University of Virginia Library

Reproductions and Scholarship
by G. Thomas Tanselle

George Parker Winship, reporting in 1913 on the activities of the John Carter Brown Library during the preceding year, said, "A purchase which has had an unexpected influence upon the development of the Library is that of a mechanical photographing machine." It may be hard for us now to imagine how the role of photocopying—once a process for it was readily available—could ever have been "unexpected," for the present academic world is a place that Gore Vidal has with justice called "Xerox-land."[1] Many, perhaps most, scholars in the humanities today would say that the widespread access to inexpensive copying facilities has transformed—for the better— their manner of working. Some have even been heard to name the Xerox Corporation along with two or three inspiring teachers or seminal thinkers in their field as the dominant influences on the way their own work has developed. Although few voices have been raised in complaint, the advent of cheap photocopying is not an unmixed blessing. One of the pejorative implications of Vidal's phrase is obviously that photocopying has enabled scholars increasingly to pack away endless bits of information without digesting it. Unquestionably for some scholars the act of placing material in a file gives them a sense of control over their subject matter; and a bulging file of xerographic copies is likely to provide them greater satisfaction than the smaller file of supposedly less reliable hand-written or typewritten notes that could have been produced in the same amount of time (even though the latter would at least have necessitated reading the material). But there have always been scholars who do not know how to make constructive use of their accumulated data and others who clearly do know how: the ubiquity of copying machines does not change that. In any case, the disturbing aspect of Xerox-land that I wish to address is quite different, but equally fundamental: the total lack of understanding of the nature of documentary evidence that is exhibited by most scholars in their use of photocopies.

Even so thoughtful a bibliographical and historical scholar as Winship could say in that 1913 report, "The machine does accurately and

more quickly what the human copyist transcribes with constant likelihood of error."[2] Winship would certainly have understood that photography is not error-free, but he chose not to weaken his enthusiastic endorsement of the new machine by mentioning any of its limitations. Later commentators, speaking both of photocopying a few pages and of microfilming entire books, have usually taken the same line, though we may be less inclined in many instances to believe that any reservations had occurred to them. One outspoken critic, however, was William A. Jackson, who addressed the Bibliographical Society of America in 1941 on "Some Limitations of Microfilm." He began by asserting that there is really nothing to say except that "microfilms are microfilms and not the original book or manuscript, and [I could] then sit down, my case being stated, and allow the implications of that statement to be worked out by each one of you according to your own experience." But, he added, "I am down for fifteen minutes and so I shall employ the rest of them in elaborating this point."[3] He was right that nothing more ought to need saying; but I, too, shall make some further comments, not because I have an allotted number of minutes or pages to fill but because so many people behave as if they do not understand the differences between reproductions and originals.

I shall be speaking specifically of reproductions of documents (manuscript or printed sheets or books) containing verbal texts, but the points I shall make are of course applicable to documents containing musical or dance-notational texts as well, and they are further applicable to reproductions of works of visual art.[4] When I use the word "reproduction" (and, more loosely, "photocopy," "photofacsimile," and so on), I mean the product of any chemical or electrostatic process that aims to represent with exactness (though perhaps on an enlarged or diminished scale) not only the text of a given document but also the details of its presentation, insofar as they can be duplicated on a different surface. Sometimes—in the past more than now—the text of a printed document is reset in the same typeface as the original and with spacing and lineation intended to imitate the original, and the result is called a "type-facsimile";[5] but such a "facsimile" is not a reproduction as here defined because it involves a new typesetting. Most people, I believe, recognize why this kind of "facsimile" cannot be regarded as the equivalent of the original, whereas they frequently do not see why the same is true of photographic and xerographic copies. I should like to explore both the practical and the theoretical reasons for maintaining that reproductions are not substitutes for originals and then examine a few prominent recent instances of confusion on this point.

I

The bulk of the writing about microfilms and other photocopies of manuscript and printed material (and there is a considerable literature of the subject)[6] does not allude to the possibility of inaccurate or misleading reproduction. Indeed, what is often stressed is the absolute fidelity of photographic copies. For example, Charles F. McCombs, in his widely circulated pamphlet The Photostat in Reference Work (first published in 1920),[7] speaks of the "unlimited possibilities" of photostats because of "the advantages of an absolutely accurate copy" (p. 4), a copy "free from errors and omissions to which the most careful copyist or typist is at times liable" (p. 5). The next year the Bibliographical Society of America devoted a whole issue of its Papers to the subject of "Photographic Copying" (the title on the cover label, though the primary topic was the photostat), and in the course of 53 pages[8] there are only two brief references to any limitations of photocopies: Henrietta C. Bartlett is quoted as saying that the photostat "should not be considered of equal value to the student with the original" (p. 15), and in the discussion of the acceptance of photostats in court "Mr. Winship and others spoke of the possibility of altering or faking photostat copies" (p. 50). Otherwise it is all praise, far less perceptive bibliographically than one would expect in the publication of a bibliographical society. Thus George Watson Cole (Huntington Library), in his survey of "valuable and interesting . . . information" from the responses to a questionnaire, quotes Frederick W. Cook (Archives Division, Commonwealth of Massachusetts) as saying that "the machine makes no mistakes" (p. 5); Alexander J. Wall (New-York Historical Society) writes in his reply that "photostat reproductions answer every purpose for the student of American history and we bind them the same as any pamphlet and treat them as an original publication in their classification" (p. 10); James Thayer Gerould (Princeton University Library) says, "In many cases the photograph is more satisfactory for the scholar's use than the original" (p. 11); and Cole, in his concluding remarks, states as a fact that a photostat "requires no reading back" to check it for accuracy (p. 16).

