauses indicated by punctuation heavier than commas (B-patterns); and lines shared by two or more speakers—split lines—(C-patterns). That the punctuation may be scribal or compositorial is recognized, but it is still considered a valid marker because its source is contemporaneous. The data are presented numerically and graphically, and although no actual dates are assigned, the patterns of Cym. and WT resemble each other more strongly than they resemble the patterns of the other works studied.

Alexander (ed. 1951, p. xv), wisely declining to be pinned down to more than an approximate order of composition, places WT, along with Per., Cym., Tmp., and H8, between 1608 and 1613. Maxwell (ed. Cym., 1960, p. xi), commenting on 1609–10 and 1610–11 for the dates of Cym. and WT, respectively: It is reasonable to associate the greater artistic assurance of The Winter’s Tale with a later date [than Cym.’s], which is also supported by the fact that Shakespeare undoubtedly knew the Boccaccio source of Cymbeline when he wrote The Winter’s Tale. I think [the 1610–11] date for The Winter’s Tale may well be a year too late. There is a fairly close verbal parallel between The Winter’s Tale [1941–7] and Philaster [4.6.2–8], which seems to me most easily explained as an echo of the former by the latter; and Philaster is not later than 8 October 1610 [see Chambers, 1923, 3:223]. To Tillyard (1938, pp. 9–10), however, the echoes seem reversed: Sh. is improving Fletcher (see n. 1941–7). In any case, as Pafford (ed. 1963, p. 97 n.) remarks, the strewing of corpses as well as bridal beds with flowers is a common idea.

As for the presence of Cym.’s source in WT, the story in question is the ninth novel of the second day of Boccaccio’s Decameron. Muir (1977, pp. 262–3) summarizes it: Some Italian merchants at an inn in Paris deride the idea of a female chastity and Bernabo of Genoa is provoked by Ambrogiuolo to bet on the chastity of his wife, Ginevra. On going to Genoa, Ambrogiuolo realizes that he cannot seduce Ginevra. Concealed in a chest, he is able to observe the pictures of her bedroom, to steal a ring and other belongings, and to observe a mole on the lady’s breast. Bernabo is convinced that he has lost the wager and he orders his servant to murder Ginevra. The servant, convinced of her innocence, spares her. She dresses in his clothes and takes service with the Soldan. One day she sees her purse and girdle in a stall in the market-place. When the truth comes to light, Ginevra reveals herself to her husband and forgives him. The villain is tortured to death: the verie same day that hee was impaled on the stake, annointed with honey, and fixed in the place appointed, to his no meane tormente: he not onely died, but likewise was deuoured to the bare [263] bones, by Flies, Waspes, and Hornets, whereof the Countrey notoriously aboundeth. As Iachimo is forgiven, this torture is not used in Cymbeline; but in The Winter’s Tale Autolycus tells the Clown [2665–72]. Smith (1972, p. 216) holds that it is generally agreed . . . that The Winter’s Tale is the later of the two [because] the use of the leftover passage about the punishment of the villain . . . suggests that The Winter’s Tale followed Cymbeline.

