University of Virginia Library

Some Editorial Principles (with special reference to Henry V)
by
Alice Walker

In what follows my object is to demonstrate how compositor analysis contributes towards the solution of editorial problems in Shakespeare. I use the Folio text of Henry V for illustration, partly because it is rightly held to be one of the most reliable Folio texts and partly because it was the work of the two Jaggard compositors, A and B, whose habits are of considerable importance to editors. I do not propose to suggest which readings require emendation, except by way of illustration, or even to consider the general problem of the transmission of this text, which is the subject of another article in this volume: my aim is merely to show how answers to a few important questions may be found.

It would be agreed, I think, that the most crucial question today is the amount and kind of corruption we may expect to find in good substantive texts. Clearly one cannot generalise, as it varies from play to play and depends on two things: (1) the character of the copy, and (2) the kind of care taken in its printing. Compositors' errors thus represent only a fraction of the evidence an editor needs to take into account in formulating the principles on which he must work, but I shall concern myself with the habits of compositors because emphasis on the printing of Shakespeare's texts is urgently needed in order to correct some of the false premises of present-day editing, often resting on past fallacies rather than present realities.

Earlier in the present century when the first gigantic strides were made towards recognition of the kind of copy from which early good quarto and Folio texts were printed, the immediate reaction was one of confidence. It seemed as if, given reasonable care over proof-correction (and this was assumed), there could not be much wrong with good substantive texts which familiarity with the secretary hand might not put right, since even casual proof-reading with copy would have provided a safeguard against omissions and interpolations, anticipations and recollections detrimental to sense

and 'absolute numbers.' The assumption that proof was read with reasonable care was the keystone of conservative editorial theory, but it has recently been badly shaken by what Hinman[1] has so far published about Jaggard's proof-reading of the Folio. If Jaggard's preoccupation, throughout the Folio, was with typographical blemishes and not with fidelity to copy, then the character of the compositors who set the type is as important as the character of the copy itself. In recent years, recognition of the need to know all that can be discovered about compositors, press-work, and proof-reading has, in fact, undermined in some ways the textual theory of the pre-war years.

In the last resort, an editor must be prepared to take responsibility for the readings of his text: for seeing that the emendations he makes are consistent with the kind of transmission he postulates and for seeing too that he does not flightily emend one kind of error in one place and then jib at emending its like in another. There may be cruxes which defy emendation and there will certainly be doubtful readings where it is difficult to know what to do for the best. But there is, in every text, a nucleus of readings which are certainly wrong and it is from these that an editor must develop the principles on which he emends other readings. Disregarding errors in foreign names and foreign words, requiring an apprehension of the matter which Jaggard's compositors plainly did not possess, there are just over 50 such readings in the dialogue of Henry V emended by the Old Cambridge and recent English editors (Dover Wilson, Alexander, Walter, and Sisson). When these are referred to the stints of the two compositors who set this play, it is clear what types of error an editor should be on the watch for and where the greater number of one type of error or another is likely to be. Spelling evidence shows that Compositor A set pp. 69-71a, 75b-87, 90-91, 92b-95 (I.i.-I.ii.135, II.ii.180-IV.iv.43, IV.vii.143-V.i.34, V.ii.12 ff.); Compositor B set up pp. 71b-75a, 88-89, 92a (I.ii.136-II.ii.179, IV.iv.44-IV.vii.142, V.i.34-V.ii.11).[2] The unanimously rejected readings are as follows:

Compositor A's stints

                 

p.71	I.ii.131	(1)	Bloods	for blood F3
75b	II.iii.16	(2)	Table	babbled Theobald
	II.iii.49	(3)	world	word Q1,3, Rowe
77	III.Ch.4	(4)	Douer	Hampton Theobald
	III.Ch.6	(5)	fayning	fanning Rowe
	III.i.7	(6)	commune	summon Rowe; conjure Walter
	III.i.17	(7)	Noblish	noblest F2
	III.i.24	(8)	me	men F4
	III.i.32	(9)	straying	straining Rowe

                                       

