University of Virginia Library

I

In 1946, Charlton Hinman applied for a patent on his new machine, then still in its earliest developmental stages (“Mechanized Collation: A Preliminary Report” 101). He was just back from the war, where he had served as an officer in a naval intelligence group based in Washington, D.C. (Tanselle, Life 33). Before entering the military, he was working as a Research Fellow at the Folger Shakespeare Library, engaged in the arduous task of collating by hand the First Folio text of Othello (Bond, Letter). This work was an extension of his dissertation, “The Printing of the First Quarto of Othello.” He had taken his doctoral degree in 1941 from the University of Virginia, where he was Fredson Bowers' first Ph.D. candidate (Tanselle, Life 33).

Hinman had chosen from the very start of his career to associated himself with projects that involved the close examination of printed texts. This was a natural choice given the exceedingly analytical, not to mention practical, turn of his mind. When he graduated from high school, his father had insisted that he take six months of vocational training before entering college. Through middle age he was a “tinkerer,” making croquet mallets on his wood lathe, building retaining walls, and taking on other around-the-house projects (Barbara Hinman). His second wife, Myra Mahlow, commented that he was

“just the kind of man who ought to be associated with printing... because he liked things down in black and white. He liked to see the eternal verities verified” (Martin).

The purpose of the machine for which he was seeking a patent was straightforward and grew directly from the needs of his research. During the Renaissance, the period of his specialty, books were proofread and corrected continually during the printing process, and early uncorrected sheets were commonly bound up with corrected ones from later in the print run. Thus the printed matter in the last book sold could, and usually did, differ substantially from that of the first, as it also could and quite often did from nearly every other copy in the printing. These variations are precisely the details the collator was developed to help detect. The operation of the device Hinman would eventually build was also straightforward. The operator sets up one book turned to a particular page on a platform on one side of the machine and another copy from the same printing turned to the same page on a platform on the other. He or she then views these items, which are superimposed via a set of mirrors, through a pair of binocular optics. After making adjustments to bring the two objects into registration, the operator activates a system of lights that alternately illuminates each page. If the pages are identical, they more or less appear as one; if they are not identical, the points of difference are called to the operator's eye by appearing to dance or wiggle about.

The work that Hinman conducted at the Folger before the war was part of a long and unglamorous tradition: centuries earlier, Samuel Johnson, for instance, had observed that collation is “dull” but “necessary” work (Preface xlviii). Collating Othello at the Folger, Hinman was carrying out his research (Hinman, “Alternate Projection”), as had generations of scholars before him, by what some editors have termed the “Wimbledon method” (McLeod, McLeod Portable Collator 2). With two copies of a book open before him, he read from each alternately, line by line and letter by letter, looking back and forth, keeping his place with the aid of a fingertip. Another time-honored technique, though there is no evidence that he used it, is sometimes called “circle collation” and requires more than one person (Oakman 333). One individual reads aloud while the others follow along silently. Done correctly, the person reading aloud calls out marks of punctuation, line breaks, dropped-letters, and other printed features in addition to the words. Collation has always been a laborious and time-consuming activity, but compared to machine collation the older techniques are particularly tedious. Partly because of this, they are also vulnerable to mistakes. In the Wimbledon method, it is very easy to overlook something as one's eyes dart back and forth between pages. In the other method, unless all readers exercise extreme care, it is difficult to differentiate homonyms, e.g., “knight” and “night.” The circle method also does not do a good job of distinguishing alternate spellings of the same word, “go” and “goe,” to take a famous example from the First Folio. Because these routines also involve the act of reading, it is easy for a

researcher to be distracted from typographical features by the content of the text. In addition, neither technique is amenable to efficient collation of multiple copies.