Earlier, however, Nosworthy (ed. Cym., 1955, pp. xvi–xvii), instead of drawing this conclusion, recognizes the possibility that Shakespeare owed the name Belarius [in Cym.] to the Bellaria of Greene’s Pandosto . . . so that the evidence can point either way. My own guess is that the composition of the two plays was more or less simultaneous or, at any rate, that both had been written, revised and prepared for the stage before either was actually performed, with consequent cross-fertilisation. This view . . . tallies with the attractive theory, put forward by [Bentley, 1948], that the impending acquisition of the Blackfriars private theatre led, in the Spring and Summer of 1608, to discussions among the King’s Players as a result of which Shakespeare was henceforth to write with the Blackfriars in mind, and not the Globe, and that [Cym., WT, and Tmp.,] in that [xvii] order, were the fruits of that decision. Bentley says nothing about their respective dates, but the application of his theory would . . . suggest 1608 for the first play of the series. Recalling that it was at the public theater that Forman saw WT, Nosworthy demurs at and not the Globe, preferring to think that after his company acquired the Blackfriars theater in 1608, Sh. wrote dual-purpose plays, for such, most emphatically and triumphantly, the Romances are. If Cym., the first play of the series, was composed in 1608, 1608–9 would presumably be the date of WT. Although Pafford (ed. 1963, p. xxiii) does not deal explicitly with Nosworthy’s idea, he comments: The language, style, and spirit of the play all point to a late date. The tangled speech, the packed sentences, speeches which begin and end in the middle of a line, and the high percentage of light and weak endings are all marks of Shakespeare’s writing at the end of his career. But of more importance than verse tests is the similarity of the last plays in spirit and themes. . . . Practically all authorities . . . accept, with minor variations, the approximate dates given by Chambers [i.e., 1610–11]. Those who do not follow Chambers may instead favor Wentersdorf—for example, Fitch (1981, p. 300), who, in a revival of sense-pause investigation, decides on 1609–10.

Summary

A comprehensive reexamination of the internal evidence of WT’s date is made by Taylor (in Wells & Taylor, pp. 93–109). Among the data included are the percentage of rhyme to verse and to prose; colloquialisms in verse (e.g., i’th’, ’em, ’ll, I’m), which show that Shakespeare’s reversion to an antiquated dramatic form [the romance] apparently coincides with some backsliding toward a less colloquial poetry (p. 101); and a revision of Wentersdorf’s metrical indices and Oras’s pause patterns. Taylor concludes (p. 107) that although minor ambiguities remain about the order of particular plays, we can be reasonably confident about the shape of the canon after about 1597. Nevertheless, recent eds. differ to some extent, as the following comparison shows (see table).

Chambers (1930) is included because his chronology was accepted by many eds. until recently; all his dates given here are reiterated by McManaway (1950). Lr. is present because Taylor holds that the text of that play included in F1 is a revision of the text that first appeared in the quarto of 1608 substantial enough to constitute a new creation. Cardenio is a play based on a story in Don Quixote and thought to be by Shakespeare and Fletcher (as are H8 and TNK). As Chambers (1930, 1:539) notes: A play of Cardenno or Cardenna was given by the King’s men at court in the winter of 1612–13. It was acted in 1727 under the title Double Falsehood and printed in 1728 as Written Originally by W. Shakespeare; And now Revised and Adapted to the Stage by Mr. Theobald. Bevington’s WT range is an innovation for him; his previous editions of the Works (1973 and 1980) substantially agree with Chambers and with Evans (c. 1610–11). He appears to have extended these dates on the chance that Taylor may be right, a question that critics of Lr. will have to decide. As it now stands, however, c. 1610–11 is as close as one can come to the date of WT.

A Comparison of Four Chronologies

Sources

Primary Source

Pandosto

The first ed. of WT to quote extensively from Pandosto to illustrate Sh.’s dependence was Malone (ed. 1790). Recognition of WT’s source preceded Malone by a century, however. Langbaine (1691, p. 466): The Plot of this Play may be read in a little Stitch-pamphlet, which is call’d, as I remember, The Delectable History of Dorastus and Fawnia, otherwise Pandosto, by Robert Greene. Rowe (ed. 1709, 1:xxvii–xxviii): The Winter’s Tale . . . contains the space of sixteen or seventeen Years, and the Scene [xxviii] is sometimes laid in Bohemia, and sometimes in Sicily, according to the original Order of the Story. Gildon (1710, p. 336): Whence I suppose the Absurdities are copyed, and the making Bohemia of an Inland, a maritime Country. The opinion of Grey (1754, 1:244) that Dorastus and Faunia is of a more modern date [than WT], and borrow’d from Shakespeare was refuted by Farmer (in Steevens, ed. 1778). Farmer reported a copy of Pandosto with a publication date of 1588, considerably before any date proposed for Sh.’s play.

