University of Virginia Library

Regularization and Normalization in Modern Critical Texts
by
Fredson Bowers [*]

A critical text differs from a diplomatic reprint, or the modification that historians call a documentary edition, in that the editor in various specified and usually recorded details has altered his chosen copy-text in order to correct error; to impose some degree of conformity by reducing inconsistency of spelling, punctuation, and other accidentals; or to insert authorial revisions from another document.[1] Such alteration is of course emendation, for strictly speaking any change that an editor makes in his copy-text comes under the head of emendation. But some emendations are more equal than others. Obviously those that concentrate on the words, the substantives, of a text are of prime importance, and the success with which an emending editor deals with these is usually taken as the ultimate test of his expertise. His treatment of the accidentals—the texture clothing these substantives—so long

as it does not attract attention by eccentricity is largely taken for granted and is seldom given notice in a review.

The difference in specialized attention allocated by many editors of modern texts to the reconstruction of authoritative substantives as against the accidentals can be accounted for by reference to the history of editing. For earlier literature there exists a clear line of demarcation between modernized editions and what are known as old-spelling texts. Yet whether modernized or unmodernized, the critically adjusted substantives of either genre should be identical; thus the distinction lies in whether the editor has preserved unmodernized the older forms of spelling in all their irregularity, erratic capitalization, and early rhetorical system of punctuation, or whether all accidentals have been modified to agree with presentday usage. When an old-spelling edition is planned, an editor is forced by the nature of his copy-text to pay scrupulous attention to the critical treatment of these accidentals on a par with the substantives.

For nineteenth- and twentieth-century writing the question of modernized versus unmodernized accidentals is not one of such practical import as to require two different types of edition, as is necessary with Shakespeare for example. Spelling has become largely codified although with some trans-Atlantic variation, the so-called emphasis capitals of the eighteenth century and earlier have almost wholly dissolved into concepts, a generally recognized syntactical system has replaced the older rhetorical punctuation, even though different varieties may be seen in its progress such as heavy or light, parenthetical or fluid, formalized or informal and flexible. And if some aspects of this punctuation, spelling, and other accidentals are not entirely current in our year of 1989, yet a presentday reader has little or no difficulty in adjusting to what is a perfectly intelligible system for the transmission of meaning even though its surface may appear to be slightly quaint.

In these circumstances once the copy-text has been selected the editors of modern writers have been inclined to neglect the accidentals except for the correction of misprints and other positive manifestations of error. This benign neglect, with its acceptance of the view that editorial attention has been satisfied in these matters by a relatively faithful reproduction of the copy-text, means that in effect an editor is applying a double standard: critical analysis of the substantives but diplomatic reprinting of the accidentals. Yet depending upon the textual situation—as whether an editor is faced with a single authority or with multiple authoritative documents—at least as many opportunities (if not more) are likely to exist for critical treatment of the accidentals texture as are found among the substantive building blocks. Thus despite the lesser claim that accidentals may have on the transmission of absolute meaning,[2] their full inclusion in the process of critical editing is as important to the authenticity, the "definitiveness," of the text of an edition as is any other factor.[3]

My overall concern is with the recovery of authorial accidentals within transmitted documents in whatever detail is possible and their controlled restoration in the copy-text as replacements for non-authorial forms produced

in the process of transmission. I hope to be concerned with a few principles that evolve from the need to deal with the evidence, and their possible application, moving from an originally conservative position to what I suppose we must call a liberal extension. For my present narrow purpose I dismiss from consideration the various editorial problems both of copy-text and its treatment that arise when an authorial manuscript and its printed result compete for general as well as specific authority. Instead I propose to deal at the start only with what can be done to reconcile variant accidentals within the same text, one form of which would seem to represent the author and the other the compositor or, later, the copyreader. The purpose of this critical attention is to achieve an authoritative uniformity in certain details of the texture, a smoothing-out of non-authorial inconsistencies that are usually not positive error but are instead simple transmissional irregularities.[4]

Why should an editor trouble himself with the copy-text's variant accidentals when positive error is not in question? One answer concerns editorial integrity. It is accepting the double standard if an editor does not engage himself to recover whatever authority he can from the available evidence in the accidentals with as much care as he devotes to the substantives. Another answer concerns the effect of irregularity on the reader. I suggest that an important distinction between a diplomatic reprint of a document and a critical edition even of that one document can be the difference between the reader's apprehension of raw material and his apprehension of an intelligently and purposefully shaped text that in distinctive characteristics reflects as much as possible the author and not the publisher's intermediaries. The one is chiefly for specialized reference; the other is chiefly for reading and general information as well as pleasure. A scholarly text is no less scholarly if it takes as its ideal the form of a reading edition that appeals as much as it legitimately can to a user's sense of pleasure as well as profit. The days are long since gone when it is practicable for a publisher to issue an edition that is designed only for the professional scholar (medieval literature necessarily excepted) and does not seek among its purchasers the general literate reader with an interest in the work being printed, including collegiate students at an early stage in their training. I conceive that a reading text must attempt the delicate balancing act of simultaneously appealing to both audiences without loss of scholarship but also without arbitrary and sometimes pedantic barriers to general intelligent appreciation.[5] That this compromise is both desirable and practicable may be illustrated by the growing tendency of publishers to issue a critical scholarly edition with all its accompanying apparatus, and then to abstract the text alone, by offset, for a general audience. The Harvard University Press has done this with William James's Pragmatism and his Principles of Psychology and proposes a paperback of Varieties of Religious Experience. With some modifications, the Library of America has also adopted this formula (except for one or two in-house editorial misjudgments), with acclaim from reviewers and one hopes from purchasers.

