A Midsummer Night's Dream arrow Created with Sketch. Appendix

Irregular, Doubtful, and Emended Accidentals in Q1

The following emendations made in the Variorum text correct obvious typographical irregularities in Q1. Mistakes that produce English words are not recorded here, but they are in the historical collation. In each note the lemma is the emended Variorum reading, and the first siglum is that of the source if any from which the emendation is drawn, followed by the rejected Q1 reading.

10 Hip.] not indented Q1
15, 27, 62, 120 The.] the. Q1
61, 179, 192 Her.] her. Q1
66 looke.] Q2;  ~ , Q1
88 Her.] Q2;  ~ , Q1
162 patience] Q2; patienee Q1
269 Quin.] Q2; Qnin. Q1
275 Bott.] not indented Q1
279 Quin.] Quin, (catchword) Q1
307 Flu.] Q2; Fla. Q1
310 cõming.] F1
(comming.)
; cõ- | (ming. (turnover) Q1
319–20 Two lines F1; one line Qq
337 Quin.] Not indented Q1
361 you,]  ~ ‸ (catchword) Q1
363 Moone-light] F1; Moone-| light Qq
372 Enter] ¶ Enter Q1
379 the greene.] the | (greene. (turnunder) Q1
414 of the night.] of | (the night. (turnunder) Q1
529 certaine] Q2; cettaine Q1
552 Pu.] Not indented Q1
690 comfort] Q2; comfor Q1
714 faire prayer] Q2
(faire praier)
; faireprayer Q1
804 thought] Q2
(-thought)
; thoughr Q1
820 Quin.] rowe1; Qnin. Q1; Peter. Q2-F4
845 Sno.] Not indented Q1
863 Bot.] catchword Q1; Bo. (full line) Q1
885 your] Q2; yonr Q1
889, 1456 Ro.] Not indented Q1
922 through bryer:] through | (bryer: (turnunder) Q1
936 trãslated. Exit.] trãslated. | (Exit. (turnover) Q1
1072 mee to.] mee | (to. (turnover) Q1
1190 Hermias] hermias Q1
1266 sweete] Q2; sweeete Q1
1279 Helen] helen Q1
1305 Hermia] hermia Q1
1312 Therefore] Q2; Thefore Q1
1314 Helena] helena Q1
1359 Hel.] Hel. Q1
1362 but] Q2; hut Q1
1368 Deme.] Q2
(Dem.)
;  ~ , Q1
1389 should] Q2; shoud Q1
1391 farre blamelesse] Q2; farr eblamelesse Q1
1435 notwithstanding] Q2; notwistanding Q1
[see Press Variants]
1451 for] Q2; [long s]or Q1
1472 shalt] Q2; shat Q1
1523 much] Q2; mueh Q1
1548 Squirils hoord,] Squirils | (hoord, (turnover) Q1
1557 Enrings] Q2; Enríngs Q1
1567 of fresh] Q2; offresh Q1
1594, 1617 Tita.] tita. Q1
1599 Ti.] not indented Q1
1601–2 eyes peepe.] eyes | (peepe. (turnover) Q1
1624, 1666 The.] the. Q1
1633 Hercules] hercules Q1
1659–60 their hornes.] their | (hornes. (turnover) Q1
1670 enmitie.] Q2;  ~ , Q1
1688 Helena] Helena Q1
1696 Helena] helena Q1
1717 found] Q2; fonnd Q1
1749 of doubt] Q2; ofd o ubt Q1
1773 Quin.] Q2; Quin, Q1
1793 Hip.] hip. Q1
1814 Hyp.] hyp. Q1
1827 wee haue,] wee | (haue, (turnover) Q1
1846 Thracian] thracian Q1
1850 Of learning] Q2; Oflearning Q1
1854, 2067, 2106, 2113 Thisby] thisby Q1
1929 Thisby] Q2; Thsby Q1
1939, 1996, 2130 Thysby] thysby Q1
1994 haire] Q2; hayire Q1
1998 Py.] py. Q1
2001 Pyra.] pyra. Q1
2003, 2005 Pyr.] pyr. Q1
2007 Thy.] thy. Q1
2063 Th.] not indented Q1
2051–2 woulde change.] woulde | (change. (turnover) Q1
2064 Lyon.] Q2; Lyon, Q1
2112 Pyramus] pyramus Q1
2180 Tita.] tita. Q1

Unadopted Conjectures

9 withering out] lithering out [= lingering or lingeringly] becket (1815, 1:265); widowing out gould (1881, p. 12); widowing on gould (1884, p. 15)
13 night] height daniel (1870, p. 30)
16 merriments] merriment furness (v1895)
31–4 Two lines ending consent . . . Duke mtby3
36 Thou, thou] Thou gould (1884, p. 15)
46 harshnesse] testiness kellner (1931, p. 14)
53 Immediatly] Immutably kellner (1925, p. 105)
59 leaue] lave [= embellish, beautify] becket (1815, 1:265)
85 earthlyer happy] earthly happy steevens in v1803; eathlier or rathelier happy marsh (1878, p. 243)
85 distild] distol’d [= live together in pairs] gould (1884, p. 56)
97 After 99 wagner & proescholdt (ed. 1881)
101 crazed] razed wilson (1873, p. 242)
103 Hermias] Hermia mtby2
(withdrawn mtby4)
, tyrwhitt (1766, p. 50)
116 Nedars] Nestor’s walker (1860, 2:30); Medon’s kellner (1931, p. 14)
119 Vpon . . . spotted] ’Pon . . . apostate wilson (1873, p. 243)
127 fancies] fancie mtby2, keightley (1867, p. 130)
140 which I could] Yet could I becket (1815, 1:265)
141 Beteeme] Bestream or Bestow wilson (1873, p. 244)
146 crosse! too high to be inthrald to loue] cross, to be enthrall’d! too high, too low becket (1815, 1:265–6)
149 friends] others mcol1
(and withdrawn)
154 a shadowe] an arrow gould (1887, p. 68)
156 spleene] sheen [= brightness, lustre] mhan1; shene [= shining] becket (1815, 1:266)
160 then] that mtby3
161 It stands] If’t stand rann
180–1 bowe . . . heade] craft . . . shaft kalepky (1928, p. 242)
192 God speede] Speed you mgrin
195 lodestarres] lode-stones kellner (1925, p. 36)
199 Your . . . Hermia,] ( ~  . . .  ~ ,) knt1
199 words I] worth I’d mtby2, wagner & proescholdt (ed. 1881)
200 eare . . . voice] fair . . . fair deighton (ed. 1891)
201 tongue] voice cartwright (1866, p. 10)
212 hateth me] follows thee mgrin
214 Hel. . . . would . . . mine] Her. Would . . . Thine! mgrin
217 Lisander] Demetrius clarkes (ed. 1864)
228 Vpon faint] ’Pon sam(i)t [= samite] kellner (1931, p. 14)
231 eyes] feet mtby2
232 strange companions] stranger companions mcal
232 companions] societies mtby2
246 vile] wild wilson (1873, p. 220)
246 quantitie] quality johnson in v1773
278 a] Om. warner in v1778
297 teare a Cat in] tear: à catin [French: like a very drab] becket (1815, 1:267); tear-coat in wh1
298 raging] ragged mtby4, white (1793, in fennell, 1853, p. 30), furness (v1895)
298 in] in and keightley (1867, p. 131)
341 you should] we should mtby3; I should mtby4
345 doue] doe bailey (1866, 2:198–9)
368 obscenely] obscurely grey (1754, 1:47 and withdrawn)
371 cut] break or not mhan1
375–6 Robin. . . . Fa.] Fairy. . . . Puck jourdain (MS c. 1860)
375 How] Why how mcap3
378 moons] Mooned mtheo1
379 And] Fairy. And jourdain (MS c. 1860)
379 orbs] herbs grey (1754, 1:48); cups wilson (1873, p. 246)
384 heere] clear daniel (1870, p. 31)
399 or] in moberly (ed. 1881)
400 square] jar or sparre peck (1740, pp. 223–4); squall anon. in peck (1740, p. 223); quarrel wilson (1873, p. 246)
413 Thou] Indeed, thou schmidt (1881, p. 3); Spirit, thou ard1
413 aright] all aright mtheo1 and wagner & proescholdt (ed. 1881)
420 bob] bab gould (1884, p. 15)
422 aunt] aunct [abbreviation of auncient] becket (1815, 1:268)
425 tailour cryes] Tail over eyes or O Lord, cryes mtby2
(and withdrawn)
; murder cries mtby4
425 tailour] tail-sore anon. in capn; tailloir [= the square stone of the capital of a pillar] becket (1815, 1:268–9); tail her bell (1852–64, 3:194); tail o’er carruthers & chambers (1861, 3:80); traitor perring (1885, pp. 67–72); faitor [= traitor] deighton (ed. 1906, p. 90); tale o’er perring in wright shakespeariana; tailer furness (v1895); hallo! kellner (1931, p. 14)
429 roome Faery] Fairy, roome, for seymour (1805, 1:43); roomer Fairy nicholson (1864, p. 49); room, fair fairy mtby4, ard1; give room or room ho oxf2
436 skippe] keep harness in col1; trip dyce1
457 the middle] this muddy wilson (1873, p. 247)
457 spring] prime wilson (1873, p. 247)
460 in] upon lettsom in dyce2
466 pelting] petling jackson (1819, p. 11)
472 murrion] murrian’d mtby3; murrain’d chedworth (1805, p. 68)
473 Morris] MORTICE anon. in johnson (ed. 1771, addenda 1:12)
475 lacke] want kinnear (1883, p. 86)
476–7 The . . . blest] reline as 484–5 elze (1867, p. 537); reline as 490–1 moberly (ed. 1881)
476 want] wants wood (ed. 1806); chant wh1 (Suppl., 1:xliii, and withdrawn); have keightley (1867, p. 132); wail kinnear (1883, p. 86)
476 winter heere] wonted cheer mtby2; wonted year john1; winter’s chear hutchesson in cam2
476 winter] Winters warburton in theo1; summer keightley (1867, p. 131)
476 heere.] Here. [from HERR . . . HEER . . . a Lord, or Master] anon. (Caribbeana, 1741, 2:75); heer. [= hard, rigorous] becket (1815, 1:270); gere. brae in cam1; hire. wilson (1873, p. 247); hoar. herr (1879, p. 91); clear edgecombe (2000–1, p. 6)
484 Hyems] Adam’s herr (1879, p.93)
484 chinne] chill theobald in mtby2 and mtheo2 (20 May 1729 [fol. 132v]), and grey (1754, 1:49)
484 chinne and Icy] icy cime, a [Cime is . . . French for top] becket (1815, 1:270–1)
487 childing] chilling or churlish herr (1879, pp. 92–3)
489 increase] inverse mhan2
503 embarked traders] traders embarked sprague (ed. 1896)
507 Following (her] Flowing (her mjenn; Fellowing (her mtol; Her fellowing becket (1815, 1:271–2); having her cartwright (1866, p. 10)
510 rich with] with rich mcol1
525 once I] I once mcol1
544 purple] purpled mtby4
557 then] whom kellner (1931, p. 6)
575 draw not Iron. For] draw no truer; for wilson (1873, p. 248); draw, not I run anon. in wilson (1873, p. 248);  ~  ~  ~ ‸ for gould (1887, p. 57)
585 loose] loathe anon. in hal; tose [= teaze, torment] kellner (1931, p. 14)
629 mee] here mtheo1
632 Quite ouercanopi’d] White clover canopied bulloch (1878, pp. 59–60)
635 these] those cartwright (1866, p. 10), wells & taylor (1987, p. 281)
637–8 intervening line lost: Upon her will I steal there as she lies keightley (1867, p. 132)
638 And] Now lettsom in dyce2
648 Cocke crowe] Cock-crow tannenbaum (1933, p. 113)
652 for] in mtby2, heath (1765, p. 51); e’er mtby3, hud2; fly kinnear (1883, p. 88)
652 minute] Minuit warb
666–7 intervening line lost gould (1887, p. 68)
678 Sentinel screams and goes to inform the other fairies parsons (1953, p. 67)
704 For lying so, Hermia] For, Hermia, lying so schmidt (1881, p. 4)
709 humane] common wilson (1873, p. 249)
717 prest] blest mgrin
726 Despised] Who despis’d mgrin
728 and dirty] bedewed gentleman (ed. 1774)
731 Churle, vpon . . . eyes] Upon . . . maiden eyes gentleman (ed. 1774)
734 Sleepe] Keep daniel (1870, p. 32)
734 thy] thine mgrin
759 Helena] Helen walker (1860, 1:230)
759 shewes] owes (owns) the moberly (ed. 1881)
768 Helena] Helen, now, seymour (1805, 1:46); now Helena mgrin; but Helen now walker (1860, 1:230)
773 ripe] rip’d mtby2 and schmidt (1881, p. 8)
777 Loues stories] Love-stories walker (1860, 1:255)
805 And] Yet thiselton (1903, p. 38)
811 Either] Or mgrin
821 things] three things walker (1860, 2:256)
825 feare] feat becket (1815, 1:272)
829 seeme] serve gould (1884, p. 57)
830 swords] sword mcol1
836–7 eight & eight] eighty-eight anon. in hal
839 I feare] I [= Ay], I fear furness (v1895)
848 necke] maske gould (1887, p. 57)
855 them] ’em anon. in cam2; hem thiselton (1903, p. 40)
867–8 great chamber window] great-chamber anon. in cam1
867 leaue] set rid
895 odious ‸] odours, or odorous ‸ col2
897 sauours sweete] savour’s vile schmidt (1881, p. 4)
898–9 intervening two lines missing mal
898 hath] not schmidt (1881, p. 4)
899 but heere a while] a while but here jackson (1819, pp. 13–4)
907 triumphant] a pungent mtby2
908 brisky] frisky clayton (1979, p. 14)
908 Iuuenall] Jew, venal bluestone (1953, p. 326)
908 Jew] jew’ [an abbreviation of jewel] becket (1815, 1:272); Joy thiselton (1904, p. 16); Juv [abbreviation of Latin iuvenum] taylor (1990, p. 61)
917 faire,] so, fair keightley (1867, p. 133); fairer schmidt (1881, p. 4); horse, jackson (2000, p. 70)
921 a Round] around furness (v1895)
922 bogge,] brook, thro’ bog peck (1740, p. 157); bog, through burn ritson in v1793; bog, through bourn ard1
924 headelesse] heedless delius (ed. 1859); curblesse gould (1884, p. 57); herdless kellner (1925, p. 36)
988 eye] tail knt1
(and withdrawn)
989 haue] show gould (1884, p. 15)
1010–1 your patience] your relations mjenn; your passions farmer in mstv1; you passing mason (1785, p. 69); your puissance rann
1022–56 Om. gentleman (ed. 1774)
1031 patches] wretches grey (1754, 1:60 and withdrawn)
1039 nole] cowl mgrin
1041 Minnick] mammock ritson (1783, p. 44)
1047 at our] at one allen in v1895; with one kellner (1931, p. 15)
1058 latcht] bath’d mjenn; lav’d or wash’d mlong; laced anon. in cam1; hatch’d [= ornamented, thinly covered] daniel (1870, pp. 32–4); streak’d or bath’d wilson (1873, p. 249); washed orger (1890, pp. 40–1); hatch’d [= stained, smeared] deighton (ed. 1891); leeched kellner (1931, p. 15)
1071 intervening part line lost schmidt (1881, p. 5)
1071 plunge] wade maginn (1837, p. 378)
1071 the deepe] more deep mlong; thigh-deep kellner (1931, p. 15)
1071 to.] too, nor leave me here to weep cuningham (ed. 1905)
1077 displease] disseise [= dispossess] annandale in irv
1080 dead] lead’n cartwright (1866, p. 10)
1089 him then?] him? Then furness (v1895)
1097 mood] word allen in v1895
1101 and if] And, if furness (v1895)
1103 I: see] I. So, | See furness (v1895)
1107 So] Since deighton (ed. 1891)
1168–9 O, . . . kisse This . . . white, this] This let me kiss, / This princess of pure white–O seal of bliss! becket (1815, pp. 273–4)
1169 Princesse of pure] purest of pure mjenn and lettsom in dyce2; quintessence of bailey (1862, p. 153; withdrawn, 1866, p. 200); essence of pure cartwright (1866, p. 11); priceless purest anon. in moberly (ed. 1881); Empress of pure irv (and withdrawn)
1169 white] whites bailey (1866, p. 200)
1174–5 doe, | But . . . ioyne, in soules,] do‸ | In souls, but . . . join ‸ malone (1780, 1:118)
1175 ioyne in] join, ill tyrwhitt (1766, p. 32)
1175 soules] scoffs or scorns mtby3; scorns or scoffs john1; scouls blackstone in malone (1780, 1:118); shoals white (1785, p. 278); soulk [= wretchedness] becket (1815, 1:274); sooth bailey (1866, p. 202); taunts elze (1867, p. 538); sport wetherell (1867, p. 582); sports wilson (1873, p. 250); insults spedding in cln1; sport gould (1884, p. 57); ieeres oxf2
1188 know I] do, I jackson (1819, p. 14)
1192 do] love cam1
1196 to] be [= by] kellner (1931, p. 15)
1201 aby] abay [= suffer] becket (1815, 1:274)
1201 deare] here walker (1860, 1:307)
1202 deare] fere cartwright (1866, p. 11)
1204 his] its wilson (1873, p. 250)
1215 oes] orbs grey (1754, 1:61)
1228 vs; O] of vs; O mtby2; vs; O, O! mtby4
1228 all] all then keightley (1867, p. 133); all this hud2
1229 All] Our mtby2
1230 two artificial] to artificer wilson (1873, p. 252)
1230 gods] buds wilson (1873, p. 252); girls gould (1884, p. 15)
1245 for it] for’t walker (1860, 3:79)
1247 your] your complaining mtby4
1249 scorne] loue mtby4
1261 vnlou’d)]  ~  . . . keightley (1867, p. 133)
1264 Perseuer,] —perceive you wilson (1873, p. 254)
1278 praise] praier thiselton (1903, p. 53); pleas oxf2
1283 to.] true. anon. in v1895
1286 Ethiop.] Ethiop, you! heath (1765, p. 53)
1287–8 intervening line or part line lost cam1
1287 No, no:] Now, now, bulloch (1878, pp. 60–2); om. rid
1287 heele] sir, no! mjenn; he’ll not stir: jackson (1819, p. 15); hell wilson (1873, p. 255); Sir! Hell’s abyss bulloch (1878, pp. 60–2); he’ll not / Forsake his love. Coward, you fear to fight; furnivall (1880) in wright shakespeariana; sir, no: schmidt (1881, p. 7); thou’lt kinnear (1883, p. 89); sir; still orson (1891, p. 153); heele kiss— anon. in wright shakespeariana; she will let you; perring in wright shakespeariana; you’ll thiselton (1903, p. 53); om. rid; he’ll only ard2
1288 Seeme] Dem. Seem joicey (1893, p. 102); Seems wilson (1873, p. 255)
1288 you would] you’d keightley (1867, p. 134) and anon. in wright shakespeariana
1288 would follow] would, fellow! wilson (1873, p. 255); would flow bulloch (1878, pp. 60–2)
1289 not] on bulloch (1878, p. 62)
1304 newes,] news‸ [= revolutionizes, v. trans.] crook (1914, p. 107)
1312 Therefore be] Therefore, | Be walker (1860, 3:49)
1312 out of hope,] out moberly (ed. 1881)
1312 of question, of doubt:] of question: anon. in cam1; of doubt, of question, schmidt (1881, p. 6); question and doubt anon. in wright shakespeariana
1366 Minimus] You minim, as [= ace] kellner (1931, p. 15)
1366 made] man’d [= maimed] kellner (1925, p. 124)
1388 shadowes] Fairies gould (1884, p. 15)
1409 his] its wilson (1873, p. 252)
1411 derision] division mtby2, guest (1838, 1:130, 147), strachey (1854, p. 680); discision kellner (1931, p. 15)
1427–8 Given to Oberon mtby2 and mtheo2 (20 May 1729 [fol. 133r])
1428 black-brow’d] endless rid
1434 salt] sea tathwell in grey (1754, 1:62)
1434 salt‸]  ~ , white (1854, p. 216)
1437 Pu.] Puck [sings]. anon. in cam1
1439 Goblin, . . . downe] Given to Oberon col1; Goblin-lead-them-up-and-down staunton (1874, p. 863)
1466 why] wherefore schmidt (1881, p. 6)
1472 buy] bide mtby4
1485 Rob.] Puck [sings]. anon. in cam1
1497 Rob.] Puck [sings]. anon. in cam1
1497 sleepe] Sleep you seymour (1805, 1:48)
1499 thou] Then thou mtheo4, mQ2fl7, mjenn, chedworth (1805, p. 70), seymour (1805, 1:48); thou now oxf2
1505 mare] mate gould (1884, p. 15)
1506 well] still steevens (v1793)
1525–6 ouerflowen] over-flow’d malone (1780, 1:118–9)
1542 desirest] desires furnivall (1880) in wright shakespeariana
1545 of hay] of a [= hay, or ale] hunter (1845, 1:296)
1547–9 four verse lines ending Fairy . . . hoard . . . wary . . . board bulloch (1878, pp. 62–3)
1549 thee] thee some mgrin; thee the walker (1860, 2:257); thee in deighton (ed. 1891)
1549 newe] mellow anon. in bullen (1907, 10:408)
1549 nuts] nuts wary To furnish forth thy board bulloch (1878, pp. 62–3)
1552 an exposition] a disposition mgrin
1554 be alwaies away.] be away.—Away! [Seeing them loiter.] upton (1746, p. 241); be always i’ th’ way. heath (1765, p. 55); be always: Away! jackson (1819, p. 15); bear all noise away. kellner (1931, p. 15)
1555–6 So . . . entwist:] given to The First Fairy farzaad (1946, pp. 53–4)
1555 woodbine] wood rine [= bark] upton (1746, pp. 241–2); weedbind steevens (v1778); wood pine gould (1887, p. 68); bindweed wray (–1892) in wright shakespeariana and in cam2
1556–7 the female . . . Elme.] given To chorus Of Fairies farzaad (1946, p. 54)
1557 fingers] fissures gould (1887, p. 68)
1579 transformed] transforming wilson (1873, p. 256)
1582 May all] All may grey (1754, 1:64 [Errata])
1604 rocke] knock whalley (1756, 5:275)
1605 new] anew dey (1901, p. 481)
1613 sad] staid daniel (1870, p. 34)
1628 Vncouple] Uncoupled malone (1780, 1:119)
1641 sanded] sounded col1
1651 Nedars] Nestor’s walker (1860, 2:30)
1691 melted] All melted stau; Immaculate bulloch (1878, pp. 63–4); So melted or Being melted schmidt (1881, p. 7); Has melted perring in wright shakespeariana
1691 as the] away like mjenn; as thaws the kinnear (1883, p.95)
1698 But] Then lettsom in walker (1860, 2:115); And mgrin; When kinnear (1883, p. 96)
1698 But like a] Belike as bulloch (1878, pp. 64–5)
1698 sicknesse] sickman oxf2
1699 But] And kellner (1931, p. 11)
1703 we more will here] more will we hear lettsom in walker (1860, 3:50)
1711 Come] Come, my mgrin; Come me perring in wright shakespeariana
1712 Deme.] Lys. capn (1779, 2:113–4)
1717 like a iewell] likewise double mtby2; like a gimmal [= Ring of double hoops] smith (1803, p. 11); like a Guille [French . . . for deception, trick] becket (1815, 1:276); like a double cartwright (1866, p. 11); like a double [= counterfeit stone] furness (v1895) [misreading batten (1876, p. 12)]
1718+1 Are . . . sure | That we . . . It . . . ] Are . . . sure we . . . it keightley (1867, p. 136)
1718+1 Are] Are well cap; Are now mal; Are yet anon. in cam1
1767 scaped] scraped grey (1754, 1:70)
1770 in] for hal
1783 preferd] proffered mtheo2 (27 May 1729 [fol. 8])
1803 a] the dodd (1752, 1:87)
1806 the formes] a mass seymour (1805, 1:49 and withdrawn)
1812–13 interpolated lines white (1854, pp. 217–9)
1812 Or] For mtby2, anon. in cam2; As taylor in cam2
1839 There] Here mtby2 and anon. in hal
1856 And wõdrous strange snow] a wonder strange enow bullen (1907, 10:408)
1856 wõdrous strange] wind-restraining wetherell (1867, p. 582); ponderous flakes of leo (1880, p. 708)
1856 wõdrous] pond’rous jortin in mtby3
1856 strange] strong mason (1785, p. 71); swarte stau and kinnear (1883, p. 97); warm chaplyn in cam2; raven or orange or azure bailey (1862, pp. 197–9); staining nicholson in cln1; strange, hot or strange, jet perring in cln1; sooty herr (1879, p. 94); scaldinge ebsworth (ed. Q2 1880, p. xviii); red perring (1885, p. 75); fiery orger (1890, p. 41); flaming mtby2 and orson (1891, p. 58); flaring scott (ed. 1898); scathing perring in wright shakespeariana; stranger cuningham (1920, p. 402); tawny kellner (1931, p. 15); flamy macintyre (–1950, fol. 2)
1856 snow] show mtby2; in hue bulloch (1878, pp. 65–6)
1868–9 play it? . . . men,] play’t? . . . men, | My noble Lord (or My gracious Duke) schmidt (1881, p. 7)
1874–5 I haue heard It ouer] as the second half of line 1876 daniel (1870, p. 35)
1874 No . . . you. I . . . heard] No . . . you, daniel (1870, p. 35)
1875–7 reline as 1877, 1875, 1878 daniel (1870, p. 35)
1875 ouer] o’er daniel (1870, p. 35)
1876–7 intervening line lost john1; transpose gould (1884, p. 16)
1877 strecht] wretch’d ulrici in v1895
1877 cond] penn’d kenrick (1765, p. 20)
1878 To . . . seruice] as the first half of line 1876 daniel (1870, p. 35)
1888–9 noble respect Takes it in might, not] respect As noble, taken not in might but richards (1892) in cam2 and in wright shakespeariana; a fault Noble respect takes in might, not lambrechts (1965, p. 164)
1888 duty] duty meaning spedding in cam1
1888 cannot] can but poorly tiessen (1877, p. 6)
1888 doe,] do, yet would or do, tho’ fain mcole; aptly do, bailey (1866, p. 203); do, but would, abbott (1870, p. 419)
1888 doe, noble respect] do aright, Respect seymour (1805, 1:52)
1888 noble] Om. bulloch (1878, p. 66)
1889 it . . . not] not . . . but john1
1889 might] mind bailey (1866, p. 203); right cartwright (1866, p. 11)
1889 might, not merit] merit, not in might seymour (1805, 1:52); noble might, not noble merit bulloch (1878, p. 66)
1895 haue] th’ave wh1
1918–19 A good . . . true.] given To Demetrius mtby4, cam1
1922 Chaine] skein anon. in cam1
1944 slaine] sleyne [= torn into threads] becket (1815, 1:278)
1966 haire] hau(l)m [= straw] kellner (1931, p. 15)
1969 discourse] in discourse farmer in v1773
2000 Helen] Heren blackstone in tomlins (1844, p. 97)
2010 Moon] Mean [= partition] kellner (1931, p. 16)
2010 Moon vsed] moon housed mtby3; monial [= dividing-post] round (1914, p. 287)
2013 heare] sheer [= get away] mhan1 (11 June 1737, fol. 20) and clayton (1979, p. 30); disappear heath (1765, p. 58); leave gould (1884, p. 16)
2020 beasts, in] beasts; e’en malone (1783, p. 11)
2020 man] moon-calf farmer in v1773
2020 and] in jackson (1819, p. 16)
2026 as . . . am] am . . . in daniel (1870, p. 35); as . . . n’am gollancz (ed. 1894)
2027 A] Nor mlong
2027 nor] none cartwright (1866, p. 11); but keightley (1867, p. 136); or daniel (1870, p. 35)
2027 no] a mason (1785, p. 72)
2027 damme] skin daniel (1870, p. 35)
2038 Moone] man anon. in cam1
2071 And] Now lettsom in dyce2
2072 Lyon] lion’s lettsom in dyce2
2090 deflour’d] devour’d mtby2
2095 hoppe] rap gould (1881, p. 12)
2098 Tongue . . . Moone] Eye . . . Moone mtby2 and scott (ed. 1898); Tongue . . . mount, mtby4; Moon . . . | Dog elze (1867, p. 538)
2109 Heere . . . Play.] given To Philostrate mtby2 and mtheo2 (27 May 1729 [fol. 69r])
2113 warnd] ward mjenn and stau
2115 meanes,] mourns mF2fl48
2120 These . . . lippes . . . nose] This . . . brow | . . . mow kinnear (1883, p. 100)
2120 lilly lippes . . . cherry nose] lips lily . . . nose cherry farmer in v1773
2120 lippes] O’s cartwright (1866, p. 12); toes bulloch (1878, pp. 67–8)
2120 this] With gould (1881, p. 12)
2120 nose] nip wh1; tips gould (1881, p. 12)
2154 hungry] Hungarian grey (1754, 1:70, in the lemma)
2175 house giue] house in john1; hall go lettsom in dyce2; hall a cartwright (1866, p. 12); housewives’ wilson (1873, p. 260); house gives kinnear (1883, p. 100–1); house, giv’n orger (1890, p. 42)
2176 By] Gives cartwright (1866, p. 12); Now kinnear (1883, p. 100–1); But orson (1891, p. 58)
2200 his] this colne
2203 Euer shall] Every hall staunton in ingleby (1855, p. 771), and stau (and withdrawn)
2204 And] So rid
2204 owner] owners wilson (1873, p. 260)
2212 more yielding] mere idling wilson (1873, p. 260)
2213 reprehend] reprobate hunter (1845, p. 282)

The Text

Authenticity

There has been less resistance to attributing all of MND to Shakespeare than with other plays in the canon. Wilson (ed. 1924, p. 94): The only presumably pre-Shakespearian play, known to us by name, which might conceivably have formed the basis of the Dream, is the mysterious King of the Fairies, scornfully linked by both Nashe and Greene with another drama called Delfrigus . . . , as part of the stock-in-trade of a travelling company. Wilson refers to Nashe’s The Gentlemen Stvdents of Both Vniversities (ed. R. B. McKerrow, 1904–10, 3:324): a company of taffaty fooles . . . might haue antickt it vntill this time vp and downe the Countrey with the King of Fairies, and dined euery day at the pease porredge ordinary with Delfrigus; and to Greene’s Groatsworth of Wit (ed. A. B. Grosart, 1881–6, 12:131), in which a player says to Roberto why, I am as famous for Delphrigus, and the king of Fairies, as euer was any of my time. Wilson continues, We have not, however, been able to trace any clues to the existence of such a play beneath the Shakespearian text, unless it be its curious connexion, or seeming connexion, with old dramas like Damon and Pythias, 1582, and Heywood’s translation of Seneca’s Hercules Furens, 1581. For the alleged allusion to Damon and Pythias, Wilson (p. 148) cites passages beginning at 2083 and at 2124, and (pp. 109–10) for that to Hercules Furens the passage beginning 297. Wilson opines that (p. 110) there seems no reason why Shakespeare should burlesque a translation ten or a dozen years old and therefore thinks perhaps the text here goes back to some pre-Shakespearian version. de la Mare (1935, pp. xxxii–xlviii), attending to the poor quality of some of the verse given the four lovers and to the absence of the vocabulary of some of these lines from the rest of the Sh. canon, concludes (p. xlvii) that the earliest draft of the Dream was not of [Sh.’s] own workmanship but a play . . . written by some more or less artless scribe. Following de la Mare, Wilson then states unequivocally that (1948, p. 29) the original play was not written by Shakespeare at all. Yet, according to the noted disintegrationist Robertson (1924, 1:440), in MND we can catch the true voice of Shakespeare. Chambers (1924, p. 10) is prepared to credit Sh. with commonplace Elizabethan dramatic carpentry, rather than disintegrate plays by attributing parts of them to other playwrights.

Blumenthal, who finds (1961, p. 116) some seven participants in the authorship of the Shakespeare canon, thinks Robertson’s claim to be (p. 30) uncertain. Among the anti-Stratfordians, Bacon (1857, p. lxxxi) attributes MND to Raleigh, Theobald (1901, passim) attributes it to Bacon, Clark (1930, pp. 435–49) and the Ogburns (1952, chs. 44–5) to Oxford, Brooks (1943, pp. 596–7) to Dyer, Titherley (1952, pp. 71–5) and Evans (1956, p. 59) to Derby, Sweet (1965, p. 71) to Queen Elizabeth, and Hoffman (1955, p. 127) to Marlowe. Ross (1939, pp. 16–17) thinks MND among the works of Anne Whateley written in association with Sh.

The First Quarto (1600)

The printing history of Sh.’s MND starts on 8 Oct. 1600 when a book with its title was approved for publication in the Stationers’ Register (Book C, fol. 65v, as transcribed by Greg, BEPD, 1:16): Tho. fyssher Entred for his copie vnder the hand[es] of mr Rodes / and the Wardens. A booke called A mydsõmer night[es] dreame According to Greg (ibid., 3:1485), this was the only copy for a play licensed by Rodes, who, Greg suggests, may have been Thomas Rhodes. Greg (1962, p. 81): A Thomas Rhodes appears in the index to Hennessy’s Reportorium without a reference. A book called Micrologia; Characters or Essays, by M. R., 1629 (STC 17146), was licensed by E. Martin on 22 Dec. 1628 as Rodes charecters.

The play appeared in print in the same year, with the following title-page: [ornament] | A | Midsommer nights | dreame. | As it hath beene sundry times pub- |lickely acted, by the Right honoura- |ble, the Lord Chamberlaine his |seruants. | Written by William Shakespeare. | [device <McKerrow>321] | ¶ Imprinted at London, for Thomas Fisher, and are to | be soulde at his shoppe, at the Signe of the White Hart, | in Fleetestreete. 1600. This first edition is often called the Fisher Quarto to distinguish it from Q2 also dated 1600 on its title page; Q2 was once known as the Roberts Quarto and is now usually called the Pavier Quarto. Chambers (1930, 1:356): The printer may be [Edward] Allde or [Richard] Bradock. Greg (BEPD, 1:276): The printer [of Q1] appears from the ornaments used to have been probably Richard Bradock. The device is Fisher’s. Turner (1962, p. 33): As far as I have been able to determine, nothing is known of Bradock which would be of significant value to us in our examination of MND Q1. He was admitted to the Livery on 1 July 1598 and for a time was actively engaged in the trade. Around the turn of the century, he probably printed several play quartos: in 1598 [Christopher Marlowe’s] Edward II Q2; in 1600 [Ben Jonson’s] Every Man out of his Humor; in 1601 [Anthony Munday’s] The Downfall and The Death of Robert Earl of Huntington; and in 1602 [John Marston’s] Antonio and Mellida [and] Antonio’s Revenge, and [Jonson’s] Poetaster. Bradock seems to have thrived as a printer between 1598 and 1608, but he also printed plays long before and some time after the turn of the century. Greg (BEPD, 3:1497) records that Bradock printed his first extant play, Nathaniel Woodes’s The Conflict of Conscience, in 1581, and did not print his last until 1616 (Beaumont and Fletcher’s The Scornful Lady). However, Pantzer (STC 3:16) identifies the printer of the latter as John Beale. Thus the last of the extant plays that Bradock printed date from 1608, including A Yorkshire Tragedy from the Shakespeare apocrypha.

Berger (ed. 1995, p. viii): Thomas Fisher’s career as a publisher and bookseller was a short one. He was freed as a draper on 8 November 1596 by Richard Smith and transferred to the Stationers’ Company in 1600. Of the three other titles associated with him, Nicholas Breton’s Pasquil’s Mistress was printed in 1600, perhaps by Richard Bradock, and John Marston’s Antonio and Mellida and Antonio’s Revenge were printed in 1602 by Bradock. In these last two Matthew Lownes appears to have had an interest as well, as his shop in St Dunstan’s Churchyard is cited on the title-pages of both volumes.

Bartlett & Pollard (1939, pp. 70–1) identify and locate eight extant copies of Q1. Modern facsimiles include that by William Griggs, ed. J. W. Ebsworth, 1880; that in Shakespeare’s Plays in Quarto, ed. Michael J. B. Allen & Kenneth Muir, 1981; and Vol. 157 of the Malone Society Reprints, ed. Thomas L. Berger, 1995. For a digital facsimile see the Shakespeare Quartos Archive (www.quartos.org).

MND Q1 was one of a number of Shakespeare’s plays to see print in 1600 with the dramatist’s name on the title-page. The others were 2H4 Q1, MV Q1, and Ado Q1. Earlier printings with Shakespeare’s name on their title-pages were LLL Q1 1598, R2 Q2 and Q3 1598, R3 Q2 1598, and 1H4 Q2 1599. Blayney (1997, p. 388): some of these plays sold very well, R2, R3, and 1H4 being in the list of the top ten best-sellers among early modern English plays. However, like Ado and 2H4 (neither of which ever saw a second quarto edition) and LLL (which was not reprinted in quarto until 1631), MV and MND (without second editions until 1619) did not sell very well, even with Shakespeare’s name on their title-pages. Ibid. (p. 385) identifies May 1600–Oct. 1601, the interval within which MND Q1 was entered, printed, and published, as one of two peak periods for the registration of plays between 1585 and 1604, the other coming at Dec. 1593–May 1595. In each peak period twenty-seven plays were registered, although only 80% of the total of fifty-four registrations issued in books. Ibid. (p. 387): MND was one of eight Lord Chamberlain’s plays registered in the second peak period. See also Blayney (2005), and Farmer and Lesser (2005, Popularity and 2005, Structures).

Quality of Printing in Q1

While on the whole positive, editors and critics exhibit a wide range of opinion about Q1’s quality as a witness to what Sh. wrote. Pope (ed. 1725, 1:xv), not attending to the question of the priority of Q1 to Q2: If any were supervised [at the press] by himself [Sh.], I should fancy the two parts of Henry the 4th, and Midsummer-Night’s Dream might have been so; because I find no other printed with any exactness; and (contrary to the rest) there is very little variation in all the subsequent editions of them. Johnson (ed. 1765, 1:176) unable to establish the priority of Q1 to Q2: Neither of the editions approach to exactness. Building on Capell’s (1783, 2.3:115) inference of Shakespearean authority for the punctuation of the mechanicals’ prologue (see n. 1906–15), Knight (ed. 1839, 1:331) writes, The original of these editions, whichever it might be, was . . . carefully superintended through the press. The text appears to us as perfect as it is possible to be, considering the state of typography in that day. There is one remarkable evidence of this. The prologue to the interlude of the Clowns, in the fifth act [1906–15], is purposely made inaccurate in its punctuation throughout. . . . ; and this is precisely one of those matters of nicety in which a printer would have failed, unless he had followed an extremely clear copy, or his proofs had been corrected by an author or an editor. Compare Hudson (ed. 1851, 2:255) and White (ed. 1857, 4:17), as well as the following writing after Q1’s priority had been demonstrated: Chambers (1930, 1:358), Parrott (ed. 1938, p. 131), Greg (1942; 1954, p. 125), Doran (ed. 1959, p. 27), Holland (ed. 1994, p. 112). Clark and Wright (ed. 1863, 2:viii), guessing right about Q1’s priority, nonetheless regard it carelessly printed. Kittredge (ed. 1936, p. 229) echoes this judgment, describing Q1 as not very carefully printed.

Furness (ed. 1895, pp. x–xii), also guessing right about Q1’s priority: the excellence of the text [of Q1] is counterbalanced by the inferiority of the typography. . . . [A]lthough the entrances of the characters are noted, the exits are often omitted, and spelling throughout is [xi] archaic, for instance, shee [241], bedde [228], dogge [589], &c., betraying merely a compositor’s peculiarity. . . . [F]onts are mixed, and the type old and battered. Believing that the Q1 compositor set type by the ear from dictation, Furness finds such errors as Dians bud, or Cupids flower [1588], instead of Dian’s bud o’er Cupid’s flower the consequence. (P. xii): [I]t is assuredly more likely that such blunders as Eagles [454] for AEgle, or Peregenia [453] for Perigouna . . . are due to the deficient hearing of a compositor. However, Furness (p. xii) concedes that compositors . . . are exposed [to such errors] when with a retentive memory they carry long sentences in their minds, not just when they set from dictation. Cuningham (ed. 1905, p. xi), in the main closely following Furness: the text . . . has reached us in a state of comparative correctness and purity, [yet] there are passages which are admittedly corrupt. Ibid. (p. xv): the text is superior, and likewise the punctuation.

Wilson (ed. 1924, p. 79): Q1 is superior to many of the other quartos. . . . The misprints are few, and the literals fewer. The compositors seem to have worked slowly, whether through inexperience or because they found the manuscript difficult to read; for the text contains a number of interesting archaic spellings which almost certainly derive from the copy. . . . On the whole the work must be pronounced as moderately competent. Its chief weaknesses are two. First it is evident that the compositors conceived it as their duty to expand most of the contractions they found in the original. Particularly instructive in this connexion is the misprint Bet it [691] in which we catch the compositor red-handed so to speak. [Wilson also compares 35 (bewitcht), 78 (Whether), 895 (of), 1231 (needles), all of which he suspects to be compositorial expansions of copy forms.] And secondly it is clear that the compositors have introduced a large number of full stops into a text which originally contained very few, and that they have also peppered the dialogue with superfluous commas. Furthermore their pointing is careless, as is shown by the numerous instances of transposition in terminal stops. Nevertheless, apart from commas and periods the punctuation of the Quarto is comparatively good on the whole, at times even beautiful. Wilson’s belief that printer’s copy must have contained few commas and periods is skewed by his thinking that the lightly punctuated Hand-D pages in the MS The Booke of Sir Thomas Moore (BL Harl. 7368) are certainly Sh.’s and that they are typical of his punctuation; see Wilson’s What Follows if Some of the Good Quarto Editions of Shakespeare’s Plays Were Printed from His Autograph Manuscripts, Transactions of the Bibliographical Society 15, 1917–19, p. 136. Evans (ed. 1974, p. 247), following Wilson, cites the excessively heavy use of phrasal commas in Q1. Differing from Wilson, Ridley (ed. 1934, p. viii): It is true that the Quarto is very heavily punctuated, very much more so, for example, than Hamlet. But it is not on the face of it likely that a compositor, who after all is a busy working man, is going to pepper his pages with commas, or any other mark of punctuation, merely for the fun of the thing. To him the insertion of marks of punctuation is merely so much more labour, and prima facie therefore there seems no reason why we should not pay as much attention to the compositor’s commas as to any of his other marks of punctuation. . . . [T]he punctuation of the Quarto very frequently produces interesting results. In a certain number of cases it makes a real difference to the sense; in more cases it makes a real difference in the emphasis which is thrown upon phrases by their becoming more isolated; and, perhaps most important of all, it greatly diversifies the rhythms. However, according to Joseph Moxon in Mechanick Exercises on the Whole Art of Printing (1683–4) (ed. Herbert Davis & Harry Carter, 2nd ed., 1962, p. 192), it is the compositor’s duty to discern and amend the bad . . . Pointing of his Copy.

Turner (1962, pp. 33–54): There is nothing very striking about the typography; on cursory examination the book seems to be a run-of-the-mine Elizabethan dramatic quarto. Turner’s primary focus is instances of erroneous line-division of dialogue, some few of which he attributes to the compositor. (P. 48): Mislineation [of verse at 490–1, where the last word of the first line is printed as the first word of the second line] may have resulted from the compositor’s carelessness; but . . . just possibly . . . the MS rather than the workman was at fault. . . . Almost certainly the compositor was juggling the text when he set a short speech of Bottom’s and one of Peter Quince’s in a single line of type at the foot of B2 [319–20, in the inner forme (i.e., B1v, 2, 3v, 4) after, Turner thinks (see here), the outer forme (i.e., B1, 2v, 3, 4v) was already set and the] limits of B1v, B2 . . . had been established. At 413–14, the compositor apparently thought he could squeeze the first complete line of verse into the same line of type with the half-line of verse which begins the speech, a calculation which, as the turn-over shows, was none too accurate. The compositor made analogous interventions at (p. 49) 552–3, 1071–2 and (p. 54) 2063–4, but his responsibility [is] doubtful for the line division of 61–2. He set prose as verse at (p. 49) 911–12 and (p. 54) 1986–90. (Werstine [2012, pp. 144–5, n. 24] suspects the compositor was perhaps responsible for further mislineation [see here].) Turner also tabulates dozens of wrong-font errors (such as roman for italic and small capitals for full capitals) both (pp. 40–5) apparently deliberate—because of shortages of type—and (p. 40, n. 8) accidental.

Brooks (ed. 1979, pp. xxxiii–iv): In the First Quarto an editor has a text of high authority to follow. . . . [I]t is a gratifyingly clean text. . . . In the present edition, it has been found necessary to depart from Q1 in just over fifty verbal readings, besides nine punctuations affecting the sense, one transposition of a pair of lines, a number of line-divisions, and a very few places where verse was set as prose or prose as verse. Apart from the authorial lapses, and the cruces, [xxxiv] the faults are unsurprising errors of the press, almost all of the kinds that compositors are prone to.

For the supposition that Q1 lacks one or more songs, see n. 2175–2206.

Press Variants and Proof-Correction in Q1

Johnson (ed. 1888, p. 39) records the press variants listed below on sig. F1v. Wright (ed. 1891, 2:295) lists only the first of the two on sig. F1v, as does Furness (ed. 1895, p. 166).

Berger (ed. 1995, pp. vi–viii):

Collation of the eight copies reveals five [that is, six] press variants in four of the sixteen formes. . . . The inner forme of sheet A exists in three states.

Copies Collated

  • BL (British Library, C.34.k.29 . . . )
  • Bodl (Bodleian Library; C3 damaged . . . )
  • TCC (Trinity College Cambridge . . . )
  • CSmH (Henry E. Huntington Library . . . )
  • CtYEC (Yale Elizabethan Club . . . )
  • DFo (Folger Shakespeare Library . . . )
  • MB (Boston Public Library . . . )
  • MH (Harvard University; lacks C2, C3, H2, H3 . . . )

Press Variants

Sheet A (inner forme)

  • Corrected:BL, MH
  • Uncorrected:Bodl, TCC, CSmH, CtYEC, DFo, MB

Sig. A2r[4]Now] Now (turned initial N)

  • Corrected:BL, Bodl, TCC, CSmH, CtYEC, MB, MH
  • Uncorrected:DFo

Sig. A2r [18]to funerals:] ro funerals:

Sheet E (outer forme)

  • Corrected:BL, Bodl, TCC, CSmH, CtYEC, DFo, MH
  • Uncorrected:MB

Sig. E3r[1254]he] be

Sheet E (inner forme)

  • Corrected:BL, Bodl, TCC, CSmH, CtYEC, MB, MH
  • Uncorrected:DFo

Sig. E1v[1159]ſwore] fwore

Sheet F (inner forme)

  • First stage [sic] corrected:BL, Bodl, CSmH, CtYEC, DFo, MH
  • Uncorrected:TCC, MB
  • Sig. F1v[1435]notwiſtanding] notwiſtandiug
  • Sig. F1v[1438]them vp & down:] them vp & dowe:
Berger lists no second stage of correction of the inner forme of sheet F; were there such a stage, he presumably would not have stated only of the inner forme of sheet A that it exists in three states.

Ibid. (p. vi, n. 4): Often it is difficult to determine when a press variant exists, and the distinction between deliberate stop-press variants and accidental shifting and bad inking can be vexing to decide. Thus, the h in both at [125] (Sig. A3v) appears to have slipped slightly in the British Library copy, producing bot h, and the space between I and know at [1667] (Sig. F4v) is so loose that the Bodleian, Folger, Huntington, Harvard and Yale Elizabethan Club copies read Iknow. Similarly, O long at [1479] (Sig. F2r) has slipped significantly, producing Ol ong in the British, Huntington, Yale Elizabethan Club, Folger, and Harvard copies. The hyphen in loue-shaft at [536] (Sig. C1r), clear in some copies, is so weakly inked as to appear almost invisible in the British Library, Trinity College Cambridge, and Yale Elizabethan Club copies. At [1159, see above in list of press variants] (Sig. E1v), I agree with W. W. Greg that the Folger copy is variant and reads fwore . . . ; but Richard Kennedy, textual editor of the New Variorum Midsummer Night’s Dream, disagrees. Kennedy and I agree (contra Stanley Wells and Gary Taylor, William Shakespeare: A Textual Companion (Oxford, 1987), p. 287) that what appears to be a comma after melody at [201] (Sig. A4v) in the Bodleian copy is an overinked full stop.

Ibid. (p. vi, n. 7): Much of the bottom half of C3 [in the Bodleian copy] has been torn off. It has been repaired with another piece of paper, and the missing quarto text added in a post seventeenth-century hand. The copyist placed a comma after an extant deere at [695], which the Oxford editors mistook for a press variant (Textual Companion, pp. 281, 287).

Compositor Identification and Order of Printing in Q1

Turner (1962, p. 33): Neither variations in spelling nor typographical abnormalities indicate that [Q1] was set up by more than one compositor.

Turner also (pp. 34–5) provides a partial analysis of the headlines or running-titles, which (p. 34) read on both recto and verso A Midsommer nightes dreame, except on H3v where we find A Midsommer nights dreame. Below his analysis is completed in square brackets, with the roman numerals that Turner assigns to each distinctively identifiable headline and with the pages on which each occurs identified by their signatures:

I [B2, C3v, D4, E3v, F3v,] G4, H4
II [B4, C2, D2, E2, F2,] G2, H2, A2v
III [B1v, C1v, D1v, E1v, F1v,] G1v, H1v, A3
IV B3, C3, D4v, E4v, F4v, G4v, H3, A4v
V [B2v,] C1, D2v, [E1, F1,] G3, H4v
VI [B3v, C4, D3v, E4, F4,] G3v
VII B4v, C4v, D3, E2v, F3, G2v, H1, A3v
VIII [B1, C2v, D1, E3, F2v,] G1, H2v, A4
IX H3v
From this analysis Turner can establish incontrovertibly that (p. 34) the book was worked in two skeleton-formes, one regularly imposing the inner [1v, 2, 3v, 4] and the other the outer forme [1, 2v, 3, 4v] and that sheet A . . . was the last sheet to go through the press. His demonstration of the latter proposition consists of observation of significant changes [that] were made in two [running-]titles during the course of printing: (1) The g appearing in the title used on B3 and C3 (IV in the [chart above]) was replaced at D4v, and the new type appears on E4v, F4v, G4v, H3, and A4v. (2) The title used on B4v and C4v (VII) is characterized by a broken r and a defective e in dreame. At D3 a break in the M also appears, and the three defects are found together on E2v and F3 (where the e’s of dreame were exchanged in position). At G2v the r seems to have been replaced, and the e prints somewhat better than usual. When the title appears on H1, only the break in the M and the new r are evident, and only these two characteristics can be observed in the title as it appears on A3v. It is clear that sheet A was printed after sheet H.

Furthermore, according to Turner, it can be shown that the outer forme of sheet H was the first of its formes to go to press and that it is possible to generalize from this practice with sheet H and to infer that the (p. 35) outer formes of all sheets but A regularly preceded inner formes through the press. The grounds provided for these two conclusions make them far from certain. The alleged priority of presswork on the outer forme of sheet H depends initially for Turner on the pattern of reappearance of headlines from sheet H in sheet A. Both of the headlines (namely VII and VIII) used for the only two pages of the inner forme of sheet A that require headlines (namely A3v and A4, because A1v is blank and A2 bears the head title) come from the outer forme of sheet H. Then a third headline from H outer (IV)—together with two headlines from H inner (II, III)—supply the three pages of A outer needing headlines (excluding the title page, A1). Knowing that H(o) was sent to press before H(i) [and thus was returned to the compositor before H(i) so that its headlines were available for use in A(i), the first forme of sheet A to be set into type] and that earlier in the book all the outer formes were imposed in the same skeleton used for H(o), we can infer that outer formes of all sheets but A regularly preceded inner formes through the press.

This demonstration of Turner’s depends not just on the evidence he presents but also on questionable assumptions that he makes—one explicit, the others not. He implicitly assumes that Bradock printed MND Q1 by itself, rather than concurrently with other books, or (Blayney, 1982, p. 92) that MND was the only work available for composition. However, McKenzie (1969, p. 18), studying the records of Cambridge University Press from the late 17th c., discovers that concurrent printing is frequent, and Blayney (1982, pp. 45, 264 n.) finds evidence of the practice among books printed in London in the decade immediately following Bradock’s work on MND Q1. Turner also implicitly assumes that (Blayney 1982, p. 92) one of Richard Bradock’s two presses was not in use. The explicit assumption underlying Turner’s analysis is that, in terms of the production of the single book Q1 on a single press, (Turner 1962, p. 46) composition and presswork could stay more-or-less in balance. . . . [The compositor planned] to compose two formes, distribute the first, set the third, distribute the second, and so on. . . . Thus the speed of the press, which barring accidents would have remained fairly constant, is established as the rate at which about four type pages could be composed. Examination of the CUP records by McKenzie (1969, pp. 8–10) shows a wide variation in the speed of both composition and presswork not only by different workmen but also by the same workmen at different times and therefore calls into question the likelihood of compositors or pressmen maintaining the balance supposed by Turner. Only by applying these assumptions can it be assumed that the single press was still occupied printing the second forme of Q1’s sheet H as the compositor was setting and then imposing the first forme of Q1’s sheet A and further assumed that the investigator’s task at this juncture is thus simply to determine which forme of sheet H supplied headlines to the first sheet-A forme to be set. With these assumptions in place, Turner concludes H(o) was the first forme of that sheet to be wrought off the press because headlines from it appear in A(i), whereas headlines from both formes of sheet H appear in A(o). If any of Turner’s assumptions fail, then his demonstration becomes inconclusive. There is no evidence that can be adduced for any of the assumptions, which therefore have the status only of hypotheses, two of which are falsified by the evidence against them provided by McKenzie and Blayney.