Such blind faith in the virtues of photography, on the part of those who should know better, has continued. Keyes D. Metcalf reported uncritically in 1938, "It is generally conceded that it [microphotography] will be used extensively by librarians in the place of the original of books and MSS." He added, even more astonishingly, that in photographs "books and MSS. which could not otherwise be found in libraries will become available and may be studied from a bibliographical point of view"—and this statement (like the ones cited by Cole earlier) appeared in

the journal of the Bibliographical Society of America.[9] A quarter-century later the same journal published a piece by Richard W. Hale, Jr., who took as his starting point "the simple thought that a photographic reproduction of bibliographical information is textually trustworthy."[10] In between, Library Trends published an article by Herman H. Fussler implying that microfilm eliminates the necessity of seeing the original: it "permits the historian in Athens, Georgia, for example, to have access, without leaving Athens, to an important collection of manuscripts in the Bancroft Library at the University of California. The cost is clearly less than that of going to California to consult the originals (unless the number of manuscripts required is very large)."[11] And James G. Hodgson, in an early pamphlet on xerography (which, its subtitle predicts, "may be a tool of great importance to libraries"), welcomed the process by asserting that "the final form is a very exact reproduction of the original."[12]

There is no need to multiply examples, for everyone knows that such statements abound—though it is disturbing to find that at least one bibliographical society and several rare-book librarians (on whom one might expect to rely for carefully considered views on this subject) have endorsed the common belief that photocopies do not lie. The reasons for being suspicious of reproductions have not gone entirely unexpressed, however; and one should not be surprised that the essential points were made at a meeting of the London Bibliographical Society as early as the mid-1920s. The March 1926 number of the Library, which prints the papers delivered at the December 1925 meeting and a summary of the ensuing discussion,[13] contains two statements that together make the primary case against reproductions and suggest how copies may appropriately be used. W. W. Greg, in his paper, said simply that "no process but in some measure obscures what it reproduces" (p. 321). And the president of the Society, Frederick George Kenyon, said in the discussion that photographic copies "should be regarded not as substitutes for the originals, but as approximations only helpful in suggesting points which must subsequently be verified" (p. 327). A decade later the Bibliographical Society published, as the tenth of the Supplements to its journal, R. B. Haselden's Scientific Aids for the Study of Manuscripts (1935), which was considered the standard treatment of its subject in the years that followed. Because the book is largely concerned with the examination of the physical characteristics of manuscripts, it naturally begins from the premise that one is studying the originals; but Haselden does say explicitly, in his chapter on photography, "No authoritative conclusion regarding a manuscript can be based on the examination of any known form of reproduction thereof; the original must always be consulted" (p. 70).

Six years later came W. A. Jackson's forceful paper, mentioned above, asserting flatly that "no photograph can take the place of the original" (p. 287). In the next decade Jerry McDonald's "The Case against Microfilming,"[14] although it deals with the filming of business firms' archives (and partly from the economic point of view),[15] makes many of the points that one would have to make in evaluating the reliability of film for scholarly research: McDonald comments on the necessity for "inspecting" microfilms (by which he implies collation and spot-checking for legibility, if not full proofreading), the loss of detail in them, the drawbacks of not seeing colors or embossings, the errors produced by fluctuations in the electric current supplied to microfilming machines, and the distortions produced by variations in the temperature of the places where microfilms are stored. In 1961 Laurence A. Cummings, writing on "Pitfalls of Photocopy Research,"[16] added to this catalogue of problems: photographers omit material through oversight or the assumption that it could not be significant enough for the customer to wish to pay for;[17] pictures may not be in sharp focus if the leaf photographed does not lie flat or if adjustments are not made when the lens-image distance shifts because of the thickness of the book being photographed; and erasures, show-through from the reverse side of a leaf, uneven surfaces, and spots on the paper or on the lens all can cause photographs to be misleading.[18] Cummings concludes, "any serious editorial work based on examination of microfilms, photostats, and other reproductions without first-hand consulting of the original must be tentative. The camera lens cannot replace the scholar's eye." The word "editorial" could be omitted from this statement, of course, because the problems enumerated would affect any serious work, whether textual or not. The same point could be made about a 1968 comment of Franklin B. Williams, Jr.: "no one questions the principle that an editor must work finally with the originals."[19] For "an editor" one could substitute "all serious readers"—though the idea that "no one" questions such a statement, or even the more limited original one, is far from being literally true (the phrase really means "no one who has thought the matter through with logical rigor"). Williams has provided the best historical survey of the production of facsimiles of pre-1641 English books,[20] accompanying it with what he calls a "cautionary" listing of photofacsimiles and indicating many instances of faulty reproductions, ranging from those based on defective copies to those that mistake type-facsimiles for originals.

Most of the criticisms of reproductions, including some of those just cited, have focused on specific examples, and it is worth mentioning several of them here to suggest the kinds of problems that actually occur. One large class of problems consists of errors arising in the planning or

production of photocopies. Sometimes such mistakes are inadvertent, but at other times they result from misguided attempts to be helpful by tampering with the images. Instances of the former are the omission of the dedication leaf from the 1931 Country Life facsimile of Tusser's Good Husbandry (reported by Williams) and that of the last page of text from the Garland facsimile of William Biggs's Narrative (reported by David Vander Meulen). Besides omitting material, reproductions can also present material in incorrect order: Glenn H. Blayney reports that the Huntington Library microfilm of A Yorkshire Tragedy presents the facing pages B3v/B4r where A3v/A4r ought to be (and vice versa)—because it was made from photostats, not directly from the original, and two of the photostats got switched. Williams calls attention to the use of Smeeton's type-facsimile of Kyd's Solimon and Perseda and Ashbee's hand-traced facsimile of Jack Jugeler, rather than the originals, as the bases for Farmer's collotype facsimiles of these titles; and he notes that Farmer's collotype of Rastell's Of Gentylnes and Nobylyte reproduces the British Library copy without indicating that four leaves in that copy are in type-facsimile, just as the Scholars' Facsimiles & Reprints series includes, without notice, a pen-facsimile title page in its reproduction of Palladis Tamia. The 1977 Arno Press facsimile of Donne's Biathanatos represents a compounding of errors (discovered by Ernest W. Sullivan II): it reproduces not the original but Hebel's 1930 facsimile (for the Facsimile Text Society), which in turn incorporated opaquing (to eliminate show-through) that wiped out more than a hundred punctuation marks (some of which were replaced by pen).[21]

