University of Virginia Library

The Text of Shakespeare's Romeo and Juliet [1]
by
G. I. Duthie

THE EDITOR WHO APPROACHES ROMEO AND Juliet must carefully consider the first quarto, Q1, published in 1597, and the second quarto, Q2, published in 1599. Q1 is in its entirety a substantive edition. It was at no point printed from any document now extant. Granted the fact that, as we shall see, Q2 depends directly on Q1 at certain points, it is nevertheless abundantly clear to the most superficial observation that, as a whole, Q2 is not a reproduction of Q1. Q2 depends also on some other, non-extant, source—presumably a manuscript. In the main Q2 is also a substantive edition. Only Q1 and Q2 have the status of substantive editions.

Q1 is a 'bad' quarto, and Q2 is a 'good' quarto. It is generally agreed, I believe, that Q2 gives a text taken in the main from an authentic documentary source. As regards the Q1 text, I subscribe to what I take to be a common view—namely, that it is a memorial reconstruction—the work of a reporter or reporters.

The first question which I should like to discuss is—what exactly was the nature of the copy for Q2?

It is well known that there are some conspicuous bibliographical links between the two quartos. They indicate that, at any rate at certain points, Q2 was printed from a copy of Q1.

Indubitable and impressive links are to be found in an extended series in the passage stretching from I ii 57 to I iii 36.[2] This passage occurs in both editions on sheet B.

Links in this passage have been pointed out and discussed by various critics—Robert Gericke, Miss Greta Hjort, Sir Walter Greg, and Mr. Sidney Thomas.[3] Mr. Thomas has shown,[4] in my view convincingly, that here we are dealing, not with independent derivation of the two printed texts from the same document, but with a direct dependence of Q2 on the printed pages of Q1.

Noting a number of links in this passage, Miss Hjort apparently sprang to the conclusion that Q2 in its entirety was printed from a copy of Q1 which had been corrected by an editor by comparison with another document. I am not the first to point out that this could not be so, since there are some pages in Q1 which simply could not have been altered in pen and ink in order to yield the Q2 text: more alterations and additions would have been required than there would have been room for on the Q1 pages. Miss Hjort's theory will not do as regards the nature of the copy for Q2 as a whole. Sir Walter Greg subsequently suggested that the copy for Q2 consisted of (i) a first quarto, corrected by an editor, in handwriting, as far as the end of sheet B, and (ii) from that point on, a transcript of a playhouse manuscript. Here are Sir Walter's own words: "It seems clear that some editor was commissioned to prepare the copy for an authorized quarto, and for this purpose was provided with the 1597 edition and a playhouse manuscript. He began by taking the printed text and elaborately correcting and expanding it by comparison with the manuscript, but when he got to the end of sheet B he decided that it would be less trouble to make a transcript of the latter. This he proceeded to do through the remainder of the play, though I will not say that he may not have used other fragments of the printed text, and I am certain that he consulted it on occasions when the manuscript was obscure."[5]

Now in the article which I have cited Mr. Thomas makes a point which seems to me sound, and which necessitates a modification of Sir Walter Greg's theory. Let us consider the relationship between the two quartos up to the end of Q1 sheet B, leaving

the Prologue out of account until further notice. From the first line of the first scene of the play up to the end of Q1 sheet B, there are no bibliographical links between the two quartos before I ii 57. There is nothing between I i 1 and I ii 56 which in any way necessitates the view that Q2 depends on Q1. From I ii 57 to I iii 36 there is a considerable amount of impressive evidence indicating a direct dependence of Q2 on Q1. Some of this evidence consists of errors in Q1 which are reproduced in Q2. On the one hand these errors were not corrected: and on the other hand there is no textual difference between the two quartos in the relevant passage which need be regarded as a correction by an editor. Between the two points with which we are here concerned, I ii 57 and I iii 36, the two quartos show only a handful of verbal differences. At I ii 68 Q1 has an "and" which Q2 lacks (Q1 "my faire Neece Rosaline and Liuia, Q2 my faire Neece Rosaline, Liuia,); but the Q2 compositor could easily omit a single word by accident, and it is highly probable that the Q1 "and" is correct, since the context appears to be verse, and the "and" is necessary for metrical smoothness. At I ii 77 we have the variant Q1 "thee", Q2 "you" (Q1 "Indeed I should haue askt thee that before", Q2 "Indeed I should haue askt you that before"). The Q2 compositor may well have made the substitution unconsciously himself as he carried a group of words in his head while setting up type. Or perhaps his eye caught the "you" in the next line. At I iii 12 we have Q1 "a", Q2 "an". Q2 is right, for the next word is "houre" (Q1 "Faith I can tell her age vnto a houre", Q2 "Faith I can tell her age vnto an houre"); but the Q2 compositor could easily make the obvious correction on his own responsibility. At I iii 33 Q2 has a definite article which is desirable from the point of view of sense, and which is absent from Q1 (Q1 "to see it teachie and fall out with Dugge", Q2 "to see it teachie and fall out with the Dugge"); but, again, the Q2 compositor could easily have supplied it on his own initiative. And these four appear to be the only verbal variations between the two quartos in the passage extending from the first to the last bibliographical link in this portion of the play.[6] The position is, then, that in this passage certain errors in Q1 were not corrected in the copy for Q2, while the few small verbal alterations that there are in Q2 can all be attributed to the Q2 compositor himself. It looks, then, as if between I ii 57 and I iii 36 Q2 was printed from Q1 unrevised by scribal intervention.

Now Mr. Thomas argues[7] that, before and after this passage, Q1, up to the end of sheet B, could hardly have been corrected to yield the Q2 text, for more alterations and additions would have been required than there is room for in Q1. I agree. And so I believe that Sir Walter Greg's theory of the relationship of the two quartos up to the end of Q1 sheet B must be modified. Q1 sheet B ends at I iii 100. My view of the copy for Q2 up to this point (excluding the Prologue) is as follows: from I i 1 to I ii 56 Q2 depends on manuscript copy; from I ii 57 to I iii 36 Q2 depends on Q1 unrevised; from that point to I iii 100 Q2 again depends on manuscript copy. This I take to be Mr. Thomas's view, and I think he is right.