Hinman was distinguished from the many collator-scholars before him by the scale and intensity of his work, for never before had the writings of Shakespeare, or any other author, been subjected to such a close textual reading in so many apparently identical copies. The scholarly scrutiny of the physical characteristics of books dated back several decades before the invention of his machine. At the beginning of the twentieth century, building on the work of a few nineteenth-century trailblazers (notably Henry Bradshaw and William Blades), W. W. Greg and other proponents of what would come to be known as the New Bibliography sought to raise the standards of their field. They advocated techniques and methods that they found lacking in most past as well as contemporary projects, namely “observation, comparison, formal rules of procedure, analysis of materials, and logic” (Maguire 32). Though these qualities are common to good research in any field, it was precisely their conspicuous absence in so many bibliographical undertakings that led Greg and his fellow reformers to encourage a more “scientific” approach. Much has been made of their association of bibliography with science, and it is not my purpose either to revive or review those discussions here. Suffice it to say that in using the term they were attempting to underscore the rigorous and systematic nature of their work in contrast to what they saw as little more than a “dilettante interest in old books merely as antiquarian objects” (Tanselle, “Bibliography” 60).[3] In the history of this movement, the name Fredson Bowers stands equal to, if not even a little ahead of, that of W. W. Greg. Bowers was a second generation New Bibliographer, but his contributions to the theory and development of the movement were enormous. It has been observed that he put the “American” in the “Anglo-American” approach to bibliography (Tanselle, Life 146). In studying under Bowers, Hinman was receiving his New Bibliographical training in situ. One commentator even referred to the collator as an “electronic Bowers” (“Mechanized” 156). Though offered in jest, this description is particularly apt. The machine descends directly from the most important aspect of the New Bibliography—its rigorous emphasis on physical evidence and specifically the evidence found in the very objects under investigation. Greg, Bowers, et al. made books the focus of research in a way that they had never been before; they were now primary evidence in their own history. There were people, as will be discussed below, who preceded Hinman in the attempt to use technology as an aid to the study books and printing, and in time this emphasis on physical evidence would lead to the introduction of photography, microscopes, beta-radiography, micrometers, and other techniques and tools to the field. However, the Hinman Collator, as the first and the most famous device invented specifically for the close and systematic analysis of printed works,

represents the spirit of this movement incarnate. In a sense, it is indeed an electronic New Bibliographer.

For Hinman as a textual critic, the underlying assumptions as well as the overall goal of his research were also the same as for his theoretical forebears, though they were the originators of neither the goal nor the assumptions. Their analytical and material-based emphasis was new, but the orientation and aim of their work was the same as it had been for generations of editors. W. W. Greg and Samuel Johnson worked under essentially the same paradigm and sought the same quarry. Both men saw the printing process as inherently corruptive and both desired to excise those corruptions to establish a text more representative of what the author in question wanted or intended. As with Johnson, the author whose intentions Hinman was pursuing was no less than the inimitable Bard himself, though Hinman came to the task with a much more sophisticated knowledge of the printed remnants before him as physical structures. The collator was invented precisely to help detect, and thus help sweep away, all the non-authorial words, letters, and even marks of punctuation standing between Hinman and the ideal exemplar he was striving to reconstruct. In words that echoed those of editors for hundreds of years before him, he described his goal as nothing less than the reconstitution of “what Shakespeare actually wrote” (Printing vii).

In addition to a practical cast of mind, a theoretical framework that emphasized the books themselves as primary evidence, an orientation toward the author as the authority in textual matters, and training at the hands of the individual who would become the foremost textual and bibliographical scholar of the postwar era, Hinman also had before him a famous example of the application of technology to bibliographical research. In 1910, William J. Neidig used composite photographs to superimpose images of title-pages in the famous Pavier quartos (Neidig) [plate 2]. By this method, Neidig confirmed Greg's earlier contention that most of these nine Shakespeare plays had been printed in 1619 rather than variously in 1600, 1608, and 1619, as their title pages are dated. When Neidig examined his superimposed images, he found that the upper half of the pages varied as the titles of the plays changed. The bottom half of each page, however, often registered exactly the same in many details, even in regard to the wear of the type and nicks in the woodcut ornament. No seventeenth-century printer would have left a title-page standing for twenty years, and neither could a compositor have successfully reproduced the pages so exactly. Therefore the plays must have been printed, as Neidig concluded and Greg had earlier argued, at the same time. Both Neidig and Hinman, though a few decades apart, were collating texts, only in slightly different ways. Neidig, perhaps because he was more interested in similarities than differences, did not take the next step of viewing the pages alternately. Nevertheless, in striking on the idea of superimposition, he carried a baton that would later be passed to Charlton Hinman, though Neidig himself would not do the passing.

