University of Virginia Library

Collateral Substantive Texts (with special reference to Hamlet)
by
Alice Walker

In my article on the Folio I Henry IV, 1 [1] I suggested that there might be as many as two hundred errors contributed by the compositor alone in the Folio texts of Lear and Othello. In what follows I shall consider what can be done to eliminate these and other errors from edited texts.

The quarto and Folio texts of Lear and Othello are substantive collateral texts and there is only one way in which collateral texts can be edited: that is eclectically. In his preface to Juvenal, Housman exposed with devastating logic 'the folly of leaning on one manuscript like Hope on her anchor and trusting to heaven that no harm will come of it.' Collateral substantive prints stand broadly in the same relationship to the 'true original' as collateral manuscripts at the head of a series to the archetype. Consequently, Housman's principles for the editing of Juvenal and Manilius hold good for the editing of those plays of Shakespeare for which we have two or more substantive prints. In the case of Juvenal, the text was corrupted in transmission by the blunders and interpolations of scribes and by interchange of lections. In the case of Shakespeare's plays, transcription, memorial contamination, printing-house errors and conflation have similarly operated against any one substantive text's preserving the whole truth and nothing but the truth. The lines of transmission for Shakespeare's plays are shorter, but a careless print could cause greater damage at a blow than a succession of conscientious copyists and we have only to think of Hamlet Q2 to realise what the damage might amount to.

The position as regards Shakespeare's plays is naturally simpler than is the case with the text of Juvenal. All the same, two of our 'better' substantive quartos, Roberts's Hamlet Q2 and Creede's Romeo and Juliet Q2, have suffered from contamination by an inferior strain as well as from

the carelessness of compositors. The third substantive text of Hamlet, the Folio, was likewise carelessly printed and the copy seems also to have been contaminated by playhouse corruptions. The quarto of Troilus and Cressida, though a good text, is at least at two removes from the foul papers; and the Folio text, which should have corrected all the quarto's errors, failed to do so and introduced further complications through printing- house errors of its own. The Lear quarto is a mongrel of obscure provenance, closer by descent (on the evidence of its length) to the foul papers than the Folio text; and the latter, though of respectable antecedents, was carelessly printed, corrupted by conflation, and, in any case, somewhat suspect since it was based on a prompt-book. Only if one of these prints had flawlessly reproduced autograph which had not been tampered with by another hand would we be justified in reliance on the better text.

No one, of course, supposes that substantive Shakespearian texts (whatever their superiority to a collateral) have the integrity of Malone Society reprints, and editors have rightly made use of an inferior collateral print (even the worst of bad quartos) when they recognised that through a scribe's or compositor's carelessness, or the editorial interference of a bookkeeper or expurgator, a word, phrase, line, or passage was lost or perverted in the superior text. Where much current textual theory and practice seems at fault is in refusing to desert the better text unless its readings are impossible.

Thus, on the false premises that it is not safe to reject a reading in the Folio Lear provided some kind of sense can be made of it, two recent editors follow the Folio in reading 'being' in the following passage (II.ii):

If the editor of Lear chooses to follow the Folio's 'being' rather than the quarto's 'bring', it should be for one of two reasons: either because he believes it is the better reading (and not because he finds it in the better text) or because there is nothing, so far as he can see, to choose between these variants. In the latter event, he must follow Housman's reasoned policy of giving priority to the reading of the better authority, not in any confidence that it is necessarily right but because the odds are that the better text will be right oftener than the inferior one. On merits, the quarto's 'bring' is unquestionably superior to 'being', and discrimination between readings on individual merit (and not according to provenance) will undoubtedly remove most of the compositor's errors from the Folio texts of Lear and Othello. Naturally, the less the damage to the inferior text in transmission the oftener the choice will prove to be a straightforward one, uncomplicated by the possibility of emending the reading of the better text in some other way.