It is interesting to speculate on the question of why, at I ii 57, Q2 should suddenly begin to depend on Q1 instead of on the copy on which it has hitherto been depending, and why, at I iii 36, it should suddenly revert to copy other than Q1. An obvious suggestion is that between these two points a manuscript ultimately or directly underlying Q2 was mutilated or illegible, so that recourse had to be had, by editor or compositor, to the previously printed edition. Mr. Thomas, the most recent scholar to discuss this matter, has advanced this explanation. "The manuscript from which Q2 was printed," he declares, "either lacked the leaves which contained the lines [in question] or was completely illegible at that point."[8] Mr. Thomas is here reiterating the explanation given by the earliest critic who noticed the connection between the two quartos here—Robert Gericke. But I am not quite happy about this explanation, for the following reason. Round about the point where Q2 here ceases to depend on Q1, in the midst of a long speech by the Nurse, the Q1 text changes its character. Up to I iii 36 the Nurse's lines in Q1 have, I think, been metrically quite satisfactory. At I iii 36 Q2 suddenly ceases to depend on Q1, and only five lines further on we find that, while Q2 is still metrically

satisfactory, Q1 shows a definite metrical breakdown.[9] Does this not at least suggest to us the idea that perhaps Q2 is dependent on Q1 in the passage with which we are concerned because that passage is in Q1 satisfactory in the main, whereas Q2 leaves Q1 at I iii 36 because Q1 becomes no longer satisfactory? That is what I tentatively suggest.

Q2 leaves Q1 at I iii 36. Almost immediately Q1 shows memorial corruption. In Q1, immediately after Q2 has left it, the Nurse speaks of how Juliet could have "wadled vp and downe". In Q2 she speaks of how Juliet could have "run and wadled all about". I believe that in Q1 we have here an anticipation of II v 51 where the same speaker, the Nurse, refers to herself, in Q2, as catching her death with "iaunsing vp and downe". Q2 has no sooner left Q1 here in I iii than Q1—immediately—shows memorial corruption. Now we must be very careful. Immediately Q2 leaves Q1, Q1 has memorial corruption. But it may be suggested that Q1 has memorial corruption in the passage in which Q2 copies it— memorial corruption not detectable because there is nothing with which to compare the Q1 text (that of Q2 being the same). We must, then, be careful. The argument I should advance is this. In Q1, from the beginning of I iii up to I iii 36 the Nurse's part is satisfactory metrically: at I iii 36 Q2 leaves Q1: within five lines of this we find that Q1 is metrically unsatisfactory, whereas the different text of Q2 continues metrically sound: this suggests that Q2 may leave Q1 because, whereas Q1 had been satisfactory as a whole heretofore for a time, it no longer was. But why should Q2 leave Q1 four lines before the metrical breakdown in Q1? Well, in the very second line after Q2 leaves it, Q1 shows a memorial error. I admit that my suggestion is without absolute proof. But I should say that it seems to me reasonable to suppose that Q2 leaves Q1 at I iii 36 not because the manuscript behind Q2 has suddenly become legible again, where before, for a considerable stretch, it has not been legible, but rather because before I iii 36 Q1 for a time gave a text in the main satisfactory, whereas after that it does not.

I should suggest, then, the following hypothesis, very tentatively. A person who may be called the Q2 editor, or Scribe E, was entrusted officially with the task of preparing copy for Q2. He

was equipped with an authentic manuscript, and also with a copy of Q1. At I i 1 he began transcribing the authentic manuscript. But he continually looked at Q1, comparing it with the authentic manuscript. When he came to I ii 57 he read ahead in Q1, comparing it with the manuscript, and he found that for some time thereafter—up to I iii 36, in fact—Q1 was of good quality. He saw that after that point Q1 deteriorated sharply. And so, having up to I ii 56 been copying the authentic manuscript, he now tore out of his Q1 the two leaves, B3 and B4, which contained the patch of text he had found to be satisfactory. He drew his pen through the text on B3r up to and including I ii 56, and he drew his pen through the text on B4v after I iii 36. Then he simply put the two printed leaves, so dealt with, in the pile of leaves, hitherto manuscript, which he was producing to form the copy for Q2. We note, of course, that, while Scribe E found the Q1 text satisfactory between the two points, he neglected to correct (presumably even to notice) certain errors which Q1 in fact contained there: it is less satisfactory to us than it seemed to him.

I cannot forbear referring to the well-known fact that a troublesome and interesting feature of this sheet B passage in which Q2 depends on Q1 is that in it the Nurse's lines are printed in italics in both quartos. The Nurse begins speaking at I iii 2, and that it is her first line in the play.

Now since Q2 is dependent on Q1 in this passage, it is easy to say that the Nurse's lines are here printed in italics in Q2 because they are so in Q1. That, I think, is correct. The awkward question that must be faced is—why were they printed in italics in Q1 in the first place? I can only assume (and I do not think I am the first to do so) that in the manuscript behind Q1 the Nurse's lines in this section of text were written in the Italian type of handwriting, the speeches of the other characters being written in the English type of handwriting. The Q1 compositor reproduced this calligraphic distinction typographically. Now why should the Nurse's lines be written in any manuscript in a different type of handwriting from those of the other characters? I cannot see that it can be held in the least probable that anyone writing the play out straight ahead would on his own initiative vary his handwriting in this way. I take it, then, that at this point there lies behind Q1 a

composite manuscript. One of the persons involved wrote the Italian type of hand. He wrote out the Nurse's lines, but not those of the other characters. What sort of manuscript would contain the speeches of only one character? Clearly, one presumes, an actor's manuscript part. The following theory would explain the facts. The persons responsible for producing the copy for Q1 had in their possession a manuscript fragment containing the beginning of the Nurse's part. This was written in an Italian hand. One of the persons involved in the production of the copy wrote the intervening speeches of the other characters into this manuscript fragment. He used an English hand. The result would be a scrap of manuscript in which the Nurse's lines appeared in one type of handwriting, and the intervening speeches of the other characters in the other type of handwriting. A compositor who decided to reproduce this difference typographically would produce a printed text such as we find in fact in Q1. Using Q1, the Q2 compositor, reproducing the typography of his copy, would produce what we find in fact in Q2. Such a manuscript fragment as I am suggesting would, one supposes, have originally contained cue-words, and these presumably were also written in Italian handwriting. But it is easy to suppose that in the composite fragment the cue-words were deleted when the complete intervening speeches were inserted in English handwriting.