The intentional creation of errors by opaquing and retouching is widespread. Of course, those responsible do not think of themselves as creating errors; but if a facsimile purports to represent an original as closely as possible, any alteration that removes distracting marks present in the original (as opposed to those introduced by the camera lens or the surface of the copying machine) produces by definition an error—to say nothing of the inadvertent erasures that often accompany any cleaning up. One of the most famous examples of what opaquing is likely to lead to is the 1954 Yale facsimile of the First Folio of Shakespeare, which contains an errata list in its second printing and some restored readings in its third and which, as Fredson Bowers reported in his thorough review, contains an enormous number of additional errors, especially erasures of line-end punctuation. Colin McKelvie's 1976 facsimile of the Armagh Public Library copy of Gulliver's Travels (for Scholars' Facsimiles & Reprints) contains, as Donald Greene has shown, alterations that go beyond erasures in the reproductions of pages with manuscript revisions in what may be Swift's hand: on one page of the original, "Red"

and "Green" are entered side by side in the margin, whereas in the facsimile "Red" appears below "Green"; on another page the marginal words "Horn and Crown" are much reduced in size from the original; and on a third a large blot in the original is replaced by a seemingly hand-lettered "and" in the facsimile. The first two changes were made because the leaf size of the facsimile would not otherwise have accommodated the annotation (and there are instances of marginal notes that do not show up at all); the third was made on the assumption that the blot was accidental, though it was arguably Swift's way of deleting the "and". Greene concludes, "It is clear that with a little judicious airbrushing and a sharp pen, and photographic enlargement and reduction, a 'facsimile' can be made to say the thing that is not." That facsimiles, like other printed books, can vary from one printing to another is illustrated not only by the Yale Folio but by the Scolar Press reproduction of Pope's 1729 Dunciad Variorum: in the first printing of this facsimile (1966) two superscript footnote letters are properly present on pages 14-15, but in the second printing (1968) a thorough cleaning of marginal spots has taken away these letters. David Vander Meulen has found still other instances of tampering in this facsimile: the errata leaf is placed in its conventional position at the end of the book, although it occurs in a different location in the British Library copy, which the facsimile claims to reproduce; three blank leaves are inserted without comment, apparently to mark divisions, where the original has none; and the pages bearing ornaments or plates are slightly enlarged, suggesting that they may have been produced by a different process.[22]

If all such intentional and inadvertent departures from the originals constitute one category of problems that reproductions present, the other large category consists of those instances in which reproductions that cannot be called inaccurate are nevertheless misleading. There are many places in manuscripts, for example, that cannot be properly interpreted unless the various inks or pencils involved can be distinguished and unless the marks that have bled through a sheet from the other side can be recognized for what they are. Yet most reproductions do not offer a broad enough range of gradation in tone to make such discriminations possible; and frequently the photographic adjustments necessary to make the faintest inscriptions show up cause distortion in the heavier inscriptions. Almost everyone who has ever used reproductions of manuscripts is likely to have misinterpreted them at one point or another, either by thinking something is illegible when it is not or by investing something with an unwarranted significance. We should not be surprised when Jerry McDonald says that "the chief engineer's 'hen scratchings' are much easier to decipher in the original than they are on film" (p. 348),

or when Laurence Cummings reports that what he thought were Sir Walter Ralegh's initials in a Renaissance manuscript examined on film turned out to be "a few random pen scratches and an interesting pattern of wrinkles" (p. 101).

A prominent instance of scholarly work vitiated (among other reasons) by reliance on photocopies that exhibit such problems is Hans Walter Gabler's edition of Joyce's Ulysses. John Kidd has demonstrated that time after time the errors in Gabler's text can be explained by comparing the published facsimiles in The James Joyce Archive (1978-79) with the originals. Gabler prints "Captain Culler" at a point where Joyce's inscription on a surviving proof sheet clearly reads "Captain Buller", presumably because the high-contrast reproduction in The James Joyce Archive makes a printer's pencil mark at this spot look something like a "C". At another point on a surviving proof, Joyce wrote "crême de la crême" in ink, and at a later stage the circumflexes were changed in blue pencil to grave accents; Gabler does not report the circumflexes, which in the facsimile are hidden by the heavy pencil marks. Kidd shows the considerable significance of these particular misreadings for interpretation, but even those of less importance demonstrate the point that facsimiles can be misleading. It is crucial for studying the development of Joyce's text to distinguish two types of black-lead pencil as well as the colors of both ink and lead markings; but the facsimiles do not permit such identification, and it is not provided by Gabler—though he does try to record erasures and inevitably misses many and invents others. Kidd concludes that the Gabler edition "is a study not of Joyce's manuscripts but of inadequate facsimiles."[23]