On 1 July 1588, A booke intitled the complaint of tyme was entered in the Register of the Company of Stationers to Thomas Orwin (Arber, 1875–94, 2:493). Pandosto. The Triumph of Time was printed by Orwin for the stationer Thomas Cadman (RSTC [3:277] 12285); The Historie of Dorastus and Fawnia is its head title. The discrepancy between the title as given in the Register and the title as published creates some uncertainty that the entry pertains to the novel (see Wells, 1988, pp. xxx–xxxi); even so, there is no reason to doubt that the novel appeared in 1588. It was frequently reprinted—in 1592, 1595, 1600 (RSTC 12287.5, a single copy in the Biblioteka Gdanska), 1607, 1609, 1614, 1619, 1621 (a single copy in the Vienna National Library, located by Wells; not in RSTC), 1629, 1632, 1632–6 (date cropped; see Wells, p. xxxviii), 1636, and c. 1640 (date cropped; see Allison, 1975, no. 84). Wells (pp. xl–xlii) identifies eleven more editions of Pandosto, including an abridgment, published from 1648 to the end of the century and seven or eight more complete eds. and abridgments from about 1700 to 1735; uncertainties in the tabulation arise from difficulties in distinguishing the eds. and in dating them. The title pages of eds. 1588 through 1632 carry as a subtitle The Triumph of Time; in eds. 1636 and following, the earlier title and subtitle are replaced by The Pleasant Historie of Dorastus and Fawnia, the title Langbaine was thinking of.

Jusserand (1890; 1966, p. 155): The novel had an immense success, much greater according to appearances than the exquisite drama of a Winter’s Tale, that Shakespeare drew from it. Collier (1836, p. 19 n.) alleges that the 1609 ed. is probably the very one used by our great Dramatist. In a subsequent work, Collier (ed. 1842, 3:476 n.) declares that in eds. of Pandosto published after 1588, the oracle’s words are the king shall live without an heire, whereas in 1588 the word is die. He settled on 1609 rather than an earlier ed. because it is the one that most immediately preceded the writing of WT. The actual situation, according to Wells (1988, p. 128), is that the eds. of 1588, 1592, and 1595 read liue, as Sh. does (WT 1315), whereas 1609 and subsequent eds. read die. Muir (1957, p. 240) asserts that Sh. used the 1588 ed., but, as Wells remarks, Muir evidently did not know that the 1592 and 1595 eds. existed. Since Sh. did not reproduce a unique feature of any of these eds., the specific one he consulted cannot be ascertained. Moreover, Coggins (1980) argues that there seems to have been a 1584 ed., no copies of which are known to have survived.

Later versions of Pandosto are found in Lennox (1753), a paraphrase; Collier, Shakespeare’s Library, vol. 1 (1843; 1875); NUC lists an ed. from the 1840s, possibly a separate issue of the Collier text, which seems not to be noticed elsewhere; an ed. of 1858, according to S. A. Tannenbaum & D. A. Tannenbaum, Elizabethan Bibliographies: Robert Greene [1939], possibly the version in Halliwell (ed. 1859); A. B. Grosart, The Life and Complete Works of Robert Greene, vol. 4 (188?); Morley (ed. 1887), an abridgment; Anon., Pandosto, or The Historie of Dorastus and Fawnia, by Robert Greene (New Rochelle, 1902); Thomas (ed. Pandosto, 1907); James Winny, The Descent of Euphues (Cambridge, 1957); Bullough (1975); and Wells (ed. Pandosto, 1988, but finished in 1962). Eds. of WT containing Pandosto include those of Halliwell, Morley (1887), Furness (ed. 1898), and Pafford (1963).

