The battle of the book: the research library today.

In seventeenth- and eighteenth-century England and France a boisterous debate, traditionally known as the "Battle of the Books" raged for many decades. The issue at stake was one of style: should we accept the "Antients" (to use Jonathan Swift's spelling) as our models and exemplars in matters literary, given their immemorial legacy of acutely expressive prose and verse, or should we rather forge a "Modern" style and manner befitting our own age and its peculiar requirements and contingencies? Charles Perrault in France in the 1695 preface to his Contes sided resolutely with the moderns, and this on moral grounds: the ancient fables taught a destructive morality. Interestingly enough, he singled out the pernicious effects of certain misogynistic classical tales on young girls' moral nature and declared: "I maintain that my fables deserve more to be related than most of the ancient tales ... if one considers them from the moral aspect."

Similar debates took place at other times and in other cultures. In ninth-century Baghdad, to name but one, poets argued strenuously over whether it was better to ape the style of those earlier desert-dwelling bards who had made the original glory of literary Arabic or to forge an idiom and manner reflective of the overly refined and courtly world in which the poets actually lived. The scurrilous wag (and great poet) Abu Nuwas went so far as to lampoon the early poets and to state that the only thing he himself sought in the ancient ruins was a good stiff drink.

Today, it seems, we are confronted with a new battle of the books. Ours, however, is not between two competing and irreconcilable types of book but between the book itself and its would-be surrogates, whether these latter take the form of CD-ROMs, video disks, digital encryptions, or indeed, formats and media not yet invented or even imagined. It is thus a battle of the book itself, rather than merely of one type of book against another. And in no institution has this battle been waged more confusingly or more protractedly over the last twenty-five years than in that supposed asylum pacis, or "haven of peace," the epithet which the great German scholar and library director Adolf von Harnack once gave the research library.

In a certain real sense, of course, this is a phony war. Books and computers work well together and have proved complementary, even symbiotic, on numerous ventures (Is publishing itself even conceivable now without automation?). But each format has come to stand for something in the minds of its adherents: if not a style, then a stance. For the book lover, it is the affection and reverence for tradition coupled with the conviction that the book as a medium is essentially unimprovable. (I should alert the reader that I share this conviction, even though I presided over the implementation of two large automated systems in libraries where I was the director.) For the advocate of automation, by contrast, tradition is itself the problem; there are bold and innovative electronic ways of building collections and of running libraries. The computer holds out the promise of resolving many of the old intractable problems: lack of space, deterioration of paper, the cumbersomeness of making bibliographic and intellectual connections using printed sources alone. The zealous computer fanatic sees the book lover as troglodytic; the staunch book lover regards the computer fanatic as barbaric. As you might suspect, both sides are right and both sides are wrong.

In developing his fable about the Library of Babel, Jorge Luis Borges noted that the library, which was a model of the universe, was infinite and unending. So, too, it often seems, are the problems underlying the history of research libraries over the past few decades. In what follows, therefore, I use a broad brush, all the while recognizing that the topics raised could each demand many pages for a full and nuanced treatment.

To understand this period of almost incessant change, two factors must be recognized. First, until quite recently, this has been an era of unprecedented budgetary crisis for libraries (and often for the universities that support them), with crunches and squeezes and freezes and clawbacks--the terminology alone is terrifying--of varying intensity affecting nearly every fiscal year since the affluent 1960s, and accompanied by staggering increases in the prices for both books and periodicals. And second, this has been the period in which, not by coincidence, automation began to be introduced swiftly--often, all too swiftly--into libraries, initially as a cost-saving strategem and then, increasingly, as an alternative to the long-accepted but expensive standard practices and services, such as original cataloging or collection-building by means of professional bibliographers, among other possible examples.

It is not easy to summarize this period dispassionately but all observers would agree, I think, that the effects of both of these factors have been far-reaching, occasionally destructive and quite often traumatic. The trouble with budget cuts is not solely that they restrict growth but also that they open up opportunities for crisis managers, who for the most part have no stake in, or love for, the libraries they administer but grasp crises as opportune moments for self-promotion or the promotion of some not-so-hidden agenda. And the trouble with automation is not that it is itself unsuitable or deleterious to research libraries but that it has furnished an irresistible pretext for sweeping change, often for its own sake rather than for the sake of the libraries or those researchers who depend on them.

