University of Virginia Library

The Historical Collation in an Old-Spelling Shakespeare Edition: Another View
by
Fredson Bowers

Paul Werstine's article "Modern Editions and Historical Collation in Old-Spelling Editions of Shakespeare" (Analytical and Enumerative Bibliography, 4 [1980], 95-106) leads me to attempt a few supplementary observations that may suggest ways around the difficulties he raises. I am the more encouraged to comment on his views in that during the development of his argument he did me the honor of quoting various of my past opinions.

Dr. Werstine bases his case that an acceptable historical collation of modern variant readings (i.e., from the eighteenth century on) is inadvisable if not impossible as part of a critical old-spelling edition of Shakespeare on (1) "the pragmatic difficulties arising from any attempt to do them justice in a historical collation." As a supplement to this, he adds (2) "It also acknowledges the difference between the editorial principles governing critical old-spelling editions and those on which modernized editions are based, as well as the divergence in editorial principles among the editors of various modernized editions." In the coda he adds (3) "So wide are these differences as to make comparison of editions in a historical collation largely meaningless without elaborate textual notes explaining the collation" (p. 95).

These three points may be viewed in order, although the first two are so interrelated as to require some occasional overlapping of discussion. Dr. Werstine takes it that the formation of a full substantive historical apparatus "implies collation of all previous texts of Shakespeare," a feat obviously impracticable since "such a monumental task must distract an editor for years from the bibliographical, textual, and philological inquiry on which his edition will finally rest" (p. 97). One cannot deny the validity of this objection provided the premise be accepted. But I expect most scholars would agree that such a comprehensive collation of variants is suitable only for the Variorum edition of Shakespeare, now happily revived, and that the minutiae on a comprehensive plan of out-of-date editorial substantive variation should remain the province of the variorum principle, with which a modern critical old-spelling edition need have very little link.

Dr. Werstine suggests, only to dismiss, the possibility of a selective list of editions to collate on historical principles. For instance, he rejects (properly, I think) the recording of all substantives even from the most prominent and influential of eighteenth-century editors, which "would swell a historical collation with many readings no modern editor could consider seriously" (p. 99). Could an editor, then, confine himself to nineteenth- and twentieth-century editions? Dr. Werstine thinks not, for "even the last century's editions are based on quite diverse editorial principles in order to satisfy a variety of audiences, and thus they necessarily differ widely among themselves and, as modernized editions, even more widely from an old-spelling edition." The conclusion is that the audience for an old-spelling edition is—for this section of the apparatus—exclusively concerned with the early editions on which the text has been based, and thus with "limiting the unnecessary interposition of editor between reader and text." This objective requires an apparatus that aims at allowing "the reader to reconstruct the copy-text from which the editor worked; hence too . . . emphasis on thorough collations of the early editions in an attempt to recover the author's final intention" (p. 100).

Here, I think, we can make a break-in to consider the value that may be placed on an historical collation composed exclusively (as I understand it) of substantive variation in early editions of Shakespeare, or of other dramatists, from the readings of the edited text. First, we must tackle the argument that this shucking-off of all extraneous material (assumed to be the editorial emendations of later editions) would enable a reader the more readily to reconstruct the copy-text from which the editor worked, this taken to be the major preoccupation of a scholar consulting the apparatus of an old-spelling edition. As applied to the historical collation, which is the sole object of Dr. Werstine's inquiry, I am inclined to doubt the validity of his statement. If an old-spelling editor records his emendations in a series of textual entries separate from the historical collation that lists rejected substantive readings in the editions chosen for collation,[1] it is the emendation notation alone that

acquaints a reader with editorial substantive changes in the text he is reading and thus enables him to construct the copy-text, provided the editor also lists either in these (foot) notes, or preferably in a separate part of the apparatus, the records of his alterations made in the copy-text accidentals. If one were confining the historical collation only to early editions, and if the copy-text were the only early edition, as in Marlowe's Jew of Malta, there could be no historical collation. That fact in itself should be sufficient to indicate that the historical collation has nothing to do with providing the reader with the information he needs to reconstruct the copy-text. Thus it would seem that Dr. Werstine has mistaken the function of the historical collation in respect to the reader's concern with reconstructing the copy-text (done entirely through the record of emendation, instead), and thus one of the bases for his argument fails.

Even so, let us pursue the matter a little further and suppose that early editions do exist, as with Shakespeare. The first requirement is to define what one means by the term. For Elizabethan plays in general (loosely so called written before 1640), a convenient terminal date is 1700, although this cut-off point need not be rigid if there are special circumstances. Any such date as 1700 is arbitrary, of course, but it does enable an editor of Elizabethan plays reprinted in the Restoration to record variants in editions which by some obscure resort to theatrical tradition or to some now lost manuscript could theoretically contain readings to which authority might be imputed.[2] There is a further advantage in that 1700, very much in general, marks for some authors like Shakespeare especially, or Beaumont and Fletcher, or Marlowe, the beginning of formally edited complete editions which start a new textual tradition.

Putting aside for the moment the question of an historical collation for variants after 1700, we may now ask, what is the point of recording variants in early editions whether for Shakespeare or other dramatists, or for any author? The circumstances divide neatly in two. In the first, as represented by Thomas Dekker or by Christopher Marlowe, only the first editions (which must of course be chosen as the copy-texts) have any authority: any later editions are mere reprints. In the second, either the author has deliberately revised a later edition, as Ben Jonson did with Sejanus and other plays in his

1616 Folio, or else as with Shakespeare's Richard III, Hamlet, Othello, or Troilus and Cressida a later edition either has had access to or was itself set up from a manuscript in a different tradition but one still deriving from authority. In this group may also be placed plays like Fletcher's Woman's Prize surviving in a manuscript but printed in the 1647 Folio from a different manuscript.

