University of Virginia Library

The Manuscripts of Dryden's The State of Innocence and the Relation of the Harvard MS to the First Quarto
by
Marion H. Hamilton

In the Author's Preface to The State of Innocence, Dryden explains that he is publishing an opera in his own defense—"many hundred Copies of it being dispers'd abroad without my knowledge or consent: so that every one gathering new faults, it became at length a Libel against me."[1]

Of the "many hundred Copies" the following five, listed here with the sigils that will be used in this study, are now extant:

• H MS Thr. 9, Harvard University
• M Add. MSS 37, 158, British Museum
• B Rawl. C. 146, Bodleian Library
• Hn1 El. 11650, Huntington Library
• Hn2 Rare Book 134219-29, Huntington Library

Although all of these manuscripts have been examined by scholars, a complete comparative study to determine their derivation from the non-extant archetype and their relative authority has not heretofore been made.[2]

All of the MSS are scribal transcripts and all, except for Hn2, which is in the same hand as Hn1, were made by different scribes. The Harvard MS, however, has the distinction of containing certain corrections and emendations made

by Dryden himself, notwithstanding the fact that the first two pages are incomplete.[3]

Textual collation of the manuscripts, in which Q1 served as the control for the study, demonstrates the following to be the ancestral tree:

The evidence in support of this tree is both statistical and analytical and will be given in that order, beginning with the following Table of Substantive Variants.[4]

                           

MS[5]	Unique readings	Readings shared with 1 MS	Readings shared with 2 MSS	Total shared with individual MSS	Total shared reading
Harvard	43	M 4	B & M 1	M 5	12
		B 2	Hn & B 4	B 7
		Hn 1	Hn & M 0	Hn 5
British	317	H 4	Hn & B 66	H 5
Museum		B 30	B & H 1	B 97	120
		Hn 19	Hn & H 0	Hn 85
Bodleian	181	H 2	H & M 1	H 7
		M 30	Hn & H 4	M 97	179
		Hn 76	Hn & M 66	Hn 146
Huntington	219	H 1	H & B 4	H 5
		M 19	H & M 0	M 85	166
Hn1	7	B 76	B & M 66	B 146
Hn2	24

On statistical evidence alone, certain probabilities are immediately apparent. First, the Harvard manuscript, with only 43 unique readings and 12 shared unique readings, [6] bears the closest relation to Q1. Since its relatively few shared unique readings are distributed almost evenly among the other MSS, there is no reason to assume either that H derived from any of them, or that it bears a closer relation to any one of the three MSS than to the remaining two. Instead, it is clearly related to Q1, although whether in radiation from a common ancestor or in a straight line of descent cannot be determined by statistical evidence.

Second, the large number of unique readings in M, B, and Hn suggest that each is a terminal text, although the 31 variants that exist between the two Huntington MSS indicate that they are terminal only in relation to the other MSS since the variation between them suggests, but cannot prove until its content is examined, that one of the Hn MSS may have derived from the other.

Third, the agreement of 66 readings in B, M, and Hn against H and Q1 suggests the possibility that each of the two groups may have derived from an exclusive common ancestor.

Finally, the agreement of 76 readings in B and Hn against readings supported by all other texts suggests the possibility that B and Hn may owe these readings to a later ancestor than the one they presumably share with M.

Since statistical evidence alone cannot prove any of these assumptions to be true, the line of descent can only be established with certainty by an analysis of the readings themselves. The results of this, too detailed to elaborate here, can be summarized under the heading of each manuscript as follows:

Harvard MS

The fragmentary state of the first two pages of this MS precludes the possibility that any other text derived from it. Furthermore, although many of the 43 unique readings are correctable scribal errors, others, consisting largely of omitted words, are not obvious errors and would have been followed by any MS descending directly from H. Of the 12 readings that H shares with one or more manuscripts, all are of minor significance since they involve only a simple variance such as the change from singular to plural. In no instance does an analysis of the readings indicate that H is any closer to one MS than to another. One can assume, however, since it is known that Dryden revised the text, that, apart from the first two pages, it represents a good version of the archetype.

British Museum MS

The 317 unique substantive readings of the MS include the omission of six lines and 11 words that appear in all other MSS. On this basis, therefore, it is evident that M is a terminal text. Of the 66 readings shared by M, B, and Hn, although many of them are minor variants of no significance, the majority involve a more meaningful change that precludes coincidence and indicates the

existence of an exclusive common ancestor for the three MSS. The following three selected readings, with Q1 providing the lemma and the variant italicized, will illustrate:

• 1. this new created Orb] new made (II.i.30.)
• 2. Enjoy and blast her] blesse (III.i.95.)
• 3. with early care] cheerful (V.i.381.)