Proceeding on these assumptions Turner plots the recurrence of distinctively damaged individual pieces of types in Q1, using this evidence to argue that the book was generally set into type by formes and not seriatim (that is, in the order in which pages are to be read). According to Turner, a forme that contains types only from one of the formes of a preceding sheet must have been set into type before a forme that contains type from both formes of that preceding sheet. This judgment is constructed by analogy to the one about headlines already discussed. It yields the following results concerning the order in which the pages of Q1 were set into type:

  • Sheet B (p. 41): not necessarily set by formes: It is a safe guess that all of B(o) was set before work began on B(i), but we cannot absolutely rule out such an order as B1–B1v-B2v-B3–B4v-B2[–B3v, B4];
  • Sheet C (p. 41): initial seriatim setting gave way to setting by formes: C1 (B[o] was almost certainly distributed by the time C2v was set and possibly before much of C1 was set [on the dubious evidence of the k at C1, 8] [But see also (p. 35): h B3,19–C1,1.]), 1v, 2, 2v, 3, 4v (distribution of B[i]), 3v, 4;
  • Sheet D (p. 41): D1 (distribution of C[o]), 2v, 3, 4v, 1v, 2 (distribution of C[i]), 3v, 4;
  • Sheet E (p. 42): E1, 2v, 3, 4v (D[o] distributed at line 12 or 13 of E4v), (D[i] distributed) E1v, 2, 3v, 4;
  • Sheet F (p. 43): F1, 2v, 3 (E[o] distributed), 4v, 1v, 2, 3v (Perhaps partway through setting F3v the compositor distributed E1v, but the rest of the standing type [in E(i)] seems to have been distributed after F(i) was imposed.), 4;
  • Sheet G (p. 44): G1, 2v, 3 (F1 and 3 distributed), 4v (F2v and 4v distributed), Iv, 2 (F1v and 2 distributed), 3v, 4 (F3v and 4 distributed);
  • Sheet H (p. 45): H1, 2v, 3, 4v (G[o] distributed), 1v, 2, 3v, 4
  • Sheet A (p. 45): A1v (blank) (G[i] distributed), 2, 3v, 4, 1 (title page; H[o] distributed at [i.e., before the setting of] A1 or A2v), 2v, 3, 4v.

Blayney (1982, pp. 92–3), explaining and questioning Turner’s analysis, focuses on the recurrence of individually distinctive types from sheet B in sheet C (question marks indicate types [described by Turner] as doubtful):

C1r 1v 2r 2v 3r 3v 4r 4v
From B(o) 1+? 4 2 2 ?
From B(i) 4 2
It is stated that when type reappears in this manner, composition cannot have been seriatim. The statement (which also applies to the similar evidence in sheet D) is completely untrue. The evidence is perfectly consistent with seriatim setting, with B(o) distributed before or during C1r and B(i) distributed after C3r. And in fact Turner then suggested that most of the sheet was set seriatim, except that C4v preceded C3v. The difficulty is that the evidence shows almost nothing. No matter what order is suggested, the failure to detect B(i) evidence in certain pages has to be trusted to indicate that no such evidence exists, whereas the absence of B(o) evidence has to be ascribed to a failure to detect what is really present. But if types from B(o) can be present but undetected in three (or four) pages, so can types from B(i). Because according to Turner’s assumptions (presented above) it is improbable that type from B(o) could have appeared in C1r, the evidence from that page was rejected. By rejecting different parts, and by filling in the gaps in other ways, the supposed evidence could be made to agree with almost any order of setting.

I do not suggest that Turner’s conclusions are wrong, since they may be right. The point is simply that the setting-order of sheets B-D of A Midsummer Night’s [93] Dream remains unestablished and that not enough evidence has been presented. Typographical evidence which is equally consistent with seriatim setting, setting by formes, and other possible methods; which can be supplemented by guesswork in selected pages from which it is absent; and which can be ignored selectively to suit the needs of an unsupported theory of work-flow, cannot be considered adequate. Nonetheless, Hinman (1965, p. 31) reported that a very general investigation of setting by formes . . . indicate[s] . . . pretty surely . . . MND was set in this manner, but he presented no evidence for his judgment.

In addition to the recurrence of distinctively damaged types, Turner also had recourse to type shortages indicated by the substitution of roman font for italic and small capitals for full capitals. Yet he lacked confidence in type shortages as a guide to establishing setting-order of pages (p. 40): By itself the testimony of shortages is, I believe, less reliable than that of any other bibliographical technique, and explained his reasons at length. He also anticipated Blayney’s criticism: However, the reliability of type shortage evidence can be increased when we evaluate it in the light of type reappearances, but even here we can be forced away from the most desirable position by occasionally having to take into account the evidence of only one or two reappearing types and sometimes having to argue from the non-appearance of type. Both are bad policies because mistakes in individual type identifications are easy to make and reappearances are easy to miss.

Turner used (1962, p. 35 n.) photostats of the Huntington Library copy of MND Q1 to identify distinctively damaged types. Weiss (1988, pp. 239–42) demonstrates the short-comings of this use of such a photostat. Checking Turner’s type identifications against the Folger copy of Q1, Weiss (p. 240) can confirm only 46% of the identifications. Replicating Turner’s analysis (with its questionable assumptions discussed above) using the reliable fraction of his type identifications together with new ones discovered in the examination of the Folger copy, Weiss revises Turner’s account of when during the composition of later formes earlier formes were distributed (pp. 241–2): No contradictions occurred with respect to the distributions of sigs. B and C. The appearance of ligature ft6 from D1:6 at E3.11 is one type-page before the suggested distribution of D(o) after E3. Similarly, k5 from E(o) (E4v:23) appears at F3:2, one page early. More significant differences occur in later sheets[:] the appearance of W4 from the last page of F(i) (F4:33) in the first page of G(o) (G1:8), a full gathering prior to the suggested distribution after the imposition of [242] G(i). . . . With respect to the claims that both formes of sheet G and H(o) were delayed four pages each . . . , r1 from G(o) (G2v:19) appears at H1:9, four pages before the suggested G(o) distribution point at H1v, and the appearance of w4 from G2:19 at H1:25, N2 from G3v:29 at H1:28, d16 from G2:12 at H1:31, ligature sh4 from G2:29 at H1v:1, and d12 from G2:29 at H2:24 indicate that G(i) was distributed before the composition of sig. H rather than after the imposition of H(i). Finally, the implication that sig. A was set in type from G(i) and H(o) without a distribution of H(i) seems incorrect. Appearing in A(o) (the second forme of sig. A to be set) are the following types from H(i): y4 from H2:32 at A2v:20, h7 from H1v:18 at A2v:8, and W4 from H2:32 at A4v:32 (and possibly h8 from H2:20, which may appear in A(i) at A2:19). In short, the evidence suggests a more or less normal sequence of distributions following the completion of each of the later sheets.

Valuable though Weiss’s study is for the quality of its type-identification evidence, it still does not free itself from the assumptions identified in Turner’s work by Blayney and McKenzie. While evidence of type shortages is consistent with the setting-order of the pages suggested to Turner by the type-recurrence, shortages cannot be used to establish setting-order. My inspection of the leaves of the Folger copy of Q1 under raking light fails to reveal any indentations in them such as might have been made by the type metal when the sheets were perfected, and thus fails to establish the order in which formes of sheets were printed. Holland (ed. 1994, p. 113): there is insufficient evidence as yet to establish the setting order.

Revision in Q1

Beginning in the middle of the 19th c., there arises a claim that Sh. revised the text of Q1 one or more times; the narrative of such revision remains somewhat consistent as it is elaborated by successive proponents, with the dialogue associated with the four lovers imagined to survive from Sh.’s earliest version of the play. Verplanck (ed. 1847, 2: Introductory Remarks [to MND], 6 [new pagination for each play and its accessories]): It seems . . . very probable . . . that [MND] was originally written in a very different form from that in which we now have it, several years before the date of the drama in its present shape—that it was subsequently remoulded after a long interval, with the addition of the heroic personages, and all the dialogue between Oberon and Titania, perhaps some alteration of the lower comedy; the rhyming dialogue and the whole perplexity of the Athenian lovers being retained, with slight change, from the more boyish comedy. White (ed. 1857, 4:16–17): It seems that [MND] was produced, in part at least, at an earlier period of Shakespeare’s life than his twenty-ninth year [i.e., 1593]. Although as a whole it . . . abounds in passages worthy even of Shakespeare in his full maturity, it also contains whole Scenes which are hardly worthy of his ’prentice hand . . . [17] and which yet seem to bear the unmistakeable marks of his unmistakeable pen. These scenes are the various interviews between Demetrius and Lysander, Hermia and Helen, in Acts II and III. . . . There seems, therefore, warrant for the opinion that this Dream was one of the very first conceptions of the young poet; . . . perhaps . . . he . . . went from Stratford up to London with it partly written; . . . when there, he found it necessary at first to forego completion of it for labor that would find readier acceptance at the theatre; and . . . afterward, when he had more freedom of choice, he reverted to his early production, and in 1594 worked it up into the form in which it was produced. . . . At least some of the additions might have been made . . . for a performance at Court. . . . Except in the play itself I have no support for this opinion, but I am willing to be alone in it.

Fleay (1878, p. 61): MND probably was recast previously to publication. Idem (1886, pp. 181–6) dates the version for the public stage to (p. 183) the winter of 1592, (p. 181) its present form to 1595. January 26, . . . the date of the marriage of William Stanley, Earl of Derby, and subsequent modifications to produce the court version to (p. 182) the winter of 1594–5. (P. 182): the traces of the play having been altered . . . are numerous. There is a double ending. Robin’s final speech [2207–22] is palpably a stage epilogue, while what precedes from Enter Puck [2153] to [2206] is very appropriate for a marriage entertainment, but scarcely suited to the stage. In Acts iv. and v., again, we find in the speech-prefixes Duke, Duchess, Clown, for Theseus, Hippolita, Bottom: such variations are nearly always marks of alteration, the unnamed characters being anterior in date. In the prose scenes speeches are several times assigned to wrong speakers, another common mark of alteration. (P. 183): wherever Robin occurs in the stage-directions or speech-prefixes scarcely any, if any, alteration has been made; Puck, on the contrary, indicates change. (P. 185): The time-analysis . . . has probably been disturbed by omissions in producing the Court version. [138–265] ought to form, and probably did, in the original play, a separate scene; it certainly does not take place in the palace. To the same cause must be attributed the confusion as to the moon’s age; cf. [222–3] with the opening lines [5–14]: the new moon was an afterthought, and evidently derived from a form of the story in which the first day of the month and the new moon were coincident after the Greek time-reckoning. Idem (1891, 2:194): The play has certainly alternative endings: one a song by Oberon for a marriage, and then Exeunt, with no mark of Puck’s remaining on the stage; the other an Epilogue by Puck, apparently for the Court (cf. gentles in [2213]). It might seem, as the Epilogue is placed last, that the marriage version was the earlier, and so I took it to be when I wrote my Life of Shakespeare [1886, quoted above]; but the compliment to Elizabeth [524–45] was certainly written for the Court; and this passage is essential to the original conduct of the play, which may have been printed from the marriage-version copy, with additions from the Court copy. This would require a date for the marriage subsequent to the Court performance. One version must date 1596, for the weather description [463–92], which can be omitted without in any way affecting the progress of the play, requires that date. I believe this passage was inserted for the Court performance in 1596, that on the public stage having taken place in 1595; but that the marriage presentation, being subsequent to this, was most likely at the union of Southampton and Elizabeth Vernon in 1598–9. Chambers (ed. 1897, pp. 14–15): There are certain indications which make me think that [MND] was also at some period [after its composition in 1594–5] slightly retouched. Two passages, [1204–1384+1] and [1793–1902], show a markedly larger proportion of feminine endings than the rest of the play. In the earlier passages, this may be due merely to the excited state of the speakers, but I cannot resist the suspicion that the opening of act v. shows some traces of later work. See also Luce (1906, p. 157). Noble (1923, p. 58 n. 1): The Quarto did not use italics for songs. My own belief is that the whole of the fairy part in the final episode is a comparatively late addition. Witness the fact that Oberon can sing and lead a chorus in Act V, a faculty of which he evinces no sign in the rest of the play.

Wilson (ed. 1924, pp. 80–153): The Q1 text emerges from three distinct episodes of Shn. composition in 1592–3, 1594–5, and 1598. What remains in Q1 of the 1592 version are the (p. 91) lovers’ scenes—those featuring Helena, Hermia, Demetrius, and Lysander—wherein the psychology is generally as crude as the verse (p. 92) is stiff and antithetical. Wilson dates these scenes to 1592–3 because they include 1849–50, which he, like Fleay (1886, p. 183) reads as an allusion to [Robert] Greene’s death [on 3 September 1592] (p. 94). To Sh.’s 1594–5 revision belong Q1’s mechanicals and Bottom scenes, most of the passages in which Robin is used in stage directions and speech prefixes to designate the character otherwise called Puck (including the epilogue—2207–22), parts of the fairy scenes (Wilson, ed. 1924, pp. 95–6), and the introduction of Hippolyta into 1.1 and 4.1. This revision includes the mechanicals’ concern over frightening the ladies with too realistic a representation of a lion (838–56), which Wilson, again like Fleay (1886, p. 185), takes to be an allusion to a spectacle from the celebration in the Scottish court of Prince Henry’s baptism (30 August 1594) for which it was prudently decided to substitute a blackamoor (Wilson ed. 1924, p. 95) for the lion that was initially to have drawn a triumphal car (ibid.). The revision also includes a description (463–92) of what Wilson takes to be the wet and chilly summer of 1594 (ibid.). Sh.’s final handling of MND in 1598 gives Q1 its mature Shakespearian verse, in which the masterly diction and vigorous sweep . . . introduce a note of intellectual energy that makes the whole glow with poetic genius (p. 183). Some such verse Wilson imagines to have been added in short passages written in the margin of the 1594–5 version, with other longer passages interpolated on additional leaves. For Wilson, as for Fleay, the use of Puck for Robin in stage directions and speech prefixes is peculiar to the 1598 revision and is the clue to its purpose, namely the introduction of the little western flower. All but two occurrences of Puck (1028 and 2153–4 being the exceptions) are associated with references to the flower, which functions as a compliment and representation of Elizabeth Vernon, daughter of Sir John Vernon of Hodnet Hall, Shropshire (p. 100) on the occasion of her marriage to the Earl of Southampton in 1598—the occasion for which Sh. added the wedding masque at 2153–206. Wilson (1962, p. 206) substitutes the wedding of Thomas Berkeley to Elizabeth Carey at Blackfriars on 19th February 1596 for the Southampton-Vernon wedding as the occasion for the second revision.

According to Wilson, the following derive from the 1592–3 version: most of 1.1 (2–265) including Helena’s entry at 24–5; 566–625 in 2.1; 686–717 and 737–811 in 2.2; 1063–1124 and 1146–1221 in 3.2; 1792–1881 (although with additions from 1598) in 5.1. To the 1594–5 belong 3–23, 131–5, and the splicing together of two 1592–3 scenes at 136–7 in 1.1; 1.2 (266–371); the beginning of 2.1 (373–523); all but one line of 3.1 (813–1020); 1021–62 (although this passage was later revised in part) and 1440–1506 in 3.2; 4.1 (1509–1745) though certain parts look like first draft material recopied (p. 131)—Wilson specifies 1624–48, 1711, and 1722–3 as 1594 additions—and three more additions were made in 1598; 4.2 (1746–89); 1882–1985 (but 1890–1904 were added in 1598), 1986–2152 (with three minor additions from 1598), and 2207–22 in 5.1. The 1598 revision consists of 153–9 in 1.1; 524–65 and 626–49 in 2.1; 650–85 and 718–36 in 2.2; 901 in 3.1; 1125–45 and 1222–1439 (which was partially recopied and revised in 1598 [p. 125]) in 3.2; 1586–90, 1604–10, 1690–2 in 4.1; 1797–1800, 1804–9, 1822–5, 1828–31, 1855–7, 1863–7, 1874–5, 1879–80, 1890–1902, 2015–16, 2138–40, 2143–4, and 2153–206 in 5.1. Craig (1931, p. 335): Revision of some sort is unmistakable in the case of A Midsummer Night’s Dream. (See also Craig 1961, p. 108.)

Wilson’s revision hypothesis is not left to collapse under the burden of its accumulated speculation. Reviewing his ed., Chambers (1925, pp. 342–4) finds the case for revision to contain inconsistencies: Professor Wilson . . . thinks that, while I.1 and IV.1 belong substantially to the [earliest version], certain awkwardnesses in the introduction of Hippolyta suggest that she was an afterthought, connected in some obscure way with the indication in [4–5] of a four-day period for the action, which is not consistent with the time-analysis. I do not suppose that he would lay much stress on this, especially as he accepts Hipployta as part of his [earliest-version] substratum of V.1. Wilson, Chambers notes (p. 344), makes much of the variation in the naming of Robin-Puck, but nothing of that of Bottom-Clown. Yet Chambers does agree that 5.1 was revised and compliments Wilson on (p. 343) a valuable bibliographical contribution in his attention to the persistent mislining by the printer of passages in [1798–1800, 1804–9, 1822–5, 1828–31, 1855–7, 1863–7, 1874–5, 1879–80], which suggests that these passages were taken in from cramped marginal alterations in the copy. However, unlike Wilson, Chambers does not think the alleged marginal alterations can be dated years later than the context into which they have been supposedly interpolated. I agree again that the fairy-mask of V.1 [2153–206] and the epilogue of [2207–22] are probably duplicate endings. Still Chambers refuses to acknowledge that the adaptation issuing in this duplication need date from as late as 1598. (See also Chambers 1930, 1:360–1.)

Greg (1942, pp. 124–5), though, altogether rejects Wilson’s theory of revision, and subsequent editors join Greg both in this rejection and in his substitution of so-called foul papers (see here for Greg’s definition of this term) as the explanation for any discrepancies in Q1. According to Greg, whether or no the two endings were written at the same time, it would not be surprising to find both in the foul papers in their present order. Nonetheless, Greg does preserve, with modifications, Wilson’s conception of revision at the beginning of 5.1 (p. 125): On the whole I think the copy for Q must have been the author’s manuscript. . . . The most important piece of evidence is at the beginning of the last act where eight passages of verse are mislined [1798–1800, 1804–9, 1822–5, 1828–31, 1855–7, 1863–7, 1874–5, 1879–80]. Wilson has pointed out that if these are removed what remains is perfectly regular and consecutive, and he argues that they represent marginal additions in the copy. . . . It seems to me quite possible that Shakespeare, coming back to his work in a fresher mood, found what he had written rather flat and sought to brighten it up. And I cannot believe there was anything like the amount of revision Wilson imagines [elsewhere in Q1]: it would certainly have left other traces of the sort, whereas the text is elsewhere particularly clean. Greg later (1955, p. 243) is even more dismissive of Wilson’s theory, calling it all very ingenious; the difficulty is to believe that this refashioning would not have left plainer bibliographical traces than are now apparent. Nonetheless Greg adheres to Wilson’s theory of alleged marginal alteration. Spencer (1930, pp. 24–5) does not agree that, if the eight mislined passages are removed what remains is perfectly regular and consecutive, as Greg put it. Instead, Spencer notes that what remains does not cohere as well as the whole existing text does with the eight passages in place. He argues that it is hardly conceivable that a reviser, expanding certain speeches, should make the new joints less conspicuous than the old, since the new represent elaboration, while the old represent the flow of his thought as originally conceived. For example (p. 25), the description of the interlude in the schedule read by Theseus [1852–3] specifies four qualities: the piece is tedious, brief, tragical, and merry, all these qualities lying within the original text as segregated by Professor Wilson. But in the next speech, which contains Philostrate’s explanation, only tedious and brief lie within Mr Wilson’s [original text: 1858–61]. For tragical and merry we must subjoin . . . one of the additions [1863–7]. Kirschbaum (1946, p. 48) supports with another example Spencer’s supposition that what Wilson calls the original text and what he calls the additions were written at the same time: it will be noticed that in [1796] the sequence is Louers first, mad men second. Omitting the so-called addition [1797–1800], we see that the sequence in [1801–3] is mad man first, louer second. Why the shift in sequence if Shakespeare wrote [1796–1803] originally without . . . [1797–1800]? But when we look at [1799–1800, a supposed addition] . . . we see that the sequence lunatics first, louer second, Poet third is the sequence followed in [the allegedly original 1801–3] . . . and [the allegedly later addition 1804–9]. . . . In other words, the sequence followed in the supposedly original version in [1801–3] is not the sequence first indicated in the supposedly original version at [1796] but the sequence indicated in the supposed marginal addition at [1799–1800]. Thus, the ensuing hypothesis is that both [alleged original and alleged later addition] were written at one and the same time and that the [allegedly additional 1799–1800] . . . was written before and not after the [the allegedly original 1801–3]. And since the sequence indicated in [1799–1800] is followed in [1801–9], it may be suggested that [1804–9], the lines on the poet [a supposed addition], were not an afterthought but were written immediately after [1801–3]. See also Lull (1998) on these alleged Shn. revisions.

Turner (1962, pp. 49–50) seeks to corroborate Wilson’s revision theory in a number of places in Q1, including where the eight passages of verse are mislined. Noting that sigs. C1–3, which, according to Turner, are unusual in being set seriatim, contain, on Wilson’s theory, passages from all three stages of composition (497–681), Turner thinks it possible that the workman was confronted here with particularly nightmarish copy. Again, on Turner’s analysis, work on F(o) and the first two pages of F(i) went slowly; and this is another part of [50] the text ([1371–1506], ending near the foot of F2v) which Wilson thinks to have been considerably worked over. Problems with Turner’s method (discussed above, here) compromise any possibility of his analysis buttressing Wilson’s theory. Other problems obtrude in Turner’s justification of Wilson’s interpretation of the eight mislined passages in sig. G (5.1) as evidence of revision. Werstine (2012, p. 145) both summarizes and criticizes this justification: Close bibliographical analysis of the quarto by Robert K. Turner, Jr., shows that quire G is peculiar not only for the frequency with which its verse is mislined, but also for containing four pages with fewer lines of type than is normal. Usually each page has 35 lines, but sigs. G1r, 1v, and 2v have only thirty-four lines each, and G2r only thirty-two ([Turner] 1962, 39). In all, then, quire G is short six lines of type. Six is also the number of lines of type that are saved as an apparently accidental consequence of the mislining of five of the eight passages of verse that are erroneously divided; three of the mislined passages occupy the same space as they would if they were properly set. No one, including Turner, has remarked on this coincidence or attempted to account for it. Turner attempts to explain away the four short pages as follows: in casting off copy for quire G, the compositor, who, he assumes, cast off his own copy, evidently counted in some material that he later did not set[;] . . . he may have failed to notice that some lines here and there were supposed to be cancelled ([ibid.], 54). Such an explanation fails to convince because throughout Q Turner can find only two other places where there may have been errors in casting off so that the compositor had to juggle the lineation of the text in order to fit copy to a predetermined space (ibid., 55)—B2r, [319–20] and H2r, [2063–4]. His explanation then forces us to believe that Quire G is the unique site not only of a considerable amount of mislined verse and short pages (which, according to Turner, bear no relation whatsoever to the mislined verse) but also of a considerable number of misleadingly cancelled lines in its copy (for which there can be, in the nature of the case, no surviving evidence). Without any adequate explanation for the short pages of quire G, Wilson’s theory about the source of the mislined verse in the quire must remain shrouded in doubt. It is instead possible that whoever cast off copy for quire G found no difficulty in printer’s copy and counted off the lines with the same meticulous accuracy found in almost all the rest of his work, and then the compositor, who sometimes unaccountably, if only occasionally, mislined the text elsewhere (e.g., . . . H1r-H1v [1986–9] . . . ), made the mistakes in dividing verse that Wilson attributes to printer’s copy. Such an explanation may not be the right one, but at least it relates the bibliographical anomaly of the short pages to the textual anomaly of the frequently mislined verse.

After Greg’s dismissal of Wilson’s revision hypothesis, Smidt (1986, pp. 123, 128, 130–4, 140, 210) is reluctant to develop in any detail a theory of revision on the basis of so-called unconformities in Q1: One might suppose that the little Indian is left as a residue from an early attempt to work out an appropriate fairy plot, before the fairies became involved with Duke Theseus and his bride. He has lost his raison d’être as a cause of the fairies’ quarrel, but he is still useful as a means of providing a solution to it. (P. 128): The variation between the names Robin and Puck may well be a sign of different stages of composition, as some scholars have thought, but there may have been merely an expansion of Puck’s character as the writing of the play advanced, not a substitution of a spirit for a gnome. And this may have occurred during a continuous process of composition, so that the fairy Puck is not necessarily a sign of revision. (P. 130): The most interesting phenomenon as far as unconformities are concerned is the fitful appearance of the moon during the night in the wood. . . . [132] It is darkness . . . , only lightened for a while by the stars, that prevails while the lovers are in the wood. . . . There is no reference to present moonlight in the rehearsal scene [813–936]. Yet (p. 133) there is in fact enough indication of a moon in the fairy scenes to make the mention of a new moon on the wedding night [by Hippolyta at 12] well-nigh impossible, in spite of our expectations. . . . [134] If the inconsistency was accidental, was it brought about by the merging of different plot components, Theseus and Hippolyta on the one hand, the fairies on the other hand, the young lovers and the artisans in between? (P. 140): The contradictions in this comedy were not brought about by changes of mind, and, with one possible exception, not by inadvertence. The exception is the moonlit fairy scenes. Only (p. 210) the repetition of Theseus’s order to Demetrius and Egeus to accompany him [123–5 and 132–5] suggests some kind of textual disturbance. For reengagement with Wilson’s theory of revision, see Hunter (1998 and 2002, pp. 3–6).

Printer’s Copy for Q1

Speculation on this question has given rise to four suggestions, all of them testifying to perceptions of the high authority of the Q1 text: (1) Sh.’s own MS; (2) Sh.’s own MS as marked up by his company for production; (3) a transcript of Sh.’s MS made and used in the playhouse; (4) a non-theatrical transcript. The cautious commentators have found language to avoid or at least to qualify precise identification.

Discussion begins with Capell, who seems to opt for the third alternative. He (ed. 1768, 1:3 ff.) identifies fourteen quarto texts, including MND, that ought to be excluded from Heminge and Condell’s characterization of all the Shakespeare quartos as diuerse stolne, and surreptitious copies, maimed, and deformed by the frauds and stealthes of iniurious imposters (p. 10): Let it then be granted, that these quarto’s [sic] are the Poet’s own copies, however they were come by; hastily written at first, and issuing from presses most of them as corrupt and licentious as can any where be produc’d, and not overseen by himself, nor by [11] any of his friends. . . . It may be true, that they were stoln; but stoln from the Author’s copies, by transcribers who found means to get at them: and maim’d they must needs be, in respect of their many alterations after the first performance. . . . [12] The very errors and faults of these quarto’s . . . are, with the editor [Capell], proofs of their genuineness; For from what hand, but that of the Author himself, could come those seemingly-strange repetitions [of passages in LLL and Tro.], . . . those imperfect entries . . . ? Capell’s use of transcribers suggests that he thinks copy to have been scribal transcripts, and his reference to their many alterations after the first performance locates transcription in the playhouse, where copy would be subject to theatrical adaptation.

Similarly Malone (in Boswell, ed. 1821, 1:203) in general recognizes the superiority of the early quartos to counterpart F1 texts, except for Wiv. and H5 (Q1 Ham. was not discovered until 1823), yet does not attempt to specify just what kind of MSS served as printer’s copy: With respect to the other thirteen copies [quartos, including MND] . . . , they in general are preferable to the exhibition of the same plays in the folio; for this plain reason, because, instead of printing these plays from a manuscript, the editors of the folio . . . printed the greater part of them from the very copies which they represented as maimed and imperfect, and frequently from a late, instead of the earliest, edition; in some instances with additions and alterations of their own.

Clark and Wright (ed. 1863, 2:viii), cautiously subjunctive: Fisher’s edition . . . may have been taken from the author’s manuscript. Furness (ed. 1895, p. xii) takes pains to imagine that Heminge and Condell may not have been guilty of a wilful untruth, as alleged by Malone, when they implied that they provided for F1 only his [Sh.’s] papers as printer’s copy, but, in fact, provided quarto copy for such plays as MND if they knew that this [quarto] text was [originally] printed directly from his manuscript. Chambers (ed. 1897, p. 8), leaving open all alternatives for printer’s copy: Q1 has been printed from a clear and authentic manuscript. Cuningham (ed. 1905, p. xiii) adds to the alternatives the possible use of actors’ parts in the creation of printer’s copy, an idea apparently borrowed ultimately from Johnson’s 1756 Proposals for Printing . . . the Dramatick Works of William Shakespeare (ed. Arthur Sherbo, 1968, 7:52): Q1 was printed in all probability, either from the authentic MS. of Shakespeare himself, or at least from an accurate copy or, perhaps, copies of the actors’ parts, transcribed in the theatre from the original MS.

Furness (ed. 1895, p. xi) misleads a number of his successors with the observation that in both Q1 and Q2 the stage directions are, as in copies used on the stage, in the imperative, such as wind horns, [1622] sleep [1484]. So Cuningham (ed. 1905, xv) notes stage-directions . . . in the imperative, as is customary in stage copies as his ground for suggesting the possible playhouse provenance of printer’s copy. And Pollard (1909, p. 72) declares that The imperative form of the stage directions, Ly doune ([1110]) and Winde hornes ([1622]) may be taken as indicating its origin from a playhouse copy. (See Idem 1920, p. 64.) Greg (1942, p. 37) properly questions the assumption that the prompter can be identified by the use of the imperative: The prompter writes directions for his own use; they are generally terse and to the point. Chambers [1930, 1:118] questions whether they are usually in the imperative. They are not: but being short and curt they tend to imperative and participial constructions. Wilson (1945, p. 67) provides examples of the imperative in Anthony Munday’s hand, not the hands of the theatrical annotator(s), in the theatrical MS of Iohn A kent & Iohn a Cumber; Werstine (2012, p. 229) gives examples of imperative SDD in authorial hands from MSS of Thomas Heywood’s The Captives and the anonymous The Waspe. Greg (1942, p. 125), contradicting himself, follows Pollard in identifying the hand of the prompter in the appearance of imperative SDD: There are however [in Q1] a few directions that suggest the prompter, such as Lie down and Wind horn. In the palpable duplication, Enter Lovers; Lysander, Demetrius, Hermia, and Helena ([1819–20]), we may see the book-keeper expanding a typically brief direction of the author’s. (Werstine [2012, p. 131] notes that actual theatrical texts rarely show a bookkeeper specifying a group in terms of their proper names, finding only one example, from the annotated quarto of Thomas Lodge and Robert Greene’s A Looking Glasse, for London and England.) Greg differs from Pollard, though, in judging printer’s copy to be, not a playhouse copy, but so-called foul papers (see here for Greg’s definition of this term) containing (1942, p. 125) notes made in preparation for the prompt-book. Greg’s belief in the possibility of such a document once having existed arises from his interpretation of a single extant MS, Thomas Heywood’s transcription of The Captives, as such a document, but as Werstine (2012, pp. 300–9) observes, this interpretation is at odds with features of the MS. Greg rules out (1942, 125) a playhouse transcript to be printer’s copy only because he believes that there could only ever have been one such transcript and that it was used to annotate printer’s copy for F, which differs from Q1 chiefly in terms of such annotations.

Pollard (1920, p. 63), while not setting aside his belief in theatrical annotation of printer’s copy, anticipated Greg in arguing as well for the possibility that such annotated copy could originally have been inscribed by Sh.: Possibly in some cases, if [a dramatist] were familiar with the theatre, he might use the same technical language as a prompter, so that Shakespeare himself, in the scene in the wood in the Midsummer Night’s Dream, may have written the directions, Enter a Faerie at one doore and Robin goodfellow at another, Enter the King of Fairies at one doore, with his traine; and the Queen at another with hers, the doors, of course, being those of the stage, not of the wood. Adams (1923, p. 519) states an opinion that can be interpreted to be identical to Pollard’s but need not necessarily be: MND is printed from authentic playhouse copy; compare Neilson & Hill, ed. 1942, p. 88. Pollard (1923, p. 7): some of the flaws in these Good Quartos [including MND Q1] are the result of imperfections in Shakespeare’s own work, and I have ventured to claim that some of these Good Quartos may actually have been set up from Shakespeare’s autograph manuscripts. Wilson (ed. 1924, p. 80) is of Pollard’s mind: beyond doubt [Q1 is] printed from a theatrical prompt-book, . . . [with] the managerial voice giving real directions to the players; he quotes 1110, 1484, and 1622. Printer’s copy also contains irregularities strongly suggestive of an author’s manuscript, and so it is both Shakespeare’s autograph manuscript and the prompt-book just as Shakespeare left it. Kittredge (ed. 1936, p. 229) and Parrott (ed. 1938, p. 131) follow Wilson. Chambers (1930, 1:358), persuaded to be more specific than in 1897: Q1 may be from the author’s manuscript. Kirschbaum (1955, pp. 172–3) stands out against the idea of theatrical annotation of a Sh. MS: Greg’s evidence for prompter’s additions does not pass muster. The directions could just as well come from the author, and there is no reason why they should not be copied by a scribe. (P. 173): The nomenclature [variation in naming of characters in SPP and SDD] . . . does not show the author in the heat of composition—does not cause the reader to assume foul papers behind the print. . . . There is little evidence of foul papers or playhouse in Q. Nothing in it rules out printing from transcript. . . . There is neither internal nor external evidence to show that the copy came to the publisher from Shakespeare’s fellows.

In 1955 Greg establishes what becomes a virtual consensus among 20th-c. editors that printer’s copy for Q1 is authorial foul papers defined as (1955, pp. 106, 142) a copy representing the play more or less as the author intended it to stand, but not itself clear or tidy enough to serve as a prompt-book, that is, a theatrical MS used to guide performance, because it contained (p. 142) loose ends and false starts and unresolved confusions. Greg presumed that book-keepers necessarily tidied away from prompt-books certain features of foul papers, including seven features still to be found in Q1 (ibid., pp. 240–2):

  • (1)multiple designations of the same character (p. 241): Oberon is King of Fairies (or simply King) or Oberon indifferently; Titania is of course named in the text [i.e., dialogue and SPP], but in directions [the proper name] appears only at [650] on her second entry [otherwise she is named Queene in SDD]; Bottom, on his most important appearance [1509] is merely Clowne; Theseus and Hippolyta, after long appearing by name, become as a rule Duke and Duchess after the play begins in Act V; lastly Robin (Goodfellow) and the generic Puck alternate;
  • (2)indefinite entrances involving speakers (p. 240): after carefully naming the rude mechanicals in I.ii, the author later contents himself with the description the Clownes (III.i) or the rabble (IV.ii); the young couples, having been named in I.i, become simply Louers at V.i . . . , for here the names [i.e., Lysander, Demetrius, Hermia, and Helena] are a palpable addition; the fairies are quite indefinite, except for the foure at [980]. Greg also quotes as allegedly bearing the characteristics of the author and thereby implicitly needing the bookkeeper’s attention SDD containing the term traine to refer to speakers attending royalty (p. 240: her [i.e., Titania’s] traine (650); all his [i.e., Theseus’s] traine (1622);
  • (3)other indefinite SDD: Greg’s characteristically authorial SDD include (p. 240) Enter Theseus, Hippolita, with others (2), where others refers to supers, rather than speakers, including Philostrate, addressed by name in this scene;
  • (4)inclusion of an unnecessary character in a SD (p. 241): Pucke is made to enter with Oberon at [1021], though in fact he only does so three lines later;
  • (5)missing entrances and exits;
  • (6)descriptive SDD: characteristic of the author, according to Greg, are, for example (p. 240), Enter Egeus and his daughter Hermia . . . and Enter Quince, the Carpenter; and Snugge, the Ioyner . . . ;
  • (7)marginal insertions of dialogue (pp. 241–2): Greg fully endorses the theory of J. Dover Wilson’s that some passages of verse at the beginning of the play’s last scene are wrongly divided in Q1 because they were marginal additions to the dialogue, their position in the margins forbidding their writer from dividing them properly as verse and the compositor following his copy in this error (see here).

Greg also mentions the following incidental features of Quarto MND as characteristic of foul papers: erroneous SPP—for Greg, one of Shakespeare’s (ibid., p. 247 Note A) oversights in composition, the author having written consecutive speeches for Flute and Thisbe, forgetting that they were the same; double entrance (p. 240): Helena enters in 1.1 both at 25 and at 191; a dialogue error in naming (ibid., p. 246 Note A): Flute for Snout at 1957; and an erroneous SD (ibid., 240): Enter Quince, Flute, Thisby and the rabble (1746)—in this scene Shakespeare had forgotten that Flute and Thisbe are one.

Werstine (2012, p. 132): The appearance of these features in the earliest printing of A Midsummer Night’s Dream leads Greg to conclude that we can hardly imagine that Q represents a finished prompt-book [1955, p. 241] and therefore, by this process of elimination of what he regards . . . as the only possible alternative, Quarto Dream must represent Shakespeare’s foul papers—in spite of the persistence of the same features in the actual theatrical texts that he calls promptbooks, which destroys his argument. Werstine reviews each of Greg’s features and compares them to what is found in the twenty-one extant texts (both MS and annotated quartos) that bear theatrical annotation and thereby show what book-keepers actually did and did not do to their playbooks (these texts are described in Werstine 2012, pp. 234–57). In these twenty-one Werstine finds (1) multiple designations of the same character (pp. 359–64), (2) indefinite entrances involving speakers, including the very terms cited by Greg to discount a theatrical MS—crewe, trayne, the rest (pp. 375–9), (3) other indefinite SDD, including uses of others identical to those cited from MND by Greg (pp. 379–82), (4) inclusion of an unnecessary character in a SD (p. 384), (5) missing entrances (pp. 374–5) and exits (pp. 386–8), (6) descriptive SDD such as, from the scribal MS Ironside: Enter Edmond and Alfricke the generall vnder the kinge:/ (1.3.332) and Enter Edricke a poore man . . . (2.2.461). (Needlessly explanatory SDD, like those that Greg quotes from the MND quarto, are also frequent in The Second Maidens Tragedy [also called The Lady’s Tragedy], another scribal MS, in further indication that such SDD are by no means peculiar to authorial MSS: Enter the new Vsurping Tirant; The Nobles of his faction, Memphonius, Sophonirus, Heluetius with others, The right heire Gouianus depos’de [1.1.1–3]; Enter L Anselmus the deposde kinges brother, wth | his Frend Votarius [1.2.257–8]; Enter the ladye of Gouianus . . . [2.1.636]; Enter Tirant wondrous discontedly: Nobles afarr of [4.2.1655–6]; Enter Votarius with Anselmus the Husband [5.1.1984]); (7) marginal insertions of dialogue (pp. 388–9). Even the minor features cited by Greg to identify foul papers as printer’s copy are located in theatrical MSS by Werstine: erroneous SPP (pp. 371–2), double entrances (pp. 385–6), a dialogue error in naming (p. 371), and erroneous SDD (p. 385).

Greg himself showed his awareness of the flaw in his reasoning when he wrote (1955, p. 142) It must, however, be recognized that owing to the casual ways of book-keepers these characteristics may persist, to some extent at least, in the prompt-book. In reproducing his argument and conclusions about copy for MND Q1, editors fail to attend to this caveat: Doran (ed. 1959, p. 174); Wells (ed. 1967, p. 165); Evans (ed. 1974, p. 247); Brooks (ed. 1979, pp. xxii–v); Foakes (ed. 1984, pp. 135–6); Wells & Taylor (1987, p. 279); Holland (ed. 1994, pp. 113–14); Berger (ed. 1995, p. ix). Horwood (ed. 1939, p. 10) accepts what would become Greg’s argument and conclusion in the form in which it was first presented in McKerrow (1935)—Greg’s inspiration—with reference only to variation in naming in SPP and SDD. Greg’s definition of foul papers has also been overturned in Werstine (2009, pp. 44–5).

A significant problem with Greg’s use of variation in naming of characters in SPP in MND Q1 and other texts as evidence of so-called foul papers had already been identified by Kennedy (1998, pp. 178–9): There may, however, be another explanation for the variation in SP’s in some of Shakespeare’s early texts. The change may not be authorial at all, but compositorial. It seems to have been a printing-house convention that a compositor did not have to follow copy in the matter of SP’s, but could choose to call characters by their first names or last names, or [179] generic names or personal names, or by their functions or peculiarities. If he needed to, he could vary the SP from Quin. to Pet[er], from The[seus] to Duke, . . . and so on. Most of the time, variant SP’s do not point to authorial foul papers, but signify compositorial change. And most of the time, variant SP’s are not signs of an author’s revising, or of an author in the heat of composition, but are rather indications of a compositor switching SP’s because of type shortage. Kennedy applies this theory in detail to the variation in naming in MND Q1’s SPP (179–90).

To support the view that printer’s copy is in Sh.’s own hand, some recent editors have followed Wilson (ed. 1924, pp. 112, 116, 121, 148) in adducing certain spellings in MND Q1 as the same as or somewhat analogous to spellings in the Hand-D pages of MS The Booke of Sir Thomas Moore (BL Harl. 7368), in the belief that Hand D is Sh.: Evans (ed. 1974, p. 247), Brooks (ed. 1979, pp. xxv–vi), Foakes (ed. 1984, p. 136), Wells & Taylor (1987, p. 279), Holland (ed. 1994, p. 114); Berger (ed. 1995, p. ix). However, recent review of Hand D’s spelling by Jackson (2007) shows that only a half-dozen are sufficiently rare to constitute acceptable evidence for attribution, and none of these is in MND Q1.

There is no evidence for a rational choice among the following alternatives for MND Q1 printer’s copy: (1) Sh.’s own MS; (2) Sh.’s own MS as marked up by his company for production; (3) a transcript of Sh.’s MS made and used in the playhouse; (4) a non-theatrical transcript. While there is nothing in the way of playhouse notes in the quarto to demonstrate alternatives 2 and 3, Werstine (2012, p. 4) shows that in some actual playhouse MSS there is so little annotation that it is possible that MSS with no annotation could have been used in production.

The Second Quarto (1619)

There is no entry for Q2 in the Stationers’ Register, as is the case with a great many books published in the late 16th and early 17th c. (see Blayney 1997, pp. 400–5). Erne (2003, pp. 255–8) suggests that Heminge and Condell were referring specifically to the whole series of plays among which MND Q2 appeared in 1619 when the two actors wrote of the stolne, and surreptitious copies in their prefatory remarks to the First Folio (1623)—if these remarks were of their composition. However, Heminge and Condell fail to supply any justification for Erne’s specification, and they provide MND Q2 as copy for the Folio printing.

Q2’s title page, as transcribed by Greg (BEPD, 1:169), reads as follows: [ornament] | A | Midsommer nights | dreame. | As it hath beene sundry times pub- |likely acted, by the Right Honoura- |ble, the Lord Chamberlaine his |seruants. | VVritten by VVilliam Shakespeare. | [device <McKerrow>283] | Printed by Iames Roberts, 1600.

Bartlett & Pollard (1939, pp. 71–3) identify and locate thirty extant copies of Q2. Modern facsimiles include that of William Griggs, ed. J. W. Ebsworth, 1880. For a digital facsimile see the Shakespeare Quartos Archive (www.quartos.org).

The Printing of Q2

Except for its slightly different top lace border, its altogether different mid-page printer’s ornament, and its substitution of reference to Roberts for that to Fisher, the wording and even the alternation of roman and italic fonts of Q2’s title page duplicate Q1’s. Thus it falsifies its date of printing and its printer, and thereby created uncertainty (see above here) and gave rise to dispute (see below). Three centuries passed before the discovery that Q2 was in fact printed in 1619 by William Jaggard for Thomas Pavier. Pollard (1909, p. 93): the mid-page printer’s ornament William Jaggard was using . . . in other books both before and after 1619. The resemblance between Q1 and Q2 title pages continues on sig. A2, the first page of the play’s text. In both quartos that page is surmounted by A | MIDSOMMER NIGHTS | DREAME. followed by a centered SD and then by a centered SP.

Because the states of formes in Q1 differ only by typographical errors that are easily noticed and corrected, it is impossible to identify the states of the formes in the copy of Q1 used as printer’s copy for Q2. Pollard (1923, p. 5): Taking each intermediate edition [like this 1619 Pavier quarto of MND] by itself, in no single instance do we find evidence of the sort of care which could lead us to believe that its overseer had obtained access to any authoritative source. . . . As evidence of the words which Shakespeare wrote or of the words which were spoken by the actors engaged in his plays these intermediate editions are absolutely worthless, except where we possess only one or two copies of the First Edition. . . . A Second Edition might . . . be printed from a copy of the First in which a correction had been made which does not appear in any copy of the First now extant. Pollard’s only one or two is optimistic; even when there are eight extant copies of Q1, we may not expect to find among them all the different states of correction of their formes and therefore all the states that may have been present in the copy of Q1 used to print Q2.

The Pavier quarto of 1619 is, for the most part, a page-for-page reprint of Q1. Clark & Wright (ed. 1863, 2:viii): On comparing these two Quartos we find that they correspond page for page, though not line for line, except in the first five pages of sheet G. In Q1 the first four pages of this sheet contain fewer typographical lines than the thirty-five found on the rest of the pages. In Q2 these four pages each contain 35 typographical lines. This regularization is effected sometimes by the transfer of lines from later pages to earlier ones, sometimes by the chopping up of verse into shorter lines, sometimes by the correct division of verse that is mislined in Q1, and sometimes by the addition of white space around SDD. Like Q1, Q2 is divided into neither acts nor scenes. Q2 has a different tailpiece on H4v from Q1’s; Greg (BEPD 1:169): The ornament on H4v [of Q2] is a copy of device 179.

Dyce (ed. 1857, 1:clxiii): Roberts’s [Q2] is the less accurate quarto. Clark & Wright (ed. 1863, 2:viii): The printer’s errors in Fisher’s [Q1] edition are corrected in that issued by Roberts [Q2], and . . . in the Roberts Quarto the Exits are more frequently marked. Ebsworth (ed. Q2 1880, pp. ix–xiii): In Fisher’s [Q1], the business [i.e., the SDD] is given (as usual) in Italic type, with exception of the proper names of the characters; which are in Roman type. But in Roberts’s [Q2], the whole line is in Italic type, names and all. (P. x): Roberts’s page [is] wider than Fisher’s to the extent of about two letters’ breadth [The measure in Q1 is 82 mm; that in Q2 87 mm.]. And it is remarkable that when . . . difference [in line-for-line reproduction] ensued . . . a recurrence has been speedily made to the former agreement. (P. xiii): The spelling of Q2 is more modern than Q1’s: We give a brief sample of these differences in corresponding places; but they are innumerable throughout: — Roberts’s Quarto: tell — Snug — else — home-spuns — perhaps — hue — eke — Iew — Snowt — do — hog — Finch — Sparrow — answer — lye — he, etc. . . . Fisher’s Quarto: tel — Snugge — els — homespunnes — perhappes — hewe — eeke — Iewe — Snowte — doe — hogge — Fynch — Sparrowe — answere — ly — hee etc. [845–952]. Ebsworth also notes some contractions such as trēble, for tremble [852]; lātern, for lantern [871]; chābre, for chamber [873]; vnderstād, for vnderstand [903]; trāslated, for translated [935–6] in Q1 that are expanded in Q2. It is by no means difficult to understand the improved clearness in typography of Roberts over that of Fisher (supposing, as we do, that Roberts had Fisher’s printed book before his eyes). For there was the additional space gained—1. By the excision of redundant letters; 2. By having a wider platform of type in his page; 3. By his gaining an occasional line in prose passages, and thus being able to afford extra leads at entrance of characters. Despite this improvement in typographical clearness, there is a marked deterioration in the minute divisions of the verse by punctuation. Commas are less frequent, either from negligence or from systematic repugnance to the scholarly and grammatical breaking-up of sentences. Furness (ed. 1895, p. xi–xv), comparing Q2 to Q1: The Second Quarto . . . has the fairer page, with type fresh and clear. (P. xv): In . . . Q1 there are about fifty-six stage-directions; in . . . Q2 about seventy-four. Chambers (ed. 1897, p. 168): Q2 is printed from Q1. . . . it is set up with greater attention to typographical details. . . . And where the typographical correspondence of the two editions gets out, the spacing of Q2 is always arranged so as to recover it as soon as possible. The printer is evidently working from a model. Cuningham (ed. 1905, p. xi): Q 2 corrects some of the mistakes in Q 1; but, on the other hand, it commits more than it corrects. Rhodes (1923, p. 64): the additions in Roberts’ are of small importance, being commonly the mark of Exit when it is quite clear from what the actor said that he was leaving the stage.

Brooks (ed. 1979, pp. xxvii–viii) cites further bibliographical evidence of [Q2’s] derivation [from Q1]: reproduction of such peculiarities as the spelling wodde for wood [261; here Brooks appears to be in error; unless there is an unrecorded press variant, Q2 reads wood] and the omission of Enter before Robin and Demetrius [1465]; its printing of The. In himselfe he is [62], Enter Snout [929], Enter Lysander [1440], Enter Thisbie [1989], and Lyon. Oh [2064] just as Q1 has them, tucked in on the same line as the end of the preceding speech; [Q2’s] agreement with Q1 in capitalizations which are not simply those to be expected; and above all its concurrence in thirty-five of the speech-prefixes where Q1 varies the form of the abbreviation. In II.i, for example, where Q1 has Ob. three times, then Oberon, Ob. four times more, then Oberon again; and, also exceptionally, for Demetrius’ sixth speech, Demet. Q2 follows suit. Full collation shows that it repeats all but thirteen of the verbal errors made in Q1. It corrects prose set as verse at [911–12], but not at [1986–9]. A [xxviii] half-hearted attempt is made to rectify some of the misdivided verse in V.i between [1797] and [1880], but most of it, like the misdivided [490–1] in II.I, is reprinted as it stands.

. . . Of [Q2’s] thirteen corrections, four eliminate obvious literal misprints. In the remainder the errors announce themselves: the misreading of waves for wanes [7], and of Cet. for Bot. [867]; the displacement of t from comfor to bet [691], the omission of an o from good [695], and of to before expound [1734] where the sense requires it; a mistake of number in gentleman [1333]; a failure to repeat is after this [1649]; an assimilation of is to knit [699]; and a catching of yet from earlier in the phrase [2104]. They needed nothing beyond the context in Q1 itself either to draw attention to them or to indicate the proper correction. Apart from the accidentals of spelling and the like, Q2 differs from Q1 only through the guesswork which furnished these corrections, and by over sixty new errors of the printing-house. Since it derives from the author only through Q1, its readings have no independent authority.

See the textual notes of this edition for all significant differences between Q1 and Q2, including SPP, SDD, lineation, punctuation, and verbal variants.

Date and Auspices of Q2

Editors struggled to determine the priority of the two editions, both dated 1600 on their title-pages. (See here.) Although most judged correctly that Fisher’s edition was the earlier, disputes arose. Halliwell-Phillipps (ed. 1856, 5:11): Perhaps Fisher’s edition, which on the whole, seems to be more correct than the other, was printed from a corrected copy of that published by Roberts. Fleay (1891, 2:178–9): The consensus of critical opinion is that Roberts pirated his copy from the earlier Fisher edition; but it would be a unique phenomenon had this been allowed to pass without inhibition or, at least, protest. All the evidence lies the other way. Better readings are usually found in later editions, whenever these are produced in the lifetime of the author. Printer’s errors are far more likely to have been introduced than corrected in a second edition. . . . It seems to me far more likely that Roberts printed the play for Fisher, who did not, for some reason unknown to us, care to put his name on the first issue; but finding the edition quickly exhausted, and the play popular, he then appended his name as publisher.

Only in the early 20th c. was Q2 MND correctly dated. Knowles (2020, pp. 1116–18): it was identified as part of a group of plays printed in the same year (1619) though bearing title pages dated from 1600 to 1619. That curious and rather shabby collection (Greg, 1955, p. 12) of plays known or sometimes thought to be by Shakespeare—Parts 1 and 2 of The Whole Contention betweene . . . Lancaster and York (2H6 and 3H6), Per., A Yorkshire Tragedy, MV, Wiv., MND, Lr., H5, and 1 Sir John Oldcastle—had been reprinted in 1619 from earlier quartos or octavos originally issued by a variety of printers and publishers; this new collection, now known as the Pavier Quartos, was printed by William Jaggard (whose shop would soon print F1) for the publisher Thomas Pavier, who was apparently planning to bring out a collection of Shakespeare’s plays. Several bound collections of these ten plays have in fact survived. The brilliant literary sleuthing that revealed the truth behind their several falsified title pages has been recounted by Richard Altick in The Scholar Adventurers (1950, pp. 189–95).

Greg (1908, pp. 113–31, 381–409) first proposed that the ten quartos bound together in a 17th-c. binding were not remainders—three were dated 1600, two 1608, three 1619—but were all actually printed in 1619 despite the differing dates on their title-pages. Eight had a printer’s device and numerals not used until 1610 and a type font not used until 1617, and all were printed on the same papers, bearing the same group of watermarks, which would not have been available over a span of nineteen years. (On rare Pavier watermarks bearing dates of either 1617 or 1619, see Stevenson [1951–2]). In each quarto the printer imitated an original edition. Greg inferred that initially three of these editions were printed in and dated 1619, but that when Pavier for some reason got nervous about his undertaking he issued others under their original dates, possibly seeming to sell off the remainders of editions printed years before by other publishers in order to avoid challenges to copyright. Pollard (1909, pp. 81–104) reports that he and Greg became suspicious of the quartos because they did not specify the printer, publisher, and publisher’s address, but rather the initials T. P. (for Thomas Pavier) on five of the title-pages; because around 1619 William Jaggard was using two of the printer’s devices appearing repeatedly in this group of quartos; because a font of Roman type used in the suspect quartos was a new kind also used in F1 in 1623; and because the spelling in each suspect quarto was generally more modern than in its (older) counterpart, evidently reflecting the habits of Jaggard’s compositors. The clinching proof was provided by Neidig (1910, pp. 145 ff.), who showed by photographic overlays that seven of the nine title-pages were printed in part from the same setting of type, parts of which were transferred from one title-page to another; these therefore were (p. 154) not printed nineteen years apart, but within a few days of each other. The order of printing that he established for these title-pages—WC, YT, Per., MV, Wiv., Lr., H5, and SJO—has been generally accepted as the order of printing of the plays themselves, with the exception that Per. follows WC, with which it shares continuous signatures. On the evidence of watermarks Blayney (1972, pp. 196–7) provisionally places MND before Lr., and Knowles (1982, p. 195) has found the supporting evidence that a number of distinctive types in the last three sheets of MND appear in the first three sheets of Lr. Wiv. is printed in a larger and different font than that used in Lr.