According to Bolte (Schlussscene, 1891, p. 90), the novel provided material for a French play and a Dutch play: Jean Puget de La Serre’s Pandoste ou la Princesse malheureuse, en deux journées (Paris, 1631), which was performed at the Hôtel de Bourgogne in 1631; and Meynert Voskuyl’s Dorastus en Fauniaas (Amsterdam, 1637). The British Library Catalogue also lists Voskuyl’s Bellaria en Pandostos (Amstrelredam, 1637). Thomas (1907, p. xix) reports that a French version by Alexandre Hardy, who wrote several pastoral plays, is now lost. Jusserand (in Lee, ed. 1907, pp. xxviii–xxix) gives further details: The first translation was printed at Paris, chez Guillaume Marette, in 1615, under the title of Histoire tragique de Pandosto roy de Bohême et de Bellaria sa femme. Ensemble les amours de Dorastus et de Faunia. . . . [xxix] Le tout traduit premièrement en Anglois de la langue Bohême, et de nouveau mis en françois par L. Regnault. . . . The translation [takes] a good many liberties . . . with the text (some voluntary, others not). . . . To the supposed Bohemian original there is no further allusion. One adaptation, Le roman d’Albanie et de Sycile par le Sr du Bail gentil-homme Poict[evin], appeared in 1626, and another, Histoire de Pandolphe, roy de Bohême et de Cellaria sa femme, in 1722. Jusserand also supplies (pp. xxxvi–xxxix) the sketches made by the stage decorator [xxxvii] Mahelot for the scenery used in the performance of Hardy’s dramatic version.

A derivative in blank verse, Francis Sabie’s The Fissher-mans Tale (RSTC 21535), appeared in 1595, some copies being issued as a part of Sabie’s Pans Pipe, three eclogues in hexameters; and The Fissher-mans Tale was followed by a second part, Flora’s Fortune (RSTC 21536), also in 1595. These poems were reprinted by Halliwell (ed. 1859, 8:127–60)—possibly from a defective copy, since the extracts, as he calls his text, run from line 632 of The Fissher-mans Tale to the end of Flora’s Fortune.

Pandosto of 1588 collates A–G4. The one surviving copy, located in the British Library, lacks the four leaves of sig. B; the missing text is supplied here by the edition of 1592, Folger Shakespeare Library. In the following reprint, the black letter of these eds. is rendered as roman and the roman as italic. The Epitaph (here) is italic in ed. 1588. Most ornaments are ignored. Macrons and abbreviations of the, and, that, and with have been expanded. The two- and three-line initials beginning certain paragraphs are reduced to regular capitals, turned letters are returned, and the white space preceding and following some lines of dialogue has been removed. Footnotes show the origin of emendations of the 1588 and 1592 texts; the spelling of the emendations has been altered occasionally to accord with that of the copy-text. Asterisks in the text call attention to the footnotes. In the notes, a wavy dash in the variant reading repeats the word of the lemma in the corresponding position; the caret indicates absence of punctuation.

Important points of comparison and contrast between Greene’s novel and Sh.’s play, and representative commentary on them, may be found in the following notes: 3370, 3371, 3378, 3379, 3381, 3382, 3388, 14–16, 50–1, 121–3, 181–92, 234, 269–72, 285–8, 288, 410, 450–1, 455–66, 461, 468–81, 565, 582, 583, 648, 715–19, 798–805, 800, 938–41, 980, 1016–17, 1062, 1111–15, 1134, 1147, 1173, 1191–5, 1202–6, 1219–21, 1222–4, 1263, 1280, 1293, 1299, 1347–8, 1366–85, 1437, 1440, 1486, 1507, 1509–10, 1512, 1559, 1596–8, 1691–2, 1795, 1798–9, 1798, 1799, 1826–36, 1828–32, 2240, 2263, 2408, 2512, 2630, 2724, 2915, 2926, 2995, 2998, 3032, 3061. Passages of the novel referred to in these commentary notes are here preceded by bracketed TLN numbers of those notes, inserted into the text. In addition, TLN numbers for some other lines in WT, which correspond to passages in Pandosto but are not discussed in the notes, are likewise inserted here into the text before the corresponding passages of the novel.