:: :: University of Virginia Library

accidental — made in the copy-text by the editor(s) of a given edition. The essential parts of each entry are simply the page and line citation, the reading of the edited text, the symbol representing the source of that reading, and the rejected copy-text reading. The general form which these items usually take includes a square bracket to signify the lemma and a semicolon to separate the source of that reading from the copy-text reading which follows:[35]

10.31 whom] W; who

Another possibility, employed in the Melville edition, eliminates the bracket and the semicolon and places the two readings in separate columns. It could perhaps be argued that this scheme makes the list slightly easier to use for purposes of surveying the nature of the emendations as a whole or constructing various kinds of statistics about them, since the source symbols would more readily show up along the right side of the first column and the copy-text readings would have a common margin in the second column. In any case, if the list is limited strictly to those readings of the copy-text which do not appear in the edited text, no symbol is required after the second reading, since in each case it is by definition the copy-text reading. (In those unusual instances in which a deficient copy-text is rectified by intercalations from another text, so that the copy-text is in effect composite, symbols following the second reading are helpful, even though a separate list of the intercalations would presumably be available.) It is important, however, to understand the reason for setting up a list restricted in this way. So long as a historical collation is to be included in the apparatus, the readings of the copy-text could be ascertained from it and would not necessarily have to be presented in a separate list. But this arrangement would be awkward and inconvenient in two ways: first, since the historical collation is normally limited to substantive variants and since the reader may legitimately wish to know all emendations, including accidentals, the absence of a separate list of emendations would cause the historical collation to become an uneven mixture, combining a complete record of substantive variants with an incomplete record of variants in accidentals; second, discovering what emendations had been made would be somewhat less easy if notation of them were imbedded in the larger historical collation. There is no question about the necessity of having at hand a record of all editorial

alterations in the basic authoritative document chosen to provide copy-text; and the greater convenience of having that record as a discrete unit, along with the resulting greater consistency of the historical collation, provides compelling reason for what might otherwise seem a superfluous or repetitive list. The situation is a good illustration of the principle that some sacrifice of economy is more than justified if the result is truly greater clarity and usefulness.

In the light of this summary of the general rationale behind the idea of a separate list of emendations, two common variations in the basic form outlined above are worth examining. One is the segregation of substantives and accidentals into two different lists. This system is used in Bowers's Dekker and the Cambridge Beaumont and Fletcher (where emendations in substantives are listed at the foot of the page and emendations in accidentals at the end of the text), as well as in the Virginia edition of Crane's Maggie and the Wesleyan edition of Fielding's Joseph Andrews (where both lists come at the end of the text). Since the purpose of a list of emendations is to make the whole range of emendations easier to examine and analyze, it follows that under certain circumstances — particularly when there is an especially large number of emendations in accidentals — the separation of substantives and accidentals will make such examination easier still. In other words, if the total number of emendations is small or even moderate, little is gained by exchanging the simplicity of one list for the complication of two; but when there are a great many emendations, with the possible result that the emendations in substantives would be obscured by being included in the same list as a large number of emendations in accidentals, the data may be much easier to use if the two categories are listed separately. To do so is only to extend the principle of convenience and clarity on which the whole list is founded in the first place. And, by a further extension, certain large categories of automatic alterations within the list of accidentals itself may be separated so as not to overwhelm the other individual alterations of probably greater significance. For example, in the Northwestern-Newberry Typee 224 words ending in "-our" in the British copy-text are changed to the American "-or"; once the policy of making this category of changes is adopted, the changes themselves are automatic, and to list all 224 instances in a list of emendations would place an unnecessary impediment in the way of using the list to trace the more important alterations. Yet it is unwise to make any textual changes silently;[36] so these

224 alterations of spelling are recorded in a footnote to the textual essay. In the Dewey edition many emendations are made in the capitalization of words standing for concepts, and these emendations are gathered into a separate list of "concept capitalization."[37] Whenever there is a large separable category of emendations, this practice is a useful way to avoid, on the one hand, overburdening the main list and, on the other, risking the dangers of silent emendation.