Chambers (1930, 1:134–7): The Contention and Pericles have continuous signatures and were clearly designed for issue together. . . . William Jaggard succeeded to the printing business of James Roberts about 1608, and by 1617 had associated in it his son Isaac Jaggard. . . . The reprinting of 1619 was no doubt done in concert with Pavier, who owned the copyright of five of the plays. . . . Presumably licence was obtained from Johnson for the use of Merry Wives of Windsor, and from Butter for that of King Lear. Of the other three, Midsummer-Night’s Dream was probably derelict, and Merchant of Venice may have been believed to be so. Blount’s registration of Pericles had already been overlooked, and there is nothing to show that Gosson had any copyright. The shortened imprints suggest that the title-pages were originally meant for half-titles in a comprehensive volume, which would naturally begin with a general and more explicit title-page. . . . It was nothing to Pavier and Jaggard that they were reprinting bad texts and ascribing to Shakespeare plays that were not his. Perhaps Shakespeare’s fellows viewed [136] such proceedings with less equanimity. On May 1619 a letter was addressed by the Lord Chamberlain to the Stationers’ Company directing that none of the King’s men’s plays should be printed without some of their consents. Its exact terms are not preserved. But they appear to be recited in a letter of similar import written on 10 June 1637 by Philip Earl of Pembroke, then Lord Chamberlain, and brother of William Earl of Pembroke, who was Lord Chamberlain in 1619, asking the Stationers to stay publication of any King’s men plays without their consents. Pavier and Jaggard [137] may have issued all the ten plays. It is perhaps more likely that they had already abandoned the continuous signatures and perhaps the idea of a comprehensive volume, had separately issued those dated 1619, and had the rest ready in print. . . . Whatever the events of 1619, they can have left no enduring malice between the King’s men and the Jaggards, since it was again from their press that the collection . . . the First Folio came. From the facts of Pavier’s life—that he was at the time publishing religious books rather than plays and was just entering the governing councils of the Stationers’ Company—Johnson (1992, pp. 35–40) concludes that Jaggard, not Pavier, was the instigator of a straightforward scheme to put into print a collection of as many plays as were available, not a complete collection, and that Pavier collaborated by lending the copyrights on the plays he owned (WC, YT, SJO, H5, apparently Per.), by negotiating permissions for Lr. and Wiv., and by assuming or appropriating rights to the more-or-less derelict MV and MND. The faked imprints, Johnson suspects, were intended not to deceive the copyright holders but to avoid protest by the acting company or their agents, who ultimately may have concluded that plays being offered as old goods offered no competition to their planned new and improved Folio. Whether Pavier and Jaggard had conceived of their enterprise as a straightforward business venture, exactly why and when and how the players and other publishers may have objected to the project, what effect Pembroke’s letter may have had, and how William Jaggard and his son Isaac were persuaded to transfer their attentions to the larger project of the First Folio of 1623 have been much speculated upon and discussed; see Pollard (1909, pp. 100–4), Greg (1924, pp. 139–44), Greg (1955, pp. 9–17), . . . Greg (BEPD, 1957, 3:1107–8). Blayney (privately): Since Jaggard’s name did not appear on any of the 1619 quartos, there’s no reason to suppose that any of the players ever guessed that the culprit was the printer they knew best (because of his playbill monopoly). So that even if it had been the players who chose Jaggard to print the Folio (as it almost certainly wasn’t), we can’t assume that they’d forgiven him for a known transgression. . . . We shouldn’t credit them with knowing all that we know.

Kirschbaum (1955, 240–1) speculates about Jaggard’s particular circumstances and strategy in printing MND Q2: The play was derelict copy [Fisher, publisher of Q1, having disappeared from the Stationers’ Company without transferring his right to publish the book to any other member]; in order to publish such copy, it was necessary to obtain the Stationers’ Company’s permission. For the stationers’ guild to grant Jaggard the right to print this play in 1619 might be construed by King’s men as an act in direct defiance of their interests. [Greg (1955, p. 24) quotes the Stationers’ Court-Book C: vppon a letter from the right honorable the Lord Chamberleyne It is thought fit & so ordered That no playes that his Maiestyes players do play shalbe printed without consent of somme of them.] (P. 241): Jaggard decided not to try to establish copyright in [MND] at Stationers’ Hall but to print the derelict copy with a false publisher and a false date. This was the only alternative to issuing it with a 1619 date, a procedure which might have led to some kind of trouble. Jaggard’s apprehensions may perhaps be gauged by the supposition that his edition of [MND] purports to be not an edition different from another published in 1600 . . . but a second issue of the edition bearing Fisher’s name. It was not uncommon for a single edition to be sold by two or more publishers, each publisher having his name only on the title page of the issue he sold. How fortunate, therefore, for Jaggard to find in his shop a large stock of unsold copies of [MND] published in 1600 by his predecessor, Roberts! According to Kirschbaum (ibid., p. 250), Jaggard made his MND look like a different issue of the authentic Fisher edition or like an edition closely copying and closely succeeding the Fisher edition. However, Kirschbaum’s speculation, like Johnson’s above, casts Jaggard in the role of Q2’s publisher, rather than, as he was, only its printer, who may therefore have had no concerns about rights in the copy or about the King’s Men.

Massai (2007, pp. 112–19), departing from earlier scholars’ emphasis on Pavier and Jaggard’s deception of their fellow stationers and the King’s Men, instead proposes that publication of the Pavier quartos may have been part of a larger marketing scheme devised by these two stationers for their mutual benefit in selling first these quartos and then, if successful, the 1623 Folio. Massai associates the Pavier quartos with several other 17th-c. nonce collections of plays by single dramatists that brought together previously published editions of plays with editions just published for the collection. For example, (p. 116) the 1607 re-issue of Sir William Alexander’s The Monarchicke Tragedies (STC 344) includes two additional plays, . . . both dated 1607, and two plays originally issued in the 1604 edition of The Monarchick Tragedies (STC 343). While one of these two plays has no individual title page, the other retains the original one and the date in the imprint is unchanged. With some title pages dated 1619 and others earlier, the 1619 Pavier quartos may resemble such collections. Jaggard’s (p. 118) advantage in leading Pavier’s prospective readers to believe that they were [119] offered the scattered remains of a recently deceased playwright whose works had not been published since 1615 can probably best be described as a pre-publicity stunt. . . . Pavier’s marketing strategy was aimed at arousing rather than satisfying a specific demand for a product that was still relatively new to the English book market—a collection in folio of plays by a dramatist writing for the commercial playhouses. Isaac Jaggard may eventually have succeeded in finding an investor for his project [the 1623 Folio] because Pavier had paved the way for it in 1619 by significantly reviving the fortunes of Shakespeare in print. . . . What would Pavier gain from it? . . . Selling his quartos both individually and as a nonce collection would minimize Pavier’s financial risk, [and] . . . Pavier would gain additional revenue from lending the right to reprint his Shakespeare plays to other stationers [namely, those in the syndicate bringing out the Folio].

Compositor Identification and Order of Printing in Q2

Kable (1970, pp. 7–18) identifies Compositor B of the 1623 Folio as the single compositor who set all the Pavier quartos. Andrews (1971, p. 320) identifies the compositor of MND Q2 as the fellow workman of Compositor B on the Pavier quartos: Compositor F alone set up the type for four plays [including] . . . A Midsummer Night’s Dream. Andrews employs the following criteria for discriminating between the two compositors: spelling, capitalization of I will contractions, punctuation of the text, placement and punctuation of marginal stage directions, and consistency in the use of italic type for proper nouns. He attributes the marked changes in punctuation between Q1 and Q2 to Compositor F (pp. 395–6): with Midsummer Night’s Dream . . . Compositor F’s average number of punctuation changes . . . jumps dramatically upward, this time to 17.0 punctuation changes per page (1050 changes in 62 pages). . . . [396] It is difficult to judge the extent to which the dramatic increase in punctuation changes . . . was affected by the nature of the punctuation in the Q1 Midusmmer copy-text. . . . however, it would appear that the increase is largely to be explained as resulting from a new degree of aggressiveness on the part of the compositor. Andrews (1973, Two Compositors, p. 5) subsequently renames his Compositor F Compositor G.

In the course of compositor identification, Andrews (1971, pp. 331–2) provides a headline analysis of MND Q2: the set of running-titles [headlines] that appear on pages B1, B2v, B3, and B4v reappear on pages B2, B1v, B4, and B3v, respectively. They then reappear on the following pages: D1, D2v, D3, D4v and D2, D1v, D4, D3v; E1, E2v, E3, E4v and E2, E1v, E4, E3v; G1, G2v, G3, G4v and G2, G1v, G4, G3v. Another set of running-titles appear in parallel sequence: C1, C2v, C3, C4v and C2, C1v, C4, C3v; F1, F2v, F3, F4v and F2, F1v, F4, F3v; H1, H2v, H3, H4v and H2, H1v, H4, H3v. It is obvious that two different skeleton frames were employed in the printing of Midsummer, one frame being used to print both the outer and inner formes of quires B, D, E, and G and a second frame being used to print both the outer and inner formes of quires A, C, F, and H. (I omitted quire A from the initial part of this discussion because it contains only five pages with running-titles; it should be noted, however, that the running-titles for these pages are identical with, and in the same relationships to each other as, the running-titles for quires C, F, and H.

While Andrews sees no significance in such headline analysis for compositor identification, Blayney (1972, 197–205) does. Andrews (1973, Supplement to Two Compositors, p. 11) summarizes Blayney’s position, while departing from it: Blayney adopts as a working hypothesis the idea that the alternating skeletons reflect alternating compositorial stints. While the idea, according to Andrews, has some application to MV, Blayney’s working hypothesis runs into difficulties when he tries to apply it to the remaining Pavier Quartos, including MND. It is at variance with the most compelling evidence based on spellings, punctuation, and other differentiae. Andrews (1973, Supplement to Unresolved Bibliographical Problems, pp. 1–5) expands these differentiae to include the spacing of medial commas and periods and of periods after SPP.

Andrews and Blayney also differ concerning when in the sequence of Pavier quartos MND Q2 was printed. Andrews (1971, p. 328), noting that MND’s title-page contains no typographical associations with other Pavier quarto title-pages, writes that MND was the very last Pavier quarto to be printed. Blayney (1972, pp. 196–7), relying on watermarks, places MND Q2 just before Lr.

Knowles (1982, pp. 191–206), accepting Blayney’s location of MND Q2, analyzes the recurrence of distinctively damaged types from the last three quires of MND Q2 in Lr. Q2 before focusing on such recurrences within the latter. He discovers that quires at the beginning of Lr. were set by two compositors at two different type cases; (p. 202) he refuses to speculate on the relation of these two Jaggard workmen to the ones who set type for F1. Although Knowles’s primary interest is Lr., he does advance a hypothetical explanation of the typographical relations of Q2 MND and Q2 Lr. Since the last three sheets of MND seem to have been set from two type cases [197], and, as Peter Blayney has shown . . . , with two skeletons used for the most part in the same unusual pattern of alternation as is found in Lr., one may safely suppose that the last three sheets of MND were set by two compositors working more or less concurrently; since the type cases are the same for both plays, one may even think it likely that the same two compositors who set the last sheets of MND continued at their cases to begin setting the early sheets of the next Pavier quarto, Q2 Lr. It looks as if Compositor 1 [at case x], after setting MND G(i), distributes the long-standing type pages from E(i) and begins to set H(o) sometime before his fellow compositor at case y begins to set H(i), the last forme for this play. He finishes while Compositor 2 is distributing types from G(o) in order to set H(i). Apparently Compositor 1 begins work right away on the first forme of the next play, A(o) of Lr., before G(i) has been unlocked. . . . Meanwhile Compositor 2 has finished H(i) and is about to begin Lr. B(o). . . . The precise details of these speculations will have to await further confirmation from the study of types throughout the whole of Q2 MND, but I think that there is already sufficient evidence from types to support Peter Blayney’s assignment of MND just prior to Lr. in the Pavier series.

Brooks (ed. 1979, pp. xxviii–xxix) notes an abnormal prefix [Peter for Quin 820] . . . on D1r. Bottom’s Peter quince [819] supports Dover Wilson’s diagnosis of the cause [Wilson, ed. 1924, p. 154], a shortage of [italic] capital Q’s; undoubtedly it prompted the resort to Peter, Pet., which continues on D2v. These are two pp. of the outer forme; in between them, D1v, D2r, belonging to the inner forme, have Quin. (eight times). Sheet D, then, was set by formes, D2v [xxix] after D1r, and not seriatim [or in reading order]. So, no doubt, was the whole of Q2; an easy method with a page-for-page reprint.

Annotated Q1 Copy for Q2

Cuningham (ed. 1905, p. xv), speculatively: In respect of the superior stage-directions of Q 2, it may not be unlawful to conjecture that Roberts [i.e., Pavier] had taken a copy of Fisher’s Quarto to a theatrical representation, or had otherwise procured a prompter’s copy and improved the stage-directions of his edition accordingly. Massai (2007, 122–9) identifies certain patterns of very occasional editorial attention across the Pavier quartos, including MND Q2.

The First Folio (1623)

On 8 Nov. 1623 the first collection of Sh.’s plays, now known as the First Folio, was entered in the Stationers’ Register (Book D, p. 69), as here transcribed in Greg (BEPD, 1:33; cf. 3:1109–12): Mr. Blounte Isaak Iaggard. Entred for their Copie vnder the hands of Mr. Dor. Worrall and Mr. Cole warden Mr. William Shakspeers Comedyes Histories, & Tragedyes soe manie of the said Copies as are not formerly entred to other men. vizt. [Here follows a list of half the plays]. Edward Blount and Isaac Jaggard were co-publishers; Isaac Jaggard had just recently inherited the printing business of his late father William. The Dor. Worrall who granted official license to publish was Thomas Worrall, chaplain to the Bishop of London, and George Cole was then Upper Warden for the Stationers. MND is not explicitly included in the entry because it had already been entred to other men, namely, Thomas Fisher, in 1600. In F1 MND is found on sigs. N1v-O3v, pp. 145–62 of the first section, the comedies. Lee’s (1902) census of extant copies has been replaced by West (2003). All earlier facsimiles, such as those of Lee (1902), Methuen (1910), and Kökeritz-Prouty (Yale, 1954), are surpassed by Hinman’s (1968; 1996) facsimile compiled from the best pages of copies in the Folger Library. For a facsimile of MND alone see West (c. 2008).

Press Variants

Hinman (1963, 1:260–1) identifies sigs. N2, N6v, and O3 as certainly or possibly indicating stop-press corrections. However, his claims that sig. N2 (where in some copies for in 331 is unevenly inked) (p. 261) almost certainly reflects proof correction; that sig. N6v (where in some copies a space prints after day-light, in 1481) thus contains a possible stop-press variant; and that sig. O3 (where a space prints after sent. in 2040) also exhibits stop-press correction have all been silently set aside by Rasmussen & West (2012, p. 875). Sig. O2 is variant (Hinman, p. 261): O2 (page 159, MND)—one non-textual variant only, as in O5v [where the page no. also varies].

1. page no. 165] 1 copy only (Folg. 60)
159] all others; Lee and Yale [facsimiles].

Compositor Identification and Order of Printing in F

Drawing upon variations in the spelling of frequently occurring words (e.g., do/doe and go/goe), Satchell (1920, p. 352) distinguishes between two compositors setting type in F1 Mac.; Willoughby (1932, pp. 56–8), applying Satchell’s method more widely in F1, concludes that these two compositors, now called Compositor A and Compositor B, must have been assisted by at least another pair, for MND, MV, and Rom. (p. 58) show no evidence . . . of having been composed by either A or B. Hinman (1957, p. 4), announcing his discovery of a new compositor in the F1 Tragedies, designates the newcomer Compositor E because not all of the material before the Tragedies was set by A and B, and C and D may later be required to designate compositors in the Comedies. Then, in his masterly study of the printing of F1, Hinman (1963, 1:193–200) uses evidence of the recurrence of distinctively damaged types as well as spellings to separate Compositors C and D from Compositors A and B. While (ibid., 2:518) Compositor D worked only on the Comedies (MM, Err., Ado, LLL, MND, MV, and AYL), Compositor C worked only most frequently on the Comedies, for Hinman finds it possible that C might be identified as Compositor B’s partner on plays in the Histories and Tragedies, particularly R2 and Ham. Hinman employs typographical evidence, rules, headlines and spellings to determine the order in which the pages of F1, including (ibid., 2:414–26) those in quires N and O (all of MND and the beginning of MV), were set into type and their compositors. Cairncross (1971, pp. 44, 47), in a rather unsystematic study, disputes a number of Hinman’s compositor attributions of particular pages. Howard-Hill (1973, pp. 83, 98) and O’Connor (1975, pp. 93–9, 117), in more thorough and orderly examinations, confirm Cairncross’s reassignment of four quire-O pages from Compositor A, to whom Hinman assigned them, to Compositor D, while disproving Cairncross’s (1971, p. 47; 1972, pp. 379, 406) other reattributions of pages from Compositor C to Compositor B and from the latter to Compositor E. It is the combined work of Hinman, Howard-Hill, and O’Connor that is now widely accepted—see, e.g., Wells & Taylor (1987, p. 149).

Hinman demonstrated that MND, like all the other F plays, was set into type by formes from cast-off copy, or copy marked up to indicate exactly which lines were to fit on each page of the quire, so that the pages could be set out of order. He identified three different compositors, each at a different case—Compositor B at case y, Compositor C at case x, and Compositor D at case z. The order of printing of the formes and the division of work among the compositors discovered by Hinman follows, but the pages marked with an asterisk are those confirmed or reassigned to Compositor D by, in succession, Cairncross, Howard-Hill, and O’Connor:

Cx Dz Cx Dz By Dz Cx Cx By Dz By Dz Dz By By Cx Cx Dz Cx By Cx Cx Dz Cx Dz Cx Dz
N3v: N4 N3: N4v N2v: N5 N2: N5v N1v: N6 N1: N6va *N6vb O3: (O4va) (O4vb) O2v: (*O5) O3v: (O4a1–50) (O4a51-b) O2: (*O5v) O1v: (*O6) O1: (*O6v)

Mistakes in casting could force compositors to alter the line division of their copy. All three of the F MND compositors change line division, but few of these changes evidently compensate for faulty casting-off. Werstine (1984, p. 114): Compositor B splits in two Q verse lines at 413–14 (correction of a Q error of running the initial half-line of speech together with the next line), 430–1, and 434–5; Compositor C at 1560–1; and Compositor D at 1441–2. Compositor B may deliberately stretch his copy for sig. N2v when he divides in two the pentameters on each side of a mid-line SD above and below which he creates white space. Both Compositor C and D, though, are simply dividing the first lines of speeches that, combined with SPP, are each too wide for the F column (p. 79). When B twice prints prose as verse on sig. O3 (2068–9, 2108–9) he neither saves nor loses space (p. 116). When he sets verse as prose at 2044–5, he is (pp. 87–8) faced with a verse line too long for his composing stick and therefore runs the end of the line together with the following verse line to set both lines as [88] prose. While the lines appear on sig. O3 in the first half of a quire, where he may need to adjust his copy to available space, his deliberately saving a line of type seems unlikely because elsewhere in the same column he allows for lavish white space around SDD. (P. 92): Compositor D divides off the last sentence of a prose speech at 999–1002 on sig. N5, thereby using an extra line of type, but because he is setting a page in the second half of a quire, space is not a factor. Instead the relineation seems designed to mark a change of address or topic and seems to be associated with other such changes he made in LLL at 128–30 and 800–2.

Most, but not all, the verbal variants between Q2 and F have been recognized and classified as corrections or errors introduced by the three compositors: for Compositor B see Werstine (1978); for Compositors C and D O’Connor (1977).

Features of F

Collier (ed. 1842, 2:cc2): The chief difference between the two quartos and the folio is, that in the latter the Acts, but not the Scenes, are distinguished. (Rhodes [1923, p. 120] addresses the possibility that the divisions [into acts in F] . . . were made in consonance with theatrical practice and connoted pauses [between acts at the Blackfriars or the new Globe]. . . . [T]he division into five acts necessitates two pauses during the game of blind man’s bluff in the woods, which is marked into three acts. Although it shows execrable stage-management, at the end of Actus Tertius is a note They sleepe all that act, meaning that the four lovers would have to lie, feigning sleep, in view of the audience while the act is playing [the act being the music between the acts]. . . . [I]t is indisputable that the division . . . was made by the prompter in consonance with theatrical practice. It cannot be entertained for a moment that They sleepe through the act [sic] was a literary or editorial note, to assist a reader in visualising the action. Foakes [ed. 1984, p. 151] believes that the act in this SD refers to the next act—Act 4.) Furness (ed. 1895, p. xv): In Roberts’s (Q2) [there are] about seventy-four [SDD]; and in the Folio, about ninety-seven.

Q2 Copy For F

Johnson (ed. 1765, 1:176): Roberts [i.e., Q2] was followed, though not without some variations, by Hemings and Condel [i.e., F]. Furness (ed. 1895, pp. xiii–xiv) demonstrates how the failure of the Q2 compositor to follow his copy precisely and set the word and in roman type in Titania’s line 979 in turn led the F compositor to create the redundant SD at 979–80: Enter Pease-blossome, Cobweb, Moth, Mustard-|seede and foure Fairies. Ebsworth (ed. Q2 1880, pp. xx–xxi) cites as proof of F’s use of Q2 as copy common errors at 183, 481, 482, 552, 1199, 213, 532, 557, 1652, 1688. Wilson (ed. 1924, p. 154): In 1619 . . . Jaggard . . . followed [his copy the Fisher Q] with suspicious exactitude. Apparently, however, the constant occurrence of Queene and Quince in dialogue, stage-direction and speech-heading, strained the resources of his compositors’ type. In any event, the italic Q seems to have given out on sig. D1r. and D2v., and accordingly the name Peter had to be resorted to in place of Quince. The fact that the F. also reads Peter in this same section of the text is a proof that it was set up from the Q. of 1619 and not from the Fisher Q. of 1600. Another proof is the reappearance in 1623 of nearly all the sixty to seventy misprints first introduced into the text in 1619. When we observe, moreover, that to these transmitted misprints the F. compositors added another sixty to seventy of their own, it will be evident that the F. version cannot claim much textual authority.

Only Craig (1961, pp. 108–9) appears to dissent from the view that F was printed from a copy of Q2: As a printed version of the same manuscript from which the fair copy had been made, Q1 would resemble the theatrical version very closely, and this may be said of both of the quarto and the folio as they stand. It does not seem necessary therefore, in view of this identity of origin, to imagine that the folio has been set from the quarto. Printing of the folio from the playhouse copy is a simpler and more satisfactory way in which to account for resemblances between these two texts. . . . [109] Although [F] has some features that may be derived from Q2, [it] actually resembles Q1 more closely than it does Q2. There are of course passages in which F differs from both Q1 and Q2. . . . In this perplexity one has to content oneself with a moderate position: the official playbook was in the hands of Jaggard and Blount and served them as copy for the body of the play, although there are in F some minor resemblances to Q2.

Brooks (ed. 1979, pp. xxviii–xix) cites evidence . . . of several kinds for the use of Q2 as F’s copy: At [1170] both [Q2 and F] have the abbreviation Hell. (Q1 Hel.), unique in each. On fifteen further occasions they have identical abbreviations in speech-prefixes, differing from Q1’s and from some of their own. Hence we can be confident that the tucking-in of Enter Snowt [929], rather than giving it the normal line of its own, and the printing of prose as verse at [1986–8], come in F from Q2, even though they originated in Q1. The Folio has several instances of progressive corruption. At [253] Q1 reads is so oft; Q2 inadvertently omitted so; F, lamely attempting to mend the metre, miscorrects to is often. Q2, at [1703], undoes the Q1 inversion more will hear, reading will hear more; F worsens the corruption with shall hear more. There are less striking instances at [1415, 1420]. Lysander’s sentence at [1677–8] is left incomplete because Egeus interrupts him; not realizing this, Q2 completes it by supplying a verb: be. The Folio repeats this and over fifty of its other corruptions: good examples are Q2’s silly foal for filly foal (misreading long s) [417], . . . and hearken for listen (a compositor’s synonym) [2038].

See the textual notes of this edition for all significant differences between Q2 and F, including SPP, SDD, lineation, punctuation, and verbal variants.

Annotation of Q2 Copy for F

Capell (1783, 2:3:111 ff.) identifies some F-only SDD as playhouse interpolations (see here). White (ed. 1857, 4:17): Printed copy [for F] had been used at the theatre for stage purposes and corrected with some care. Ebsworth (ed. Q2 1880, p. xix) dissents: It is idle to talk of the Folio editors having access to any manuscript authority for [MND]. We hold it indisputable that they used Roberts’s printed Quarto, sometimes increasing the defects, sometimes guessing commonplace variations; but they give absolutely nothing of such improvements as would have been gained from a genuine manuscript, or even from a certified revised and corrected prompt-book. Halliwell-Phillipps (1884, p. 255) confirms the theatrical provenance of the copy of Q2 or the MS used to annotate that copy by identifying Tawyer (1924) as a subordinate actor in the Globe Theatre in the pay of Heminge’s [sic]. For more on Tawyer, see G. E. Bentley, Jacobean and Caroline Stage, 2:590. Furness (ed. 1895, pp. xiv–xv) traces the appearance of Egeus for Philostrate in F’s Act 5 to the doubling of their parts and identifies the Asse head (927) as a prompter’s term: (p. xv) the prompter of Shakespeare’s stage, knowing well enough that there was among the scanty properties but one Asse-head, inserted in the text with the Asse head—the only one they had. Brooks (ed. 1979, p. xxxii n. 1), objecting to the possibility of doubling Egeus and Philostrate, points out that there is no time for an actor to re-enter as Egeus just after he has exited as Philostrate in 1.1. See also Greg, 1955, p. 243, imagining Philostrate unavailable for Act 5 because he doubled another role that also needed to be performed then, and Hodgdon, 1986, p. 536. Smidt (1986, pp. 121–2) also disagrees about the doubling: When the Folio substitutes Egeus for Philostrate as master of ceremonies at the wedding feast this could be explained as a way of saving an actor’s part, but there are no great number of men’s parts in [MND], and it is more likely that at some point in the stage history of the play someone objected to the absence of Egeus at the feast and thought he ought to join the party once he had been admitted to the comedy. Brooks seems right. More likely the appearance of Egeus for Philostrate in Act 5 arises from the telescoping of the roles. When bookkeepers subsumed one role under another in actual playhouse MSS, they often failed systematically to record the disappearance of the subsumed role in SDD and SPP; hence perhaps the persistence of Philostrate in Act 1 and once in a SP in Act 5 (1874) in F (see Werstine 2012, pp. 164–72). For the idea that Egeus, rather than Philostrate, appeared in 5.1 in the allegedly earliest (1594) version of the play, see Hunter (1998, pp. 8–9, and 2002, p, 6). For the idea that John Heminge annotated the copy of Q2 with notes in which he recalled a 1594 performance of the play, see Hunter (2002, pp. 7–10). For the application of literary and/or performance criticism to the Q1/F variants, particularly Philostrate/Egeus, see Hodgdon (1986), Wells (1991, MND Revisited, p. 22), Calderwood (1991, p. 428, n. 40), Wiles (1993, p. 174), Holland (ed. 1994, pp. 265–8), McGuire (1988, 1989), Pollack-Pelzner (2009). Taylor (2002, p. 52, n. 31) contends that F’s substitution of Egeus for Philostrate in Act 5 is inexplicable and therefore certainly wrong.

Cuningham (ed. 1905, p. xvi): the text of the Folio has its direct origin in a stage copy of Q 2. He cites as evidence ll. 1924, 927, and 2107, alleging in connection with the last that the early entrance of Thisby is an indication that printer’s copy was a stage copy . . . indicating that the actor was to be ready before he has to make his actual appearance on stage. Such an observation about the F SD as a warning direction is fanciful because Flute as Thisby comes onstage fewer than a half-dozen lines later; actual theatrical texts almost always mark warnings much earlier.

By the 1920s confidence that the copy of Q2 used in the playhouse must also have served as printer’s copy for F begins to slip. Adams (1923, pp. 538–9): MND, like R2, 1H4, Tit., and Ado, was printed in F from the actors’ special copies of . . . quartos which had been converted at the theatre into prompt-books or from (p. 539) the most available editions of these quartos [after they had been compared to] . . . the actors’ prompt-books . . . ; these collated quartos [would have been placed] in the hands of the [F] compositors. Wilson (ed. 1924, p. 78), in addition to suggesting an annotated quarto that served as printer’s copy may have served as a playhouse promptbook also allows only that Q2 copy for F could have been corrected for the purpose of publication, by some scribe working with the prompt-book before him. Wilson is led to suggest such an alternative by his belief that a copy of Q1 may have served the acting company in the playhouse; the Q1 punctuation of Theseus’s speech at 1841–57 is the slender reed on which he builds: (p. 157, n. 1) Now each item in the brief [read by Lysander in F] in Q. 1600 is followed by a question-mark, as if it were a query put to some one who replies with the comments [the only parts of the speech given Theseus in F], and it looks very probable that it was these queries which suggested the F. arrangement. If so, then the theatre prompt-book was almost certainly a copy of Q. 1600 seeing that all the queries but two towards the end of the speech, have been eliminated in Q. 1619. Acquainted as he is with playhouse MSS, Wilson also attempts to locate SDD first printed in F in particular places on the pages of the quarto prompt-book, suggesting (p. 156) that shifting places [1460] appeared in the margin of sig. F2, where it governed the action represented on the page as a whole, rather than simply in the line opposite which it is printed in F, to which it is irrelevant. He also imagines (p. 157) that Enter Pucke, printed in F over twenty lines before he needs to enter [865], was in the playhouse quarto a warning SD, noted atop sig. D2. Although Greg (1942, pp. 125–6) accepts the tradition that F’s new directions undoubtedly originated in the playhouse, he denies that the copy of Q2 from which F was printed could have served to guide performance because F is not sufficiently consistent and correct (p. 126): I should have expected to find more of the book-keeper’s notes in the original prompt-book, and therefore in Q; and if Q had itself been used as a prompt-book I should have expected to find certain anomalies removed in F. If the book-keeper found it necessary to specify the Lovers in v.i [see above, here] in the original prompt-book, why did he not the Clowns in III.i [813] either there or in the prompt quarto, especially since in the latter he took the trouble to translate the rabble into Snout and Starveling in IV.ii [1746]? Surely the errors in I.i whereby two half-lines of text appear as stage directions [30, 33] would have been corrected. Why does the entrance of the translated Bottom appear out of place [927]? The duplication in V.i, whereby we have Exit all but Wall [1951] followed three lines later by the exit of Lion, Thisbe, and Moonshine [1955], could hardly have been overlooked in performance. The second is the original direction of Q; the first must have been introduced from a manuscript. No doubt some confusion might have occurred in transferring the prompter’s notes from the copy Q1 to one of Q2; but on the whole the theory that a quarto was used as prompt copy seems to raise more difficulties than it solves. Idem (1955, pp. 244–5) also cites as additional examples of the incompetence and clumsiness of his imagined editor of Q2 copy for F: 865, 888, 1385, 1509, 1541, 1559, 1661–2, 1746, 1819, 2009, as they appear in F. Greg’s idea that Q2 copy for F was not itself annotated and used in the playhouse is followed by Doran (ed. 1959, p. 174), Wells (ed. 1967, p. 165); Evans (ed. 1974, p. 247); Brooks (ed. 1979, pp. xxix–xxx); Foakes (ed. 1984, p. 147); Wells & Taylor (1987, p. 279); Holland (ed. 1994, p. 115); Berger (ed. 1995, pp. x–xi). Nonetheless, quartos were annotated for playhouse use (see Werstine 2012, pp. 314–17, 335–42) and continued to be throughout the 17th c., and Greg’s expectations of thoroughness, consistency, and correctness of annotation in early modern theatrical texts are denied by the contents of such actual texts (see Werstine 2012, pp. 107–99, 234–391).

Authority and Revision in F

To judge from Capell’s comments on particular variants in F’s dialogue and SDD, he attributes no authority to F, but Malone has somewhat higher regard for F, and by the latter half of the 19th c., a number of editors are prepared to grant Shn. authority to readings in it. The New Bibliographers return to Capell’s positon, but near the end of the 20th c. there is a revival of the 19th-c. belief in F.

Capell (1783, 2:3:111) thinks F’s SD Musicke Tongs, Rurall Musicke (1541) simply wrong: it is certainly an interpolation of the players; as no such direction appears in either quarto, and Titania’s reply is a clear exclusion of it. He denies (2.3:113–14) the authority of F’s cut at 1718+1 of Are you sure / That we are awake? He is equally dismissive of (2.3:115) F’s redistribution of some of 1841–57 to Lysander— this reading and commenting of two persons, alternately, has something aukward in it: and seems a change of the players, calculated for the ease of the actor who presented the latter character—and of the player editors’ error in making Egeus enter in an act [Act 5] he has no concern in . . . (probably) from their laying Philostrate’s character in this act upon the player who had finish’d that of Egeus. It is not clear if Capell is suggesting that the Egeus actor doubled the role of Philostrate throughout the play or only in the last act. (Capell [1783, 2.3:116] thinks Sh.’s own revision can be recovered in small part from F2, in particular in the reading streames at 2076.)

Malone (in Boswell, ed. 1821, 1:203), though, calls for more respect for F: Thus therefore the first folio, as far as respects the plays above enumerated [including MND], labours under the disadvantage of being at least a second, and in some cases a third, edition of these quartos. I do not, however, mean to say, that many valuable corrections of passages undoubtedly corrupt in the quartos are not found in the folio copy; or that a single line of these plays should be printed by a careful editor without a minute examination, and collation of both copies [i.e., Q and F].

For White (ed. 1857, 4:17) though, Neither quarto . . . is to be regarded in any other light than as an assistant in eliminating such corruptions as may have crept into the folio itself; though Fisher’s enables us to correct some errors which were passed over in the copy of the quarto furnished to the printers by Heminge and Condell. The quartos sometimes concur in a reading different from that in the folio; but this is of little moment: it merely shows (unless in the case of a palpable corruption of the press) that in the copy from which the folio was printed, an error is corrected which had appeared in both the previous editions. The presumption is especially in favor of the authorized edition [i.e., F], when we know that it was printed from a copy that had been corrected in Shakespeare’s theatre, and probably under his own eye, if not by his own hand. (In particular White cites the F readings at 700, 1247, 1384+, 1718+1, 1994 [corrected by Shakespeare or someone else in his theatre], and 2010 as authoritative; however, he thinks 1812–13, which are common to Q and F, an interpolation and the substitution of Egeus for Philostrate in Act 5 wrong, both not Sh.’s.) So Furness (ed. 1895, p. xii): It may be that in using a printed text [for F, namely Q2, Heminge and Condell] were virtually using Shakespeare’s manuscript if they knew that this text . . . had been for years used in their theatre as a stage copy, with possible additional stage-business marked on the margin for the use of the prompter, and here and there sundry emendations, noted possibly by the author’s own hand, who, by these changes, theoretically authenticated all the rest of the text. Adams (1923, pp. 539): These printed prompt-copies [such as the copy of Q2 used to print F MND] would receive corrections (from the author, or from the actors), alterations, and additions and such stage-directions as were found necessary.

However, the New Bibliographer Wilson (ed. 1924, p. 159), generalizing about the changes in SDD and SPP introduced into F, returns to Capell’s position: it should not be necessary to argue that Shakespeare himself had nothing whatever to do with them. Even less easy is it to imagine him in any way responsible for the F. corrections in the dialogue. [He lists those unlikely to be due to the compositors: 759, 1247, 1287, 1719, 1829, 1994, 2010.] These variants are almost certainly due to the scribe who gave us the F. stage-directions. Some of them are good, some indifferent, and some definitely bad; but all are assuredly guesses. Greg (1942, p. 125–6) on F’s text: The new directions undoubtedly originated in the playhouse. (P. 126): Such changes in the text as are not either misprints or corrections of misprints seem to be the editor’s and do not imply any independent source.

Brooks (ed. 1979, pp. xxx–xxxiii), following Greg’s questionable presumption that the copy of Q2 from which F was set could not itself have been the playhouse text used to guide performance: The promptbook consulted in preparing copy for the Folio was clearly not without alterations from the text in the prompt-copy as [xxxi] originally transcribed from Shakespeare’s autograph [assuming without grounds that that Shn. copy could not itself have been used in the playhouse]. Theoretically, some changes may have been authorized by him; but at least the majority were no doubt made without authority, whether in the original prompt-book or in a new one, if a new one was transcribed from it. But whatever unauthentic changes had accumulated in it, the prompt-copy which supplied some Folio readings did derive by a process of transcription from Shakespeare’s autograph. That process was independent of Q1. Accordingly, in respect of readings which the Folio can be presumed to have taken from the prompt-copy, F is an independent witness to what may have stood in the autograph. In the line of descent described it is the earliest extant witness, and in respect of those readings, and of those alone, it is therefore a substantive (that is, an evidential) text—as Q1 is for the play as a whole. Such authority as F therefore has is weakened, however, by the annotator’s demonstrable negligence and clumsiness. He cites the misplaced SD at 927 and the duplicate SDD introduced at 865 and 1951. His neglect of dialogue further limits the possible authority of F. This is significant for F’s readings at [1247 and 1994]: passionate where Q1 has a palpable omission, and knit up in thee where Q1 has the impossible knit now againe. . . . Yet if they are retrievals from prompt-copy, why are there not more? That the annotator’s eye might fall upon dialogue may be suggested by F’s choise of merit for the Quartos’ choise of friends [149]. Brooks (pp. 154–5) makes a case that merit could have been Sh.’s first choice of reading in his initial composition, one that he later replaced with friends but one that nonetheless found its way into the playhouse text from Sh.’s own papers. In spite of his confidence in the authority of this single F-only reading, Brooks thinks that his annotator of Q2 (p. xxxii) was perfectly prepared to guess, even when he could have consulted the prompt-book, and passionate and up in thee may be other guesses of his, though there the contexts offered little hint.

A further subtraction has to be made from the authority of F’s text, even where its source is prompt-copy. The prompt-copy itself is unlikely still to have represented in all respects the kind of performance for which Shakespeare designed the play, or to which he may have adapted it. The substitution of Egeus for Philostrate in Act V, at odds with Theseus’ enquiry for our usual manager of mirth and damaging the metre at [1833] was made apparently to save a speaking part (Philostrate is mute in I.i): it is a change Shakespeare cannot have wished for, though he might acquiesce in it as an expedient. The same may be said of the one or more intervals [at the ends of acts] introduced in a play conceived and originally performed as a continuous action. . . . [xxxiii] If the revival of the Dream matched by the prompt-book was in 1609 or later [the approximate date at which Sh.’s company began to perform at the Blackfriars and observe intervals between acts for the first time], Shakespeare may not have been closely associated with it.

With Foakes (ed. 1984, p. 148), Wells & Taylor (1987, pp. 279–80), and Holland (ed. 1994, pp. 115–17, 257–68) the 19th-c. understanding of F’s authority makes a return. Foakes interprets as Sh.’s revisions the few corrections made to Q2’s dialogue in F: 149, 1041, 1247, 1994, and 2010. He also thinks that Puck’s early entrance in F at 865 in 3.1 records prompt-copy, suggesting that Puck should overhear rather more of the dialogue between Quince and his crew than his entry in the quartos would permit. Wells and Taylor present the F text’s possible censorship as a consequence of the 1606 Acte to Restraine Abuses, shown by the loss of 2113–2113+1 as further evidence of its theatrical provenance. They adopt the following editorial policy: Without strong evidence to the contrary, one must therefore assume that the prompt-book is the authority for all added or substantially altered Folio directions and speech prefixes. Some of these variants might derive from late revivals, over which Shakespeare had no control; but none certainly do [sic], and only the act divisions and Tawyer’s name can be confidently associated with performances later than those in the mid 1590s. Although each direction has been considered on its merits, we have found no reason to doubt that the bulk of the Folio directions represent the play as originally and authoritatively staged. Those directions which clearly envisage a different staging from that implied by Q seem to us to be dramatic improvements for which Shakespeare was probably responsible. Such an editorial policy forces justification of F SDD that were long thought to be erroneous, such as the F entrance of Pucke in 3.1. [at 865] over twenty lines before Q1’s entrance for him (which is also reproduced in F, 888) and twenty lines before, for all one can tell, he has business onstage: (p. 281) an editor committed to entertain possible authorial revision must consider the F alternative. (Pp. 281–2): Following Greg’s unwarranted assumption that F had to have been printed from a copy of Q2 annotated with reference to a playhouse MS (rather than from a copy of Q2 annotated for use in the playhouse), Wells and Taylor also assume that the annotator must have been right to add the F SD from the playhouse MS. They justify this second assumption by imagining that F records accurately a production in which Puck entered silently and unnoticed to supply Quince with the almanac he was requesting at the point of the F SD. (Werstine [2012, pp. 173–6], however, shows that there is no reason to suppose bookkeepers’ additions of entrances necessarily inerrant because in actual theatrical texts some such additions can be shown to be erroneous in context; consequently, playhouse texts, such as the one inferred to lie behind F MND, need not be reliable records of any performance.) Holland (ed. 1994, p. 117) attributes to some other authority than the compositor’s or editor’s ingenuity the five readings adopted by Foakes as Shn. revisions. He writes (pp. 257–68) of Shakespeare’s Revisions of Act 5, accepting Wilson’s account of the mislined verse at 1798–1880 and counting as a second revision the substitution of Egeus for Philostrate and the hiving off of pieces of Theseus’s 1841–57 speech for Lysander. Ioppolo (1991, p. 113) had associated Sh.’s alleged revision of 1824–5 in Q1’s printer’s copy with the transfer of some of Theseus’s lines to Lysander in the F printer’s copy, but Holland (ed. 1994, p. 266) demurred.

Authenticity

There has been less resistance to attributing all of MND to Shakespeare than with other plays in the canon. Wilson (ed. 1924, p. 94): The only presumably pre-Shakespearian play, known to us by name, which might conceivably have formed the basis of the Dream, is the mysterious King of the Fairies, scornfully linked by both Nashe and Greene with another drama called Delfrigus . . . , as part of the stock-in-trade of a travelling company. Wilson refers to Nashe’s The Gentlemen Stvdents of Both Vniversities (ed. R. B. McKerrow, 1904–10, 3:324): a company of taffaty fooles . . . might haue antickt it vntill this time vp and downe the Countrey with the King of Fairies, and dined euery day at the pease porredge ordinary with Delfrigus; and to Greene’s Groatsworth of Wit (ed. A. B. Grosart, 1881–6, 12:131), in which a player says to Roberto why, I am as famous for Delphrigus, and the king of Fairies, as euer was any of my time. Wilson continues, We have not, however, been able to trace any clues to the existence of such a play beneath the Shakespearian text, unless it be its curious connexion, or seeming connexion, with old dramas like Damon and Pythias, 1582, and Heywood’s translation of Seneca’s Hercules Furens, 1581. For the alleged allusion to Damon and Pythias, Wilson (p. 148) cites passages beginning at 2083 and at 2124, and (pp. 109–10) for that to Hercules Furens the passage beginning 297. Wilson opines that (p. 110) there seems no reason why Shakespeare should burlesque a translation ten or a dozen years old and therefore thinks perhaps the text here goes back to some pre-Shakespearian version. de la Mare (1935, pp. xxxii–xlviii), attending to the poor quality of some of the verse given the four lovers and to the absence of the vocabulary of some of these lines from the rest of the Sh. canon, concludes (p. xlvii) that the earliest draft of the Dream was not of [Sh.’s] own workmanship but a play . . . written by some more or less artless scribe. Following de la Mare, Wilson then states unequivocally that (1948, p. 29) the original play was not written by Shakespeare at all. Yet, according to the noted disintegrationist Robertson (1924, 1:440), in MND we can catch the true voice of Shakespeare. Chambers (1924, p. 10) is prepared to credit Sh. with commonplace Elizabethan dramatic carpentry, rather than disintegrate plays by attributing parts of them to other playwrights.

Blumenthal, who finds (1961, p. 116) some seven participants in the authorship of the Shakespeare canon, thinks Robertson’s claim to be (p. 30) uncertain. Among the anti-Stratfordians, Bacon (1857, p. lxxxi) attributes MND to Raleigh, Theobald (1901, passim) attributes it to Bacon, Clark (1930, pp. 435–49) and the Ogburns (1952, chs. 44–5) to Oxford, Brooks (1943, pp. 596–7) to Dyer, Titherley (1952, pp. 71–5) and Evans (1956, p. 59) to Derby, Sweet (1965, p. 71) to Queen Elizabeth, and Hoffman (1955, p. 127) to Marlowe. Ross (1939, pp. 16–17) thinks MND among the works of Anne Whateley written in association with Sh.

The First Quarto (1600)

The printing history of Sh.’s MND starts on 8 Oct. 1600 when a book with its title was approved for publication in the Stationers’ Register (Book C, fol. 65v, as transcribed by Greg, BEPD, 1:16): Tho. fyssher Entred for his copie vnder the hand[es] of mr Rodes / and the Wardens. A booke called A mydsõmer night[es] dreame According to Greg (ibid., 3:1485), this was the only copy for a play licensed by Rodes, who, Greg suggests, may have been Thomas Rhodes. Greg (1962, p. 81): A Thomas Rhodes appears in the index to Hennessy’s Reportorium without a reference. A book called Micrologia; Characters or Essays, by M. R., 1629 (STC 17146), was licensed by E. Martin on 22 Dec. 1628 as Rodes charecters.

The play appeared in print in the same year, with the following title-page: [ornament] | A | Midsommer nights | dreame. | As it hath beene sundry times pub- |lickely acted, by the Right honoura- |ble, the Lord Chamberlaine his |seruants. | Written by William Shakespeare. | [device <McKerrow>321] | ¶ Imprinted at London, for Thomas Fisher, and are to | be soulde at his shoppe, at the Signe of the White Hart, | in Fleetestreete. 1600. This first edition is often called the Fisher Quarto to distinguish it from Q2 also dated 1600 on its title page; Q2 was once known as the Roberts Quarto and is now usually called the Pavier Quarto. Chambers (1930, 1:356): The printer may be [Edward] Allde or [Richard] Bradock. Greg (BEPD, 1:276): The printer [of Q1] appears from the ornaments used to have been probably Richard Bradock. The device is Fisher’s. Turner (1962, p. 33): As far as I have been able to determine, nothing is known of Bradock which would be of significant value to us in our examination of MND Q1. He was admitted to the Livery on 1 July 1598 and for a time was actively engaged in the trade. Around the turn of the century, he probably printed several play quartos: in 1598 [Christopher Marlowe’s] Edward II Q2; in 1600 [Ben Jonson’s] Every Man out of his Humor; in 1601 [Anthony Munday’s] The Downfall and The Death of Robert Earl of Huntington; and in 1602 [John Marston’s] Antonio and Mellida [and] Antonio’s Revenge, and [Jonson’s] Poetaster. Bradock seems to have thrived as a printer between 1598 and 1608, but he also printed plays long before and some time after the turn of the century. Greg (BEPD, 3:1497) records that Bradock printed his first extant play, Nathaniel Woodes’s The Conflict of Conscience, in 1581, and did not print his last until 1616 (Beaumont and Fletcher’s The Scornful Lady). However, Pantzer (STC 3:16) identifies the printer of the latter as John Beale. Thus the last of the extant plays that Bradock printed date from 1608, including A Yorkshire Tragedy from the Shakespeare apocrypha.

Berger (ed. 1995, p. viii): Thomas Fisher’s career as a publisher and bookseller was a short one. He was freed as a draper on 8 November 1596 by Richard Smith and transferred to the Stationers’ Company in 1600. Of the three other titles associated with him, Nicholas Breton’s Pasquil’s Mistress was printed in 1600, perhaps by Richard Bradock, and John Marston’s Antonio and Mellida and Antonio’s Revenge were printed in 1602 by Bradock. In these last two Matthew Lownes appears to have had an interest as well, as his shop in St Dunstan’s Churchyard is cited on the title-pages of both volumes.

Bartlett & Pollard (1939, pp. 70–1) identify and locate eight extant copies of Q1. Modern facsimiles include that by William Griggs, ed. J. W. Ebsworth, 1880; that in Shakespeare’s Plays in Quarto, ed. Michael J. B. Allen & Kenneth Muir, 1981; and Vol. 157 of the Malone Society Reprints, ed. Thomas L. Berger, 1995. For a digital facsimile see the Shakespeare Quartos Archive (www.quartos.org).

MND Q1 was one of a number of Shakespeare’s plays to see print in 1600 with the dramatist’s name on the title-page. The others were 2H4 Q1, MV Q1, and Ado Q1. Earlier printings with Shakespeare’s name on their title-pages were LLL Q1 1598, R2 Q2 and Q3 1598, R3 Q2 1598, and 1H4 Q2 1599. Blayney (1997, p. 388): some of these plays sold very well, R2, R3, and 1H4 being in the list of the top ten best-sellers among early modern English plays. However, like Ado and 2H4 (neither of which ever saw a second quarto edition) and LLL (which was not reprinted in quarto until 1631), MV and MND (without second editions until 1619) did not sell very well, even with Shakespeare’s name on their title-pages. Ibid. (p. 385) identifies May 1600–Oct. 1601, the interval within which MND Q1 was entered, printed, and published, as one of two peak periods for the registration of plays between 1585 and 1604, the other coming at Dec. 1593–May 1595. In each peak period twenty-seven plays were registered, although only 80% of the total of fifty-four registrations issued in books. Ibid. (p. 387): MND was one of eight Lord Chamberlain’s plays registered in the second peak period. See also Blayney (2005), and Farmer and Lesser (2005, Popularity and 2005, Structures).

Quality of Printing in Q1

While on the whole positive, editors and critics exhibit a wide range of opinion about Q1’s quality as a witness to what Sh. wrote. Pope (ed. 1725, 1:xv), not attending to the question of the priority of Q1 to Q2: If any were supervised [at the press] by himself [Sh.], I should fancy the two parts of Henry the 4th, and Midsummer-Night’s Dream might have been so; because I find no other printed with any exactness; and (contrary to the rest) there is very little variation in all the subsequent editions of them. Johnson (ed. 1765, 1:176) unable to establish the priority of Q1 to Q2: Neither of the editions approach to exactness. Building on Capell’s (1783, 2.3:115) inference of Shakespearean authority for the punctuation of the mechanicals’ prologue (see n. 1906–15), Knight (ed. 1839, 1:331) writes, The original of these editions, whichever it might be, was . . . carefully superintended through the press. The text appears to us as perfect as it is possible to be, considering the state of typography in that day. There is one remarkable evidence of this. The prologue to the interlude of the Clowns, in the fifth act [1906–15], is purposely made inaccurate in its punctuation throughout. . . . ; and this is precisely one of those matters of nicety in which a printer would have failed, unless he had followed an extremely clear copy, or his proofs had been corrected by an author or an editor. Compare Hudson (ed. 1851, 2:255) and White (ed. 1857, 4:17), as well as the following writing after Q1’s priority had been demonstrated: Chambers (1930, 1:358), Parrott (ed. 1938, p. 131), Greg (1942; 1954, p. 125), Doran (ed. 1959, p. 27), Holland (ed. 1994, p. 112). Clark and Wright (ed. 1863, 2:viii), guessing right about Q1’s priority, nonetheless regard it carelessly printed. Kittredge (ed. 1936, p. 229) echoes this judgment, describing Q1 as not very carefully printed.

Furness (ed. 1895, pp. x–xii), also guessing right about Q1’s priority: the excellence of the text [of Q1] is counterbalanced by the inferiority of the typography. . . . [A]lthough the entrances of the characters are noted, the exits are often omitted, and spelling throughout is [xi] archaic, for instance, shee [241], bedde [228], dogge [589], &c., betraying merely a compositor’s peculiarity. . . . [F]onts are mixed, and the type old and battered. Believing that the Q1 compositor set type by the ear from dictation, Furness finds such errors as Dians bud, or Cupids flower [1588], instead of Dian’s bud o’er Cupid’s flower the consequence. (P. xii): [I]t is assuredly more likely that such blunders as Eagles [454] for AEgle, or Peregenia [453] for Perigouna . . . are due to the deficient hearing of a compositor. However, Furness (p. xii) concedes that compositors . . . are exposed [to such errors] when with a retentive memory they carry long sentences in their minds, not just when they set from dictation. Cuningham (ed. 1905, p. xi), in the main closely following Furness: the text . . . has reached us in a state of comparative correctness and purity, [yet] there are passages which are admittedly corrupt. Ibid. (p. xv): the text is superior, and likewise the punctuation.

Wilson (ed. 1924, p. 79): Q1 is superior to many of the other quartos. . . . The misprints are few, and the literals fewer. The compositors seem to have worked slowly, whether through inexperience or because they found the manuscript difficult to read; for the text contains a number of interesting archaic spellings which almost certainly derive from the copy. . . . On the whole the work must be pronounced as moderately competent. Its chief weaknesses are two. First it is evident that the compositors conceived it as their duty to expand most of the contractions they found in the original. Particularly instructive in this connexion is the misprint Bet it [691] in which we catch the compositor red-handed so to speak. [Wilson also compares 35 (bewitcht), 78 (Whether), 895 (of), 1231 (needles), all of which he suspects to be compositorial expansions of copy forms.] And secondly it is clear that the compositors have introduced a large number of full stops into a text which originally contained very few, and that they have also peppered the dialogue with superfluous commas. Furthermore their pointing is careless, as is shown by the numerous instances of transposition in terminal stops. Nevertheless, apart from commas and periods the punctuation of the Quarto is comparatively good on the whole, at times even beautiful. Wilson’s belief that printer’s copy must have contained few commas and periods is skewed by his thinking that the lightly punctuated Hand-D pages in the MS The Booke of Sir Thomas Moore (BL Harl. 7368) are certainly Sh.’s and that they are typical of his punctuation; see Wilson’s What Follows if Some of the Good Quarto Editions of Shakespeare’s Plays Were Printed from His Autograph Manuscripts, Transactions of the Bibliographical Society 15, 1917–19, p. 136. Evans (ed. 1974, p. 247), following Wilson, cites the excessively heavy use of phrasal commas in Q1. Differing from Wilson, Ridley (ed. 1934, p. viii): It is true that the Quarto is very heavily punctuated, very much more so, for example, than Hamlet. But it is not on the face of it likely that a compositor, who after all is a busy working man, is going to pepper his pages with commas, or any other mark of punctuation, merely for the fun of the thing. To him the insertion of marks of punctuation is merely so much more labour, and prima facie therefore there seems no reason why we should not pay as much attention to the compositor’s commas as to any of his other marks of punctuation. . . . [T]he punctuation of the Quarto very frequently produces interesting results. In a certain number of cases it makes a real difference to the sense; in more cases it makes a real difference in the emphasis which is thrown upon phrases by their becoming more isolated; and, perhaps most important of all, it greatly diversifies the rhythms. However, according to Joseph Moxon in Mechanick Exercises on the Whole Art of Printing (1683–4) (ed. Herbert Davis & Harry Carter, 2nd ed., 1962, p. 192), it is the compositor’s duty to discern and amend the bad . . . Pointing of his Copy.