In connection with the italics in the Nurse's part a further point must be noted. Q2 leaves Q1 at I iii 36. Scribe E doubtless reverted to the production of manuscript copy. After I iii 36, sig. B4v of Q1 could hardly have been corrected by hand so as to produce the Q2 text—there is not enough room. But Q2 continues to set up the Nurse's lines in italics for a while (up to the end of sheet B). That, I imagine,—and the suggestion does not originate with me,—is to be attributed to the Q2 compositor deciding to go on, on his own responsibility, making a typographical distinction no longer rooted in his copy.

Italics for the Nurse cease in Q1 at the end of sheet C. Again, of course, it is not necessary to assume that, for the Nurse's speeches, Q1 depends on copy in Italian handwriting right up to the end of sheet C. The fragmentary manuscript part which we have postulated may have broken off before that, and the Q1

compositor may have continued his italics on his own responsibility until the absurdity of the procedure occurred to him or was pointed out to him.

I have spoken of a fragment of a manuscript part: but I would point out that I have not claimed that it was authentic at any point. Scribe E accepted the Nurse's first 28 lines or so from Q1, and, as we have seen, they are metrically satisfactory; but that may be attributable simply to good reporting.

A final word remains to be said about another detail of the Q1 italics problem in I iii. In Q1 the scene ends with a three-line speech by the Clown, and it is printed in italics. It is not so printed in Q2. The question is—does Q1, as regards this three-line speech by the Clown, rely on copy in Italian handwriting? I do not think it necessary to assume that it does. The Clown's italics may be laid solely to the charge of the Q1 compositor. For quite a time he has been setting alternate passages in italic and roman. The Nurse's final speech in the scene is in Q1 followed by a single-line speech by Lady Capulet, in roman, and then by a three-line speech by Juliet, also in roman. Then comes the Clown's three-line speech, in italics. It is quite conceivable that the Q1 compositor simply decided, on his own initiative, that it was time for italics again.

Q1 sheet B finishes at I iii 100. Sir Walter Greg's theory was that up to this point Q2 was printed from a copy of Q1 corrected by hand. That theory we must, I think, reject; and we must substitute the following. From I i 1 up to I ii 56 Q2 was printed from manuscript copy: from I ii 57 to I iii 36 Q2 was printed from Q1 uncorrected by Scribe E: and from that point to I iii 100 Q2 was printed again from manuscript copy.

It will be recalled that Sir Walter Greg's theory had it that after the point in the text which coincides with the end of Q1 sheet B the copy for Q2 consisted of a transcription of a playhouse manuscript—"though," Sir Walter declares, "I will not say that [the scribe] may not have used other fragments of the printed text, and I am certain that he consulted it on occasions when the manuscript was obscure."[10] We are now concerned with the question—what was the nature of the copy for Q2 after that point in the text which coincides with the end of Q1 sheet B?

In 1919 Messrs. Pollard and Dover Wilson produced a set of articles on the Shakespearian bad quartos.[11] In these articles they discussed the origin of the bad quarto texts, and they advanced a highly ingenious theory. It was a very gallant attempt at explaining a very puzzling group of texts, and it was based on some acute observation. But it was wrong, as Mr. Dover Wilson is now the first to admit. Yet, as regards Romeo and Juliet, in enunciating and endeavouring to uphold their theory, Messrs. Pollard and Dover Wilson noted something which is undeniably true—that there are bibliographical links between the two quartos in two passages subsequent to the end of Q1 sheet B. These links are to be found at II iv 36-43 and III v 26-35. The passages occur in Q1 on leaves E1 and G3 respectively. In these two passages, admittedly short, there is between the two quartos a remarkable degree of agreement in the use of initial capitals and in the placing of colons; yet there are verbal differences in each case. Mr. Thomas suggests[12] that in these two passages we have examples of the Q2 editor consulting Q1 because the manuscript he was copying was illegible. I cannot myself think that this is a very likely explanation here. Consider the first of the two passages, II iv 36-43. The two texts run as follows:

Mr. Thomas's theory assumes that when he came to the words "Without his Roe, like a dryed Hering" Scribe E found his manuscript difficult to read. So he looked at Q1. He copied these words from Q1, reproducing Q1's capitalisation. He then proceeded with the speech, copying Q1, copying its capitals and its colons. But he must have had his eye on his manuscript too, for he made changes in the text of the Q1 passage. Of course, it might be suggested that not all the alterations are his work. Perhaps in a given case the Q2 compositor made an alteration owing to carelessness. But we could not possibly hold that the Q2 compositor was responsible for all the changes in this short passage: there are too many changes in too short a passage to make that in the least likely. For at least some of the changes Scribe E must be responsible. But again, it might, I suppose, be suggested that Scribe E, copying Q1 here because his manuscript was illegible, altered Q1 memorially. That is certainly possible. But there is no positive evidence to indicate this hypothesis here, and it would, I think, be unsafe to have recourse to it. I think we must assume that some, if not indeed all, of the Q2 alterations in this passage came from the authentic manuscript used in the preparation of the copy for Q2. Now what of the suggestion that in this passage Scribe E consulted Q1 because the manuscript he was copying was illegible? If we accept that suggestion we must, I think, be prepared to take the following, for example, as sufficiently likely to be acceptable. Consider the portion of text beginning with the words "Laura to his Lady". Owing to illegibility in the manuscript before him, Scribe E applied himself to the task of copying out of Q1 the words "Laura to his Lady was but a kitchin drudg, yet she had a better loue to berime her:". He followed Q1 sufficiently closely to reproduce the capitalisation and the colon; but at the ninth and tenth words the manuscript before him suddenly became sufficiently legible to show him that instead of Q1's "drudg" he must write "wench", and instead of Q1's "yet" he must write "marrie". Having taken these words from the manuscript which was allegedly difficult to read, he returned to Q1, continuing to accept its capitals and its colons, but introducing other verbal alterations from the manuscript. I cannot say that this appears to me a very likely theory. I should rather say that the most reasonable explanation of the relationship between the two

quartos in this passage is that Q2 was printed from Q1 itself, the latter having been corrected by Scribe E by comparison with his manuscript (which was not illegible).

