Why is bibliography important? If it is to be more than a list of titles, what use is it? The question has acquired new pertinence now that texts have become both more available and less trustworthy, thanks to the Internet. Students usually download texts from computers without asking where they came from, and they frequently get garbage. But the problem is not new.

Here is a passage from the first printed version of King Lear (the quarto edition of 1608, Act III, scene 4): “swithald footed thrice the old a nellthu night more and her nine fold bid her, O light and her troth plight and arint thee, with arint thee.”1 It is seventeenth-century garbage, which probably was almost as incomprehensible to seventeenth-century readers as it is to us. To make sense of it, textual editors have drawn on folklore, philology, paleography, the history of religion, and their own intuition. They have concluded that Shakespeare meant to evoke the notion of Saint Withold driving away a female demon and her brood during a tempestuous night. This kind of textual criticism, accompanied by commentary and variants in footnotes and appendices, is familiar to any reader of Shakespeare. What can bibliography contribute to it?

Consider another example of trash in Shakespeare, the quarto edition of The Merchant of Venice published in 1619. In Act I, scene 3, Antonio asks Bassanio whether Shylock knows how much Antonio wants to borrow: “are you resolu’d, How much he would haue?” In the original edition of The Merchant of Venice, a quarto of 1600, the lines read: “is hee yet possest How much ye would?” Which to prefer? We cannot know what Shakespeare intended, because no manuscripts of his plays survive—except perhaps three pages, written in his hand, from the unperformed tragedy Sir Thomas More. But we can identify the most corrupt passages in the early printed versions.

By analysis of the physical copies, bibliographers have determined that the type of the 1619 quarto was set by the same compositor, a particularly slipshod workman whom they call Compositor B, who set nine other quartos of Shakespearean or pseudo-Shakespearean plays in the same year, using earlier editions as his copy. When he came upon a phrase that he considered deficient, he “improved” it. So the 1619 version of those lines is pure Compositor B, and the text of the play as a whole (it has an average of one significant error in every twenty-three lines) is very impure Shakespeare. Moreover, B also composed about half the text of the First Folio, our main source for reconstructing Shakespeare’s oeuvre. To make sense of Shakespeare, therefore, it is not enough to be a literary critic. One must also be a bibliographer—or at least understand enough of bibliography to know how books came into being in the late sixteenth and early seventeenth centuries.2

This kind of bibliography—usually called “descriptive” or “analytical” bibliography to distinguish it from the book-listing or “enumerative” variety—became a powerful force in the humanities during the first half of the twentieth century. But what exactly was it, and did it have implications for anything besides the editing of texts? Sir Walter Greg, the ultimate authority in these matters, defined bibliography as “the science of the material transmission of literary documents.”3 His formulation was contested by some who found “science” too positivistic and “literary” too narrow, since bibliographical analysis could be applied in principle to any kind of text and any form of communication. But the emphasis on materiality appealed to all bibliographers, because all of them studied books as physical objects. By learning how texts became imbedded in paper as typographical signs and transmitted to readers as pages bound in books, they hoped to understand a fundamental aspect of literature itself.

Greg and R.B. McKerrow began to work out the basic concepts and techniques of this “new bibliography,” as it came to be called, when they were students at Trinity College, Cambridge, in the 1890s. With the publication of An Introduction to Bibliography for Literary Students by McKerrow in 1928 and Principles of Bibliographical Description by Fredson Bowers in 1949, bibliography emerged as a coherent discipline with standards that had coalesced into an orthodoxy. By 1950, bibliography became a requirement for the Ph.D. in many English departments. Along with philology and other professional skills, graduate students learned how to recognize formats, collate signatures, detect cancels (leaves inserted to replace leaves with errors or potentially offensive passages), distinguish typefaces, trace watermarks, analyze art work, and identify bindings.