That each of these readings is coherent in its variant form and would have been accepted by any scribe finding it in his exemplar is self-evident. It is obvious, then, that they must have originated in the manuscript that served as the exclusive common ancestor for M, B, and Hn. The identity of that ancestor, however, must be determined on the basis of two possible alternatives: (1) the readings originated in the archetype and represent the author's intentions; or (2) they originated in an non-extant MS as corruptions of the archetype and were retained by all descendents of that text. If the first possibility is accepted, the 66 readings as they are in H and in Q1 become corruptions of the original text. Since, however, the majority of the readings shared by M, B, and Hn against H and Q1 destroy the point of the line or upset the metre, this alternative must be rejected in favor of the second. On this basis one can postulate the hypothetical manuscript β, the originator of the 66 variant readings and the exclusive common ancestor of the family of M, B, and Hn.

In addition to the 66 variants shared by the three MSS, it will be remembered that statistical evidence shows further agreement between M and B of 30 shared unique readings, and 19 between M and Hn. Examination of the readings, however, proves them to be without significance since they are either minor variants such as "over" for "o're", or readings that have further variation in the other MSS that obviates their importance.

Bodleian MS

Among the 181 unique substantive readings of B there are three hiatuses, and 15 words and one speech prefix omitted from the text but supplied by all other MSS. These omissions, together with the many unique readings that are sufficiently coherent in B to have been accepted by any derived text, offer conclusive evidence that it is a terminal MS. With the family relationship of B, M, and Hn already established, there is still the relationship between B and Hn to consider. This statistical evidence indicates, on the basis of 76 shared unique readings, might be closer than that of the family directly descended from β.

Examination of the shared variants shows that they include one hiatus, the omission of four words and one speech heading supplied by M, H, and Q1, and one stage direction that differs from the other texts. Since B is a terminal manuscript, and, as will be shown later, Hn is also terminal in so far as the other MSS are concerned, none of these variants could have originated in either B or Hn and descended to the other. Three examples, selected from over 50 possible ones, will illustrate that these variants could only have originated in a non-extant MS from which only B and Hn derived:

• 1. I shall turn] wing (II.i.41.)
• 2. to smell the flavour] odour (III.i.62.)
• 3. for on this ground] I from (IV.ii.44.)

Although the remaining 22 agreements between B and Hn are minor variants of no significance, since the other 54 cannot be attributed to coincidence without defying common sense, the hypothetical MS γ is postulated as the immediate descendent of β and the originator of the significant agreements between B and Hn, which become a sub-family descended from γ.

Huntington MSS

The 219 unique readings in the two Huntington MSS include three hiatuses, the omission of 11 lines, 19 words, and one speech prefix, and the addition of 11 words not present in the other MSS. On this evidence, it is unnecessary to present further proof that no extant MS could have derived from them. Furthermore, since the relationship between Hn and the other MSS has already been established, it is now only necessary to restore the two Huntington MSS, until now considered as one, to their original status of two documents in order to demonstrate the descent of one from the other.

Statistical evidence shows 31 variant readings between them: seven unique in Hn1; 24 unique in Hn2. All that can be determined from these facts is that one descended from the other or that both descended collaterally from the same source. Analysis of the readings, however, demonstrates that they are not collateral but did, in fact, descend one from the other with Hn1 serving as the parent of Hn2. None of the seven unique readings in Hn1 is of significance since each is a correctable scribal error. On the other hand, among the 24 unique readings in Hn2 there is the omission of one whole line, four half-lines, and two words, as well as the addition of one word. Since all of these readings are supported not only by Hn1 but also by all other texts, it is evident that Hn2 is a terminal text. Had Hn1 and Hn2 derived independently from an exclusive common ancestor, the incidence of variation between them would have been far greater. Since Hn1 could not have descended from Hn2, it is reasonable to assume that the reverse obtains and that Hn2, although correcting or modifying the seven unique readings of Hn1, descended directly from the latter, adding 24 unique readings in the process.

With the derivation of the five extant MSS established, their relative authority is also indicated. Hn2, at the bottom of the manuscript tree, is four removes from the archetype; Hn1 and B, both descended from γ, are three; and M, descended from β, is two. Only H is directly descended from the archetype and becomes, therefore, the most authoritative of the five extant MSS. As such, the relationship of H to Q1 is of paramount importance.

Although the Harvard MS contains the least number of variants from Q1 of all the extant MSS, it did not serve as copy for the printed text. Instead, a manuscript now lost (presumably retained and destroyed by the printer) was transcribed under Dryden's supervision to be used by the printer. This manuscript, hereafter to be referred to as "q", derived independently from the archetype

which itself had been revised and augmented by the author for the purpose of publication.

The evidence found in H to support this hypothesis falls into the two categories of

• 1. Corrections and emendations in H.
• 2. Readings in H that differ from Q1 but are supported by all other MSS.