It is important that a reader trust the text he is using, that he should

feel a symbiotic relationship with its editor. In this matter scholars of course emphasize accuracy of transcription and the full record in apparatus of departures from the copy-text so that editorial intervention can be checked and analyzed. The nearer the text approaches the methods of diplomatic transcription, the more a scholar is prepared by his training to accept the faithful reproduction of the warts usually found in documents that are close to their original source. On the contrary, the nearer a text approaches the methods of a reading edition, the more it contrives to make uniform certain accidentals features that in their irregularity may create some unease in a user who is unaccustomed to accept without momentary interruption of concentration what are surface inconsistencies in the text. If in the fringes of his consciousness a reader is even slightly distracted by wondering whether he is faced with a misprint or whether there is a hidden significance he has missed in various anomalies—and whether these are authorial or transmissional—then the interruption of unselfconscious absorption diminishes his confidence, and certainly his pleasure, and may affect his future use of such texts.

I agree that a reading edition need not and should not require the rigorous mechanical smoothing-out that the usual professional copyreader is likely to apply to a commercial manuscript about to be sent to the printer. But a critical editor can learn a lesson from those intelligent, though rare, examples of copy-editing—in Britain often by persons who may themselves be writers—which attempt to make the presentation uniform in terms of the particular author and his idiom and idiosyncrasies of style with a minimum of housestyling. Sympathetic assistance of this kind (sometimes found in The New Yorker magazine) does not impose an alien texture on a writer's work. But neither does it allow potentially troublesome variance in the accidentals that is not part of the authorial design. A critical editor as enlightened copyreader is not an immediately appealing concept, but within rational (meaning flexible) bounds it is not only true but even desirable.

The two means by which a critical editor creates some order from anomalous irregularities I label regularization and normalization. I apologize for the jargonistic flavor of these terms but can think of no better ones. I construe regularization as the bringing of inconsistent elements in a text into conformity by the adjustment of variants to some one regular form already present and assumed to be authorial. Normalization I conceive as imposing an external standard of regularity without the evidence of some specific precedent in the text being edited, but one that is guided by evidence derived from similar authorial documents. Insofar as by definition nothing in a text (except perhaps analogy) encourages normalization by offering a precedent as does regularization, normalization is usually less required for readability in modern writing than in older literature, and as a consequence it is to be used with greater caution.[6] Publishers' housestyle, the imposition of which copyreaders are paid for, is normalization with a vengeance and has no place in critical editing, of course. (In an edition of older literature it would be called modernization.)

Variant spelling and word-division bulk large among the inconsistencies one may find in uncopyread texts, especially those of the turn of the century. The origin of such variation is usually compositorial even though authors were not always themselves consistent.[7] Especially after Webster developments were in progress in which certain spellings characteristic of the United States broke off from the older-fashioned ways derived ultimately from Samuel Johnson's Dictionary through British models. Words such as honour or colour with the -our ending were being changed to the alternate spelling with -or. The x-spelling in connexion was giving place to connection, and the e-spelling in develope and developement or in judgment was being dropped. Words like towards and forwards were becoming toward, forward, and the like. I am no historian and so cannot date the popularization of these changed conventions, but I am conscious of a growing irregular transition in the United States from one system to another as the nineteenth century wears on and the twentieth comes into being. Although progressive printers adopted the newer forms earlier than many authors, not all of the compositors were in agreement or were able to make their styling uniform. Moreover, the influence of copy and thence of an author's older-fashioned spelling and word-division habits might occasionally tilt a compositor toward a variant spelling or division he would not consciously, let us say, have set.

I suggest that within a literary unit there is no reason for a critical reading text to perpetuate transmissional variant spelling and word-division systems, and every reason not to trouble a reader by requiring him to adjust and then re-adjust his sensibility to variance between what he becomes accustomed to in a particular text and what is at odds when this variance has no point and is generally unauthoritative. Given a choice between two spellings or word-divisions in a unit it seems obvious that a critical editor should select as his accepted form that variant that agrees with the authorial preference whenever this can be established from his manuscripts, including letters.[8] The authorial form chosen as the basis for regularization will ordinarily be less current and more idiosyncratic than the compositorial alternative, but this uncommonness is actually an argument in favor of its adoption.[9] As remarked, it is the duty of a critical editor to do his best to preserve authorial characteristics in the accidentals of a print with as much care as he devotes to establishing the authority of the substantives. In respect to a reading edition, moreover, it is psychologically true that if a system is uniform within a text, the reader will readily adjust to less familiar though recognizable spellings and compoundings. It is inconsistency alone that is troubling.

The question arises of the limits within which an editor can regularize certain inconsistent accidentals. The conservative view is that regularization can best be defended when it is confined to some unit that is complete in itself, like an article, an essay, a short story, or a self-contained book, whether of exposition or of fiction. Obviously, chapters of a unified book should not be held to represent independent units that may be variously treated: the evidence from one chapter is transferable to every other. On the other hand, if the principle of regularization be upheld rigorously, the conservative view

would require that an after-the-event collection of essays or short stories be allowed different sets of accidentals characteristics from units to unit on the ground that the copy was almost certain to vary. A more liberal view would offer an editor the right to regularize, at least, throughout a collection sponsored by the author.

The theory of limitation by unit is based on what documentary evidence is present to suggest the nature of the accidentals in whatever happened to be the copy underlying the print. If this is known or hypothesized to be a holograph or authorial typescript, the problem is solved. But in most circumstances an editor can reconstruct only the copy directly underneath the print. If this is, say, a lost professional typescript, the odds are minimal that all of the authorial accidentals will have been preserved. When to this distorting layer of transmission is added compositorial styling, the only documentary evidence that can be trusted for emendation back to authorial forms is variation in the print, one form of which should be compositorial but the other the reading of the copy. It is then a separate problem requiring collateral evidence to establish whether either form is or is not authorial. Except when the print reproduces marked authorial eccentricities like William James's simplified spelling or Stephen Crane's adjectival punctuation, coincidence with established authorial preferences may be only fortuitous even though the most plausible assumption is that they have been directly transmitted. The nature of the copy therefore defines and limits the nature of the unit. Units set from different copy are likely to exhibit different inconsistencies, the evidence of which, in such circumstances, cannot usually be transferred from one unit to another except by the different process of normalization.