In the first case, where no later edition has authority, the editor's private collation establishes the fact for all examples where (as it has in the Second, Third, and Fourth Shakespeare Folios) previous scholarship had not collected and analyzed the evidence. Although a later simple reprint may furnish a corrected but never a revised reading (one backed by an authoritative source), it is customary for an editor to credit the earliest document with the emendation (or correction) he has adopted. The reason for the custom is obvious. First, it establishes historically the precise details of the refinement of the textual tradition; second, the possibility always exists, as McKerrow observed, that a compositor (or some elementary form of editor such as the one who figures in the third edition of Dekker's Shoemakers' Holiday), being nearer to the linguistic background of the author may guess rightly the usage and correct reading with some slight advantage over the modern editor and so his alteration might be given some modicum of respect. I may say that McKerrow's tentative hope is seldom shown to have any validity: modern standards of emendation, and modern philological resources such as the O.E.D. (when not misused) ordinarily mend a faulty text with more acumen than the rather innocent sophisticating guesses one often finds in early textual transmissions, as for instance in the Beaumont and Fletcher Second Folio. I rather think we can reassert, these days, that there is no 'secondary authority' in simple early reprints that in any generic way assists a scholarly editor to emend a text. In my experience, however, it has occasionally proved useful to be alerted from the evidence of a designed early variant that some near-contemporary thought the reading of one's copy-text was faulty. As often as not one may ponder the text with benefit, even though one may end by retaining and defending the original reading. On occasion, also, the signal that something could be wrong is legitimate but the editor may choose a different emendation from that of the variant early reprint. Usually, of course, since the reprint designedly corrects only the most obvious errors and with the most obvious readings, one may agree and assign that source for the change with no prejudice intended in favor of the correction. It is important to emphasize that all emendations made from non-authoritative sources are on an equal basis; hence a reading drawn from an early reprint has no more intrinsic validity than one suggested by an eighteenth-century editor—or by a twentieth-century one for that matter, including the editor presently at work. Thus the identification of the source of an emendation is partly a courtesy but chiefly an act of historical scholarship.

This being so, one may logically inquire whether the usefulness of an historical collation, even of the early variants, under these circumstances warrants the effort of reading and the cost of printing. To be specific, are

modern editions of Shakespeare like the New Arden 1 Henry IV which engage themselves to a collation of a series of early reprint Quartos as well as the First Folio in relation to the First Quarto copy-text, are these performing any useful function in their lengthy records of rejected unauthoritative readings? One part of the answer has already been given. As in The Comedy of Errors, for example, when no question can exist of any authority reentering the text, the record of the degenerative substantive variants in the three later Folios—like the record of the later octavos of Marlowe's Tamburlaine —provides no information whatever that is of use to a scholar evaluating the First Folio copy-text; and, of course, nothing that will enable him to reconstruct it if we take 'reconstruct' in the strict sense of the recovery of the exact details of the First Folio copy-text from the editorial ministrations in the old-spelling edition. Such a recovery is the function of the list of emendations, both substantive and accidental, the latter of course of no value in a modernized text.

The most that can be said, perhaps, is that the collation of these early editions provides a reader with the opportunity to evaluate the later Folio variants that have been rejected, since it is possible that the reader may find some reason to second-guess the editor and to take it that the rejection was ill-advised and that a particular variant should have been adopted as an emendation. This is a legitimate object, without question, although it must be remarked that if the historical collation is limited to the early editions, the reader who fancies some rejected reading and would like to reconsider its validity as an emendation has no means, short of his independent investigation, of knowing whether some later editors have adopted the reading and even whether it may not have become a fixture in the modern textual tradition, despite its rejection by the old-spelling editor in question. One may also remark that for a newly edited author the record of the historical collation enables a reader to evaluate the editor's estimate of the authority or non-authority in the recorded early editions: the collation provides the full evidence of the early textual history on which the editor has based his judgment of the copy-text and its treatment. Thus, in the phrase, his cards are all on the table. However, this situation, applicable as it may be to Beaumont and Fletcher, does not apply to Shakespeare where the non-authority of the Second, Third, and Fourth Folios has long since been established.

Finally, an historical collation of early editions can be justified, perhaps, even for Shakespeare, as a handy means to illustrate to readers, whether of modernized or of old-spelling editions, the various ways in which a text can degenerate, and be modernized and sophisticated as well as subject to inadvertent departures from copy. The more a scholarly reader studies the evidence of textual transmission, the wiser he may be in the ways of texts and their editing. Even a naive reader may be interested, for example, in the copious New Arden recording of the numerous Quarto as well as Folio variants in 1 Henry IV, and especially in the manner in which the Folio text was corrupted by its use of Q5 as setting-copy.

However, for some plays of Shakespeare and a few other dramatists there is a gray area between the well-defined polarities of this first class of unauthoritative variants from mere Folio reprints as in The Comedy of Errors and the second major class where, as in Troilus and Cressida, the First Folio may indeed introduce fresh authority to be compared with that of the Quarto, and the proper copy-text and its eclectic emendation comes in question. Within this intermediate gray area, in the Folio setting-copy minor reference, more like touching up, has manifestly been made, derived from some other document, very likely a manuscript, although the Folio editor's knowledge of theatrical tradition cannot always be ruled out. But the authority of this document, or source, is moot in its details since we may not know or cannot reconstruct with certainty the details of its textual history or, most important, the exact use of it made by the annotator. In this category comes the attention to details of speech-prefixes given by the editor of the Folio Merchant of Venice as against the Quarto setting-copy. A clearer-cut example occurs in Richard II, Act I, scene i. In the Quarto it seems reasonably certain that Gaunt leaves the stage as part of the general exeunt at the end of the scene. But an exit in this place violates the principle of the clear stage, for the next scene begins with his immediate entrance in another place and at another time with the Duchess of Gloucester.[3] Almost certainly on promptbook authority, the First Folio inserts an exit for Gaunt at line 195 in the midst of the hot dialogue between Richard and Bolingbroke. The exit in this position is unmotivated and extremely awkward in various respects, including the fact that no exit line is provided by any revision of the text. Gaunt has not asked permission of the king to withdraw, nor is there any reason for his removing himself at this point. It is clear that the exigencies of staging created this exit in some document which the Folio editor utilized, one which in this and in other details affected the Folio text.