79	III.iii.32	(10)	headly	heady F2
	III.iii.35	(11)	Desire	Defile Rowe
80	III.v.46	(12)	Kings	knights Theobald conj., Pope ii
81	III.vi.107	(13)	Leuitie	lenity Qq, Rowe
	III.vii.12	(14)	postures	pasterns F2
83	IV.Ch.16	(15)	nam'd	name Tyrwhitt conj., Steevens
84	IV.i.94	(16)	Iohn	Thomas Theobald conj., Pope ii
85	IV.i.241	(17)	Odoration	adoration F2
	IV.i.271	(18)	Hiperio	Hyperion F2
	IV.i.287	(19)	reckning of	reckoning if tyrwhitt conj., Steevens; . . . or anon conj., D.W.
	IV.i.304	(20)	friend	friends Qq, Theobald
86	IV.iii.11-4	(21)	Line order wrong	Trans. Thirlby conj., Theobald
	IV.iii.48	(22)	Line omitted	Qq, Malone
91	IV.viii.111	(23)	me	we F2
	V.Ch.10	(24)	Wiues	with wives F2
92b	V.ii.12	(25)	Ireland	England F2
	V.ii.50	(26)	withall	all Rowe
93	V.ii.77	(27)	curselarie	cursorary Q3, Pope; cursitory D.W.
94	V.ii.316	(28)	hath	hath never Rowe
95	V.ii.356	(29)	pation	paction Theobald

Compositor B's stints[3]

                                   

71b	I.ii.163	(1)	their	her Johns. conj., Capell; your Qq
72	I.ii.212	(2)	And	End Qq, Pope
73	II.Ch.32	(3)	force a play	Corrupt
	II.i.23	(4)	name	mare Qq, Theobald
	II.i.77	(5)	enough to	enough Qq, Pope
	II.i.99	(6)	Coporall	Corporal F2
74	II.i.113	(7)	that	that's F2
	II.ii.87	(8)	furnish	furnish him F2
	II.ii.107	(9)	an naturall	a natural F2
75a	II.ii.139	(10)	make	mark Theobald
	ibid.	(11)	thee	the Pope
	II.ii.147	(12)	Thomas	Henry Qq, Theobald
	II.ii.148	(13)	Marsham	Masham Qq, Rowe
	II.ii.159	(14)	Which	Which I F2
	II.ii.176	(15)	you	you have Qq, Knight; you three F2
88	IV.v.11	(16)	dye in	die in honour (Qq); . . . harness D.W.; . . . arms Walter
	IV.v.15	(17)	a base slaue	by a slave Qq, Pope
	IV.vi.34	(18)	mixtfull	mistful Warburton conj., Theobald

         

89	IV.vii.75	(19)	with	their Pope
	IV.vii.107	(20)	Countrymen	countryman Qq, F2
	IV.vii.112	(21)	Good	God Qq, F3
92a	V.i.64	(22)	began	begun Capell
	V.i.83	(23)	swore	swear Qq, F3

A's stints thus contain 29 certain errors in 20 pages and B's 23 in 6 1/2. I shall now analyse the errors, including in the category 'misreadings' all that might have been due to misreading of copy but querying those which might have been due to some other kind of error:

Compositor A's stints

• Misreadings: nos. 1?, 2, 5, 6?, 8?, 9?, 11, 13?, 14, 15, 17?, 18?, 20?, 23?, 27, 29
• Letters added: 3, 10
• Words, phrases etc. omitted: 22, 24, 28
• Words interpolated: 26
• Errors due to copy: 4, 16, 21, 25[4]
• Repetitions: 19
• Assimilations: 7
• Misinterpretation: 12[5]

Compositor B's stints

• Misreadings: 1?, 2?, 4, 20?, 21?, 22?, 23?
• Letters omitted: 6, 7, 10
• Letters added: 9, 11, 13
• Words, phrases etc. omitted: 8, 14, 15, 16
• Words interpolated: 5
• Errors due to copy: 3?
• Repetitions: 17
• Anticipations: 12, 19
• Inexplicable: 18[6]

It is evident that in A's stints there are certainly 7 instances of misreading (nos. 2, 5, 11, 14, 15, 27, 29[7] ). Among the readings I have queried as possibly due to other causes, no. 1 might have occurred through assimilation to 'sword' ('blood and sword'); no. 6 presents a problem but, as I have no doubt that Walter's 'conjure'[8] will have wide appeal, it should, I judge, be

numbered at least provisionally among misreadings, bringing the total to 8; nos. 8, 9, 18, 20, may have been due to the accidental omission of a letter (especially in view of the graver omissions, nos. 22, 24, 28); no. 13 may be a turned 'n'; no. 17 might have occurred through assimilation to the following 'o' (cf. no. 7), and no. 23 ('Come, goe me' for 'Come, goe we') certainly looks like an error of this kind. The number of misreadings in A's stints therefore lies between (say) 8 and 16.