When regularization is strictly reserved for units, some may be so uniform that regularization is impossible in the absence of the variation that permits choice from among internal precedents. (In such cases whether normalization or extended regularization is appropriate is a separate problem to be considered later in connection with The Scarlet Letter.) A not untypical example of a smallish unit where the internal uniformity offers only limited opportunity to regularize a word according to its authoritative form is William James's "Human Immortality" (the copy almost certainly holograph), which was printed as a pamphlet in 1893 by Houghton Mifflin (edited in Essays in Religion and Morality, 1982).[10] We know from manuscripts that James almost invariably did not divide words like some one and any one but here they appear divided, and since nowhere in this essay are they undivided as evidence of the forms in the printer's copy, an editor who rejects normalization must accept them as printed. Similarly this essay contains non-Jamesian forms such as naïvely with a dieresis that James never used, connection where he preferred connexion, and fullness where he usually wrote fulness. But none of these are variant, and hence a strict editor would leave them untouched. We may be thankful that the printer did not filter out James's characteristic whilst, farther, develope, and especially his lowercase christian, an idiosyncratic treatment of certain capitals. We may be thankful, also, for the single opportunity offered by a compositorial slip.

Although the printer twice set preëxist with a non-Jamesian dieresis, in the same footnote (89, fn.5) comes the true Jamesian form with a hyphen, preexists, which gives a regularizing editor his evidence on which to alter the two forms with dieresis.[11] Another opportunity to emend the uncharacteristic dieresis comes in James's "The Sense of Dizziness in Deaf Mutes" (Essays in Psychology, 1983). Here at 138.4 the journal printed the conventional coördination, which James must have missed in proof since within a few lines we find three times (158.10, 18, 21) his characteristic co-ordination with a hyphen. Similarly, in Principles of Psychology (edited 1981) the various times in which revery appears with a y, as in James's manuscripts, permits an editor to regularize the variant reverie (548.6) to the spelling revery which in fact comes only a page later (549.11).

One of James's quirks was his custom, remarked on by his brother Henry, of reducing to lower case the normal capital in adjectives derived from proper names, his practice also in the corresponding nouns though less frequently. This eccentricity was, of course, subject to housestyling almost completely, but every once in a while the copy form will slip through and on its evidence one can regularize. For instance, in Lecture I of Pragmatism (edited 1975) we find Darwinism with a capital (16.22) passed on to the book from its copy in the Popular Science Monthly; but an editor should emend to James's characteristic darwinism after this form is observed in Lecture II (39.27) in the same transmission. One does not need the extra confidence to make the emendation, but it is useful to observe that in this book James thought enough of preserving his practice to alter magazine Hegelian to hegelian either in copy or proof in the same Lecture I (16.15), and so with euclidean and aristotelian in Lecture VI (107.6). Interestingly, in Lecture VII, set from manuscript, James inadvertently wrote Latin with a capital, but (surely in proof) it appears in the book as lower-case latin (116.6) and in lower case both in manuscript and book a few lines later (116.10). These are just a few examples from single-text regularization in which the evidence is self-contained and internal.

Something of a halfway house between regularization and normalization comes in the fortunate instance of multiple authority when one or other text reveals the true Jamesian preferences as against housestyling in the other. The classic case exists whenever the setting manuscript for the print is preserved, although a precedent draft manuscript, and even a collection of notes, is not to be ignored. For instance, Lectures VII and VIII of Pragmatism were set in the first edition of 1907 from a preserved holograph that not only demonstrates authoritative forms for certain words but also illustrates how the book compositor(s) altered copy away from James's characteristics. As examples, in these chapters the book altered manuscript everyone to every one, anyone to any one, englishman to Englishman, italian to Italian, russian to Russian, french to French, european to European, greek to Greek, naively to naïvely, cannot to can not, and criticizes to criticises, as well as a range of variants in compounded words.[12] Lecture III (in part) and all of Lectures IV and V must have existed in similar manuscripts, now lost but

written at substantially the same time in series. As one would expect, the book text exhibits the same non-Jamesian forms that had been altered from the manuscript of Lectures VII and VIII. It follows that on the unitary evidence of Lectures VII and VIII, all identified Jamesian spellings and word-division forms found in the manuscript should be regularized throughout Lectures III-V set from the lost section of the same overall manuscript.

But one should logically carry this principle further (or as James would correctly say, farther). Lectures I and II were set in the book from annotated pages of essays published in the Popular Science Monthly in March and April of 1907, and Lecture VI was similarly set from an essay printed in March in the Journal of Philosophy, Psychology, and Scientific Methods. The manuscripts sent to these two periodicals would have been written in series with the manuscripts used to print the other lectures in the book and would therefore have had the same characteristic accidentals. That an extra layer of housestyling was added by pre-publication prints used later as book copy does not alter the principle that the evidence from Lectures VII and VIII is as valid for I, II, and VI as it was for Lectures IV, V, and part of III. This being so, it would be splitting hairs to assert that the part of Lecture III that was set in the 1907 book from a few annotated pages of a printed lecture in 1898 should not also be made to conform by regularization, even though no holograph manuscript for it existed in 1907 as was true for the other lectures.[13]

In sum, a fair number of accidentals in the book when it serves as copytext can be regularized by starting with the demonstrable evidence of the manuscript copy for VII and VIII, then enlarged by internal variation of other forms not present in the lecture manuscripts but one of the options identifiable as Jamesian from other sources. In this manner, with every emendation supported by documentary evidence within the unit of the book, a clean texture of spelling and word-division according to Jamesian authority can be reconstructed throughout the whole of Pragmatism.

Similarly, the setting manuscript for Chapters XIX and part of Chapter XX has been preserved for The Varieties of Religious Experience (edited 1985). In this area the book compositor(s) housestyled away from James's characteristics in the manuscript as follows: encrusted to incrusted, everyone to every one, anyone to any one, someone to some one, excentric to eccentric, connexion to connection, criticize to criticise, and farther to further. This evidence permits regularization in the parts of Chapter XX where the manuscript is missing of eccentric to excentric, any one to anyone, connection to connexion, and every one's to everyone's. Since there is reason to suppose that the earlier chapters of the book were set up from copy containing (at least ultimately) these same preferential forms, one can regularize the forms of the eight words throughout (with analogues) as well as what can be derived from different forms of compoundings.[14] To this extent, small as it is, a critical editor can reconstruct the authoritative original and repair these specific accidentals from the book's intrusive housestyling.