The question then arises, if the Folio editor in this place compared the Quarto selected as setting-copy for the printer and altered it by reference to the reading of the promptbook, how many of the other variants in the two texts may not also have been inserted by comparison with the promptbook readings? This question brings up the probability of some form of authority introduced in the Folio from an independent source that could affect at least some of the variant readings, not necessarily because Shakespeare revised the copy behind the promptbook (although an editor must deal with this possibility) but because the book may have preserved authentic readings corrupted in the printing of the Quarto, that is, provided the Folio editor paid significant attention to readings within the text itself and not just the directions (an important proviso).

Plays like these are, then, textually in a somewhat different category

from, say, 1 Henry IV in which no certain signs of intervention in the Folio from fresh authority have seemingly been detected.[4] And the case is very different from that of The Comedy of Errors where the text first appeared in the Folio, and its reprints in the succeeding Folios can carry no authority at all. So far as providing the evidence for an estimate of authority, then, an historical collation of the early editions of The Comedy of Errors is worthless, whereas to a scholar using an edition of Richard II or The Merchant of Venice or Romeo and Juliet and endeavoring to follow the editor's choices among variant readings, the collation of the early editions has sufficient reason to justify its existence.

The argument becomes more difficult when the general assumption holds that the Folio is a mere reprint of some Quarto which (whether or not 'edited') by common consent is taken to contain no reference to any other authority. May we not say that in such a case the historical collation of the Quartos and then on to the Folio and its successors is as worthless as that of The Comedy of Errors in F2-4 as evidence by which a reader can evaluate an editor's treatment of texts in which customary opinion takes it that no variants have any status except as they result from textual transmission? In one sense, yes, but in another, no. In fact, not all Quarto-Folio relationships need be quite so cut and dried as current opinion may suggest. Although we may have confidence in the judgment of presentday scholarship, past scholars felt the same about the scholarly estimate of the texts made by the Old Cambridge editors which in certain plays we now reject. It was not so long ago that J. Dover Wilson's arguments for authority in the Folio text of 1 Henry IV were thought to be worth consideration. At least they carried enough weight to provoke the scrupulous examination, backed by evidence of a full historical collation up to the First Folio, made by A. R. Humphreys in his exemplary New Arden edition to support his contrary view that no Folio authority exists. Thus in an old-spelling edition with some pretensions to definitiveness caution[5] might suggest the advisability of including historical collations for any subsequent reprints up to and including the First Folio even where present opinion declares them valueless as applied to the foundation of the text itself.[6] It may be that we think we know the Folio

plays in which the editor touched up the copy by manuscript reference as distinguished from those that he neglected or in which he acted on his own initiative; but future generations may have a different opinion. Very much in general, opinion at the present time rests on an incomplete series of independent studies of the evidence of individual plays and we lack any authoritative analytical overview of what it is proper to call the editing of the Folio. Thus it may be argued that even for the present generation of critics an historical collation provides the material for evaluating the modern editor's judgment formed from his original research or from common opinion.[7] And, of course, consistency suggests the advisability—indeed the necessity—of providing all Quarto plays with identical apparatus, regardless of questions of authority.

In the second group no difference of opinion can exist as to the need to record in an historical collation the substantive variants in Shakespeare editions that derive from two independent authorities. These separate into three fairly distinct classes. First, examples like Richard III in which it is demonstrable that the Folio editor or some surrogate (whether or not in the printing house[8]) by annotation brought one or more Quarto editions into general conformity with a manuscript in cases where the printed copy represented a so-called 'Bad Quarto,' or corrupt text. Second, examples like Troilus and Cressida where the same operation was performed but the Quarto represented a good text though one in a different tradition from that of the manuscript. Thirdly, examples where the Folio printed a good text direct from a manuscript although a good Quarto was in existence (as may have happened with 2 Henry IV) or where the Bad Quarto was too distant from the good text to serve as the basis for annotation, as with The Merry Wives of Windsor.[9] For the purposes of this discussion, however, the exact nature

of the Folio copy is of little importance, fortunately: the whole matter revolves on the single question of the need (and sometimes the practicability) for an historical collation when double authority is present.

The Shakespearean texts with two authorities are like such plays as Beaumont and Fletcher's Beggars Bush or Bonduca or The Womans Prize (preserved in manuscript as well as in print from a different source) in that a critical editor is forced into creating an eclectic text that will combine what in his opinion are the more authoritative readings drawn from the two documents, with due regard for what can be reconstructed of their textual prehistory.[10] In such situations where every variant has presumptive authority, and a choice of readings depends upon the editor's coherent theory of the histories, possible corruption in transmission, and the degree and nature of revision, whether authoritative or unauthoritative,[11] editorial opinion is bound to vary widely and it is essential that an historical listing reproduce the rejected readings of the document that was not chosen as copy-text but from which emendations have been levied as recorded in the emendations list. Only by means of such an historical collation can the full substantive facts of the authoritative texts be laid open to the inquiring reader.[12] This principle holds as much for the Bad-Quarto as for the Good-Quarto authorities, especially when an annotated Bad Quarto served as the setting-copy for the Folio revised text. As remarked, uncollatable Bad Quartos with no direct link to the Folio version will need in the main to stay uncollatable except insofar as their readings may be drawn on for emendation or as it is proper to record them as well when emendation of the Folio is made from the agreement of some later source.[13]

Dr. Werstine is not specific about his recommended collation of early editions, but he tacitly seems to aim at including all four Folios: "Instead

the rationale for excluding all editions after the Folios rest on the pragmatic difficulties . . ." (p.95). If this view be accepted, the question then arises for Shakespeare whether an old-spelling editor should rest content to stop the historical collation with the Fourth Folio, or whether he should agree to the arbitrary 1700 cut-off date (which thus would introduce all early editions before the seminal first collected edition by Nicholas Rowe), even at the expense on some occasions of detailed substantive variants recorded from a number of Restorations Quartos as in the five (actually four) of Julius Caesar that Dr. John Velz has so painstakingly identified and ordered.[14]