Reproductions of printed material can be equally misleading. Sometimes light inking or worn spots on a printed page cause type impressions not to show up at all in reproductions, and such reproductions can then give rise to the belief that variants exist where in fact there are none. Williams mentions that three facsimiles of the 1597 Richard III show no paragraph sign beginning the imprint because they used the Devonshire (now Huntington) copy, in which the title page is worn and the paragraph sign faint. Sometimes type impressions, like handwriting, are unclear even in the original, and a facsimile interposes yet another barrier in the way of interpreting such ambiguous spots. A case in point is William Biggs's Narrative (cited in another connection above): the title-page date is unclear (1825 or 1826), particularly so in the Newberry Library copy used for the Garland facsimile, in which the volume is identified as representing an 1825 edition. Uncritical users of this facsimile will assume that the 1825 date has been definitely established and that the indistinct date on the title page probably reflects a flaw in the photography,

whereas in truth this book is regularly catalogued (following Cecil K. Byrd's A Bibliography of Illinois Imprints, 1814-58 [1966]) as from 1826. An even more common situation is that reproductions of printed matter are misleading because the copy chosen is defective or contains variants that make it a poor representative of an edition as a whole. According to Ernest Sullivan, the Hebel facsimile of Biathanatos is based on "a probably unique and textually suspect copy"; and Franklin Williams discovered that the title-page date is missing from the Scholars' Facsimiles & Reprints reproduction of The Beggars Ape because it was made from the British Library copy, which is cropped. Rollo Silver has reported that the Readex Microprint reproduction of the first English Bible printed in America was apparently made from an imperfect copy that does not contain the Congressional resolutions authorizing the printing. And Charlyne Dodge describes a photocopy of Harold Frederic's "Preface to a Uniform Edition" in In the Sixties (1897) that fails to show Frederic's initials at the end because it was made from a copy that had been printed on a torn sheet: the flap created by the tear was turned back during printing so that one of the initials fell on it and one struck the hole; when the flap was later put in place, no initials therefore appeared on the recto (and one askew initial was on the verso). Without seeing the original one would have no way of knowing that this photocopy of the "Preface" does not represent a variant state or impression. This printing error is unusual, but the lesson it teaches applies to every use of a reproduction.[24]

II

These examples show a few of the ways in which erroneous and misleading reproductions occur. Clearly it is difficult to imagine a situation so far-fetched that it cannot be seriously considered as the cause for one or another problematic photocopy. Accidents do happen, and reproductions do mislead. Everyone knows that; and everyone knows (though many people act as if they do not know) that every form of reproduction can lie, by providing a range of possibilities for interpretation that is different from the one offered by the original. What is less well understood is that even if the production of copies were always accurately handled and even if the reproductions themselves were never distorted or misleading in their representation of the originals, they would still be unsatisfactory. The reciting of examples is actually irrelevant, because even if no one had ever found any problem in any previous reproduction, there would still be no reason to trust reproductions or to let them serve as substitutes for originals. The essential fact one must come back to is that every reproduction is a new document, with characteristics of

its own, and no artifact can be a substitute for another artifact. This point is more widely recognized in some fields than in others; as far as verbal compositions are concerned, many people think that the words can easily be transferred from one physical surface to another with no loss, because they do not understand the role of physical evidence in interpreting what communications in fact consist of. Any reproduction, whether clear or indistinct, must be suspect simply because it is not the ultimate source: documentary texts, like all other artifacts, must be examined first-hand if one is serious about approaching them as historical evidence.

Even those persons who have shown themselves in print to be critical of photocopies have generally not alluded to this underlying reason for the inadequacy of all reproductions; instead they have often suggested that copies could indeed replace originals if only the technology of copying could be improved so as to eliminate its present defects (though how the possibility of error can ever be eradicated is hard to see). Frank Weitenkampf, for example, discussing facsimiles of all kinds, says that despite the advent of photographic processes "there is still some need for caution in accepting results, particularly those of the earlier years of these processes."[25] The last nine words suggest that the growth of technology can make reproductions respectable. Weitenkampf's next sentence reinforces the point: "Even some of the later ones may not quite come up to the mark"—in other words, a mark does exist (even if it has not been reached) at which copies can be substitutes for originals. At the end he asserts, "We cannot carry on certain studies with copies that are not 'exact'" (p. 130). He places "exact" in quotation marks, recognizing that no copy reproduces every feature of the original, and yet he seems to believe that copies can reach a level of exactness adequate for "certain studies"—those studies apparently being the more demanding ones, whereas less exact copies are seemingly satisfactory for other studies.

The same misunderstanding of the nature of documentary evidence is shown by another class of critics of reproductions: those who complain that reproductions, particularly microfilms and microfiche, are difficult to use. Indeed, most people who raise objections to photocopies are thinking not of the accuracy of the reproduction but of their own discomfort in sitting in front of a microfilm or microfiche reader. The triviality of this point of view would make it scarcely worth noting if it were not for the prominence of many of the people who have expressed such sentiments in recent years. Their comments suggest that there is nothing wrong with microfacsimiles except the awkwardness of using them; but that is enough (since research should presumably be comfortable)

to provide grounds for protest. The shallowness of this general line is well represented by a 1958 statement of Lawrence S. Thompson's: There is no hint here of the serious reasons for preferring original books to photocopies or of why bibliophily is important. An almost equally superficial view of the relation between book collecting and photocopying had much earlier been expressed by Willard Austen, librarian at Cornell: Such a comment trivializes both historical research and book collecting, showing no recognition of how the "substance" of a verbal artifact is connected to its physical form. Unfortunately the problems inherent in this statement would not be noticed even today by many people.