A second variation from the most basic form of an emendations list is the inclusion of at least some of the further history of the rejected copy-text readings. That is, instead of providing simply the copy-text reading at those points where the copy-text has been emended, the entry includes the sigla for certain other editions which agree with the copy-text and sometimes includes the full history of the reading in the collated editions. One often-used plan,[38] following Bowers's Dekker, is to trace, at each point of emendation, the readings of all collated editions (that is, all which might contain textual authority) down to the earliest which can serve as the source of the emendation. Thus in the entry

IV.iii.19 we] Q3; me Q1-2

the earliest edition to contain the adopted emendation is Q3, and the history of the reading down to that point is given (Q1 and Q2), rather than just the copy-text (Q1) reading; what the history of the reading in any collated editions after Q3 may have been is not revealed here

but can be ascertained from the historical collation. In other editions (such as the Howells and the Irving), the complete history (in the collated editions) of the readings at certain points of emendation is given, either explicitly or through a specified system of implication; when this plan is followed, none of these entries reappears in the historical collation, which is then limited to rejected substantives. In both of these arrangements, the distinction between the historical collation and the list of emendations has been blurred to some extent; as a result, the functions of these lists are less clear-cut, and therefore more cumbersome for the editor to explain and less easy for the reader to comprehend. Including in the emendations list the history of the readings down to the point of emendation means that the emendations list becomes partly historical in function and repeats part of the material from the historical collation; but the presence of some historical information in the emendations list does not obviate the need for turning to the historical collation, since anyone wishing to examine the evidence available to the editor at a given point of substantive emendation must look at the historical collation in any case to see if there were variants in editions later than the one from which the emendation is drawn.[39] In the other system, the emendations list takes over even more of the function of the historical collation — indeed, it becomes the historical collation for certain emended readings.[40] And though none of this material is repeated in the other historical list (now containing only rejected substantives), there is no one place where the reader can go to survey all the evidence at the editor's disposal relating to substantive variants. So long as it is agreed in the first place that there is value in having a separate list of emendations, the simplest way of dividing the data is to make one list strictly a record of emendations and the other strictly a historical record. The functions of the two lists are then easier to understand, and the lists are correspondingly easier to refer to and work with.

The question of distinguishing between substantives and accidentals is often not an easy one, because some alterations of punctuation, for instance, do have an effect on meaning; but unless the substantives and accidentals are to be placed in separate lists, the question does not arise in constructing the record of emendations. Two other basic

problems of definition always have to be faced, however: since the list aims to enumerate emendations in the copy-text, the editor must have precise definitions of what constitutes an "emendation" and what is meant by "copy-text" if he is to have a firm basis for deciding what to include in the list and what to leave out. In practice, defining the two concepts becomes a single problem, for however one is defined affects the definition of the other. Editors of critical editions[41] generally agree that there is no point listing as emendations such changes as those in the display capitals at the opening of chapters, in the typographical layout of chapter headings, in the length of lines, or in the wording of running titles. Whether an editor defines "emendation" so as to exclude changes concerned with styling or design, or whether he defines "copy-text" to exclude purely typographical features of the text, the result comes to the same thing in the end. Technically, of course, an "emendation" is simply a correction or alteration, and it is the qualifying phrase "in the copy-text" which through precise definition serves to delimit the kinds of alterations to be listed. There should be no difficulty in defining "copy-text," if the distinction between "text" and "edition" is observed: "text" is an abstract term, referring to a particular combination of words, spelled and punctuated a particular way; "edition" is a concrete term referring to all copies of a given printed form of a text. Thus a "copy-text" is that authoritative text chosen as the basic text to be followed by an editor in preparing his own text, and it does not include the formal or typographical design of the document which embodies that text.[42] The type-face, the width and height of the type-page, the arrangement of headings and ornaments, and the like, are all parts of the design of an edition but are not elements of the text which is contained in that edition; similarly, the formation of letters, the spacing between words, the color of the ink, and the like, are not parts of the text embodied in a manuscript. It does no harm for an editor to enumerate certain features of design which he regards as nontextual, but it is not actually necessary for him to do so if he has defined "copy-text" carefully, for his definition will have excluded such details as external to the text.[43] Omission of any

notice of alterations in design does not constitute a category of silent emendations in the copy-text, since the design is not a part of that text at all.