This question of where to stop if an editor were to confine his record of rejected substantive variants to the so-called 'early editions' is affected by what may reasonably be taken as the special needs of a user of an old-spelling edition. (Since we are concerned with the apparatus alone, the problem of what is needed in the text itself may be put aside). One primary purpose of the apparatus being to enable a reader to reconstruct the copy-text, obviously all emendation, whether substantive or accidental, must be recorded, with its earliest source and the rejected copy-text reading. However, as has been suggested, the historical collation may be as important to the reader as the emendations list for those plays of double authority like Hamlet or Othello where an eclectic text is in order and the reader will be as concerned with what has been rejected from the alternative authority as with what has been accepted as emendations. This urgency to know immediately the alternative rejected readings along with those that have been retained from the copy-text lessens when the play has been only touched up by the Folio editor, even though a scrupulous reader will still want a complete substantive record of the variation; and it begins to approach a formality when the Folio is generally judged to be a mere unedited reprint and its variants of historical, not of critical interest. Nevertheless, since the status of the two texts may in the future be subject to re-evaluation, it would be anomalous and unwise for an old-spelling editor not to be consistent in always providing as complete a record for such plays as for those with double authority. Some scholar is sure to want the information for his own purposes that cannot be foreseen by an editor. The problem of recording Bad Quarto variants in an historical collation is too complex to discuss here in any detail. A full record would appear to be practicable for such plays as Richard III and King Lear but impossible for others. It may be that an editor can put together some combination of historical-collation recording with reprinted verbatim passages that are uncollatable placed in an appendix, roughly such as was attempted in dealing with the A and B texts of Doctor Faustus in my edition of Marlowe's Works (1973, revised 1982).

Two major problems then face the editor. First, when Quarto reprints intervene between the Folio and the Quarto copy-text (normally Q1), consistency requires the inclusion in the historical collation of all intervening editions. Unauthoritative as such editions may be, they can assume a textual

importance when the Folio was set from a Quarto other than a copy-text Q1. Without such a record, some Folio readings would be misunderstood as original variants when in fact they derive from the later setting-copy, or some ancestor Quarto, as happens in 1 Henry IV, for example, set in the Folio from the fifth of six earlier Quartos. Moreover, complications may ensue when, as with Richard III, more than one Quarto was used as Folio copy.

The second problem is what to do about the Second through the Fourth Folios since in no case are these authoritative[*]. The answer here rests on one's estimate of the highly varied uses to which scholars may put the evidence contained in an old-spelling edition and its apparatus. (One should not fall victim to the all-too-common notion that one's own peculiar and often narrow needs are universal.) It seems reasonable to take it that the owner of an old-spelling edition should require it to be as definitive in its historical records as the editor has attempted to make it definitive in its text. That is, for many scholars the early historical textual records may have an independent interest. One major reason is this. The foundations of the edited textual transmission go back to Rowe in 1709, the first official editor of a collected edition, one that was influential in its day—but also one that was based on the Fourth Folio and its inferior text. Some Rowe readings, many of these being later corrupt Folio variants, persist through the eighteenth century and after as successive editors annotated some convenient earlier edition to make up their printer's copy. Moreover, subsequent eighteenth-century editors for some time did not begin to use the First Folio or the original Quartos systematically, in the process beginning the long haul of eliminating historically unauthoritative readings from edited texts. It follows that a presentday scholar should be able to utilize the historical records of all editions before 1700 as the universal foundation for any independent research into later textual tradition whether of the eighteenth, nineteenth, or twentieth centuries. This is not to urge a variorum principle: instead it is to insist that a scholarly old-spelling edition would fail in one of its important functions if it did not provide the complete pre-Rowe textual history for the substantives as well as for the important semi-substantives in which meaning has been subject to differences of opinion.[15]

At this point (actually with F4) Dr. Werstine, as I take it, for a number of stated reasons would stop the historical collation. Although I do not agree,

I have no serious quarrel with his view and I should not automatically fault an old-spelling edition that concluded its historical record short of 1700, a date which I should prefer to that of the Fourth Folio. With such a record the foundations of the pre-Rowe editorial tradition have been laid for whatever purposes different scholars will wish to utilize the information. Nevertheless, if no information is given in this historical record of variants after 1700, serious deficiencies would exist in the total provision that a scholar might reasonably require of an old-spelling edition. It seems to me to be an open question whether a post-1700 proposed solution to an important crux that ought to be considered by any serious textual (and even critical) scholar should be concealed merely because the old-spelling editor did not choose to adopt it and thereby to record it and its source in his emendations list. The problem that Dr. Werstine finds insoluble is how to include substantive variations in editions after the Fourth Folio (or after 1700, let us say) without turning the edition into a Variorum, a practice that should not be the intent for an old-spelling edition. Other dramatists present fewer problems. Thomas Dekker, for instance (whose edition Dr. Werstine cites) was scarcely edited at all in later times except for some small attention paid to The Shoemakers' Holiday in the nineteenth century, the selected-play Mermaid edition of no scholarly consequence, and a scattering of twentieth-century drama anthology reproductions of the texts of one or two plays. It seemed appropriate, therefore, to conclude the historical collation with 1700. On the other hand, Beaumont and Fletcher had a relatively active editorial history in the eighteenth and nineteenth centuries, although a negligible one in the twentieth. Even if one mars completeness by omitting a few collected editions that were substantially reprints without fresh editing, a minimum of five major editions that helped shape the text appeared between Langbaine the Younger in 1711 and Alexander Dyce's revised edition of 1877, and for some plays the aborted Variorum must be included as well as a scattering of modern texts, chiefly dissertations. For Christopher Marlowe from Oxberry's incomplete edition of c. 1820 to Roma Gill in 1971 (omitting the Pendry-Maxwell revision of 1976) no less than eleven post-1700 editions of some textual consequence appeared. Collation of these, in addition to whatever early reprints existed was not an insuperable task in the Cambridge Marlowe, nor did the recording of variants in the historical collation require an exorbitant amount of space. The effort involved was thought worth while since it gave a scholarly reader a relatively full conspectus of Marlowe's textual history up to the present.