In the 1930s, Edwin Eliott Willoughby recognized in the seventy-nine

copies of the First Folio at the Folger a treasure trove of textual and bibliographical data. The Folio had been the subject of his graduate work, and in 1932 he revised and published his research in a short monograph on the subject (Printing). Willoughby realized, as Hinman would a few years later, that by traditional methods the collation of all those copies of the Folio would take more than one working lifetime. In 1933 he proposed a project to collate the First Folios and set out to find an optical instrument which would “superimpose the image of a page of the Folio in one copy upon the same page in another copy” (Willoughby, Uses 95). He apparently located two such devices, though where he found them and specifically how they worked he did not say. Neither did he use them for very long. One generated too much heat to be safe for books and the other caused severe eyestrain. Furthermore, Willoughby, like Neidig, only pursued the idea of mechanical collation as far as superimposition. The advantages of viewing images alternately did not occur to him. However, in attempting to apply collation by superimposition to regular pages of text rather than just title-pages, Willoughby carried the baton a bit further than Neidig and, more importantly, passed it on to Hinman. In 1941, during Hinman's term as a Research Fellow at the Folger, Willoughby, who was on the Library staff, told Hinman about his earlier experiments (Uses 95).

There were other runners before and between Neidig and Willoughby, however. In 1931, G. A. E. Bogeng, librarian of the Herzog August Bibliothek in Wolfenbüttel, described the use of the stereoscope to superimpose printed texts in the search for typographical variants (Bogeng 138-140). Stereoscopes have been around since the early nineteenth century and have been used for many purposes. They were a popular Victorian parlor novelty, and they were also used by astronomers (“Stereoscope”). Bogeng, in fact, was not the first to envision a textual application for the device. In 1894, the Frenchman F. Drouin proposed the stereoscope for the detection of forged banknotes. He also suggested that the device might be useful in the examination of printed books, though, not being a bibliographer, the idea of using it to discover typographical variants within editions did not occur to him. He suggested instead that the device might be useful for distinguishing between similarly typeset copies of different editions (Drouin 102-103). Drouin described several different types and configurations of the device, though he did not specify which would be best for working with books. Neither is it possible to tell from Bogeng's account what kind of stereoscope he was describing. It does not appear that he was writing about one designed exclusively for bibliographical work, however. He spent a good deal of effort explaining the difficulties produced by the device's restricted field of vision. If he or someone else had put one together specifically for textual collation, it seems likely that they would have designed an apparatus capable of displaying an area adequate for the texts they were studying. But if Bogeng was not describing a stereoscope built or adapted especially for the study of books or other printed documents, his description also does not offer any clues as to whether

he was describing the use of one borrowed from an observatory or some other source. Whether Hinman knew of Bogeng's work (or Bogeng of Drouin's suggestions) is not known, but there is some evidence that Hinman may have experimented with a stereoscope just after the war.[4]

More generally, there were also other scholars and individuals who had used technology for bibliographical or closely related purposes. In 1935, R. B. Haselden published a short monograph on the use of scientific devices for the study of manuscripts. One of the instruments he recommended was the comparison microscope. Though it neither overlays nor alternates images, this device is suggestive of optical collation (Haselden 50-52). Haselden also drew on works that discussed the study of paper and books in law enforcement, specifically for the detection of fakes and forgeries. The forensic investigation of documents has a long tradition outside of but not completely unknown within bibliographical circles. In his history of papermaking, Dard Hunter cited many early twentieth-century legal cases that utilized the examination, and in some instances the “collation,” of watermarks (408-427). Hunter's examples involved a different kind of evidence from what Hinman used, but they provide examples of and precedents for a material-based approach that pre-dates Hinman by several decades. The influence of forensics is most evident in Carter and Pollard's study of the Wise forgeries, the most brilliant example of analytical bibliography to precede Hinman. They drew on the technical expertise of individuals in the fields of papermaking and of type design and manufacture, and used tools such as the microscope to analyze paper fibers as well as to compare typographical impressions. The extent to which any of these individuals knew of or were influenced by each others' work, or how much if anything Hinman owed to any of them other than Willoughby, is unknown. Carter and Pollard, Neidig, and even Haselden (given that his monograph was published by the Bibliographical Society) would have certainly been known to Hinman. The important point, however, is that individuals before him pondered similar problems and attempted to solve them in very similar ways.[5]