One problem in the specification of copy-text is raised by the existence of variations within an impression. Such variations may be caused by stop-press corrections or by type which slipped or shifted during the course of printing. The precise definition of copy-text in terms of particular states of the variations obviously determines which of these readings qualify for inclusion in the list of emendations: thus if uncorrected formes are taken as copy-text, the only press-variants which would turn up in the emendations would be those adopted from corrected formes; and if the correct spacings at points where letters shift around are regarded as characteristics of the copy-text, the only variants of this kind which would be reported in the emendations list would be those for which no copy with correct spacings had been found. The decision as to whether correct or incorrect states are taken as copy-text may vary in individual circumstances, but the point is that the copy-text must be defined in terms of the specific variants within the impression which embodies it; for this abstract "text" must have one and only one reading at any given point,[44] and to define a copy-text merely in terms of an impression is not sufficiently rigorous, since more than one reading may exist at many points within various copies of that impression.[45] Because sheets embodying corrected states of some formes (or correct spacings of letters) will be bound with sheets embodying uncorrected states of other formes (or incorrect spacings of other letters), it is unlikely — when more than a few press-variants occur — that any single physical copy can be found which contains the entire copy-text.[46] Emendations in the copy-text, therefore, are not simply emendations in the text of a particular copy; and the copy-text remains an authoritative documentary form of the text, even though no one existing physical entity (or even no one physical entity that ever existed)

happens to preserve it. The exigencies of producing a book — the fact that the forme is the unit in printing and the sheet the unit in gathering a copy of a book together for binding[47] — makes it natural that the finished product may contain a mixture of states. One may have to examine a large number of copies of a given impression to discover the press-variants in it, and one can never be sure that any copies left unexamined do not contain additional variants. In the Ohio State Scarlet Letter, for instance, collation of eight copies of the first impression produced five variants, all examples of loosened type which either shifted position or failed to print. At four of the points of variance, some copies carried the correct reading; but in the remaining case one copy read "t obelieve," and the others read "tobelieve." Since the correct spacing in this one instance did not occur in any of the examined copies, it had to be listed as an emendation, whereas the other four variants do not enter the emendations list at all, since the correct form of each did appear in at least one copy.[48] If, however, another copy were to be collated in which "to believe" appeared correctly, that form would no longer be an emendation and should not appear on the emendations list. As with any other research, the conclusions must be based on the evidence at hand; and that evidence, in any inductive investigation, is probably incomplete. If the number of surviving copies is small, one can examine all the available evidence and still be far from the truth; if the number is large, one may reasonably wish to set some practical limits on the extent of the investigation. But in either case the results are liable to modification by the next copy which turns up. The danger is unavoidable; but at least one can operate with precision and rigor within the limits of the located evidence. Part of what that entails is defining the copy-text in terms of press-variants (saying, for example, "the text in a copy of the 1850 impression with x at 172.15, y at 234.21, and z 278.11"), for only in this way can one know what constitutes an emendation and belongs in the list.

There are some variations among copies of a given impression which are nontextual and need not be reported, any more than differences in design between the copy-text print and the critical edition

need be specified. Usually it is not difficult to distinguish between these nontextual press-variants and the press-variants of textual significance just discussed. They are frequently due to differences in inking or in the amount of damage which a particular piece of type (or letter in a plate) has suffered. Variations in inking need not be reported if all the letters are visible, but if the inking is so poor that some letters do not show up at all in any copy examined, the variation is in effect a textual one of the kind described above. Battered letters or marks of punctuation — whether or not the batter varies from copy to copy — can be silently corrected without involving textual emendation, so long as there is no question what letters or marks are intended. But if the damage is great enough to raise possible doubt about their identity, any attempt at correction becomes a textual emendation and must be listed. Thus if a dot appears in the middle of a sentence at a place where it could be the upper half of either a colon or a semicolon, and if no examined copy shows enough of the lower half for identification, a textual decision is required to correct the punctuation; or if a small mark appears between two words where it could perhaps be a hyphen, and if no examined copy clears up the matter, the editor's decision to consider it a hyphen rather than, say, a part of the damaged preceding letter is a textual one; or if a letter which ought to be "e" appears to be a "c" in every examined copy, the correction is a textual emendation. The importance of having access to a large number of copies for this kind of checking is obvious. Most editors rightly feel that it is unfortunate if their lists of emendations have to be overburdened with entries which are probably not really emendations at all and which might be eliminated if more copies were available for examination. But without those copies, there is no alternative to recording them as emendations, since there is no documentary proof that the copy-text contained the correct readings.