However, Shakespeare is another story because of the multitude of editions that flooded the eighteenth and nineteenth centuries, various of them serious attempts at establishing the text at least according to contemporary standards. The considerable activity in the twentieth century has given rise to what is familiarly known as 'the Shakespeare industry.' It is this flood that causes Dr. Werstine to deny the possibility of an effective historical collation short of a variorum. If an editor, he suggests, were to attempt a conspectus of emendations, he would be forced to "turn to those editors responsible

for most of the emendations still found in twentieth-century editions —Pope, for example, and Theobald, Hanmer and Capell. . . . Yet these early editors and scholars were so convinced of the unskillfulness and negligence of Shakespeare's printers and thus so prolific of emendations that they would swell a historical collation with many readings no modern editor could consider seriously" (p. 99). If one were to print, instead, a collation of nineteenth- and twentieth-century editions, the problem would be equally complex although different in the kind of variants, he continues. "Yet even the last century's editions are based on quite diverse editorial principles in order to satisfy a variety of audiences, and thus they necessarily differ widely among themselves and, as modernized editions, even more widely from an old-spelling edition. . . . Collation of an old-spelling edition with recent modernized texts seems merely to record the thoroughly predictable effects on the Shakespeare text of such differences in editorial policy and illustrates the influence that time, taste, and audience have exerted on modernized editions. . . . Variation simply results from individual editorial policies, each quite different from an old-spelling editor's policy. Of what use is the comparison of recent modernized editions to each other as a part of an old-spelling edition?"

At this point I begin to part company with Dr. Werstine's logic. I agree completely that even a selection of eighteenth-century editions (and let us add nineteenth-century editions as well) could not be managed for a full substantive collation without the listing of more readings now presently in discard than could justify the expense and time save in a variorum edition. Moreover, the selection process would be so arbitrary and the number of recorded editions so limited by practical considerations that about as much useful information would be omitted as was provided, and the results could never be utilized by a scholar to attribute certain emendations to definite editors. Any attempt to provide only a partial collation of significant variants from important editors would satisfy no one. (Who takes care of the caretaker's daughter . . . . What is partial? What significant? Who important?) The question then arises, granted that the owner of an old-spelling edition may reasonably require a full substantive collation of early editions before 1700, granted that scholars concerned with Shakespearean textual matters would find that such a record obviated the time necessary for them independently to make the identical collations for any special text in which they had an interest,[16] it still may follow that this owner will find that actually

he needs more information. The key to these reasonable requirements, I take it, is the fact that an old-spelling edition should not be a facsimile but instead a critically edited reading text, as well as a reference source for important textual information. The implications of the difference between a facsimile and an edited text are wide ranging. Once we abandon the concept that the primary reason for the edition's existence is not to serve as a reference work pure and simple like a Variorum, and admit that it has a double function, and that the offering of a conservatively edited but attractively presented text in its original dress is in fact more important than the non-selective record of variation, even among early editions, then we may progress to another view of the proper function of the apparatus that should accompany such a text.

The essential apparatus is in two parts: first, the complete record of emendation, both substantive and accidental; second, the record of rejected substantive readings in a group of editions selected according to a principle that most economically and efficiently serves the need of the critically edited text. This last requirement is the heart of the problem that an editor faces; but even so, it cannot be tackled independent of the form taken by the emendations, for the two parts of the apparatus must mesh. In my own experience I have found it most practicable, and less intrusive for a reader, to separate the record of the more important emendations from those of lesser immediate interest; that is, to make one group of the editorial changes in the substantives, and another group of those in the so-called accidentals, or incidentals, or whatever one chooses to call them.[17] In addition to the reasons cited in footnote 17 immediately above, there are also other practical considerations. In the listing of substantive emendations to the copy-text it is vital that the editor provide the original (that is, the earliest) source, whether from an early edition or as a borrowing from a modern editor, and append, of course, the

rejected copy-text reading. The reasons and I hope the virtues of this procedure have already been discussed. It instantly provides a reader with an historical perspective on the sources of the emendations a particular old-spelling editor feels are required to remove corruption, these being the primary concern with all texts in the simplest cases of single authority. However, when multiple authority is present the historical perspective includes the history of the eclectic reconstruction of what seems to the editor to be the best adjustment that should be made between conflicting authorities to insure a text that will attempt to represent not only the author's most correct but also his final intentions insofar as these can be determined by principled analysis of the evidence.

The case seems to me to differ essentially when one comes to the assignment of a source for changes in the copy-text accidentals, the emendation of which is ordinarily found more often than not in the punctuation. If an editor chose, I suspect he could take full responsibility and provide no source whatever for the accidental changes that he felt were advisable or positively required, especially since it is his positive duty in an old-spelling edition to alter the accidentals strictly in accord with the general system of its time and, whenever possible, with the specific system in the work being edited.[18] Obviously, the rewritten accidentals imposed on a text by a modernizing editor after 1700 are so completely unauthoritative that to credit Pope, say, with what by Elizabethan standards in the copy-text is a clarifying change from a semicolon to a colon would be absurd, and even more absurd to credit Kittredge, or Dover Wilson, or Alexander with the same if any of them had been the first to adopt in a modernized text what by chance proved to be the specific form of punctuation that an emending old-spelling editor felt was consistent with the original early system of the copy-text. On the other hand, although it is true that each successive early reprint is bound to represent a form of printing-house modernization, and although it is also true that by 1700 the more Elizabethan rhetorical system of punctuation had been greatly modified in the direction of syntactical pointing, nevertheless it may seem useful to a reader to know the source of any editorial accidentals emendation when it has been provided by an early edition since there may be some slight editorial comfort in being backed by the opinion of a near contemporary, no matter how casual. It follows, then, that an old-spelling editor may usefully list the earliest source of any accidental emendation when it is drawn from pre-1700 editions but ignore any assignment of source after that arbitrary date when new-broom editors have taken over with their modernized reading editions according to a different system. Whatever assistance early editions may be thought to provide, an old-spelling editor's own principled

opinion needs no support from post-1700 editors in the matter of accidentals save, perhaps, in some special examples of semi-substantives, which are not here in question. If this distinction be observed between providing the source for all substantive emendations but only a pre-1700 source for the accidentals, the use of two ranges of apparatus is almost required for consistency's sake; that is, footnotes to the text (preferably), readily visible to the reader for his instant reference by line number when substantive emendations have been made, but a separate appendix list for accidentals emendations containing for post-1700 alterations no information about the source of the change, which must be taken as reflecting only the editor's opinion of its necessity or advisability.