The only certain misreading in B's stints is no. 4, though no. 1 may be an error of this kind; no. 2 might have been a misreading (assuming the line began with a minuscule), but it looks more like a typical error of anticipation ('End in one purpose, and . . .'), like nos. 12, 19; and it would seem unrealistic to regard nos. 20-23 as anything but typical perversions, such as abound in B's work in 1 Henry IV, in the absence of any certain misreading in B's stints except (perhaps) no. 1 and (certainly) no. 4. The number of misreadings in B's stints thus lies between (say) 2 and 7. This means that between 10 and 23 out of 52 certain errors were misreadings—less than half, even if we unrealistically accept the higher figure.

These two lists of indubitable errors need to be related to the conclusions reached by Mr. Cairncross elsewhere in this volume, and it is not my purpose to discuss them in detail or in relation to the question of copy except to remark that Compositor A had the lion's share of the setting from manuscript and B from corrected quarto[9]. The point I want to make is the general one that a start from the indubitable errors, analysed on a compositor basis, makes it easier to judge which of the many emendations that have been made, or suggested, is warranted on the evidence; and, as one of the main tasks of an editor today is that of separating the grain from the chaff in three centuries of critical opinion, it is important to have some systematic means of winnowing. I am not suggesting that a survey such as I have made need be applied rigidly. There may be types of error frequently made by a given compositor which are not clearly exemplified in the manifest errors of every text; further, my survey gives only an inkling (A, no. 3) of the certainty of contamination from Q2 and Q3 opened up by Mr. Cairncross's conclusions. But in spite of inevitable limitations, these lists provide four useful pieces of information: (1) that the ductus literarum as the solution to textual difficulties represents half measures at the best even when the copy was difficult manuscript[10]; (2) that misreading may be more appropriately invoked in explanation of errors in A's stints than in B's; (3) that errors in B's stints may be twice as numerous as in A's; and

(4) that B's errors may include a very high proportion due to the kind of carelessness exemplified in his stints in I Henry IV.[10a]

This nucleus of certain errors touches, of course, only such corruptions as would seem obvious to anyone attentively following the matter. This is clear from the fact that (as the lists reveal) most of the errors of this kind in Henry V had been corrected by the mid-eighteenth century. But, starting from this nucleus, it is easier for an editor to test past opinion as well as his own. Thus, at II.iv.107, the emendation of 'priuy' to 'pining' is warranted on the evidence of difficulty with minims[11]; the evidence of no. 19 warrants the assumption that 'by' at V.Ch.29 was A's interpolation due to repetition; and the evidence of no. 23 suggests that 'for' for 'or' at IV.iv.14 (clearly wanted, since Pistol's threat is 'death or ransom') was an error of assimilation to the preceding 'fortie'. Compositor B's 'Maiesties surueyes' (I.ii.197) is on a par with no. 9 (the addition of a final 's' was, in any case, one of B's most pernicious habits[12]) and 'come' for 'came' at II.i.114 looks like a characteristic error of anticipation. These are among the doubtful readings which some, but not all, of the five editions emend. There are 36 such readings in all: 22 of them in A's stints (6 or 7 attributable to misreading), 14 in B's (of which only 1 is explicable as a misreading) and most of the emendations are in accordance with what the indubitable errors suggest about likely types of error in their stints. There might have been greater uniformity of opinion over these doubtful readings (which are by no means all that are questionable) had the ductus literarum not proved a stumbling block.