Shakespearean studies especially flourished in this environment, because the early editions of the plays, published at a low point in the history of printing, are full of errors and cannot be corrected against an original manuscript. As far as we know, Shakespeare took no part in their publication. For him, it seems, the performance was what counted, and he probably modified the scripts as the action evolved on the stage or rewrote them for reruns. We can imagine his “foul papers” (early drafts) and prompt books, but to come up with texts we have to find a way through the faulty editions thrown together in the printing shops of his day. Hamlet appeared first in a primitive quarto of 1603, next in a quarto of 1604–1605, which is twice as long, and then in the folio of 1623, which has eighty-five new lines and differs greatly from both of the earlier editions. King Lear presents so many puzzles that its most recent editors printed two versions of it.4 They are radically different, yet each conforms to the most exacting bibliographical standards and each may represent a version that Shakespeare considered at one point to be final. So we now have two King Lears, as well as older, conflated editions, and we are richer for it, thanks to bibliography.


Textual conundrums of this sort inspired generations of scholars to feats of ever-greater virtuosity. By poring over early editions, they have traced typographical clues of every variety—inconsistent spelling, irregularities in spacing, chipped type, anything that could help them reconstruct the production processes of Elizabethan printing shops and therefore get closer to Shakespeare’s missing copy. Many learned to set type themselves and turned into amateur letterpress printers. In their imaginations, Ph.D.s became companions of the workers who first turned Shakespeare’s words into books. It was an intoxicating idea, and it did not last.

Bibliography has not disappeared, but it has been pushed aside and ignored by more recent trends in literary scholarship. From the New Criticism of the 1940s to the deconstructionism of the 1960s and the new historicism of the 1980s, the study of texts became increasingly detached from their embodiment in books. Bibliography began to look like an arcane discipline that might have uses for editing Shakespeare but little relevance for understanding modern literature. Some modern works, from Pamela to Ulysses, posed important bibliographical problems, but most could be edited with minimal notes about textual variants. In 1968 Edmund Wilson raised a storm by denouncing editions sponsored by the Modern Language Association for bibliographical overkill—he mentioned a project in which eighteen editorial workers were preparing an edition of Tom Sawyer by reading the text backward. When the polemics died down, bibliography had lost much of its appeal. It disappeared from graduate programs and even from most library schools. To a generation that had witnessed the collapse of the canon and the rise of the Internet, fine-grained analysis of old books no longer was attractive.

In the midst of this self-questioning, the inevitable occurred: heresy. All orthodoxies generate heretics, but the Martin Luther of bibliography, Donald F. McKenzie, was especially threatening to the old guard, because he could beat the best of them at their own game. Having assimilated the Bowers principles and developed into an expert printer himself, McKenzie left his native New Zealand for Cambridge, England, where he wrote a Ph.D. under Philip Gaskell, a master bibliographer. The book that resulted, The Cambridge University Press, 1696– 1712 (1966), was hailed as one of the most rigorous works ever written in the tradition of Greg and McKerrow. It had a disquieting aspect, however, because not only did McKenzie provide a bibliographical analysis of every book produced by the CUP during those sixteen years, but he also related the physical evidence to manuscripts from the archives of the press, and the manuscripts revealed that things had not taken place as they should have, according to the conventional wisdom.

Compositors did not supply pressmen with “formes” (pages of type arranged inside an iron frame and locked in place so as to be ready for printing) in a consistent pattern. On the contrary, a compositor would send a completed forme to whatever press was free. So at one point or other, all the pressmen of the shop often ran off copies of a particular book. Moreover, compositors also switched frequently from one job to another. They might set the type for one forme of a treatise like Newton’s Principia, published by the CUP in 1713, then compose a bill of lading or a receipt, and later take up a book of sermons. Some tasks took longer than others and some were more urgent, so the foreman distributed them in the most efficient way, and several books were always moving through production at the same time, each following its own, erratic rhythm. The regularity of output at the shop level compensated for the irregularities in the labor of each man, a way of organizing work that McKenzie called “concurrent production.” It sounded innocent enough as an idea; but when he developed all its implications, he seemed to sap the foundations of orthodox bibliography.