Turner (1962, pp. 33–54): There is nothing very striking about the typography; on cursory examination the book seems to be a run-of-the-mine Elizabethan dramatic quarto. Turner’s primary focus is instances of erroneous line-division of dialogue, some few of which he attributes to the compositor. (P. 48): Mislineation [of verse at 490–1, where the last word of the first line is printed as the first word of the second line] may have resulted from the compositor’s carelessness; but . . . just possibly . . . the MS rather than the workman was at fault. . . . Almost certainly the compositor was juggling the text when he set a short speech of Bottom’s and one of Peter Quince’s in a single line of type at the foot of B2 [319–20, in the inner forme (i.e., B1v, 2, 3v, 4) after, Turner thinks (see here), the outer forme (i.e., B1, 2v, 3, 4v) was already set and the] limits of B1v, B2 . . . had been established. At 413–14, the compositor apparently thought he could squeeze the first complete line of verse into the same line of type with the half-line of verse which begins the speech, a calculation which, as the turn-over shows, was none too accurate. The compositor made analogous interventions at (p. 49) 552–3, 1071–2 and (p. 54) 2063–4, but his responsibility [is] doubtful for the line division of 61–2. He set prose as verse at (p. 49) 911–12 and (p. 54) 1986–90. (Werstine [2012, pp. 144–5, n. 24] suspects the compositor was perhaps responsible for further mislineation [see here].) Turner also tabulates dozens of wrong-font errors (such as roman for italic and small capitals for full capitals) both (pp. 40–5) apparently deliberate—because of shortages of type—and (p. 40, n. 8) accidental.

Brooks (ed. 1979, pp. xxxiii–iv): In the First Quarto an editor has a text of high authority to follow. . . . [I]t is a gratifyingly clean text. . . . In the present edition, it has been found necessary to depart from Q1 in just over fifty verbal readings, besides nine punctuations affecting the sense, one transposition of a pair of lines, a number of line-divisions, and a very few places where verse was set as prose or prose as verse. Apart from the authorial lapses, and the cruces, [xxxiv] the faults are unsurprising errors of the press, almost all of the kinds that compositors are prone to.

For the supposition that Q1 lacks one or more songs, see n. 2175–2206.

Press Variants and Proof-Correction in Q1

Johnson (ed. 1888, p. 39) records the press variants listed below on sig. F1v. Wright (ed. 1891, 2:295) lists only the first of the two on sig. F1v, as does Furness (ed. 1895, p. 166).

Berger (ed. 1995, pp. vi–viii):

Collation of the eight copies reveals five [that is, six] press variants in four of the sixteen formes. . . . The inner forme of sheet A exists in three states.

Copies Collated

  • BL (British Library, C.34.k.29 . . . )
  • Bodl (Bodleian Library; C3 damaged . . . )
  • TCC (Trinity College Cambridge . . . )
  • CSmH (Henry E. Huntington Library . . . )
  • CtYEC (Yale Elizabethan Club . . . )
  • DFo (Folger Shakespeare Library . . . )
  • MB (Boston Public Library . . . )
  • MH (Harvard University; lacks C2, C3, H2, H3 . . . )

Press Variants

Sheet A (inner forme)

  • Corrected:BL, MH
  • Uncorrected:Bodl, TCC, CSmH, CtYEC, DFo, MB

Sig. A2r[4]Now] Now (turned initial N)

  • Corrected:BL, Bodl, TCC, CSmH, CtYEC, MB, MH
  • Uncorrected:DFo

Sig. A2r [18]to funerals:] ro funerals:

Sheet E (outer forme)

  • Corrected:BL, Bodl, TCC, CSmH, CtYEC, DFo, MH
  • Uncorrected:MB

Sig. E3r[1254]he] be

Sheet E (inner forme)

  • Corrected:BL, Bodl, TCC, CSmH, CtYEC, MB, MH
  • Uncorrected:DFo

Sig. E1v[1159]ſwore] fwore

Sheet F (inner forme)

  • First stage [sic] corrected:BL, Bodl, CSmH, CtYEC, DFo, MH
  • Uncorrected:TCC, MB
  • Sig. F1v[1435]notwiſtanding] notwiſtandiug
  • Sig. F1v[1438]them vp & down:] them vp & dowe:
Berger lists no second stage of correction of the inner forme of sheet F; were there such a stage, he presumably would not have stated only of the inner forme of sheet A that it exists in three states.

Ibid. (p. vi, n. 4): Often it is difficult to determine when a press variant exists, and the distinction between deliberate stop-press variants and accidental shifting and bad inking can be vexing to decide. Thus, the h in both at [125] (Sig. A3v) appears to have slipped slightly in the British Library copy, producing bot h, and the space between I and know at [1667] (Sig. F4v) is so loose that the Bodleian, Folger, Huntington, Harvard and Yale Elizabethan Club copies read Iknow. Similarly, O long at [1479] (Sig. F2r) has slipped significantly, producing Ol ong in the British, Huntington, Yale Elizabethan Club, Folger, and Harvard copies. The hyphen in loue-shaft at [536] (Sig. C1r), clear in some copies, is so weakly inked as to appear almost invisible in the British Library, Trinity College Cambridge, and Yale Elizabethan Club copies. At [1159, see above in list of press variants] (Sig. E1v), I agree with W. W. Greg that the Folger copy is variant and reads fwore . . . ; but Richard Kennedy, textual editor of the New Variorum Midsummer Night’s Dream, disagrees. Kennedy and I agree (contra Stanley Wells and Gary Taylor, William Shakespeare: A Textual Companion (Oxford, 1987), p. 287) that what appears to be a comma after melody at [201] (Sig. A4v) in the Bodleian copy is an overinked full stop.

Ibid. (p. vi, n. 7): Much of the bottom half of C3 [in the Bodleian copy] has been torn off. It has been repaired with another piece of paper, and the missing quarto text added in a post seventeenth-century hand. The copyist placed a comma after an extant deere at [695], which the Oxford editors mistook for a press variant (Textual Companion, pp. 281, 287).

Compositor Identification and Order of Printing in Q1

Turner (1962, p. 33): Neither variations in spelling nor typographical abnormalities indicate that [Q1] was set up by more than one compositor.

Turner also (pp. 34–5) provides a partial analysis of the headlines or running-titles, which (p. 34) read on both recto and verso A Midsommer nightes dreame, except on H3v where we find A Midsommer nights dreame. Below his analysis is completed in square brackets, with the roman numerals that Turner assigns to each distinctively identifiable headline and with the pages on which each occurs identified by their signatures:

I [B2, C3v, D4, E3v, F3v,] G4, H4
II [B4, C2, D2, E2, F2,] G2, H2, A2v
III [B1v, C1v, D1v, E1v, F1v,] G1v, H1v, A3
IV B3, C3, D4v, E4v, F4v, G4v, H3, A4v
V [B2v,] C1, D2v, [E1, F1,] G3, H4v
VI [B3v, C4, D3v, E4, F4,] G3v
VII B4v, C4v, D3, E2v, F3, G2v, H1, A3v
VIII [B1, C2v, D1, E3, F2v,] G1, H2v, A4
IX H3v
From this analysis Turner can establish incontrovertibly that (p. 34) the book was worked in two skeleton-formes, one regularly imposing the inner [1v, 2, 3v, 4] and the other the outer forme [1, 2v, 3, 4v] and that sheet A . . . was the last sheet to go through the press. His demonstration of the latter proposition consists of observation of significant changes [that] were made in two [running-]titles during the course of printing: (1) The g appearing in the title used on B3 and C3 (IV in the [chart above]) was replaced at D4v, and the new type appears on E4v, F4v, G4v, H3, and A4v. (2) The title used on B4v and C4v (VII) is characterized by a broken r and a defective e in dreame. At D3 a break in the M also appears, and the three defects are found together on E2v and F3 (where the e’s of dreame were exchanged in position). At G2v the r seems to have been replaced, and the e prints somewhat better than usual. When the title appears on H1, only the break in the M and the new r are evident, and only these two characteristics can be observed in the title as it appears on A3v. It is clear that sheet A was printed after sheet H.

Furthermore, according to Turner, it can be shown that the outer forme of sheet H was the first of its formes to go to press and that it is possible to generalize from this practice with sheet H and to infer that the (p. 35) outer formes of all sheets but A regularly preceded inner formes through the press. The grounds provided for these two conclusions make them far from certain. The alleged priority of presswork on the outer forme of sheet H depends initially for Turner on the pattern of reappearance of headlines from sheet H in sheet A. Both of the headlines (namely VII and VIII) used for the only two pages of the inner forme of sheet A that require headlines (namely A3v and A4, because A1v is blank and A2 bears the head title) come from the outer forme of sheet H. Then a third headline from H outer (IV)—together with two headlines from H inner (II, III)—supply the three pages of A outer needing headlines (excluding the title page, A1). Knowing that H(o) was sent to press before H(i) [and thus was returned to the compositor before H(i) so that its headlines were available for use in A(i), the first forme of sheet A to be set into type] and that earlier in the book all the outer formes were imposed in the same skeleton used for H(o), we can infer that outer formes of all sheets but A regularly preceded inner formes through the press.

This demonstration of Turner’s depends not just on the evidence he presents but also on questionable assumptions that he makes—one explicit, the others not. He implicitly assumes that Bradock printed MND Q1 by itself, rather than concurrently with other books, or (Blayney, 1982, p. 92) that MND was the only work available for composition. However, McKenzie (1969, p. 18), studying the records of Cambridge University Press from the late 17th c., discovers that concurrent printing is frequent, and Blayney (1982, pp. 45, 264 n.) finds evidence of the practice among books printed in London in the decade immediately following Bradock’s work on MND Q1. Turner also implicitly assumes that (Blayney 1982, p. 92) one of Richard Bradock’s two presses was not in use. The explicit assumption underlying Turner’s analysis is that, in terms of the production of the single book Q1 on a single press, (Turner 1962, p. 46) composition and presswork could stay more-or-less in balance. . . . [The compositor planned] to compose two formes, distribute the first, set the third, distribute the second, and so on. . . . Thus the speed of the press, which barring accidents would have remained fairly constant, is established as the rate at which about four type pages could be composed. Examination of the CUP records by McKenzie (1969, pp. 8–10) shows a wide variation in the speed of both composition and presswork not only by different workmen but also by the same workmen at different times and therefore calls into question the likelihood of compositors or pressmen maintaining the balance supposed by Turner. Only by applying these assumptions can it be assumed that the single press was still occupied printing the second forme of Q1’s sheet H as the compositor was setting and then imposing the first forme of Q1’s sheet A and further assumed that the investigator’s task at this juncture is thus simply to determine which forme of sheet H supplied headlines to the first sheet-A forme to be set. With these assumptions in place, Turner concludes H(o) was the first forme of that sheet to be wrought off the press because headlines from it appear in A(i), whereas headlines from both formes of sheet H appear in A(o). If any of Turner’s assumptions fail, then his demonstration becomes inconclusive. There is no evidence that can be adduced for any of the assumptions, which therefore have the status only of hypotheses, two of which are falsified by the evidence against them provided by McKenzie and Blayney.

Proceeding on these assumptions Turner plots the recurrence of distinctively damaged individual pieces of types in Q1, using this evidence to argue that the book was generally set into type by formes and not seriatim (that is, in the order in which pages are to be read). According to Turner, a forme that contains types only from one of the formes of a preceding sheet must have been set into type before a forme that contains type from both formes of that preceding sheet. This judgment is constructed by analogy to the one about headlines already discussed. It yields the following results concerning the order in which the pages of Q1 were set into type:

  • Sheet B (p. 41): not necessarily set by formes: It is a safe guess that all of B(o) was set before work began on B(i), but we cannot absolutely rule out such an order as B1–B1v-B2v-B3–B4v-B2[–B3v, B4];
  • Sheet C (p. 41): initial seriatim setting gave way to setting by formes: C1 (B[o] was almost certainly distributed by the time C2v was set and possibly before much of C1 was set [on the dubious evidence of the k at C1, 8] [But see also (p. 35): h B3,19–C1,1.]), 1v, 2, 2v, 3, 4v (distribution of B[i]), 3v, 4;
  • Sheet D (p. 41): D1 (distribution of C[o]), 2v, 3, 4v, 1v, 2 (distribution of C[i]), 3v, 4;
  • Sheet E (p. 42): E1, 2v, 3, 4v (D[o] distributed at line 12 or 13 of E4v), (D[i] distributed) E1v, 2, 3v, 4;
  • Sheet F (p. 43): F1, 2v, 3 (E[o] distributed), 4v, 1v, 2, 3v (Perhaps partway through setting F3v the compositor distributed E1v, but the rest of the standing type [in E(i)] seems to have been distributed after F(i) was imposed.), 4;
  • Sheet G (p. 44): G1, 2v, 3 (F1 and 3 distributed), 4v (F2v and 4v distributed), Iv, 2 (F1v and 2 distributed), 3v, 4 (F3v and 4 distributed);
  • Sheet H (p. 45): H1, 2v, 3, 4v (G[o] distributed), 1v, 2, 3v, 4
  • Sheet A (p. 45): A1v (blank) (G[i] distributed), 2, 3v, 4, 1 (title page; H[o] distributed at [i.e., before the setting of] A1 or A2v), 2v, 3, 4v.

Blayney (1982, pp. 92–3), explaining and questioning Turner’s analysis, focuses on the recurrence of individually distinctive types from sheet B in sheet C (question marks indicate types [described by Turner] as doubtful):

C1r 1v 2r 2v 3r 3v 4r 4v
From B(o) 1+? 4 2 2 ?
From B(i) 4 2
It is stated that when type reappears in this manner, composition cannot have been seriatim. The statement (which also applies to the similar evidence in sheet D) is completely untrue. The evidence is perfectly consistent with seriatim setting, with B(o) distributed before or during C1r and B(i) distributed after C3r. And in fact Turner then suggested that most of the sheet was set seriatim, except that C4v preceded C3v. The difficulty is that the evidence shows almost nothing. No matter what order is suggested, the failure to detect B(i) evidence in certain pages has to be trusted to indicate that no such evidence exists, whereas the absence of B(o) evidence has to be ascribed to a failure to detect what is really present. But if types from B(o) can be present but undetected in three (or four) pages, so can types from B(i). Because according to Turner’s assumptions (presented above) it is improbable that type from B(o) could have appeared in C1r, the evidence from that page was rejected. By rejecting different parts, and by filling in the gaps in other ways, the supposed evidence could be made to agree with almost any order of setting.

I do not suggest that Turner’s conclusions are wrong, since they may be right. The point is simply that the setting-order of sheets B-D of A Midsummer Night’s [93] Dream remains unestablished and that not enough evidence has been presented. Typographical evidence which is equally consistent with seriatim setting, setting by formes, and other possible methods; which can be supplemented by guesswork in selected pages from which it is absent; and which can be ignored selectively to suit the needs of an unsupported theory of work-flow, cannot be considered adequate. Nonetheless, Hinman (1965, p. 31) reported that a very general investigation of setting by formes . . . indicate[s] . . . pretty surely . . . MND was set in this manner, but he presented no evidence for his judgment.

In addition to the recurrence of distinctively damaged types, Turner also had recourse to type shortages indicated by the substitution of roman font for italic and small capitals for full capitals. Yet he lacked confidence in type shortages as a guide to establishing setting-order of pages (p. 40): By itself the testimony of shortages is, I believe, less reliable than that of any other bibliographical technique, and explained his reasons at length. He also anticipated Blayney’s criticism: However, the reliability of type shortage evidence can be increased when we evaluate it in the light of type reappearances, but even here we can be forced away from the most desirable position by occasionally having to take into account the evidence of only one or two reappearing types and sometimes having to argue from the non-appearance of type. Both are bad policies because mistakes in individual type identifications are easy to make and reappearances are easy to miss.

Turner used (1962, p. 35 n.) photostats of the Huntington Library copy of MND Q1 to identify distinctively damaged types. Weiss (1988, pp. 239–42) demonstrates the short-comings of this use of such a photostat. Checking Turner’s type identifications against the Folger copy of Q1, Weiss (p. 240) can confirm only 46% of the identifications. Replicating Turner’s analysis (with its questionable assumptions discussed above) using the reliable fraction of his type identifications together with new ones discovered in the examination of the Folger copy, Weiss revises Turner’s account of when during the composition of later formes earlier formes were distributed (pp. 241–2): No contradictions occurred with respect to the distributions of sigs. B and C. The appearance of ligature ft6 from D1:6 at E3.11 is one type-page before the suggested distribution of D(o) after E3. Similarly, k5 from E(o) (E4v:23) appears at F3:2, one page early. More significant differences occur in later sheets[:] the appearance of W4 from the last page of F(i) (F4:33) in the first page of G(o) (G1:8), a full gathering prior to the suggested distribution after the imposition of [242] G(i). . . . With respect to the claims that both formes of sheet G and H(o) were delayed four pages each . . . , r1 from G(o) (G2v:19) appears at H1:9, four pages before the suggested G(o) distribution point at H1v, and the appearance of w4 from G2:19 at H1:25, N2 from G3v:29 at H1:28, d16 from G2:12 at H1:31, ligature sh4 from G2:29 at H1v:1, and d12 from G2:29 at H2:24 indicate that G(i) was distributed before the composition of sig. H rather than after the imposition of H(i). Finally, the implication that sig. A was set in type from G(i) and H(o) without a distribution of H(i) seems incorrect. Appearing in A(o) (the second forme of sig. A to be set) are the following types from H(i): y4 from H2:32 at A2v:20, h7 from H1v:18 at A2v:8, and W4 from H2:32 at A4v:32 (and possibly h8 from H2:20, which may appear in A(i) at A2:19). In short, the evidence suggests a more or less normal sequence of distributions following the completion of each of the later sheets.

Valuable though Weiss’s study is for the quality of its type-identification evidence, it still does not free itself from the assumptions identified in Turner’s work by Blayney and McKenzie. While evidence of type shortages is consistent with the setting-order of the pages suggested to Turner by the type-recurrence, shortages cannot be used to establish setting-order. My inspection of the leaves of the Folger copy of Q1 under raking light fails to reveal any indentations in them such as might have been made by the type metal when the sheets were perfected, and thus fails to establish the order in which formes of sheets were printed. Holland (ed. 1994, p. 113): there is insufficient evidence as yet to establish the setting order.

Revision in Q1

Beginning in the middle of the 19th c., there arises a claim that Sh. revised the text of Q1 one or more times; the narrative of such revision remains somewhat consistent as it is elaborated by successive proponents, with the dialogue associated with the four lovers imagined to survive from Sh.’s earliest version of the play. Verplanck (ed. 1847, 2: Introductory Remarks [to MND], 6 [new pagination for each play and its accessories]): It seems . . . very probable . . . that [MND] was originally written in a very different form from that in which we now have it, several years before the date of the drama in its present shape—that it was subsequently remoulded after a long interval, with the addition of the heroic personages, and all the dialogue between Oberon and Titania, perhaps some alteration of the lower comedy; the rhyming dialogue and the whole perplexity of the Athenian lovers being retained, with slight change, from the more boyish comedy. White (ed. 1857, 4:16–17): It seems that [MND] was produced, in part at least, at an earlier period of Shakespeare’s life than his twenty-ninth year [i.e., 1593]. Although as a whole it . . . abounds in passages worthy even of Shakespeare in his full maturity, it also contains whole Scenes which are hardly worthy of his ’prentice hand . . . [17] and which yet seem to bear the unmistakeable marks of his unmistakeable pen. These scenes are the various interviews between Demetrius and Lysander, Hermia and Helen, in Acts II and III. . . . There seems, therefore, warrant for the opinion that this Dream was one of the very first conceptions of the young poet; . . . perhaps . . . he . . . went from Stratford up to London with it partly written; . . . when there, he found it necessary at first to forego completion of it for labor that would find readier acceptance at the theatre; and . . . afterward, when he had more freedom of choice, he reverted to his early production, and in 1594 worked it up into the form in which it was produced. . . . At least some of the additions might have been made . . . for a performance at Court. . . . Except in the play itself I have no support for this opinion, but I am willing to be alone in it.

Fleay (1878, p. 61): MND probably was recast previously to publication. Idem (1886, pp. 181–6) dates the version for the public stage to (p. 183) the winter of 1592, (p. 181) its present form to 1595. January 26, . . . the date of the marriage of William Stanley, Earl of Derby, and subsequent modifications to produce the court version to (p. 182) the winter of 1594–5. (P. 182): the traces of the play having been altered . . . are numerous. There is a double ending. Robin’s final speech [2207–22] is palpably a stage epilogue, while what precedes from Enter Puck [2153] to [2206] is very appropriate for a marriage entertainment, but scarcely suited to the stage. In Acts iv. and v., again, we find in the speech-prefixes Duke, Duchess, Clown, for Theseus, Hippolita, Bottom: such variations are nearly always marks of alteration, the unnamed characters being anterior in date. In the prose scenes speeches are several times assigned to wrong speakers, another common mark of alteration. (P. 183): wherever Robin occurs in the stage-directions or speech-prefixes scarcely any, if any, alteration has been made; Puck, on the contrary, indicates change. (P. 185): The time-analysis . . . has probably been disturbed by omissions in producing the Court version. [138–265] ought to form, and probably did, in the original play, a separate scene; it certainly does not take place in the palace. To the same cause must be attributed the confusion as to the moon’s age; cf. [222–3] with the opening lines [5–14]: the new moon was an afterthought, and evidently derived from a form of the story in which the first day of the month and the new moon were coincident after the Greek time-reckoning. Idem (1891, 2:194): The play has certainly alternative endings: one a song by Oberon for a marriage, and then Exeunt, with no mark of Puck’s remaining on the stage; the other an Epilogue by Puck, apparently for the Court (cf. gentles in [2213]). It might seem, as the Epilogue is placed last, that the marriage version was the earlier, and so I took it to be when I wrote my Life of Shakespeare [1886, quoted above]; but the compliment to Elizabeth [524–45] was certainly written for the Court; and this passage is essential to the original conduct of the play, which may have been printed from the marriage-version copy, with additions from the Court copy. This would require a date for the marriage subsequent to the Court performance. One version must date 1596, for the weather description [463–92], which can be omitted without in any way affecting the progress of the play, requires that date. I believe this passage was inserted for the Court performance in 1596, that on the public stage having taken place in 1595; but that the marriage presentation, being subsequent to this, was most likely at the union of Southampton and Elizabeth Vernon in 1598–9. Chambers (ed. 1897, pp. 14–15): There are certain indications which make me think that [MND] was also at some period [after its composition in 1594–5] slightly retouched. Two passages, [1204–1384+1] and [1793–1902], show a markedly larger proportion of feminine endings than the rest of the play. In the earlier passages, this may be due merely to the excited state of the speakers, but I cannot resist the suspicion that the opening of act v. shows some traces of later work. See also Luce (1906, p. 157). Noble (1923, p. 58 n. 1): The Quarto did not use italics for songs. My own belief is that the whole of the fairy part in the final episode is a comparatively late addition. Witness the fact that Oberon can sing and lead a chorus in Act V, a faculty of which he evinces no sign in the rest of the play.

Wilson (ed. 1924, pp. 80–153): The Q1 text emerges from three distinct episodes of Shn. composition in 1592–3, 1594–5, and 1598. What remains in Q1 of the 1592 version are the (p. 91) lovers’ scenes—those featuring Helena, Hermia, Demetrius, and Lysander—wherein the psychology is generally as crude as the verse (p. 92) is stiff and antithetical. Wilson dates these scenes to 1592–3 because they include 1849–50, which he, like Fleay (1886, p. 183) reads as an allusion to [Robert] Greene’s death [on 3 September 1592] (p. 94). To Sh.’s 1594–5 revision belong Q1’s mechanicals and Bottom scenes, most of the passages in which Robin is used in stage directions and speech prefixes to designate the character otherwise called Puck (including the epilogue—2207–22), parts of the fairy scenes (Wilson, ed. 1924, pp. 95–6), and the introduction of Hippolyta into 1.1 and 4.1. This revision includes the mechanicals’ concern over frightening the ladies with too realistic a representation of a lion (838–56), which Wilson, again like Fleay (1886, p. 185), takes to be an allusion to a spectacle from the celebration in the Scottish court of Prince Henry’s baptism (30 August 1594) for which it was prudently decided to substitute a blackamoor (Wilson ed. 1924, p. 95) for the lion that was initially to have drawn a triumphal car (ibid.). The revision also includes a description (463–92) of what Wilson takes to be the wet and chilly summer of 1594 (ibid.). Sh.’s final handling of MND in 1598 gives Q1 its mature Shakespearian verse, in which the masterly diction and vigorous sweep . . . introduce a note of intellectual energy that makes the whole glow with poetic genius (p. 183). Some such verse Wilson imagines to have been added in short passages written in the margin of the 1594–5 version, with other longer passages interpolated on additional leaves. For Wilson, as for Fleay, the use of Puck for Robin in stage directions and speech prefixes is peculiar to the 1598 revision and is the clue to its purpose, namely the introduction of the little western flower. All but two occurrences of Puck (1028 and 2153–4 being the exceptions) are associated with references to the flower, which functions as a compliment and representation of Elizabeth Vernon, daughter of Sir John Vernon of Hodnet Hall, Shropshire (p. 100) on the occasion of her marriage to the Earl of Southampton in 1598—the occasion for which Sh. added the wedding masque at 2153–206. Wilson (1962, p. 206) substitutes the wedding of Thomas Berkeley to Elizabeth Carey at Blackfriars on 19th February 1596 for the Southampton-Vernon wedding as the occasion for the second revision.

According to Wilson, the following derive from the 1592–3 version: most of 1.1 (2–265) including Helena’s entry at 24–5; 566–625 in 2.1; 686–717 and 737–811 in 2.2; 1063–1124 and 1146–1221 in 3.2; 1792–1881 (although with additions from 1598) in 5.1. To the 1594–5 belong 3–23, 131–5, and the splicing together of two 1592–3 scenes at 136–7 in 1.1; 1.2 (266–371); the beginning of 2.1 (373–523); all but one line of 3.1 (813–1020); 1021–62 (although this passage was later revised in part) and 1440–1506 in 3.2; 4.1 (1509–1745) though certain parts look like first draft material recopied (p. 131)—Wilson specifies 1624–48, 1711, and 1722–3 as 1594 additions—and three more additions were made in 1598; 4.2 (1746–89); 1882–1985 (but 1890–1904 were added in 1598), 1986–2152 (with three minor additions from 1598), and 2207–22 in 5.1. The 1598 revision consists of 153–9 in 1.1; 524–65 and 626–49 in 2.1; 650–85 and 718–36 in 2.2; 901 in 3.1; 1125–45 and 1222–1439 (which was partially recopied and revised in 1598 [p. 125]) in 3.2; 1586–90, 1604–10, 1690–2 in 4.1; 1797–1800, 1804–9, 1822–5, 1828–31, 1855–7, 1863–7, 1874–5, 1879–80, 1890–1902, 2015–16, 2138–40, 2143–4, and 2153–206 in 5.1. Craig (1931, p. 335): Revision of some sort is unmistakable in the case of A Midsummer Night’s Dream. (See also Craig 1961, p. 108.)

Wilson’s revision hypothesis is not left to collapse under the burden of its accumulated speculation. Reviewing his ed., Chambers (1925, pp. 342–4) finds the case for revision to contain inconsistencies: Professor Wilson . . . thinks that, while I.1 and IV.1 belong substantially to the [earliest version], certain awkwardnesses in the introduction of Hippolyta suggest that she was an afterthought, connected in some obscure way with the indication in [4–5] of a four-day period for the action, which is not consistent with the time-analysis. I do not suppose that he would lay much stress on this, especially as he accepts Hipployta as part of his [earliest-version] substratum of V.1. Wilson, Chambers notes (p. 344), makes much of the variation in the naming of Robin-Puck, but nothing of that of Bottom-Clown. Yet Chambers does agree that 5.1 was revised and compliments Wilson on (p. 343) a valuable bibliographical contribution in his attention to the persistent mislining by the printer of passages in [1798–1800, 1804–9, 1822–5, 1828–31, 1855–7, 1863–7, 1874–5, 1879–80], which suggests that these passages were taken in from cramped marginal alterations in the copy. However, unlike Wilson, Chambers does not think the alleged marginal alterations can be dated years later than the context into which they have been supposedly interpolated. I agree again that the fairy-mask of V.1 [2153–206] and the epilogue of [2207–22] are probably duplicate endings. Still Chambers refuses to acknowledge that the adaptation issuing in this duplication need date from as late as 1598. (See also Chambers 1930, 1:360–1.)

Greg (1942, pp. 124–5), though, altogether rejects Wilson’s theory of revision, and subsequent editors join Greg both in this rejection and in his substitution of so-called foul papers (see here for Greg’s definition of this term) as the explanation for any discrepancies in Q1. According to Greg, whether or no the two endings were written at the same time, it would not be surprising to find both in the foul papers in their present order. Nonetheless, Greg does preserve, with modifications, Wilson’s conception of revision at the beginning of 5.1 (p. 125): On the whole I think the copy for Q must have been the author’s manuscript. . . . The most important piece of evidence is at the beginning of the last act where eight passages of verse are mislined [1798–1800, 1804–9, 1822–5, 1828–31, 1855–7, 1863–7, 1874–5, 1879–80]. Wilson has pointed out that if these are removed what remains is perfectly regular and consecutive, and he argues that they represent marginal additions in the copy. . . . It seems to me quite possible that Shakespeare, coming back to his work in a fresher mood, found what he had written rather flat and sought to brighten it up. And I cannot believe there was anything like the amount of revision Wilson imagines [elsewhere in Q1]: it would certainly have left other traces of the sort, whereas the text is elsewhere particularly clean. Greg later (1955, p. 243) is even more dismissive of Wilson’s theory, calling it all very ingenious; the difficulty is to believe that this refashioning would not have left plainer bibliographical traces than are now apparent. Nonetheless Greg adheres to Wilson’s theory of alleged marginal alteration. Spencer (1930, pp. 24–5) does not agree that, if the eight mislined passages are removed what remains is perfectly regular and consecutive, as Greg put it. Instead, Spencer notes that what remains does not cohere as well as the whole existing text does with the eight passages in place. He argues that it is hardly conceivable that a reviser, expanding certain speeches, should make the new joints less conspicuous than the old, since the new represent elaboration, while the old represent the flow of his thought as originally conceived. For example (p. 25), the description of the interlude in the schedule read by Theseus [1852–3] specifies four qualities: the piece is tedious, brief, tragical, and merry, all these qualities lying within the original text as segregated by Professor Wilson. But in the next speech, which contains Philostrate’s explanation, only tedious and brief lie within Mr Wilson’s [original text: 1858–61]. For tragical and merry we must subjoin . . . one of the additions [1863–7]. Kirschbaum (1946, p. 48) supports with another example Spencer’s supposition that what Wilson calls the original text and what he calls the additions were written at the same time: it will be noticed that in [1796] the sequence is Louers first, mad men second. Omitting the so-called addition [1797–1800], we see that the sequence in [1801–3] is mad man first, louer second. Why the shift in sequence if Shakespeare wrote [1796–1803] originally without . . . [1797–1800]? But when we look at [1799–1800, a supposed addition] . . . we see that the sequence lunatics first, louer second, Poet third is the sequence followed in [the allegedly original 1801–3] . . . and [the allegedly later addition 1804–9]. . . . In other words, the sequence followed in the supposedly original version in [1801–3] is not the sequence first indicated in the supposedly original version at [1796] but the sequence indicated in the supposed marginal addition at [1799–1800]. Thus, the ensuing hypothesis is that both [alleged original and alleged later addition] were written at one and the same time and that the [allegedly additional 1799–1800] . . . was written before and not after the [the allegedly original 1801–3]. And since the sequence indicated in [1799–1800] is followed in [1801–9], it may be suggested that [1804–9], the lines on the poet [a supposed addition], were not an afterthought but were written immediately after [1801–3]. See also Lull (1998) on these alleged Shn. revisions.

Turner (1962, pp. 49–50) seeks to corroborate Wilson’s revision theory in a number of places in Q1, including where the eight passages of verse are mislined. Noting that sigs. C1–3, which, according to Turner, are unusual in being set seriatim, contain, on Wilson’s theory, passages from all three stages of composition (497–681), Turner thinks it possible that the workman was confronted here with particularly nightmarish copy. Again, on Turner’s analysis, work on F(o) and the first two pages of F(i) went slowly; and this is another part of [50] the text ([1371–1506], ending near the foot of F2v) which Wilson thinks to have been considerably worked over. Problems with Turner’s method (discussed above, here) compromise any possibility of his analysis buttressing Wilson’s theory. Other problems obtrude in Turner’s justification of Wilson’s interpretation of the eight mislined passages in sig. G (5.1) as evidence of revision. Werstine (2012, p. 145) both summarizes and criticizes this justification: Close bibliographical analysis of the quarto by Robert K. Turner, Jr., shows that quire G is peculiar not only for the frequency with which its verse is mislined, but also for containing four pages with fewer lines of type than is normal. Usually each page has 35 lines, but sigs. G1r, 1v, and 2v have only thirty-four lines each, and G2r only thirty-two ([Turner] 1962, 39). In all, then, quire G is short six lines of type. Six is also the number of lines of type that are saved as an apparently accidental consequence of the mislining of five of the eight passages of verse that are erroneously divided; three of the mislined passages occupy the same space as they would if they were properly set. No one, including Turner, has remarked on this coincidence or attempted to account for it. Turner attempts to explain away the four short pages as follows: in casting off copy for quire G, the compositor, who, he assumes, cast off his own copy, evidently counted in some material that he later did not set[;] . . . he may have failed to notice that some lines here and there were supposed to be cancelled ([ibid.], 54). Such an explanation fails to convince because throughout Q Turner can find only two other places where there may have been errors in casting off so that the compositor had to juggle the lineation of the text in order to fit copy to a predetermined space (ibid., 55)—B2r, [319–20] and H2r, [2063–4]. His explanation then forces us to believe that Quire G is the unique site not only of a considerable amount of mislined verse and short pages (which, according to Turner, bear no relation whatsoever to the mislined verse) but also of a considerable number of misleadingly cancelled lines in its copy (for which there can be, in the nature of the case, no surviving evidence). Without any adequate explanation for the short pages of quire G, Wilson’s theory about the source of the mislined verse in the quire must remain shrouded in doubt. It is instead possible that whoever cast off copy for quire G found no difficulty in printer’s copy and counted off the lines with the same meticulous accuracy found in almost all the rest of his work, and then the compositor, who sometimes unaccountably, if only occasionally, mislined the text elsewhere (e.g., . . . H1r-H1v [1986–9] . . . ), made the mistakes in dividing verse that Wilson attributes to printer’s copy. Such an explanation may not be the right one, but at least it relates the bibliographical anomaly of the short pages to the textual anomaly of the frequently mislined verse.

After Greg’s dismissal of Wilson’s revision hypothesis, Smidt (1986, pp. 123, 128, 130–4, 140, 210) is reluctant to develop in any detail a theory of revision on the basis of so-called unconformities in Q1: One might suppose that the little Indian is left as a residue from an early attempt to work out an appropriate fairy plot, before the fairies became involved with Duke Theseus and his bride. He has lost his raison d’être as a cause of the fairies’ quarrel, but he is still useful as a means of providing a solution to it. (P. 128): The variation between the names Robin and Puck may well be a sign of different stages of composition, as some scholars have thought, but there may have been merely an expansion of Puck’s character as the writing of the play advanced, not a substitution of a spirit for a gnome. And this may have occurred during a continuous process of composition, so that the fairy Puck is not necessarily a sign of revision. (P. 130): The most interesting phenomenon as far as unconformities are concerned is the fitful appearance of the moon during the night in the wood. . . . [132] It is darkness . . . , only lightened for a while by the stars, that prevails while the lovers are in the wood. . . . There is no reference to present moonlight in the rehearsal scene [813–936]. Yet (p. 133) there is in fact enough indication of a moon in the fairy scenes to make the mention of a new moon on the wedding night [by Hippolyta at 12] well-nigh impossible, in spite of our expectations. . . . [134] If the inconsistency was accidental, was it brought about by the merging of different plot components, Theseus and Hippolyta on the one hand, the fairies on the other hand, the young lovers and the artisans in between? (P. 140): The contradictions in this comedy were not brought about by changes of mind, and, with one possible exception, not by inadvertence. The exception is the moonlit fairy scenes. Only (p. 210) the repetition of Theseus’s order to Demetrius and Egeus to accompany him [123–5 and 132–5] suggests some kind of textual disturbance. For reengagement with Wilson’s theory of revision, see Hunter (1998 and 2002, pp. 3–6).

Printer’s Copy for Q1

Speculation on this question has given rise to four suggestions, all of them testifying to perceptions of the high authority of the Q1 text: (1) Sh.’s own MS; (2) Sh.’s own MS as marked up by his company for production; (3) a transcript of Sh.’s MS made and used in the playhouse; (4) a non-theatrical transcript. The cautious commentators have found language to avoid or at least to qualify precise identification.

Discussion begins with Capell, who seems to opt for the third alternative. He (ed. 1768, 1:3 ff.) identifies fourteen quarto texts, including MND, that ought to be excluded from Heminge and Condell’s characterization of all the Shakespeare quartos as diuerse stolne, and surreptitious copies, maimed, and deformed by the frauds and stealthes of iniurious imposters (p. 10): Let it then be granted, that these quarto’s [sic] are the Poet’s own copies, however they were come by; hastily written at first, and issuing from presses most of them as corrupt and licentious as can any where be produc’d, and not overseen by himself, nor by [11] any of his friends. . . . It may be true, that they were stoln; but stoln from the Author’s copies, by transcribers who found means to get at them: and maim’d they must needs be, in respect of their many alterations after the first performance. . . . [12] The very errors and faults of these quarto’s . . . are, with the editor [Capell], proofs of their genuineness; For from what hand, but that of the Author himself, could come those seemingly-strange repetitions [of passages in LLL and Tro.], . . . those imperfect entries . . . ? Capell’s use of transcribers suggests that he thinks copy to have been scribal transcripts, and his reference to their many alterations after the first performance locates transcription in the playhouse, where copy would be subject to theatrical adaptation.

Similarly Malone (in Boswell, ed. 1821, 1:203) in general recognizes the superiority of the early quartos to counterpart F1 texts, except for Wiv. and H5 (Q1 Ham. was not discovered until 1823), yet does not attempt to specify just what kind of MSS served as printer’s copy: With respect to the other thirteen copies [quartos, including MND] . . . , they in general are preferable to the exhibition of the same plays in the folio; for this plain reason, because, instead of printing these plays from a manuscript, the editors of the folio . . . printed the greater part of them from the very copies which they represented as maimed and imperfect, and frequently from a late, instead of the earliest, edition; in some instances with additions and alterations of their own.

Clark and Wright (ed. 1863, 2:viii), cautiously subjunctive: Fisher’s edition . . . may have been taken from the author’s manuscript. Furness (ed. 1895, p. xii) takes pains to imagine that Heminge and Condell may not have been guilty of a wilful untruth, as alleged by Malone, when they implied that they provided for F1 only his [Sh.’s] papers as printer’s copy, but, in fact, provided quarto copy for such plays as MND if they knew that this [quarto] text was [originally] printed directly from his manuscript. Chambers (ed. 1897, p. 8), leaving open all alternatives for printer’s copy: Q1 has been printed from a clear and authentic manuscript. Cuningham (ed. 1905, p. xiii) adds to the alternatives the possible use of actors’ parts in the creation of printer’s copy, an idea apparently borrowed ultimately from Johnson’s 1756 Proposals for Printing . . . the Dramatick Works of William Shakespeare (ed. Arthur Sherbo, 1968, 7:52): Q1 was printed in all probability, either from the authentic MS. of Shakespeare himself, or at least from an accurate copy or, perhaps, copies of the actors’ parts, transcribed in the theatre from the original MS.

Furness (ed. 1895, p. xi) misleads a number of his successors with the observation that in both Q1 and Q2 the stage directions are, as in copies used on the stage, in the imperative, such as wind horns, [1622] sleep [1484]. So Cuningham (ed. 1905, xv) notes stage-directions . . . in the imperative, as is customary in stage copies as his ground for suggesting the possible playhouse provenance of printer’s copy. And Pollard (1909, p. 72) declares that The imperative form of the stage directions, Ly doune ([1110]) and Winde hornes ([1622]) may be taken as indicating its origin from a playhouse copy. (See Idem 1920, p. 64.) Greg (1942, p. 37) properly questions the assumption that the prompter can be identified by the use of the imperative: The prompter writes directions for his own use; they are generally terse and to the point. Chambers [1930, 1:118] questions whether they are usually in the imperative. They are not: but being short and curt they tend to imperative and participial constructions. Wilson (1945, p. 67) provides examples of the imperative in Anthony Munday’s hand, not the hands of the theatrical annotator(s), in the theatrical MS of Iohn A kent & Iohn a Cumber; Werstine (2012, p. 229) gives examples of imperative SDD in authorial hands from MSS of Thomas Heywood’s The Captives and the anonymous The Waspe. Greg (1942, p. 125), contradicting himself, follows Pollard in identifying the hand of the prompter in the appearance of imperative SDD: There are however [in Q1] a few directions that suggest the prompter, such as Lie down and Wind horn. In the palpable duplication, Enter Lovers; Lysander, Demetrius, Hermia, and Helena ([1819–20]), we may see the book-keeper expanding a typically brief direction of the author’s. (Werstine [2012, p. 131] notes that actual theatrical texts rarely show a bookkeeper specifying a group in terms of their proper names, finding only one example, from the annotated quarto of Thomas Lodge and Robert Greene’s A Looking Glasse, for London and England.) Greg differs from Pollard, though, in judging printer’s copy to be, not a playhouse copy, but so-called foul papers (see here for Greg’s definition of this term) containing (1942, p. 125) notes made in preparation for the prompt-book. Greg’s belief in the possibility of such a document once having existed arises from his interpretation of a single extant MS, Thomas Heywood’s transcription of The Captives, as such a document, but as Werstine (2012, pp. 300–9) observes, this interpretation is at odds with features of the MS. Greg rules out (1942, 125) a playhouse transcript to be printer’s copy only because he believes that there could only ever have been one such transcript and that it was used to annotate printer’s copy for F, which differs from Q1 chiefly in terms of such annotations.

Pollard (1920, p. 63), while not setting aside his belief in theatrical annotation of printer’s copy, anticipated Greg in arguing as well for the possibility that such annotated copy could originally have been inscribed by Sh.: Possibly in some cases, if [a dramatist] were familiar with the theatre, he might use the same technical language as a prompter, so that Shakespeare himself, in the scene in the wood in the Midsummer Night’s Dream, may have written the directions, Enter a Faerie at one doore and Robin goodfellow at another, Enter the King of Fairies at one doore, with his traine; and the Queen at another with hers, the doors, of course, being those of the stage, not of the wood. Adams (1923, p. 519) states an opinion that can be interpreted to be identical to Pollard’s but need not necessarily be: MND is printed from authentic playhouse copy; compare Neilson & Hill, ed. 1942, p. 88. Pollard (1923, p. 7): some of the flaws in these Good Quartos [including MND Q1] are the result of imperfections in Shakespeare’s own work, and I have ventured to claim that some of these Good Quartos may actually have been set up from Shakespeare’s autograph manuscripts. Wilson (ed. 1924, p. 80) is of Pollard’s mind: beyond doubt [Q1 is] printed from a theatrical prompt-book, . . . [with] the managerial voice giving real directions to the players; he quotes 1110, 1484, and 1622. Printer’s copy also contains irregularities strongly suggestive of an author’s manuscript, and so it is both Shakespeare’s autograph manuscript and the prompt-book just as Shakespeare left it. Kittredge (ed. 1936, p. 229) and Parrott (ed. 1938, p. 131) follow Wilson. Chambers (1930, 1:358), persuaded to be more specific than in 1897: Q1 may be from the author’s manuscript. Kirschbaum (1955, pp. 172–3) stands out against the idea of theatrical annotation of a Sh. MS: Greg’s evidence for prompter’s additions does not pass muster. The directions could just as well come from the author, and there is no reason why they should not be copied by a scribe. (P. 173): The nomenclature [variation in naming of characters in SPP and SDD] . . . does not show the author in the heat of composition—does not cause the reader to assume foul papers behind the print. . . . There is little evidence of foul papers or playhouse in Q. Nothing in it rules out printing from transcript. . . . There is neither internal nor external evidence to show that the copy came to the publisher from Shakespeare’s fellows.

In 1955 Greg establishes what becomes a virtual consensus among 20th-c. editors that printer’s copy for Q1 is authorial foul papers defined as (1955, pp. 106, 142) a copy representing the play more or less as the author intended it to stand, but not itself clear or tidy enough to serve as a prompt-book, that is, a theatrical MS used to guide performance, because it contained (p. 142) loose ends and false starts and unresolved confusions. Greg presumed that book-keepers necessarily tidied away from prompt-books certain features of foul papers, including seven features still to be found in Q1 (ibid., pp. 240–2):

  • (1)multiple designations of the same character (p. 241): Oberon is King of Fairies (or simply King) or Oberon indifferently; Titania is of course named in the text [i.e., dialogue and SPP], but in directions [the proper name] appears only at [650] on her second entry [otherwise she is named Queene in SDD]; Bottom, on his most important appearance [1509] is merely Clowne; Theseus and Hippolyta, after long appearing by name, become as a rule Duke and Duchess after the play begins in Act V; lastly Robin (Goodfellow) and the generic Puck alternate;
  • (2)indefinite entrances involving speakers (p. 240): after carefully naming the rude mechanicals in I.ii, the author later contents himself with the description the Clownes (III.i) or the rabble (IV.ii); the young couples, having been named in I.i, become simply Louers at V.i . . . , for here the names [i.e., Lysander, Demetrius, Hermia, and Helena] are a palpable addition; the fairies are quite indefinite, except for the foure at [980]. Greg also quotes as allegedly bearing the characteristics of the author and thereby implicitly needing the bookkeeper’s attention SDD containing the term traine to refer to speakers attending royalty (p. 240: her [i.e., Titania’s] traine (650); all his [i.e., Theseus’s] traine (1622);
  • (3)other indefinite SDD: Greg’s characteristically authorial SDD include (p. 240) Enter Theseus, Hippolita, with others (2), where others refers to supers, rather than speakers, including Philostrate, addressed by name in this scene;
  • (4)inclusion of an unnecessary character in a SD (p. 241): Pucke is made to enter with Oberon at [1021], though in fact he only does so three lines later;
  • (5)missing entrances and exits;
  • (6)descriptive SDD: characteristic of the author, according to Greg, are, for example (p. 240), Enter Egeus and his daughter Hermia . . . and Enter Quince, the Carpenter; and Snugge, the Ioyner . . . ;
  • (7)marginal insertions of dialogue (pp. 241–2): Greg fully endorses the theory of J. Dover Wilson’s that some passages of verse at the beginning of the play’s last scene are wrongly divided in Q1 because they were marginal additions to the dialogue, their position in the margins forbidding their writer from dividing them properly as verse and the compositor following his copy in this error (see here).

Greg also mentions the following incidental features of Quarto MND as characteristic of foul papers: erroneous SPP—for Greg, one of Shakespeare’s (ibid., p. 247 Note A) oversights in composition, the author having written consecutive speeches for Flute and Thisbe, forgetting that they were the same; double entrance (p. 240): Helena enters in 1.1 both at 25 and at 191; a dialogue error in naming (ibid., p. 246 Note A): Flute for Snout at 1957; and an erroneous SD (ibid., 240): Enter Quince, Flute, Thisby and the rabble (1746)—in this scene Shakespeare had forgotten that Flute and Thisbe are one.

Werstine (2012, p. 132): The appearance of these features in the earliest printing of A Midsummer Night’s Dream leads Greg to conclude that we can hardly imagine that Q represents a finished prompt-book [1955, p. 241] and therefore, by this process of elimination of what he regards . . . as the only possible alternative, Quarto Dream must represent Shakespeare’s foul papers—in spite of the persistence of the same features in the actual theatrical texts that he calls promptbooks, which destroys his argument. Werstine reviews each of Greg’s features and compares them to what is found in the twenty-one extant texts (both MS and annotated quartos) that bear theatrical annotation and thereby show what book-keepers actually did and did not do to their playbooks (these texts are described in Werstine 2012, pp. 234–57). In these twenty-one Werstine finds (1) multiple designations of the same character (pp. 359–64), (2) indefinite entrances involving speakers, including the very terms cited by Greg to discount a theatrical MS—crewe, trayne, the rest (pp. 375–9), (3) other indefinite SDD, including uses of others identical to those cited from MND by Greg (pp. 379–82), (4) inclusion of an unnecessary character in a SD (p. 384), (5) missing entrances (pp. 374–5) and exits (pp. 386–8), (6) descriptive SDD such as, from the scribal MS Ironside: Enter Edmond and Alfricke the generall vnder the kinge:/ (1.3.332) and Enter Edricke a poore man . . . (2.2.461). (Needlessly explanatory SDD, like those that Greg quotes from the MND quarto, are also frequent in The Second Maidens Tragedy [also called The Lady’s Tragedy], another scribal MS, in further indication that such SDD are by no means peculiar to authorial MSS: Enter the new Vsurping Tirant; The Nobles of his faction, Memphonius, Sophonirus, Heluetius with others, The right heire Gouianus depos’de [1.1.1–3]; Enter L Anselmus the deposde kinges brother, wth | his Frend Votarius [1.2.257–8]; Enter the ladye of Gouianus . . . [2.1.636]; Enter Tirant wondrous discontedly: Nobles afarr of [4.2.1655–6]; Enter Votarius with Anselmus the Husband [5.1.1984]); (7) marginal insertions of dialogue (pp. 388–9). Even the minor features cited by Greg to identify foul papers as printer’s copy are located in theatrical MSS by Werstine: erroneous SPP (pp. 371–2), double entrances (pp. 385–6), a dialogue error in naming (p. 371), and erroneous SDD (p. 385).

Greg himself showed his awareness of the flaw in his reasoning when he wrote (1955, p. 142) It must, however, be recognized that owing to the casual ways of book-keepers these characteristics may persist, to some extent at least, in the prompt-book. In reproducing his argument and conclusions about copy for MND Q1, editors fail to attend to this caveat: Doran (ed. 1959, p. 174); Wells (ed. 1967, p. 165); Evans (ed. 1974, p. 247); Brooks (ed. 1979, pp. xxii–v); Foakes (ed. 1984, pp. 135–6); Wells & Taylor (1987, p. 279); Holland (ed. 1994, pp. 113–14); Berger (ed. 1995, p. ix). Horwood (ed. 1939, p. 10) accepts what would become Greg’s argument and conclusion in the form in which it was first presented in McKerrow (1935)—Greg’s inspiration—with reference only to variation in naming in SPP and SDD. Greg’s definition of foul papers has also been overturned in Werstine (2009, pp. 44–5).

A significant problem with Greg’s use of variation in naming of characters in SPP in MND Q1 and other texts as evidence of so-called foul papers had already been identified by Kennedy (1998, pp. 178–9): There may, however, be another explanation for the variation in SP’s in some of Shakespeare’s early texts. The change may not be authorial at all, but compositorial. It seems to have been a printing-house convention that a compositor did not have to follow copy in the matter of SP’s, but could choose to call characters by their first names or last names, or [179] generic names or personal names, or by their functions or peculiarities. If he needed to, he could vary the SP from Quin. to Pet[er], from The[seus] to Duke, . . . and so on. Most of the time, variant SP’s do not point to authorial foul papers, but signify compositorial change. And most of the time, variant SP’s are not signs of an author’s revising, or of an author in the heat of composition, but are rather indications of a compositor switching SP’s because of type shortage. Kennedy applies this theory in detail to the variation in naming in MND Q1’s SPP (179–90).

To support the view that printer’s copy is in Sh.’s own hand, some recent editors have followed Wilson (ed. 1924, pp. 112, 116, 121, 148) in adducing certain spellings in MND Q1 as the same as or somewhat analogous to spellings in the Hand-D pages of MS The Booke of Sir Thomas Moore (BL Harl. 7368), in the belief that Hand D is Sh.: Evans (ed. 1974, p. 247), Brooks (ed. 1979, pp. xxv–vi), Foakes (ed. 1984, p. 136), Wells & Taylor (1987, p. 279), Holland (ed. 1994, p. 114); Berger (ed. 1995, p. ix). However, recent review of Hand D’s spelling by Jackson (2007) shows that only a half-dozen are sufficiently rare to constitute acceptable evidence for attribution, and none of these is in MND Q1.

There is no evidence for a rational choice among the following alternatives for MND Q1 printer’s copy: (1) Sh.’s own MS; (2) Sh.’s own MS as marked up by his company for production; (3) a transcript of Sh.’s MS made and used in the playhouse; (4) a non-theatrical transcript. While there is nothing in the way of playhouse notes in the quarto to demonstrate alternatives 2 and 3, Werstine (2012, p. 4) shows that in some actual playhouse MSS there is so little annotation that it is possible that MSS with no annotation could have been used in production.