If an editor chose (and why not?) this list could be extended from a draft manuscript for Lecture I in which we have the typical Jamesian whilst,

centre, forever, the lower-case hegelianism and scottish, a number of spellings in z instead of s such as baptize, characterize, philosophize, and verbalizing (sufficient by analogy to remove all other s-forms in the book),[15] as well as the important distinction James made between so-called with a hyphen when used as an adjective preceding a noun as in so-called philosophy and without the hyphen when placed in such verbal phrases as logic commonly so called. Extending the list from a draft for Lecture II, we should have, with some duplication, greek, symbolize, centre, kantian, hegelian, german, christian, and anyone. Among the evidences for James's preferences that appear in a series of notes for the lecture series are colour, connexion, pre-existent, and cannot as well as more lower-case adjectives from personal names and various spellings in z.

Multiple authority as a source for regularization is not confined to the evidence of manuscripts, however. When one document served in whole or in part as annotated printer's copy for another, we may have a situation in which the text being edited could be consistent in the non-Jamesian form but the other text, set from this copy, printed the known Jamesian characteristic. In reverse, if the original text has the proper form but the reprint the housestyled variant, we may put this change down to the printer away from what had been earlier set from James's copy. A better case, of course, comes when a revised text by its frequent use of preferential forms indicates that James had annotated the copy with some care to secure what he wanted. An interesting example comes in the essay "Brute and Human Intellect" (edited in Essays in Psychology, 1983) in which the word imbedded in its American spelling is invariant in the original printing in the Journal of Speculative Philosophy. James's own spelling was the older-fashioned British embedded, which we find in his Principles of Psychology irregularly mixed with imbedded. However, in his Principles Chapter XXII, "Reasoning," where the annotated "Brute and Human Intellect" served in part as copy for the book, we find that journal imbedded appears as James's known spelling embedded, almost certainly by annotation (971.8-9;974.18). Despite three cases in this same chapter where James apparently did not annotate and the Principles repeats the journal imbedded (982.14;987.15;988.36), the significant double occurrence of embedded offers the opportunity to emend the journal in reverse when the article is separately edited and from these emended readings to regularize the remaining words in places where the Principles did not utilize annotated pages from the journal. Also, the evidence here for the annotation in the copy for the Principles to secure the spelling embedded more than justifies the regularizing throughout the Principles to the embedded Jamesian spelling.

In the same manner the Jamesian forms in the Principles can be used to emend (and thence to regularize) in various other journal articles used as copy such words as further to farther ("Brute" 30.17), for ever to forever ("Are We Automata?" 43.8), some one's to someone's ("The Spatial Quale 75.37), outward to outwards ("The Feeling of Effort" 96.6), paralysed to paralyzed (ibid., 96.8), while to whilst ("The Hidden Self" 255.38;256.17;257.4;

262.37;263.38;264.17,19,22), upward to upwards (ibid., 256.16), toward to towards (ibid., 259.26), nowise to no wise (ibid., 259.32), and reverie to revery ("What the Will Effects" 220.39).

Admittedly debatable but nevertheless logical are special cases when the actual text of the Principles repeats the non-Jamesian form throughout the section where the journal article is the copy, but on evidence elsewhere within the Principles the critical editor has been enabled to regularize by emendation to restore the Jamesian forms of these particular words. In such circumstances, one may argue, the emended Principles may serve as authority to restore Jamesian forms to the separately edited underlying journal texts. Examples from Essays in Psychology would be toward to towards ("Brute and Human Intellect" 50.33), any one to anyone (ibid., 31.29, and "Are We Automata?" 40.5, and "The Feeling of Effort" 87.32;93.23), fore finger to forefinger (ibid., 96.4), some one to someone (ibid., 105.2), every one to everyone ("On Some Omissions of Introspective Psychology," 155.11) and some one's to someone's (ibid., 156.13).

A more speculative but still logical procedure would be to group in a collection of essays, or the like, those pieces that had appeared in the same journal or magazine and to emend non-Jamesian forms by reference to the occurrence of their Jamesian counterparts in different articles but from the same journal and therefore subjected to the same housestyling. If this were done, one would then emend as partial regularization the imbedded spelling in "The Spatial Quale" (65.2) by reference to the two embedded spellings in "Brute and Human Intellect" regularized from imbedded on the authority of the alteration to embedded when this section was reprinted in the Principles. Or if this example seems too remotely authorized, one could, say, take a stray towards or anyone that had slipped through in some one journal article and regularize these particular words throughout other articles printed in the same journal. It could be argued that there is documentary authority, even though at one remove, for such changes and that they may thus be regarded as regularizations (not normalizations) on the ground that the journal would then be the unit and not the individual essay. An analogy would be an authorial collection of previously unpublished works which, being set from manuscript, could be viewed as a single-unit book and therefore subject to regularization throughout. As the examples of Pragmatism and The Varieties of Religious Experience illustrate, the preservation of the manuscript for even one of such essays would be highly useful and would add to the number of regularizing forms that could be extended to the others. But these two books also provide examples of variants from internal evidence elsewhere than in the manuscript sections that an editor was able to take advantage of in the process of regularization; hence in a collection of the kind envisaged some precedents would be found to serve as the basis for limited regularization even though no manuscripts were known.[16]

The authority of a holograph manuscript is of course greater for accidentals than that of a print. But holographs are not necessarily uniform in all respects of accidentals, and their inconsistencies while authoritative in a technical

sense should be subject in a critical edition (not a diplomatic reprint) using a manuscript as copy-text to the same process of regularization suitable for a print. In such manuscript variants needing regularization the number of precedents is useful for determining the norm to which variants of any word should be made to conform; but in case of doubt, and especially when numbers are ambiguous, an editor needs to assure himself that his selection agrees with the authorial preference in similar but not identical forms. When evidence in the copy-text is still insufficient for certainty, the forms in other manuscripts of roughly similar date will need to be consulted.