Though Hinman's military service interrupted his work at the Folger, ultimately it only hastened the progress of his research, for separated from all those copies of the First Folio he had the time to seriously think about mechanical collation and its labor-saving potential. He was trained as a Navy cryptanalyst and assigned to a code-breaking unit in Washington, D.C., though he would spend most of the war on special assignment in Australia (Bond, Letter). Other members of his intelligence unit no doubt encouraged his wartime ruminations. Fredson Bowers was his commanding officer, and William H. Bond, Giles Dawson, and Ray O. Hummell, all of who would go on to distinguished bibliographical careers after the war, served in the same group. In the course of his service Hinman heard a story about the analysis of aerial photographs that guided his thoughts on mechanical collation. He later reported it in two slightly different versions, the best known of which was published in the pages of Papers of the Bibliographical Society of America. There he states that the military used aerial photography as an aid in the evaluation of target areas—“gun emplacements and the like.” The process involved taking aerial photographs before and after a bombing run. The images were later overlaid on a screen and then viewed alternately: “first one picture for a fraction of a second, then the other picture for the same brief period, then the first picture again, and so on. The result was—or at any rate was supposed to be—that wherever there had been no change in the target area since it had first been photographed the screen showed only a single, perfectly motionless picture of that area; but that wherever there had been a change the picture on the screen flickered or wobbled” (“Mechanized Collation: A Preliminary Report” 103). The other version was delivered as a lecture in June of 1947 to the Bibliographical Society of the University of Virginia. It is essentially the same except that he does not explicitly mention bombing. Instead, he recounts that aerial photography was used to evaluate the movement of enemy supplies and other resources. He also now sets the scene in the Pacific and specifically identifies the Japanese as the object of this observation (“Why 79” 12).

In both instances he states that the story was not completely true. The military did use aerial photography in its intelligence gathering efforts. Many so-called “pre-” and “post-strike” photographs survive in archival collections

[Description: Charlton Joseph Kadio Hinman]

[Description: Composite photograph produced by William J. Neidig]

[Description: Blink comparator at the Harvard College Observatory]

[Description: Patent drawing of Hinman's prototype collator]

[Description: Second, re-designed Hinman Collator]

[Description: Delivery of the Ohio State University collator]

[Description: University of Iowa collator]

[Description: University of Houston collator]

around the country. Furthermore, the military did experiment with various techniques to heighten the effectiveness of aerial photography. For example, Harold E. Edgerton, an MIT engineer and pioneer of high-speed photography, was employed by the Army Air Force to build a better electronic flash device for night time surveillance. The part of the story that was false involved the superimposition and alternate viewing of pictures. The military, as Hinman later pointed out, conducted experiments along these lines (Gwinn) but to his knowledge had no “success in this particular kind of photographic reconnaissance” (“Why 79” 12) for “several good reasons” (“Mechanized Collation: A Preliminary Report” 103). Hinman did not enumerate these reasons, but at least one of them is not difficult to imagine. Using World War II technology, it simply would not have been possible to photograph the same patch of ground twice from exactly the same altitude and position in the sky under the same atmospheric conditions, and without being able to replicate those conditions more or less exactly, the two photographs would not have registered well enough to make their superimposition possible, much less meaningful. The importance of the story for Hinman was that the “underlying principles” were sound and thus helped steer him in the right direction (“Mechanized Collation: A Preliminary Report” 102).[6]