Over the years discussions of photocopying have intermittently addressed the question of the admissability of photocopies in courts of law, and in the process they have exhibited yet again a failure to focus on basic issues. In 1921 both John Clement Fitzpatrick (of the Library of Congress Manuscript Division) and John S. Greene (of the Photostat Corporation) stated that photostats are accepted as evidence in court, whereas photographs are not, because (in Fitzpatrick's words) "nothing intervenes between the original and the photostat print, which is not the case with a photograph, which has the developed plate between the original and the finished reproduction" and therefore (in Greene's words) "it is an easy matter to rearrange a photograph."[28] Greene was so rash as to say that "the photostat print cannot be changed," but there are now millions of users of xerographic copying machines who would understand why that statement is not the whole truth.[29] Jerry McDonald, in the course of his critical discussion of microfilm, says, "The legality of microfilm is pretty well established," although judges can decide in individual cases what is acceptable: "If you encounter one who has had a bad experience with film, he may reject it as primary evidence." All one can say is that any judges who do accept reproductions are lowering the

evidentiary standards of their courts; certainly no reproduction can offer "primary evidence" unless the reproduction itself is the document at issue. Nor does certification help: Edward Tweedell noted in 1921 that the Crerar Library had provided many photocopies in patent cases, "for which certified copies are furnished"—as if certification were what gave them legal standing.[30] But certification only means that the person doing the certifying believes the copy to be faithful; if the original exists, second-hand testimony is unnecessary. That courts sometimes accept photographic and xerographic copies should affect no one's thinking about the appropriateness of reproductions as evidence; the definition of primary evidence ought to be the same in scholarly research and in courts of law, but neither scholars nor judges are immune from lapses in logic. All the comments I have quoted here about the legal uses of photocopies miss the essential point: that photocopies are different documents from the originals. It is ironic that legal procedure, which is founded on a critical approach to evidence and which does not tolerate a confusion of hearsay and primary evidence where human witnesses are concerned, can sometimes fail to recognize the same confusion when the witnesses are documents.[31] That this confusion does occur in court is a measure of how deeply seated is the belief in the transferability of documentary texts from one document to another—a belief that has probably intensified with the increase in the use of copying machines, in spite of the undeniably wide recognition of the alteration of documentary texts that takes place in copying centers everywhere.[32]

Although different methods of reproduction may offer different opportunities for error and for intentional alteration, all are alike in producing new documents that in one degree or another are not identical with the documents supposedly being copied. And whether those documents contain handwritten or printed texts is of course irrelevant. I have therefore made no distinctions here among the various copying processes or between manuscript and printed materials. There is one way, however, in which printed materials do pose a special problem. Nearly all such materials were originally printed in editions of more than a single copy; and, if more than one copy of an edition survives today, anyone wishing to make a reproduction of that edition must decide which copy to use. Copies of an edition (that is, copies printed from the same type-setting) cannot be assumed to be identical to each other for the same reason that reproductions cannot be assumed to be identical to the originals: they are separate physical objects, separate documents. Indeed, as analytical bibliographers have been showing for a century or so, copies of printed editions from the fifteenth, sixteenth, and seventeenth centuries can be expected to vary as a result of stop-press alterations made

during the course of printing; and copies of later editions often vary because of the excision and substitution of leaves or entire gatherings and because of changes made between press runs. A reproduction of any one copy of a printed item represents only that copy, not the edition as a whole; yet both producers and users of photofacsimiles of printed material often assume that the facsimile can stand for the edition. This problem, to be sure, is not limited to reproductions, for the same people would obviously not realize that in using an original they would need to collate it against other copies.

Whenever one needs to talk about an entire edition—as in a scholarly critical edition, or a descriptive bibliography, or a critical essay—one must be aware of the differences among copies of the edition.[33] If a published facsimile is to be of service to scholarship, it ought to be accompanied by a record of the ways in which the copy photographed varies in text from other copies. When a reproduction of printed material does not include such information, one therefore has further reason for being cautious, in addition to the reasons that apply to all reproductions. It is in fact not necessary for a photofacsimile of printed material to be limited to the pages of a single copy of an edition, unless the aim is to reproduce a particular copy. Fredson Bowers, who has given this question its most thoughtful treatment, believes that "the ideal photographic facsimile—containing the necessary apparatus—should consist of a collection of formes from any number of copies, these formes being chosen first according to the principle of their textual state, and second according to clarity and fidelity of the inking."[34] The most prominent use of this approach thus far is Charlton Hinman's The Norton Facsimile: The First Folio of Shakespeare (1968), in which the corrected state of each forme is represented. Hinman's apparatus identifies the copies used, as well as the "substantive and semisubstantive" variant readings of the uncorrected states. Such a reproduction does in a sense represent the edition as a whole because it draws on evidence present in a large number of the surviving copies and rises above the idiosyncrasies of individual copies, with their fortuitous assemblages of sheets. It is a product of scholarship and serves a purpose that no unannotated facsimile of a single copy, and indeed no single copy in the original, could serve. As a collection of photographs, it cannot escape the problems inherent in all reproductions; but, given that limitation, it shows how the difficulties presented by printed editions can be responsibly accommodated in a facsimile.

Reproductions do have their uses, as long as one understands why they must always be approached with caution and why they can never be thought to obviate examination of originals.[35] There is an enormous

difference between the attitude underlying the careful production of facsimile editions, involving thorough proofreading and the writing of notes commenting on potentially misleading spots,[36] and the frame of mind exhibited by persons who order photocopies and proceed to use them as the equivalent of the originals, never making any direct comparisons. Some of the large commercial projects for supplying extensive series of microfilm, microfiche, or xerographic copies are unfortunately much nearer to the latter than the former. If one Los Angeles company, after examining the results of its order for the microfilming of 2,300,000 documents, had to require the refilming of 35,000,[37] one can imagine how often the microfilms that scholars use may be defective, for they are not always subjected even to the kind of inspection (which presumably did not include proofreading) that this company undertook. Every reproduction of a written or printed text, whether prepared by a publisher for public distribution or by an individual for personal use, should be proofread against the original just as carefully as if it were a newly typeset text, and notes should be made describing the original at those places where the reproduction misrepresents it enough to cause a reader possibly to misinterpret the text.[38] Even a reproduction of a manuscript that is to be used, or published, with an accompanying transcription should be supplemented with such notes, which in effect explain how some of the words or punctuation marks in the transcription were arrived at.