Simple regularization may be illustrated by an example from Nathaniel Hawthorne's Blithedale Romance. At 148.32 the setting-copy manuscript reads buttonwood tree without the hyphen supplied in the first edition. Authority in the print as against the manuscript could result only from authorial proofreading, the likelihood of which in this case is minimal. Fortunately, buttonwood-tree with a hyphen comes in the next paragraph of the manuscript (149.6) and again in the next (149.29). In the same paragraph as the unhyphenated word comes hyphenated apple-trees (148.15), peach-trees (148.16), and shortly fruit-trees (149.29) and pine-trees (156.20). This accumulation of evidence that Hawthorne strongly preferred to hyphenate the names of trees can also be applied as the decisive factor in The House of the Seven Gables as between the manuscript's unhyphenated pear tree (88.2) and its precedent pear-trees (87.3), aided of course by hyphenated damson-trees in the same sentence. Likewise, current bushes in Seven Gables (153.9) can be regularized to the hyphenated word not only on the evidence of current-bushes earlier (87.4) but on such analogues as hyphenated grape-vine. Analogy again is useful when the only precedent for ambiguous manuscript white pine-tree in the Romance (208.11) is the same hyphenation earlier (98.4). The book corrects both to white-pine tree and an editor should do the same, recognizing how the force of Hawthorne's preference for hyphenating tree names led him into a situation from which he should be extricated. Numbers are of little use in Seven Gables where in the manuscript unhyphenated Pyncheon house occurring four times (153.28;173.3;184.11;192.5) is surrounded by the hyphenated form five times (5.6;6.5-6;28.28;38.26;182.25). Thus it is helpful that we have hyphenated Pyncheon-street, Pyncheon-elm, and Pyncheon-garden. By good analogy, the single unhyphenated Pyncheon bull (231.2-3) can be normalized to Pyncheon-bull.

Hawthorne's Marble Faun poses some difficult problems in regularization in the spelling, owing to his attempt in the copy-text manuscript to assist the British printer who set the book by spelling certain words in what he took to be the English form instead of in his native American style. Unfortunately, his intentions were better than his practice, for his ordinary usage often peeps through the self-imposed alien system. The problem for a critical editor is which system he should choose as the basis for his regularization. At first sight it might seem that one could flip a coin. Convenience and presentday readability suggest the reversion to Hawthorne's normal American preferences. On the other hand, authorial intention (even though imperfectly carried

out) suggests regularizing what are clearly his lapses and adopting on a regular basis the system he was trying to achieve.[17] For better or worse that was the decision of the textual editor of the Centenary edition. Thus the -our spellings of such variants as colour-color, favoured-favored, honour-honor, endeavour-endeavor, and even terrour-terror, errour-error, and inferiour-inferior (rejected by the printer) were regularized according to Hawthorne's intention.[18] In this manuscript Hawthorne's spelling of some words was in transition, so that (the anglicizing aside) his own preferences wavered between the older-fashioned and the more modern forms that mix in this document, as in canvass-canvas where regularization seemed desirable to canvas as the spelling that he was beginning to adopt. Also in The Marble Faun Hawthorne used as part of his normal practice the doublet subtle-subtile interchangeably; but since no trend could be discovered (as in canvas) regularization was not attempted (see Textual Note to 45.13). Hawthorne wrote Signor and a few Signore without regard for their distinction in Italian. Since he was not writing in Italian, regularization to conventional Signor seemed more logical than an attempt to correct him.

In the absence of a manuscript, Hawthorne's early Fanshawe (edited with Blithedale Romance, Centenary 1968) with its numerous inconsistencies of accidentals poses an odd problem in determining the base for regularization, a problem requiring bibliographical analysis to solve. Nine spelling, capitalization, and punctuation variants, with a few extra minor matters, break down to definite pages assignable to five different compositors: endings in -our or -or, the spelling gray-grey, the doublets Doctor-doctor, Widow-widow, Dame-dame, Master-master, Inn-inn, Heaven-heaven, and the presence or absence of quotation marks about the epigraphs for each chapter, plus a punctuation formality or so. In addition, the use or non-use of dashes at the end of sentences of narrative prose to justify the lines at the right margin was valuable as well as the variable placement of a final quotation mark in relation to other punctuation. Without manuscripts there might have been problems with the spellings, but external evidence readily provided the information to regularize them. With an editor's knowledge of Hawthorne's strong preference for hyphenating tree names, it needed only the single appearance of hyphenated elm-tree in order to hyphenate elm, birch, and oak elsewhere with confidence. A clue to the capitalization of Inn as authorial was contained in the printer's error the Sun set by Compositor C, corrected to an Inn in the errata list. It seems evident that C misread Inn as Sun, and that only an ill-formed capital I could have been mistaken for a capital S. Capitalized Doctor was authenticated as the manuscript form since of the five workmen only Compositor D used lower case. Three of the five workmen set capitalized Widow. A number of accidentals variants yielded to authority by linking the evidence of compositorial preferences with the forms in Hawthorne's manuscripts. Thus in the end internal evidence authenticated by external produced a relatively uniform and authoritative texture that replaced the mishmash that would have come from a social-contract faithful reprint of the first edition.[19]

The regularization of punctuation is perhaps the most difficult and least successful, the more especially because authors themselves are not likely to be consistent. For example, William James often employed the English system of omitting a comma before the and in a series of three or more, but his manuscripts have it often enough to make an attempt so dangerous to regularize that any copy-text diversity must be preserved. Stephen Crane deliberately avoided separating by commas adjectives in a series before a noun with enough frequency that it can be marked down as a definite idiosyncrasy that must be preserved; on the other hand, his manuscripts do sometimes contain the conventional commas so that any attempt to remove all such commas in prints would be speculative normalization not to be ventured. The only exception would be if in a series of three such adjectives he wrote a comma after the first or the second but either carelessly did not insert the other, or else the comma itself was carelessly inserted by the printer. Regularization here, according to the main custom in the copy-text, would seem to be permissible.