Quality of Printing in Q1

While on the whole positive, editors and critics exhibit a wide range of opinion about Q1’s quality as a witness to what Sh. wrote. Pope (ed. 1725, 1:xv), not attending to the question of the priority of Q1 to Q2: If any were supervised [at the press] by himself [Sh.], I should fancy the two parts of Henry the 4th, and Midsummer-Night’s Dream might have been so; because I find no other printed with any exactness; and (contrary to the rest) there is very little variation in all the subsequent editions of them. Johnson (ed. 1765, 1:176) unable to establish the priority of Q1 to Q2: Neither of the editions approach to exactness. Building on Capell’s (1783, 2.3:115) inference of Shakespearean authority for the punctuation of the mechanicals’ prologue (see n. 1906–15), Knight (ed. 1839, 1:331) writes, The original of these editions, whichever it might be, was . . . carefully superintended through the press. The text appears to us as perfect as it is possible to be, considering the state of typography in that day. There is one remarkable evidence of this. The prologue to the interlude of the Clowns, in the fifth act [1906–15], is purposely made inaccurate in its punctuation throughout. . . . ; and this is precisely one of those matters of nicety in which a printer would have failed, unless he had followed an extremely clear copy, or his proofs had been corrected by an author or an editor. Compare Hudson (ed. 1851, 2:255) and White (ed. 1857, 4:17), as well as the following writing after Q1’s priority had been demonstrated: Chambers (1930, 1:358), Parrott (ed. 1938, p. 131), Greg (1942; 1954, p. 125), Doran (ed. 1959, p. 27), Holland (ed. 1994, p. 112). Clark and Wright (ed. 1863, 2:viii), guessing right about Q1’s priority, nonetheless regard it carelessly printed. Kittredge (ed. 1936, p. 229) echoes this judgment, describing Q1 as not very carefully printed.

Furness (ed. 1895, pp. x–xii), also guessing right about Q1’s priority: the excellence of the text [of Q1] is counterbalanced by the inferiority of the typography. . . . [A]lthough the entrances of the characters are noted, the exits are often omitted, and spelling throughout is [xi] archaic, for instance, shee [241], bedde [228], dogge [589], &c., betraying merely a compositor’s peculiarity. . . . [F]onts are mixed, and the type old and battered. Believing that the Q1 compositor set type by the ear from dictation, Furness finds such errors as Dians bud, or Cupids flower [1588], instead of Dian’s bud o’er Cupid’s flower the consequence. (P. xii): [I]t is assuredly more likely that such blunders as Eagles [454] for AEgle, or Peregenia [453] for Perigouna . . . are due to the deficient hearing of a compositor. However, Furness (p. xii) concedes that compositors . . . are exposed [to such errors] when with a retentive memory they carry long sentences in their minds, not just when they set from dictation. Cuningham (ed. 1905, p. xi), in the main closely following Furness: the text . . . has reached us in a state of comparative correctness and purity, [yet] there are passages which are admittedly corrupt. Ibid. (p. xv): the text is superior, and likewise the punctuation.

Wilson (ed. 1924, p. 79): Q1 is superior to many of the other quartos. . . . The misprints are few, and the literals fewer. The compositors seem to have worked slowly, whether through inexperience or because they found the manuscript difficult to read; for the text contains a number of interesting archaic spellings which almost certainly derive from the copy. . . . On the whole the work must be pronounced as moderately competent. Its chief weaknesses are two. First it is evident that the compositors conceived it as their duty to expand most of the contractions they found in the original. Particularly instructive in this connexion is the misprint Bet it [691] in which we catch the compositor red-handed so to speak. [Wilson also compares 35 (bewitcht), 78 (Whether), 895 (of), 1231 (needles), all of which he suspects to be compositorial expansions of copy forms.] And secondly it is clear that the compositors have introduced a large number of full stops into a text which originally contained very few, and that they have also peppered the dialogue with superfluous commas. Furthermore their pointing is careless, as is shown by the numerous instances of transposition in terminal stops. Nevertheless, apart from commas and periods the punctuation of the Quarto is comparatively good on the whole, at times even beautiful. Wilson’s belief that printer’s copy must have contained few commas and periods is skewed by his thinking that the lightly punctuated Hand-D pages in the MS The Booke of Sir Thomas Moore (BL Harl. 7368) are certainly Sh.’s and that they are typical of his punctuation; see Wilson’s What Follows if Some of the Good Quarto Editions of Shakespeare’s Plays Were Printed from His Autograph Manuscripts, Transactions of the Bibliographical Society 15, 1917–19, p. 136. Evans (ed. 1974, p. 247), following Wilson, cites the excessively heavy use of phrasal commas in Q1. Differing from Wilson, Ridley (ed. 1934, p. viii): It is true that the Quarto is very heavily punctuated, very much more so, for example, than Hamlet. But it is not on the face of it likely that a compositor, who after all is a busy working man, is going to pepper his pages with commas, or any other mark of punctuation, merely for the fun of the thing. To him the insertion of marks of punctuation is merely so much more labour, and prima facie therefore there seems no reason why we should not pay as much attention to the compositor’s commas as to any of his other marks of punctuation. . . . [T]he punctuation of the Quarto very frequently produces interesting results. In a certain number of cases it makes a real difference to the sense; in more cases it makes a real difference in the emphasis which is thrown upon phrases by their becoming more isolated; and, perhaps most important of all, it greatly diversifies the rhythms. However, according to Joseph Moxon in Mechanick Exercises on the Whole Art of Printing (1683–4) (ed. Herbert Davis & Harry Carter, 2nd ed., 1962, p. 192), it is the compositor’s duty to discern and amend the bad . . . Pointing of his Copy.

Turner (1962, pp. 33–54): There is nothing very striking about the typography; on cursory examination the book seems to be a run-of-the-mine Elizabethan dramatic quarto. Turner’s primary focus is instances of erroneous line-division of dialogue, some few of which he attributes to the compositor. (P. 48): Mislineation [of verse at 490–1, where the last word of the first line is printed as the first word of the second line] may have resulted from the compositor’s carelessness; but . . . just possibly . . . the MS rather than the workman was at fault. . . . Almost certainly the compositor was juggling the text when he set a short speech of Bottom’s and one of Peter Quince’s in a single line of type at the foot of B2 [319–20, in the inner forme (i.e., B1v, 2, 3v, 4) after, Turner thinks (see here), the outer forme (i.e., B1, 2v, 3, 4v) was already set and the] limits of B1v, B2 . . . had been established. At 413–14, the compositor apparently thought he could squeeze the first complete line of verse into the same line of type with the half-line of verse which begins the speech, a calculation which, as the turn-over shows, was none too accurate. The compositor made analogous interventions at (p. 49) 552–3, 1071–2 and (p. 54) 2063–4, but his responsibility [is] doubtful for the line division of 61–2. He set prose as verse at (p. 49) 911–12 and (p. 54) 1986–90. (Werstine [2012, pp. 144–5, n. 24] suspects the compositor was perhaps responsible for further mislineation [see here].) Turner also tabulates dozens of wrong-font errors (such as roman for italic and small capitals for full capitals) both (pp. 40–5) apparently deliberate—because of shortages of type—and (p. 40, n. 8) accidental.

Brooks (ed. 1979, pp. xxxiii–iv): In the First Quarto an editor has a text of high authority to follow. . . . [I]t is a gratifyingly clean text. . . . In the present edition, it has been found necessary to depart from Q1 in just over fifty verbal readings, besides nine punctuations affecting the sense, one transposition of a pair of lines, a number of line-divisions, and a very few places where verse was set as prose or prose as verse. Apart from the authorial lapses, and the cruces, [xxxiv] the faults are unsurprising errors of the press, almost all of the kinds that compositors are prone to.

For the supposition that Q1 lacks one or more songs, see n. 2175–2206.

Press Variants and Proof-Correction in Q1

Johnson (ed. 1888, p. 39) records the press variants listed below on sig. F1v. Wright (ed. 1891, 2:295) lists only the first of the two on sig. F1v, as does Furness (ed. 1895, p. 166).

Berger (ed. 1995, pp. vi–viii):

Collation of the eight copies reveals five [that is, six] press variants in four of the sixteen formes. . . . The inner forme of sheet A exists in three states.

Copies Collated

  • BL (British Library, C.34.k.29 . . . )
  • Bodl (Bodleian Library; C3 damaged . . . )
  • TCC (Trinity College Cambridge . . . )
  • CSmH (Henry E. Huntington Library . . . )
  • CtYEC (Yale Elizabethan Club . . . )
  • DFo (Folger Shakespeare Library . . . )
  • MB (Boston Public Library . . . )
  • MH (Harvard University; lacks C2, C3, H2, H3 . . . )

Press Variants

Sheet A (inner forme)

  • Corrected:BL, MH
  • Uncorrected:Bodl, TCC, CSmH, CtYEC, DFo, MB

Sig. A2r[4]Now] Now (turned initial N)

  • Corrected:BL, Bodl, TCC, CSmH, CtYEC, MB, MH
  • Uncorrected:DFo

Sig. A2r [18]to funerals:] ro funerals:

Sheet E (outer forme)

  • Corrected:BL, Bodl, TCC, CSmH, CtYEC, DFo, MH
  • Uncorrected:MB

Sig. E3r[1254]he] be

Sheet E (inner forme)

  • Corrected:BL, Bodl, TCC, CSmH, CtYEC, MB, MH
  • Uncorrected:DFo

Sig. E1v[1159]ſwore] fwore

Sheet F (inner forme)

  • First stage [sic] corrected:BL, Bodl, CSmH, CtYEC, DFo, MH
  • Uncorrected:TCC, MB
  • Sig. F1v[1435]notwiſtanding] notwiſtandiug
  • Sig. F1v[1438]them vp & down:] them vp & dowe:
Berger lists no second stage of correction of the inner forme of sheet F; were there such a stage, he presumably would not have stated only of the inner forme of sheet A that it exists in three states.

Ibid. (p. vi, n. 4): Often it is difficult to determine when a press variant exists, and the distinction between deliberate stop-press variants and accidental shifting and bad inking can be vexing to decide. Thus, the h in both at [125] (Sig. A3v) appears to have slipped slightly in the British Library copy, producing bot h, and the space between I and know at [1667] (Sig. F4v) is so loose that the Bodleian, Folger, Huntington, Harvard and Yale Elizabethan Club copies read Iknow. Similarly, O long at [1479] (Sig. F2r) has slipped significantly, producing Ol ong in the British, Huntington, Yale Elizabethan Club, Folger, and Harvard copies. The hyphen in loue-shaft at [536] (Sig. C1r), clear in some copies, is so weakly inked as to appear almost invisible in the British Library, Trinity College Cambridge, and Yale Elizabethan Club copies. At [1159, see above in list of press variants] (Sig. E1v), I agree with W. W. Greg that the Folger copy is variant and reads fwore . . . ; but Richard Kennedy, textual editor of the New Variorum Midsummer Night’s Dream, disagrees. Kennedy and I agree (contra Stanley Wells and Gary Taylor, William Shakespeare: A Textual Companion (Oxford, 1987), p. 287) that what appears to be a comma after melody at [201] (Sig. A4v) in the Bodleian copy is an overinked full stop.

Ibid. (p. vi, n. 7): Much of the bottom half of C3 [in the Bodleian copy] has been torn off. It has been repaired with another piece of paper, and the missing quarto text added in a post seventeenth-century hand. The copyist placed a comma after an extant deere at [695], which the Oxford editors mistook for a press variant (Textual Companion, pp. 281, 287).

Compositor Identification and Order of Printing in Q1

Turner (1962, p. 33): Neither variations in spelling nor typographical abnormalities indicate that [Q1] was set up by more than one compositor.

Turner also (pp. 34–5) provides a partial analysis of the headlines or running-titles, which (p. 34) read on both recto and verso A Midsommer nightes dreame, except on H3v where we find A Midsommer nights dreame. Below his analysis is completed in square brackets, with the roman numerals that Turner assigns to each distinctively identifiable headline and with the pages on which each occurs identified by their signatures:

I [B2, C3v, D4, E3v, F3v,] G4, H4
II [B4, C2, D2, E2, F2,] G2, H2, A2v
III [B1v, C1v, D1v, E1v, F1v,] G1v, H1v, A3
IV B3, C3, D4v, E4v, F4v, G4v, H3, A4v
V [B2v,] C1, D2v, [E1, F1,] G3, H4v
VI [B3v, C4, D3v, E4, F4,] G3v
VII B4v, C4v, D3, E2v, F3, G2v, H1, A3v
VIII [B1, C2v, D1, E3, F2v,] G1, H2v, A4
IX H3v
From this analysis Turner can establish incontrovertibly that (p. 34) the book was worked in two skeleton-formes, one regularly imposing the inner [1v, 2, 3v, 4] and the other the outer forme [1, 2v, 3, 4v] and that sheet A . . . was the last sheet to go through the press. His demonstration of the latter proposition consists of observation of significant changes [that] were made in two [running-]titles during the course of printing: (1) The g appearing in the title used on B3 and C3 (IV in the [chart above]) was replaced at D4v, and the new type appears on E4v, F4v, G4v, H3, and A4v. (2) The title used on B4v and C4v (VII) is characterized by a broken r and a defective e in dreame. At D3 a break in the M also appears, and the three defects are found together on E2v and F3 (where the e’s of dreame were exchanged in position). At G2v the r seems to have been replaced, and the e prints somewhat better than usual. When the title appears on H1, only the break in the M and the new r are evident, and only these two characteristics can be observed in the title as it appears on A3v. It is clear that sheet A was printed after sheet H.

Furthermore, according to Turner, it can be shown that the outer forme of sheet H was the first of its formes to go to press and that it is possible to generalize from this practice with sheet H and to infer that the (p. 35) outer formes of all sheets but A regularly preceded inner formes through the press. The grounds provided for these two conclusions make them far from certain. The alleged priority of presswork on the outer forme of sheet H depends initially for Turner on the pattern of reappearance of headlines from sheet H in sheet A. Both of the headlines (namely VII and VIII) used for the only two pages of the inner forme of sheet A that require headlines (namely A3v and A4, because A1v is blank and A2 bears the head title) come from the outer forme of sheet H. Then a third headline from H outer (IV)—together with two headlines from H inner (II, III)—supply the three pages of A outer needing headlines (excluding the title page, A1). Knowing that H(o) was sent to press before H(i) [and thus was returned to the compositor before H(i) so that its headlines were available for use in A(i), the first forme of sheet A to be set into type] and that earlier in the book all the outer formes were imposed in the same skeleton used for H(o), we can infer that outer formes of all sheets but A regularly preceded inner formes through the press.

This demonstration of Turner’s depends not just on the evidence he presents but also on questionable assumptions that he makes—one explicit, the others not. He implicitly assumes that Bradock printed MND Q1 by itself, rather than concurrently with other books, or (Blayney, 1982, p. 92) that MND was the only work available for composition. However, McKenzie (1969, p. 18), studying the records of Cambridge University Press from the late 17th c., discovers that concurrent printing is frequent, and Blayney (1982, pp. 45, 264 n.) finds evidence of the practice among books printed in London in the decade immediately following Bradock’s work on MND Q1. Turner also implicitly assumes that (Blayney 1982, p. 92) one of Richard Bradock’s two presses was not in use. The explicit assumption underlying Turner’s analysis is that, in terms of the production of the single book Q1 on a single press, (Turner 1962, p. 46) composition and presswork could stay more-or-less in balance. . . . [The compositor planned] to compose two formes, distribute the first, set the third, distribute the second, and so on. . . . Thus the speed of the press, which barring accidents would have remained fairly constant, is established as the rate at which about four type pages could be composed. Examination of the CUP records by McKenzie (1969, pp. 8–10) shows a wide variation in the speed of both composition and presswork not only by different workmen but also by the same workmen at different times and therefore calls into question the likelihood of compositors or pressmen maintaining the balance supposed by Turner. Only by applying these assumptions can it be assumed that the single press was still occupied printing the second forme of Q1’s sheet H as the compositor was setting and then imposing the first forme of Q1’s sheet A and further assumed that the investigator’s task at this juncture is thus simply to determine which forme of sheet H supplied headlines to the first sheet-A forme to be set. With these assumptions in place, Turner concludes H(o) was the first forme of that sheet to be wrought off the press because headlines from it appear in A(i), whereas headlines from both formes of sheet H appear in A(o). If any of Turner’s assumptions fail, then his demonstration becomes inconclusive. There is no evidence that can be adduced for any of the assumptions, which therefore have the status only of hypotheses, two of which are falsified by the evidence against them provided by McKenzie and Blayney.

Proceeding on these assumptions Turner plots the recurrence of distinctively damaged individual pieces of types in Q1, using this evidence to argue that the book was generally set into type by formes and not seriatim (that is, in the order in which pages are to be read). According to Turner, a forme that contains types only from one of the formes of a preceding sheet must have been set into type before a forme that contains type from both formes of that preceding sheet. This judgment is constructed by analogy to the one about headlines already discussed. It yields the following results concerning the order in which the pages of Q1 were set into type:

  • Sheet B (p. 41): not necessarily set by formes: It is a safe guess that all of B(o) was set before work began on B(i), but we cannot absolutely rule out such an order as B1–B1v-B2v-B3–B4v-B2[–B3v, B4];
  • Sheet C (p. 41): initial seriatim setting gave way to setting by formes: C1 (B[o] was almost certainly distributed by the time C2v was set and possibly before much of C1 was set [on the dubious evidence of the k at C1, 8] [But see also (p. 35): h B3,19–C1,1.]), 1v, 2, 2v, 3, 4v (distribution of B[i]), 3v, 4;
  • Sheet D (p. 41): D1 (distribution of C[o]), 2v, 3, 4v, 1v, 2 (distribution of C[i]), 3v, 4;
  • Sheet E (p. 42): E1, 2v, 3, 4v (D[o] distributed at line 12 or 13 of E4v), (D[i] distributed) E1v, 2, 3v, 4;
  • Sheet F (p. 43): F1, 2v, 3 (E[o] distributed), 4v, 1v, 2, 3v (Perhaps partway through setting F3v the compositor distributed E1v, but the rest of the standing type [in E(i)] seems to have been distributed after F(i) was imposed.), 4;
  • Sheet G (p. 44): G1, 2v, 3 (F1 and 3 distributed), 4v (F2v and 4v distributed), Iv, 2 (F1v and 2 distributed), 3v, 4 (F3v and 4 distributed);
  • Sheet H (p. 45): H1, 2v, 3, 4v (G[o] distributed), 1v, 2, 3v, 4
  • Sheet A (p. 45): A1v (blank) (G[i] distributed), 2, 3v, 4, 1 (title page; H[o] distributed at [i.e., before the setting of] A1 or A2v), 2v, 3, 4v.

Blayney (1982, pp. 92–3), explaining and questioning Turner’s analysis, focuses on the recurrence of individually distinctive types from sheet B in sheet C (question marks indicate types [described by Turner] as doubtful):

C1r 1v 2r 2v 3r 3v 4r 4v
From B(o) 1+? 4 2 2 ?
From B(i) 4 2
It is stated that when type reappears in this manner, composition cannot have been seriatim. The statement (which also applies to the similar evidence in sheet D) is completely untrue. The evidence is perfectly consistent with seriatim setting, with B(o) distributed before or during C1r and B(i) distributed after C3r. And in fact Turner then suggested that most of the sheet was set seriatim, except that C4v preceded C3v. The difficulty is that the evidence shows almost nothing. No matter what order is suggested, the failure to detect B(i) evidence in certain pages has to be trusted to indicate that no such evidence exists, whereas the absence of B(o) evidence has to be ascribed to a failure to detect what is really present. But if types from B(o) can be present but undetected in three (or four) pages, so can types from B(i). Because according to Turner’s assumptions (presented above) it is improbable that type from B(o) could have appeared in C1r, the evidence from that page was rejected. By rejecting different parts, and by filling in the gaps in other ways, the supposed evidence could be made to agree with almost any order of setting.

I do not suggest that Turner’s conclusions are wrong, since they may be right. The point is simply that the setting-order of sheets B-D of A Midsummer Night’s [93] Dream remains unestablished and that not enough evidence has been presented. Typographical evidence which is equally consistent with seriatim setting, setting by formes, and other possible methods; which can be supplemented by guesswork in selected pages from which it is absent; and which can be ignored selectively to suit the needs of an unsupported theory of work-flow, cannot be considered adequate. Nonetheless, Hinman (1965, p. 31) reported that a very general investigation of setting by formes . . . indicate[s] . . . pretty surely . . . MND was set in this manner, but he presented no evidence for his judgment.

In addition to the recurrence of distinctively damaged types, Turner also had recourse to type shortages indicated by the substitution of roman font for italic and small capitals for full capitals. Yet he lacked confidence in type shortages as a guide to establishing setting-order of pages (p. 40): By itself the testimony of shortages is, I believe, less reliable than that of any other bibliographical technique, and explained his reasons at length. He also anticipated Blayney’s criticism: However, the reliability of type shortage evidence can be increased when we evaluate it in the light of type reappearances, but even here we can be forced away from the most desirable position by occasionally having to take into account the evidence of only one or two reappearing types and sometimes having to argue from the non-appearance of type. Both are bad policies because mistakes in individual type identifications are easy to make and reappearances are easy to miss.

Turner used (1962, p. 35 n.) photostats of the Huntington Library copy of MND Q1 to identify distinctively damaged types. Weiss (1988, pp. 239–42) demonstrates the short-comings of this use of such a photostat. Checking Turner’s type identifications against the Folger copy of Q1, Weiss (p. 240) can confirm only 46% of the identifications. Replicating Turner’s analysis (with its questionable assumptions discussed above) using the reliable fraction of his type identifications together with new ones discovered in the examination of the Folger copy, Weiss revises Turner’s account of when during the composition of later formes earlier formes were distributed (pp. 241–2): No contradictions occurred with respect to the distributions of sigs. B and C. The appearance of ligature ft6 from D1:6 at E3.11 is one type-page before the suggested distribution of D(o) after E3. Similarly, k5 from E(o) (E4v:23) appears at F3:2, one page early. More significant differences occur in later sheets[:] the appearance of W4 from the last page of F(i) (F4:33) in the first page of G(o) (G1:8), a full gathering prior to the suggested distribution after the imposition of [242] G(i). . . . With respect to the claims that both formes of sheet G and H(o) were delayed four pages each . . . , r1 from G(o) (G2v:19) appears at H1:9, four pages before the suggested G(o) distribution point at H1v, and the appearance of w4 from G2:19 at H1:25, N2 from G3v:29 at H1:28, d16 from G2:12 at H1:31, ligature sh4 from G2:29 at H1v:1, and d12 from G2:29 at H2:24 indicate that G(i) was distributed before the composition of sig. H rather than after the imposition of H(i). Finally, the implication that sig. A was set in type from G(i) and H(o) without a distribution of H(i) seems incorrect. Appearing in A(o) (the second forme of sig. A to be set) are the following types from H(i): y4 from H2:32 at A2v:20, h7 from H1v:18 at A2v:8, and W4 from H2:32 at A4v:32 (and possibly h8 from H2:20, which may appear in A(i) at A2:19). In short, the evidence suggests a more or less normal sequence of distributions following the completion of each of the later sheets.

Valuable though Weiss’s study is for the quality of its type-identification evidence, it still does not free itself from the assumptions identified in Turner’s work by Blayney and McKenzie. While evidence of type shortages is consistent with the setting-order of the pages suggested to Turner by the type-recurrence, shortages cannot be used to establish setting-order. My inspection of the leaves of the Folger copy of Q1 under raking light fails to reveal any indentations in them such as might have been made by the type metal when the sheets were perfected, and thus fails to establish the order in which formes of sheets were printed. Holland (ed. 1994, p. 113): there is insufficient evidence as yet to establish the setting order.

Revision in Q1

Beginning in the middle of the 19th c., there arises a claim that Sh. revised the text of Q1 one or more times; the narrative of such revision remains somewhat consistent as it is elaborated by successive proponents, with the dialogue associated with the four lovers imagined to survive from Sh.’s earliest version of the play. Verplanck (ed. 1847, 2: Introductory Remarks [to MND], 6 [new pagination for each play and its accessories]): It seems . . . very probable . . . that [MND] was originally written in a very different form from that in which we now have it, several years before the date of the drama in its present shape—that it was subsequently remoulded after a long interval, with the addition of the heroic personages, and all the dialogue between Oberon and Titania, perhaps some alteration of the lower comedy; the rhyming dialogue and the whole perplexity of the Athenian lovers being retained, with slight change, from the more boyish comedy. White (ed. 1857, 4:16–17): It seems that [MND] was produced, in part at least, at an earlier period of Shakespeare’s life than his twenty-ninth year [i.e., 1593]. Although as a whole it . . . abounds in passages worthy even of Shakespeare in his full maturity, it also contains whole Scenes which are hardly worthy of his ’prentice hand . . . [17] and which yet seem to bear the unmistakeable marks of his unmistakeable pen. These scenes are the various interviews between Demetrius and Lysander, Hermia and Helen, in Acts II and III. . . . There seems, therefore, warrant for the opinion that this Dream was one of the very first conceptions of the young poet; . . . perhaps . . . he . . . went from Stratford up to London with it partly written; . . . when there, he found it necessary at first to forego completion of it for labor that would find readier acceptance at the theatre; and . . . afterward, when he had more freedom of choice, he reverted to his early production, and in 1594 worked it up into the form in which it was produced. . . . At least some of the additions might have been made . . . for a performance at Court. . . . Except in the play itself I have no support for this opinion, but I am willing to be alone in it.

Fleay (1878, p. 61): MND probably was recast previously to publication. Idem (1886, pp. 181–6) dates the version for the public stage to (p. 183) the winter of 1592, (p. 181) its present form to 1595. January 26, . . . the date of the marriage of William Stanley, Earl of Derby, and subsequent modifications to produce the court version to (p. 182) the winter of 1594–5. (P. 182): the traces of the play having been altered . . . are numerous. There is a double ending. Robin’s final speech [2207–22] is palpably a stage epilogue, while what precedes from Enter Puck [2153] to [2206] is very appropriate for a marriage entertainment, but scarcely suited to the stage. In Acts iv. and v., again, we find in the speech-prefixes Duke, Duchess, Clown, for Theseus, Hippolita, Bottom: such variations are nearly always marks of alteration, the unnamed characters being anterior in date. In the prose scenes speeches are several times assigned to wrong speakers, another common mark of alteration. (P. 183): wherever Robin occurs in the stage-directions or speech-prefixes scarcely any, if any, alteration has been made; Puck, on the contrary, indicates change. (P. 185): The time-analysis . . . has probably been disturbed by omissions in producing the Court version. [138–265] ought to form, and probably did, in the original play, a separate scene; it certainly does not take place in the palace. To the same cause must be attributed the confusion as to the moon’s age; cf. [222–3] with the opening lines [5–14]: the new moon was an afterthought, and evidently derived from a form of the story in which the first day of the month and the new moon were coincident after the Greek time-reckoning. Idem (1891, 2:194): The play has certainly alternative endings: one a song by Oberon for a marriage, and then Exeunt, with no mark of Puck’s remaining on the stage; the other an Epilogue by Puck, apparently for the Court (cf. gentles in [2213]). It might seem, as the Epilogue is placed last, that the marriage version was the earlier, and so I took it to be when I wrote my Life of Shakespeare [1886, quoted above]; but the compliment to Elizabeth [524–45] was certainly written for the Court; and this passage is essential to the original conduct of the play, which may have been printed from the marriage-version copy, with additions from the Court copy. This would require a date for the marriage subsequent to the Court performance. One version must date 1596, for the weather description [463–92], which can be omitted without in any way affecting the progress of the play, requires that date. I believe this passage was inserted for the Court performance in 1596, that on the public stage having taken place in 1595; but that the marriage presentation, being subsequent to this, was most likely at the union of Southampton and Elizabeth Vernon in 1598–9. Chambers (ed. 1897, pp. 14–15): There are certain indications which make me think that [MND] was also at some period [after its composition in 1594–5] slightly retouched. Two passages, [1204–1384+1] and [1793–1902], show a markedly larger proportion of feminine endings than the rest of the play. In the earlier passages, this may be due merely to the excited state of the speakers, but I cannot resist the suspicion that the opening of act v. shows some traces of later work. See also Luce (1906, p. 157). Noble (1923, p. 58 n. 1): The Quarto did not use italics for songs. My own belief is that the whole of the fairy part in the final episode is a comparatively late addition. Witness the fact that Oberon can sing and lead a chorus in Act V, a faculty of which he evinces no sign in the rest of the play.

Wilson (ed. 1924, pp. 80–153): The Q1 text emerges from three distinct episodes of Shn. composition in 1592–3, 1594–5, and 1598. What remains in Q1 of the 1592 version are the (p. 91) lovers’ scenes—those featuring Helena, Hermia, Demetrius, and Lysander—wherein the psychology is generally as crude as the verse (p. 92) is stiff and antithetical. Wilson dates these scenes to 1592–3 because they include 1849–50, which he, like Fleay (1886, p. 183) reads as an allusion to [Robert] Greene’s death [on 3 September 1592] (p. 94). To Sh.’s 1594–5 revision belong Q1’s mechanicals and Bottom scenes, most of the passages in which Robin is used in stage directions and speech prefixes to designate the character otherwise called Puck (including the epilogue—2207–22), parts of the fairy scenes (Wilson, ed. 1924, pp. 95–6), and the introduction of Hippolyta into 1.1 and 4.1. This revision includes the mechanicals’ concern over frightening the ladies with too realistic a representation of a lion (838–56), which Wilson, again like Fleay (1886, p. 185), takes to be an allusion to a spectacle from the celebration in the Scottish court of Prince Henry’s baptism (30 August 1594) for which it was prudently decided to substitute a blackamoor (Wilson ed. 1924, p. 95) for the lion that was initially to have drawn a triumphal car (ibid.). The revision also includes a description (463–92) of what Wilson takes to be the wet and chilly summer of 1594 (ibid.). Sh.’s final handling of MND in 1598 gives Q1 its mature Shakespearian verse, in which the masterly diction and vigorous sweep . . . introduce a note of intellectual energy that makes the whole glow with poetic genius (p. 183). Some such verse Wilson imagines to have been added in short passages written in the margin of the 1594–5 version, with other longer passages interpolated on additional leaves. For Wilson, as for Fleay, the use of Puck for Robin in stage directions and speech prefixes is peculiar to the 1598 revision and is the clue to its purpose, namely the introduction of the little western flower. All but two occurrences of Puck (1028 and 2153–4 being the exceptions) are associated with references to the flower, which functions as a compliment and representation of Elizabeth Vernon, daughter of Sir John Vernon of Hodnet Hall, Shropshire (p. 100) on the occasion of her marriage to the Earl of Southampton in 1598—the occasion for which Sh. added the wedding masque at 2153–206. Wilson (1962, p. 206) substitutes the wedding of Thomas Berkeley to Elizabeth Carey at Blackfriars on 19th February 1596 for the Southampton-Vernon wedding as the occasion for the second revision.

According to Wilson, the following derive from the 1592–3 version: most of 1.1 (2–265) including Helena’s entry at 24–5; 566–625 in 2.1; 686–717 and 737–811 in 2.2; 1063–1124 and 1146–1221 in 3.2; 1792–1881 (although with additions from 1598) in 5.1. To the 1594–5 belong 3–23, 131–5, and the splicing together of two 1592–3 scenes at 136–7 in 1.1; 1.2 (266–371); the beginning of 2.1 (373–523); all but one line of 3.1 (813–1020); 1021–62 (although this passage was later revised in part) and 1440–1506 in 3.2; 4.1 (1509–1745) though certain parts look like first draft material recopied (p. 131)—Wilson specifies 1624–48, 1711, and 1722–3 as 1594 additions—and three more additions were made in 1598; 4.2 (1746–89); 1882–1985 (but 1890–1904 were added in 1598), 1986–2152 (with three minor additions from 1598), and 2207–22 in 5.1. The 1598 revision consists of 153–9 in 1.1; 524–65 and 626–49 in 2.1; 650–85 and 718–36 in 2.2; 901 in 3.1; 1125–45 and 1222–1439 (which was partially recopied and revised in 1598 [p. 125]) in 3.2; 1586–90, 1604–10, 1690–2 in 4.1; 1797–1800, 1804–9, 1822–5, 1828–31, 1855–7, 1863–7, 1874–5, 1879–80, 1890–1902, 2015–16, 2138–40, 2143–4, and 2153–206 in 5.1. Craig (1931, p. 335): Revision of some sort is unmistakable in the case of A Midsummer Night’s Dream. (See also Craig 1961, p. 108.)

Wilson’s revision hypothesis is not left to collapse under the burden of its accumulated speculation. Reviewing his ed., Chambers (1925, pp. 342–4) finds the case for revision to contain inconsistencies: Professor Wilson . . . thinks that, while I.1 and IV.1 belong substantially to the [earliest version], certain awkwardnesses in the introduction of Hippolyta suggest that she was an afterthought, connected in some obscure way with the indication in [4–5] of a four-day period for the action, which is not consistent with the time-analysis. I do not suppose that he would lay much stress on this, especially as he accepts Hipployta as part of his [earliest-version] substratum of V.1. Wilson, Chambers notes (p. 344), makes much of the variation in the naming of Robin-Puck, but nothing of that of Bottom-Clown. Yet Chambers does agree that 5.1 was revised and compliments Wilson on (p. 343) a valuable bibliographical contribution in his attention to the persistent mislining by the printer of passages in [1798–1800, 1804–9, 1822–5, 1828–31, 1855–7, 1863–7, 1874–5, 1879–80], which suggests that these passages were taken in from cramped marginal alterations in the copy. However, unlike Wilson, Chambers does not think the alleged marginal alterations can be dated years later than the context into which they have been supposedly interpolated. I agree again that the fairy-mask of V.1 [2153–206] and the epilogue of [2207–22] are probably duplicate endings. Still Chambers refuses to acknowledge that the adaptation issuing in this duplication need date from as late as 1598. (See also Chambers 1930, 1:360–1.)

Greg (1942, pp. 124–5), though, altogether rejects Wilson’s theory of revision, and subsequent editors join Greg both in this rejection and in his substitution of so-called foul papers (see here for Greg’s definition of this term) as the explanation for any discrepancies in Q1. According to Greg, whether or no the two endings were written at the same time, it would not be surprising to find both in the foul papers in their present order. Nonetheless, Greg does preserve, with modifications, Wilson’s conception of revision at the beginning of 5.1 (p. 125): On the whole I think the copy for Q must have been the author’s manuscript. . . . The most important piece of evidence is at the beginning of the last act where eight passages of verse are mislined [1798–1800, 1804–9, 1822–5, 1828–31, 1855–7, 1863–7, 1874–5, 1879–80]. Wilson has pointed out that if these are removed what remains is perfectly regular and consecutive, and he argues that they represent marginal additions in the copy. . . . It seems to me quite possible that Shakespeare, coming back to his work in a fresher mood, found what he had written rather flat and sought to brighten it up. And I cannot believe there was anything like the amount of revision Wilson imagines [elsewhere in Q1]: it would certainly have left other traces of the sort, whereas the text is elsewhere particularly clean. Greg later (1955, p. 243) is even more dismissive of Wilson’s theory, calling it all very ingenious; the difficulty is to believe that this refashioning would not have left plainer bibliographical traces than are now apparent. Nonetheless Greg adheres to Wilson’s theory of alleged marginal alteration. Spencer (1930, pp. 24–5) does not agree that, if the eight mislined passages are removed what remains is perfectly regular and consecutive, as Greg put it. Instead, Spencer notes that what remains does not cohere as well as the whole existing text does with the eight passages in place. He argues that it is hardly conceivable that a reviser, expanding certain speeches, should make the new joints less conspicuous than the old, since the new represent elaboration, while the old represent the flow of his thought as originally conceived. For example (p. 25), the description of the interlude in the schedule read by Theseus [1852–3] specifies four qualities: the piece is tedious, brief, tragical, and merry, all these qualities lying within the original text as segregated by Professor Wilson. But in the next speech, which contains Philostrate’s explanation, only tedious and brief lie within Mr Wilson’s [original text: 1858–61]. For tragical and merry we must subjoin . . . one of the additions [1863–7]. Kirschbaum (1946, p. 48) supports with another example Spencer’s supposition that what Wilson calls the original text and what he calls the additions were written at the same time: it will be noticed that in [1796] the sequence is Louers first, mad men second. Omitting the so-called addition [1797–1800], we see that the sequence in [1801–3] is mad man first, louer second. Why the shift in sequence if Shakespeare wrote [1796–1803] originally without . . . [1797–1800]? But when we look at [1799–1800, a supposed addition] . . . we see that the sequence lunatics first, louer second, Poet third is the sequence followed in [the allegedly original 1801–3] . . . and [the allegedly later addition 1804–9]. . . . In other words, the sequence followed in the supposedly original version in [1801–3] is not the sequence first indicated in the supposedly original version at [1796] but the sequence indicated in the supposed marginal addition at [1799–1800]. Thus, the ensuing hypothesis is that both [alleged original and alleged later addition] were written at one and the same time and that the [allegedly additional 1799–1800] . . . was written before and not after the [the allegedly original 1801–3]. And since the sequence indicated in [1799–1800] is followed in [1801–9], it may be suggested that [1804–9], the lines on the poet [a supposed addition], were not an afterthought but were written immediately after [1801–3]. See also Lull (1998) on these alleged Shn. revisions.

Turner (1962, pp. 49–50) seeks to corroborate Wilson’s revision theory in a number of places in Q1, including where the eight passages of verse are mislined. Noting that sigs. C1–3, which, according to Turner, are unusual in being set seriatim, contain, on Wilson’s theory, passages from all three stages of composition (497–681), Turner thinks it possible that the workman was confronted here with particularly nightmarish copy. Again, on Turner’s analysis, work on F(o) and the first two pages of F(i) went slowly; and this is another part of [50] the text ([1371–1506], ending near the foot of F2v) which Wilson thinks to have been considerably worked over. Problems with Turner’s method (discussed above, here) compromise any possibility of his analysis buttressing Wilson’s theory. Other problems obtrude in Turner’s justification of Wilson’s interpretation of the eight mislined passages in sig. G (5.1) as evidence of revision. Werstine (2012, p. 145) both summarizes and criticizes this justification: Close bibliographical analysis of the quarto by Robert K. Turner, Jr., shows that quire G is peculiar not only for the frequency with which its verse is mislined, but also for containing four pages with fewer lines of type than is normal. Usually each page has 35 lines, but sigs. G1r, 1v, and 2v have only thirty-four lines each, and G2r only thirty-two ([Turner] 1962, 39). In all, then, quire G is short six lines of type. Six is also the number of lines of type that are saved as an apparently accidental consequence of the mislining of five of the eight passages of verse that are erroneously divided; three of the mislined passages occupy the same space as they would if they were properly set. No one, including Turner, has remarked on this coincidence or attempted to account for it. Turner attempts to explain away the four short pages as follows: in casting off copy for quire G, the compositor, who, he assumes, cast off his own copy, evidently counted in some material that he later did not set[;] . . . he may have failed to notice that some lines here and there were supposed to be cancelled ([ibid.], 54). Such an explanation fails to convince because throughout Q Turner can find only two other places where there may have been errors in casting off so that the compositor had to juggle the lineation of the text in order to fit copy to a predetermined space (ibid., 55)—B2r, [319–20] and H2r, [2063–4]. His explanation then forces us to believe that Quire G is the unique site not only of a considerable amount of mislined verse and short pages (which, according to Turner, bear no relation whatsoever to the mislined verse) but also of a considerable number of misleadingly cancelled lines in its copy (for which there can be, in the nature of the case, no surviving evidence). Without any adequate explanation for the short pages of quire G, Wilson’s theory about the source of the mislined verse in the quire must remain shrouded in doubt. It is instead possible that whoever cast off copy for quire G found no difficulty in printer’s copy and counted off the lines with the same meticulous accuracy found in almost all the rest of his work, and then the compositor, who sometimes unaccountably, if only occasionally, mislined the text elsewhere (e.g., . . . H1r-H1v [1986–9] . . . ), made the mistakes in dividing verse that Wilson attributes to printer’s copy. Such an explanation may not be the right one, but at least it relates the bibliographical anomaly of the short pages to the textual anomaly of the frequently mislined verse.

After Greg’s dismissal of Wilson’s revision hypothesis, Smidt (1986, pp. 123, 128, 130–4, 140, 210) is reluctant to develop in any detail a theory of revision on the basis of so-called unconformities in Q1: One might suppose that the little Indian is left as a residue from an early attempt to work out an appropriate fairy plot, before the fairies became involved with Duke Theseus and his bride. He has lost his raison d’être as a cause of the fairies’ quarrel, but he is still useful as a means of providing a solution to it. (P. 128): The variation between the names Robin and Puck may well be a sign of different stages of composition, as some scholars have thought, but there may have been merely an expansion of Puck’s character as the writing of the play advanced, not a substitution of a spirit for a gnome. And this may have occurred during a continuous process of composition, so that the fairy Puck is not necessarily a sign of revision. (P. 130): The most interesting phenomenon as far as unconformities are concerned is the fitful appearance of the moon during the night in the wood. . . . [132] It is darkness . . . , only lightened for a while by the stars, that prevails while the lovers are in the wood. . . . There is no reference to present moonlight in the rehearsal scene [813–936]. Yet (p. 133) there is in fact enough indication of a moon in the fairy scenes to make the mention of a new moon on the wedding night [by Hippolyta at 12] well-nigh impossible, in spite of our expectations. . . . [134] If the inconsistency was accidental, was it brought about by the merging of different plot components, Theseus and Hippolyta on the one hand, the fairies on the other hand, the young lovers and the artisans in between? (P. 140): The contradictions in this comedy were not brought about by changes of mind, and, with one possible exception, not by inadvertence. The exception is the moonlit fairy scenes. Only (p. 210) the repetition of Theseus’s order to Demetrius and Egeus to accompany him [123–5 and 132–5] suggests some kind of textual disturbance. For reengagement with Wilson’s theory of revision, see Hunter (1998 and 2002, pp. 3–6).

Printer’s Copy for Q1

Speculation on this question has given rise to four suggestions, all of them testifying to perceptions of the high authority of the Q1 text: (1) Sh.’s own MS; (2) Sh.’s own MS as marked up by his company for production; (3) a transcript of Sh.’s MS made and used in the playhouse; (4) a non-theatrical transcript. The cautious commentators have found language to avoid or at least to qualify precise identification.

Discussion begins with Capell, who seems to opt for the third alternative. He (ed. 1768, 1:3 ff.) identifies fourteen quarto texts, including MND, that ought to be excluded from Heminge and Condell’s characterization of all the Shakespeare quartos as diuerse stolne, and surreptitious copies, maimed, and deformed by the frauds and stealthes of iniurious imposters (p. 10): Let it then be granted, that these quarto’s [sic] are the Poet’s own copies, however they were come by; hastily written at first, and issuing from presses most of them as corrupt and licentious as can any where be produc’d, and not overseen by himself, nor by [11] any of his friends. . . . It may be true, that they were stoln; but stoln from the Author’s copies, by transcribers who found means to get at them: and maim’d they must needs be, in respect of their many alterations after the first performance. . . . [12] The very errors and faults of these quarto’s . . . are, with the editor [Capell], proofs of their genuineness; For from what hand, but that of the Author himself, could come those seemingly-strange repetitions [of passages in LLL and Tro.], . . . those imperfect entries . . . ? Capell’s use of transcribers suggests that he thinks copy to have been scribal transcripts, and his reference to their many alterations after the first performance locates transcription in the playhouse, where copy would be subject to theatrical adaptation.

Similarly Malone (in Boswell, ed. 1821, 1:203) in general recognizes the superiority of the early quartos to counterpart F1 texts, except for Wiv. and H5 (Q1 Ham. was not discovered until 1823), yet does not attempt to specify just what kind of MSS served as printer’s copy: With respect to the other thirteen copies [quartos, including MND] . . . , they in general are preferable to the exhibition of the same plays in the folio; for this plain reason, because, instead of printing these plays from a manuscript, the editors of the folio . . . printed the greater part of them from the very copies which they represented as maimed and imperfect, and frequently from a late, instead of the earliest, edition; in some instances with additions and alterations of their own.

Clark and Wright (ed. 1863, 2:viii), cautiously subjunctive: Fisher’s edition . . . may have been taken from the author’s manuscript. Furness (ed. 1895, p. xii) takes pains to imagine that Heminge and Condell may not have been guilty of a wilful untruth, as alleged by Malone, when they implied that they provided for F1 only his [Sh.’s] papers as printer’s copy, but, in fact, provided quarto copy for such plays as MND if they knew that this [quarto] text was [originally] printed directly from his manuscript. Chambers (ed. 1897, p. 8), leaving open all alternatives for printer’s copy: Q1 has been printed from a clear and authentic manuscript. Cuningham (ed. 1905, p. xiii) adds to the alternatives the possible use of actors’ parts in the creation of printer’s copy, an idea apparently borrowed ultimately from Johnson’s 1756 Proposals for Printing . . . the Dramatick Works of William Shakespeare (ed. Arthur Sherbo, 1968, 7:52): Q1 was printed in all probability, either from the authentic MS. of Shakespeare himself, or at least from an accurate copy or, perhaps, copies of the actors’ parts, transcribed in the theatre from the original MS.

Furness (ed. 1895, p. xi) misleads a number of his successors with the observation that in both Q1 and Q2 the stage directions are, as in copies used on the stage, in the imperative, such as wind horns, [1622] sleep [1484]. So Cuningham (ed. 1905, xv) notes stage-directions . . . in the imperative, as is customary in stage copies as his ground for suggesting the possible playhouse provenance of printer’s copy. And Pollard (1909, p. 72) declares that The imperative form of the stage directions, Ly doune ([1110]) and Winde hornes ([1622]) may be taken as indicating its origin from a playhouse copy. (See Idem 1920, p. 64.) Greg (1942, p. 37) properly questions the assumption that the prompter can be identified by the use of the imperative: The prompter writes directions for his own use; they are generally terse and to the point. Chambers [1930, 1:118] questions whether they are usually in the imperative. They are not: but being short and curt they tend to imperative and participial constructions. Wilson (1945, p. 67) provides examples of the imperative in Anthony Munday’s hand, not the hands of the theatrical annotator(s), in the theatrical MS of Iohn A kent & Iohn a Cumber; Werstine (2012, p. 229) gives examples of imperative SDD in authorial hands from MSS of Thomas Heywood’s The Captives and the anonymous The Waspe. Greg (1942, p. 125), contradicting himself, follows Pollard in identifying the hand of the prompter in the appearance of imperative SDD: There are however [in Q1] a few directions that suggest the prompter, such as Lie down and Wind horn. In the palpable duplication, Enter Lovers; Lysander, Demetrius, Hermia, and Helena ([1819–20]), we may see the book-keeper expanding a typically brief direction of the author’s. (Werstine [2012, p. 131] notes that actual theatrical texts rarely show a bookkeeper specifying a group in terms of their proper names, finding only one example, from the annotated quarto of Thomas Lodge and Robert Greene’s A Looking Glasse, for London and England.) Greg differs from Pollard, though, in judging printer’s copy to be, not a playhouse copy, but so-called foul papers (see here for Greg’s definition of this term) containing (1942, p. 125) notes made in preparation for the prompt-book. Greg’s belief in the possibility of such a document once having existed arises from his interpretation of a single extant MS, Thomas Heywood’s transcription of The Captives, as such a document, but as Werstine (2012, pp. 300–9) observes, this interpretation is at odds with features of the MS. Greg rules out (1942, 125) a playhouse transcript to be printer’s copy only because he believes that there could only ever have been one such transcript and that it was used to annotate printer’s copy for F, which differs from Q1 chiefly in terms of such annotations.

Pollard (1920, p. 63), while not setting aside his belief in theatrical annotation of printer’s copy, anticipated Greg in arguing as well for the possibility that such annotated copy could originally have been inscribed by Sh.: Possibly in some cases, if [a dramatist] were familiar with the theatre, he might use the same technical language as a prompter, so that Shakespeare himself, in the scene in the wood in the Midsummer Night’s Dream, may have written the directions, Enter a Faerie at one doore and Robin goodfellow at another, Enter the King of Fairies at one doore, with his traine; and the Queen at another with hers, the doors, of course, being those of the stage, not of the wood. Adams (1923, p. 519) states an opinion that can be interpreted to be identical to Pollard’s but need not necessarily be: MND is printed from authentic playhouse copy; compare Neilson & Hill, ed. 1942, p. 88. Pollard (1923, p. 7): some of the flaws in these Good Quartos [including MND Q1] are the result of imperfections in Shakespeare’s own work, and I have ventured to claim that some of these Good Quartos may actually have been set up from Shakespeare’s autograph manuscripts. Wilson (ed. 1924, p. 80) is of Pollard’s mind: beyond doubt [Q1 is] printed from a theatrical prompt-book, . . . [with] the managerial voice giving real directions to the players; he quotes 1110, 1484, and 1622. Printer’s copy also contains irregularities strongly suggestive of an author’s manuscript, and so it is both Shakespeare’s autograph manuscript and the prompt-book just as Shakespeare left it. Kittredge (ed. 1936, p. 229) and Parrott (ed. 1938, p. 131) follow Wilson. Chambers (1930, 1:358), persuaded to be more specific than in 1897: Q1 may be from the author’s manuscript. Kirschbaum (1955, pp. 172–3) stands out against the idea of theatrical annotation of a Sh. MS: Greg’s evidence for prompter’s additions does not pass muster. The directions could just as well come from the author, and there is no reason why they should not be copied by a scribe. (P. 173): The nomenclature [variation in naming of characters in SPP and SDD] . . . does not show the author in the heat of composition—does not cause the reader to assume foul papers behind the print. . . . There is little evidence of foul papers or playhouse in Q. Nothing in it rules out printing from transcript. . . . There is neither internal nor external evidence to show that the copy came to the publisher from Shakespeare’s fellows.

In 1955 Greg establishes what becomes a virtual consensus among 20th-c. editors that printer’s copy for Q1 is authorial foul papers defined as (1955, pp. 106, 142) a copy representing the play more or less as the author intended it to stand, but not itself clear or tidy enough to serve as a prompt-book, that is, a theatrical MS used to guide performance, because it contained (p. 142) loose ends and false starts and unresolved confusions. Greg presumed that book-keepers necessarily tidied away from prompt-books certain features of foul papers, including seven features still to be found in Q1 (ibid., pp. 240–2):

  • (1)multiple designations of the same character (p. 241): Oberon is King of Fairies (or simply King) or Oberon indifferently; Titania is of course named in the text [i.e., dialogue and SPP], but in directions [the proper name] appears only at [650] on her second entry [otherwise she is named Queene in SDD]; Bottom, on his most important appearance [1509] is merely Clowne; Theseus and Hippolyta, after long appearing by name, become as a rule Duke and Duchess after the play begins in Act V; lastly Robin (Goodfellow) and the generic Puck alternate;
  • (2)indefinite entrances involving speakers (p. 240): after carefully naming the rude mechanicals in I.ii, the author later contents himself with the description the Clownes (III.i) or the rabble (IV.ii); the young couples, having been named in I.i, become simply Louers at V.i . . . , for here the names [i.e., Lysander, Demetrius, Hermia, and Helena] are a palpable addition; the fairies are quite indefinite, except for the foure at [980]. Greg also quotes as allegedly bearing the characteristics of the author and thereby implicitly needing the bookkeeper’s attention SDD containing the term traine to refer to speakers attending royalty (p. 240: her [i.e., Titania’s] traine (650); all his [i.e., Theseus’s] traine (1622);
  • (3)other indefinite SDD: Greg’s characteristically authorial SDD include (p. 240) Enter Theseus, Hippolita, with others (2), where others refers to supers, rather than speakers, including Philostrate, addressed by name in this scene;
  • (4)inclusion of an unnecessary character in a SD (p. 241): Pucke is made to enter with Oberon at [1021], though in fact he only does so three lines later;
  • (5)missing entrances and exits;
  • (6)descriptive SDD: characteristic of the author, according to Greg, are, for example (p. 240), Enter Egeus and his daughter Hermia . . . and Enter Quince, the Carpenter; and Snugge, the Ioyner . . . ;
  • (7)marginal insertions of dialogue (pp. 241–2): Greg fully endorses the theory of J. Dover Wilson’s that some passages of verse at the beginning of the play’s last scene are wrongly divided in Q1 because they were marginal additions to the dialogue, their position in the margins forbidding their writer from dividing them properly as verse and the compositor following his copy in this error (see here).

Greg also mentions the following incidental features of Quarto MND as characteristic of foul papers: erroneous SPP—for Greg, one of Shakespeare’s (ibid., p. 247 Note A) oversights in composition, the author having written consecutive speeches for Flute and Thisbe, forgetting that they were the same; double entrance (p. 240): Helena enters in 1.1 both at 25 and at 191; a dialogue error in naming (ibid., p. 246 Note A): Flute for Snout at 1957; and an erroneous SD (ibid., 240): Enter Quince, Flute, Thisby and the rabble (1746)—in this scene Shakespeare had forgotten that Flute and Thisbe are one.

Werstine (2012, p. 132): The appearance of these features in the earliest printing of A Midsummer Night’s Dream leads Greg to conclude that we can hardly imagine that Q represents a finished prompt-book [1955, p. 241] and therefore, by this process of elimination of what he regards . . . as the only possible alternative, Quarto Dream must represent Shakespeare’s foul papers—in spite of the persistence of the same features in the actual theatrical texts that he calls promptbooks, which destroys his argument. Werstine reviews each of Greg’s features and compares them to what is found in the twenty-one extant texts (both MS and annotated quartos) that bear theatrical annotation and thereby show what book-keepers actually did and did not do to their playbooks (these texts are described in Werstine 2012, pp. 234–57). In these twenty-one Werstine finds (1) multiple designations of the same character (pp. 359–64), (2) indefinite entrances involving speakers, including the very terms cited by Greg to discount a theatrical MS—crewe, trayne, the rest (pp. 375–9), (3) other indefinite SDD, including uses of others identical to those cited from MND by Greg (pp. 379–82), (4) inclusion of an unnecessary character in a SD (p. 384), (5) missing entrances (pp. 374–5) and exits (pp. 386–8), (6) descriptive SDD such as, from the scribal MS Ironside: Enter Edmond and Alfricke the generall vnder the kinge:/ (1.3.332) and Enter Edricke a poore man . . . (2.2.461). (Needlessly explanatory SDD, like those that Greg quotes from the MND quarto, are also frequent in The Second Maidens Tragedy [also called The Lady’s Tragedy], another scribal MS, in further indication that such SDD are by no means peculiar to authorial MSS: Enter the new Vsurping Tirant; The Nobles of his faction, Memphonius, Sophonirus, Heluetius with others, The right heire Gouianus depos’de [1.1.1–3]; Enter L Anselmus the deposde kinges brother, wth | his Frend Votarius [1.2.257–8]; Enter the ladye of Gouianus . . . [2.1.636]; Enter Tirant wondrous discontedly: Nobles afarr of [4.2.1655–6]; Enter Votarius with Anselmus the Husband [5.1.1984]); (7) marginal insertions of dialogue (pp. 388–9). Even the minor features cited by Greg to identify foul papers as printer’s copy are located in theatrical MSS by Werstine: erroneous SPP (pp. 371–2), double entrances (pp. 385–6), a dialogue error in naming (p. 371), and erroneous SDD (p. 385).