In The Battle of the Books, Jonathan Swift could speak, rather disingenuously, of "the publick Peace of Libraries." Whatever mood may reign in libraries, it is fair to say that over the past few decades, peace, public or private, has not been predominant. Rather, a stubborn engagement, with numerous skirmishes, several protracted sieges, and not a few premature cries of victory, has swirled around books and libraries even though the precise issues have often been unclear, even to--perhaps especially to--the participants. In all this, the book has not merely survived but has flourished even as myriad forms of automation, like upstart pretenders to the epistemological throne, have been touted and installed, sometimes with the quite blatant purpose of supplanting the book and all it represents. Millions upon millions of dollars have been expended, hundreds of careers have been embroiled (and not a few destroyed), and yet any final resolution remains in doubt. If we can indeed call this a "battle of the books" perhaps it is more of a brawl, a brawl carried out in slow motion and under cover of a rather ferocious gentility (only a seeming paradox in the world of librarianship). In fact, we have more books than ever (though whether libraries can keep up with the flow remains a troubling question), and we have ever more sophisticated, and ever costlier, automated systems, ostensibly to enable us to retrieve the "information" in those books.

In reality, alas, information has become its own raison d'etre with the book considered, by surprisingly many members of the profession, as little more than a clumsy, outmoded and luxurious obstacle to that quest. Small wonder that in my own thirteen years as a research library director, though I came to know many well-read and cultured directors, I never heard the book referred to at professional meetings, such as the biannual gatherings of the influential Association for Research Libraries, as anything but a kind of parasitic encumbrance, one that ate up space, generated noxious dust, deteriorated in various unseemly ways, and of which there yet never seemed to be enough to satisfy the ravening appetites of both students and faculty.

Formal public expressions of this are of course circumspect. In a statement of the priorities for the year 2000 of the 121 top-ranked American and Canadian libraries that are members of the Association of Research Libraries, we read of the need to "help research libraries and their constituencies move into a transformed and increasingly diverse environment through the development of human resources, programs, and services." Is this some sort of code or just terminal fuzziness? Judging from my own experience I would say that the key phrase is "a transformed and increasingly diverse environment." These weasel words mean almost nothing, to be sure, but to me at least the melange of smarmy condescension and prissy buzz-words ("transformed" "increasingly diverse") suggests that these prestigious libraries are girding themselves for yet more disruptive change. Note too that the one thing the "constituencies" (those pesky faculty, students, and other researchers) will not be offered is larger or better book collections, but only "human resources, services, and programs." No doubt the "human resources" will take the form not of dowdy old librarians but of "cybrarians"--a new and ghastly coinage--which are not some race of alien mutants but librarians with the libri removed.

The next ARL priority pretty much gives the game away: "Ensure that research and learning will flourish through the development of advanced networking applications and Internet." This insurance, be it noted, is to be accorded not through the provision of books, journals, documents, microfilm, or even databases and other online resources, but rather through "advanced networking applications and Internet." This claptrap too is code; the sanctimonious phrase "research and learning" is intended to divert attention from the alternatives being recommended. Translated, this means that when your "constituencies" (librarians used to call them "users" so maybe this is an improvement) grumble that you're not purchasing the books and periodicals they need, offer them the smokescreen panacea of "networking." In other words, less misleadingly put: send them anywhere else but here.

At the other extreme of this brawl of the book have been those bug-eyed and splenetic observers of the library scene, such as Nicholson Baker, whose recent onslaught (in his book The Double Fold) on conservation practices in research libraries has stirred so much acrimony. One reason, of course, why Baker is viewed quite dimly by the library establishment is that he, or his type, has long been familiar: the aggrieved and expostulating professor who appears one day, frothing with rage, in the librarian's office waving some discarded tome and demanding to know how on earth so indispensable a title could have been discarded (or "de-accessioned" as librarians coyly put it). There is probably no book, pamphlet, leaflet, magazine, newspaper, comic book, or indeed, printed object of any sort, that does not have some research value, and in this, at least, the raving professor is right. I have known one librarian-collector of ephemera who dutifully amassed hoards of subway ticket stubs, empty shampoo bottles and even designer sunglasses because all of these objects bore print on their surfaces. But libraries are institutions in which a principle of selection should reign, and this for very sound epistemological rather than pragmatic reasons. Selection, when carried out skillfully in a research library, is an art as well as a science, and it is this art, along with the collections it can shape, that we are in danger of losing in the current "battle," if indeed we have not already lost it. Much that Baker complains of is, of course, all too true, but the trouble is that he should have been proclaiming it twenty-five years ago when it could have made a difference and when many librarians themselves might have welcomed such an ally. His fulminations are just too late.