In general, regularization of punctuation can deal with some authority only in the more formal features of punctuation that are most subject to compositorial styling. One example would be the compositorial custom of putting a dash after James's colon introducing an inset quotation. But in some years James himself wrote-in such a dash, rather at random; thus regularization when precedents are available within a printed unit is possible but may in a few instances be unauthoritative. Nevertheless, when the opportunity comes, with James the effort is worth while, and when a manuscript is the copy-text regularization to remove compositorial dashes will usually have the benefit of numbers. Another styling is generally resolved with ease. During the time that the present American system of placing punctuation in relation to quotation marks was slowly coming into being, old-fashioned compositors would set a semicolon ending a quotation inside instead of outside the quotation mark. For instance, Compositor A of Fanshawe twice set the closing quotation mark inside the semicolon, the only two occurrences in the book. Some authors agreed, but most seemed to have used something like the present system though with such carelessness that they were as likely to place punctuation outside as inside quotation marks and often in their manuscripts the mark is directly above the punctuation, leaving it up to the compositor to straighten out according to the accepted styling that had actually been intended. In such circumstances it may seem best to impose the modern system on the prints of American authors after the eighteenth century whether or not a manuscript is the copy-text.

Hawthorne illustrates another problem of placement. In his manuscripts he regularly puts a comma before an opening parenthesis and another comma before the closing. The compositors of The Scarlet Letter agreed, so that we read at 65.30-32, for example, 'But he opposes to me, (with a young man's oversoftness, albeit wise beyond his years,) that it were wronging the very nature of woman', just as the lost manuscript must have had it. The compositors of another printinghouse who set up the Seven Gables from a manuscript

with similar characteristics ignored the manuscript and regularly placed the comma after the closing parenthesis. In Fanshawe compositorial inconsistency in this matter offered the opportunity to regularize according to the occasional occurrence of Hawthorne's own preference.

Similar to this problem is the varying use of commas before the parenthetical use of dashes. William James, as one example, usually set off parenthetical matter by simple dashes although the compositors of his works almost invariably added the conventional commas of the time even if irregularly. Moreover, compositors were likely to insert a comma before a single dash that was not internally parenthetical but that marked a syntactical break at the end of a sentence. It is a moot point how far an editor should go in regularizing this comma-dash convention. If an author's manuscripts show that he is invariable in one or other system, any precedent in the print (no manuscript being preserved) may be used as the evidence for regularizing. These commas are often found when they have no syntactical purpose, that is, when the matter between the dashes being removed no comma would be natural or, at least, required. Thus any editorial attempt to regularize according to the interpretation of the syntax is almost sure to end in failure. When an author seems to have no marked preferences, it would be conservatively wisest to follow the vagaries of the print (and more certainly of a manuscript copytext) unless an editor is prepared to act as copyreader in this respect. But an author without any preference must be rare. If like William James a writer at some period develops a preference but occasionally slips, even a manuscript copy-text could be regularized with some confidence. Given a printed copytext, and an intimate acquaintance with the author, regularization will probably reconstruct the lost manuscript more often than it sophisticates it.

With Hawthorne the problem slightly clarifies. Hawthorne habitually wrote parenthetical dashes without commas: in the whole Marble Faun manuscript I doubt there are more than one or two such commas.[20] Under these circumstances it was probably timid in the Centenary Scarlet Letter to reproduce the compositors' variable mixture of commas versus no commas accompanying dashes.[21] If it had been possible to break down the compositorial shares in this book with any exactitude and so to identify the true characteristics of each workman, regularization would have had a firmer basis. Nonetheless, it should have been done, anyway, for Hawthorne's preference was clear from his manuscripts intended for publication and the book text followed this preference often enough to indicate that sentences with the comma-dash combination were compositorially styled.[22]

Doubtless editors will discover other formal features of punctuation that can legitimately be regularized on sufficient evidence. Unfortunately, ordinary syntactical punctuation in discourse is subject to so much inconsistent authorial variation that no writer can be regularized by second-guessing the probabilities that he would punctuate a particular passage in any precisely recoverable manner. Marked idiosyncrasies can be utilized from time to time, of course, although with caution, as for instance in Crane's inclination not to

punctuate an adjectival series before a noun. But it would be a bold editor who would regularize by removing conventional commas without evidence from an underlying manuscript of some sort.

I conclude with an illustration that may be of interest as dealing with a problem set by dialect in narrative. Readers of Stephen Crane's Red Badge of Courage when edited from the book-text must have been puzzled ever since the first edition by patches of singular incongruity in the speech of the tall soldier (Conklin), the loud soldier (Wilson), and even the protagonist youth. In Chapter I Conklin speaks in a heavy dialect (3.18-20). He is answered by Wilson in correct English (3.28-4.2). Conklin responds (4.19-23) in dialect to the questions of other dialect-speaking common soldiers, but when he next appears his dialect has been dropped and he speaks good English (10.26-29), this correct normal speech continuing through Chapters II and III. On the other hand, following Conklin's first non-dialect speech he is answered by Wilson who has shifted to dialect (10.31-32), but shortly (12.04) Wilson returns to correct speech, which is continued in Chapter II and again in Chapter III (18.25 et seq.; 21.19 et. seq.). When he next appears, in Chapter XIII (75.24), Wilson speaks dialect, which is then invariable to the end of the novel. Similarly, after Chapter III Conklin, who had continued correct speech in Chapter III (21.26-28.34), returns to dialect in Chapter IX (55.7 et seq.) until his death (58.5). The youth has spoken ordinary English in Chapters I-III, but beginning with Chapter IX he has patches of dialect, as in his speech to Conklin at 57.9 although normal speech resumes in the next line (57.10). In Chapter XIII the youth incongruously shifts to dialect in his dialogue with Wilson (75.31-76.5;76.18) and so continues to the end of the chapter (79.24-29) but with one interlude of normal speech (76.25-26). In Chapter XIV, at the start the youth's speech reverts to normal but he lapses into dialect again (83.3-6), immediately returns to normal speech (83.9-11), and then back to dialect to end the chapter (83.13; 84.8-9). Beginning with Chapter XV the youth speaks ordinary English consistently.[23]

To summarize: Conklin begins with dialect in Chapter I, but toward the end of the chapter and throughout Chapters II and III his speech is normal. Dialect returns in Chapter IX. On the other hand, Wilson starts off in Chapter I with proper English, changes to dialect, and then back to English, which is then maintained in Chapters II-III, until dialect starts again in Chapter XIII. In contrast, the youth begins by speaking ordinary English in Chapters I-III but inconsistently falls into dialect in some passages in Chapters IX, XIII, and XIV.