It might perhaps be asked 'why then all this bother over problems of transmission if the choice of reading rests on merits ?' There are two main reasons, which I have already hinted at. The first is that we need to know which is the more authoritative text because an editor must fall back on this when the merits of readings are so evenly balanced that he finds discrimination impossible. Fortunately, in great literature, and especially in the work of a connoisseur of words like Shakespeare, it does not often happen that an editor needs to gamble in this way. The second reason is that we need to know all we can discover about the transmission of substantive texts in order to formulate coherent principles for emendation. Many of our substantive collateral texts are memorially contaminated and for this reason it may at times be necessary to emend a blunder in the superior text rather than to accept what appears at first sight the superior reading in an inferior one. We need, that is, to know not only what is an error (a matter of literary judgment) but also what kind of error it is (a question of transmission). This is why we need the fullest possible information about the history of the printer's copy (whether foul papers, a transcript, a memorially contaminated text, or 'mixed' print and manuscript) and all that can be discovered by bibliographical means about its printing (the habits of the compositor or compositors who set it and everything that the press-work and proof-correction can tell us).

From my own experience in editing these texts, I should say that, when the variants are considered on their merits (artistic, dramatic, acoustic[2] ), the Folio versions of Lear and Othello contain well over two hundred errors apiece (mostly compositor's errors) and that the number is nearer three hundred in Lear. This is, of course, a personal judgment. On what is a matter of editorial principle based on reason, like Housman's logical procedure for the editing of collateral texts, we should be able to reckon on editors seeing eye to eye. Agreement is also to be expected over principles of emendation, provided there is basic agreement on transmission. But where the merits of individual readings are in question, though responsible editors are fairly certain to concur in rejecting most of the poorer readings, we must expect difference of opinion over a few; and this is only

proper if the aim of English studies is to foster the exercise of an informed but independent judgment on literary matters.

Consequently, I am not afraid like one bibliographer[3] that, since what is known of Jaggard's proof-reading makes the emendation of Folio texts not merely allowable but desirable, editors will give their fancy too free a rein. No one would begin to edit Shakespeare without some feeling of the matter or without giving careful thought to the work of enlightened editors who worked on genuinely eclectic lines-notably the Old Cambridge editors. We know a great deal more about the transmission of Shakespeare's plays than they did, but a great deal less about how to choose a good reading from a bad. They paid Shakespeare the compliment (I think rightly) of supposing that the better reading, even if it occurred in an inferior text, was his. This was sensible, since it is obviously easier to mar what is good than to improve on it. The fashion is now, on the contrary, to suppose that the errors in the better texts were what Shakespeare wrote and that the better readings in inferior texts were the improvements of book-keepers and actors. We ought to give our greatest poet and dramatist the benefit of any doubt.

Nor, like another bibliographer,[4] do I regard the kind of conflation which I postulate in the transmission of the three Hamlet texts as a complication one would gladly be rid of: it would be far better to be rid of common errors. Why should we regret the loss of security if the security was merely security in error? Editors of classical texts have surmounted the obstacles of conflation, 'troublesome to the truth-seeker and annoying to the formulist yet not on the whole disadvantageous to the author.'[5] If there are common errors due to conflation they should be removed.

To suggest that we need to re-consider the principles on which collateral texts are edited is not therefore to suggest putting the clock back to the mid-nineteenth century. Eclecticism today would be (and is) controlled by the great advances due to twentieth-century research into the big problem of transmission which defeated the Old Cambridge editors. Eclectically edited texts of Hamlet today are, of course, grounded on Q2 and the recognised procedure (where merits are evenly matched) is to give Q2 the benefit of the doubt and (if possible) to emend the Q2 errors