Handled in this way, reproductions can serve as a useful stopgap, until one can return to the originals for a final check. They can thus be a true convenience, whereas without these precautions their helpfulness is illusory. There is no way that reproductions—regardless of what technology is developed in the future[39]—can ever be the equal of originals as documentary evidence, for there is no way of getting around the fact that they are one step (at least) removed from those originals. And there is no way that the existence of reproductions, however high their quality, can justify the destruction of originals. No one seems to have trouble understanding why a reproduction of a vase cannot replace the original for any serious study; but many people apparently fail to see that a paper with written or printed words on it is also an artifact, containing an unreproducible assemblage of clues to its own genesis.[40] Originals are clearly necessary for the study of the distribution or publishing of verbal works; but they are also essential for the study of the texts contained in those distributed objects, whether manuscripts or printed books. The words that come to us from the past, transmitted by paper and ink, cannot be assumed to reflect accurately what their authors intended; in order to assess how the words that are present in documents came to be

there, and indeed to try to make sure that we know what words are in fact there, we must avail ourselves of all the evidence that comes with them. The study of the past requires artifacts from the past; reproductions are the products of a different time from that of the originals they attempt to duplicate, and they therefore transport us to a different time.

III

That these considerations have not always been understood by those to whom we ought to be able to turn for guidance in such matters has been repeatedly demonstrated in recent years. For example, in A Guide to Documentary Editing (1987), written by Mary-Jo Kline for the Association for Documentary Editing, the principal discussion of the use of originals consists of the following: "Whenever possible, transcriptions should be perfected against the originals of their source texts, not merely against photocopied versions. When this is not feasible, the edition's introduction should make this omission clear" (p. 178).[41] The second sentence is certainly true: readers of an edition should always be informed when a transcription has not been read against the original. But the first sentence, prescribing a reading against the original "Whenever possible," fails to convey a sense of the importance of the procedure, suggesting only that it is desirable, not that it is essential. A fuller statement on this matter is a conspicuous lack in a book that places considerable emphasis on the use of photocopies.[42] Near the beginning we are told, "Modern scholarly editing was made a practical possibility by technological advances in one area—photoduplication" (p. 23); and the book treats in some detail the collecting and cataloguing of photocopies in the editorial office. The equivalence of originals and copies is implied by such statements as this: "The manuscript or a reliable photocopy is to be preferred over any later scribal copies or transcriptions as the source text" (p. 82). In the section on "Microform Supplements" (i.e., to letterpress editions), the choice between film and fiche is addressed (pp. 70-71), but nothing is said about the proofreading that such facsimile publications require. There is even the assertion that the "fathers of expanded transcription," Julian Boyd and Lyman Butterfield, did not record in letterpress editions certain kinds of details from manuscript texts because they assumed "that microform editions of their projects' archives would make facsimiles of these source texts available to a wide audience" (p. 128). Perhaps they did; but surely some further comment is called for, in an introductory guide of this sort, explaining not only the limitations of microfilm but also the contribution made by a full record in print. The term "source text," rather than "source document," is repeatedly used, implying that the text is easily extractable from the

artifact that preserved it. In the context of this book, the absence of a detailed warning about the problems presented by reproductions is positively misleading.

The treatment of reproductions in this Guide should be contrasted with that in an earlier comparable guide, the 1967 Statement of Editorial Principles of the Center for Editions of American Authors. This Statement insists at the outset that "if the editor is using photocopies of manuscripts, he must read his working copy against the originals to be certain that he has not missed changes or additions or cancellations that do not show up in photocopy" (p. 2). A few pages later a more detailed and forceful directive is issued:

Both these statements are repeated in the revised edition of 1972 (entitled Statement of Editorial Principles and Procedures), with the first one expanded to include a warning against reading "as punctuation in the photocopies what are actually smudges, specks, or holes in the manuscript" (p. 1). The Committee on Scholarly Editions, which succeeded the CEAA, endorses the same procedures and issued a set of "Guiding Questions" for testing editions, including the question (quoted here from the April 1977 version), "Where copy-text is manuscript, how have the transcriptions or copies of manuscript been verified against the original?" The CEAA/CSE requirement of checking transcriptions against originals, which had thus been in effect for twenty years at the time A Guide to Documentary Editing was published, is of course the out-growth of a longer scholarly tradition. Against this background the statements about photocopies in the Guide appear particularly weak and disappointing, and the Guide in this respect takes a large step backward.