One of the most troubling aspects of our battle of the books as it has been conducted, mostly behind the scenes, in our major university libraries is that there exists so little common ground between proponents of the traditional book with all its offshoots and the self-appointed apostles of a new approach. Swift could declare of his own conflict that "the present Quarrel is so enflamed by the warm Heads of either Faction, and the Pretensions somewhere or other so exorbitant, as not to admit the least Overtures of Accommodation." Worse still, few indeed have been those participants in our own battle who have had the courage or the foresight or the plain common sense to proclaim that the conflict is probably moot: indeed, we and our libraries--and by extension, that means our culture, and the survival of culture--require urgently all modes of knowledge, for in the end they are complementary, not antithetical. The invention of printing did not obviate the need for manuscripts or indeed for papyrus and palm-leaf scrolls and cuneiform tablets, all of which libraries still collect, or should be collecting.

One of my own frustrations as a library director lay in the difficulty, if not the patent impossibility, of persuading one side or the other that both books and computers were essential. On occasion, this took farcical turns. University administrators routinely cherish the plaintive hope that automation will somehow, in the near future, obviate the need for books, libraries, and traditional library buildings, the costs of which increase exponentially year after year (it is a rule of thumb that book prices rise by a year or more ahead of the Consumer Price Index).

Researchers and students are, however, usually just as desperate to obtain print materials in addition to online resources; those who use the sources, especially in the sciences, often find both print and online formats necessary since they consult them at different times for different purposes; the online version in a lab in the heat of an experiment and the print version for writing up results. In the midst of such troubles, I found myself lunching with a university vice president who had been effectively blocking the construction of a new special library for over two years because, as he put it, "books will be extinct in five years and so, why construct a new building to house them?" Suddenly, between the soup and the main course, he lost all patience and exclaimed, "Yes, yes, I know you book lovers all too well! The book is a sexual object for you. Think about it: those creamy white pages you love to spread wide, like a woman's thighs, the textures of the paper, even the smell of a fresh book ...! It is classically Freudian! Oh yes, I have you pegged!" The fact that my interlocutor was French-born and still possessed a Chevalieresque accent gave his words incredibly lurid emphasis.

Startled by this outburst, I yet had to admit upon later reflection that he had a point. A book is intensely physical, even sensuous. Reading is not a disincarnate, cerebral activity but a solidly physical process in which we deploy almost all our senses, and no doubt a Freudian pleasure-principle is at work while we read. After all, we are at least subliminally conscious of the weight of the book in our hands, the design and layout of the pages engage our eye, the typeface is pleasing or annoying or diverting, marginalia or underscorings may arrest our attention, we can smell the ink at times and graze the texture of the pages, the binding, the dust jacket with our fingertips. In her memoir One Writer's Beginnings, Eudora Welty put it this way: "I cannot remember a time when I was not in love with them [books]--with the books themselves, cover and binding and the paper they were printed on, with their smell and their weight and with their possession in my arms, captured and carried off to myself."

When we open an old book, or one we read in childhood, the scent of the past that rises up can bring back whole Proustian realms in its gust. By contrast, nothing could be less sensuous than a computer monitor with that cool filtering glow that emanates from it; even the touch of fingers on the keyboard is abstract compared to the feel of a cover or of pages. I do not write this as a nostalgic paean to the book as such (though I see nothing wrong in that), but I do wonder whether the knowledge gained from books may stick with us longer in part because it comes wreathed in these sensuous associations. In any case, the senses, if not the attention (try to distract a teenager from a computer screen!), seem more fully engaged by a book than by a monitor, however seductively the latter may twitter.

No doubt I am biased, but it strikes me that a covert complicity exists between book and reader that does not obtain between computer and user. Reading a book becomes an experience in one's life in a way that consulting a computer cannot be (or, at least, cannot be yet). The computer is unsurpassable for the transmission of facts, of raw information, as well as for its miraculous indexing properties, but it does not--again, perhaps, does not yet--engage our imaginations and intellects in quite the same way a book does. I believe that this has to do with the serial and sequential nature of a book, its succession of pages in linear alignment, as opposed to the scroll-like nature of the computer screen. There is a reason, after all, why the codex superseded the scroll; there is an affinity between the way in which information is presented in a book and the way in which we learn that has no analogue in the world of automation. As far as libraries are concerned, the most visible effect of this change, of this reversion to the scroll that computers represent, has been the relinquishment of the library card catalog. This too has been the subject of a much-discussed and rather silly essay by Nicholson Baker; however piquant Baker's complaints, he somehow missed the entire point of what the shift from card to online record betokened, for better or for worse.