Obviously something is wrong, and a reader should be baffled why Conklin and Wilson sometimes speak like the rest of the common soldiers and sometimes like educated men. Similarly, there is no apparent reason why the youth should arbitrarily turn to dialect and then back to proper speech. By any standards this is a flaw in a matter in which consistency is important to establish, but editors without a sense of regularization have ignored the

problem. Fortunately, an editor need not invent dialect for the soldiers' normal speech although by normalization he must translate the youth's uncorrected dialect into ordinary English. Examination of the manuscript used to make a lost typescript from which the first edition was typeset in 1895[24] demonstrates that the non-dialect speech of all three characters is the result of alteration that Crane started as a result of Hamlin Garland's disapproval of dialect-speaking characters. With typical carelessness Crane not only did not consistently alter all the dialect in Chapters I-III, as he intended, but also after he had decided that Conklin and Wilson should retain dialect speech he did not go back to remove the interlineations of normal speech in these chapters. Correspondingly, although he tried more successfully to revise the youth's dialect throughout the manuscript, he inadvertently skipped some passages. As a result, the typescript and then the first edition faithfully reproduced the faults of the manuscript. Crane's intentions are clear, however, and an editor should observe them. (For an account of Garland's reading the manuscript, the alterations, and the transmission to print, see "The Text: History and Analysis," in the University Press of Virginia edition, 1975, especially pp. 199-234.)

Normalization, as suggested in this paper, can be defined as emendation not from documentary evidence directly bearing on the text in question such as is used for regularization but instead from the imposition of an external standard without specific precedent in the text or its associated documents. The most obvious example, which no critical editor could employ, would be the modernization of all accidentals according to the current system. Necessary normalization is applied more often by the critical edition's publisher (one hopes with the knowledge and consent of the editor) in respect to various typographical details and other mechanics of presentation. For instance, heading capitals are usually normalized in the first line of any section or chapter, headings may be set in different fonts from the original and have their periods removed. In fact, all fonts are normalized (preserving of course emphasis italic or words set all in capitals). The fonts of punctuation are normalized, especially punctuation in relation to italic. It is convenient to set all punctuation in italic within italic passages. On the other hand, careless printers (and authors) may italicize punctuation after any italic word(s) immediately followed by roman text. The only proper printing practice is to set such punctuation in roman to agree with the following word. Authorial preference is not involved in these matters, and it is conventional for such alterations to be made silently.

Authors are traditionally careless about or ignorant of conventions applied to ellipsis dots. Some authors may string five or six of these dots when three or four would be correct; silent normalizing to the conventional three or four again involves no authorial preferences. Spacing between the dots may vary in different prints, but it is conventional to separate dots by an enor em- space also placed before and after the sequence. Printers (and authors) are also likely to be careless about a matter that has some relation to meaning.

If occurring in mid-sentence ellipsis to be indicated by dots is normalized to three. If the ellipsis ends a sentence and the next text begins a sentence, the first of the four dots should be placed closeup to correspond to the period that it is. The following three dots then follow, perhaps after a double space as an option. If instead the ellipsis begins in mid-sentence and the resumed text starts with a capitalized new sentence, then the four dots are evenly spaced, the first and last treated in the same manner as three mid-sentence dots, except that the fourth in this case is the period that ends the elided sentence. Whether a dash should be printed closeup both fore and aft or else spaced is a typographical convention that varies without significance. It would seem easiest to normalize in the modern manner to closeup. An editor needs to mark his copy for the printer in these respects at the minimum and to observe the proofs carefully lest he in turn be the victim of housestyling where it is unwanted. This is particularly true for the font of quotation marks and of question and exclamation marks, italic or roman.

Some odds and ends of normalization may trend toward regularization. Correction of misspellings, for instance, could be called either, as could the linking of the digraph ae or oe in classical words. If the copy-text is consistent, the indication in German of umlauted ü by ue, for example, is worth preserving, but otherwise the umlauted form may be the basis for regularization. Idiosyncratic spellings ought to be preserved so long as they are not positive misspellings. Authors like William James who may at some period use various assorted simplified spellings ought not to be normalized; whether regularization is indicated may be determined by the document being edited. However, if tho or tho' are frequent enough to be regularized, normalization to the more correct tho' with an apostrophe should not interfere with authorial preference for simple tho.

Among the odds and ends one must mention the convention of spacing contractions like did n't or didn't. This can be partly an authorial preference and partly compositorial styling. My own view is that it is simplest to regularize or if necessary to normalize to the modern forms. This treatment may seem ruthless and insensitive to historicity, but it will save problems of consistency, such as whether did n't is consistent but doesn't equally so in the same unit. Almost certainly analogous forms will vary in their treatment in a copy-text, authors might space some contractions but not others, and in short it could be a mess. If it is not a mess, then an editor might well preserve this convention. Otherwise, a note in the textual introduction could take care of the matter. Hyphenated words broken at the end of a line in the copy-text need regularization or normalization, depending upon the circumstances, when transcribed, and an editor should be careful to clarify the status of hyphenated words after the printer has set his own edited text.

William James was often sketchy in the forms of his footnotes, abbreviating titles at will, providing cited authors with initials, or full names, or surnames only, not always indicating the date and especially the particular edition used for citations, and at various times incomplete in the notation of pages cited. Normalization here is particularly of service to a reader, the

editor identifying the exact edition quoted from and providing full and accurate bibliographical information, especially since James's memory sometimes betrayed him and his spelling of authors' names was not always exact.