Another egregious recent instance of misunderstanding the nature of reproductions is provided by a policy of the Harry Ransom Humanities Research Center of the University of Texas. Decherd Turner, a few months after he became its director, proclaimed his views in "An HRC Decalogue," published in the personal newsletter he had established (HRC Notes, No. 3, Thanksgiving 1980). The sixth item in his decalogue reads in full as follows:

The patent absurdity of this statement is compounded when one reads the second item of the decalogue: When one puts these two statements together, the incredible incoherence of the position becomes apparent. If it would be a waste of money to buy manuscript materials of which "copies exist elsewhere," then the copies must be fully the equal of the originals. In that case, why spend money on the conservation of the original documents already on hand, when inexpensive photocopies of them could be made? Indeed, why should good money be spent on originals at all, even virgin documents that have never been violated by camera or copying machine? Let other libraries, foolish enough not to object to materials that have been reproduced, buy the originals; Texas could then for comparatively little money build up a magnificent collection of photocopies.[43]

It is difficult to believe that, once Turner's decalogue was in print,

there was not enough forceful opposition to this stand on reproduction to result in an alteration of the policy. But the stand continued to be enunciated. In the ninth newsletter, dated 31 March 1984, one section is entitled "The Copy Machine." After asserting, "Perhaps no development like the Xerox copy has so identified the distinctions between the needs of the librarian and those of the private collector," Turner makes his principal point: Turner here places "the librarian" in the unenviable position of believing that the existence of a Xerox copy of a document—or a published text of it!—drains the original of scholarly value. Yet a few sentences earlier he had said, "The object of a librarian's dollar is to gather unique materials into one place for purposes of research." To insure that they are unique, one must insist on certification that they have not been copied: "Without certification, the librarian is in serious jeopardy of spending resources for materials which are not unique, thus calling into question his judgment." But if copies are as good as originals, what is the point of assembling "unique materials" instead of encouraging their multiplication?[44] And if anyone is ever allowed to publish the texts of the "unique materials," was the "librarian's dollar" well spent after all? Does not this approach to a research collection lead to a situation in which materials are gathered to be hidden from view rather than made available for scholarly dissemination?[45] Turner's position is by no means representative of that of other special-collections librarians; indeed, it is so extreme that it is perhaps not taken seriously by any of them. Yet the belief from which it springs—that copies can take the place of originals— is widespread, and the implications of it are frightening, as is the fact that even one director of a major library can hold it.

From the point of view of the number of people involved, the most significant recent instances of misunderstanding the relation between copies and originals have occurred in connection with the book-preservation movement. The immense problems posed by the deterioration of books printed on acidic paper in the nineteenth and twentieth centuries are finally being addressed seriously by many institutions and individuals, and we must be grateful for that. But the public statements by the

persons active in this effort nearly always fail to recognize the limitations of photocopies as documentary evidence. The debates over the usage of the two terms conservation and preservation at least call attention to two distinct kinds of activity: operations intended to extend the life of the original physical book and operations intended to transfer the text to a new physical object. In the library world the treatment of originals is increasingly coming to be called conservation, with preservation used as the broader term that encompasses the transfer of texts. One person who has been vocal in making this distinction is Pamela W. Darling, but in the process she illustrates a common confusion. "In the museum world," she says, "where every item is unique, conservation is—quite properly in my view—the dominant term since physical care is virtually the only option. Microfilming the paintings or recording the appearance of woven baskets on an optical disc and discarding the originals would hardly do!"[46] The implication is that books are different: that, in contrast to all other human products, they are not unique and that, once their texts have been copied, nothing is lost by discarding them. Verbal works are indeed different from paintings in that they do not exist on paper in the way paintings exist on canvas, for language is an abstract medium; but those who believe that verbal works can simply be copied on film are failing to recognize the contribution physical evidence makes toward assessing the reliability of a given text as a representation of a particular verbal work. In addition, of course, every book is a piece of evidence for the study of publishing history.[47]

Preservation is much in the news, and librarians have frequently been interviewed on the subject, very often with unfortunate results. For example, the Yale Alumni Magazine for Summer 1987 reported on the preservation program in the Yale libraries in these terms: "The decisions on what to save and what to reformat aren't made lightly. As an overall rule, books valued for their information are given a new format. Books and documents valued as objects are conserved."[48] This distinction is in fact nonexistent: all books are potentially valuable for their "information" (their texts), and all are worth saving as artifacts, as evidences of past human activity directed to the transmission of texts. What this statement is in practice likely to mean is that books of high market value will receive expensive conservation treatment, and other books will be microfilmed or photocopied and then thrown out. Such a policy is not worthy of a research library. A few months after this article on Yale, a similar piece about Columbia appeared in the Columbia alumni magazine, with the same false distinction between "intrinsic value" and "content": "For the most part, Columbia conserves only books that have some intrinsic value, such as those with marginal notes, elaborate bindings, or

excellent illustrations that would not microfilm well. Works valuable only for their content and for which replacement copies are not available are usually given a new format, such as microfilm, microfiche, or photocopy."[49] Microfilm is several times referred to in such terms as "the only proven medium" (p. 16) or "a technology that we know works" (p. 18); what should have been said instead is that we know it does not work. Librarians admit, according to the article, that "few scholars like to use microfilm" (p. 18); but the basis of the complaint is the expected one, that microfilms are not as easy to use as books, and nothing is said about the status of reproductions as secondary, not primary, evidence.

To raise these issues is not to object to the microfilming of endangered books: there is no question that vast quantities of books are crumbling apart and that having texts on microfilm is better than not having them at all. But the widespread misunderstanding of documentary evidence leads to the unnecessary destruction of books in the name of textual "preservation." An article in the New York Times describes the usual procedure:

The "special cases" referred to are by definition uncommon, and the general rule in preservation-microfilming operations is to discard what is left of the books after microfilming. An article in the Washington Post describes the fate of a 1909 book by the American explorer Fanny Bullock Workman after a reader at the Library of Congress called for it and thus brought its condition to the attention of the library staff: "Once the filming was complete, the physical remains of the book were taken to the Exchange and Gifts division where they were boxed up with other library waste paper and shipped to a pulping company in Baltimore to be turned into pulp."[51] Although this article takes more seriously than it should the complaints about the inconvenience of microfilm, it is unusual in being generally critical of the aftermath of microfilming and puts the problem concisely when it says, "Spines are still being split and pages pulped as books disappear into information."[52]