A further point worth remark concerning the indubitable errors in this text is that on at least a dozen occasions the Quarto preserved the truth where the better text failed. Most (though not all) of the Folio errors which can be set right by reference to the Quarto are trivial and could have been corrected with certainty in the absence of a collateral text. All the same, it is not reasonable to suppose that chance will have operated so that a good text and a bad will have corrupted the same readings, and because Pope was too partial to the Quarto is no reason for flying to the other extreme by trying to avoid its aid. There is no reason for rejecting two Quarto readings,

followed by the (Old) Cambridge editors at IV.vi.15—'And cries aloud "Tarry, dear cousin Suffolk"'. The Folio's substitution of 'He' and 'my' for the italicised words is entirely characteristic of Compositor B, making first an error of repetition and then one of anticipation. So too is the substitution of 'black and white' (II.ii.104) for 'black from white', the Quarto reading and a better complement to 'stands off'. The Folio reading is again Compositor B's, who presumably anticipated 'and' at the beginning of the following line ('Treason and murder'). Fear of bad texts is only justified if we can be quite certain that there were no errors in either the copy or the printing of the more authoritative text. Provided that an editor can explain how the corruption came about (and B's certain errors justify our assuming similar errors in the above two readings) it would be dangerous, in view of the superficial character of Jaggard's proof correction, not to accept the better readings.

In an article in the last volume of Studies in Bibliography [12a] I argued the need for eclecticism in the editing of collateral substantive texts with the reservation that eclectic editing today was something very different from that of the eighteenth century. Under the impression that there was grave corruption of readings both in the printer's copy and in its printing, far greater freedom of choice was then exercised than would now be warranted, but this does not mean that there must be no choice at all. Thus, Capell's rejection of the Folio's 'tomb' for the Quarto's 'grave' at I.ii.103 would now seem inadmissible on two counts: first, because it presupposes a memorial error of the kind we can postulate in the Quarto copy but not in the copy for the Folio; and secondly, because the reading was Compositor A's, whose errors in this text give no warrant for assuming he was prone to inexplicable verbal substitutions of this kind. On the other hand, at III.iii.47, where the Folio reads

there is much to be said for Capell's substitution of the Quarto's 'dread'[13] for the second 'great', since there is grave suspicion that Compositor A made a similar error at IV.i.287 (no. 19). The emendation of the Folio is here the more admissible because the Governor's speech seems very well reported in the Quarto. Consequently, Capell's emendation of the Folio's 'Succours' to 'succour' (Q) two lines earlier also merits serious consideration. So too does Capell's 'contaminate' (Q 'contamuracke', one of its oddest misreadings) at IV.v.16 for B's 'contaminated'.[14] The tendency is now to draw on the Quarto only when the Folio is manifestly in error. But we cannot suppose that all compositors' errors resulted in an obvious breakdown in the sense, and although we know that something is missing at IV.v.11 (B, no. 16), the Quarto's 'dread' at III.iii.47 is, in fact, a reading in which greater confidence can be placed than its 'honour' as a stop-gap at IV.v.11, since the Quarto seriously garbled the dialogue of the latter scene.[15] Where there are errors of repetition and anticipation in plays for which we have a single substantive text, there is often nothing to be done about them. Editors have beaten their brains in vain over 'Cleanse the stufft bosome, of that perillous stuffe' (Macbeth, V.iii.44)—a B crux to which a collateral text might have provided a satisfactory solution. Where we have a collateral text, even a bad one, we are indeed fortunate, and a number of Quarto readings in Henry V merit more consideration than they have recently had.

The quandary in which the variant readings at IV.vi.15 ('And cries . . . Suffolk') place an editor was pursued to its logical conclusion by Greg (Principles of Emendation, p. 47):

The logic of this is applicable to every case where an inferior text has a reading which is manifestly superior to its counterpart in a better text, but the dilemma can probably be resolved in most cases by an examination of what the compositor's errors reveal, since confidence in Jaggard's proof correction is not now so strong as it was in 1928. B's unfortunate habit of

attempting to memorise more than he could reproduce accurately, resulting in errors of recollection and anticipation, certainly seems to leave an editor freedom of choice at IV.vi.15. The only escape from the 'rather revolutionary' theory above (involving an editor in a choice between Shakespeare's first and second thoughts) is, in fact, a counter-revolution towards eclectic editing (involving a choice between truth and falsehood in collateral texts) or the abandonment of literary values—and what serious risks editors would be taking if they accepted all but the most blatant errors of the more authoritative text is evident from Compositor B's performance in 1 Henry IV. Because the copy for a particular text was more authoritative than that for a collateral it does not follow that it was proof against carelessness in its printing.