in accordance with what can be inferred concerning transmission rather than to accept (as the Old Cambridge editors did) Folio variants which may be no more than bodges. But this, and the emendation of common errors where there has been conflation due to the use of mixed copy, is not the end of the matter. Hitherto we have generalised about texts, describing The Merchant of Venice Q1 and the Folio Henry V, for instance, as 'clean' texts and Hamlet Q2 and Coriolanus as 'careless.' We can no longer generalise in this way. The two Roberts quartos were the work of the same pair of compositors, as has been shown elsewhere in this volume; the two Folio texts were the work of the same pair of Jaggard compositors, and how much more careful Jaggard A might be than Jaggard B was demonstrated in my article on I Henry IV. Hence the superiority of Henry V may have been due to its having been mainly the work of Jaggard A, whereas Jaggard B was largely responsible for Coriolanus. There is certainly a marked increase in the number of errors when B takes over in the former (pp. 71b-75a,88-9,92a) and a decrease when A's hand is evident in Coriolanus (pp. 7,9-11,13- 4,16a,18a). Hence it might not follow that the copy for Coriolanus was inferior in any way to the copy for Henry V. The basis for comparison is, therefore, in the first instance compositorial and, if it happens that the work of one compositor was inferior to that of his fellow, generalisations about a text as a whole and conclusions drawn from such generalisations as to either the extent of printing-house corruption or the character of the copy may seriously distort the facts.

Since we now know that the same pair of Roberts compositors set The Merchant of Venice Q1 and Hamlet Q2, it is natural to ask what emerges from a comparison of their work in these two texts and what the editorial implications are. This is especially important in the case of Hamlet. Dover Wilson's pioneer enquiry into the problem of transmission led him to the conclusion that the Q2 text was set up by a single compositor, who was both incompetent and hurried. The conclusion has not until recently been questioned and the eclectic tradition in the editing of Hamlet, though it has shifted ground a little, has not lost face: Dover Wilson's, Alexander's, and Sisson's are all eclectic texts. I have already argued that the eclectic editing of collateral texts is soundly based on common sense, but common sense will not, of course, explain whether the supposed disparity between the two Roberts prints is real or imaginary nor, if the belief is justified, account for the difference.

Many factors have to be borne in mind. I have mentioned one already that inferences about the transmission of a text cannot fairly be made until its errors have been analysed in relation to the compositors' stints. This therefore needs to be done before we can judge whether copy was a factor

to be reckoned with. The second is, that it is dangerous to assume that texts set by the same compositor or pair of compositors were set under equally favourable conditions. Because Compositor B was careless in I Henry IV and his work is manifestly more careless than A's in Henry V, it by no means follows that he was invariably responsible for an average of eight errors to a page in setting from print, his average in the former play, and five errors to a page in setting from manuscript, his average in the latter (so far, of course, as we can judge from the manifest errors in a text for which we have a single authority). Julius Caesar, for instance, was mainly B's work (A setting only pp. 112-4 and the lower half of 129b), but it is, to all appearances, a cleaner text than Henry V; and it would be rash to argue that this was due to more legible copy for Julius Caesar on account of the considerable number of errors in I Henry IV where B was working from print. In the present state of our knowledge we can only guess what caused this apparent inequality of workmanship in Folio texts. The work of the normally accurate Compositor A went to pieces, for instance, in Troilus and Cressida, and haste may therefore have affected the quality of the work.

Thirdly, we have to remember that the editing of texts for which we have a single authority is conservative. This was the policy which the Old Cambridge editors deliberately adopted on the grounds that they knew too little about transmission to emend, with confidence, corruptions of unknown origin.[6] Where they had, or thought they had, collateral texts, they selected what, in their judgment, was the true reading; where they had a single authority, they only emended when the Folio reading was impossible and the emendation adopted was the only possible remedy. If the defect could be made good in more ways than one, then they made no alteration but left it to the reader to exercise his own judgment on the emendations recorded in their notes (a fact perhaps forgotten by those who are no less conservative but provide no critical apparatus). Since no appreciable change is evident in the present-day editing of texts based on a single authority, comparison between the number of readings normally emended in The Merchant of Venice and Hamlet Q2 is significant only if we postulate the editing of the latter on conservative lines, disregarding the evidence of Q1 and the Folio.