Greg himself showed his awareness of the flaw in his reasoning when he wrote (1955, p. 142) It must, however, be recognized that owing to the casual ways of book-keepers these characteristics may persist, to some extent at least, in the prompt-book. In reproducing his argument and conclusions about copy for MND Q1, editors fail to attend to this caveat: Doran (ed. 1959, p. 174); Wells (ed. 1967, p. 165); Evans (ed. 1974, p. 247); Brooks (ed. 1979, pp. xxii–v); Foakes (ed. 1984, pp. 135–6); Wells & Taylor (1987, p. 279); Holland (ed. 1994, pp. 113–14); Berger (ed. 1995, p. ix). Horwood (ed. 1939, p. 10) accepts what would become Greg’s argument and conclusion in the form in which it was first presented in McKerrow (1935)—Greg’s inspiration—with reference only to variation in naming in SPP and SDD. Greg’s definition of foul papers has also been overturned in Werstine (2009, pp. 44–5).

A significant problem with Greg’s use of variation in naming of characters in SPP in MND Q1 and other texts as evidence of so-called foul papers had already been identified by Kennedy (1998, pp. 178–9): There may, however, be another explanation for the variation in SP’s in some of Shakespeare’s early texts. The change may not be authorial at all, but compositorial. It seems to have been a printing-house convention that a compositor did not have to follow copy in the matter of SP’s, but could choose to call characters by their first names or last names, or [179] generic names or personal names, or by their functions or peculiarities. If he needed to, he could vary the SP from Quin. to Pet[er], from The[seus] to Duke, . . . and so on. Most of the time, variant SP’s do not point to authorial foul papers, but signify compositorial change. And most of the time, variant SP’s are not signs of an author’s revising, or of an author in the heat of composition, but are rather indications of a compositor switching SP’s because of type shortage. Kennedy applies this theory in detail to the variation in naming in MND Q1’s SPP (179–90).

To support the view that printer’s copy is in Sh.’s own hand, some recent editors have followed Wilson (ed. 1924, pp. 112, 116, 121, 148) in adducing certain spellings in MND Q1 as the same as or somewhat analogous to spellings in the Hand-D pages of MS The Booke of Sir Thomas Moore (BL Harl. 7368), in the belief that Hand D is Sh.: Evans (ed. 1974, p. 247), Brooks (ed. 1979, pp. xxv–vi), Foakes (ed. 1984, p. 136), Wells & Taylor (1987, p. 279), Holland (ed. 1994, p. 114); Berger (ed. 1995, p. ix). However, recent review of Hand D’s spelling by Jackson (2007) shows that only a half-dozen are sufficiently rare to constitute acceptable evidence for attribution, and none of these is in MND Q1.

There is no evidence for a rational choice among the following alternatives for MND Q1 printer’s copy: (1) Sh.’s own MS; (2) Sh.’s own MS as marked up by his company for production; (3) a transcript of Sh.’s MS made and used in the playhouse; (4) a non-theatrical transcript. While there is nothing in the way of playhouse notes in the quarto to demonstrate alternatives 2 and 3, Werstine (2012, p. 4) shows that in some actual playhouse MSS there is so little annotation that it is possible that MSS with no annotation could have been used in production.

The Second Quarto (1619)

There is no entry for Q2 in the Stationers’ Register, as is the case with a great many books published in the late 16th and early 17th c. (see Blayney 1997, pp. 400–5). Erne (2003, pp. 255–8) suggests that Heminge and Condell were referring specifically to the whole series of plays among which MND Q2 appeared in 1619 when the two actors wrote of the stolne, and surreptitious copies in their prefatory remarks to the First Folio (1623)—if these remarks were of their composition. However, Heminge and Condell fail to supply any justification for Erne’s specification, and they provide MND Q2 as copy for the Folio printing.

Q2’s title page, as transcribed by Greg (BEPD, 1:169), reads as follows: [ornament] | A | Midsommer nights | dreame. | As it hath beene sundry times pub- |likely acted, by the Right Honoura- |ble, the Lord Chamberlaine his |seruants. | VVritten by VVilliam Shakespeare. | [device <McKerrow>283] | Printed by Iames Roberts, 1600.

Bartlett & Pollard (1939, pp. 71–3) identify and locate thirty extant copies of Q2. Modern facsimiles include that of William Griggs, ed. J. W. Ebsworth, 1880. For a digital facsimile see the Shakespeare Quartos Archive (www.quartos.org).

The Printing of Q2

Except for its slightly different top lace border, its altogether different mid-page printer’s ornament, and its substitution of reference to Roberts for that to Fisher, the wording and even the alternation of roman and italic fonts of Q2’s title page duplicate Q1’s. Thus it falsifies its date of printing and its printer, and thereby created uncertainty (see above here) and gave rise to dispute (see below). Three centuries passed before the discovery that Q2 was in fact printed in 1619 by William Jaggard for Thomas Pavier. Pollard (1909, p. 93): the mid-page printer’s ornament William Jaggard was using . . . in other books both before and after 1619. The resemblance between Q1 and Q2 title pages continues on sig. A2, the first page of the play’s text. In both quartos that page is surmounted by A | MIDSOMMER NIGHTS | DREAME. followed by a centered SD and then by a centered SP.

Because the states of formes in Q1 differ only by typographical errors that are easily noticed and corrected, it is impossible to identify the states of the formes in the copy of Q1 used as printer’s copy for Q2. Pollard (1923, p. 5): Taking each intermediate edition [like this 1619 Pavier quarto of MND] by itself, in no single instance do we find evidence of the sort of care which could lead us to believe that its overseer had obtained access to any authoritative source. . . . As evidence of the words which Shakespeare wrote or of the words which were spoken by the actors engaged in his plays these intermediate editions are absolutely worthless, except where we possess only one or two copies of the First Edition. . . . A Second Edition might . . . be printed from a copy of the First in which a correction had been made which does not appear in any copy of the First now extant. Pollard’s only one or two is optimistic; even when there are eight extant copies of Q1, we may not expect to find among them all the different states of correction of their formes and therefore all the states that may have been present in the copy of Q1 used to print Q2.

The Pavier quarto of 1619 is, for the most part, a page-for-page reprint of Q1. Clark & Wright (ed. 1863, 2:viii): On comparing these two Quartos we find that they correspond page for page, though not line for line, except in the first five pages of sheet G. In Q1 the first four pages of this sheet contain fewer typographical lines than the thirty-five found on the rest of the pages. In Q2 these four pages each contain 35 typographical lines. This regularization is effected sometimes by the transfer of lines from later pages to earlier ones, sometimes by the chopping up of verse into shorter lines, sometimes by the correct division of verse that is mislined in Q1, and sometimes by the addition of white space around SDD. Like Q1, Q2 is divided into neither acts nor scenes. Q2 has a different tailpiece on H4v from Q1’s; Greg (BEPD 1:169): The ornament on H4v [of Q2] is a copy of device 179.

Dyce (ed. 1857, 1:clxiii): Roberts’s [Q2] is the less accurate quarto. Clark & Wright (ed. 1863, 2:viii): The printer’s errors in Fisher’s [Q1] edition are corrected in that issued by Roberts [Q2], and . . . in the Roberts Quarto the Exits are more frequently marked. Ebsworth (ed. Q2 1880, pp. ix–xiii): In Fisher’s [Q1], the business [i.e., the SDD] is given (as usual) in Italic type, with exception of the proper names of the characters; which are in Roman type. But in Roberts’s [Q2], the whole line is in Italic type, names and all. (P. x): Roberts’s page [is] wider than Fisher’s to the extent of about two letters’ breadth [The measure in Q1 is 82 mm; that in Q2 87 mm.]. And it is remarkable that when . . . difference [in line-for-line reproduction] ensued . . . a recurrence has been speedily made to the former agreement. (P. xiii): The spelling of Q2 is more modern than Q1’s: We give a brief sample of these differences in corresponding places; but they are innumerable throughout: — Roberts’s Quarto: tell — Snug — else — home-spuns — perhaps — hue — eke — Iew — Snowt — do — hog — Finch — Sparrow — answer — lye — he, etc. . . . Fisher’s Quarto: tel — Snugge — els — homespunnes — perhappes — hewe — eeke — Iewe — Snowte — doe — hogge — Fynch — Sparrowe — answere — ly — hee etc. [845–952]. Ebsworth also notes some contractions such as trēble, for tremble [852]; lātern, for lantern [871]; chābre, for chamber [873]; vnderstād, for vnderstand [903]; trāslated, for translated [935–6] in Q1 that are expanded in Q2. It is by no means difficult to understand the improved clearness in typography of Roberts over that of Fisher (supposing, as we do, that Roberts had Fisher’s printed book before his eyes). For there was the additional space gained—1. By the excision of redundant letters; 2. By having a wider platform of type in his page; 3. By his gaining an occasional line in prose passages, and thus being able to afford extra leads at entrance of characters. Despite this improvement in typographical clearness, there is a marked deterioration in the minute divisions of the verse by punctuation. Commas are less frequent, either from negligence or from systematic repugnance to the scholarly and grammatical breaking-up of sentences. Furness (ed. 1895, p. xi–xv), comparing Q2 to Q1: The Second Quarto . . . has the fairer page, with type fresh and clear. (P. xv): In . . . Q1 there are about fifty-six stage-directions; in . . . Q2 about seventy-four. Chambers (ed. 1897, p. 168): Q2 is printed from Q1. . . . it is set up with greater attention to typographical details. . . . And where the typographical correspondence of the two editions gets out, the spacing of Q2 is always arranged so as to recover it as soon as possible. The printer is evidently working from a model. Cuningham (ed. 1905, p. xi): Q 2 corrects some of the mistakes in Q 1; but, on the other hand, it commits more than it corrects. Rhodes (1923, p. 64): the additions in Roberts’ are of small importance, being commonly the mark of Exit when it is quite clear from what the actor said that he was leaving the stage.

Brooks (ed. 1979, pp. xxvii–viii) cites further bibliographical evidence of [Q2’s] derivation [from Q1]: reproduction of such peculiarities as the spelling wodde for wood [261; here Brooks appears to be in error; unless there is an unrecorded press variant, Q2 reads wood] and the omission of Enter before Robin and Demetrius [1465]; its printing of The. In himselfe he is [62], Enter Snout [929], Enter Lysander [1440], Enter Thisbie [1989], and Lyon. Oh [2064] just as Q1 has them, tucked in on the same line as the end of the preceding speech; [Q2’s] agreement with Q1 in capitalizations which are not simply those to be expected; and above all its concurrence in thirty-five of the speech-prefixes where Q1 varies the form of the abbreviation. In II.i, for example, where Q1 has Ob. three times, then Oberon, Ob. four times more, then Oberon again; and, also exceptionally, for Demetrius’ sixth speech, Demet. Q2 follows suit. Full collation shows that it repeats all but thirteen of the verbal errors made in Q1. It corrects prose set as verse at [911–12], but not at [1986–9]. A [xxviii] half-hearted attempt is made to rectify some of the misdivided verse in V.i between [1797] and [1880], but most of it, like the misdivided [490–1] in II.I, is reprinted as it stands.

. . . Of [Q2’s] thirteen corrections, four eliminate obvious literal misprints. In the remainder the errors announce themselves: the misreading of waves for wanes [7], and of Cet. for Bot. [867]; the displacement of t from comfor to bet [691], the omission of an o from good [695], and of to before expound [1734] where the sense requires it; a mistake of number in gentleman [1333]; a failure to repeat is after this [1649]; an assimilation of is to knit [699]; and a catching of yet from earlier in the phrase [2104]. They needed nothing beyond the context in Q1 itself either to draw attention to them or to indicate the proper correction. Apart from the accidentals of spelling and the like, Q2 differs from Q1 only through the guesswork which furnished these corrections, and by over sixty new errors of the printing-house. Since it derives from the author only through Q1, its readings have no independent authority.

See the textual notes of this edition for all significant differences between Q1 and Q2, including SPP, SDD, lineation, punctuation, and verbal variants.

Date and Auspices of Q2

Editors struggled to determine the priority of the two editions, both dated 1600 on their title-pages. (See here.) Although most judged correctly that Fisher’s edition was the earlier, disputes arose. Halliwell-Phillipps (ed. 1856, 5:11): Perhaps Fisher’s edition, which on the whole, seems to be more correct than the other, was printed from a corrected copy of that published by Roberts. Fleay (1891, 2:178–9): The consensus of critical opinion is that Roberts pirated his copy from the earlier Fisher edition; but it would be a unique phenomenon had this been allowed to pass without inhibition or, at least, protest. All the evidence lies the other way. Better readings are usually found in later editions, whenever these are produced in the lifetime of the author. Printer’s errors are far more likely to have been introduced than corrected in a second edition. . . . It seems to me far more likely that Roberts printed the play for Fisher, who did not, for some reason unknown to us, care to put his name on the first issue; but finding the edition quickly exhausted, and the play popular, he then appended his name as publisher.

Only in the early 20th c. was Q2 MND correctly dated. Knowles (2020, pp. 1116–18): it was identified as part of a group of plays printed in the same year (1619) though bearing title pages dated from 1600 to 1619. That curious and rather shabby collection (Greg, 1955, p. 12) of plays known or sometimes thought to be by Shakespeare—Parts 1 and 2 of The Whole Contention betweene . . . Lancaster and York (2H6 and 3H6), Per., A Yorkshire Tragedy, MV, Wiv., MND, Lr., H5, and 1 Sir John Oldcastle—had been reprinted in 1619 from earlier quartos or octavos originally issued by a variety of printers and publishers; this new collection, now known as the Pavier Quartos, was printed by William Jaggard (whose shop would soon print F1) for the publisher Thomas Pavier, who was apparently planning to bring out a collection of Shakespeare’s plays. Several bound collections of these ten plays have in fact survived. The brilliant literary sleuthing that revealed the truth behind their several falsified title pages has been recounted by Richard Altick in The Scholar Adventurers (1950, pp. 189–95).

Greg (1908, pp. 113–31, 381–409) first proposed that the ten quartos bound together in a 17th-c. binding were not remainders—three were dated 1600, two 1608, three 1619—but were all actually printed in 1619 despite the differing dates on their title-pages. Eight had a printer’s device and numerals not used until 1610 and a type font not used until 1617, and all were printed on the same papers, bearing the same group of watermarks, which would not have been available over a span of nineteen years. (On rare Pavier watermarks bearing dates of either 1617 or 1619, see Stevenson [1951–2]). In each quarto the printer imitated an original edition. Greg inferred that initially three of these editions were printed in and dated 1619, but that when Pavier for some reason got nervous about his undertaking he issued others under their original dates, possibly seeming to sell off the remainders of editions printed years before by other publishers in order to avoid challenges to copyright. Pollard (1909, pp. 81–104) reports that he and Greg became suspicious of the quartos because they did not specify the printer, publisher, and publisher’s address, but rather the initials T. P. (for Thomas Pavier) on five of the title-pages; because around 1619 William Jaggard was using two of the printer’s devices appearing repeatedly in this group of quartos; because a font of Roman type used in the suspect quartos was a new kind also used in F1 in 1623; and because the spelling in each suspect quarto was generally more modern than in its (older) counterpart, evidently reflecting the habits of Jaggard’s compositors. The clinching proof was provided by Neidig (1910, pp. 145 ff.), who showed by photographic overlays that seven of the nine title-pages were printed in part from the same setting of type, parts of which were transferred from one title-page to another; these therefore were (p. 154) not printed nineteen years apart, but within a few days of each other. The order of printing that he established for these title-pages—WC, YT, Per., MV, Wiv., Lr., H5, and SJO—has been generally accepted as the order of printing of the plays themselves, with the exception that Per. follows WC, with which it shares continuous signatures. On the evidence of watermarks Blayney (1972, pp. 196–7) provisionally places MND before Lr., and Knowles (1982, p. 195) has found the supporting evidence that a number of distinctive types in the last three sheets of MND appear in the first three sheets of Lr. Wiv. is printed in a larger and different font than that used in Lr.

Chambers (1930, 1:134–7): The Contention and Pericles have continuous signatures and were clearly designed for issue together. . . . William Jaggard succeeded to the printing business of James Roberts about 1608, and by 1617 had associated in it his son Isaac Jaggard. . . . The reprinting of 1619 was no doubt done in concert with Pavier, who owned the copyright of five of the plays. . . . Presumably licence was obtained from Johnson for the use of Merry Wives of Windsor, and from Butter for that of King Lear. Of the other three, Midsummer-Night’s Dream was probably derelict, and Merchant of Venice may have been believed to be so. Blount’s registration of Pericles had already been overlooked, and there is nothing to show that Gosson had any copyright. The shortened imprints suggest that the title-pages were originally meant for half-titles in a comprehensive volume, which would naturally begin with a general and more explicit title-page. . . . It was nothing to Pavier and Jaggard that they were reprinting bad texts and ascribing to Shakespeare plays that were not his. Perhaps Shakespeare’s fellows viewed [136] such proceedings with less equanimity. On May 1619 a letter was addressed by the Lord Chamberlain to the Stationers’ Company directing that none of the King’s men’s plays should be printed without some of their consents. Its exact terms are not preserved. But they appear to be recited in a letter of similar import written on 10 June 1637 by Philip Earl of Pembroke, then Lord Chamberlain, and brother of William Earl of Pembroke, who was Lord Chamberlain in 1619, asking the Stationers to stay publication of any King’s men plays without their consents. Pavier and Jaggard [137] may have issued all the ten plays. It is perhaps more likely that they had already abandoned the continuous signatures and perhaps the idea of a comprehensive volume, had separately issued those dated 1619, and had the rest ready in print. . . . Whatever the events of 1619, they can have left no enduring malice between the King’s men and the Jaggards, since it was again from their press that the collection . . . the First Folio came. From the facts of Pavier’s life—that he was at the time publishing religious books rather than plays and was just entering the governing councils of the Stationers’ Company—Johnson (1992, pp. 35–40) concludes that Jaggard, not Pavier, was the instigator of a straightforward scheme to put into print a collection of as many plays as were available, not a complete collection, and that Pavier collaborated by lending the copyrights on the plays he owned (WC, YT, SJO, H5, apparently Per.), by negotiating permissions for Lr. and Wiv., and by assuming or appropriating rights to the more-or-less derelict MV and MND. The faked imprints, Johnson suspects, were intended not to deceive the copyright holders but to avoid protest by the acting company or their agents, who ultimately may have concluded that plays being offered as old goods offered no competition to their planned new and improved Folio. Whether Pavier and Jaggard had conceived of their enterprise as a straightforward business venture, exactly why and when and how the players and other publishers may have objected to the project, what effect Pembroke’s letter may have had, and how William Jaggard and his son Isaac were persuaded to transfer their attentions to the larger project of the First Folio of 1623 have been much speculated upon and discussed; see Pollard (1909, pp. 100–4), Greg (1924, pp. 139–44), Greg (1955, pp. 9–17), . . . Greg (BEPD, 1957, 3:1107–8). Blayney (privately): Since Jaggard’s name did not appear on any of the 1619 quartos, there’s no reason to suppose that any of the players ever guessed that the culprit was the printer they knew best (because of his playbill monopoly). So that even if it had been the players who chose Jaggard to print the Folio (as it almost certainly wasn’t), we can’t assume that they’d forgiven him for a known transgression. . . . We shouldn’t credit them with knowing all that we know.

Kirschbaum (1955, 240–1) speculates about Jaggard’s particular circumstances and strategy in printing MND Q2: The play was derelict copy [Fisher, publisher of Q1, having disappeared from the Stationers’ Company without transferring his right to publish the book to any other member]; in order to publish such copy, it was necessary to obtain the Stationers’ Company’s permission. For the stationers’ guild to grant Jaggard the right to print this play in 1619 might be construed by King’s men as an act in direct defiance of their interests. [Greg (1955, p. 24) quotes the Stationers’ Court-Book C: vppon a letter from the right honorable the Lord Chamberleyne It is thought fit & so ordered That no playes that his Maiestyes players do play shalbe printed without consent of somme of them.] (P. 241): Jaggard decided not to try to establish copyright in [MND] at Stationers’ Hall but to print the derelict copy with a false publisher and a false date. This was the only alternative to issuing it with a 1619 date, a procedure which might have led to some kind of trouble. Jaggard’s apprehensions may perhaps be gauged by the supposition that his edition of [MND] purports to be not an edition different from another published in 1600 . . . but a second issue of the edition bearing Fisher’s name. It was not uncommon for a single edition to be sold by two or more publishers, each publisher having his name only on the title page of the issue he sold. How fortunate, therefore, for Jaggard to find in his shop a large stock of unsold copies of [MND] published in 1600 by his predecessor, Roberts! According to Kirschbaum (ibid., p. 250), Jaggard made his MND look like a different issue of the authentic Fisher edition or like an edition closely copying and closely succeeding the Fisher edition. However, Kirschbaum’s speculation, like Johnson’s above, casts Jaggard in the role of Q2’s publisher, rather than, as he was, only its printer, who may therefore have had no concerns about rights in the copy or about the King’s Men.

Massai (2007, pp. 112–19), departing from earlier scholars’ emphasis on Pavier and Jaggard’s deception of their fellow stationers and the King’s Men, instead proposes that publication of the Pavier quartos may have been part of a larger marketing scheme devised by these two stationers for their mutual benefit in selling first these quartos and then, if successful, the 1623 Folio. Massai associates the Pavier quartos with several other 17th-c. nonce collections of plays by single dramatists that brought together previously published editions of plays with editions just published for the collection. For example, (p. 116) the 1607 re-issue of Sir William Alexander’s The Monarchicke Tragedies (STC 344) includes two additional plays, . . . both dated 1607, and two plays originally issued in the 1604 edition of The Monarchick Tragedies (STC 343). While one of these two plays has no individual title page, the other retains the original one and the date in the imprint is unchanged. With some title pages dated 1619 and others earlier, the 1619 Pavier quartos may resemble such collections. Jaggard’s (p. 118) advantage in leading Pavier’s prospective readers to believe that they were [119] offered the scattered remains of a recently deceased playwright whose works had not been published since 1615 can probably best be described as a pre-publicity stunt. . . . Pavier’s marketing strategy was aimed at arousing rather than satisfying a specific demand for a product that was still relatively new to the English book market—a collection in folio of plays by a dramatist writing for the commercial playhouses. Isaac Jaggard may eventually have succeeded in finding an investor for his project [the 1623 Folio] because Pavier had paved the way for it in 1619 by significantly reviving the fortunes of Shakespeare in print. . . . What would Pavier gain from it? . . . Selling his quartos both individually and as a nonce collection would minimize Pavier’s financial risk, [and] . . . Pavier would gain additional revenue from lending the right to reprint his Shakespeare plays to other stationers [namely, those in the syndicate bringing out the Folio].

Compositor Identification and Order of Printing in Q2

Kable (1970, pp. 7–18) identifies Compositor B of the 1623 Folio as the single compositor who set all the Pavier quartos. Andrews (1971, p. 320) identifies the compositor of MND Q2 as the fellow workman of Compositor B on the Pavier quartos: Compositor F alone set up the type for four plays [including] . . . A Midsummer Night’s Dream. Andrews employs the following criteria for discriminating between the two compositors: spelling, capitalization of I will contractions, punctuation of the text, placement and punctuation of marginal stage directions, and consistency in the use of italic type for proper nouns. He attributes the marked changes in punctuation between Q1 and Q2 to Compositor F (pp. 395–6): with Midsummer Night’s Dream . . . Compositor F’s average number of punctuation changes . . . jumps dramatically upward, this time to 17.0 punctuation changes per page (1050 changes in 62 pages). . . . [396] It is difficult to judge the extent to which the dramatic increase in punctuation changes . . . was affected by the nature of the punctuation in the Q1 Midusmmer copy-text. . . . however, it would appear that the increase is largely to be explained as resulting from a new degree of aggressiveness on the part of the compositor. Andrews (1973, Two Compositors, p. 5) subsequently renames his Compositor F Compositor G.

In the course of compositor identification, Andrews (1971, pp. 331–2) provides a headline analysis of MND Q2: the set of running-titles [headlines] that appear on pages B1, B2v, B3, and B4v reappear on pages B2, B1v, B4, and B3v, respectively. They then reappear on the following pages: D1, D2v, D3, D4v and D2, D1v, D4, D3v; E1, E2v, E3, E4v and E2, E1v, E4, E3v; G1, G2v, G3, G4v and G2, G1v, G4, G3v. Another set of running-titles appear in parallel sequence: C1, C2v, C3, C4v and C2, C1v, C4, C3v; F1, F2v, F3, F4v and F2, F1v, F4, F3v; H1, H2v, H3, H4v and H2, H1v, H4, H3v. It is obvious that two different skeleton frames were employed in the printing of Midsummer, one frame being used to print both the outer and inner formes of quires B, D, E, and G and a second frame being used to print both the outer and inner formes of quires A, C, F, and H. (I omitted quire A from the initial part of this discussion because it contains only five pages with running-titles; it should be noted, however, that the running-titles for these pages are identical with, and in the same relationships to each other as, the running-titles for quires C, F, and H.

While Andrews sees no significance in such headline analysis for compositor identification, Blayney (1972, 197–205) does. Andrews (1973, Supplement to Two Compositors, p. 11) summarizes Blayney’s position, while departing from it: Blayney adopts as a working hypothesis the idea that the alternating skeletons reflect alternating compositorial stints. While the idea, according to Andrews, has some application to MV, Blayney’s working hypothesis runs into difficulties when he tries to apply it to the remaining Pavier Quartos, including MND. It is at variance with the most compelling evidence based on spellings, punctuation, and other differentiae. Andrews (1973, Supplement to Unresolved Bibliographical Problems, pp. 1–5) expands these differentiae to include the spacing of medial commas and periods and of periods after SPP.

Andrews and Blayney also differ concerning when in the sequence of Pavier quartos MND Q2 was printed. Andrews (1971, p. 328), noting that MND’s title-page contains no typographical associations with other Pavier quarto title-pages, writes that MND was the very last Pavier quarto to be printed. Blayney (1972, pp. 196–7), relying on watermarks, places MND Q2 just before Lr.

Knowles (1982, pp. 191–206), accepting Blayney’s location of MND Q2, analyzes the recurrence of distinctively damaged types from the last three quires of MND Q2 in Lr. Q2 before focusing on such recurrences within the latter. He discovers that quires at the beginning of Lr. were set by two compositors at two different type cases; (p. 202) he refuses to speculate on the relation of these two Jaggard workmen to the ones who set type for F1. Although Knowles’s primary interest is Lr., he does advance a hypothetical explanation of the typographical relations of Q2 MND and Q2 Lr. Since the last three sheets of MND seem to have been set from two type cases [197], and, as Peter Blayney has shown . . . , with two skeletons used for the most part in the same unusual pattern of alternation as is found in Lr., one may safely suppose that the last three sheets of MND were set by two compositors working more or less concurrently; since the type cases are the same for both plays, one may even think it likely that the same two compositors who set the last sheets of MND continued at their cases to begin setting the early sheets of the next Pavier quarto, Q2 Lr. It looks as if Compositor 1 [at case x], after setting MND G(i), distributes the long-standing type pages from E(i) and begins to set H(o) sometime before his fellow compositor at case y begins to set H(i), the last forme for this play. He finishes while Compositor 2 is distributing types from G(o) in order to set H(i). Apparently Compositor 1 begins work right away on the first forme of the next play, A(o) of Lr., before G(i) has been unlocked. . . . Meanwhile Compositor 2 has finished H(i) and is about to begin Lr. B(o). . . . The precise details of these speculations will have to await further confirmation from the study of types throughout the whole of Q2 MND, but I think that there is already sufficient evidence from types to support Peter Blayney’s assignment of MND just prior to Lr. in the Pavier series.

Brooks (ed. 1979, pp. xxviii–xxix) notes an abnormal prefix [Peter for Quin 820] . . . on D1r. Bottom’s Peter quince [819] supports Dover Wilson’s diagnosis of the cause [Wilson, ed. 1924, p. 154], a shortage of [italic] capital Q’s; undoubtedly it prompted the resort to Peter, Pet., which continues on D2v. These are two pp. of the outer forme; in between them, D1v, D2r, belonging to the inner forme, have Quin. (eight times). Sheet D, then, was set by formes, D2v [xxix] after D1r, and not seriatim [or in reading order]. So, no doubt, was the whole of Q2; an easy method with a page-for-page reprint.

Annotated Q1 Copy for Q2

Cuningham (ed. 1905, p. xv), speculatively: In respect of the superior stage-directions of Q 2, it may not be unlawful to conjecture that Roberts [i.e., Pavier] had taken a copy of Fisher’s Quarto to a theatrical representation, or had otherwise procured a prompter’s copy and improved the stage-directions of his edition accordingly. Massai (2007, 122–9) identifies certain patterns of very occasional editorial attention across the Pavier quartos, including MND Q2.

The Printing of Q2

Except for its slightly different top lace border, its altogether different mid-page printer’s ornament, and its substitution of reference to Roberts for that to Fisher, the wording and even the alternation of roman and italic fonts of Q2’s title page duplicate Q1’s. Thus it falsifies its date of printing and its printer, and thereby created uncertainty (see above here) and gave rise to dispute (see below). Three centuries passed before the discovery that Q2 was in fact printed in 1619 by William Jaggard for Thomas Pavier. Pollard (1909, p. 93): the mid-page printer’s ornament William Jaggard was using . . . in other books both before and after 1619. The resemblance between Q1 and Q2 title pages continues on sig. A2, the first page of the play’s text. In both quartos that page is surmounted by A | MIDSOMMER NIGHTS | DREAME. followed by a centered SD and then by a centered SP.

Because the states of formes in Q1 differ only by typographical errors that are easily noticed and corrected, it is impossible to identify the states of the formes in the copy of Q1 used as printer’s copy for Q2. Pollard (1923, p. 5): Taking each intermediate edition [like this 1619 Pavier quarto of MND] by itself, in no single instance do we find evidence of the sort of care which could lead us to believe that its overseer had obtained access to any authoritative source. . . . As evidence of the words which Shakespeare wrote or of the words which were spoken by the actors engaged in his plays these intermediate editions are absolutely worthless, except where we possess only one or two copies of the First Edition. . . . A Second Edition might . . . be printed from a copy of the First in which a correction had been made which does not appear in any copy of the First now extant. Pollard’s only one or two is optimistic; even when there are eight extant copies of Q1, we may not expect to find among them all the different states of correction of their formes and therefore all the states that may have been present in the copy of Q1 used to print Q2.

The Pavier quarto of 1619 is, for the most part, a page-for-page reprint of Q1. Clark & Wright (ed. 1863, 2:viii): On comparing these two Quartos we find that they correspond page for page, though not line for line, except in the first five pages of sheet G. In Q1 the first four pages of this sheet contain fewer typographical lines than the thirty-five found on the rest of the pages. In Q2 these four pages each contain 35 typographical lines. This regularization is effected sometimes by the transfer of lines from later pages to earlier ones, sometimes by the chopping up of verse into shorter lines, sometimes by the correct division of verse that is mislined in Q1, and sometimes by the addition of white space around SDD. Like Q1, Q2 is divided into neither acts nor scenes. Q2 has a different tailpiece on H4v from Q1’s; Greg (BEPD 1:169): The ornament on H4v [of Q2] is a copy of device 179.

Dyce (ed. 1857, 1:clxiii): Roberts’s [Q2] is the less accurate quarto. Clark & Wright (ed. 1863, 2:viii): The printer’s errors in Fisher’s [Q1] edition are corrected in that issued by Roberts [Q2], and . . . in the Roberts Quarto the Exits are more frequently marked. Ebsworth (ed. Q2 1880, pp. ix–xiii): In Fisher’s [Q1], the business [i.e., the SDD] is given (as usual) in Italic type, with exception of the proper names of the characters; which are in Roman type. But in Roberts’s [Q2], the whole line is in Italic type, names and all. (P. x): Roberts’s page [is] wider than Fisher’s to the extent of about two letters’ breadth [The measure in Q1 is 82 mm; that in Q2 87 mm.]. And it is remarkable that when . . . difference [in line-for-line reproduction] ensued . . . a recurrence has been speedily made to the former agreement. (P. xiii): The spelling of Q2 is more modern than Q1’s: We give a brief sample of these differences in corresponding places; but they are innumerable throughout: — Roberts’s Quarto: tell — Snug — else — home-spuns — perhaps — hue — eke — Iew — Snowt — do — hog — Finch — Sparrow — answer — lye — he, etc. . . . Fisher’s Quarto: tel — Snugge — els — homespunnes — perhappes — hewe — eeke — Iewe — Snowte — doe — hogge — Fynch — Sparrowe — answere — ly — hee etc. [845–952]. Ebsworth also notes some contractions such as trēble, for tremble [852]; lātern, for lantern [871]; chābre, for chamber [873]; vnderstād, for vnderstand [903]; trāslated, for translated [935–6] in Q1 that are expanded in Q2. It is by no means difficult to understand the improved clearness in typography of Roberts over that of Fisher (supposing, as we do, that Roberts had Fisher’s printed book before his eyes). For there was the additional space gained—1. By the excision of redundant letters; 2. By having a wider platform of type in his page; 3. By his gaining an occasional line in prose passages, and thus being able to afford extra leads at entrance of characters. Despite this improvement in typographical clearness, there is a marked deterioration in the minute divisions of the verse by punctuation. Commas are less frequent, either from negligence or from systematic repugnance to the scholarly and grammatical breaking-up of sentences. Furness (ed. 1895, p. xi–xv), comparing Q2 to Q1: The Second Quarto . . . has the fairer page, with type fresh and clear. (P. xv): In . . . Q1 there are about fifty-six stage-directions; in . . . Q2 about seventy-four. Chambers (ed. 1897, p. 168): Q2 is printed from Q1. . . . it is set up with greater attention to typographical details. . . . And where the typographical correspondence of the two editions gets out, the spacing of Q2 is always arranged so as to recover it as soon as possible. The printer is evidently working from a model. Cuningham (ed. 1905, p. xi): Q 2 corrects some of the mistakes in Q 1; but, on the other hand, it commits more than it corrects. Rhodes (1923, p. 64): the additions in Roberts’ are of small importance, being commonly the mark of Exit when it is quite clear from what the actor said that he was leaving the stage.

Brooks (ed. 1979, pp. xxvii–viii) cites further bibliographical evidence of [Q2’s] derivation [from Q1]: reproduction of such peculiarities as the spelling wodde for wood [261; here Brooks appears to be in error; unless there is an unrecorded press variant, Q2 reads wood] and the omission of Enter before Robin and Demetrius [1465]; its printing of The. In himselfe he is [62], Enter Snout [929], Enter Lysander [1440], Enter Thisbie [1989], and Lyon. Oh [2064] just as Q1 has them, tucked in on the same line as the end of the preceding speech; [Q2’s] agreement with Q1 in capitalizations which are not simply those to be expected; and above all its concurrence in thirty-five of the speech-prefixes where Q1 varies the form of the abbreviation. In II.i, for example, where Q1 has Ob. three times, then Oberon, Ob. four times more, then Oberon again; and, also exceptionally, for Demetrius’ sixth speech, Demet. Q2 follows suit. Full collation shows that it repeats all but thirteen of the verbal errors made in Q1. It corrects prose set as verse at [911–12], but not at [1986–9]. A [xxviii] half-hearted attempt is made to rectify some of the misdivided verse in V.i between [1797] and [1880], but most of it, like the misdivided [490–1] in II.I, is reprinted as it stands.

. . . Of [Q2’s] thirteen corrections, four eliminate obvious literal misprints. In the remainder the errors announce themselves: the misreading of waves for wanes [7], and of Cet. for Bot. [867]; the displacement of t from comfor to bet [691], the omission of an o from good [695], and of to before expound [1734] where the sense requires it; a mistake of number in gentleman [1333]; a failure to repeat is after this [1649]; an assimilation of is to knit [699]; and a catching of yet from earlier in the phrase [2104]. They needed nothing beyond the context in Q1 itself either to draw attention to them or to indicate the proper correction. Apart from the accidentals of spelling and the like, Q2 differs from Q1 only through the guesswork which furnished these corrections, and by over sixty new errors of the printing-house. Since it derives from the author only through Q1, its readings have no independent authority.

See the textual notes of this edition for all significant differences between Q1 and Q2, including SPP, SDD, lineation, punctuation, and verbal variants.

Date and Auspices of Q2

Editors struggled to determine the priority of the two editions, both dated 1600 on their title-pages. (See here.) Although most judged correctly that Fisher’s edition was the earlier, disputes arose. Halliwell-Phillipps (ed. 1856, 5:11): Perhaps Fisher’s edition, which on the whole, seems to be more correct than the other, was printed from a corrected copy of that published by Roberts. Fleay (1891, 2:178–9): The consensus of critical opinion is that Roberts pirated his copy from the earlier Fisher edition; but it would be a unique phenomenon had this been allowed to pass without inhibition or, at least, protest. All the evidence lies the other way. Better readings are usually found in later editions, whenever these are produced in the lifetime of the author. Printer’s errors are far more likely to have been introduced than corrected in a second edition. . . . It seems to me far more likely that Roberts printed the play for Fisher, who did not, for some reason unknown to us, care to put his name on the first issue; but finding the edition quickly exhausted, and the play popular, he then appended his name as publisher.

Only in the early 20th c. was Q2 MND correctly dated. Knowles (2020, pp. 1116–18): it was identified as part of a group of plays printed in the same year (1619) though bearing title pages dated from 1600 to 1619. That curious and rather shabby collection (Greg, 1955, p. 12) of plays known or sometimes thought to be by Shakespeare—Parts 1 and 2 of The Whole Contention betweene . . . Lancaster and York (2H6 and 3H6), Per., A Yorkshire Tragedy, MV, Wiv., MND, Lr., H5, and 1 Sir John Oldcastle—had been reprinted in 1619 from earlier quartos or octavos originally issued by a variety of printers and publishers; this new collection, now known as the Pavier Quartos, was printed by William Jaggard (whose shop would soon print F1) for the publisher Thomas Pavier, who was apparently planning to bring out a collection of Shakespeare’s plays. Several bound collections of these ten plays have in fact survived. The brilliant literary sleuthing that revealed the truth behind their several falsified title pages has been recounted by Richard Altick in The Scholar Adventurers (1950, pp. 189–95).

Greg (1908, pp. 113–31, 381–409) first proposed that the ten quartos bound together in a 17th-c. binding were not remainders—three were dated 1600, two 1608, three 1619—but were all actually printed in 1619 despite the differing dates on their title-pages. Eight had a printer’s device and numerals not used until 1610 and a type font not used until 1617, and all were printed on the same papers, bearing the same group of watermarks, which would not have been available over a span of nineteen years. (On rare Pavier watermarks bearing dates of either 1617 or 1619, see Stevenson [1951–2]). In each quarto the printer imitated an original edition. Greg inferred that initially three of these editions were printed in and dated 1619, but that when Pavier for some reason got nervous about his undertaking he issued others under their original dates, possibly seeming to sell off the remainders of editions printed years before by other publishers in order to avoid challenges to copyright. Pollard (1909, pp. 81–104) reports that he and Greg became suspicious of the quartos because they did not specify the printer, publisher, and publisher’s address, but rather the initials T. P. (for Thomas Pavier) on five of the title-pages; because around 1619 William Jaggard was using two of the printer’s devices appearing repeatedly in this group of quartos; because a font of Roman type used in the suspect quartos was a new kind also used in F1 in 1623; and because the spelling in each suspect quarto was generally more modern than in its (older) counterpart, evidently reflecting the habits of Jaggard’s compositors. The clinching proof was provided by Neidig (1910, pp. 145 ff.), who showed by photographic overlays that seven of the nine title-pages were printed in part from the same setting of type, parts of which were transferred from one title-page to another; these therefore were (p. 154) not printed nineteen years apart, but within a few days of each other. The order of printing that he established for these title-pages—WC, YT, Per., MV, Wiv., Lr., H5, and SJO—has been generally accepted as the order of printing of the plays themselves, with the exception that Per. follows WC, with which it shares continuous signatures. On the evidence of watermarks Blayney (1972, pp. 196–7) provisionally places MND before Lr., and Knowles (1982, p. 195) has found the supporting evidence that a number of distinctive types in the last three sheets of MND appear in the first three sheets of Lr. Wiv. is printed in a larger and different font than that used in Lr.

Chambers (1930, 1:134–7): The Contention and Pericles have continuous signatures and were clearly designed for issue together. . . . William Jaggard succeeded to the printing business of James Roberts about 1608, and by 1617 had associated in it his son Isaac Jaggard. . . . The reprinting of 1619 was no doubt done in concert with Pavier, who owned the copyright of five of the plays. . . . Presumably licence was obtained from Johnson for the use of Merry Wives of Windsor, and from Butter for that of King Lear. Of the other three, Midsummer-Night’s Dream was probably derelict, and Merchant of Venice may have been believed to be so. Blount’s registration of Pericles had already been overlooked, and there is nothing to show that Gosson had any copyright. The shortened imprints suggest that the title-pages were originally meant for half-titles in a comprehensive volume, which would naturally begin with a general and more explicit title-page. . . . It was nothing to Pavier and Jaggard that they were reprinting bad texts and ascribing to Shakespeare plays that were not his. Perhaps Shakespeare’s fellows viewed [136] such proceedings with less equanimity. On May 1619 a letter was addressed by the Lord Chamberlain to the Stationers’ Company directing that none of the King’s men’s plays should be printed without some of their consents. Its exact terms are not preserved. But they appear to be recited in a letter of similar import written on 10 June 1637 by Philip Earl of Pembroke, then Lord Chamberlain, and brother of William Earl of Pembroke, who was Lord Chamberlain in 1619, asking the Stationers to stay publication of any King’s men plays without their consents. Pavier and Jaggard [137] may have issued all the ten plays. It is perhaps more likely that they had already abandoned the continuous signatures and perhaps the idea of a comprehensive volume, had separately issued those dated 1619, and had the rest ready in print. . . . Whatever the events of 1619, they can have left no enduring malice between the King’s men and the Jaggards, since it was again from their press that the collection . . . the First Folio came. From the facts of Pavier’s life—that he was at the time publishing religious books rather than plays and was just entering the governing councils of the Stationers’ Company—Johnson (1992, pp. 35–40) concludes that Jaggard, not Pavier, was the instigator of a straightforward scheme to put into print a collection of as many plays as were available, not a complete collection, and that Pavier collaborated by lending the copyrights on the plays he owned (WC, YT, SJO, H5, apparently Per.), by negotiating permissions for Lr. and Wiv., and by assuming or appropriating rights to the more-or-less derelict MV and MND. The faked imprints, Johnson suspects, were intended not to deceive the copyright holders but to avoid protest by the acting company or their agents, who ultimately may have concluded that plays being offered as old goods offered no competition to their planned new and improved Folio. Whether Pavier and Jaggard had conceived of their enterprise as a straightforward business venture, exactly why and when and how the players and other publishers may have objected to the project, what effect Pembroke’s letter may have had, and how William Jaggard and his son Isaac were persuaded to transfer their attentions to the larger project of the First Folio of 1623 have been much speculated upon and discussed; see Pollard (1909, pp. 100–4), Greg (1924, pp. 139–44), Greg (1955, pp. 9–17), . . . Greg (BEPD, 1957, 3:1107–8). Blayney (privately): Since Jaggard’s name did not appear on any of the 1619 quartos, there’s no reason to suppose that any of the players ever guessed that the culprit was the printer they knew best (because of his playbill monopoly). So that even if it had been the players who chose Jaggard to print the Folio (as it almost certainly wasn’t), we can’t assume that they’d forgiven him for a known transgression. . . . We shouldn’t credit them with knowing all that we know.

Kirschbaum (1955, 240–1) speculates about Jaggard’s particular circumstances and strategy in printing MND Q2: The play was derelict copy [Fisher, publisher of Q1, having disappeared from the Stationers’ Company without transferring his right to publish the book to any other member]; in order to publish such copy, it was necessary to obtain the Stationers’ Company’s permission. For the stationers’ guild to grant Jaggard the right to print this play in 1619 might be construed by King’s men as an act in direct defiance of their interests. [Greg (1955, p. 24) quotes the Stationers’ Court-Book C: vppon a letter from the right honorable the Lord Chamberleyne It is thought fit & so ordered That no playes that his Maiestyes players do play shalbe printed without consent of somme of them.] (P. 241): Jaggard decided not to try to establish copyright in [MND] at Stationers’ Hall but to print the derelict copy with a false publisher and a false date. This was the only alternative to issuing it with a 1619 date, a procedure which might have led to some kind of trouble. Jaggard’s apprehensions may perhaps be gauged by the supposition that his edition of [MND] purports to be not an edition different from another published in 1600 . . . but a second issue of the edition bearing Fisher’s name. It was not uncommon for a single edition to be sold by two or more publishers, each publisher having his name only on the title page of the issue he sold. How fortunate, therefore, for Jaggard to find in his shop a large stock of unsold copies of [MND] published in 1600 by his predecessor, Roberts! According to Kirschbaum (ibid., p. 250), Jaggard made his MND look like a different issue of the authentic Fisher edition or like an edition closely copying and closely succeeding the Fisher edition. However, Kirschbaum’s speculation, like Johnson’s above, casts Jaggard in the role of Q2’s publisher, rather than, as he was, only its printer, who may therefore have had no concerns about rights in the copy or about the King’s Men.

Massai (2007, pp. 112–19), departing from earlier scholars’ emphasis on Pavier and Jaggard’s deception of their fellow stationers and the King’s Men, instead proposes that publication of the Pavier quartos may have been part of a larger marketing scheme devised by these two stationers for their mutual benefit in selling first these quartos and then, if successful, the 1623 Folio. Massai associates the Pavier quartos with several other 17th-c. nonce collections of plays by single dramatists that brought together previously published editions of plays with editions just published for the collection. For example, (p. 116) the 1607 re-issue of Sir William Alexander’s The Monarchicke Tragedies (STC 344) includes two additional plays, . . . both dated 1607, and two plays originally issued in the 1604 edition of The Monarchick Tragedies (STC 343). While one of these two plays has no individual title page, the other retains the original one and the date in the imprint is unchanged. With some title pages dated 1619 and others earlier, the 1619 Pavier quartos may resemble such collections. Jaggard’s (p. 118) advantage in leading Pavier’s prospective readers to believe that they were [119] offered the scattered remains of a recently deceased playwright whose works had not been published since 1615 can probably best be described as a pre-publicity stunt. . . . Pavier’s marketing strategy was aimed at arousing rather than satisfying a specific demand for a product that was still relatively new to the English book market—a collection in folio of plays by a dramatist writing for the commercial playhouses. Isaac Jaggard may eventually have succeeded in finding an investor for his project [the 1623 Folio] because Pavier had paved the way for it in 1619 by significantly reviving the fortunes of Shakespeare in print. . . . What would Pavier gain from it? . . . Selling his quartos both individually and as a nonce collection would minimize Pavier’s financial risk, [and] . . . Pavier would gain additional revenue from lending the right to reprint his Shakespeare plays to other stationers [namely, those in the syndicate bringing out the Folio].

Compositor Identification and Order of Printing in Q2

Kable (1970, pp. 7–18) identifies Compositor B of the 1623 Folio as the single compositor who set all the Pavier quartos. Andrews (1971, p. 320) identifies the compositor of MND Q2 as the fellow workman of Compositor B on the Pavier quartos: Compositor F alone set up the type for four plays [including] . . . A Midsummer Night’s Dream. Andrews employs the following criteria for discriminating between the two compositors: spelling, capitalization of I will contractions, punctuation of the text, placement and punctuation of marginal stage directions, and consistency in the use of italic type for proper nouns. He attributes the marked changes in punctuation between Q1 and Q2 to Compositor F (pp. 395–6): with Midsummer Night’s Dream . . . Compositor F’s average number of punctuation changes . . . jumps dramatically upward, this time to 17.0 punctuation changes per page (1050 changes in 62 pages). . . . [396] It is difficult to judge the extent to which the dramatic increase in punctuation changes . . . was affected by the nature of the punctuation in the Q1 Midusmmer copy-text. . . . however, it would appear that the increase is largely to be explained as resulting from a new degree of aggressiveness on the part of the compositor. Andrews (1973, Two Compositors, p. 5) subsequently renames his Compositor F Compositor G.

In the course of compositor identification, Andrews (1971, pp. 331–2) provides a headline analysis of MND Q2: the set of running-titles [headlines] that appear on pages B1, B2v, B3, and B4v reappear on pages B2, B1v, B4, and B3v, respectively. They then reappear on the following pages: D1, D2v, D3, D4v and D2, D1v, D4, D3v; E1, E2v, E3, E4v and E2, E1v, E4, E3v; G1, G2v, G3, G4v and G2, G1v, G4, G3v. Another set of running-titles appear in parallel sequence: C1, C2v, C3, C4v and C2, C1v, C4, C3v; F1, F2v, F3, F4v and F2, F1v, F4, F3v; H1, H2v, H3, H4v and H2, H1v, H4, H3v. It is obvious that two different skeleton frames were employed in the printing of Midsummer, one frame being used to print both the outer and inner formes of quires B, D, E, and G and a second frame being used to print both the outer and inner formes of quires A, C, F, and H. (I omitted quire A from the initial part of this discussion because it contains only five pages with running-titles; it should be noted, however, that the running-titles for these pages are identical with, and in the same relationships to each other as, the running-titles for quires C, F, and H.

While Andrews sees no significance in such headline analysis for compositor identification, Blayney (1972, 197–205) does. Andrews (1973, Supplement to Two Compositors, p. 11) summarizes Blayney’s position, while departing from it: Blayney adopts as a working hypothesis the idea that the alternating skeletons reflect alternating compositorial stints. While the idea, according to Andrews, has some application to MV, Blayney’s working hypothesis runs into difficulties when he tries to apply it to the remaining Pavier Quartos, including MND. It is at variance with the most compelling evidence based on spellings, punctuation, and other differentiae. Andrews (1973, Supplement to Unresolved Bibliographical Problems, pp. 1–5) expands these differentiae to include the spacing of medial commas and periods and of periods after SPP.

Andrews and Blayney also differ concerning when in the sequence of Pavier quartos MND Q2 was printed. Andrews (1971, p. 328), noting that MND’s title-page contains no typographical associations with other Pavier quarto title-pages, writes that MND was the very last Pavier quarto to be printed. Blayney (1972, pp. 196–7), relying on watermarks, places MND Q2 just before Lr.

Knowles (1982, pp. 191–206), accepting Blayney’s location of MND Q2, analyzes the recurrence of distinctively damaged types from the last three quires of MND Q2 in Lr. Q2 before focusing on such recurrences within the latter. He discovers that quires at the beginning of Lr. were set by two compositors at two different type cases; (p. 202) he refuses to speculate on the relation of these two Jaggard workmen to the ones who set type for F1. Although Knowles’s primary interest is Lr., he does advance a hypothetical explanation of the typographical relations of Q2 MND and Q2 Lr. Since the last three sheets of MND seem to have been set from two type cases [197], and, as Peter Blayney has shown . . . , with two skeletons used for the most part in the same unusual pattern of alternation as is found in Lr., one may safely suppose that the last three sheets of MND were set by two compositors working more or less concurrently; since the type cases are the same for both plays, one may even think it likely that the same two compositors who set the last sheets of MND continued at their cases to begin setting the early sheets of the next Pavier quarto, Q2 Lr. It looks as if Compositor 1 [at case x], after setting MND G(i), distributes the long-standing type pages from E(i) and begins to set H(o) sometime before his fellow compositor at case y begins to set H(i), the last forme for this play. He finishes while Compositor 2 is distributing types from G(o) in order to set H(i). Apparently Compositor 1 begins work right away on the first forme of the next play, A(o) of Lr., before G(i) has been unlocked. . . . Meanwhile Compositor 2 has finished H(i) and is about to begin Lr. B(o). . . . The precise details of these speculations will have to await further confirmation from the study of types throughout the whole of Q2 MND, but I think that there is already sufficient evidence from types to support Peter Blayney’s assignment of MND just prior to Lr. in the Pavier series.

Brooks (ed. 1979, pp. xxviii–xxix) notes an abnormal prefix [Peter for Quin 820] . . . on D1r. Bottom’s Peter quince [819] supports Dover Wilson’s diagnosis of the cause [Wilson, ed. 1924, p. 154], a shortage of [italic] capital Q’s; undoubtedly it prompted the resort to Peter, Pet., which continues on D2v. These are two pp. of the outer forme; in between them, D1v, D2r, belonging to the inner forme, have Quin. (eight times). Sheet D, then, was set by formes, D2v [xxix] after D1r, and not seriatim [or in reading order]. So, no doubt, was the whole of Q2; an easy method with a page-for-page reprint.

Annotated Q1 Copy for Q2

Cuningham (ed. 1905, p. xv), speculatively: In respect of the superior stage-directions of Q 2, it may not be unlawful to conjecture that Roberts [i.e., Pavier] had taken a copy of Fisher’s Quarto to a theatrical representation, or had otherwise procured a prompter’s copy and improved the stage-directions of his edition accordingly. Massai (2007, 122–9) identifies certain patterns of very occasional editorial attention across the Pavier quartos, including MND Q2.