The modern library, as it developed, especially in North America, over the past one hundred years or so, depended above all upon a precise and rather elaborate classification of human knowledge. The best-known of these systems of classification is probably the Dewey Decimal System, devised by Melvil Dewey around the end of the nineteenth century and which depended upon decimal increments to parcel knowledge by subjects. The system was much hated by school kids (myself among them) because it compelled you to use the card catalog, at least until you had a good enough grasp of the classifications to browse. Dewey's system rested upon an ancient conception: knowledge is not only classifiable but hierarchical. Some types of knowledge are more general (the domain of reference books such as encyclopaedias, gazeteers, atlases, etc.) and so occupy the first class. Thence we proceed from the general to the more and more particular along a scale of increasing precision and ever finer distinctions.

The point of all this is to say that in order to employ such a system as the Dewey with the utmost effectiveness, you have to have something of a general, if vague, understanding of a certain taxonomy of knowledge; there is an order between, and within, classifications and a strict logic obtains. Books (except fiction) are ordered by subject, not by author. If a given writer has composed a treatise on geology, it will be in a different classification from his or her slim pamphlet of lyrics. To be sure, this is but one principle of classification; books could be classified by author, as in the old Richardson system, or by other means. From the reader's viewpoint what obviously matters is being able to find the book. And yet, in using Dewey, one is necessarily reminded that a given book falls into one category and not another. Dewey and its analogues are, if you will, vertical principles of arrangement, they are arborescent systems from which sprigs and runners and branches are continually unfurling, at predetermined points along the trunk and nowhere else. The more elaborate certain sub-disciplines become, the more vertiginous the classifications and sub-classifications themselves become with call numbers spiralling into ever longer and unwieldier combinations. (Imagine the difficulty of classifying particle physics or quantum mechanics in a system developed in what was still a Newtonian world!)

Though Dewey, in its various revisions and permutations, is used worldwide and is probably the most successful and practical classification scheme yet devised, with that of the Library of Congress not far behind, nowadays all such systems have been rendered irrelevant and even otiose, at least for the average reader, by the advent of automation. The systems are still in use--books have to be analyzed and classified and shelved somehow and librarians are inveterate systematizers--but they have been effectively gutted of their epistemological structures. What librarians term "subject access" is no longer crucial in carrying out research in a library and the old, massive, crimson-buckram-bound volumes of the Library of Congress index of subject headings, a concordance of Babel on an infinite Borgesian scale, is now more of a curiosity than anything else. Once it became feasible to conduct Boolean searches or simply to employ keywords, the entire edifice of subject classification and taxonomy began to totter. The old stuffy hierarchical categories have gone the way of the Great Auk, and anyone can now find something in the library as long as he has even the dimmest idea of what he is looking for. It would be intolerably elitist, I suppose, to lament this apparent dumbing down of the library. As opposed to the old vertical and hierarchical way of articulating knowledge, the new keyword approach is resolutely horizontal and "egalitarian"; not only can anyone use it, practically anything is connected to everything else. The notion of moving from the general or universal to the particular is made ridiculous. Who needs logic when you can enjoy random access?

It would be foolish to deny the usefulness of keyword searches and their analogues. At the same time, I have observed over the last two decades that research in libraries, among undergraduate and graduate students, is becoming shallower and shallower. If a title or an author is not instantaneously located in a database or an online catalog, the assumption is made (and is quite hard to shake) that the title or author does not exist. Gone are the days of tedious and often fruitless hours spent toiling through bibliographies and bibliographies of bibliographies for a single nugget of fact. Fair enough, and I will not lament their passing. Nor am I arguing that knowledge should be made difficult to obtain, far from it. But, for better or worse, true knowledge is by its very nature hard to get; indeed, it is not just hard to get but tough, knotty, obdurate, resistant, and even harder to hold on to than to get, and we do no one a favor by pretending otherwise. In their tiresome stress on information over the last two decades, librarians have tended to place too great an importance on what should in fact be the beginning of a search for knowledge--the mere fact, the datum--and not an end in itself. A facility in surfing the internet is no substitute for the struggle to understand. If genuine learning were available at the stroke of a keyboard, by now we would all be Leonardos and Einsteins. The computer is good not because it makes learning simpler (it does not). It is good because it is fast and because it enables us to tackle tasks and challenges that are otherwise beyond our computational abilities. It is good because through computers we can address the incredible complexity of problems that before we could not even have approached. And in libraries the computer is especially good because it permits almost instantaneous connections to be made between an infinitude of possible subjects. It articulates what would otherwise be Ovid's rudis indigestaque moles, a "rough disorderly lump" of chaotic data.