Finally, a few minor points about form should be noted. (1) First, the list will be clearer in the end if each lemma consists simply of the word or words which constitute the emendation, without any of the surrounding words.[49] Occasionally an editor will feel that it would be helpful to the reader to have an additional word or two of the context, to enable him to see more clearly the nature of the alteration involved, while he is looking at the list. It is difficult to say, however, what would be sufficient context for this purpose, but generally a few words would not be enough; and as the cited readings become longer, the actual

emendations become less easy to pick out, with the result that this approach makes the list more difficult to use (as well as less consistent, since there would be no way of defining objectively how much should be cited). The only times when a word in addition to the actual emendation should be reported are when the same word as the emendation appears elsewhere in the same line (so that one of the two words adjacent to the emendation is required to identify it), when a mark of punctuation is emended (so that the word preceding the punctuation—or after it, in the case of opening quotation marks — is convenient, and sometimes essential, for locating the emendation), and when something is deleted from the copy-text (so that the point of deletion can be located). (Even the first of these can be eliminated if one adopts Greg's device of using prefixed superscript numbers to indicate which of two or more identical words is at issue, but this system is perhaps somewhat less easy for the reader to follow.) (2) A second formal matter which might cause difficulty is the notation of a missing letter (or letters). When loosened type causes letters to shift, without any letters failing to print, there is of course no problem because the usual between-word spacing can be used (as in "t obelieve"); but when loosened type or a damaged plate results in the complete disappearance of letters, it is important to show that space for these letters exists. It clearly makes a difference whether a reading is reported as "race" or as "[]race," for the second shows that a letter has dropped out and that the original word was "brace," "grace," or "trace." These empty spaces can be noted in various ways. The Hawthorne edition simply uses a blank space, which works well enough between words but is less clear if the missing letter is at the beginning or end of a word; Kable's edition of The Power of Sympathy [50] employs a caret to mark the space, creating an ambiguity since the caret is also used to signify the absence of punctuation; and the Melville edition uses square brackets, which may be somewhat cumbersome but are fairly suggestive and do not conflict with another symbol. (3) Another question of notation concerns those emendations which are in fact additions to the copy-text — that is, words or passages for which there is no counterpart in the copy-text. One common editorial device is to use the abbreviation "Om." or "om." to signify the lack of corresponding text at a given point in the copy-text. If the abbreviation is specifically defined in this way, it is clear enough; but if it is not explicitly defined and is allowed to suggest "omitted," it can be misleading, since the omission of anything implies

that something was available to be omitted, whereas the additions to a copy-text are often passages not yet written at the time the copy-text was completed. A phrase like "[not present]," which suggests no direction of change, would avoid the problem and would require no explanation.[51] (4) There remains the question of adjusting the symbols for editions and impressions to take variant states into account. If one of the uncorrected formes of a particular sheet is taken as copy-text but requires emendation at several points from the corresponding corrected forme, the symbol indicating the source of the emendations must note the state involved. For hand-printed books the conventional method is to attach a "u" or "c" in parentheses to the symbol for the edition — "Q1 (u)," "F2 (c)" — though of course superscript letters could also be employed. For later books, if the symbol for a given impression ends with figures, states can be represented by suffixed letters ("A55a," "A55b") or — regardless of the makeup of the symbol — by superscript letters ("A55a"); these letters signify the sequence of presently known states of individual readings within an impression (not necessarily "uncorrected" and "corrected" states of formes).[52] Because no single copy of a book may contain all the uncorrected or corrected formes, or all the earliest or latest states of variants, these attached letters — for books of any period — must be understood to refer, not to physical "books" (that is, not to entire copies of a given impression), but to readings that may or may not be present in any individual copy of the proper impression. A copy containing one Q1 (u) reading may contain other Q1 (c) readings, or a copy containing some A55a readings may have other A55b readings. For this reason the superscript letter may have an advantage over the suffixed one in emphasizing the fact that it is essentially a different kind of symbol — referring to a stage of variation at a particular point within an impression, not to the whole impression (or edition), as does the basic symbol to which it is attached. Many formal matters such as these may seem of minor consequence in themselves, but, taken together, the decisions regarding them may make the difference between a list of emendations

which is cumbersome and perhaps misleading and one which is convenient, logical, and easily understood.