The First Folio (1623)

On 8 Nov. 1623 the first collection of Sh.’s plays, now known as the First Folio, was entered in the Stationers’ Register (Book D, p. 69), as here transcribed in Greg (BEPD, 1:33; cf. 3:1109–12): Mr. Blounte Isaak Iaggard. Entred for their Copie vnder the hands of Mr. Dor. Worrall and Mr. Cole warden Mr. William Shakspeers Comedyes Histories, & Tragedyes soe manie of the said Copies as are not formerly entred to other men. vizt. [Here follows a list of half the plays]. Edward Blount and Isaac Jaggard were co-publishers; Isaac Jaggard had just recently inherited the printing business of his late father William. The Dor. Worrall who granted official license to publish was Thomas Worrall, chaplain to the Bishop of London, and George Cole was then Upper Warden for the Stationers. MND is not explicitly included in the entry because it had already been entred to other men, namely, Thomas Fisher, in 1600. In F1 MND is found on sigs. N1v-O3v, pp. 145–62 of the first section, the comedies. Lee’s (1902) census of extant copies has been replaced by West (2003). All earlier facsimiles, such as those of Lee (1902), Methuen (1910), and Kökeritz-Prouty (Yale, 1954), are surpassed by Hinman’s (1968; 1996) facsimile compiled from the best pages of copies in the Folger Library. For a facsimile of MND alone see West (c. 2008).

Press Variants

Hinman (1963, 1:260–1) identifies sigs. N2, N6v, and O3 as certainly or possibly indicating stop-press corrections. However, his claims that sig. N2 (where in some copies for in 331 is unevenly inked) (p. 261) almost certainly reflects proof correction; that sig. N6v (where in some copies a space prints after day-light, in 1481) thus contains a possible stop-press variant; and that sig. O3 (where a space prints after sent. in 2040) also exhibits stop-press correction have all been silently set aside by Rasmussen & West (2012, p. 875). Sig. O2 is variant (Hinman, p. 261): O2 (page 159, MND)—one non-textual variant only, as in O5v [where the page no. also varies].

1. page no. 165] 1 copy only (Folg. 60)
159] all others; Lee and Yale [facsimiles].

Compositor Identification and Order of Printing in F

Drawing upon variations in the spelling of frequently occurring words (e.g., do/doe and go/goe), Satchell (1920, p. 352) distinguishes between two compositors setting type in F1 Mac.; Willoughby (1932, pp. 56–8), applying Satchell’s method more widely in F1, concludes that these two compositors, now called Compositor A and Compositor B, must have been assisted by at least another pair, for MND, MV, and Rom. (p. 58) show no evidence . . . of having been composed by either A or B. Hinman (1957, p. 4), announcing his discovery of a new compositor in the F1 Tragedies, designates the newcomer Compositor E because not all of the material before the Tragedies was set by A and B, and C and D may later be required to designate compositors in the Comedies. Then, in his masterly study of the printing of F1, Hinman (1963, 1:193–200) uses evidence of the recurrence of distinctively damaged types as well as spellings to separate Compositors C and D from Compositors A and B. While (ibid., 2:518) Compositor D worked only on the Comedies (MM, Err., Ado, LLL, MND, MV, and AYL), Compositor C worked only most frequently on the Comedies, for Hinman finds it possible that C might be identified as Compositor B’s partner on plays in the Histories and Tragedies, particularly R2 and Ham. Hinman employs typographical evidence, rules, headlines and spellings to determine the order in which the pages of F1, including (ibid., 2:414–26) those in quires N and O (all of MND and the beginning of MV), were set into type and their compositors. Cairncross (1971, pp. 44, 47), in a rather unsystematic study, disputes a number of Hinman’s compositor attributions of particular pages. Howard-Hill (1973, pp. 83, 98) and O’Connor (1975, pp. 93–9, 117), in more thorough and orderly examinations, confirm Cairncross’s reassignment of four quire-O pages from Compositor A, to whom Hinman assigned them, to Compositor D, while disproving Cairncross’s (1971, p. 47; 1972, pp. 379, 406) other reattributions of pages from Compositor C to Compositor B and from the latter to Compositor E. It is the combined work of Hinman, Howard-Hill, and O’Connor that is now widely accepted—see, e.g., Wells & Taylor (1987, p. 149).

Hinman demonstrated that MND, like all the other F plays, was set into type by formes from cast-off copy, or copy marked up to indicate exactly which lines were to fit on each page of the quire, so that the pages could be set out of order. He identified three different compositors, each at a different case—Compositor B at case y, Compositor C at case x, and Compositor D at case z. The order of printing of the formes and the division of work among the compositors discovered by Hinman follows, but the pages marked with an asterisk are those confirmed or reassigned to Compositor D by, in succession, Cairncross, Howard-Hill, and O’Connor:

Cx Dz Cx Dz By Dz Cx Cx By Dz By Dz Dz By By Cx Cx Dz Cx By Cx Cx Dz Cx Dz Cx Dz
N3v: N4 N3: N4v N2v: N5 N2: N5v N1v: N6 N1: N6va *N6vb O3: (O4va) (O4vb) O2v: (*O5) O3v: (O4a1–50) (O4a51-b) O2: (*O5v) O1v: (*O6) O1: (*O6v)

Mistakes in casting could force compositors to alter the line division of their copy. All three of the F MND compositors change line division, but few of these changes evidently compensate for faulty casting-off. Werstine (1984, p. 114): Compositor B splits in two Q verse lines at 413–14 (correction of a Q error of running the initial half-line of speech together with the next line), 430–1, and 434–5; Compositor C at 1560–1; and Compositor D at 1441–2. Compositor B may deliberately stretch his copy for sig. N2v when he divides in two the pentameters on each side of a mid-line SD above and below which he creates white space. Both Compositor C and D, though, are simply dividing the first lines of speeches that, combined with SPP, are each too wide for the F column (p. 79). When B twice prints prose as verse on sig. O3 (2068–9, 2108–9) he neither saves nor loses space (p. 116). When he sets verse as prose at 2044–5, he is (pp. 87–8) faced with a verse line too long for his composing stick and therefore runs the end of the line together with the following verse line to set both lines as [88] prose. While the lines appear on sig. O3 in the first half of a quire, where he may need to adjust his copy to available space, his deliberately saving a line of type seems unlikely because elsewhere in the same column he allows for lavish white space around SDD. (P. 92): Compositor D divides off the last sentence of a prose speech at 999–1002 on sig. N5, thereby using an extra line of type, but because he is setting a page in the second half of a quire, space is not a factor. Instead the relineation seems designed to mark a change of address or topic and seems to be associated with other such changes he made in LLL at 128–30 and 800–2.

Most, but not all, the verbal variants between Q2 and F have been recognized and classified as corrections or errors introduced by the three compositors: for Compositor B see Werstine (1978); for Compositors C and D O’Connor (1977).

Features of F

Collier (ed. 1842, 2:cc2): The chief difference between the two quartos and the folio is, that in the latter the Acts, but not the Scenes, are distinguished. (Rhodes [1923, p. 120] addresses the possibility that the divisions [into acts in F] . . . were made in consonance with theatrical practice and connoted pauses [between acts at the Blackfriars or the new Globe]. . . . [T]he division into five acts necessitates two pauses during the game of blind man’s bluff in the woods, which is marked into three acts. Although it shows execrable stage-management, at the end of Actus Tertius is a note They sleepe all that act, meaning that the four lovers would have to lie, feigning sleep, in view of the audience while the act is playing [the act being the music between the acts]. . . . [I]t is indisputable that the division . . . was made by the prompter in consonance with theatrical practice. It cannot be entertained for a moment that They sleepe through the act [sic] was a literary or editorial note, to assist a reader in visualising the action. Foakes [ed. 1984, p. 151] believes that the act in this SD refers to the next act—Act 4.) Furness (ed. 1895, p. xv): In Roberts’s (Q2) [there are] about seventy-four [SDD]; and in the Folio, about ninety-seven.

Q2 Copy For F

Johnson (ed. 1765, 1:176): Roberts [i.e., Q2] was followed, though not without some variations, by Hemings and Condel [i.e., F]. Furness (ed. 1895, pp. xiii–xiv) demonstrates how the failure of the Q2 compositor to follow his copy precisely and set the word and in roman type in Titania’s line 979 in turn led the F compositor to create the redundant SD at 979–80: Enter Pease-blossome, Cobweb, Moth, Mustard-|seede and foure Fairies. Ebsworth (ed. Q2 1880, pp. xx–xxi) cites as proof of F’s use of Q2 as copy common errors at 183, 481, 482, 552, 1199, 213, 532, 557, 1652, 1688. Wilson (ed. 1924, p. 154): In 1619 . . . Jaggard . . . followed [his copy the Fisher Q] with suspicious exactitude. Apparently, however, the constant occurrence of Queene and Quince in dialogue, stage-direction and speech-heading, strained the resources of his compositors’ type. In any event, the italic Q seems to have given out on sig. D1r. and D2v., and accordingly the name Peter had to be resorted to in place of Quince. The fact that the F. also reads Peter in this same section of the text is a proof that it was set up from the Q. of 1619 and not from the Fisher Q. of 1600. Another proof is the reappearance in 1623 of nearly all the sixty to seventy misprints first introduced into the text in 1619. When we observe, moreover, that to these transmitted misprints the F. compositors added another sixty to seventy of their own, it will be evident that the F. version cannot claim much textual authority.

Only Craig (1961, pp. 108–9) appears to dissent from the view that F was printed from a copy of Q2: As a printed version of the same manuscript from which the fair copy had been made, Q1 would resemble the theatrical version very closely, and this may be said of both of the quarto and the folio as they stand. It does not seem necessary therefore, in view of this identity of origin, to imagine that the folio has been set from the quarto. Printing of the folio from the playhouse copy is a simpler and more satisfactory way in which to account for resemblances between these two texts. . . . [109] Although [F] has some features that may be derived from Q2, [it] actually resembles Q1 more closely than it does Q2. There are of course passages in which F differs from both Q1 and Q2. . . . In this perplexity one has to content oneself with a moderate position: the official playbook was in the hands of Jaggard and Blount and served them as copy for the body of the play, although there are in F some minor resemblances to Q2.

Brooks (ed. 1979, pp. xxviii–xix) cites evidence . . . of several kinds for the use of Q2 as F’s copy: At [1170] both [Q2 and F] have the abbreviation Hell. (Q1 Hel.), unique in each. On fifteen further occasions they have identical abbreviations in speech-prefixes, differing from Q1’s and from some of their own. Hence we can be confident that the tucking-in of Enter Snowt [929], rather than giving it the normal line of its own, and the printing of prose as verse at [1986–8], come in F from Q2, even though they originated in Q1. The Folio has several instances of progressive corruption. At [253] Q1 reads is so oft; Q2 inadvertently omitted so; F, lamely attempting to mend the metre, miscorrects to is often. Q2, at [1703], undoes the Q1 inversion more will hear, reading will hear more; F worsens the corruption with shall hear more. There are less striking instances at [1415, 1420]. Lysander’s sentence at [1677–8] is left incomplete because Egeus interrupts him; not realizing this, Q2 completes it by supplying a verb: be. The Folio repeats this and over fifty of its other corruptions: good examples are Q2’s silly foal for filly foal (misreading long s) [417], . . . and hearken for listen (a compositor’s synonym) [2038].

See the textual notes of this edition for all significant differences between Q2 and F, including SPP, SDD, lineation, punctuation, and verbal variants.

Annotation of Q2 Copy for F

Capell (1783, 2:3:111 ff.) identifies some F-only SDD as playhouse interpolations (see here). White (ed. 1857, 4:17): Printed copy [for F] had been used at the theatre for stage purposes and corrected with some care. Ebsworth (ed. Q2 1880, p. xix) dissents: It is idle to talk of the Folio editors having access to any manuscript authority for [MND]. We hold it indisputable that they used Roberts’s printed Quarto, sometimes increasing the defects, sometimes guessing commonplace variations; but they give absolutely nothing of such improvements as would have been gained from a genuine manuscript, or even from a certified revised and corrected prompt-book. Halliwell-Phillipps (1884, p. 255) confirms the theatrical provenance of the copy of Q2 or the MS used to annotate that copy by identifying Tawyer (1924) as a subordinate actor in the Globe Theatre in the pay of Heminge’s [sic]. For more on Tawyer, see G. E. Bentley, Jacobean and Caroline Stage, 2:590. Furness (ed. 1895, pp. xiv–xv) traces the appearance of Egeus for Philostrate in F’s Act 5 to the doubling of their parts and identifies the Asse head (927) as a prompter’s term: (p. xv) the prompter of Shakespeare’s stage, knowing well enough that there was among the scanty properties but one Asse-head, inserted in the text with the Asse head—the only one they had. Brooks (ed. 1979, p. xxxii n. 1), objecting to the possibility of doubling Egeus and Philostrate, points out that there is no time for an actor to re-enter as Egeus just after he has exited as Philostrate in 1.1. See also Greg, 1955, p. 243, imagining Philostrate unavailable for Act 5 because he doubled another role that also needed to be performed then, and Hodgdon, 1986, p. 536. Smidt (1986, pp. 121–2) also disagrees about the doubling: When the Folio substitutes Egeus for Philostrate as master of ceremonies at the wedding feast this could be explained as a way of saving an actor’s part, but there are no great number of men’s parts in [MND], and it is more likely that at some point in the stage history of the play someone objected to the absence of Egeus at the feast and thought he ought to join the party once he had been admitted to the comedy. Brooks seems right. More likely the appearance of Egeus for Philostrate in Act 5 arises from the telescoping of the roles. When bookkeepers subsumed one role under another in actual playhouse MSS, they often failed systematically to record the disappearance of the subsumed role in SDD and SPP; hence perhaps the persistence of Philostrate in Act 1 and once in a SP in Act 5 (1874) in F (see Werstine 2012, pp. 164–72). For the idea that Egeus, rather than Philostrate, appeared in 5.1 in the allegedly earliest (1594) version of the play, see Hunter (1998, pp. 8–9, and 2002, p, 6). For the idea that John Heminge annotated the copy of Q2 with notes in which he recalled a 1594 performance of the play, see Hunter (2002, pp. 7–10). For the application of literary and/or performance criticism to the Q1/F variants, particularly Philostrate/Egeus, see Hodgdon (1986), Wells (1991, MND Revisited, p. 22), Calderwood (1991, p. 428, n. 40), Wiles (1993, p. 174), Holland (ed. 1994, pp. 265–8), McGuire (1988, 1989), Pollack-Pelzner (2009). Taylor (2002, p. 52, n. 31) contends that F’s substitution of Egeus for Philostrate in Act 5 is inexplicable and therefore certainly wrong.

Cuningham (ed. 1905, p. xvi): the text of the Folio has its direct origin in a stage copy of Q 2. He cites as evidence ll. 1924, 927, and 2107, alleging in connection with the last that the early entrance of Thisby is an indication that printer’s copy was a stage copy . . . indicating that the actor was to be ready before he has to make his actual appearance on stage. Such an observation about the F SD as a warning direction is fanciful because Flute as Thisby comes onstage fewer than a half-dozen lines later; actual theatrical texts almost always mark warnings much earlier.

By the 1920s confidence that the copy of Q2 used in the playhouse must also have served as printer’s copy for F begins to slip. Adams (1923, pp. 538–9): MND, like R2, 1H4, Tit., and Ado, was printed in F from the actors’ special copies of . . . quartos which had been converted at the theatre into prompt-books or from (p. 539) the most available editions of these quartos [after they had been compared to] . . . the actors’ prompt-books . . . ; these collated quartos [would have been placed] in the hands of the [F] compositors. Wilson (ed. 1924, p. 78), in addition to suggesting an annotated quarto that served as printer’s copy may have served as a playhouse promptbook also allows only that Q2 copy for F could have been corrected for the purpose of publication, by some scribe working with the prompt-book before him. Wilson is led to suggest such an alternative by his belief that a copy of Q1 may have served the acting company in the playhouse; the Q1 punctuation of Theseus’s speech at 1841–57 is the slender reed on which he builds: (p. 157, n. 1) Now each item in the brief [read by Lysander in F] in Q. 1600 is followed by a question-mark, as if it were a query put to some one who replies with the comments [the only parts of the speech given Theseus in F], and it looks very probable that it was these queries which suggested the F. arrangement. If so, then the theatre prompt-book was almost certainly a copy of Q. 1600 seeing that all the queries but two towards the end of the speech, have been eliminated in Q. 1619. Acquainted as he is with playhouse MSS, Wilson also attempts to locate SDD first printed in F in particular places on the pages of the quarto prompt-book, suggesting (p. 156) that shifting places [1460] appeared in the margin of sig. F2, where it governed the action represented on the page as a whole, rather than simply in the line opposite which it is printed in F, to which it is irrelevant. He also imagines (p. 157) that Enter Pucke, printed in F over twenty lines before he needs to enter [865], was in the playhouse quarto a warning SD, noted atop sig. D2. Although Greg (1942, pp. 125–6) accepts the tradition that F’s new directions undoubtedly originated in the playhouse, he denies that the copy of Q2 from which F was printed could have served to guide performance because F is not sufficiently consistent and correct (p. 126): I should have expected to find more of the book-keeper’s notes in the original prompt-book, and therefore in Q; and if Q had itself been used as a prompt-book I should have expected to find certain anomalies removed in F. If the book-keeper found it necessary to specify the Lovers in v.i [see above, here] in the original prompt-book, why did he not the Clowns in III.i [813] either there or in the prompt quarto, especially since in the latter he took the trouble to translate the rabble into Snout and Starveling in IV.ii [1746]? Surely the errors in I.i whereby two half-lines of text appear as stage directions [30, 33] would have been corrected. Why does the entrance of the translated Bottom appear out of place [927]? The duplication in V.i, whereby we have Exit all but Wall [1951] followed three lines later by the exit of Lion, Thisbe, and Moonshine [1955], could hardly have been overlooked in performance. The second is the original direction of Q; the first must have been introduced from a manuscript. No doubt some confusion might have occurred in transferring the prompter’s notes from the copy Q1 to one of Q2; but on the whole the theory that a quarto was used as prompt copy seems to raise more difficulties than it solves. Idem (1955, pp. 244–5) also cites as additional examples of the incompetence and clumsiness of his imagined editor of Q2 copy for F: 865, 888, 1385, 1509, 1541, 1559, 1661–2, 1746, 1819, 2009, as they appear in F. Greg’s idea that Q2 copy for F was not itself annotated and used in the playhouse is followed by Doran (ed. 1959, p. 174), Wells (ed. 1967, p. 165); Evans (ed. 1974, p. 247); Brooks (ed. 1979, pp. xxix–xxx); Foakes (ed. 1984, p. 147); Wells & Taylor (1987, p. 279); Holland (ed. 1994, p. 115); Berger (ed. 1995, pp. x–xi). Nonetheless, quartos were annotated for playhouse use (see Werstine 2012, pp. 314–17, 335–42) and continued to be throughout the 17th c., and Greg’s expectations of thoroughness, consistency, and correctness of annotation in early modern theatrical texts are denied by the contents of such actual texts (see Werstine 2012, pp. 107–99, 234–391).

Authority and Revision in F

To judge from Capell’s comments on particular variants in F’s dialogue and SDD, he attributes no authority to F, but Malone has somewhat higher regard for F, and by the latter half of the 19th c., a number of editors are prepared to grant Shn. authority to readings in it. The New Bibliographers return to Capell’s positon, but near the end of the 20th c. there is a revival of the 19th-c. belief in F.

Capell (1783, 2:3:111) thinks F’s SD Musicke Tongs, Rurall Musicke (1541) simply wrong: it is certainly an interpolation of the players; as no such direction appears in either quarto, and Titania’s reply is a clear exclusion of it. He denies (2.3:113–14) the authority of F’s cut at 1718+1 of Are you sure / That we are awake? He is equally dismissive of (2.3:115) F’s redistribution of some of 1841–57 to Lysander— this reading and commenting of two persons, alternately, has something aukward in it: and seems a change of the players, calculated for the ease of the actor who presented the latter character—and of the player editors’ error in making Egeus enter in an act [Act 5] he has no concern in . . . (probably) from their laying Philostrate’s character in this act upon the player who had finish’d that of Egeus. It is not clear if Capell is suggesting that the Egeus actor doubled the role of Philostrate throughout the play or only in the last act. (Capell [1783, 2.3:116] thinks Sh.’s own revision can be recovered in small part from F2, in particular in the reading streames at 2076.)

Malone (in Boswell, ed. 1821, 1:203), though, calls for more respect for F: Thus therefore the first folio, as far as respects the plays above enumerated [including MND], labours under the disadvantage of being at least a second, and in some cases a third, edition of these quartos. I do not, however, mean to say, that many valuable corrections of passages undoubtedly corrupt in the quartos are not found in the folio copy; or that a single line of these plays should be printed by a careful editor without a minute examination, and collation of both copies [i.e., Q and F].

For White (ed. 1857, 4:17) though, Neither quarto . . . is to be regarded in any other light than as an assistant in eliminating such corruptions as may have crept into the folio itself; though Fisher’s enables us to correct some errors which were passed over in the copy of the quarto furnished to the printers by Heminge and Condell. The quartos sometimes concur in a reading different from that in the folio; but this is of little moment: it merely shows (unless in the case of a palpable corruption of the press) that in the copy from which the folio was printed, an error is corrected which had appeared in both the previous editions. The presumption is especially in favor of the authorized edition [i.e., F], when we know that it was printed from a copy that had been corrected in Shakespeare’s theatre, and probably under his own eye, if not by his own hand. (In particular White cites the F readings at 700, 1247, 1384+, 1718+1, 1994 [corrected by Shakespeare or someone else in his theatre], and 2010 as authoritative; however, he thinks 1812–13, which are common to Q and F, an interpolation and the substitution of Egeus for Philostrate in Act 5 wrong, both not Sh.’s.) So Furness (ed. 1895, p. xii): It may be that in using a printed text [for F, namely Q2, Heminge and Condell] were virtually using Shakespeare’s manuscript if they knew that this text . . . had been for years used in their theatre as a stage copy, with possible additional stage-business marked on the margin for the use of the prompter, and here and there sundry emendations, noted possibly by the author’s own hand, who, by these changes, theoretically authenticated all the rest of the text. Adams (1923, pp. 539): These printed prompt-copies [such as the copy of Q2 used to print F MND] would receive corrections (from the author, or from the actors), alterations, and additions and such stage-directions as were found necessary.

However, the New Bibliographer Wilson (ed. 1924, p. 159), generalizing about the changes in SDD and SPP introduced into F, returns to Capell’s position: it should not be necessary to argue that Shakespeare himself had nothing whatever to do with them. Even less easy is it to imagine him in any way responsible for the F. corrections in the dialogue. [He lists those unlikely to be due to the compositors: 759, 1247, 1287, 1719, 1829, 1994, 2010.] These variants are almost certainly due to the scribe who gave us the F. stage-directions. Some of them are good, some indifferent, and some definitely bad; but all are assuredly guesses. Greg (1942, p. 125–6) on F’s text: The new directions undoubtedly originated in the playhouse. (P. 126): Such changes in the text as are not either misprints or corrections of misprints seem to be the editor’s and do not imply any independent source.

Brooks (ed. 1979, pp. xxx–xxxiii), following Greg’s questionable presumption that the copy of Q2 from which F was set could not itself have been the playhouse text used to guide performance: The promptbook consulted in preparing copy for the Folio was clearly not without alterations from the text in the prompt-copy as [xxxi] originally transcribed from Shakespeare’s autograph [assuming without grounds that that Shn. copy could not itself have been used in the playhouse]. Theoretically, some changes may have been authorized by him; but at least the majority were no doubt made without authority, whether in the original prompt-book or in a new one, if a new one was transcribed from it. But whatever unauthentic changes had accumulated in it, the prompt-copy which supplied some Folio readings did derive by a process of transcription from Shakespeare’s autograph. That process was independent of Q1. Accordingly, in respect of readings which the Folio can be presumed to have taken from the prompt-copy, F is an independent witness to what may have stood in the autograph. In the line of descent described it is the earliest extant witness, and in respect of those readings, and of those alone, it is therefore a substantive (that is, an evidential) text—as Q1 is for the play as a whole. Such authority as F therefore has is weakened, however, by the annotator’s demonstrable negligence and clumsiness. He cites the misplaced SD at 927 and the duplicate SDD introduced at 865 and 1951. His neglect of dialogue further limits the possible authority of F. This is significant for F’s readings at [1247 and 1994]: passionate where Q1 has a palpable omission, and knit up in thee where Q1 has the impossible knit now againe. . . . Yet if they are retrievals from prompt-copy, why are there not more? That the annotator’s eye might fall upon dialogue may be suggested by F’s choise of merit for the Quartos’ choise of friends [149]. Brooks (pp. 154–5) makes a case that merit could have been Sh.’s first choice of reading in his initial composition, one that he later replaced with friends but one that nonetheless found its way into the playhouse text from Sh.’s own papers. In spite of his confidence in the authority of this single F-only reading, Brooks thinks that his annotator of Q2 (p. xxxii) was perfectly prepared to guess, even when he could have consulted the prompt-book, and passionate and up in thee may be other guesses of his, though there the contexts offered little hint.

A further subtraction has to be made from the authority of F’s text, even where its source is prompt-copy. The prompt-copy itself is unlikely still to have represented in all respects the kind of performance for which Shakespeare designed the play, or to which he may have adapted it. The substitution of Egeus for Philostrate in Act V, at odds with Theseus’ enquiry for our usual manager of mirth and damaging the metre at [1833] was made apparently to save a speaking part (Philostrate is mute in I.i): it is a change Shakespeare cannot have wished for, though he might acquiesce in it as an expedient. The same may be said of the one or more intervals [at the ends of acts] introduced in a play conceived and originally performed as a continuous action. . . . [xxxiii] If the revival of the Dream matched by the prompt-book was in 1609 or later [the approximate date at which Sh.’s company began to perform at the Blackfriars and observe intervals between acts for the first time], Shakespeare may not have been closely associated with it.

With Foakes (ed. 1984, p. 148), Wells & Taylor (1987, pp. 279–80), and Holland (ed. 1994, pp. 115–17, 257–68) the 19th-c. understanding of F’s authority makes a return. Foakes interprets as Sh.’s revisions the few corrections made to Q2’s dialogue in F: 149, 1041, 1247, 1994, and 2010. He also thinks that Puck’s early entrance in F at 865 in 3.1 records prompt-copy, suggesting that Puck should overhear rather more of the dialogue between Quince and his crew than his entry in the quartos would permit. Wells and Taylor present the F text’s possible censorship as a consequence of the 1606 Acte to Restraine Abuses, shown by the loss of 2113–2113+1 as further evidence of its theatrical provenance. They adopt the following editorial policy: Without strong evidence to the contrary, one must therefore assume that the prompt-book is the authority for all added or substantially altered Folio directions and speech prefixes. Some of these variants might derive from late revivals, over which Shakespeare had no control; but none certainly do [sic], and only the act divisions and Tawyer’s name can be confidently associated with performances later than those in the mid 1590s. Although each direction has been considered on its merits, we have found no reason to doubt that the bulk of the Folio directions represent the play as originally and authoritatively staged. Those directions which clearly envisage a different staging from that implied by Q seem to us to be dramatic improvements for which Shakespeare was probably responsible. Such an editorial policy forces justification of F SDD that were long thought to be erroneous, such as the F entrance of Pucke in 3.1. [at 865] over twenty lines before Q1’s entrance for him (which is also reproduced in F, 888) and twenty lines before, for all one can tell, he has business onstage: (p. 281) an editor committed to entertain possible authorial revision must consider the F alternative. (Pp. 281–2): Following Greg’s unwarranted assumption that F had to have been printed from a copy of Q2 annotated with reference to a playhouse MS (rather than from a copy of Q2 annotated for use in the playhouse), Wells and Taylor also assume that the annotator must have been right to add the F SD from the playhouse MS. They justify this second assumption by imagining that F records accurately a production in which Puck entered silently and unnoticed to supply Quince with the almanac he was requesting at the point of the F SD. (Werstine [2012, pp. 173–6], however, shows that there is no reason to suppose bookkeepers’ additions of entrances necessarily inerrant because in actual theatrical texts some such additions can be shown to be erroneous in context; consequently, playhouse texts, such as the one inferred to lie behind F MND, need not be reliable records of any performance.) Holland (ed. 1994, p. 117) attributes to some other authority than the compositor’s or editor’s ingenuity the five readings adopted by Foakes as Shn. revisions. He writes (pp. 257–68) of Shakespeare’s Revisions of Act 5, accepting Wilson’s account of the mislined verse at 1798–1880 and counting as a second revision the substitution of Egeus for Philostrate and the hiving off of pieces of Theseus’s 1841–57 speech for Lysander. Ioppolo (1991, p. 113) had associated Sh.’s alleged revision of 1824–5 in Q1’s printer’s copy with the transfer of some of Theseus’s lines to Lysander in the F printer’s copy, but Holland (ed. 1994, p. 266) demurred.

Press Variants

Hinman (1963, 1:260–1) identifies sigs. N2, N6v, and O3 as certainly or possibly indicating stop-press corrections. However, his claims that sig. N2 (where in some copies for in 331 is unevenly inked) (p. 261) almost certainly reflects proof correction; that sig. N6v (where in some copies a space prints after day-light, in 1481) thus contains a possible stop-press variant; and that sig. O3 (where a space prints after sent. in 2040) also exhibits stop-press correction have all been silently set aside by Rasmussen & West (2012, p. 875). Sig. O2 is variant (Hinman, p. 261): O2 (page 159, MND)—one non-textual variant only, as in O5v [where the page no. also varies].

1. page no. 165] 1 copy only (Folg. 60)
159] all others; Lee and Yale [facsimiles].

Compositor Identification and Order of Printing in F

Drawing upon variations in the spelling of frequently occurring words (e.g., do/doe and go/goe), Satchell (1920, p. 352) distinguishes between two compositors setting type in F1 Mac.; Willoughby (1932, pp. 56–8), applying Satchell’s method more widely in F1, concludes that these two compositors, now called Compositor A and Compositor B, must have been assisted by at least another pair, for MND, MV, and Rom. (p. 58) show no evidence . . . of having been composed by either A or B. Hinman (1957, p. 4), announcing his discovery of a new compositor in the F1 Tragedies, designates the newcomer Compositor E because not all of the material before the Tragedies was set by A and B, and C and D may later be required to designate compositors in the Comedies. Then, in his masterly study of the printing of F1, Hinman (1963, 1:193–200) uses evidence of the recurrence of distinctively damaged types as well as spellings to separate Compositors C and D from Compositors A and B. While (ibid., 2:518) Compositor D worked only on the Comedies (MM, Err., Ado, LLL, MND, MV, and AYL), Compositor C worked only most frequently on the Comedies, for Hinman finds it possible that C might be identified as Compositor B’s partner on plays in the Histories and Tragedies, particularly R2 and Ham. Hinman employs typographical evidence, rules, headlines and spellings to determine the order in which the pages of F1, including (ibid., 2:414–26) those in quires N and O (all of MND and the beginning of MV), were set into type and their compositors. Cairncross (1971, pp. 44, 47), in a rather unsystematic study, disputes a number of Hinman’s compositor attributions of particular pages. Howard-Hill (1973, pp. 83, 98) and O’Connor (1975, pp. 93–9, 117), in more thorough and orderly examinations, confirm Cairncross’s reassignment of four quire-O pages from Compositor A, to whom Hinman assigned them, to Compositor D, while disproving Cairncross’s (1971, p. 47; 1972, pp. 379, 406) other reattributions of pages from Compositor C to Compositor B and from the latter to Compositor E. It is the combined work of Hinman, Howard-Hill, and O’Connor that is now widely accepted—see, e.g., Wells & Taylor (1987, p. 149).

Hinman demonstrated that MND, like all the other F plays, was set into type by formes from cast-off copy, or copy marked up to indicate exactly which lines were to fit on each page of the quire, so that the pages could be set out of order. He identified three different compositors, each at a different case—Compositor B at case y, Compositor C at case x, and Compositor D at case z. The order of printing of the formes and the division of work among the compositors discovered by Hinman follows, but the pages marked with an asterisk are those confirmed or reassigned to Compositor D by, in succession, Cairncross, Howard-Hill, and O’Connor:

Cx Dz Cx Dz By Dz Cx Cx By Dz By Dz Dz By By Cx Cx Dz Cx By Cx Cx Dz Cx Dz Cx Dz
N3v: N4 N3: N4v N2v: N5 N2: N5v N1v: N6 N1: N6va *N6vb O3: (O4va) (O4vb) O2v: (*O5) O3v: (O4a1–50) (O4a51-b) O2: (*O5v) O1v: (*O6) O1: (*O6v)

Mistakes in casting could force compositors to alter the line division of their copy. All three of the F MND compositors change line division, but few of these changes evidently compensate for faulty casting-off. Werstine (1984, p. 114): Compositor B splits in two Q verse lines at 413–14 (correction of a Q error of running the initial half-line of speech together with the next line), 430–1, and 434–5; Compositor C at 1560–1; and Compositor D at 1441–2. Compositor B may deliberately stretch his copy for sig. N2v when he divides in two the pentameters on each side of a mid-line SD above and below which he creates white space. Both Compositor C and D, though, are simply dividing the first lines of speeches that, combined with SPP, are each too wide for the F column (p. 79). When B twice prints prose as verse on sig. O3 (2068–9, 2108–9) he neither saves nor loses space (p. 116). When he sets verse as prose at 2044–5, he is (pp. 87–8) faced with a verse line too long for his composing stick and therefore runs the end of the line together with the following verse line to set both lines as [88] prose. While the lines appear on sig. O3 in the first half of a quire, where he may need to adjust his copy to available space, his deliberately saving a line of type seems unlikely because elsewhere in the same column he allows for lavish white space around SDD. (P. 92): Compositor D divides off the last sentence of a prose speech at 999–1002 on sig. N5, thereby using an extra line of type, but because he is setting a page in the second half of a quire, space is not a factor. Instead the relineation seems designed to mark a change of address or topic and seems to be associated with other such changes he made in LLL at 128–30 and 800–2.

Most, but not all, the verbal variants between Q2 and F have been recognized and classified as corrections or errors introduced by the three compositors: for Compositor B see Werstine (1978); for Compositors C and D O’Connor (1977).

Features of F

Collier (ed. 1842, 2:cc2): The chief difference between the two quartos and the folio is, that in the latter the Acts, but not the Scenes, are distinguished. (Rhodes [1923, p. 120] addresses the possibility that the divisions [into acts in F] . . . were made in consonance with theatrical practice and connoted pauses [between acts at the Blackfriars or the new Globe]. . . . [T]he division into five acts necessitates two pauses during the game of blind man’s bluff in the woods, which is marked into three acts. Although it shows execrable stage-management, at the end of Actus Tertius is a note They sleepe all that act, meaning that the four lovers would have to lie, feigning sleep, in view of the audience while the act is playing [the act being the music between the acts]. . . . [I]t is indisputable that the division . . . was made by the prompter in consonance with theatrical practice. It cannot be entertained for a moment that They sleepe through the act [sic] was a literary or editorial note, to assist a reader in visualising the action. Foakes [ed. 1984, p. 151] believes that the act in this SD refers to the next act—Act 4.) Furness (ed. 1895, p. xv): In Roberts’s (Q2) [there are] about seventy-four [SDD]; and in the Folio, about ninety-seven.

Q2 Copy For F

Johnson (ed. 1765, 1:176): Roberts [i.e., Q2] was followed, though not without some variations, by Hemings and Condel [i.e., F]. Furness (ed. 1895, pp. xiii–xiv) demonstrates how the failure of the Q2 compositor to follow his copy precisely and set the word and in roman type in Titania’s line 979 in turn led the F compositor to create the redundant SD at 979–80: Enter Pease-blossome, Cobweb, Moth, Mustard-|seede and foure Fairies. Ebsworth (ed. Q2 1880, pp. xx–xxi) cites as proof of F’s use of Q2 as copy common errors at 183, 481, 482, 552, 1199, 213, 532, 557, 1652, 1688. Wilson (ed. 1924, p. 154): In 1619 . . . Jaggard . . . followed [his copy the Fisher Q] with suspicious exactitude. Apparently, however, the constant occurrence of Queene and Quince in dialogue, stage-direction and speech-heading, strained the resources of his compositors’ type. In any event, the italic Q seems to have given out on sig. D1r. and D2v., and accordingly the name Peter had to be resorted to in place of Quince. The fact that the F. also reads Peter in this same section of the text is a proof that it was set up from the Q. of 1619 and not from the Fisher Q. of 1600. Another proof is the reappearance in 1623 of nearly all the sixty to seventy misprints first introduced into the text in 1619. When we observe, moreover, that to these transmitted misprints the F. compositors added another sixty to seventy of their own, it will be evident that the F. version cannot claim much textual authority.

Only Craig (1961, pp. 108–9) appears to dissent from the view that F was printed from a copy of Q2: As a printed version of the same manuscript from which the fair copy had been made, Q1 would resemble the theatrical version very closely, and this may be said of both of the quarto and the folio as they stand. It does not seem necessary therefore, in view of this identity of origin, to imagine that the folio has been set from the quarto. Printing of the folio from the playhouse copy is a simpler and more satisfactory way in which to account for resemblances between these two texts. . . . [109] Although [F] has some features that may be derived from Q2, [it] actually resembles Q1 more closely than it does Q2. There are of course passages in which F differs from both Q1 and Q2. . . . In this perplexity one has to content oneself with a moderate position: the official playbook was in the hands of Jaggard and Blount and served them as copy for the body of the play, although there are in F some minor resemblances to Q2.

Brooks (ed. 1979, pp. xxviii–xix) cites evidence . . . of several kinds for the use of Q2 as F’s copy: At [1170] both [Q2 and F] have the abbreviation Hell. (Q1 Hel.), unique in each. On fifteen further occasions they have identical abbreviations in speech-prefixes, differing from Q1’s and from some of their own. Hence we can be confident that the tucking-in of Enter Snowt [929], rather than giving it the normal line of its own, and the printing of prose as verse at [1986–8], come in F from Q2, even though they originated in Q1. The Folio has several instances of progressive corruption. At [253] Q1 reads is so oft; Q2 inadvertently omitted so; F, lamely attempting to mend the metre, miscorrects to is often. Q2, at [1703], undoes the Q1 inversion more will hear, reading will hear more; F worsens the corruption with shall hear more. There are less striking instances at [1415, 1420]. Lysander’s sentence at [1677–8] is left incomplete because Egeus interrupts him; not realizing this, Q2 completes it by supplying a verb: be. The Folio repeats this and over fifty of its other corruptions: good examples are Q2’s silly foal for filly foal (misreading long s) [417], . . . and hearken for listen (a compositor’s synonym) [2038].

See the textual notes of this edition for all significant differences between Q2 and F, including SPP, SDD, lineation, punctuation, and verbal variants.

Annotation of Q2 Copy for F

Capell (1783, 2:3:111 ff.) identifies some F-only SDD as playhouse interpolations (see here). White (ed. 1857, 4:17): Printed copy [for F] had been used at the theatre for stage purposes and corrected with some care. Ebsworth (ed. Q2 1880, p. xix) dissents: It is idle to talk of the Folio editors having access to any manuscript authority for [MND]. We hold it indisputable that they used Roberts’s printed Quarto, sometimes increasing the defects, sometimes guessing commonplace variations; but they give absolutely nothing of such improvements as would have been gained from a genuine manuscript, or even from a certified revised and corrected prompt-book. Halliwell-Phillipps (1884, p. 255) confirms the theatrical provenance of the copy of Q2 or the MS used to annotate that copy by identifying Tawyer (1924) as a subordinate actor in the Globe Theatre in the pay of Heminge’s [sic]. For more on Tawyer, see G. E. Bentley, Jacobean and Caroline Stage, 2:590. Furness (ed. 1895, pp. xiv–xv) traces the appearance of Egeus for Philostrate in F’s Act 5 to the doubling of their parts and identifies the Asse head (927) as a prompter’s term: (p. xv) the prompter of Shakespeare’s stage, knowing well enough that there was among the scanty properties but one Asse-head, inserted in the text with the Asse head—the only one they had. Brooks (ed. 1979, p. xxxii n. 1), objecting to the possibility of doubling Egeus and Philostrate, points out that there is no time for an actor to re-enter as Egeus just after he has exited as Philostrate in 1.1. See also Greg, 1955, p. 243, imagining Philostrate unavailable for Act 5 because he doubled another role that also needed to be performed then, and Hodgdon, 1986, p. 536. Smidt (1986, pp. 121–2) also disagrees about the doubling: When the Folio substitutes Egeus for Philostrate as master of ceremonies at the wedding feast this could be explained as a way of saving an actor’s part, but there are no great number of men’s parts in [MND], and it is more likely that at some point in the stage history of the play someone objected to the absence of Egeus at the feast and thought he ought to join the party once he had been admitted to the comedy. Brooks seems right. More likely the appearance of Egeus for Philostrate in Act 5 arises from the telescoping of the roles. When bookkeepers subsumed one role under another in actual playhouse MSS, they often failed systematically to record the disappearance of the subsumed role in SDD and SPP; hence perhaps the persistence of Philostrate in Act 1 and once in a SP in Act 5 (1874) in F (see Werstine 2012, pp. 164–72). For the idea that Egeus, rather than Philostrate, appeared in 5.1 in the allegedly earliest (1594) version of the play, see Hunter (1998, pp. 8–9, and 2002, p, 6). For the idea that John Heminge annotated the copy of Q2 with notes in which he recalled a 1594 performance of the play, see Hunter (2002, pp. 7–10). For the application of literary and/or performance criticism to the Q1/F variants, particularly Philostrate/Egeus, see Hodgdon (1986), Wells (1991, MND Revisited, p. 22), Calderwood (1991, p. 428, n. 40), Wiles (1993, p. 174), Holland (ed. 1994, pp. 265–8), McGuire (1988, 1989), Pollack-Pelzner (2009). Taylor (2002, p. 52, n. 31) contends that F’s substitution of Egeus for Philostrate in Act 5 is inexplicable and therefore certainly wrong.

Cuningham (ed. 1905, p. xvi): the text of the Folio has its direct origin in a stage copy of Q 2. He cites as evidence ll. 1924, 927, and 2107, alleging in connection with the last that the early entrance of Thisby is an indication that printer’s copy was a stage copy . . . indicating that the actor was to be ready before he has to make his actual appearance on stage. Such an observation about the F SD as a warning direction is fanciful because Flute as Thisby comes onstage fewer than a half-dozen lines later; actual theatrical texts almost always mark warnings much earlier.

By the 1920s confidence that the copy of Q2 used in the playhouse must also have served as printer’s copy for F begins to slip. Adams (1923, pp. 538–9): MND, like R2, 1H4, Tit., and Ado, was printed in F from the actors’ special copies of . . . quartos which had been converted at the theatre into prompt-books or from (p. 539) the most available editions of these quartos [after they had been compared to] . . . the actors’ prompt-books . . . ; these collated quartos [would have been placed] in the hands of the [F] compositors. Wilson (ed. 1924, p. 78), in addition to suggesting an annotated quarto that served as printer’s copy may have served as a playhouse promptbook also allows only that Q2 copy for F could have been corrected for the purpose of publication, by some scribe working with the prompt-book before him. Wilson is led to suggest such an alternative by his belief that a copy of Q1 may have served the acting company in the playhouse; the Q1 punctuation of Theseus’s speech at 1841–57 is the slender reed on which he builds: (p. 157, n. 1) Now each item in the brief [read by Lysander in F] in Q. 1600 is followed by a question-mark, as if it were a query put to some one who replies with the comments [the only parts of the speech given Theseus in F], and it looks very probable that it was these queries which suggested the F. arrangement. If so, then the theatre prompt-book was almost certainly a copy of Q. 1600 seeing that all the queries but two towards the end of the speech, have been eliminated in Q. 1619. Acquainted as he is with playhouse MSS, Wilson also attempts to locate SDD first printed in F in particular places on the pages of the quarto prompt-book, suggesting (p. 156) that shifting places [1460] appeared in the margin of sig. F2, where it governed the action represented on the page as a whole, rather than simply in the line opposite which it is printed in F, to which it is irrelevant. He also imagines (p. 157) that Enter Pucke, printed in F over twenty lines before he needs to enter [865], was in the playhouse quarto a warning SD, noted atop sig. D2. Although Greg (1942, pp. 125–6) accepts the tradition that F’s new directions undoubtedly originated in the playhouse, he denies that the copy of Q2 from which F was printed could have served to guide performance because F is not sufficiently consistent and correct (p. 126): I should have expected to find more of the book-keeper’s notes in the original prompt-book, and therefore in Q; and if Q had itself been used as a prompt-book I should have expected to find certain anomalies removed in F. If the book-keeper found it necessary to specify the Lovers in v.i [see above, here] in the original prompt-book, why did he not the Clowns in III.i [813] either there or in the prompt quarto, especially since in the latter he took the trouble to translate the rabble into Snout and Starveling in IV.ii [1746]? Surely the errors in I.i whereby two half-lines of text appear as stage directions [30, 33] would have been corrected. Why does the entrance of the translated Bottom appear out of place [927]? The duplication in V.i, whereby we have Exit all but Wall [1951] followed three lines later by the exit of Lion, Thisbe, and Moonshine [1955], could hardly have been overlooked in performance. The second is the original direction of Q; the first must have been introduced from a manuscript. No doubt some confusion might have occurred in transferring the prompter’s notes from the copy Q1 to one of Q2; but on the whole the theory that a quarto was used as prompt copy seems to raise more difficulties than it solves. Idem (1955, pp. 244–5) also cites as additional examples of the incompetence and clumsiness of his imagined editor of Q2 copy for F: 865, 888, 1385, 1509, 1541, 1559, 1661–2, 1746, 1819, 2009, as they appear in F. Greg’s idea that Q2 copy for F was not itself annotated and used in the playhouse is followed by Doran (ed. 1959, p. 174), Wells (ed. 1967, p. 165); Evans (ed. 1974, p. 247); Brooks (ed. 1979, pp. xxix–xxx); Foakes (ed. 1984, p. 147); Wells & Taylor (1987, p. 279); Holland (ed. 1994, p. 115); Berger (ed. 1995, pp. x–xi). Nonetheless, quartos were annotated for playhouse use (see Werstine 2012, pp. 314–17, 335–42) and continued to be throughout the 17th c., and Greg’s expectations of thoroughness, consistency, and correctness of annotation in early modern theatrical texts are denied by the contents of such actual texts (see Werstine 2012, pp. 107–99, 234–391).

Authority and Revision in F

To judge from Capell’s comments on particular variants in F’s dialogue and SDD, he attributes no authority to F, but Malone has somewhat higher regard for F, and by the latter half of the 19th c., a number of editors are prepared to grant Shn. authority to readings in it. The New Bibliographers return to Capell’s positon, but near the end of the 20th c. there is a revival of the 19th-c. belief in F.

Capell (1783, 2:3:111) thinks F’s SD Musicke Tongs, Rurall Musicke (1541) simply wrong: it is certainly an interpolation of the players; as no such direction appears in either quarto, and Titania’s reply is a clear exclusion of it. He denies (2.3:113–14) the authority of F’s cut at 1718+1 of Are you sure / That we are awake? He is equally dismissive of (2.3:115) F’s redistribution of some of 1841–57 to Lysander— this reading and commenting of two persons, alternately, has something aukward in it: and seems a change of the players, calculated for the ease of the actor who presented the latter character—and of the player editors’ error in making Egeus enter in an act [Act 5] he has no concern in . . . (probably) from their laying Philostrate’s character in this act upon the player who had finish’d that of Egeus. It is not clear if Capell is suggesting that the Egeus actor doubled the role of Philostrate throughout the play or only in the last act. (Capell [1783, 2.3:116] thinks Sh.’s own revision can be recovered in small part from F2, in particular in the reading streames at 2076.)

Malone (in Boswell, ed. 1821, 1:203), though, calls for more respect for F: Thus therefore the first folio, as far as respects the plays above enumerated [including MND], labours under the disadvantage of being at least a second, and in some cases a third, edition of these quartos. I do not, however, mean to say, that many valuable corrections of passages undoubtedly corrupt in the quartos are not found in the folio copy; or that a single line of these plays should be printed by a careful editor without a minute examination, and collation of both copies [i.e., Q and F].

For White (ed. 1857, 4:17) though, Neither quarto . . . is to be regarded in any other light than as an assistant in eliminating such corruptions as may have crept into the folio itself; though Fisher’s enables us to correct some errors which were passed over in the copy of the quarto furnished to the printers by Heminge and Condell. The quartos sometimes concur in a reading different from that in the folio; but this is of little moment: it merely shows (unless in the case of a palpable corruption of the press) that in the copy from which the folio was printed, an error is corrected which had appeared in both the previous editions. The presumption is especially in favor of the authorized edition [i.e., F], when we know that it was printed from a copy that had been corrected in Shakespeare’s theatre, and probably under his own eye, if not by his own hand. (In particular White cites the F readings at 700, 1247, 1384+, 1718+1, 1994 [corrected by Shakespeare or someone else in his theatre], and 2010 as authoritative; however, he thinks 1812–13, which are common to Q and F, an interpolation and the substitution of Egeus for Philostrate in Act 5 wrong, both not Sh.’s.) So Furness (ed. 1895, p. xii): It may be that in using a printed text [for F, namely Q2, Heminge and Condell] were virtually using Shakespeare’s manuscript if they knew that this text . . . had been for years used in their theatre as a stage copy, with possible additional stage-business marked on the margin for the use of the prompter, and here and there sundry emendations, noted possibly by the author’s own hand, who, by these changes, theoretically authenticated all the rest of the text. Adams (1923, pp. 539): These printed prompt-copies [such as the copy of Q2 used to print F MND] would receive corrections (from the author, or from the actors), alterations, and additions and such stage-directions as were found necessary.

However, the New Bibliographer Wilson (ed. 1924, p. 159), generalizing about the changes in SDD and SPP introduced into F, returns to Capell’s position: it should not be necessary to argue that Shakespeare himself had nothing whatever to do with them. Even less easy is it to imagine him in any way responsible for the F. corrections in the dialogue. [He lists those unlikely to be due to the compositors: 759, 1247, 1287, 1719, 1829, 1994, 2010.] These variants are almost certainly due to the scribe who gave us the F. stage-directions. Some of them are good, some indifferent, and some definitely bad; but all are assuredly guesses. Greg (1942, p. 125–6) on F’s text: The new directions undoubtedly originated in the playhouse. (P. 126): Such changes in the text as are not either misprints or corrections of misprints seem to be the editor’s and do not imply any independent source.

Brooks (ed. 1979, pp. xxx–xxxiii), following Greg’s questionable presumption that the copy of Q2 from which F was set could not itself have been the playhouse text used to guide performance: The promptbook consulted in preparing copy for the Folio was clearly not without alterations from the text in the prompt-copy as [xxxi] originally transcribed from Shakespeare’s autograph [assuming without grounds that that Shn. copy could not itself have been used in the playhouse]. Theoretically, some changes may have been authorized by him; but at least the majority were no doubt made without authority, whether in the original prompt-book or in a new one, if a new one was transcribed from it. But whatever unauthentic changes had accumulated in it, the prompt-copy which supplied some Folio readings did derive by a process of transcription from Shakespeare’s autograph. That process was independent of Q1. Accordingly, in respect of readings which the Folio can be presumed to have taken from the prompt-copy, F is an independent witness to what may have stood in the autograph. In the line of descent described it is the earliest extant witness, and in respect of those readings, and of those alone, it is therefore a substantive (that is, an evidential) text—as Q1 is for the play as a whole. Such authority as F therefore has is weakened, however, by the annotator’s demonstrable negligence and clumsiness. He cites the misplaced SD at 927 and the duplicate SDD introduced at 865 and 1951. His neglect of dialogue further limits the possible authority of F. This is significant for F’s readings at [1247 and 1994]: passionate where Q1 has a palpable omission, and knit up in thee where Q1 has the impossible knit now againe. . . . Yet if they are retrievals from prompt-copy, why are there not more? That the annotator’s eye might fall upon dialogue may be suggested by F’s choise of merit for the Quartos’ choise of friends [149]. Brooks (pp. 154–5) makes a case that merit could have been Sh.’s first choice of reading in his initial composition, one that he later replaced with friends but one that nonetheless found its way into the playhouse text from Sh.’s own papers. In spite of his confidence in the authority of this single F-only reading, Brooks thinks that his annotator of Q2 (p. xxxii) was perfectly prepared to guess, even when he could have consulted the prompt-book, and passionate and up in thee may be other guesses of his, though there the contexts offered little hint.

A further subtraction has to be made from the authority of F’s text, even where its source is prompt-copy. The prompt-copy itself is unlikely still to have represented in all respects the kind of performance for which Shakespeare designed the play, or to which he may have adapted it. The substitution of Egeus for Philostrate in Act V, at odds with Theseus’ enquiry for our usual manager of mirth and damaging the metre at [1833] was made apparently to save a speaking part (Philostrate is mute in I.i): it is a change Shakespeare cannot have wished for, though he might acquiesce in it as an expedient. The same may be said of the one or more intervals [at the ends of acts] introduced in a play conceived and originally performed as a continuous action. . . . [xxxiii] If the revival of the Dream matched by the prompt-book was in 1609 or later [the approximate date at which Sh.’s company began to perform at the Blackfriars and observe intervals between acts for the first time], Shakespeare may not have been closely associated with it.