I shall approach Hamlet through The Merchant of Venice, tabulating the readings in The Merchant which any or all of the four following editions have emended: the Old Cambridge edition, Dover Wilson's, Alexander's, and Sisson's. I disregard Q2 readings followed by the Old Cambridge editors in the belief that it was the earlier of the two editions dated 1600, except in a few cases where the choice may have been made on merits.[7] I give the line numbering of the 1891-3 Old Cambridge edition and italicise trivial errors (including a few mis-spellings which may have been due to misreading), giving the correction (when necessary for clarification) in brackets. In the case of more serious errors I give first the Q1 reading and then the emendation in modern spelling, followed by the sigla of the editions which have adopted it. If there is more than one emendation, I place the best first. I disregard everything immaterial to verbal accuracy (including grammatical normalisations, expansions and contractions for the sake of the metre) and also the provenance of emendations unless they originated in one of the later quartos or Folios.[8] Many of the emendations (as will be seen) are of the kind that anyone following the matter intelligently might make and a few may have been accidentally introduced. It is of some interest that Jaggard B set the 1619 quarto (a compound of intelligence, highhandedness, and carelessness) and also pp. 162a (= 164a),178-180a,183 of the Folio text so that the Folio emendations between III.v.75 and IV.i.225 (inclusive) are probably his.

A conspectus of this kind can, of course, give only an inkling of what the corruption may amount to,[9] but it seems reasonably likely to give a fair picture of the comparative merits of the two compositors and what it suggests is that Roberts X was more prone to error than Roberts Y. What we have is 19 trivial and 57 more serious errors,[10] of which 11 trivial and 36 more serious are in X's stints ( just over 34 pages) and 8 trivial and 21

more serious in the work of Y (38 pages).[11] That the error at III.iv.49 was Shakespeare's and the mistake at IV.i.393 possibly due to a conjectural attempt to supply a missing speech prefix does not affect the picture. No full sheet set by X is (apparently) so free from errors as Y's sheet B; nor is any sheet set by Y so manifestly careless as X's sheet G. On the other hand, there is seemingly not much to choose between X's work on sheet C and Y's on sheet D. Misreadings are not numerous: 'docks' at I.i.27, 'wit' at II.i.18, 'timber' at II.vii.62, 'heere' at III.i.92, 'Tranect' at III.iv.53, 'meane it, it' at III.v.68, 'bleake' at IV.i.74 are errors of this kind, but the majority of the errors suggest the accidental interpolation or omission of letters and words rather than difficulties with copy.

Roberts's Hamlet Q2 gives much the same sort of impression of greater care on the part of Compositor Y. If we isolate from consideration the first three sheets (the work of X) set up, I believe, from mixed copy and, therefore, not so relevant for comparison with The Merchant of Venice, what we find on a survey of Q2 readings rejected by all four editions in favour of the Folio readings is that the number (including trivial errors) is about 35 on every full sheet except on sheets E and K (both the work of Y), where the number is about halved, and 24 on sheet I (the work of X and his best stint when setting from manuscript). But these figures afford no real basis for comparison with The Merchant of Venice and what we need, in order to get the matter into perspective, is a conspectus of what the postion might have been if editors had been dependent on Q2 for the text of Hamlet.

What a difference the survival of a collateral text makes can be demonstrated from Hamlet itself. I have mentioned sheet K as one of the two sheets set by Y which exemplify (on the evidence of the four editions used as yardstick) more reliable workmanship than is characteristic of sheets E-O in general. The four editions reject unanimously one reading on K1, three on K1v, one on K2, two on K2v and one at the head of K3. Two-thirds of K3 and nearly the whole of K3v are occupied by Hamlet's conversation with the Captain and his soliloquy on the little plot of Polish ground (IV.iv.) Here editors accept the Q2 text, though it has manifestly garbled both matter and metre here and there, having no collateral authority by which to correct its readings since the Folio omits nearly all of the scene. On K4, where the Folio comes to the rescue again, one reading is corrected and three are corrected, with Folio aid, on K4v What we need, therefore, is to bring the editing of Hamlet Q2 into line with the editing of these two pages and The Merchant of Venice.