Of course, if any critic feels that the agreements in capitalisation and punctuation between the two quartos in the passage we are dealing with can be explained as simply a matter of coincidence, then he will not be much impressed by the above suggestion. I myself cannot regard them as a matter of pure coincidence. To postulate pure coincidence would be to strain credibility too far. For one thing, in the passage we have been examining there is a remarkable agreement between Q1 and Q2 in the matter of the setting up of proper names in italic or roman type. It is not only that the same words have initial capitals in both texts, and that the colons appear in exactly the same places: we also have to reckon with the fact that in both texts all the proper names in the passage are italicised except three—the same three in both texts—namely, Petrarch, Dido, and Cleopatra. In Q1 the first letter of "Dido" is in italic, the other three in roman: in Q2 all four letters of the name are in roman: but it is an essential agreement. I am sure that the name "Dido" appears in roman in Q2 because it does (essentially) in Q1, and that the names "Petra (r) ch" and "Cleopatra" appear in roman in Q2 because they do in Q1. Q2 surely depends directly on Q1 in this passage; but the Q1 passage has been altered editorially.

I have not the space to examine the other passage I referred to—III v 26-35. I can only say that a comparison of the two versions of it seems to me to indicate the same conclusion.

Now, having decided that in these two short passages Q2 is dependent on an edited copy of Q1, we naturally look carefully at the two leaves of Q1 involved—E1 and G3. We look at each in its entirety. And I should claim that it is possible, in pen and ink, to correct these two Q1 leaves in their entirety so as to give the Q2 text. I think it reasonable, then, to suggest, and I do suggest, that part of the copy for Q2 consisted of leaves E1 and G3 of Q1, corrected by hand in accordance with an authoritative manuscript, and then torn out of the particular first quarto used, and placed in the bundle of papers that formed the copy for Q2.

It may be suggested that certain other leaves of Q1 were also

torn out and, in an edited state, used as copy for Q2. I should like to suggest that Q1 leaf D1 may well be a case in point. This leaf includes the passage II ii 40-43. In Q1 it runs— In Q2 we have— The Q2 version seems hardly satisfactory. There is nothing necessarily suspicious about the fact that in Q2 the half-line "Belonging to a man" stands as a metrically incomplete line. But some readers—and I am among them—consider that the phrase "some other name Belonging to a man" seems rather odd. I suggest that Q2 here depends on the Q1 page corrected by Scribe E. He wanted to indicate on his Q1 page the following version: To indicate this, he had to insert a complete line into his first quarto—"Belonging to a man. O be some other name.". I suggest that he put in the words "Belonging to a man." directly under the words "Nor arme, nor face,". I suggest that he inserted the words "O be some other name." in the right-hand margin. He drew a line connecting the beginning of this right-hand marginal insertion to the end of the interlinear insertion. This connecting stroke accidentally scored through the words "nor any other part", and the result was that the Q2 compositor thought that the words "O be some other name" were intended to replace the words "nor any other part", and he thought that these words, "O be some other name", were intended to be followed by the interlinear insertion, "Belonging to a man". The conflation, traditional since Malone, is I believe, correct. This may perhaps be regarded as tenuous evidence for the theory that here Q2 depends on an edited leaf of Q1. But that seems to me the best explanation of the textual conditions. The whole of Q1 leaf D1 can be corrected by hand in such a way as to produce the Q2 text—there is sufficient room for all the necessary alterations.

I have suggested that part of the copy for Q2 consisted of leaves D1, E1, and G3 of Q1, corrected by Scribe E in the light of an authentic manuscript which was part of his equipment. Here I would mention the Prologue to the play. The text differs substantially in the two quartos, but there are similarities in the setting up. Both versions are set up in italics with a large initial letter in line 1; and both have brackets round line 2. I take it that this is a case of a page printed in Q2 from an edited copy of Q1. The Q1 text would have had to be altered extensively: but there is plenty of room on the page in Q1, even if the Prologue were to be almost completely rewritten; and I imagine that that was done.

In dealing with the text up to I iii 100 (the end of Q1 sheet B) we contended that part of that stretch was in Q2 printed not from Q1, edited or unedited, but from manuscript copy. It is abundantly clear that at certain other points, later on, Q2 was again set up from manuscript copy. There are points where we can say with confidence that there would not be nearly enough room on the Q1 pages for the insertion of the necessary alterations and additions.

It should be noted, however, that it is possible that at a given point, writing out manuscript copy, Scribe E may have consulted Q1 and imported into his transcription a Q1 reading. A case in point occurs, I believe, at II i 13. The two texts of II i 10-14 run as follows:

(In Q1 the passage occurs on sig. C4v.)