With Foakes (ed. 1984, p. 148), Wells & Taylor (1987, pp. 279–80), and Holland (ed. 1994, pp. 115–17, 257–68) the 19th-c. understanding of F’s authority makes a return. Foakes interprets as Sh.’s revisions the few corrections made to Q2’s dialogue in F: 149, 1041, 1247, 1994, and 2010. He also thinks that Puck’s early entrance in F at 865 in 3.1 records prompt-copy, suggesting that Puck should overhear rather more of the dialogue between Quince and his crew than his entry in the quartos would permit. Wells and Taylor present the F text’s possible censorship as a consequence of the 1606 Acte to Restraine Abuses, shown by the loss of 2113–2113+1 as further evidence of its theatrical provenance. They adopt the following editorial policy: Without strong evidence to the contrary, one must therefore assume that the prompt-book is the authority for all added or substantially altered Folio directions and speech prefixes. Some of these variants might derive from late revivals, over which Shakespeare had no control; but none certainly do [sic], and only the act divisions and Tawyer’s name can be confidently associated with performances later than those in the mid 1590s. Although each direction has been considered on its merits, we have found no reason to doubt that the bulk of the Folio directions represent the play as originally and authoritatively staged. Those directions which clearly envisage a different staging from that implied by Q seem to us to be dramatic improvements for which Shakespeare was probably responsible. Such an editorial policy forces justification of F SDD that were long thought to be erroneous, such as the F entrance of Pucke in 3.1. [at 865] over twenty lines before Q1’s entrance for him (which is also reproduced in F, 888) and twenty lines before, for all one can tell, he has business onstage: (p. 281) an editor committed to entertain possible authorial revision must consider the F alternative. (Pp. 281–2): Following Greg’s unwarranted assumption that F had to have been printed from a copy of Q2 annotated with reference to a playhouse MS (rather than from a copy of Q2 annotated for use in the playhouse), Wells and Taylor also assume that the annotator must have been right to add the F SD from the playhouse MS. They justify this second assumption by imagining that F records accurately a production in which Puck entered silently and unnoticed to supply Quince with the almanac he was requesting at the point of the F SD. (Werstine [2012, pp. 173–6], however, shows that there is no reason to suppose bookkeepers’ additions of entrances necessarily inerrant because in actual theatrical texts some such additions can be shown to be erroneous in context; consequently, playhouse texts, such as the one inferred to lie behind F MND, need not be reliable records of any performance.) Holland (ed. 1994, p. 117) attributes to some other authority than the compositor’s or editor’s ingenuity the five readings adopted by Foakes as Shn. revisions. He writes (pp. 257–68) of Shakespeare’s Revisions of Act 5, accepting Wilson’s account of the mislined verse at 1798–1880 and counting as a second revision the substitution of Egeus for Philostrate and the hiving off of pieces of Theseus’s 1841–57 speech for Lysander. Ioppolo (1991, p. 113) had associated Sh.’s alleged revision of 1824–5 in Q1’s printer’s copy with the transfer of some of Theseus’s lines to Lysander in the F printer’s copy, but Holland (ed. 1994, p. 266) demurred.

The Date of Composition

The current consensus of scholarly opinion is that MND was written around 1595–96, during the same period as LLL, R2, and Rom.

The latest date for the composition and first performance of MND is set by the reference to it in Francis Meres’s Palladis Tamia (1598, f. 282r), which was entered in the Stationers’ Register 7 Sept. 1598. MND was entered in the Stationers’ Register 8 Oct. 1600, reaching print in the same year, when, according to the title page of Q1, it had been sundry times publickely acted, by the Right honourable, the Lord Chamberlaine his seruants. There are no such clear indications of the earliest date for composition or performance.

Efforts to determine an earliest date are guided by opinions (mostly impressionistic and subjective) of the development of Sh.’s style and artistry, and by attempts to link perceived allusions in the text to other works and to external events. The most popular area in the latter for the exercise of ingenuity is the identification of the wedding for which it is supposed, without evidence, that MND was composed, but various other circumstances have also claimed attention.

Style and Development

Chetwood (1750, pp. 12–13) exercises his best Endeavours to place the Dramatic Works of our Author, in the order of Time they were acted; but he gives no reasons for his order, dating MND 1595, 1600, 1610, and placing it ninth, after Rom. (fourth) and Tmp. (seventh). Griffith (1775, p. 2), while rejecting the notion which she believes prevalent that Tmp. and MND were the first and second of his writing, articulates the most common opinion as to why MND was written early in Sh.’s career, an opinion founded . . . on the idea, that his youthful imagination must naturally be thought to have been more sportive and exuberant, than his riper judgment might have permitted the indulgence of. Malone’s oft reprinted Attempt was influential not only in recognizing the prevalence of rhyme and the embellished style as marks of early work, but also in condemning the meagre and uninteresting fable and insignificance of the chief characters as attributable to Sh.’s genius being in its minority (in Steevens ed. 1778, 1:285–7; see also here, here). In 1778, and in 1785, this opinion led to his placing MND tenth in the canon, with Rom. at eleventh, both in 1595; Tmp. he placed next to last in 1612. However, in his own edition of 1790, believing the mourning muses (1849–50) to refer to Spenser’s poem The Tears of the Muses (1591) rather than to the poet’s death, he moved MND back to fourth, in 1592. The variorum edds. of 1793, 1803, and 1813 followed suit, but after Malone’s death the variorum of 1821 changed the date to 1594. Hurdis (1792, p. 17) finds Malone’s date of 1592 very reasonable but rates the play more highly, thinking it to have been the production of a judgment considerably matured and that there were undoubtedly many plays written before it. Tieck ([1793] 1796; in Bate, 1992, pp. 62, 565), while questioning Malone’s 17-year gap between MND and Tmp. (1595/1612) is nevertheless certain that the latter was written a great deal later than the former, for one might say that The Tempest is a lovelier and more perfect reprise of MND. Dibdin ([1797–1800], 3:29) adopts Malone’s 1778 chronology as generally admitted to be correct; though I cannot help confessing that I have seen no authority by which I am convinced that it is so.

Drake (1817, 2:261, 298–302) objects to Malone’s date of 1592 (in ed. 1790) because he considers it [2:298] a gross violation of probability to place three or four plays in the same year; he therefore dates MND 1593 (together with Rom., after Err. and LLL both in 1591, but before Shr., TGV, R3 and R2 in 1594–6). Though he repeats Malone’s strictures on the play (assigning them to Meres, either mistakenly or writing carelessly), he nevertheless refutes them vigorously in his subsequent analysis; (see, in part, here). Hallam (1839, 2:387, 390): MND’s superiority to [Err., TGV, and LLL] affords some presumption that it was written after them. But it evidently belongs to the earlier period of Shakspeare’s genius; poetical as we account it, more than dramatic, . . . [though not] from any deficiency of dramatic excellence. (2:390) Were I to judge by internal evidence, I should be inclined to date [Rom.] before MND. Knight (ed. 1839, pp. 331–2), presumably using Boswell (ed. 1821), believes that Malone’s date of 1594 has pretty exactly indicated the precise year. . . . But we entirely object to the reasons upon which Malone attempts to show that it was one of our author’s earliest attempts in comedy. He considers MND vastly superior to TGV, LLL, and Err. He answers Malone’s strictures point by point, concluding (p. 332): If any single composition were required to exhibit the power of the English language for the purposes of poetry, that composition would be MND. In 1849 (p. 39) he places MND in Sh.’s Second Period, 1594–1600, listed after KJ and before Rom.; in a second list (p. 40) he puts MND last of the Second early period comedies, 1589–93, and finally (p. 208) he repeats the opinion of his ed. 1839. Verplanck (ed. 1845, 2:6) believes the play, as it first appeared in print, must belong to a period about 1595, or 1596. While conceding that some stylistic features associated with the lovers deserve Malone’s strictures, yet in the other poetic scenes, the strain we hear is of a higher mood, and belongs to a period of fuller and more conscious power than the more juvenile comedies LLL and TGV. He therefore believes it was originally written in a very different form from that in which we now have it, several years before the date of the drama in its present shape. Hudson (ed. 1851, 2:257): The best conclusion we can form is, that the play was written somewhere between 1594 and 1598. Yet we have to concur with Mr. Verplanck, that there are some passages which relish strongly of an earlier period; . . . Perhaps, however, what seem the defects of [the part of the Athenian lovers], the far-fetched conceits and artificial elegances, were wisely designed, in order to invest the part with such an air of dreaminess and unreality as would better sort with the scope and spirit of the piece, . . . So that we cannot quite go along with the judicious critic last mentioned, in thinking the part in question to be the remains of a juvenile effort. Lloyd (in Singer (ed. 1856, 2:436): I cannot admit for a moment that this play exhibits the slightest signs of juvenility, as implying inferiority, as compared with [MV and H4]. Comparing it with [Rom.], I think there are some marks of a more perfectly developed taste, and of more free as well as skilful execution. Staunton (ed. 1857, 1:339): MND was written in the full vigour of Shakespeare’s youthful genius, and subsequent, there is every probability, to TGV, LLL, Err., Shr., and Rom. White (ed. 1858, 4:15–17) believes the play (4:16) produced, in part at least, at an earlier period than 1593, thinking such passages as 763–4 and 780–1 unworthy of the author even of LLL, TGV, Err., and Ven. Sh. (4:17) went from Stratford up to London with it partly written; [later] he reverted to his early production, and in 1594 worked it up into the form in which it was produced. It seems to me that, in spite of the silence of the quarto title-page on the subject, this might have been done, or at least that some additions might have been made to the play, for a performance at Court. Ulrici (1868–9; 1876, 1:223, 2:81): 1592 to 1597–98 . . . may . . . be termed the second period, or, so to say, the adolescence of Shakspeare’s genius. If we assume that during this time [R3, AWW, Rom., Shr., KJ, R2, MND, H4, and MV] were all brought to light in the above succession, . . . it seems astounding with what rapid, powerful, and safe steps Shakspeare proceeded through his career. (2:81): From internal evidence I am inclined to assume that 1596–97 was the year in which the piece was composed. For, in spirit and character, it agrees so entirely with the works belonging to the close of the second period of Shakspeare’s career that it would be difficult for any one to separate it from these. The great number of passages in rhyme, . . . as well as the many interspersed poems and songs are naturally explained by the lyrical character of the whole and by the subject of the conversations. Fleay’s (1874, pp. 10–16) attempts to apply scientific tests to versification to determine chronology, which led to his placing MND in 1592, second in the canon following LLL, elicited objections from his Shakspere Society audience as well as ridicule from others (cf. pp. 17–23, and Murphy (2003, pp. 210–11); on the uses and difficulties of metrical tests in determining chronology cf. Chambers, (1930, 1:255–69, 2:397–408). In 1876 (p. 26), echoing Malone’s opinion of poetry, plot and characterization, Fleay adheres to 1592, but in 1877 (p. 20) he lists MND c. 1593, claiming: This play as we have it, is a revised edition made for publication in 1600. It may have been added to, as well as revised; his later shifts to accommodate lunar or wedding theories are recorded below. In 1881 (pp. 50–1, 100), he attacks critics of his metrical tests; he presents his table for MND, describing it as (p. 100) Written 1592: revised as in Q 1, 2, 1600. Ward (1875, 1:380): The general character of the piece allows the supposition that it was written somewhere between 1593 and 1597; the abundance of rhymes and the paucity of feminine endings point to an early date; the construction of the play is likewise slight; yet there is an obvious growth of dramatic power beyond the very earliest period of Shakspere’s dramatic activity. Stokes (1878, pp. 53–4) believes MND contemporary with Rom. and should be dated about 1595. He agrees with Hallam that it is superior to LLL, Err., and TGV. Ebsworth (ed. Q1 1880, p. xi) rejects what are called verse tests, but remarks on the absence of light-ending or weak-ending and run-on lines, and comments: The continuity of rhyme . . . in Titania’s and Oberon’s speeches adds to their musical impressiveness; he dates MND probably 1593–94, at earliest; and not later than 1596. Finkenbrink (1884, pp. 4–20) dates MND about 1594 partly because of its similarities with LLL, Err., and TGV, which he dates 1590–93, and partly through (p. 17) the peculiarities of verse and metre, style and diction, specifying the great progress made over the earlier plays in the latter, defending the use of rhyme and of alliteration as appropriate dramatically, and praising the development in the structure of verse, and even in the use of classical allusion, characterization, and plot design. White (1886, p. 14): MND was written, or at least completed, some three or four years later than [LLL and Err.]. . . . [I]n its execution it shows, both in thought and in structure, and no less in poetical form, a marked mental development. Barnett (1887, p. 10): A critical examination of the play, and a comparison with others proves that M.N.D. is amongst Shakespeare’s earlier plays, and was most probably written between 1591 and 1593. Rolfe (1889, pp. 185–7): (p. 186) The internal evidence of style etc., is in favor of two dates, . . . though . . . I doubt whether the play was revised for a nuptial ceremony. In its present form it must be at least a year or two earlier than [MV] (which can hardly have been written before 1596 or 1597); and portions of it appear to be considerably earlier than the rest. He quotes 763–4 and 780–1 as instances of crudity. Chambers (ed. 1897, pp. 8–15): (p. 11) 1594–5 seems to me to suit admirably with the character and style of the play. It clearly belongs to the earliest group of Shakespeare’s comedies. It abounds with rhyme, with strained conceits, with antithesis and other rhetorical devices. The blank verse is far more regular and monotonous than that of any of the later plays; . . . Then, again, the interest of character is very slight. However, he considers MND betrays in many ways a notable advance over LLL, Err., and TGV, suggesting (p. 12) it is the last of that group, and that the chief advantage of dating it 1594–5 is that it brings it into closer neighbourhood to R2 and Rom. Craig (ed. 1903, pp. vi–vii) finds MND stylistically later than Err., LLL, and TGV, but earlier than MV. Ainger (1905, pp. 20–1) places both MND and Rom. in Sh’s first period together with LLL on stylistic grounds, but dates MND 1591–93 and Rom. 1595 or 1596. Brooke (1905, p. 1): MND belongs, probably, to the winter of 1595, . . . About four years before, in 1591, Shakespeare had written Rom. Cuningham (ed. 1905, p. xxii): It is noteworthy . . . that the play stands fifth in Meres’s list of comedies, affording some slight indication of Meres’s belief, knowledge, or recollection that it was not amongst the very earliest of Sh.’s plays. He believes the play was composed in the autumn of 1594–95. Furnivall (ed. 1908, p. vii): MND is his (probably) third comedy following LLL and Err. Structurally and metrically the comedy belongs to Shakspere’s early time of mistaken identity and cross purposes, of more than two sets of lovers, of ryme and of doggerel. Black (in Hudson & Black, ed. 1910, pp. xxvi–xxxi): (p. xxvii) The weight of evidence is in favor of 1594–1595. Placing it among the earliest plays (p. xxx) does not allow for the marked growth of dramatic and formative power, the imaginative insight, the spontaneity, and the mastery of expression both sympathetic and creative, which the play shows as compared with [LLL, Err., and TGV]. . . . [xxxi] It is possible that what seem defects and immaturities, the fanciful quirks and far-fetched conceits, and the seeming weakness and juvenility in characterization, were designed to invest the play with such an air of dreaminess and unreality as would better sort with its general scope and spirit; nevertheless the play is marked as early by the puns, etc., the prevalence of rhyme, and the monotonously regular blank verse, though the prose of the dialogue in the comic passages indicates growth and development. Hemingway (1911, pp. 78–80) believes (p. 79) the first version of [Rom.] appeared about 1591 and the first version of the Dream was written soon, perhaps immediately, after Rom., in 1592–3. He suggests that Sh. finds the emotionalism and sentimentalism of his tragedy . . . a trifle exaggerated and ridiculous, . . . and so, shaking himself free of romantic ideals of love, he somewhat quizzically allies lovers, lunatics, and poets; (p. 80) the Queen Mab speech was added to Rom. after MND was written. MacCracken et al. (1912, p. 151): Stylistic features argue an earlier date than students who notice only the skillful plot structure are willing to assign. Perhaps 1593–5 would indicate this variation in authorities. Armstrong (1913, p. 33): MND was probably the first play that followed the sonnets, and its first production is assigned to the winter of 1595. Cunliffe (in Brooke ed. 1914, p. 2) stylistically assigns it to 1594 or 1595: There are curious echoes from [Rom. to MND], as if Shakespeare had both in mind, or was actually engaged in writing both, at the same time. Rickert (1923, pp. 143–6) analyses metrical and other peculiarities reaching the conclusion (p. 144) that the greater part of the first three acts belongs to an early play revised, and the greater part of Act IV and practically all of Act V are later work. Through comparison with TGV and MV, she assigns a date of 1592 to the former and 1595 to the latter part. Therefore MND (p. 147) was begun early, finished hastily for a special purpose several years later, and later still [for public performance before 1598] revised to avoid possible offense. Robertson (1923, p. 13) considers MND’s versification so far developed that it might be doubted whether as it stands it can be even so early as 1594. Noble (1923, p. 52), believing that singing children are necessary in MND, associates the play with Wiv. and AYL: To my mind if it is maintained that [MND], as we have it now, was first produced in a public theatre, then 1595 is an impossibly early date; he prefers 1598. Thorndike (1929, p. 101): It was probably written at nearly the same time as Rom. Kittredge (ed. 1936, p. 229): So far as style and metre testify, the play might be dated anywhere from 1594 to 1596. Muir (1937, p. 77): MND was written immediately after [Rom.], though the New Cambridge editors (as everywhere) see evidence of subsequent revision. Holzknecht & McLure (ed. 1937, p. 197): Scholars generally date the play somewhat uncertainly 1594–8, with the majority leaning to an early date. Whether it preceded or followed [Rom.] is not known, though the parallels between the plays suggest that it came after. de la Mare (1940, pp. 296–305), noting that scholars have suggested dates of composition from 1590 to 1598, provisionally endorses the suggestion that the Dream, as we have it now, was composed at different times . . . ; or that, having been completed, it was redrafted and revised. He argues further from the lovers’ verse, (p. 300) odd little errors, varying speech headings, bad jokes and unusual vocabulary, that the play as we have it is primarily not Sh.’s. (P. 303): Surely, to accept as Shakespeare’s, at any age, what is provably not merely scamped or heedless but poverty-stricken verse . . . is more extravagant than to discredit its being his at [304] all? Brooke (1948, p. 35) believes Rom.’s Queen Mab speech preceded [MND] and contains the germ of that play. Schanzer (Midsummer, 1955, pp. 13–14) as evidence that Rom. preceded MND (both written in 1595), cites the reconciliation of the fathers (2135–6), a feature lacking in most source versions; writing with Rom. in mind, Sh. (p. 14) unwittingly added this touch to the traditional story. Brunner (1957, p. 66) believes MND belongs to the period 1594–6, a possible parody of Rom. in the interlude suggesting a later date for MND (Ger.). Munro (ed. 1957, p. 340) finds the versification, style, [and] plot . . . suited by a tentative date of 1595. Baldwin (1959, pp. 472–92): MND is the last probable representative of early work. (P. 477): The external fifth act parallels the structure of LLL. (P. 492): There are minor touches of staging, etc. which apparently came in later, but there was clearly no fundamental revision. Wells (ed. 1967, pp. 11–12): It is generally thought of as more mature, and therefore probably later, than four other comedies [TGV, Shr., Err., LLL]. . . . [MV] is reasonably thought of as later in date. Certainty would perhaps be most welcome as to whether [MND] came before or after [Rom.]. . . . [12] The richness and complexity of [Rom.] cause it to be more usually regarded as the later work. Waller (1966, pp. 4–6) tabulates selected linguistic features such as frequency of doth/does, hath/has; MND is placed with R2 in 1595–6, after Rom. in 1594–5. Fergusson (1977, p. 122): It is probable that [MND] is the closest in time to [MV]; perhaps it was written directly after it. Brooks (ed. 1979, pp. xlii–liii) presents stylistic evidence for dating the Dream c. 1595, in close proximity to plays which in style it resembles. He dates MV, 1 and 2H4, and Wiv. between 1596 and 1598, claiming that only the Belmont scene shows the lyricism characteristic of (pp. xliii–xlv) R2, Rom., and LLL, which he dates c. 1594–5. (P. xliii) I have found more parallels in the Dream with [LLL] and [Rom.] than with any other plays. Whether [Rom.] precedes or follows the Dream cannot be firmly determined, though there are signs that it is the earlier. (P. xlv): Stylistically, the group is linked not only by the lyricism from which the critics have named it . . . but also by its rhetoric, the art of which is not concealed but displayed. He proceeds to give examples of rhetorical figures from MND, with footnotes providing parallels in R2, Rom. and LLL. (P. liii): There is every reason to suppose that the formal rhetoric in the Dream was seen by Shakespeare and those for whom he wrote as the right partner for the lyricism. . . . Together, they characterize his style in the mid-1590s. Foakes (ed. 1984, pp. 1–4): The best evidence for dating this play remains . . . its nature and style, for it shares with a group of plays written about 1594–7 the mastery of lyrical drama achieved by Shakespeare in the mid 1590s. He lists LLL 1594–5, R2 1595, Rom. 1595–6, MND 1595–6, MV 1596–7. In all of these plays there is a conscious display of poetic and rhetorical skills and devices. Wells & Taylor (1987, pp. 118–19) date MND 1595. (P. 119): Most scholars would agree that, given the extraordinary parallels between them at every level of style and structure, Dream and Romeo were written at about the same time, though there is no consensus about which came first: . . . More generally, Dream belongs stylistically to a group of plays which includes Rom., LLL, and R2. Knutson (1991, pp. 60, 143, 196) places MND in the Chamberlain’s Men repertory in 1595–6, after Err., LLL and Rom., though also assigning its composition to 1594–5, when there was (p. 143) a flurry of interest in plays using magic. Riess & Williams (1992, pp. 214–18) argue that Rom. predates MND, and that in writing the latter, Sh. used the events and the language of tragedy to increase the mirth of comedy especially in the Interlude. Holland (ed. 1994, pp. 110–12): MND was written and first performed in 1595 or 1596. . . . Stylistically the play fits with other plays of the same date, especially Romeo and Juliet. Though he leans towards believing Rom. the earlier, [i]n the final analysis, all that matters is that the two plays were clearly being worked on at roughly the same moment. Kermode (2000, p. 52): Rom. is a kind of twin to MND; both of them are fairly securely dated 1595. Others pointing to affinities with the lyrical group of plays as best evidence of date include Kittredge & Ribner (ed. 1966, pp. ix–x), Traversi (1968, 1:109, 139), Bevington (ed. 1988, pp. 81–2), Halio (1996, pp. 155–6), and Greenblatt (ed. 1997, p. 805).

Too delightful to ignore is Bather (1887, pp. 69–75), who divides Sh.’s works into four periods, (p. 72) I, the period of apprenticeship; II, a period of manly vigour; III, a gloom-period; IV, a period of final calm. He orders the plays by percentage of puns, by which measure MND, at 0.53, comes 24th. Neither gloom nor calm seeming entirely suitable, Bather hedges (p. 75): Yet all critics, except Chalmers, give it a very early date, 1592–4. The play is however very different in character from any other play of Shakespeare’s. It is more in the style of a masque; it is a poem, and written in rhyme for the most part. The rhymes can be no argument for a very early date. In any case the play is so anomalous that we need not regard it as upsetting any of our results. Perhaps the eminent biologist should have stuck to fossil echinoderms.

Allusions to Other Writings and to Topical Events

Attempts to determine the play’s date through perceived (or imagined) references in the text to writings of other authors or to current events have been tied to particular lines or speeches, to the Pyramus and Thisbe interlude, or to the general impression of the play as a whole. Of particular lines the most frequently commented are 337–9, 376, 385, 463–489, 841–2, 1849–50 (see nn. 376–7, 384–5, 456–92, 841–2, 1849–50). On possible references to other writings in the interlude, see also here. Most voluminous is the debate arising from the effort to attach the play to a specific occasion, especially that of a noble wedding. As with the supposed unevenness of the style of the play, pursuing links to external references often necessitates positing rewriting, revisions, or additions.

Often taken as an important indicator is the possibility that Titania’s lament on destructive weather refers to actual conditions. Halliwell (1841, pp. 6–10): We suppose this play to have been written in the autumn of [1594], and we believe we can bring better evidence than has yet been adduced. He quotes Simon Forman’s observations on the rain and floods of summer 1594 (Ashm. 384), drawing attention not only to the bad weather, but to Forman’s mention of the plenty of small nuts, which may have suggested Titania’s offer to Bottom of new nuts (1549); in support he quotes Stowe and Churchyard. In 1855 (pp. 17–23; repeated in ed. 1856, 5:4–7) he adds Dr. King’s description of the weather (first noticed by Blakeway in Boswell ed. 1821, 5:342). Cf. n. 456–92. At this later date he slightly modifies his position, believing it first produced either towards the close of . . . 1594, or early in 1595. He also takes into account Greene’s death in 1592 and Nashe’s Greenes Funeralls (1594) probably supporting this date, but dismisses any reference to Spenser’s death (see n. 1849–50). In 1879 (pp. 6–7), influenced by the possible echo at 375–6 of Spenser’s FQ Bk 6 which was entered on the Stationers’ Register January 20, 1596, he suggests that MND must have been composed later that year.

Others taking the reports of 1594 bad weather as evidence to date MND at the end of 1594 or possibly 1595 or early 1596 include Collier (ed. 1842, 2:387–8), Verplanck (ed. 1845, 2:6), Delius (ed. 1859, 5:II [2]), Hudson (ed. 1851, 2:255–6), White (ed. 1858, 4:15–17) but only in the play’s revised form, and Fleay (1886, p. 126) who believes the date should be fixed in the winter of 1594–5 . . . the allusions to the remarkable weather of 1594 being too marked to be put aside contemptuously. Finkenbrink (1884, pp. 14–15) also finds the similarity too remarkable to be accidental, countering Wright’s (ed. 1877, p. vi) objection by claiming that the fair harvest mentioned by Stow refers only to local pockets of favorable weather. Morley (ed. 1886, pp. 5–7) finds the connection (p. 7) quite possible. Boas (1896, p. 182) sees a very probable reference. Chambers (ed. 1897, p. 10) avers the speech primarily suggests a date of 1594. Cuningham (ed. 1905, pp. xxii–xxvi) gives weight to this indicator for 1594–5. For Cunliffe (in Brooke ed. 1914, p. 2) it is a plausible conjecture. Chambers (1930, 1:246): Despite tempests being common phenomena an allusion is fairly plausible . . . to the rather unusual bad weather of 1594–5. Kittredge (ed. 1936, p. 229) calls the weather the only topical allusion that has any probability. Harrison (ed. 1937, p. 12) cites the weather as support for his estimate of composition between the autumn of 1594 and the spring of 1595. Alexander (1939, p. 105) concurs. Neilson & Hill (ed. 1942, p. 88): The speech reflects in all probability the weather of 1594. Thomas (1949, pp. 319–21) argues that the weather in 1596 was worse than 1594, and therefore the play belongs to the later year. Arnold (1955, p. 100; 1977, p. 93) cites Stow on the bad weather of 1594 (Fr.). Halliday (1961, pp. 120–1) suggests Sh. invented the fairy story of MND to cheer everyone up after the summer of 1596 repeated the bad weather of 1594.

Some are unconvinced. Dyce (ed. 1857, 1:clxiii) dismisses the suggestion as ridiculous. Ward (1875, 1:379–80) is dismissive of attempts to fix MND’s date by references to the weather, or to the death of learning. Wright (ed. 1877, pp. iv–vi) after quoting King and Stowe and citing Forman, points out that Stowe declares (p. vi) a faire haruest followed the bad summer weather of 1594, which contradicts Titania’s words and negates efforts to assign MND to 1594. I am even sceptical enough to think that Titania’s speech not only does not describe the events of the year 1594, but that it is purely the product of the poet’s own imagination. Barnett (1887, p. 10) dismisses such claims as suppositious. Craig (ed. 1903, p. vi): To lay much stress on such things as these in a climate like ours is, I think, injudicious. Wells (ed. 1967, p. 12) dismisses topical allusion to weather as a guide to date. Wells & Taylor (1987, p. 19): The allusion to unseasonable weather [456–92] would fit the second half of 1594, 1595, or 1596, and hence is of little value in narrowing the range of composition. Similarly, Bevington (ed. 1988, p. 81–2): The description of unruly weather [463–89] has been related to the bad summer of 1594, but complaints about the weather are perennial. Greenblatt (ed. 1997, p. 805) concludes that attempts to fix date by royal Progresses or bad weather have been defeated by the frequency of both; similarly Holland (ed. 1994, p. 111).

Meteorological evidence of date is also sought through references to the moon. Apart from a couple of unconvincing and contradictory suggestions by Fleay (1879, pp. 56–7), as reported in Robinson’s Epitome, and Sarrazin (1900, pp. 67–8; see below), the first detailed effort to use astrology to date the play is by Draper (1938, pp. 266–8). Citing 1084 and 1421, he computes that Venus as the morning star was at its brightest on (p. 267) May 2 according to the Julian calendar in 1595, when it was bright and very obvious . . . from the latter part of April into June, the only year between 1592 and 1598 when this was the case. Noting other references in the play to May Day (1653–4), to the new moon (12, 92), and to moonlight, sometimes faint (cf. 222–3, 434–5, 861–73), he calculates (p. 268) that there was an astronomical new moon on April 29, 1595, O. S.; and the thin crescent might be dimly visible on the following evening and more clearly on May first. He concludes that the play was written for an occasion on May 1, 1595. Cambillard (1939, pp. 118–26) is concerned with an esoteric theory of astrological myth rather than date of composition, but an editorial footnote on the first page assures the reader that nothing he argues contradicts Draper’s conclusions (Fr.). (Cf. Richer, 1974.) Wood (1966, pp. 128–30) argues that references to the position of the moon and the planet Venus agree with astrological data only for the year 1595 of the years between 1592 and 1598. Objections to Wood’s interpretations of the play’s references to the moon or moonshine, and to some of his astronomical findings are raised by Stevenson (1968, pp. 131–2) and Taylor (1971, pp. 134–6), leading to an exchange between Wood and Taylor (1971, pp. 464–5); no objection is raised to Wood’s initial speculation about date. The subject is revisited by Brown (1980, pp. 162–5), who feels that the astronomical references support the supposition (p. 164) that the play was written in 1595, probably in the Spring of the year, to be performed on or to celebrate a marriage on 1 May 1595, and designed so that specific reference is made to that date and to astronomical midsummer day of the previous year, 1594. Moreover, Sh.’s use of calendar symbolism may have been suggested by the appearance, early in 1595, of Spenser’s Amoretti and Epithalamion. He concludes that MND’s explicit and numerous astronomical references are intended to point Sh’s audience to associate the midsummer night of the title and the May Day mentioned in the text with specific dates in particular years, 11 June 1594 and 1 May 1595. For Hunter (1983, p. 96) and Honigmann (1985, p. 151) see below, here, here. The title of Wiles’s 1993 book Shakespeare’s Almanac: A Midsummer Night’s Dream, Marriage and the Elizabethan Calendar proclaims its main preoccupation, and Ch. 8, Weddings and Popular Astrology examines other poems and plays shaped by association with the almanac; the whole work is dedicated to connecting MND with 19 February 1596, the date of the Carey/Berkeley wedding (see further below). Holland (ed. 1994, p. 111) is dismissive of attempts to fix on the exact date of the first performance by . . . the play’s references to the phases of the moon.

The lines on the mourning muses lamenting the death of learning in beggary (see 1849–50 and n.) have also been taken as a guide to date of composition, which sometimes serves as an earliest date but is seldom considered precise. Warburton’s (ed. 1747) belief that the allusion is to Spenser’s The Teares of the Muses (in Complaints, 1591) was often taken as suggesting a date for MND close to that time, but Wright (ed. 1877, p. viii) represents the position now still held: I am inclined to think that Spenser’s poem may have suggested a title for the piece submitted to Theseus, and that we need not press for any closer parallel between them. Cf. Brooks (ed. 1979, p. 107). who believes Sh. may have had it in mind without necessarily intending a recognizable allusion.

Some have also sought a guide to date in the death of an actual poet (see n. 1849–50). Spenser was an early favorite, but when it was ascertained that he died in 1599, it was conceded that either the reference was not to him, or the lines were inserted between 1599 and 1600 (cf. Malone in Steevens ed. 1778, 1:288). Greene (d. 1592) was an even more popular candidate, but there is disagreement over whether the possible allusion necessarily means a date of composition close to the date of death. The suggestion of Marlowe (d. 1593) has not garnered much support. Tasso (d. 1595) is a late entrant, but his death date corresponds more attractively with the most widely accepted date of composition for MND.

The topical event now considered the most reliable indicator is the baptism of Prince Henry at the Scottish court in 1594 (see n. 841–2). Brooks (ed. 1979, pp. xxxiv–xxxv) relies solely on this incident to determine the terminus a quo, thinking it highly probable that Sh. knew of it. If so, . . . the Dream could not be earlier than the baptismal feast of Prince Henry, 30 August 1594, and probably not than the account of it in A True Reportarie, registered with the Stationers on 24 October. Among those who agree with finding this an acceptable indicator (besides Knight ed. 1839 and Rickert 1923, p. 67) are Chambers (ed. 1897, pp. 10–11, and 1930, 1:247), Cuningham (ed. 1905, p. xxxi), Black in Hudson & Black (ed. 1910, p. xxviii), Harrison (ed. 1937, p. 12 and ed. 1948, p. 269), Alexander (1939, pp. 105–6), Bednarz (1983, p. 82), Honigmann (1985, p. 153), Wells & Taylor (1987, p. 118); expressing reservations are Neilson & Hill (ed. 1942, p. 88) followed by Munro (ed. 1957, p. 339); entirely unconvinced is Foakes (ed. 1984, p. 125 n.). Furness (ed. 1895, p. 248) does not include this episode in his list of lines and allusions furnishing evidence of date, apparently endorsing Malone’s characterization of it as an odd coincidence.

Similarities between individual lines in MND and other literary works have also been advanced as indicators of date. Unfortunately these suggestions are often rendered dubious either by the commonplace nature of the thought expressed, or by mistaken or doubtful dating of the supposedly earlier work, or a date of publication that is at odds with other indications of MND’s date. Of the first kind are the echo of Spenser’s FQ 6.8.32 (see n. 376–7), and of The Wisdom of Dr. Dodypoll (see n. 384–5). The anonymous play is also misdated by for example Chalmers (1799, pp. 360–7), who calls several other writings in evidence for his preferred date of early 1598: Lodge’s Wits Miserie (see n. 1801), Gale’s Pyramus and Thisbe (see here), and a parliamentary report concerning (p. 367) the stealing away of men’s children without the assent of their parents. Lawrence (1920, p. 826) finds echoes of Dekker’s Hercules plays, of a lost play of Phaeton and of the university play Lingua to suggest a date of late 1597. Acheson (1922, pp. 188–93) argues through references to the bad weather of 1594, and through claimed links to Chapman’s Hymns to the Shadow of Night (1594) and to the second edition of Roydon’s Willobie his Avisa (1594) in 1596, that MND was revised in 1596 to incorporate satire of Roydon as Peter Quince and was published in 1600 to answer Roydon’s attack on him in the third edition of Willobie his Avisa. McCloskey (1931, pp. 389–91), arguing for a lost edition of The Arbour of Amorous Devices in 1594, claims that possible echoes of A poem of a Mayde forsaken (see n. 942–50) support a date of composition in 1595. More plausibly, Stokes (1878, p. 52) speculates that the reissue of North’s Plutarch in 1595 may have directed Sh.’s attention to it; the speculation gains credibility from the fact of the reissue being published by Richard Field, formerly of Stratford-on-Avon (see here).

Weddings Etc.

It would be agreeable to limit this section to the words of Holland (ed. 1994, p. 112): The wedding occasion theory appeals to critics who like the concept of a site-specific play, with fairies running through the noble house to bless the real wedding of members of the audience, and to those who wish to rescue the play from the clutches of the popular theatre audience. I fail to see the need to want either. However, it seems a necessary duty to trace the way conjecture has hardened into dogma.

Some critics have made valiant attempts to dispatch the hydra. Wells (ed. 1967, pp. 12–14): The suggestion has been offered that the play as we have it is a revision made for public performance, and even that Theseus and Hippolyta are stand-ins for the pair whose wedding is supposed to be celebrated. The belief that the wedding blessing of the last Act had [13] some extra-dramatic significance encourages a loose assumption that it is superfluous. . . .

[The play] is, certainly, much concerned with marriage; but so are many comedies. . . . [A]n allusion to [the Queen] does not imply that she was . . . present at the play’s first, or any other performance. . . . If Shakespeare’s company could at any time muster enough boys for public performances, we have no reason to doubt that it could have done so from the start. Thus the suggestion that the roles of the fairies were intended to be taken by children of the hypothetical noble house seems purely whimsical. The stage directions of the first edition . . . show no essential differences from [14] those in his other plays. . . . Furthermore, . . . the first play certainly known to have been written for such an occasion is Samuel Daniel’s Hymen’s Triumph [1614]. Price (1983, pp. 17–18) Many critics . . . assume . . . that it was written for performance at an aristocratic wedding celebration—without, perhaps, fully realising that this notion originated as a tentative conjecture by . . . Tieck, . . . and that it was much disputed in his own country and century. The principal difficulty is that it has never been possible to decide which particular wedding . . . the play was designed to celebrate. Nevertheless, the majority of recent critics concludes that . . . the play certainly owes its genesis to some such courtly occasion. . . .

[This] . . . has been disputed by [18] Stanley Wells, and Alfred Harbage [Harbage 1962, pp. 19–20], on the grounds of the complete lack of evidence that any play was written for a special private performance before 1614, and that the costs, time and effort involved in writing, rehearsing and performing a new play exclusively for such an audience would not have been worth the company’s while. Williams (Moonlight, 1997, pp. 1–18, 263–5) provides an extensive review of The Wedding-play Myth and some of the arguments against it.

Despite such herculean efforts, wedding theories and their proponents keep raising their heads, even to the point that some well-reputed scholars continue to present the theory as undisputed fact; cf., for example, Bloom (1998, pp. 148, 152–3), Duncan-Jones (2001, pp. 10, 87–9.).

Following is a list of the weddings proposed, with a short form, and the name of the first proposer:

  • 1.Robert Devereux Earl of Essex and Frances Lady Sidney, April or May 1590; Essex/Sidney; Elze 1868; 1874.
  • 2.Robert Carey, later Earl of Monmouth and Elizabeth Trevannion, 1592; Carey/Trevannion; Fleay 1876.
  • 3.Sir Thomas Heneage and Mary Countess of Southampton, 2 May 1594; Heneage/Southampton; Sarrazin 1900.
  • 4.Henry Percy Earl of Northumberland and Lady Dorothy Devereux, May 1595(?); Percy/Devereux; Draper 1972.
  • 5.William Stanley Earl of Derby and Elizabeth Vere, 26(?) January 1595; Stanley/Vere; Fleay 1876.
  • 6.Edward Russell Earl of Bedford and Lucy Harrington, 1595; Russell/Harrington; Fleay 1876.
  • 7.Thomas Berkeley, Lord Hunsdon-to-be and Elizabeth Carey, 19 February 1596; Berkeley/Carey; Chambers, Occasion, 1916.
  • 8.Henry Guildford and William Petre [or Peter, or Petrie] to Lady Elizabeth and Lady Katherine Somerset, daughters of the Earl of Worcester, 8 November 1596; Guildford/Petre/Somerset (Spenser’s Prothalamion.); Martin 1935.
  • 9.Henry Wriothesley Earl of Southampton and Elizabeth Vernon, February or August 1598; Southampton/Vernon; Tieck (in Schlegel 1830).
  • 10.Roger Manners Earl of Rutland and Elizabeth Sidney, 1599; Manners/Sidney; Acheson 1922.
  • 11.Henry Herbert and Anne Russell, 16 June 1600; Herbert/Russell; Lawrence 1922, see also Chambers Occasion, 1916.

Tieck (in Schlegel’s Shakespeare’s dramatische Werke, 1830, 3:353; tr. Furness, ed. 1895, pp. 259–60): In 1598, the friend of the poet, the Earl of Southampton, espoused his beloved Mistress Vernon, to whom he had been long betrothed. Perhaps the germ, or the first sketch, of the drama was a felicitation to the newly-married pair, in the shape of a so-called Mask, in which Oberon, Titania, and their fairies wished and prophesied health and happiness to the bridal couple. The comic antistrophe, the scene with the rude mechan-[260]icals, formed what was termed the anti-mask. . . . Thus to this Occasional Poem there were added subsequently the other scenes of the comedy.

Ulrici (1839; 1846, p. 275): Tieck . . . conjectures, I think without reason, that the piece did not receive its present form before 1600, when it was first printed. It is not easy to see how the title of Midsummer Night’s Dream . . . could ever have suited The Mask of Oberon and Titania, with its anti-mask—the play of the artizans—in short, a mere piece composed for a marriage festival. Nevertheless, Tieck’s suggestion is supported by Mézières (1860, pp. 432–3), who believes it plausible given the practices of the nobility that MND was commissioned for wedding festivities, and asks what could be more natural than to attach Sh.’s work to an event [the marriage of the poet’s best friend] which must have touched him so keenly? (Fr.); and Massey (1866, p. 481) who has no doubts, but adds: The play was probably composed some time before the marriage took place, at a period when it may have been thought the Queen’s consent could be obtained, but not so early as the commentators have imagined. I have ventured the date of 1595.

Tieck had opened the door, and the first to respond with a counterproposal is Elze (1868; tr. 1874, pp. 40–60): All indications point to the fact that the Midsummer Night’s Dream was written for and performed at the marriage of the Earl of Essex in the year 1590. . . . [MND] is evidently the production of that happy period of life when fancy is most lively and unrestrained in its creations; everything in it is lyrical effusion, unclouded cheerfulness, exempt from reflection; in a word, all is youth. He traces the interrelationships of the Essex and Sidney families, the connections with Leicester and with Sh., and endorses Halpin’s explication of Oberon’s Vision (see here). Ulrici in his third edition (1868–9; tr. 1876, 2:82–3) takes issue with Elze’s proposal, objecting that the date is unlikely on the grounds of style, and that since both the Southampton and Essex marriages took place clandestinely the queen could not have been present. He rejects Elze’s interpretation of the concluding remarks, and Halpin’s (2:83) frosty and forced allegory, and reasserts that the connection between MND and the so-called masques . . . seems to me . . . to be a very distant one, from which little or nothing can be deduced in favour of the marriage-hypothesis. In 1874 in an addendum to the MND essay as revised for Schmitz’s translation (pp. 61–6), Elze dismisses Ulrici’s recent criticism of his hypothesis (see above) and expands upon the support he has received from Kurz (1869, pp. 268 ff.). Kurz gives as a reason for believing that the festive occasion at which MND was performed must have been before 1591 that the play could not have appeared after Spenser’s FQ which in that year had presented an idealized Queen Elizabeth (p. 278): After that could Shakespeare let his fairy queen, albeit called Titania and the spouse of Oberon, fall in love with an ass? A question not to be lightly tossed aside. This translation of Kurz is from Furness (ed. 1895, pp. 261–2), who warns that a vein of quiet humour running through his Essay . . . makes it difficult to say whether or not he is anywhere really in earnest; Furness summarizes (p. 262): In short, Kurz reaches the positive conclusion (p. 289) that [MND] was performed, for the first time, at a banquet on the occasion of the unheralded festivities accompanying the marriage of Essex, and in conjunction with the observances of May in 1590, as a masque with significant characters, or as a masque-like comedy with a masque especially introduced, and all of it designed to conceal the object for which the festivities were given.

Dowden (1875, p. 67) confidently asserts that MND was written on the occasion of the marriage of some noble couple but offers no new supposition, satisfied with the possibility of either Southampton/Vernon or the earlier date of the Essex/Sidney wedding. In the introduction to MND in Craig (ed. 1911–12, 2:246), the wedding theory (attributed to Dowden) is doubted because no evidence is forthcoming to support any of the weddings proposed. Neil (ed. 1878, p. 34) does not claim a wedding occasion, but promotes the notion that MND was composed as a masque for Court at Christmas 1590, and after a run of a winter or two, it was revised about 1596, and reintroduced to the stage in a more developed form.

Fleay (1886, pp. 18–19, 26, 126, 181–6): MND was produced in its first form c. June 1592; he believes it was produced while Sh. (p. 19) worked as a journeyman or with a coadjutor and was later revised. (P. 26): On 26th January 1594–5, [MND] was, I conjecture, acted at Greenwich at the marriage of W. Stanley, Earl of Derby, and afterwards on the public stage; it was evidently written for a marriage, but . . . had been altered for this special occasion. Its original production was probably in 1592, at the marriage of Robert Carey, afterwards Earl of Monmouth. (P. 181): 1595. January 26 was the date of the marriage of William Stanley, Earl of Derby, . . . . About the same time Edward Russel, Earl of Bedford, married Lucy Harrington. Both marriages may have been enlivened by this performance. . . . [182] The date of the Court performance must be in the winter of 1594–5. But the traces of the play having been altered from a version for the stage are numerous. (1891, 2:194): [T]he weather description [463–93], which can be omitted without in any way affecting the progress of the play, . . . I believe . . . . was inserted for the Court performance in 1596, that on the public stage having taken place in 1595; but . . . the marriage presentation, being subsequent to this, was most likely at the union of Southampton and Elizabeth Vernon in 1598–9. See also Krauss (1876, p. 241).

Initial reactions were at best skeptical. Wright (ed. 1877, pp. ix–xi), prefacing his remarks by saying that in seeking a specific occasion we embark upon a wide sea of conjecture, with neither star nor compass to guide us, discusses skeptically the theories of Massey, Elze, and Kurz, concluding (p. xi): In such questions it would be well to remember the maxim of the ancient rabbis, Teach thy tongue to say, I do not know. Brink (1877–9, p. 55) is inclined to believe in a wedding occasion, but without speculating on which. Stokes (1878, p. 48): It has frequently been suggested that [MND] was composed to grace some marriage festivities; and the supposition has been supported by referring to its lyrical and almost operatic tone, to its masque-like form, and to Oberon’s song at the conclusion. But this suggestion may perhaps be answered by noting the difficulty that has been experienced in finding any nuptial event to tally with the supposed date of its composition, by the unlikelihood of so unique an undertaking on Shakespeare’s part being unrecorded, by the inappropriateness of such phrases as Bottom’s statement in [quotes 960–1], and by the promise in Puck’s epilogue that we will make amends ere long. Halliwell (1879, pp. 7–8) endorses Stokes, calling such conjectures (p. 8) gratuitous and silly. Finkenbrink (1884, pp. 4–14) argues forcefully against wedding theorists, (pp. 4–11) engaging particularly with Kreyssig (Vorlesungen über Shakspere, Berlin, 1880, 3:81 ff.), Massey (1866), Elze (1868), and Kurz, (1869). He dismisses the claim that MND is a masque (see here), or occasional play. Wedding conjectures (p. 11) are to be rejected, and withal that by which they are principally suggested, viz. that our comedy is a mask. See also Rolfe (1889, p. 186) above. Gollancz (ed. 1894, pp. v–vii) mentions that the Southampton/Vernon and the Essex/Sidney marriages have been proposed as the occasion of the play; (p. vi) there is, however, absolutely no authority for the statement, and the probabilities are strongly opposed to the supposition. However, in the highly fanciful introduction to the 1895 ed. with illustrations by Robert Anning Bell, Gollancz (p. xxxv) imagines that Sh., circa 1594–5, while working on Rom., was invited to write a new play to grace some grand wedding. From approximately this time acceptance of some form of wedding theory becomes increasingly common, though dissenters are still to be found, such as Craig (ed. 1903, p. vi) who cannot find any evidence . . . of the least value for wedding claims; more dissenting voices are again raised in the later 20th c., especially from Wells’s 1967 ed. (see above) on.

Chambers (ed. 1897, pp. 12–14, 179–80) seeks to determine the precise occasion [13] on which it was first presented. He approves Fleay’s suggestion of the Stanley/Vere wedding, 26 January 1595 (p. 14): Lord Derby, like all the Stanleys, was interested in the drama . . . , and it is worth noting that the very company to which Shakespeare belonged had been . . . the servants of his elder brother and predecessor, Ferdinando. He devotes a brief appendix to detailing William Stanley’s career and associations with plays. Lee (1898, p. 161) supports either Russell/Harrington or Stanley/Vere. Sarrazin (1900, pp. 67–71) believes that MND was written for a noble wedding on 2 May 1594 (i.e. the Heneage/Southampton marriage). He engages with the dating through the moon references, concluding that the poet may not have been overly strict with the calendar and that possibly new moon refers to the months rather than the heavenly body. On 2 or 4 May 1594, the actors would have played under a waning moon, which agrees with Theseus’s exclamation 6–7. He thinks it obvious that the scenery and setting refer to a night performance at a great house or castle, at which the Queen was present. He argues against claims for the Stanley/Vere wedding (Ger.).

Cuningham (ed. 1905, pp. xxix–xxxi) It is not improbable that it was, at least eventually, intended for the celebration of the marriage of some nobleman of Elizabeth’s court; but I rather incline to the belief that it was not so in the first instance; and that, marriage or no marriage, we should have had [MND], though, perhaps, not exactly in its present form. He discusses and tentatively approves Fleay’s claim that the play was performed on 26 January 1595 at the Stanley/Vere wedding. So too Rothschild (1906, pp. 82–3). Furnivall (ed. 1908, p. ix) demurs, believing that if the play had been performed at the Stanley/Vere wedding, Stowe (Annales, p. 1279) would have taken note of it.

Among those at this period prepared to accept the idea of a wedding occasion but without believing it can be identified, and so provide a firm date of composition, are Baker (1907, pp. 182–6), Black (in Hudson & Black, ed. 1910, pp. xxix–xxxi), Gordon (ed. 1910, p.xxvii) and Herford (1912, p. 31). Armstrong (1913, p. 33) would fain believe it was written for the Russell/Harrington match.

Chambers (Occasion, 1916, pp. 154–60), though he reviews and expands arguments for the Stanley/Vere wedding, believes the more likely occasion is the Berkeley/Carey wedding of February 1596, an opinion that he is the first to propose and that has gained significant support in recent decades. Nevertheless (1923, 1:214 n. 2, 4:109), he lists the Stanley/Vere wedding as an occasion at which Elizabeth might have been present. Later (1930, 1:358–62) he returns to the wedding question, reviewing six proposed, and again giving his judgment that the Berkeley/Carey wedding is more likely than the Stanley/Vere. Durham (ed. 1918, p. 89): Some have attempted, without conspicuous success, to determine whose wedding MND was written to honor. Lawrence (1920, p. 826) argues against MND having been commissioned for a noble wedding: This idea of the acceptance of a commission outside the playhouse by a workaday actor-dramatist is highly questionable. There was absolutely no precedent for such a course. Nevertheless, later (1922, pp. 836–40), still maintaining late 1597 as the original date of composition, he argues that MND was extensively revised for the Herbert/Russell wedding in 1600. Still later (1927, p. 138) he dismisses arguments for some noble wedding about the year 1595 as contemptible guesswork. Mathew (1922, p. 122) asserts that MND celebrated the Russell/Harrington wedding on 12 Dec. 1594 and was performed for the Queen at Greenwich later that month.

Acheson (1922, pp. 186–97) dismisses (p. 187) arguments for the 1590 Essex/Sidney and Southampton/Vernon matches as mere guesses that show lamentable ignorance of easily ascertainable historical facts. Despite possible later performances at weddings (p. 188) and however it may have been later revised, I am satisfied that it was first produced for the occasion of the marriage of Sir Thomas Heneage to Lady Southampton, mother of Shakespeare’s friend and patron, in the year 1594. He claims that this date not only coincides with the strongest internal evidence advanced by past commentators, but gives us also a bridal couple that match the advanced ages and social dignity of Theseus and Hippolyta. He gives arguments for inferring that (p. 199) the Queen graced this marriage with her presence. He sees further revisions to MND when (as he surmises, p. 195) the services of Shakespeare and his company were retained for the marriage festivities of the Earl of Rutland’s proposed marriage in the spring of [1599] and not long after the death of Spenser. So also Stopes (1922, p. 75).

Wilson (ed. 1924, pp. ix–x, 85–100) and his co-editor Quiller-Couch are convinced that MND at whatever date written, . . . was composed to celebrate a marriage—possibly for private performance at some great house, possibly even at Court, but most certainly for a wedding somewhere. Their suggestions for date(s) and occasion(s) are influenced by their belief that the play underwent several stages of revision. (P. 99): It has, of course, long been recognised that [MND] is a wedding play, while its length—it . . . is the fourth shortest play in the canon—suggests that it was intended primarily for a private rather than a public performance. . . . For instance, the marriage of the Earl of Essex in 1590 . . . was proposed by Elze and supported by Kurz. This date is not now seriously entertained in any quarter, though, if parts of the text go back to some year before 1592, it cannot be dismissed as impossible. Curiously enough, no one, as far as we know, has quoted a wedding from 1592 or 1593 to fall in with the allusion to Robert Greene’s death, and it may be that in his first draft Shakespeare had no particular wedding in view; certainly that draft did not contain the wedding-masque with which the transmitted text concludes. Nevertheless, even without the masque, the fable, as Dr Johnson would say, is so appropriate to a wedding-celebration, that it is hard to believe the play was not originally plotted to that end. And if so the revision of 1594 is likely to have been undertaken for a similar purpose. Now Jan. 26, 1595, the date of the marriage of William Stanley, Earl of Derby, to Elizabeth Vere, daughter of the Earl of Oxford, fits in very well with the 1594 allusions, and since Fleay first put it forward this match has been commonly regarded by critics as the most likely occasion for the composition of the play. We have nothing better to offer for the revision of 1594, but we do not ourselves favour the Stanley [100] wedding, since the Queen is reputed to have been present at it, and we very much doubt whether the Dream, in any form, can have been played before her.

Finally, there is the marriage of the Earl of Southampton in 1598 to Elizabeth Vernon, . . . an occasion which . . . has of late found few supporters. Indeed an apparently insuperable obstacle has hitherto stood in the way of its acceptance, namely the patent absurdity, as it would seem to anyone in the least acquainted with the development of Shakespeare’s powers and style, of supposing that he could be writing this play, as a whole, so late as the year 1598. But this obstacle vanishes directly the fact of revision be admitted. It seems to us, therefore, at least possible that Shakespeare undertook the last revision, to which we owe nearly all the finest poetry of the play, in celebration of his friend and patron’s marriage. Some chronological difficulties attached to this supposition are not regarded as fatal to the possibility. The editors hope to have established a presumption in favour of [MND] having been first handled by Shakespeare in 1592 or before, rehandled in 1594, and rehandled once again in 1598. Empson (1994, pp. 198–201), in a posthumously published essay, is influenced by Wilson’s theories, considering MND first drafted in the period of plague closures in 1592–4 for the entertainment of Southampton, revised after 1594 probably for the Berkeley/Carey wedding, and revised yet again for public performance.

Chambers (Rev., 1925, pp. 341–2) reviewing Wilson (ed. 1