This seemingly slight shift in the way we now seek for information in libraries--briskly and rapidly with apparently infinite "connectivity," as opposed to the old, laborious drudgery of toiling through card-files and reference shelves--has itself brought consequences, both good and bad, in its train and has affected not only the way in which research libraries function but also the very profession of librarian. "Information technicians" and re-tooled reference librarians may have thrived but at least two once-essential disciplines, that of cataloger and that of bibliographer, have been weakened, if not rendered obsolete. This has been one inconspicuous but destructive side effect of automation.

The bibliographer selects books; the cataloger puts them in logical and retrievable order. This sounds simple but is not. Incidentally, of all library specializations, these two are probably the most ancient, dating back to Ur of the Chaldees, if not beyond. Of course, catalogers are invariably (and often rightly) lampooned as fussbudgets and hairsplitters of the most exasperating sort, but then it is their nobly picayune craft to discern ever finer and more accurate footholds along the slippery ladder of classification. And cataloging, if not much honored today, can number among its distinguished practitioners many illustrious names, including the fourth-century B.C. poet and scholar Callimachus whose fabled Pinakes (literally "tablets") brought order to the immense holdings of the great library at Alexandria. In more recent times, Leibniz, Lessing, and Strindberg, as well as Philip Larkin (and Mao Zedong!), to name but these, served as librarians and devoted themselves to the creation of catalogs; and think of the prestige of lists and inventories in the works of Jorge Luis Borges, another librarian by profession. (To be sure, it is reported that Callimachus chafed at his post and was perhaps bitter because Ptolemy II never named him director; and Lessing quit the profession in disgust because he was sick and tired, he said, of acting both as the watchdog which guards the hay it cannot eat as well as the stable boy who must provide the cows with fodder on demand.)

As for the bibliographer, in bygone days most often a scholar-librarian with academic credentials the equal of any professor, he or she has been rechristened the "collection development" librarian or "subject specialist" and is expected to "network" and elaborate ever goofier and more unworkable schemes of "resource sharing." Most meretricious of these, in the turbulent quarter-century just past, was surely the ill-fated "Conspectus" aggressively promoted for years, over the objections of many bibliographers, by both the Research Libraries Group and the Association of Research Libraries. The Conspectus was a cockamamie project by which libraries graded their own research collections, on a scale of one to five, for the purpose of facilitating a purely mythical "shared access"; since the Conspectus inevitably entailed highly subjective and not disinterested self-assessments, it finally collapsed under its own absurdity, but not before having frittered away millions of dollars that might have been spent on enlarging and improving the very collections libraries intended to share. (One flaw in the process was illustrated for me by an exchange with a colleague at the New York Public Library who after having read my ranking of Princeton's Persian holdings declared: "Since Princeton is so good in Persian, we can stop collecting it.")

Forgotten in all these grandiose (and hugely wasteful) sideshows is the fact that the truly wonderful and incomparable American collections, those, say, at Harvard's Widener Library or the New York Public Library or at Yale, Toronto, Indiana, Berkeley, Chicago, Princeton, and on and on, were in fact formed by rampant and often highly individualistic curators and bibliographers who were as much swashbuckling buccaneers in the book business as men and women of learning and scholarly shrewdness. (Their prototype, if not their patron saint, is J. P. Morgan whose bibliographic acumen coupled with business ruthlessness created one of the glories of the world of libraries.) It goes without saying, of course, that few library administrators look with fondness on such potentially maverick staff members. Enter the computer. By holding out the rosy but specious possibility of "shared access" automation also offered a marvelous way of eliminating or "phasing out" such professionals under the snappy banner of "cost effectiveness."

Again it is not nostalgia that prompts me to lament the decline of the bibliographer and the original cataloger in research libraries. Both disciplines are as much needed now as ever before. Why? The original cataloger (that is, the cataloger who does not draw on the cataloging information provided by the Library of Congress or other libraries) works with rare, unusual, often esoteric material, e.g., Rabbinic commentaries on the Talmud with all their numerous super-commentaries, glosses and super-glosses, marginalia and excurses, etc. To catalog books of this sort may require hours or even days; all the authors, many of them obscure, must be traced and the authoritative versions of their names established, the relation of one commentary to a text or a gloss to a commentary must be teased out and elaborated, and so forth. But the result is a minute, almost microtomic dissection of a work and its intellectual and physical history that is of incalculable help to scholars. Naturally, such a specialty is both time-consuming and costly, and yet, is not this, and exactly this, what a research library should be providing? Who else will do this? True, the so-called "brief cataloging" that is now in favor does at least give researchers some grasp on complex collections, but it cannot be an adequate substitute for the full analysis and description intricate items demand.