Nothing would be gained by listing for the whole of Hamlet the Q2 readings rejected in favour of the Folio's by the four editions. Samples

will serve to show where the difference lies between the two quartos and what it amounts to. I shall give, therefore, the readings of three sheets (E,F,G) for comparison with The Merchant of Venice tables above, marking with an asterisk the Q2 readings which I judge all four editions would have emended and with an asterisk in brackets readings over which I judge opinion would have been divided, bearing in mind, of course, editorial procedure concerning similar errors in The Merchant of Venice. The second reading given in my tables is that of the Folio. I obelize readings where some larger omission would have been suspected.

What emerges from this hypothetical view of Hamlet Q2 (E-G) as edited on conservative lines without reference to Q1 or the Folio is as follows:

       

	E	F	G
Trivial errors	4	5	5
Readings postulated as corrected in all 4 editions	5	12	8
Readings over which opinion would have divided	4	7	5

It would therefore appear that the disparity in workmanship between Roberts's Merchant of Venice and Hamlet is not nearly so striking if we disregard the evidence of collateral texts, though it is plain that the workmanship in Hamlet is certainly poorer. Y's sheet E (on the hypothesis above) with 4 trivial errors and 5-9 manifest errors of a more serious kind is among the best of Y's stints in Hamlet and compares unfavourably with what would appear to be his best sheet (B) in The Merchant of Venice. Nor is it quite as good as sheet D in The Merchant of Venice, seemingly his worst stint in the earlier quarto.

What occasioned the difference would appear in some measure to be haste, since difficulties with the copy (though they might have occasioned the haste) cannot be directly responsible for the trivial slips: Y's 9 on sheets E and G, for instance, is in excess of his total in The Merchant of Venice [12] and that X's sheet F was hurried work is plain from the unsightly failure to range speech-prefixes, evident on most pages of this sheet, as well as from the large number of errors, though the number of manifest errors does not compare too unfavourably with his sheet G in The Merchant of Venice.

Since Y appears from both texts to have been the more careful workman, it seems likely that we shall find the best reflection of the character of the copy in the sheets set by him. I shall therefore consider briefly what the Q2 readings on sheets E and G rejected by all four editions suggest about the manuscript. On E there are three misreadings ('wit' at II.i.38, 'sallies' at II.i.39, and 'about' at II.ii.125) and one misunderstanding ('or tooke' at II.i.58, which, if the compositor had interpreted the copy rightly, would have appeared as 'ore-tooke'); there are five errors due to the omission or addition of a letter ('addistion' at II.i.47, 'stokins' at II.i.79, 'passions' at II.i.105, 'shone' at II.ii.76, and 'wath' at II.ii.147); five omissions (at

II.i.28,52 and II.ii.57,90,148) and the substitution of 'her' for 'his' (at II.ii.142) through repetition of a pronoun. On sheet G there are seven misreadings ('deale' at II.ii.595 (an error repeated), 'lowlines' at III.i.46, 'quietas' at III.i.75, 'euocutat' at III.i.118, 'stature' at III.i.159, 'vnmatcht' at III.i.188, and praysd' at III.ii.29); three errors due to the omission of a letter ('An' at III.i.1, 'sickled' at III.i.85, 'pronoun'd' at fifteen omissions (at II.ii-577, III.i.32, 55, 83, 92, 107, 121, 129, 137, 141, 142, 145 and III.ii.22,26,36) and seven further errors due to misapprehension or faulty memorisation ('two' at III.i.28, 'Wee'le' at III.i.33, 'these' at III.i.99,[13] 'list' at III.i.144, 'expectation' at III.i.152, 'what' at III.i.157, and 'detected' at III.ii.87).

We get a very different impression from these two sheets. The errors on sheet E suggest lack of care rather than difficulties with the copy and that the manuscript is unlikely to have presented very serious problems seems clear from the fact that this was Y's first stint and, therefore, his initial effort with an unfamiliar script. He may, of course, have started on this sheet with time in hand, but we can at least safely postulate that, given leisure, the copy could be deciphered. What occasioned the more numerous errors on sheet G may therefore have been haste, possibly partly due to an increase in the amount of prose; and it is worth remarking that X's poor performance on F (mostly prose) and better work on I (all verse, and his best sheet set from manuscript) seems to point the same way.