The above lists also show how frequently one of the later Folios corrected a Folio error. Such corrections have, of course, no authority and there is no reason why an editor should not reject F2's 'noblest' (A, no. 7) in favour of Malone's 'noble' if he thinks the latter the more Shakespearian reading—just as he is at liberty to reject Malone's 'and their' for Pope's 'while their' at IV.vii.75 (B, no. 19) if he thinks that B's garbling involved more than a single word. The corrections of the later Folios, like the conjectural emendations of later editors, must stand on their merits—and this holds for F2's emendations for the sake of metre, now often slightingly dismissed as 'metrical tinkering.' But accidental omissions that sometimes made a breach in the sense are just as likely, at other times, to have potholed the even road of a blank verse, and there is no reason why the one kind of error should be emended and not the other.[16] If a line is halting and there is an obvious remedy, even though it was conjecturally made by F2, why should an editor reject it? The very fact that F2 took an interest in this kind of error is surely evidence that metre mattered in the early seventeenth century; and if printing-house editors or compositors interested themselves in 'numbers,' they presumably recognised their importance and knew how often omission and interpolation had made them less 'absolute' than they should have been. Mostly, there is no hope of recovering a missing word or phrase, but metrical emendations are entitled to the same kind of consideration as any other kind of emendation made in derivative texts. F2's emendation of 'against' to ''gainst' at IV.ii.25, though lacking the finesse of Dyce's 'Ay, ay' for 'I' at IV.i.302, should not be rejected because it seems so simple and because F2's emendations for metre were sometimes senseless stop-gaps.

What is undoubtedly needed is a conspectus of the certain errors in all good substantive texts. No one who edits Antony and Cleopatra, for instance,

can afford to neglect the evidence for the number and kind of errors made by Compositor B in Othello; nor can the editor of King Lear safely disregard the evidence of Hamlet on the one hand and of Othello on the other, though it would be rash to infer that the rot that set in during the printing of the Tragedies was noticeably present in the Histories. Ultimately, any assessment of the trustworthiness of Folio texts will need to be related to what Hinman's collation of the Folger First Folios reveals about proof-reading, the order of formes through the press, and the order of their setting. In the meantime, an editor's only resource is to consider one Folio text in relation to another, set by the same compositor (or compositors) about the same time, as a check on his findings.

I mentioned above that it might happen that a type of error commonly found in a compositor's work might not be exemplified in the unanimously rejected readings of a single text. Thus, for instance, there is no case of word transposition in my lists above; but the lists exclude errors in foreign words, and a transposition does, in fact, occur in A's 'bien parlas' (Q 'parte fort bon') at III.iv.1, which editors unanimously correct to 'parles bien'. Further, most editors rightly transpose B's 'defie thee' to 'thee defy' at II.i.70 (following Q). Apart from the merits of the latter reading, it is useful to remember that there are 6 transpositions in B's stints of Richard II, and as quire c of this play (according to Willoughby) immediately preceded Henry V in order of printing, the Folio variants can profitably be considered in relation to the errors and suspect readings of Henry V.

The Folio text of Richard II merits, incidentally, more attention than it has recently had. It was printed from an example of Q3 which had to some extent been brought into line with a prompt book[17]. The Folio omissions are generally explained as 'cuts' in accordance with the acting version (as in the Folio Hamlet and Lear) and the very thorough overhaul of Q3's stage directions certainly suggests reference to prompt copy. The Folio text also exemplifies some mechanical changes in speech prefixes (e.g. the substitution of 'Ser.' for 'Man.') and a few of the variants introduced for the worse suggest a collator's misreadings ('soules' for 'smiles', 'placed' for 'plated', 'Rainston' for 'Ramston', 'Beares' for 'beards'). We have, there fore, in this Folio text some evidence for the kind of alteration a collator might make in preparing Folio copy. The text is also interesting, like 1 Henry IV, as a sustained example of the more stylish kind of Folio reprint[18], shedding light, in particular, on the way in which Compositors A

and B altered speech prefixes[19] and on their use of italics. To consider the Folio Richard II as a whole would take me too far from my purpose here, since the situation is complicated by edited copy and requires full discussion of a number of variants. I shall therefore use two characteristic pages from quire c, pp. 31-32, set by B and A respectively, as a sidelight on Henry V.