What to do about normalizing quotations has become something of a contentious matter. When a writer like James does not indicate ellip es by dots but prints continuous text, correction by dots would seem to be indicated in fairness to the author quoted. But even that may be denied by critics who believe that quotations should be handled precisely as if they were the author's own text with no editorial intervention save for misprints. As I understand it, the argument goes that changes, whether deliberate or inadvertent, are a part of the authorial complex being presented and that interference with his treatment of quotations is as serious a matter as editorial interference with his own text. In high theory I agree with this proposition but in practice I query it except in certain circumstances. For instance, a writer reading a misquotation from his works is not necessarily pleased at having his punctuation and his syntax altered, if British his spelling americanized, and even his own words sometimes replaced by others without notice so that what he reads in a quotation may not, he feels, fairly represent him to the world—which may indeed know no more of him than is represented by the quotation.

I believe that it is usually possible to distinguish between inadvertent faults in transmission from original author to printed quotation and the quoter's deliberate changes. Except to show that a particular writer was a careless transcriber (or employed amanuenses who were careless) the point of reproducing a distorted version is obscure since the reader cannot know what differences exist except by consulting the record in the apparatus (which may or may not list the accidentals differences). If so, the same information is transmitted if the quotation is brought into conformity with the original and the author's inadvertent changes are, instead, the subject of the apparatus listing. Moreover, since it is as possible for the compositor to alter the accidentals of a quotation by introducing his own styling, what one reads in print is not necessarily what the author transcribed in the printer's copy.[25] This is one side of the coin. As for the other, we may note that James, who quoted extensively, was likely to speed up the quoted matter by small omissions or paraphrases that he did not trouble to indicate; occasionally he felt he could phrase a point he was making by quotation more felicitously than the original, especially in the matter of sharpening the idea he was concerned with. These are legitimate authorial differences and may be viewed as so integral a part of the argument being presented as to be retained in the text, the apparatus of course doing justice to the quoted author by noting what he had in fact written. Even when the sense has not been deliberately altered in this manner James might condense a quotation by a combination of omissions here and there, perhaps only a sentence apart, and changes to link such omissions, as well as paraphrases to speed up or get over the hump in a quotation. For an editor to sprinkle a quotation of this sort with ellipsis dots to indicate every such omission, and not just the major ones

(legitimate emendations) would be to annoy a reader more than to enlighten him. The apparatus is the proper place for such disruptive notation. A critical editor may well try to strike a balance between justice to the writer quoted and to the writer quoting and treat quotations critically in a manner least confusing to the reader and as fair as possible to all parties. The principles involved are not very complex.[26]

One other contentious question of normalization may be mentioned briefly since it has complex reasons for which elaboration here would prove too lengthy. Roughly, it concerns the normalization chiefly of spelling but also of a few punctuational conventions when an American author is published in England and it is advisable to choose the English print as the copytext. Stephen Crane offers a number of examples, but the following are typical. Between July 31 and August 14, 1897, Crane published eight "London Impressions" (collected in Tales, Sketches, and Reports, 1973) in the English Saturday Review and, of course, styled in the English manner. The only American version was the unauthoritative reprint that Cora collected for Last Words from the Review copy in her own typing made from the print. This situation emphasizes the problem whether an American author should be presented to a reader in a dress completely alien to his nationality and one that could not possibly represent the printer's copy. Yet no authority exists in the proper sense of the word for the re-nationalizing of his spelling, at least, and possibly other conventions. Rigorous adherents to the theory of copy-text would reprint the English version as the only authority we know. Others may agree with the editor of Crane that recorded normalization here does no harm except to surface authority, and that an American author should wear his native clothes in the matter of spelling, as indeed Crane did in the unauthoritative American edition.

A more complex case comes when multiple authority is involved. Crane's "Price of the Harness" was simultaneously printed in England in Blackwood's Edinburgh Review and in the United States in the Cosmopolitan Magazine. The copy for both was a professional typescript and its carbon made up from Crane's now lost manuscript. Crane was in Havana and had no opportunity to check the typescript or to read proof. Technically, each print has equal authority, and under these conditions a critical editor may attempt to reconstruct the lost typescript (which is as far back as he can go in the line of transmission) by his selection from one or the other according as the readings reflect his acquaintance with Crane's accidentals as well as substantives, although in the matter of spelling the American Cosmopolitan was obviously the superior authority. On the other hand, a critical assessment of the text indicated that on the whole the English text was more faithful to Crane's substantives and to many of his characteristic accidentals (spelling aside) than its heavily styled American counterpart. Either copy-text, thus, needed relatively frequent emendation from the authority of the other. It was probably the tyranny of the copy-text—in a manner of speaking—that dictated the choice of the Cosmopolitan in the collection printed as Tales of War (1970) simply because it was the American publication. In fact, so much

emendation was required from Blackwood's that the apparatus might have been shorter if the copy-text had been reversed and Blackwood's modified by what seemed more authoritative (including the spelling) in the Cosmopolitan.

When multiple authority is involved, even though not international, as in the newspaper syndication of Crane's stories, the eclectic text that results from a critical analysis of the relative fidelity of the various equal authorities to what is established as Crane's general system of accidentals—with due regard for the nature of the basic copy from which the authorities derive—this pure reconstruction of the lost copy might be called regularization or normalization according to one's point of view. Whatever it is, it may seem to represent some sort of sophisticated blend of the two, and in such special circumstances definition is of no major importance. However, one must recognize the much more common circumstance of multiple authority where both manuscript and its print are preserved, and a combination of regularization and normalization seems to be required to sift the evidence for imperfection and inconsistency in the manuscript accidentals needing attention as set against the attempt to separate the authority of the writer's proofreading alterations from the unauthoritative printinghouse or copyreader's styling as represented in the accidentals of the print. This is to assume the simplest case when the preserved manuscript is the setting copy. Instead, when the manuscript or typescript is collateral or farther back in the direct line of transmission, the case for copy-text and for the treatment of the accidentals by regularization of this copy-text from other authoritative sources becomes more complex and less subject to generalization.

Notes