Since the list of substantive emendations and the historical collation of rejected substantive readings are only two parts of a whole, the form of one must inevitably affect the form of the other. However, since the emendations list is the more important of the two, considerations affecting its form should be overriding and the form of the historical collation must be adjusted to it. For instance, if the historical collation were to be confined to early editions alone, two forms of an emendations note are possible. (Incidentally the two systems must not be mixed.) Suppose in The Comedy of Errors the editor accepts the F2 reading 'unhappie' for F1 'unhappie a', one form of note would read: I.ii.40 unhappie] F2; unhappie a F1. This is uncommunicative about the readings of F3-4 (silence about later editions cannot be taken to represent assent in emendation formulas); thus an historical collation entry would be required that would continue the information about the readings of early editions beyond the emender F2: I:ii.40 unhappie] unhappie a F1. This entry conforms to the standard convention that in this historical part of the apparatus all collated editions agree with the lemma save for those specifically excepted; F3-4, therefore, would be understood as agreeing with the F2 variant that had been adopted by the editor.

However, if the historical record were to be strictly confined to the early editions, it would be practicable for most plays to omit the collational entry provided the necessary information were given in the footnote, as, for instance: I.ii.40 unhappie] F2-4; unhappie a F1. Emendation by a modern editor could still be handled similarly—II.ii.186 offer'd] Capell; free'd F1—with the collational entry; offer'd] free'd F1-4. Or with omission of the collational entry: offered] Capell; free'd F1-4. In such cases it is to be understood that the modern source cited was the first editor or commentator to adopt this emendation. Because of the limitation of the apparatus in these examples to the early editions, no information is given in the emendations note of the readings to be found in editors earlier than Capell even if they provided some other suggestion for 'free'd'. All that it is necessary for the reader to know here is that no edition before Capell read 'offer'd'.

When two early collatable editions have authority, the rejected variants of the copy-text are, of course, recorded as part of the emendations entry, including the readings of both early authorities in cases where the emendation

comes from a later source. However, for the convenience of the reader in these special situations of peculiar textual interest an editor ought to add regularly in the emendations notes (even when no emendation has been made in the copy-text) the rejected readings of the non-copy-text authority for immediate scrutiny instead of reserving them for the historical collation, viz (with or without the warning stet): for an edition of Hamlet using Q2 as copy-text. Such an entry calls for an historical listing that would fill in the readings of the remaining Quartos and Folios (including Q1 in this case) provided these were not introduced as part of the emendations entry in the expanded form.

The forms of the emendations list for accidentals differ somewhat, owing to the nature of the material. Typical entries would be:

The first entry indicates that the old-spelling editor drew his emendation from his own initiative, or possibly from the suggestion of some post-1700 edition. It would be otiose to track down and identify the source of accidentals to distinguish whether they had been found in modern editions or were of the editor's own devising. In the second of these the readings of F3-4 are not provided; whether or not they differed from F1 or F2 is of no consequence because this is an emendations list of accidentals that contents itself with providing only the earliest source for the adopted emendation, not an historical collation or a substantive emendations note with historical overtones.[19] In the third example convention requires the readings of all early editions to be provided prior to that chosen for emendation. The fourth illustrates a convention that is of some occasional use in cases of very mixed and minor variation but where the predominant reading is that cited. It is not particularly appropriate for the few items F1-4, which can be handled in exact detail; but the convention has its virtues when dealing with a large number of editions of no authority where a precise record is of little consequence and only the general consensus is of interest.

After this survey of how much (or how little) historical information an

emendation entry can carry, we come at last to the vexed question of the form and contents of a useful and practicable historical collation suitable for an old-spelling Shakespeare edition. The first problem for an editor is whether to duplicate in the historical record the information provided in the substantive emendation entries. (Accidentals except for the semi-substantive variety are excluded from the historical collation.) If we were to confine the collation to the four Folios as Dr. Werstine suggests (or at least to pre-1700 editions), and if the editor handled all information up to his last collatable edition in an expanded form of the emendations entry, then it would be a relatively simple matter to omit all substantive emendations entries from the historical collation since the identical information would otherwise have been provided in both places. On the other hand, if the preferred abbreviated emendations form of entry be employed, or if the amount of historical information the editor wishes to provide in the collation exceeds what is practicable for an emendation entry, then there is no escape from a certain amount of partial duplication.

Duplication in the historical collation is of two varieties. In the most elementary form an editor notes the specific editions within the complete range of his collation that both agree and disagree with the lemma, which is, of course, the reading of the edited text whether the original or an emendation. For the moment let us suppose that we end the collated editions with F4. In the full form, then, an historical entry for 1 Henry IV, III.i. 231, might read: thou have] Q1-2; have Q3-F4 or thou] Q1-2; omit Q3-F4. But it is also an established, and indeed preferable, convention of historical collations that within the range of the collated editions any edition that agrees with the lemma need not be noted. Thus by this convention

is perfectly clear in stating that the copy-text Q1, followed by Q2, read 'thou' but that all early editions beginning with Q3 left out the word. Moreover, since the lemma must agree with the edited text, the information is also provided that the editor has rejected the unauthoritative Q3 variant and retained the copy-text Q1 reading. When the number of editions collated grows in number, the full form (which is needlessly repetitious) may grow unwieldy; hence most editors justly prefer the condensed form of entry.