The Folio divergencies from Q3 make it evident that we must postulate (1) alterations in the example of Q3 prior to composition, and (2) errors introduced by the compositors. Since the following features are common to the stints of A and B, it is reasonable to assume they stood in their copy: act and scene divisions; a thorough revision of stage directions; the substitution of 'Rich.' for 'King.' as speech prefix (though this was not made on the first leaf printed, according to Willoughby, two years earlier); some cutting; the occasional restoration of a Q1 reading corrupted in Q2-3; some alterations of Q1-3 readings; some expurgating of oaths; and a number of necessary metrical improvements with also some conjectural bodging, presumably editorial.[20]

B set p. 31 (II.ii.32-II.iii.6). Readings for which he was probably responsible are:

               

II.ii. 51	Rauenspurg	for Rauenspurgh
52	?O	Ah
72	hopes linger	Hope lingers
77	Line omitted
89	Herfords[21]	Herefords
99	Come	Comes
103	om. two
108	om. go

         

122	?Exit	Exeunt Duke, & Queene, manent Bush. & Greene
134	?haue beene euer	euer haue beene
148	S.N.Bush.(misplaced)	149 S.N.Bush.
II.iii.S.D.	?Enter the Duke of Hereford, and Northumberland	Enter Hereford: Northumberland
6	our	your

I have queried readings which might have been due to copy. I assume that the three additional entries for Green, York, and the Servant (after II.ii.40, 72, 85) and the Exit at the end of the scene were in his copy, though I am more doubtful as to where responsibility lay for the alteration in the stage direction at II.ii.122[22] and in the heading to II.iii. I credit the copy also with the addition of 'Castle' at II.ii.119, the correction of 'for' to 'to' at II.ii.123[23] and the restoration of Q1's 'here' (om. Q2-3) at II.iii.3.

It is plainly not easy to draw a firm line between editorial alterations in B's copy and his errors. The situation is clearer in A's work where, if we similarly put down to copy three additional stage directions, one correction of a Q1-3 reading and four of corruptions introduced in Q2-3, we have four variants which look like compositor's errors:

       

II.iii.29	we last	for last we
35	direction	directions
87	om. no uncle
90	these	those

Two variants remain unaccounted for: A's 'kinsman' for 'Coosin' (Q1-3) at II.iii.125, almost certainly a correction from the prompt book,[24] and 'H. Percie' (II.iii.20 S.D.) for Q3's 'Harry Persie', which may similarly have been the collator's alteration, since Compositor A was not given to scamping his task.

This picture of the greater care of Compositor A is in accordance with the evidence of Richard II as a whole and suggests that we may be right in finding twice as many errors in B's work as in A's in Henry V. A single page of a compositor's work will not necessarily reveal all its facets. Thus, on p. 31, B lost one of Q3's two parentheses and added only one, whereas

A on p. 32 added two; and it was B who indubitably put the metre wrong (II.ii.92-94), whereas A generally showed the less sensitive ear. Nor does p. 31 exemplify B's habit of ravelling the wording, through recollection, anticipation or inexplicable substitutions (like 'just' for 'right', rhyming with 'fight', at I.iii.55). But the range and distribution of Folio variants (mostly errors) can easily be seen from an analysis, on a compositor basis, of Pollard's list on pp. 77-79 of his edition of Q3. Some of the items he included might have been disregarded, while on the other hand some additions need to be made to his record (which omits, for instance, the B errors at II.ii.103, 108, 134 mentioned above); but since his account of Folio errors can easily be referred to, I summarise on the basis of that. The stints of the two compositors were as follows:

• B pp. 23-31, 38, 42-5 (I.i.-II.iii.6, III.iv.54-IV.i.71, V.ii.7 ff.)
• A pp. 32-7, 39-41 (II.iii.7-III.iv.53, IV.i.72-V.ii.6)