Here, I think, Q2 was printed from manuscript copy. The

word "prouaunt" is a misreading of "pronounc (e)" handwritten, and "day" is a misreading of "doue" handwritten. Now we can easily imagine Scribe E correcting the printed Q1 in the light of his authentic manuscript—misreading the word "doue" in the manuscript as "day"—scoring through the perfectly correct "Doue" in Q1, and substituting "day". But take the Q2 "prouaunt", a non-existent word. Suppose that Scribe E, correcting Q1, comes to the word "Pronounce" in it. He looks at the authentic manuscript. It is badly written, and it seems to him to read "prouaunt". But that is a non-existent word. Can we suppose that he drew his pen through the familiar word "Pronounce" in Q1, and substituted the nonsense-form "prouaunt"? It is possible, certainly, but, I think, hardly very likely. It is more probable, in my opinion, that, thinking at first that the manuscript read "prouaunt", and not recognising that as an existent word, he would be guided by Q1 to pause, examine the manuscript word very carefully, and realise that in fact it was "pronounce." He would not, then, alter Q1. But if the Q2 compositor were at this point setting up from handwritten copy, he might easily decide that a badly written "pronounce" was in fact "prouaunt", and so he would set that up. Elizabethan and Jacobean compositors sometimes did set up nonsense-words, owing to misreading of manuscript copy. For instance, in Q1 King Lear (1608), at III iv 119, the compositor originally set up "nellthu night more" instead of "mett the night mare". What he set up can have meant absolutely nothing to him: yet he set it up, assuming that he was following his copy. So, I imagine, as regards "prouaunt" in Q2 Romeo. I doubt if Scribe E would do this sort of thing. Later on in this paper it will appear that I think of him as a member of the company of actors to which Shakespeare belonged. Doubtless he was an actor. He would not cancel "Pronounce" in Q1 and substitute the absurd "prouaunt". I think that "prouaunt" represents a misreading by the Q2 compositor of manuscript copy written out by Scribe E. Scribe E wrote "pronounce" and the Q2 compositor misread it. Yet the Q2 "Abraham: Cupid" doubtless comes straight from Q1. "Abraham" may or may not be correct; but the curious colon between "Abraham" and "Cupid" certainly constitutes a link between the two quartos which, to anyone accepting the position taken up in this paper, must in Q2 indicate direct derivation from Q1, presumably by Scribe E. In this passage, it seems to me, Q2 depends on manuscript copy written by Scribe E: the latter was copying an authentic manuscript; but for "Abraham: Cupid" he was indebted to a consultation of Q1.

At certain points, then, I think Scribe E furnished the Q2 compositor with copy in the form of leaves of Q1 corrected by comparison with an authentic manuscript. At certain other points he furnished him with copy which consisted of his own transcription of the same authentic manuscript.[13] What was the nature of this authentic manuscript which underlies Q2?

In all probability it was a manuscript, in Shakespeare's handwriting, which had not itself been used as a prompt-book. In Q2 we find that in stage-directions and speech-headings Lady Capulet is referred to by half a dozen different titles (in full or abbreviated) — Wife, Capulet's Wife, Lady, Old Lady, Lady of the House, Mother. The late R. B. McKerrow pointed out that this sort of thing would tend to confuse a prompter, and would not be likely to appear in an official prompt-book.[14] In a prompt-book a given character would tend to be indicated consistently by one designation. A diversity of titles would, however, be quite natural in a playwright's own manuscript. At one point Shakespeare, thinking of the character in her relationship with the heroine, would call her "Mother"; at another point, thinking of her in her relationship with Capulet, he would call her "Wife"; and so on. Uniformity of designation would probably be substituted for this diversity in a transcription of the author's manuscript (with adaptation) which would constitute the prompt-book. Other suggestions that the manuscript behind Q2 was an author's manuscript which had not itself been used as a prompt-book are to be found in the fact that some necessary indications of entrances and exits are wanting in Q2, and in the fact that on occasion Q2 is capable of printing two different versions of the same passage.

To sum up as regards the nature of the copy for Q2: the person who produced it, Scribe E, furnished the compositor with a

pile of papers, some printed, some handwritten. The handwritten leaves were probably transcribed from a Shakespearian autograph manuscript. The printed leaves were torn from a copy of Q1. Some of them Scribe E corrected by comparison with the same Shakespearian autograph manuscript. As regards one passage, contained in two contiguous leaves of Q1, he left the Q1 text unaltered, contenting himself with the cancellation of the material which in the Q1 leaves preceded and followed the passage he accepted. Thus behind Q2 there lie three different types of copy—(i) Q1 unaltered, (ii) Q1 edited by comparison with a Shakespearian manuscript, and (iii) direct transcription of that Shakespearian manuscript.

I should greatly like to discuss Q1 at length, but I must be brief. I have no doubt that Mr. Hoppe is in most respects correct in his recent book about this text.[15] His book is a most valuable contribution. The Q1 text is full of the kinds of error that we associate with reported texts—anticipations and recollections, inversions, vulgarisations, weak synonym-substitutions, paraphrase, metrical breakdown, and so on. Mr. Hoppe has marshalled much of this evidence very impressively. This or that individual example of corruption, we may say, might in itself be scribal or compositorial. But there are far too many egregious memorial corruptions to make it possible to avoid the formula of memorial reconstruction. The play has no sooner begun than we have in Q1 a remarkable example of the absurdity in which memorial corruption can on occasion result. In the first line of the Prologue, Q2 speaks, sensibly, of "Two housholds". Q1 has, instead, "Two houshold Frends", which in the context clearly makes nonsense. There is little doubt in my mind that behind the absurd reading of Q1 there lies a reporter's memory of a much later passage—III i 161. From the authentic version of the Prologue the reporter remembered "Two housholds", and the fact that enmity between them was involved. His mind sprang forward to III i, where, describing an affray between the two enemy households, Benvolio speaks in Q2 of how Romeo called aloud "Hold friends, friends part". The syllable "hold" constituted a memorial link which had the effect of bringing the word "friends" from Act III into the Q1 version of the first line of the Prologue,

with nonsensical results. The Q1 text is not always so bad as this. There is some good reporting. Indeed Mr. Thomas, noting that "the first two acts and part of the third are remarkably well reported", has stated that he is inclined to believe that "the pirate had access to a manuscript there".[16] But even where the Q1 text is at its best, it seems to me that there is rather too much memorial corruption to allow of the theory of direct derivation from an authentic manuscript; and I am content with the theory of remarkably good reporting at some points, made up for by some exceedingly bad reporting elsewhere, notably towards the end of the play.