The determination of which books to microfilm (or to microfilm

first) and then to destroy is a central issue in preservation-microfilming circles, and the thinking about it further illustrates the confusion between documents and "information" (or "statements," or "works"). The mentality that does not distinguish copies from originals is likely also to treat different editions of a work as unnecessary duplications of information and is ready for the next step of believing that some works, having apparently been superseded by others, need no longer exist. Deciding to microfilm great quantities of material in a given field during a particular period "saves time"—according to the article on Columbia quoted just above—"but risks wasting resources on books that will never be used" (p. 19). What will be used, of course, can never be predicted. But this way of thinking has led to the formation of groups of scholars to select the most important titles in their fields for microfilming. In 1984 the American Philological Association began a project to place on microfiche the texts of the most important classical studies published between 1850 and 1918, and a 1987 report of this undertaking[53] states near the beginning that we must not "waste our resources on materials that are unimportant." Commenting on the alternative "vacuum cleaner approach" ("preserving on a wholesale basis everything from a particular range of dates or place of publication"), the authors recognize the argument that "there might someday be a use even for materials whose importance is not evident at present"; but they proceed to say in the next sentence, rather incoherently, that the major weakness of this approach is "that materials which may never be needed by scholars take up time and money and thus displace more important materials that aren't in the chosen group" (p. 141). If titles must be selected, asking a group of scholars for advice is appropriate; but the committee should not fool itself by thinking that some works lack usefulness, for there is no product of the past that is not useful in studying the past.

The APA report is noteworthy for showing that even a group of specialists in an important area, giving protracted thought to the details of a large microfiche project, can neglect entirely the question of the status of reproductions as documentary evidence. Although, the authors say, the "involvement of scholars in preservation decision making has sharpened our sense of some of the key issues in preservation, both philosophical and pragmatic" (p. 144), the philosophical considerations do not include this most basic one. When the authors report that "most preservation programs concentrate on preserving the contents of brittle materials with little artifactual value" (p. 140), there is no criticism of the last phrase; and although the project policy allows for the retention of some books after filming, it "aims at minimizing the retention of badly damaged books" (p. 145). Rather than saving books for scholars who

dislike fiche, a "deliberate strategy" of the project is to work to overcome "scholarly resistance" to microforms—with the result that "scholarly attitudes toward preservation filming will be improved" (p. 142). This goal seems so important to the authors as to constitute the subject of their final paragraph: A different program of education would seem to be needed. Scholars should not be scolded for being "obstructions" to the discarding of artifacts, but they should be taught that there is a far more important reason for fighting that battle than the mere discomfort of having to sit in front of machines to do their reading. That scholars are in need of this lesson is clear from a February 1988 "Summary Report on Preservation Initiative among ACLS Societies," based on responses from constituent societies to a request for information from the president of the American Council of Learned Societies. It seems apparent from this report that only two of the responding societies, the American Antiquarian Society and the Bibliographical Society of America, called attention to the connections between "physical form and intellectual content"; and the report treated their concern as applying to a special category of material rather than to all material.[54]

An appropriate mechanism for the education of scholars (and other readers as well) regarding the relation of copies to originals would now seem to exist, in the Commission on Preservation and Access. This Commission was established in 1986 by the Council on Library Resources to confront on the national level the problem of "capturing and making accessible the content of brittle books" (in the words of its report for the year ending 30 June 1987). The goal is, "in effect, to form a new national collection of preserved materials" by coordinating the filming of at least three million volumes over a twenty-year period. This ambitious effort has already achieved a great deal of visibility, and its active program of public education includes a widely distributed film, Slow Fires, that aims to help form a "cohesive sense of a preservation ethic for the product of mankind's accumulated learning and experience."[55] The undertaking is a noble one, but thus far the Commission has not made an emphatic statement about the importance of the physical evidence present in every artifact.[56] This project could be nobler still if it

included in its educational objectives the teaching of this fundamental truth. The value of the Commission's microfilming program would not be undercut—quite the reverse—by a frank acknowledgment of the limitations of microfilm and a careful explanation of the meaning of documentary evidence. If the Commission would then put these ideas into practice and direct that the remains of every microfilmed book be saved for whatever bibliographical evidence it still offers, the result would be an even greater contribution to the avowed goal of preserving our intellectual heritage.[57] A central repository could be established for receiving the books, if the libraries that possessed them before microfilming did not wish to keep them. By these actions the Commission could be a powerful influence in demonstrating to the general public and scholars alike that every scrap of artifactual evidence is worth saving, that all books (not just "rare" books) are important as objects, even to persons who are not particularly concerned with publishing history and whose only interest is in understanding the texts in books. No one has ever before been in such a favorable position as the Commission on Preservation and Access for publicizing these ideas, and by doing so it could contribute immeasurably to the cause of historical scholarship, without adding greatly to the cost of its endeavor as a whole.

If this moment is not seized, the nightmare vision of microfilming that William A. Jackson depicted nearly fifty years ago may become a reality:

Jackson might be surprised—but, on second thought, probably would not be—by the significance that his phrase "devouring monster" has taken on as the years have passed. Microfilming equipment need not be the monster it sometimes is, and books need not be abandoned after their encounter with it. But if librarians are to protect those books, they must come to understand that their treasures are all the books in their charge, not just those in "treasure rooms." Microfilms and other reproductions can be helpful to scholarship if their proper use is recognized; but equating them with originals undermines scholarship by allowing precision to be replaced with approximation and secondary evidence to be confused with primary. The texts of many documents that once existed are now lost forever, and the texts of others are known only in copies. We use whatever there is; but when there are originals, we must not let substitutes supplant them as the best evidence we can have for recovering statements from the past.

Notes