Previous bibliographers had assumed that each book would move through the chain of production according to a consistent, linear pattern: a certain compositor would feed formes to the printers at a certain press, who would run off the edition, frequently leaving traces of their activity in the pattern of headlines at the top of the page, direction lines at the bottom, or press figures (usually numbers added at the bottom to identify the work of individual pressmen). It therefore would be possible to construct a se-ries of inferences, moving backward through the production process from the physical copy, to a press, a compositor, and, at least to some extent, the original manuscript, even if it were missing, as in the case of Shakespeare. Above all, Shakespeare. The search for reliable texts of his plays drove the whole discipline.

The greatest Shakespearean bibliographers, notably Greg and Charlton Hinman, allowed for irregularities. The supreme study of a book from the era of Shakespeare, Hinman’s The Printing and Proof-Reading of the First Folio of Shakespeare (1963), showed how the First Folio came into being, forme by forme, while other books were being printed at the same shop. At one point Hinman even used the term “concurrent production.” But most bibliographers took the individual book rather than the output of the entire shop as the unit of analysis, and this line of reasoning, valid enough within its own limits, led them to string together questionable hypotheses about the men who produced the first printed copies of Shakespeare. In place of workers made of flesh and blood—pre-industrial artisans who worked in erratic spurts and knocked off for bouts with the likes of Falstaff and Mistress Quickly—they substituted ghostly abstractions like compositor B surrounded by A, C, and others, who were deemed to have turned out quartos and folios at regular rhythms in accordance with the principles of bibliographical science.

Not that these imaginary beings worked like robots. On the contrary, it seemed possible to demonstrate that one man had unusually erratic spelling, that another frequently mixed up homonyms, that a third worked from an inadequate font of type, and that all of them scattered idiosyncratic marks on the pages in patterns that revealed their hands at work as distinct from Shakespeare’s. By identifying the passages that they mangled—in the case of Compositor B’s Merchant of Venice, forty words or phrases of his own that he substituted for Shakespeare’s—bibliographers hoped to isolate the alien elements in the greatest works in the English language. It was a process of elimination, essentially negative in its findings, but it brought the modern reader closer to what Shakespeare had actually written.

If the printing shop operated according to the principle of concurrent production, however, it would be difficult to determine precise patterns of production; one could not link specific passages to specific compositors with complete confidence; and the chain of infer- ences could break at crucial points. A, B, C, and the rest might be figments of overheated bibliographical imaginations, mere “printers of the mind.” That was the title McKenzie fixed to an essay of 1969, which shook the world of rare book rooms as if it were an earthquake. For the next decade scholars debated the principles of bibliography with all the passion that academics are capable of injecting into academic questions. They were generally ignored by the rest of the world, which had other things on its mind during those years. But to bibliographers, the stakes were enormous. McKenzie seemed to expose a seismic fault that ran right through their discipline.

The orthodox bibliographers defended their position with two arguments: first, that the Cambridge University Press, a small, specialized bus- iness in a provincial town at the beginning of the eighteenth century, could not be taken to typify operations in the larger printing shops of London nearly a hundred years earlier; second, that archival evidence did not invalidate the basic principle of using the analysis of books as physical objects to reach conclusions about the process of their printing—notably in the case of the early editions of Shakespeare, for that is what generated most of the heat in the polemics. If bibliography could not provide a reliable method for editing Shakespeare’s texts, what good was it?

McKenzie dealt with the first argument by drawing on evidence from the papers of William Bowyer, an important London printer, which were discovered in 1963. They confirmed the principle of concurrent production and showed even more complex and irregular patterns in the flow of work, which was frequently shared by several shops as well as by several workmen. A few years later, Jacques Rychner demonstrated that McKenzie’s analysis also held true for the production of books in the printing shop of the Société Typographique de Neuchâtel in Switzerland. To be sure, the archives in Cambridge, London, and Neuchâtel all came from the eighteenth century. But there were no significant changes in the technology of printing from 1500 (or perhaps even in Gutenberg’s time) to 1800. All three manuscript sources—and additional documentation from two other eighteenth- century London printers, William Strahan and Charles Ackers—proved that McKenzie was right: production in early-modern printing shops did not follow the regular pattern ascribed to it by orthodox bibliography.