Since in Q1 we are dealing with a memorially transmitted text, we must be very careful before we say that a given passage in Q1 represents a Shakespearian first draft, and the corresponding Q2 passage a Shakespearian revision. We cannot say this unless at a given point we are confident that the Q1 version can hardly be a corrupt version, indebted for its phrasing to the reporter or reporters. Now I confess that there are just one or two cases (no more than that) where Q1 contains a word (different from the Q2 reading) which seems to me rather unlikely as a reportorial substitution. For instance, at I iv 56 (in the "Queen Mab" speech) Q1 has "Burgomaster", Q2 "Alderman". I am not sure that the word "Burgomaster" would be likely to be substituted by a reporter on his own initiative. It could have been, I dare say. But some people might suggest rather that Shakespeare originally wrote "Burgomaster" and then, in a revision, changed it to "Alderman", that being a word more familiar to English audiences. Yet, if Shakespeare did originally write "Burgomaster", it may well be that "Alderman" is a change made, not by him, but by Scribe E (or someone else), intent on making the text fully clear to English readers. At no point in the play can I say that it seems to me necessary to suppose that there was any Shakespearian revision between the texts represented by Q1 and Q2.

The theory that Q1 represents, accurately or inaccurately, a Shakespearian first draft has been not infrequently advanced by critics in the past. There are passages in which the two quartos diverge completely or almost completely, and such passages might

at first sight tempt one to entertain the revision theory. But I am greatly attracted by Mr. Hoppe's suggestion[17] that the passages in Q1 which diverge widely from the phrasing of Q2 are to be explained as the work of a reporter who had at one time known the full version of the play—who had more recently known an abridged version, with these passages excised, and who therefore had a much dimmer memory of them than of the rest of the play— and who was in them, therefore, thrown back on his own phraseological invention more than elsewhere.

Mr. Hoppe in his book suggests the possibility that Henry Chettle may be regarded as a "candidate for the office of reporter-versifier".[18] Independently of Mr. Hoppe, Mr. Thomas has also advanced the view that certain of the passages in Q1 which diverge radically from Q2 are the work of Chettle.[19] I cannot discuss the evidence here. Suffice it to say that I think it very possible that Chettle may have been involved in, for example, the Q1 version of II vi. But, if he was, he must be regarded as just as liable to be called a "reporter" as the person who produced the non-Shakespearian verse in the bad quarto of Hamlet: for, at certain points at any rate, the same procedure is followed. As Mr. Hoppe showed so well in 1938,[20] parts of Q1 II vi are to be explained as mosaics of recollections of different passages scattered throughout the authentic text of the play. Take for example the following lines in the Q1 version of II vi—

Here Juliet refers to herself as "Day" and to Romeo as the "Sunne". I believe that the person who wrote this passage had at the back of his mind two passages in other scenes in the Q2 version, and he inverted the Romeo and Juliet roles in both cases. In Q2, at II ii 3, Romeo, seeing Juliet at her window, exclaims "Iuliet is the Sun". In the Q1 passage under discussion Juliet refers to Romeo as her "Sunne". Then secondly, at III ii 17 in Q2, Juliet, in soliloquy, exclaims "Come night, come Romeo, come thou day in night". Here Juliet refers to Romeo as "day", and, in the passage in Q1 under discussion, she refers to herself as "Day". I venture to suppose that there can be little doubt that in the passage under discussion in Q1, which here diverges completely from the corresponding Q2 text, we have to deal with the work of some person who, knowing the play, brings together reminiscences of two widely separated passages, and combines them with his own invention into new composition. We find this sort of thing in the bad quarto of Hamlet: we find it in The Taming of A Shrew:[21] we find it in other reported texts also. It is a not unusual kind of reporter-work. Chettle may have been responsible for it here. If so, as I say, Chettle is as much a "reporter" as the actor of the parts of Marcellus and Lucianus in Hamlet who produced the copy for the bad quarto of that play.[22]

Mr. Thomas thinks that Q1 II vi, after the entry of Juliet, is from the aesthetic point of view good, and indeed even better than the Q2 version. Speaking of the final line and a half of our last quotation from Q1—

Grant White avowed that they have "a touch of poetry more exquisite and more dramatic than is to be found in the" corresponding Q2 version.[23] Yet I am convinced that the explanation I have given of the genesis of this line and a half of Q1 is correct. If it is not, then we have to reckon with a couple of remarkable coincidences.

A moment or two ago, I spoke of the theory that a Shakespearian revision lies between the texts represented by Q1 and Q2, Q1 giving us the text of a Shakespearian first draft. I do not believe this theory. Now in connection with the revision theory, I should mention, briefly, that there is of course a case for regarding not Q1

but Q2 as giving the first version, and for regarding Q1 as giving a reflection (admittedly distorted, but at a given point possibly accurate owing to good reporting) of a Shakespearian revision. Mr. Hoppe refers to this possibility in the introduction to his "Crofts Classics" edition of the play.[24] He thinks of only two lines, present in Q1 but not in Q2, as possibly "a later insertion by Shakespeare"—the lines containing the phrase "without booke Prologue" (I iv 7-8). But of course, regarding these lines as genuine (on account of their literary quality), or as reflections of genuine lines, we may easily suppose their omission from Q2 to be accidental. I know of nothing which positively indicates a revision theory either way.

Having spoken of the nature of the copy for Q2, and of the nature of the transmission of the Q1 text, let us now consider the final question of what is to be the procedure of a modern editor of this play. I am at present engaged on the production of a modern-spelling edition of the play. That being so, and my space here being limited, I shall concentrate on the problems facing a modern-spelling editor, leaving out of account the special problems confronting an old-spelling editor.

Let us take first the passage in which Q2 is dependent on Q1 as it stands without interference by Scribe E (I ii 57-I iii 36). Q2 here ranks as a reprint of Q1. In fact in this passage Q2 is not substantive at all, but derivative. Not a single alteration in Q2 in this passage need be referred to anyone other than the Q2 compositor. Clearly, then, where Q2 contains a reading different from that of Q1, we must accept the Q1 reading, unless in a given case we feel inclined to accept that of Q2 as a desirable conjectural emendation.