Fascination with Hinman's wartime experience has led to a glossing over or exaggeration of certain of its aspects and, in turn, to some misunderstanding about the development of mechanical collation. It should be emphasized that Hinman's unit specialized in cryptanalytic work and not photographic analysis (Bond, Telephone, 30 March 2001). Thus, the aerial photography experiment was something that Hinman “heard,” perhaps from a neighboring unit or a colleague elsewhere in the intelligence community, and not something that he observed directly. This fact by itself may account for the slight differences in the two versions of the story and some of the ambiguity that has surrounded it. He possessed no first-hand knowledge that the military had a method of examining photographs similar to the one that he used to study books. Another detail that has been misunderstood (and Hinman never

made the point clearly himself) is whether machine collation first occurred to him during the war and as a result of his intelligence work.[7] The war certainly played an important part, but not because it gave him the idea of the collator in the first place. Rather, the war provided the time to ponder an idea he had been considering for some time, to speculate “on ways to avoid such labors [i.e. textual collation by sight]” (“Why 79” 1). The war also put Hinman in an environment, surrounded by like-minded scholars and supervised by his dissertation advisor, unusually conducive to bibliographical musings. And, as I have shown, Hinman was also not the first person to harness technology in an effort to study texts generally or even to compare copies from within the same edition.

Unfortunately the aerial reconnaissance story has been repeated so often that it has eclipsed a more important technological antecedent, and one that actually worked. Arthur M. Johnson, who would take over the commercial manufacture of the collator, wrote that Hinman developed the basic design of his machine after studying something called the “astronomer's microscope” (Johnson, Letter to William P. Barlow, 21 Nov. 1973; Johnson, Hinman Collator [1972?] I). The device to which Johnson was referring is properly known as the blink comparator [plate 3] and was invented in 1904 by the German instrument-maker Carl Pulfrich (Drummeter 14). The basic principle behind the blink comparator is the same as that of the Hinman. Two objects, in this case photographs of the same star field taken on different dates, are set up in the machine, superimposed, and then viewed alternately. Any difference between the images calls attention to itself by appearing, just as on the Hinman, to dance or move about. The most famous use of the comparator was made by C. W. Tombaugh, who discovered the planet Pluto with it in 1930 (Moore 81; Abbott 158).

Hinman knew of this machine and may have even investigated adapting it for books (Bond, Letter). A primitive blink comparator had been used at the McCormick Observatory at the University of Virginia from the 1920s through the 1940s (Virtual Museum). It is not known if Hinman ever saw the comparator at Virginia, but we do know that he spent time studying, if not an actual device, then at least the principles behind its operation. The military's unsuccessful wartime experiments may have encouraged Hinman, but the blink comparator was the real technological model for his machine. The collator is therefore better viewed not as a completely new invention but rather as a new application of an older technology. Hinman himself said his device was based on “conventional principles” put to use “in what patent lawyers call `a new mechanical environment'” (“Mechanized Collation at the Houghton Library” 132).[8]

When commenting on Hinman's demonstration of the prototype to the Bibiographical Society of America in 1947, the editor of PBSA was careful to place the word “machine” in quotation marks, suggesting perhaps that the operation of the device was a good deal less precise at this point than the word normally merits (“Mechanized Collation: A Preliminary Report” 99). The primary components of the prototype, as Hinman famously described them, were a “pair of ordinary microfilm projectors (borrowed from the Navy), some pieces of wooden apple box (abstracted from a trash pile), some heavy cardboard (begged from the Folger bindery), and parts of a rusty Erector set (more or less hi-jacked from the small son of a close personal friend)” (“Mechanized Collation: A Preliminary Report” 102).[9] An unacknowledged contributor to this prototype was John Cook Wyllie, Curator of Rare Books at the University of Virginia. Very early on, Hinman thought that the use of “color differentiation” might help in the collation process (Hinman, Letter to John Cook Wyllie, 21 Mar. 1946). The idea was to heighten differences between the pages being compared by preparing the microfilm images of them in contrasting colors. When the images were then overlaid and alternated, any variants would call attention to themselves by not only appearing to dance about but also by showing themselves in different colors, or so the theory went. Wyllie was the point man on this idea. He conducted experiments for Hinman at Virginia using the Eastman Kodak “wash-off relief process” (Hinman, Letter to John Cook Wyllie, 28 Feb. 1946). Although Wyllie achieved beautiful reproductions with this method, they appear not to have worked for collational purposes. Apparently the “scale of reduction necessary for efficient folio work” made the color differences meaningless (Wyllie, Letter to Charlton Hinman, 18 Apr. 1946), and the whole scheme was rendered moot when Hinman redesigned the machine to work primarily with originals rather than photographic reproductions. Wyllie remained an interested and enthusiastic observer, however. He attended Hinman's 1946 demonstration of the prototype for the Modern Language Association (Wyl-