There are 66 corrections or emendations in H which can be divided into five classifications:[7]

• a. Scribal corrections obviously made during the process of transcription which bring the reading into agreement with all other MSS and Q1. 32 examples.
• b. Scribal and authorial corrections apparently inserted after transcription which bring the reading into agreement with the other MSS and Q1. 18 examples.
• c. Emendations, which I believe to be authorial, that depart from the reading shared by all MSS and are retained in Q1. Seven examples.
• d. Emendations, which I believe to be authorial, that depart from the reading shared by all MSS and are not found in Q1, which restores the original reading. Five examples.
• e. Emendations, both scribal and authorial, that show further revision in Q1. Four examples.

The significance of the first and second groups rests in the fact that both the scribe and the author corrected the manuscript with considerable care in order to restore its readings to what must have been, on the evidence of agreement in the other MSS, their original form. That Dryden corrected it at all is testimony that the manuscript was sufficiently close to the archetype to warrant such action by the author.

The readings in the third group are important because they are clearly deliberate improvements made in H and retained in Q1 either to correct or to emend the archetype.

The fourth group, more complicated than the rest, consists of five readings which I shall reproduce here as they appear in the manuscript.

None of the readings in their emended form appears in Q1 but all of the crossed-out readings are not only in Q1 but also in the other MSS.[8] All that can be determined with certainty is that the crossed-out readings derived from the archetype and were revised in H by the author. Why the revisions are not retained in Q1 is impossible to establish with absolute conviction, although it is safe to assume that they were not present in the MS that served as printer's copy. It becomes necessary, therefore, to make a final decision concerning the nature of q, the MS behind Q1, and its relation to H. There are two alternatives: (1) it derived from H in the main, but was further revised and augmented by Dryden before being given to the printer; or (2) it is a MS independently derived from the archetype, revised and augmented by Dryden shortly before the play was published in 1677.

In the first explanation, H becomes the initial step in the process of revision that was to culminate in Q1; and the five readings emended in H that revert back to their original form in Q1 are merely readings rejected by the author in his final revision of the play. Unfortunately, however, this theory does not take into consideration the unique readings in H which, although relatively few in number, are still sufficient to establish the MS as a terminal text as has already been demonstrated. Moreover, it is impossible to overlook the fragmentary state of the first two pages of H. Only if one assumes that the scribe of q was presented with a freshly written leaf containing the lines omitted in H, as well as all the additional material that appears in Q1, which will shortly be discussed, can even partial derivation of q from H be accepted. The obvious difficulties to this theory make the second alternative the more probable.

In this, H becomes a MS that was sufficiently close to the archetype to warrant partial revision by Dryden at some time. When he decided to print the play, three years after it was written, he also decided to augment it with various embellishments designed to please an audience. One can postulate, then, that the Harvard manuscript was no longer in his possession and that a new manuscript was prepared from the archetype, which had already been revised by the author to include the several changes and additions that appear in Q1. This manuscript, deriving independently from the archetype, would not contain the unique errors of H, but would retain some of the emendations found in H either because Dryden had already noted them in his foul papers when he altered H, or because he remembered them and made them again. The group of five readings that do not appear in Q1 as emended in H were either rejected by the author in his final revision of the play or forgotten. On the evidence of the readings themselves, the first conjecture seems the more likely.

The readings in the fifth group, emendations both scribal and authorial that show further revision in Q1, offer additional evidence of the care with which Dryden revised his play, first in H and later in q.

Considered collectively, the five groups of corrections and emendations in H provide further proof that H did not serve as copy for Q1. The manuscript

behind Q1 is q, in all probability a transcript of the revised archetype. Conclusive evidence of the existence of this manuscript is to be found by examining the readings in H that differ from Q1 but are supported by all other MSS. This material can be divided into the following eight groups:

1. The title of the play as it appears in Q1 is The State of Innocence, and Fall of Man. In all the manuscripts, with slight modification, it is The Fall of Angells and Man in Innocence, the title under which the play was registered by Herringman in 1674. Since the emphasis of the play is on Adam and Eve and their fall from grace, rather than on the fall of the angels, which is important only as it affects and parallels the human situation, the title as given in Q1 is the appropriate one, undoubtedly authorized by Dryden.

2. The character of Asmoday in Q1 is called Asmodell in all manuscripts. In view of the fact that the names of all the angels in the play end in "el", Dryden may have realized that Asmodell carried an unsuitable angelic connotation for a devil and made the change accordingly.

3. The speech prefix to Lucifer's short line on B1v, line 26 of Q1 is omitted in all MSS. Whether this is an authorial correction made for Q1 or an inadvertent compositorial addition is debatable. On the literary evidence of the entire passage, however, I believe it to be authorial.