The second, and more important, problem of duplication concerns those entries where full duplication exists between the emendations note and the historical section. For instance, at III.i.239 of 1 Henry IV the Rowe emendation has been universally accepted: Come] Rowe; Hot. Come Q1-F4. Under such circumstances it is possible to query the usefulness of the duplication in the historical collation entry: Come] Hot. Come Q1-F4. The arguments for omission are chiefly economic in that the apparatus is shortened. On the other hand, there are advantages to scholars if the historical collation contains for easy reference in one place all substantive variation from the readings of the edited text without the necessity for a reader to conflate

certain of the emendations entries with the historical collation in order to make the record complete. Whatever system is chosen should be capable of efficient consistency no matter what the textual situation for a given play. It would be abhorrent (because misleading) to have different forms of apparatus for different Shakespeare plays within an old-spelling edition. Thus an editor will need to consider, for example, the problem of overloading his emendation notes with an excess of unnecessary information that can result from the use of the expanded form of emendation note required if the historical collation entry is to be removed. For instance, in 1 Henry IV we would find the simplified emendations which provide only the earliest source of the adopted reading; these notes would require the completion of the information by the historical collation entries which tell the reader that, in the first, Q5, F2-4 agree with the lemma; in the second that Q6, F1-4 agree. To avoid this partial duplication in the historical collation of the details of the emendation entry, the expanded system of emendation notation would need to be consistently used, as in the forms similes] Q5, F2-4; smiles Q1-4, 6; F1 and respect] Q6,F1-4; the respect Q1-5, which may be thought to provide more information about the readings in unauthoritative editions after the origination of the emendation than a reader at this point would require. More complex situations would arise in other plays like Hamlet, for example, where both Q2 and F1 have authority but where three unauthoritative reprints intervene between the two, the readings of which must be recorded somewhere. That somewhere for Q3-5 and then F2-4 is certainly most conveniently managed in the historical collation so that the essential information about emendation may be slimmed down in its entry and not obscured or confused by notation of the agreement or disagreement (and sometimes the appearance of new readings) in unauthoritative editions.

The problem is accentuated if in the historical collation an editor boldly ventures beyond 1700 and proposes to include, according to some principle, what he regards as significant editions in the textual tradition. Here a full record in the emendations note of all preceding readings in collated editions up to the earliest emender[20] could seriously overload the entry, particularly if the emendation were drawn from a source comparatively late in the tradition.

On the whole, then, for early texts (more modern texts than Shakespeare may be another matter) it seems simplest and safest to accept the consistent principle of duplication, which enables an editor to write a more succinct and businesslike substantive emendations entry and then with flexibility according to the circumstances to provide the additional fuller information in the historical collation.

It is now proper to consider the most important aspect of a proposal to include modern editions in the historical collation provided some adequate working principle can be evolved for their selection, for, as already remarked, the principle of selection is paramount in facing up to the problem. Indeed, it is possible that the more pertinent objections to including post-1700 editions would disappear if simultaneously with the selection of editions an additional principle were proposed governing the selection of the variants to record.

One may start by freely accepting the proposition that any attempt to select a workable number of key editions from the eighteenth century to the present would be doomed to failure: the gaps between them in the history of Shakespeare's text could suppress almost as much important historical information as might be furnished by the selected editions, and any truly useful textual conspectus would involve too large a number of editions to record than would be manageable except on variorum principles.

Some years ago I was asked to formulate detailed proposals for an old-spelling Shakespeare edition, although one that unfortunately never got off the ground. After much reflection and experimentation I arrived at the concept of an historical collation that would provide complete records of editions up to 1700 but would confine its records of modern editions to those in the major twentieth-century textual tradition, leaving all eighteenth- and nineteenth-century editions to the Variorum. The justification went as follows. First, the substantive emendations footnotes would include the earliest known source for any accepted variant from the copy-text, no matter where found, and thus every emendation adopted in the old-spelling edition would identify the eighteenth- or nineteenth-century editor who originated the reading whenever it had not been found in pre-1700 editions. Since it was inevitable that a conservatively edited old-spelling text would hew, very much in general, to the most tried and accepted line of emendations, the post-1700 significant variant readings would have their originator named in the emendations note. Thus an historical conspectus of the most commonly accepted Shakespearean emendations would be provided automatically, sufficient in this specific respect for the purposes of a non-variorum edition. Of course, to confine the historical record only to the accepted emendations would leave a number of debatable, but sometimes equally popular, traditional emendations unrecorded when they had not been adopted by the old-spelling editor. However, in considerable part this gap was to be covered by the historical collation of the most significant twentieth-century editions, which (along with any new emendations proposed by recent editors) would also record all their substantive readings that differed from the newly edited

copy-text. If proposed emendations from the past of this sort were of such dubious account as not to have been accepted by at least one of the collated modern editors and thus to enter the record, not very much would be lost, perhaps, for the purposes of the kind of edition proposed, and its readers.[21]

Second, the restriction of historical-collation entries to pre-1700 editions and then to selected twentieth-century ones removed a great deal of rubbish. It cannot be said, by any means, that every modern edition chosen represented the state of the art in presentday textual criticism; but at least a conspectus was given in historical perspective of what the more recently edited texts were like. This information is of as much concern for a reader of an old-spelling text as is the record of unauthoritative pre-1700 editions. As A. R. Humphreys most nicely remarks in his introduction to the New Arden 1 Henry IV, "the variants in later Qq and F have authority no greater than that of later editors, the authority merely of the light of nature" (p. lxxv). We may hope that the light of nature shines somewhat more brightly in recent years, for even a dwarf may see a sunrise more clearly when perched on a giant's shoulders.[22]

The exact editions to choose can scarcely be prescribed. My own choice was to start with the Globe, since—as a slightly corrected derivative of the Old Cambridge text—it stands as the edition above all others on which subsequent textual tradition was built, and it is not unknown today in various reprints. Peter Alexander had to be chosen for the intellectual acumen with which (though somewhat unevenly) he revised a botched text that had been handed him. Moreover, his edition is justly admired and used in England. In the United States G. L. Kittredge's was standard for years although in fact it made very little contribution to the development of a new text despite Kittredge's vast erudition, largely confined to the notes in his splendid Sixteen Plays edition. J. Dover Wilson's pioneering New Cambridge texts, though often fantastic in their guesswork, made sincere attempts to apply

bibliographical considerations (even if idiosyncratic) and cannot be ignored. We are then confronted with plays generally edited by a consortium, such as the New Arden, the Pelican, and the New Penguin, but nevertheless series that were under good editorial direction. The results are mixed but the texts must be recorded as making important contributions to the presentday tradition. The popularity of the much criticized and perhaps overconservative New Riverside, edited by G. Blakemore Evans, and also that of the David Bevington edition (revised) for Scott Foresman require their collation, the more especially since the comprehensive Spevack Shakespeare concordance is keyed to the New Riverside.