Let us now briefly glance at two modern editions and see what they do as regards the four verbal variants between Q1 and Q2 which we enumerated earlier in this paper. The two modern editions we shall look at are the old Cambridge edition and Mr. Hoppe's recent "Crofts Classics" edition. (They are both modernspelling editions.) Both follow Q2 in adopting the phrases "an hour" (I iii 12) and "with the dug" (I iii 33) instead of "a hour" and "with dug" (Q1). They are right to do so. In these two cases

the Q2 compositor has, on his own responsibility, made necessary corrections in Q1. Next: as regards the line which in Q1 runs "my faire Neece Rosaline and Liuia," (I ii 68). Q2 lacks the "and". The old Cambridge edition, wrongly, follows Q2 and omits the "and". Mr. Hoppe, rightly, follows Q1 and inserts the "and" (printing the passage, correctly, as verse). Finally: as regards the variant Q1 "thee" Q2 "you" in I ii 77. Both the old Cambridge edition and Mr. Hoppe follow Q2 and read "you" instead of "thee". Now I regard the Q2 "you" as purely compositorial, and I confess that I can see nothing in the least objectionable about Q1's "thee". The Q2 compositor's "you" is not, in my opinion, a necessary emendation. I should read "thee". Mr. Hoppe doubtless had a reason or reasons for reading "you", but any possible reasons that occur to me can, I think, be successfully refuted. It is a small point, of course; but it is a matter of principle.

In this passage, then, circumstances have resulted in our having only one substantive text, and in our having to include in our editions of Romeo a bad quarto passage of not inconsiderable length—material which at one stage has been transmitted purely memorially. Obvious errors we can emend. Corruption is not, however, always self-evident: and at a given point in this passage it is possible that Shakespeare's intentions are irrevocably lost, and that our texts will remain corrupt to the end of time.

Nowhere else in the play does an editor have to reckon with the same state of affairs as in this passage. Let us leave it now, and concentrate on the rest of the play—the great bulk of the play.

I have suggested that some other leaves of Q1 were torn out and, in an edited state, used in the copy for Q2. As regards these passages we have in Q2 a good text which stands in the same relationship to an antecedent bad text as does the folio version of King Lear. The folio text of Lear was printed from a copy of the quarto which had been corrected by an editor.[25] Now, as Sir Walter Greg has pointed out,[26] there is in such a case a greater presumption of genuineness in a good text reading where it differs from the bad text reading than where it agrees with it. That is true; but I want to emphasise that even where the good text reading differs

from that of the bad text, we must carefully scrutinise the pair of variants and consider whether the good text reading does in fact seem to us to be the right one.

I have said that I think that part of the copy for Q2 Romeo was leaf D1 of Q1, corrected in handwriting. Now at one point in Q1 leaf D1 (on the verso side) we have the following passage:

The Q2 text agrees with this verbally in the main, but for the Q1 "lasie pacing cloudes" Q2 has "lazie puffing Cloudes". Now at first thought one might suggest the following: the Q2 text was printed from an edited copy of the Q1 leaf—the corrector, Scribe E, altered the Q1 "pacing" to "puffing" because "puffing" was the reading of his authentic manuscript—and so we must accept "puffing" as the genuine reading. But I cannot do this: I think that Q1's "pacing" is the correct reading.

The image of lazy pacing clouds is a very good image. Now it may be said that the image of lazy puffing clouds is also a good image. One can, poetically, imagine a cloud, in a lazy mood, puffing from one point to another, pausing there, lazily, and then after a time puffing on to yet another point, and so on. But I believe that the context shows that Shakespeare is thinking of a much smoother motion than that would necessitate. "When he bestrides the lazy—something—clouds And sails upon the bosom of the air." Sailing upon the bosom of the air suggests a very smooth motion. There is something awkwardly incongruous in proceeding from lazy puffing clouds to smooth sailing on the bosom of the air. I repeat: consider the passage carefully, and it must surely seem that Q1's "pacing" (a smooth enough motion) fits in better with the context than does Q2's "puffing". The Q2 "puffing" suggests something rather jerky, which jars in a context otherwise devoted to smoothness.[*]

Now, assuming that Q2's "puffing" is corrupt, I believe it is possible to show how the corruption arose. I think it is a memorial corruption. At I iv 96 Mercution talks of dreams which are begot of nothing but vain fantasy; and this vain fantasy is

That is the Q2 version; but the word "puffes" occurs in the passage in Q1 also. This word, with its suggestion of a jerky movement, is entirely appropriate here, for the wind is angered. We have the notion of a petulant, jerky movement: it is admirable.

Now we are dealing with a passage in a later scene and a passage in an earlier scene. In the passage in the later scene I suspect that the Q2 word "puffing" is wrong. In the earlier scene the word "puffes" occurs (in both texts, note) with admirable appropriateness. I think that the presence of "puffing" in the later scene in Q2 is due to Scribe E having remembered "puffes" from the earlier scene. And I can claim that it is very likely that the two passages, the one in the later scene and the one in the earlier scene, could have been associated in the mind of a scribe or editor who knew the play;[27] for there is a clear phraseological link between the two passages: in the later passage we have the phrase "the bosome of the ayre", and in the earlier passage we have the "ayre", and, two lines later, the "bosome of the North". This suggests that the Q2 "puffing" is not a straightforward misreading of the word "passing" in handwriting, as, in his "Crofts Classics" edition, Mr. Hoppe apparently considers it to be.[28]

The passage involving the "lasie pacing cloudes" occurs in

Q1 on leaf D1. I think that Q2 was set up from copy which included Q1 leaf D1 corrected by comparison with an authentic manuscript. But, if I am right, in our passage Q1 already contained the true reading. The corrector, Scribe E, must have substituted "lazie puffing Cloudes", not because his manuscript read that (for it did not), but because he himself had an impression that that was the right reading. He was wrong: his impression was based on faulty memory. But he had that impression, and it induced him to alter Q1's perfectly correct reading, and substitute a wrong reading. If his authentic manuscript was at this point difficult to read he would have some excuse for relying on his own (admittedly defective) knowledge of the play. Perhaps it was.