But could manuscript material from the eighteenth century disprove arguments based on the physical analysis of Shakespearean quartos and folios? McKenzie never went that far. In fact, he produced the most thorough account of how Compositor B botched The Merchant of Venice. There was nothing wrong in principle with ascribing particular passages to compositors who could be called A or B or anything else. We even know a little bit about the actual men who worked in William Jaggard’s printing shop where the First Folio was produced in 1622– 1623—including the fact that a John Shakespeare, apparently no relation to William, had served as an apprentice to Jaggard from 1610 to 1617. By minute study of the Folio, quire by quire, Hinman thought he had found a way to identify the compositors behind the text and thus get “a little closer to the truth about what Shakespeare actually wrote.”5

Forty years after he published The Printing and Proof-Reading of the First Folio of Shakespeare, it seems that he, too, was right. The latest study of the Folio, by Peter Blayney, a partisan of McKenzie, has confirmed nearly all of Hinman’s conclusions. Blayney has identified a few more compositors and modified Hinman’s account of proof-reading. It now appears that the actors from Shakespeare’s troupe had corrected the proof before the compositors added stop-press corrections during the printing. The first edition included three distinct issues: one had thirty-five plays; one had thirty-six, including Troilus and Cressida but without its prologue; and one had thirty-six with Troilus, prologue and all. The printers scattered clues to these irregularities by marks left in the text. In some cases they crossed out a redundant page of Romeo and Juliet. In others, they left in corrections added by hand during the final proofreading. The text was always changing, always slipping morphologically from one state to another.


That lesson, from “the most important single book in English literature,” as Helen Gardner put it,6 bore on a larger issue raised by McKenzie’s apparent heresies. Bibliography might help resolve some difficulties peculiar to the editing of Shakespeare, but what could it contribute to the general understanding of literature? McKenzie himself addressed this problem in an essay of 1977, “Typography and Meaning: The Case of William Congreve,” which proved to be almost as influential as “Printers of the Mind.”

Congreve made a particularly interesting case to study, because he straddled two typographical eras. The first editions of his plays, casually printed quartos from the 1690s, were nearly as crude as the quartos of Shakespeare, whereas the three-volume, octavo edition of his works in 1710 exuded the stateliness of a classic. Which to prefer, the seventeenth- or the eighteenth-century Congreve? McKenzie faced this choice in preparing a critical edition of the works. He began by rejecting Greg’s famous distinction between “substantives,” or the basic text of a play, and “accidentals,” typographical ingredients such as ornaments or extra spacing added by the printer to separate the scenes of a play. To Greg, accidentals were merely a matter of presentation, not one that affected the meaning of a text. To McKenzie they were crucial in mediating the difference between two kinds of experience: watching a performance on a stage as opposed to reading a text on a page. Whatever effects the playwright originally had in mind when he composed a script, his play took on new meaning when it became a book, because at that point the dramatic action could only be imagined by readers working from typographical clues.

Congreve participated consciously in the shift from one medium to another, because by 1710 he had stopped writing for the stage and was concentrating on the publication of his plays. The octavo edition of his works set a standard for the new form of the book that came to prevail in the eighteenth century. Unlike the cumbersome folios and slap-dash quartos of the earlier era, it was small enough to be held comfortably in the hand and elegant enough to appeal to the tastes of a new consumer society. Congreve purged some of the bawdiest passages, but he retained most of the original texts. What gave them new meaning was the design of the book, a collaborative project worked out by Congreve with his close friend and publisher, Jacob Tonson, and Tonson’s highly skilled printer, John Watts.