lie, Letter to Charlton Hinman, 9 Jan. 1947). A few years later he also spear- headed the effort to bring one of the earliest commercially produced machines to Virginia.[10]

Hinman submitted detailed drawings of his prototype with his patent application in 1946, and these drawings show a much more sophisticated design than his PBSA description suggests [plate 4]. The “Alternate Projection Collating Device” consisted of two microfilm projectors placed at right angles in very close proximity to one another. One projector (1) cast an image onto the reflective side of a small two-way mirror (2), which then deflected the image onto a screen (3). The other projector (4) was positioned so that its image was thrown upon the same screen (3) after passing through the back or non-reflective side of the same two-way mirror (2), and in this manner the two images were superimposed. Two occulting discs (11 and 7), one positioned in front of each projector, turned so as to interrupt the projection alternately, thus casting the images on the screen one after the other rather than simultaneously. The speed at which the discs turned, and therefore also the rapidity with which the two images were alternately cast upon the screen, was regulated by means of a rheostat (12). The whole apparatus was mounted on a kind of “bread board” platform (Hinman, Letter to John Cook Wyllie, [Mar. 1955]). Hinman carried out his first experimental collations with this device at the Folger in the summer of 1946. He was granted a patent in 1949,

though by that time he had abandoned this design and completely reconfigured his idea into the machine we know today. The prototype suffered from four serious flaws: the researcher could not closely study discrepancies without getting in the way of the projected light; taking notes was difficult since the machine worked best in a darkened room (Johnson, Letter to William P. Barlow); the machine was noisy and thus disrupted other researchers (Mason, Personal); and in contrast to the final version, the prototype was not designed to work with the original texts but only with microfilm copies.

The last of these shortcomings was the most serious, for collating from copies always involves, as Hinman well knew, some inherent disadvantages. A photographic image cannot always be relied upon to reproduce the original accurately. Fly specks, for example, are easily transformed into punctuation marks, and very faintly printed marks may not show up at all. Photographs are also expensive, and with a book of any length the time required to take and develop them would exceed that required to perform the collations. What was needed, he suggested, were two machines—one, which would be more common and more frequently used, for photographs, and a second, less common and less often used, for the originals (“Mechanized Collation: A Preliminary Report” 105). The device that Hinman would ultimately build could do both, though it was primarily intended for and has been predominantly used with original texts. Of course, gathering enough original copies of the book he needed was no problem. The Folger, with seventy-nine First Folios, had more than an adequate supply for his purpose. Mechanical collation owes its invention perhaps as much to the rich resources of the Folger as it does to the ingenuity of Charlton Hinman. Certainly without the availability of such a large sample of so important a book, comparison of copies, especially on such a large scale, would not have seemed nearly so necessary, let alone possible.

By July 1949, Hinman had abandoned his prototype for a completely redesigned collator, though a few more years of tinkering would pass before he would consider the device perfected (Altick 188; Hinman, “Variant Readings” 280-281). This was the machine he would use to make his study of the First Folio [plate 5]. Within a few weeks of installing it at the Folger, he had already collated his way through “18 formes” that included “the whole of Othello and parts of Lear and of Anthony and Cleopatra” in all of the Library's copies of the volume (Hinman, “Mark III” 150). Though his work was interrupted for two more years when he was called back to military service for the Korean conflict, in 1951 he was able to report the results of these collations, thus laying the first stone in the scholarly monument that would come to be considered an “irreplaceable landmark in Shakespeare studies” (Blayney 5). At the end of 1952 he was back at the Folger, and the next year he announced that his new tool was, more or less, complete (Hinman, “Variant Readings” 280).