These nine comprise what for the moment might be regarded as a master list, to which must shortly be added the Wells-Taylor Clarendon edition. Other editions or series have their points but need not have been edited with equal care or expertise, or else the volumes may not have had such wide circulation. The odds greatly favor that significant emendations from the past that have had any scholarly acceptance will be represented at least once among these editions and thus will be represented in the historical collation.[23] The combination of the information in emendation entries with that in the historical collation of an old-spelling edition would offer a true survey of substantive alteration (whether accepted or rejected by the old-spelling editor) that is representative of presentday scholarship in historical perspective starting with what may properly be called the first modern edition, the Globe derivative of the Old Cambridge.

If the problem of the selection of post-1700 editions to collate may be solved thus, the second problem of what to collate demands attention, for 'substantives' is a slippery term. Here I think Dr. Werstine, as I read him, is overconcerned that a simple historical record of rejected readings in modern editions would fail to provide a reader with the rationale of the editors who adopted the emendations, at least without the old-spelling editor being forced to provide an impossible array of textual notes discussing divergences in most disputed readings.[24] My own view is that the textual notes in an old-spelling

edition should be governed exclusively by the editor's concern to explain his own choice between debatable readings of most interest and importance. The reasons for preceding editors having adopted one or other reading (if known) may be of interest in such a note only in respect to the old-spelling editor's analysis of the bibliographical and philological evidence as justifying the acceptance or rejection of a reading in his own text. However true it may be that many presentday commercial editions are likely to follow a textual tradition alien in its modernization to that of the old-spelling, and hence one that is of little significance to record, nevertheless such a view does an injustice to those scrupulous new-school editors who have, on the evidence, surveyed the text with a fresh eye and whose major substantive readings are as individually considered as they would need to be in the more patently scholarly old-spelling edition.

More important, perhaps, is Dr. Werstine's implication that the varieties of readings caused by different editorial policies (in the nineteenth and presumably in the twentieth centuries) either muddy the waters or are not worth recording. Here, it seems to me, his thesis tends to confuse true substantive variation with simple variation caused by modernization. This latter category, I take it, can be handled by a few arbitrary rules that will not reduce the value of what matters in the historical collation but will concentrate the entries on the only important variants from the point of view of an old-spelling edition. Only a few examples need be given to illustrate the kind of modernizations that the collation can omit. We may agree, I think, that modernizers who read twenty pounds or twenty years for the old idiomatic usages twenty pound and twenty year need not be recorded. Usually there is little consistency in editorial treatment of the you-ye doublet or the hath-has, doth-does, thy-thine, toward-towards, 'em-them, a-an, and-an. What modernizing editors do to such small-fry readings is of no concern to an old-spelling reader.[25] Of no concern, also, are modernized spellings or elisions intended to lead a presentday reader to the correct metrical value, or pronunciation, as of ne'er for old-spelling never, and the like.

A middle ground is perhaps present in the general tendency of modernizing editors to restore agreement of subject and verb. This is the sort of variant that probably needs recording since it does involve a substantive form (usually a plural or singular of the verb) and one cannot count on unanimity of opinion among editors. Since no blanket statement can be made that would

cover all editions, and since there will always be ambiguous cases where it may be a matter of opinion whether the subject is a plural or a compound singular, listing in the historical collation may seem useful even if the alterations first occurred in post-1700 editions and the entry did not carry on a pre-1700 variant with the record of modern editions. Moreover, in case of doubt as to the exact meaning an old-spelling reader can always be alerted to refer to the historical collation for a conspectus of modern opinion on the authorial intention.

If we stick very narrowly to a definition of substantive reading as excluding ordinary modernizations,[26] some doubt may be cast on the application to problems of the historical collation of Dr. Werstine's distinction between the treatment of text found in modernized and in old-spelling editions: "The modernizing editor is free to emend as extensively as is necessary for his audience's comprehension of the text.[27] . . . Yet an old-spelling editor may vary from copy-text only when he can demonstrate in his notes that the text is probably corrupt, can identify convincingly the source of the corruption, and can substitute an emendation likely to repair the corruption.[28] Collation of an old-spelling edition with recent modernized texts serves merely to record the thoroughly predictable effects on the Shakespeare text of such differences in editorial policy and illustrates the influence that time, taste, and audience have exerted on modernized editions. . . . Variation simply results from individual editorial policies, each quite different from an old-spelling editor's policy" (pp. 100-101). This is certainly an overstatement if (a) we exclude the small change of modernizations that can be brought under the heading of substantives only by a wrench and that are not the concern of a sensible and practicable historical collation geared to the needs of an old-spelling reader; (b) we confine the post-1700 editions collated to a selected group of the most influential and respected of the twentieth-century modernized texts.

It is a curious concept that when real substantives are concerned the editor of a careful modernized text is inclined to emend more freely than

an old-spelling editor. After all, both are equally concerned to present what they believe Shakespeare actually wrote; their concerns differ only in adjusting the forms that these words may take. It is very likely that Dr. Werstine's views about the policy proper for an old-spelling text are unduly conservative (as evidenced by his repetition of Greg's dicta that are sometimes impossible to implement), or else he worries unduly about listing in an old-spelling edition's collation some of the more far-out guesses in which, say, Dover Wilson or Charles Sisson occasionally indulged. Moreover, some of his concerns about overloading the collation stem from a few examples of nineteenth-century editors having mistakenly used the wrong copy-text, although even here—since the alternative text would have authority—its readings must be fully entered and there is no problem merely in appending the sigla for the modern editors who have followed such readings.

A properly constructed historical collation of Shakespearean texts, and one suitable for an old-spelling edition and the needs of its readers short of the variorum principle, lists no more variants than are of real concern to a reader intent on reconstructing the presentday textual tradition bearing on the actual words that different editors have taken Shakespeare to have used, despite the fact that this tradition is at present and in the foreseeable future, unfortunately, confined exclusively to modernized texts.[29] Even so, the better modernized texts are as closely concerned with true substantives, in a lexical sense, as would be any series of old-spelling editions, and so are as worthy of record.

Notes