Using larger sheets (but smaller pages, because an octavo sheet was folded three times and a quarto twice before being assembled into a volume) and more balanced spacing, they gave the book a refined symmetry. In place of the minimal directions of the old quartos—usually nothing more than an “enter” or “exit” to signal new scenes—they set off scenes by numbers, typographical ornaments, and lists of characters. The reader could therefore imagine who was on stage at any given time and could see how all the parts fit within the whole. Scenes, plays, the entire oeuvre were clearly articulated, as in neoclassical architecture. Congreve took his place next to Shakespeare—who had appeared in similar typographical dress a year earlier—in what was beginning to emerge as a canon of classics.

At this point, McKenzie’s argument converged with themes that had been developed in an adjoining field of research, the history of books. Unlike bibliographers, book historians studied all aspects in the production and diffusion of the printed word, including its connections with political and social change. For them, 1710 stood out as a turning point in the history of copyright. It was the year when Parliament passed the first copyright law, entitled “A Bill for the Encouragement of Learning by Vesting the Copies of Printed books in the Authors, or Purchasers, of such Copies, during the Times therein Mentioned.” As its title indicated, the law gave authors a new prominence. Although it did not actually mention them in its text, it recognized their proprietary right to the products of their imaginations. Alexander Pope showed that authors could support themselves from the sale of those rights. By mid-century Samuel Johnson epitomized the professional writer, who lived from his pen instead of from patronage, and gloried in his role of supplying demand on the literary marketplace. Literature itself was emerging as a semi-autonomous system organized around the printed book, in contrast to the world of letters from the sixteenth and seventeenth centuries. Under the Tudors and Stuarts, communication in the public sphere took place primarily through performances—on stage, from pulpits, at court, and in the streets. In Georgian England the printed word predominated, even though manuscript books continued to flourish (if fewer than one hundred copies were published, a book could be produced more economically by scribes than by printers) and news still spread by word of mouth.

The publication of Congreve therefore belonged to a general process, the transformation of letters into literature, and McKenzie announced that it needed to be understood from a broad perspective, what he termed “the sociology of texts.” In moving from science to sociology, McKenzie’s perspective could hardly have been further from the discipline of Greg and McKerrow; but it opened a way for Anglo-American bibliography to make a juncture with French histoire du livre, the wide-ranging variety of book history developed by Lucien Febvre and Henri-Jean Martin. In L’Apparition du livre (1958), they related the impact of Gutenberg’s invention to long-term social and economic phenomena such as the organization of scriptoria, the price of rags and parchment, and the development of trade routes. They stressed the need for quantitative evidence in order to measure continuity against change. And as partisans of the Annales school of history, they detected long-lasting patterns of structural stability, which led them to challenge accepted wisdom, including the belief that Gutenberg produced an immediate revolution in the publishing industry.

McKenzie attempted something similar by shifting from fine-grained analysis of individual books to the study of the London book trade as a whole, which he surveyed by making sweeps through all the surviving evidence from three years: 1644, 1668, and 1689. Research on this scale required a prodigious amount of labor, because McKenzie combined quantification from his main source, D.G. Wing’s short-title catalog of books printed between 1641 and 1700, with the examination of every copy that he could locate in major research libraries. By counting the number of sheets in each copy, he was able to make a better estimate of total output than by simply counting titles, and he could look at the entire literary landscape from the perspective of productivity and economics.

For 1668, Wing and a few additional sources provided a total of 491 titles, of which McKenzie physically studied 458. He could not produce a full analytical description of each of them, but his expert eye detected all sorts of trends and oddities. The names of the printers did not appear on more than half the title pages. Reprints accounted for nearly a third of the total output. And only fifty-two books bore some form of license or official permission to publish, despite the requirements of the licensing act of 1662. The booksellers’ main concern was to protect their copyrights, and this they could do by informal “combinations” among themselves, such as joint arrangements for marketing and sales. It looked as if printers and booksellers went about their business without paying much attention to politics and without developing much of an appetite for innovation.