It would be unreasonable to suppose that Roberts's manuscript was a careless transcript of the foul papers, since a playhouse transcript prior to the prompt-book would have served no useful purpose if it had not been both legible and passably accurate. And the possibility that any kind of transcript lay behind Q2 seems to be ruled out because, although a copyist might not achieve the same level of accuracy throughout, it involves the double coincidence that good and bad patches corresponded with the sheets of a printed book and that the better patches twice fell to the lot of the better of a pair of compositors. We have seen that there is a marked increase in the number of errors on F and there is a similarly noticeable difference between X's sheet I (where twenty-six readings are condemned by all four editions) and Y's sheet K (twelve readings condemned on the six pages for which the Folio provides a check). The errors in Q2 must therefore be laid to the charge of the compositors. On the evidence of sheets E and K, the text could have been much better, though Y's work on these sheets precludes our supposing that, even if Hamlet Q2 had been set at greater leisure, it would have seemed as clean a text as The Merchant of Venice.

To postulate that the compositors were mostly hurried may not, therefore, go all the way to explain the difference, and a contributory factor may have been that Roberts's manuscript was not as fairly written as the manuscript for The Merchant of Venice. Obvious misreadings, as we have seen, are infrequent in the earlier quarto. There are, in fact, more misreadings on sheet G of Hamlet Q2 than there are (seemingly) in the whole of The Merchant of Venice. The likelihood would therefore, seem to be that the manuscript for Hamlet was more cursively written and much like the manuscripts from which Henry V and Coriolanus were printed. Given time, or comprehension of the meaning, it was readable (there is, for instance, only one misreading, 'ground' for 'grave' on sheet K[14] ), but what difficulties Shakespeare's hand presented when transliterated without understanding is evident from some of the errors in French words in Henry V. Even though the spelling is often quasi-phonetic, 'saaue' for 'suivez', 'perdia' for 'perdu', 'buisse' for 'baiser' suggest misreadings of the kind of writing that gave rise to 'deale' for 'devil' in Hamlet Q2-a word which had presented no difficulties in The Merchant of Venice and might have been inferred from the context in Hamlet if the compositor had had time to reflect on the meaning.

The general verdict on the number of errors in Hamlet Q2 is (I think rightly) that it is very high. All the same, regrettable as it is, there is nothing extraordinary about this state of affairs, and we have only to think on the one hand of the Folio Romeo and Juliet to realise what garbling could result from hasty work (even when the copy presented no difficulties) and on the other hand of Okes's Lear to realise what blunders ill- written copy might lead to. The most disturbing class of Q2 Hamlet errors is the many omissions. Bearing in mind Compositor B's interpolations in I Henry IV and the fact that he set the greater part of the Folio Hamlet, this is where vigilance is especially necessary, even though recognition that there is playhouse corruption in the Folio text has already provided a safeguard.

A few facts of general interest emerge from the preceding survey. The first is that genuine eclecticism in the editing of Hamlet has not resulted in wide divergence of opinion.[15] On the other hand, there is no established consensus of opinion over the seemingly few errors and suspect readings

of The Merchant of Venice. The second is that the eclectic editing of collateral texts not only enables an editor to locate and emend the errors of the better text with greater assurance but it also throws a great deal of light on one aspect of printing-house transmission. If proof had been scrupulously read with copy, there would have been no need to use the evidence of collateral texts in this way; but Roberts's proof-reading for Hamlet was evidently haphazard and even literal errors escaped correction. If the evidence of the Folio text for Roberts's compositors' errors is disregarded, we have nothing to substitute except a personal judgment which is, in every way, more partial; and how imperfectly it must operate is evident from The Merchant of Venice. Fuller knowledge about transmission gives greater security, and the study of collateral texts has as much to contribute to fuller knowledge as the study of reprints. What can be deduced from the Folio Hamlet about Roberts's compositors should, in fact, assist in the editing of The Merchant of Venice.

Notes