4. The opening stage direction of Q1 omits the two phrases "thunder is heard and flashes of lightning seen" and "A shower of fire rains over the stage". Q1 adds the direction "and a hymn sung." Since the omitted phrases also appear in the opening stage direction of Shadwell's operatic version of the Dryden-Davenant The Tempest, Dryden may have decided against repeating a device already in use at the rival theatre where The Tempest remained popular for ten years. The addition of a hymn to the printed text is clearly an attempt to meet the growing demand of the public for more music in the theatre.

5. The long stage direction at the end of Act I, the short directions on C1v and D2r, and the aside on C4v of Q1 do not appear in any manuscript. Each of these is theatrical in purpose, provided in order to please an audience by offering a more colorful spectacle on the one hand, and to instruct an actor on the other. As such they are clear indications that Dryden revised his play carefully for publication not only to defend his literary reputation, but because he hoped that the play would finally be performed.

6. There are no scene headings apart from the first of each act in Q1. All manuscripts divide Act IV into two scenes. Whether this unimportant omission is authorial or compositorial cannot be determined, but it was clearly intended to preserve uniformity of scene headings.

7. Q1 lacks five lines that appear in all manuscripts: two pairs of couplets; and a single line. The first is a line that occurs early in Act II when Raphael is explaining to Adam that his single state on earth is comparable to that of the Almighty who, like Adam, has no equals, only servants. In the printed text Adam somewhat abruptly replies, "Why did he Reason in my Soul implant?" (II.i.57.) The additional line in the manuscripts provides not only a smoother but a less arrogant speech which reads

It is significant that the omission of the line in the quarto breaks the pattern of rhymed couplets consistent thus far in the act and leaves a line that stands alone. The MS reading provides the couplet but the word "wish" destroys the rhyme. Since, however, the inclusion of the line is preferable both metrically and dramatically, one is forced to believe that its omission from Q1 was caused by compositorial eye skip rather than by authorial deletion.[9]

The first of the two couplets is in Act III, appended to the end of Lucifer's soliloquy in which he vows vengeance on Adam and Eve. In Q1 the passage ends

In the MSS these lines follow immediately: Since the lines in the MSS have no bearing on the subject of the speech as a whole, which is concerned with Lucifer's terrible jealousy of Adam and Eve, it seems safe to assume that Dryden himself deleted the lines in recognition of their irrelevancy.

The second of the two couplets is also in Act III in the scene of Eve's vision. Here the woman in the dream, protesting that she will die if she tastes the fruit, asks, "and who can Heav'n withstand?" (III.ii.24.) To which in Q1 the angel replies, "Why was it made so fair, why plac'd in sight?" In the MSS, however, the angel answers

The result is incoherent since the words "Believe not so", if they refer to anything, go back to the woman's first statement concerning her fear of death. The deletion of the lines is a marked improvement that is certainly authorial.

8. Q1 differs from all manuscripts in 41 additional substantive readings. These readings, varying in significance, can be divided into two main categories:

• (a.) Q1 changes in words intended to improve the sense, the metre, or the grammar of the line that can be demonstrated as authorial. There are 26 in this category.
• (b.) Variants that are due to compositorial error in Q1. Of these there are 15.

As it is impossible here to discuss each of these readings, two representative examples from each group must suffice. Early in Act III Eve, describing her nuptial bower, lists the flowers that grow around it:

and a few lines later,

The manuscripts, however, read:

The change from "Bridegroom" to "Myrtle" could not have been compositorial since the two words are so unlike that neither misreading nor memorial failure could have occasioned it. Far more probable is that in going over the text, Dryden realized that the original image of the orange as the bridegroom and the rose as the bride was strained and artificial. Accordingly, he substituted "myrtle," the flower sacred to Venus, and the subsequent change from "both" to "all" followed logically. The word "Thy" in Q1 is, of course, a misprint for "The".

From the second category, those 15 substantive variants in Q1 which I consider compositorial errors, both examples chosen for illustration are to be found in Act III. The first comes in Lucifer's soliliquy as he listens in anguish to Adam and Eve discuss their wedding night:

The MS reading, which is consonant with the Miltonic doctrine of matter from which all creation was formed, was obviously misunderstood by the compositor who "improved" the reading by adding an erroneous "s".

The second, which refers to the flowers in the arbor, reads

A close examination of Q1 reveals that this is a clear example of a turned letter "d" printed as "p". Despite the nonsense that results, the compositor's error remained undetected through all the early texts and has been retained in all modern editions.

When one considers all the variants between Q1 and the unities of the manuscripts, it is evident that, except for a few compositorial errors, the Q1 variants can be attributed to the author. Since the manuscript that contained them represents the author's final intentions concerning the words and action of the play, Q1, deriving from this manuscript which itself descended independently from the archetype, is clearly the text of highest substantive authority.

Notes