Conservative, commercial interests even dominated the trade during revolutionary times. By surveying nearly everything published in 1644 at the height of the English Civil War, McKenzie found a surprising degree of continuity in overall production. He rejected the argument advanced by Christopher Hill and Keith Thomas that an unprecedented explosion of political literature took place in the early 1640s as a result of the freedom of the press. Neither the end of state control in 1641 nor the restoration of it in 1643 had much effect on the book trade, McKenzie argued, because booksellers continued to pursue profits in familiar ways without concern for changes in the law. Even Milton’s Areopagitica, commonly celebrated as a manifesto for a free press, was not a protest against the licensing act of 1643 but rather a response to harassment concerning his divorce tracts.

When the Revolution of 1688 produced another change in the rules of the game and prepublication censorship ended in 1695, McKenzie again saw the prevalence of continuity and economic interests rather than the triumph of liberty. The Stationers’ Company lost its monopoly of the book trade, but its members continued to dominate the business through combinations known as “congers.” Even authors remained oblivious to changes in the political climate when it came to taking the risk of appearing openly before the public by putting their names on title pages. Only 40 percent of the titles carried the author’s name in 1644 and only 43 percent in 1668. In England as in France, quantification led to revisionist results: long-term socioeconomic trends seemed more important than momentary shifts in politics.

McKenzie was the only bibliographer who could challenge accepted views by working in two registers, enumerative as well as analytical bibliography. But he did not have the last word. He would not have wanted to. Two books published after his death in March 1999 provide a measure of what he had accomplished and of what his work opened up for others to continue. The first, Making Meaning: “Printers of the Mind” and Other Essays, edited by two of his former students, Peter D. McDonald and Michael F. Suarez, S.J., gathers his most important essays in a single volume, arranged artfully by theme and introduced in a way that brings out their originality. They show McKenzie’s mind at work, undoing idées fixes and extracting new ideas from the most intractable material. They also raise the issue of bibliography’s importance beyond the field of textual criticism, where it originated.

The second work, Books and Bibliography: Essays in Commemoration of Don McKenzie, shows how that issue has been addressed by the latest generation of bibliographers and book historians. In it, they chase hares set loose by McKenzie over the last thirty years. They pursue the study of book production deep into the printing shops of the nineteenth century, analyze the interplay of oral and printed means of communication, and investigate the transmission of “texts” in the broadest sense of the term—in music, photography, and architecture. McKenzie taught that bibliography can go beyond books. By following his leads, his successors have shown that it offers a way to understand the reproduction of cultural forms of all kinds, provided that they lend themselves to rigorous description.

Meanwhile, book historians have continued to penetrate more of the mysteries that go back to Gutenberg. In 2000, when they celebrated the six hundredth anniversary of his birth—his putative birth in 1400: we know much less about him than the little we know about Shakespeare—a burst of publications attested to the vitality of bibliographical scholarship. By new techniques of analyzing paper, ink, and type, experts such as Paul Needham, Richard Schwab, and Blaise Agüera y Arcas have transformed our knowledge of how the first printed books were produced. In 1991, the Folger Library produced an exhibition of its treasures, and Peter Blayney explained them in a short book, The First Folio of Shakespeare, which synthesized the most advanced Shakespearean scholarship in language that could be understood by any layman. Bibliography, he showed, had not run out of energy, and it could speak to the general public.

It seems clear in retrospect that the boundary disputes of the 1970s did not damage the discipline and that bibliographers have everything to gain by joining textual editors and book historians in collaborative efforts to break through boundaries. The problems to be solved today extend far beyond the texts of Shakespeare. They appear in communication systems of all varieties, including the Internet, where digitized texts are detached from their moorings in printed books and e-mail messages leave trails that can easily evaporate. Those were the kinds of problems that fascinated Donald McKenzie when he died, too young, in 1999. He did not undermine bibliography, far from it. His heresies have given it new life.

This Issue

May 29, 2003