As regards passages of the play, then, in which Q2 was printed from edited leaves of Q1, what is a modern-spelling editor to do? Where Q2 agrees with Q1 he must be watchful. In a given case Scribe E may have negligently failed to make a necessary correction. The modern editor must be prepared, where he considers it necessary, readily to emend a reading in which the two quartos agree. He will of course accede to the necessity of doing so where the reading in question in itself gives evidence in its context of being corrupt. Where there is no self-evident corruption, the modern editor will doubtless accept readings in which the two quartos agree. But corruption is not always self-evident, and the modern editor will agree that in a given case in which he accepts a reading exhibited by both Q1 and Q2 he may in fact be accepting a non-Shakespearian reading. It is not possible to produce a text which we can guarantee to be free from corruption.

Now, as regards passages printed in Q2 from edited leaves of Q1, the modern editor must, where Q2 disagrees with Q1, assume that the Q2 reading comes from the authentic manuscript which Scribe E had as part of his equipment, unless in a given case there seems a strong enough reason for supposing that the Q2 reading is an error introduced by Scribe E or by the compositor. If a Q2 reading which differs from Q1 does appear to a modern editor, for good reason, to be corrupt, he must decide whether to accept the Q1 reading or whether to propose a reading differing not only from Q2 but also from Q1. There are various possibilities. Scribe E may have altered a perfectly correct Q1 reading owing to his

having misread the same word in the authentic manuscript before him. Scribe E may have failed to decipher the manuscript, and, through faulty memory, on his own responsibility, he may have altered a perfectly correct Q1 reading. Scribe E may even on his own responsibility have sophisticated a reading in which Q1 and his authentic manuscript agreed. Scribe E may have altered an erroneous Q1 reading by substituting, in the Q1 page, a misreading of a different word in his authentic manuscript. Scribe E may have replaced an erroneous Q1 reading by a word which he wrote so badly that the Q2 compositor misread it. All these various possibilities add up to this—that the modern editor must scrutinise each variant as a case in itself. He will be prejudiced in favour of Q2. If no reason for rejecting Q2 appears to him, he will accept Q2. But he must always be prepared to accept a Q1 reading, if good reason can be shown. A reporter may, by chance, correctly reproduce a Shakespearian reading which the authentic transmitters corrupt.

Finally, at certain points Q2 was printed from a transcript of Shakespeare's manuscript, and is independent of Q1 except for possible occasional consultation of Q1 by Scribe E or the Q2 compositor. Apart from such cases of possible consultation, agreement between the two quartos is now a presumptive guarantee of authenticity. In a given case of agreement here, we may say that Scribe E, copying Shakespeare's manuscript, derived a certain reading from it. Quite independently, the Q1 reporter, relying only on his memory, conveyed the same reading. Scribe E and the reporter thus corroborate each other. And yet really the position is not so good as that. For there is always that troublesome factor of possible occasional consultation of Q1 by Scribe E or the Q2 compositor. Any given agreement between Q2 and Q1 might be explained as a case of such consultation. Thus a modern editor should still scrutinise every word of the text, where the two quartos agree, in order to determine whether emendation is desirable.

Next: where Q2 (set up from manuscript copy) differs from Q1, we must assume, unless in a given case there is any reason against doing so, that the Q2 readings are derived from the Shakespearian manuscript, and are thus genuine. This is so. Yet we must still reckon with the possibility that in a given case the Q1

reporter's memory was good, so that he correctly remembered the Shakespearian reading, whereas Scribe E or the Q2 compositor corrupted it.

Just as on one occasion when he was correcting Q1 Scribe E put in a reading ("puffing") which was in fact a memorial corruption, so I believe he did on one occasion when he was transcribing the Shakespearian manuscript. I think it would be difficult to suppose that Q1 leaf D3 was corrected by hand to form copy for Q2. I think that here Q2 depends on manuscript copy. Now at the end of II ii (on that leaf) Romeo says, according to Q1, "Now will I to my Ghostly fathers Cell". In Q2 he says "Hence will I to my ghostly Friers close cell". I am concerned with the latter half of the line. The Q2 version seems to me odd. The words "ghostly Friers" are awkwardly tautologous. Sir Walter Greg agrees that the Q2 line is corrupt. He says that its ending "can only have originated, one supposes, in inability to decipher the playhouse manuscript"[29] i.e., in our belief, the Shakespearian autograph manuscript. Whether it was a matter of Scribe E failing to decipher the authentic manuscript or not (and it may well have been), I believe that the words he penned in his transcript owed their existence in his mind to a memorial anticipation of V iii 254, where the same Friar referred to speaks of having meant to keep Juliet "closely at my Cell" (Q2 only).[30]

If, in a portion of the play printed in Q2 from manuscript copy written out by Scribe E, a modern editor can argue cogently that a given reading in Q1 is good, and that the corresponding reading in Q2 is corrupt, and if he can explain how the allegedly corrupt reading in Q2 might have arisen in the light of his general theory

of how the Q2 text was transmitted, then he is, I think, entitled to adopt the Q1 reading, unless it can be suggested that a better case can be made out for a third reading, consisting of an emendation of the Q2 reading to a reading different from that of Q1.

It seems to me, then, that in editing Romeo one has to have Q1, an edition of little authority in general, on one's desk all the time, as well as Q2, an edition of very much greater authority in general. I do not think that, editing the play in accordance with the principles I have been discussing, one will find oneself introducing very many readings from Q1—just a handful, I imagine. The Q1 text is in the main a manifestly inferior text. But I submit that as one proceeds with the work one must be prepared to consider the adoption of a Q1 reading in a given case, even though Q1 is a bad quarto. If one adopts a Q1 reading in preference to a Q2 reading, one must be prepared to state strong and cogent reasons for so doing. These reasons may in a given case be literary. I do not think that a textual critic can ignore literary considerations. But I repeat—the reasons must be strong and cogent. Yet, if there are strong and cogent reasons, the modern editor should not ignore them on grounds of timidity.

Notes