An analysis on a compositor basis gives the following results:                    

	A	B
Letters omitted	3	9
Letters added	2	8
Letters substituted	1	4
Words omitted	3	2
Words transposed	2	5
Words substituted	18	43
	--	--
	29	71
Pages set	9	14

The most interesting category of variants is Pollard's last, which includes almost certainly a few corrections of Q1 errors, prompt book changes (such as are evident in some Q1 and Folio agreements in Hamlet), collator's misreadings, editorial bodges, and a very generous contribution of compositors' errors, mainly from B. There are, for instance, in B's stints 12 cases of garbling due to repetition (8 instances) and anticipation (4 instances) to one error of repetition and one of anticipation in A's.

I have so far exemplified what the analysis of compositors' errors in Henry V suggests about the kinds of error we must expect to find and where the greater concentration of errors is likely to be; and I have shown how the basis of enquiry can be broadened by considering the evidence of immediately antecedent text set by the same compositors. But the basis needs to be as broad as possible and, as I have already suggested, Richard II and 1 Henry IV are important as shedding light on the kind of alteration that a collator, an editor, or a compositor might make. In this way, they are fundamental to the understanding of Folio texts printed from more

elaborately corrected quartos: 2 Henry IV, Richard III, Hamlet, Lear, Othello, Troilus and Cressida, and—we should now add, in view of Mr. Cairncross's conclusions—Henry V. It is therefore the more surprising that Professor Jenkins in a recent article on the Folio Hamlet [24a] makes no use of their evidence and argues for conclusions about the Folio copy on mere assumptions about what a collator or compositor might do. On his premises, we should have to postulate in the Folio copy for 1 Henry IV, for instance, the speech prefix 'Falst.' (which occurs only once in Q5, where the prefixes are 'Fal.' or 'Fals.') and the spelling 'Bardolph' (which never occurs); we should further have to assume that a scribe and not the compositor was responsible for the free handling of B's stage directions and, therefore, that the Folio text was printed not from Q5 but from a transcript. Professor Jenkins's assumption that compositors reproduced copy word for word and point for point is, of course, at variance with the facts. By accident and design they altered the features of their copy very considerably and, although it seems fairly clear what the designs were, very few were executed consistently.[25]

The fact is that an editor today cannot afford to take the risk of concentrating on one text. He needs to be familiar with the whole canon, so that he can work on comparative lines, for the reason given at the outset of this article: that Jaggard's failure to correct proof with copy is a factor that textual critics of the present century have not systematically reckoned with; it undermines conservative editorial policy, since fidelity to a Folio text may be no virtue but merely the condonement of vicious errors, and it complicates more than need be the problem of transmission. The elimination of compositors' errors requires attack from within, on the lines suggested with reference to the indubitable errors in Henry V, and from without, by relating what one text has to contribute to the elucidation of another. The editor of Troilus and Cressida, for instance, faced with the Folio's 'just' for 'right' (Q) at I.iii.164 should connect it with B's substitution in Richard II, already mentioned, and he should recognise A's 'emulations, factions' (II.iii.69) for 'emulous factions' (Q) as an error of assimilation, like his 'Noblish English' in Henry V. B's habit of carrying in his head more than he could reproduce accurately (leading to paraphrase and memorial substitutions), his meddling with stage directions, and justifying lines by altering the wording are always a possibility to be reckoned with. Above all, ramifications outside the Folio should be followed up, especially the Pavier quartos, which shed a great deal of light on the sophistication of

texts in Jaggard's printing house.[26] If evidence is not co-ordinated in this way, the danger is that compositors' errors will be interpreted as authoritative corrections (derived from the prompt book) in the Folio Lear, as suggesting far more corrupt copy than need be for the Folio Hamlet, and as Shakespeare's first shots in the Folio Troilus and Cressida (thus suggesting that he revised the manuscript from which the quarto was printed). The number of arbitrary alterations on the first page of the Folio Romeo and Juliet should also act as a warning against taking too seriously all the variants in Hamlet, Lear, Othello, and Troilus and Cressida. Falsification of readings in the printing of Folio texts has done much to ravel the lines of transmission.

When an editor has settled the substantive readings of his text, taking into account what can be deduced about the copy, the compositors who set it, and the proof-reading (if any), he is not nowadays out of the wood. To start with, does he head his play 'Henry the Fift' (preserving the Folio's old ordinal numeral) or does he modernise to 'Henry the Fifth'? What must he do about 'accompt', 'wrack', 'venter', 'murther', 'apparence', 'Dolphin', 'eech', 'lyms', 'creeple', 'banqu'rout', 'strooke', 'vawting', 'cursie'—to give only a selection of the problems that face him? Obviously, if he is not a phonologist he is severely handicapped, since it is one thing for an old spelling editor to preserve all copy-text spellings which are not errors but a very different matter to preserve a selection of old spellings and to justify one's choice. An old spelling editor, for instance, would not hesitate over 'strooke' (p.p.) in the above list. Even if he cannot give a phonological account of the spelling, he will have seen it often enough. But what must the partially modernising editor do about it? He has to interpret the spelling in terms of a pronunciation related, by one means or another, to the O.E. strong verb which, by normal sound development, would have resulted in 'strike', 'stroke' (pret.), 'stricken', and he must canvas the possibilities of analogy (with, for instance, 'stick', 'stuck') or dialectal influence (as suggested by Kökeritz) before he can decide whether 'strooke' indicated a different sound from 'struck', which was in process of superseding it. An old spelling editor is concerned with old spelling; partial modernisation involves discrimination between one spelling and another on a phonological basis. This is a far more tricky business, which must inevitably lead to arbitrary measures, since even experts may often be put to it to decide what sounds some spellings represent; nor will experts always agree.[27] An appreciation of the ways in which Elizabethan speech habits differed

from our own is, of course, important; but an impression of what the differences amount to can only be given by phonetic transcription and a survey of the influences that have affected the spoken language since the sixteenth century. The arbitrary preservation of a few seemingly significant spellings is of negligible value, since so many Elizabethan spellings identical with ours (e.g. stream, join, George, hostess, realm, throne, adieu, suit, etc.) represented a pronunciation different from that of today. As a reflection of Elizabethan speech habits, even an old spelling edition will give a totally inadequate picture. The linguistic interest of an old spelling Shakespeare will, in fact, be mainly orthographical: if we want to know how actors spoke their lines, we shall find a truer record in the unsophisticated spelling of Henslowe or John of Bordeaux than in the conventional spelling of printed books.

It has often been said that 'Shakespeare' has one spelling oftener than another, as if the spelling of printed books was the author's and the antiquarian zeal for partial modernisation is perhaps prompted by the belief that compositors reproduced the spelling of their copy. But a preponderance of 'murther' and 'shew' spellings in the Folio, for instance, was merely due to the fact that Jaggard A and B preferred these spellings to 'murder' and 'show'—and compositors' preferences tell us nothing about the writer's. Though a start has been made on compositor identification on the basis of differential spellings (a very different matter from the consideration of a compositor's spelling as a whole),[28] nothing is known about conventions in printing-house spelling during the thirty years between the publication of Venus and Adonis and the First Folio: the whole subject awaits analytical

investigation on a basis of printing house, date, and compositor. All that seems clear is that, in the metamorphosis of manuscript into print, compositors largely followed their own orthographical bent and that the trade showed no inclination to make spelling more phonetic. Most of the changes introduced or established during the sixteenth century tended, in fact, to widen the breach between orthography and pronunciation (e.g. the restoration of 'b' in 'debt', 'l' in 'realm', and 's' in 'isle'), so that we should beware of supposing that, when Jaggard A and B substituted 'murther' for 'murder', as they usually plainly did, they were expressing anything more than a preference for one spelling over another.

The significance of the spelling of printed books in Shakespeare's day remains to be seen. In the meantime, the advantage lies with the traditional practice of bringing spelling into line with current usage, since a gallimaufrey of ancient and modern contributes nothing to the understanding of Shakespeare's meaning or to a critical appreciation of the differences between Elizabethan speech habits and ours. In accidentals, as in substantive readings, we need to be chary of supposing that compositors reproduced copy with the conservatism that has sometimes been assumed.

Notes