As for the bibliographer, he not only scouts out and orders books, if he is any good, he also has a comprehensive grasp of the collection. He knows its lacunae; he has a long and detailed list of desiderata; he works to anticipate shifts in a scholarly field and to provide for them; above all, he strives to form a collection that is at once internally consistent and articulated: what may to the casual glance appear whimsical or superfluous will yet have its intrinsic place within the collection, and for good reason. Unlike the cataloger's work, however, that of the bibliographer is not easily quantified. Libraries must accept with a certain degree of trust that over time a good bibliographer will succeed in creating a rich and deep collection, much as universities trust that the professors to whom they grant tenure will eventually produce work of substance and significance. Such intangible covenants count for little in the library of today, and more's the pity.

Furthermore, not only have many of the most distinguished American bibliographers and curators literally been hounded out of their positions (yes, at the same Harvards, Princetons, and Yales, whose collections they toiled to build), but the acquisition of books and journals has been widely relegated to automated "approval plans." Under such plans, believe it or not, it is the bookdealers themselves who select and often catalog the very books which they sell to libraries. Multinational bookdealers now routinely create "collection profiles" for research libraries and then proceed to ship books as they, the bookdealers, see fit.

Libraries have the right to return unwanted books but rarely do; it is simply too much bother. Despite the obvious risk of conflict of interest, these plans are not all bad since certain imprints, such as titles from university presses, are almost invariably acquired by libraries; why waste the time of a highly paid bibliographer on such routine purchases? And yet, the titles which fall through the cracks of the plans can only be retrieved by those very specialists who are now more and more being replaced by rank-and-file librarians. If this is bad enough in the realm of English-language publishing, imagine how it affects foreign and special collections for which subject expertise is indispensable. And the damage is usually irreparable; all the titles the library should have been acquiring when it was buying blindly quickly go out-of-print and become unobtainable; the special collection is diminished, usually forever.

Needless to say, this does not form any part of the rosy view of many university officials or library directors, most of whom yodel from their particular fence posts the latest eudemonistic cock-cry of yet another false dawn. The administrators have been more to blame than rank-and-file librarians, most of whom are committed to an ideal of service, to provide whatever readers need. But library "managers" have almost universally seized upon, and pushed, automation as a magical cost-saving expedient.

In one sense, this is understandable. Over the past twenty-five years, book and journal prices have soared to previously unimagined levels. I have my own bench-mark for the surge in book prices: in 1971, almost overnight, the price of E. G. Browne's Literary History of Persia--published by Cambridge University Press and for over fifty years around $15 a volume--abruptly quadrupled and made havoc of my meager graduate student book budget. But even if we look at the very recent past, we can see the increases that have devastated acquisitions budgets. A law book which in 1984 cost $44, cost $82 in 1998; a book on literature cost an average of $24 in 1984, but $44 in 1998. (I should point out that there has been some abatement of book prices in the last two years but it is probably too soon to know whether this trend will continue.)

And journal prices, especially for scientific periodicals, have risen to truly obscene levels. A chemistry journal which in 1984 averaged $228.90 per year (a hefty enough price, one would have thought!) rose to $1,302.00 in the year 2000; a journal in mathematics which cost $107.00 annually in 1984 had quintupled to cost $517.00 in 2000. (In Europe the situation became even worse: the average German chemistry journal, for example, experienced a six-fold increase for the same period.) The history of scientific journal pricing and its horrendous rise over the last twenty years deserves discussion in its own right. Suffice it to say here that one of the chief pioneers of punitive price increases in academic journals--a particularly cruel and absurd cycle since publishers effectively sell back to university libraries at exorbitant prices the research which their own professors have produced--was none other than the late, unlamented Robert Maxwell, the self-styled "bouncing Czech." His progeny, particularly in England and Holland, continue to thrive at the expense of libraries.

Beginning in the 1970S, such constraints forced library directors and their bosses to cast about desperately for budgetary relief. The computer seemed the ideal answer. (It did not immediately become apparent that computers were, if anything, even costlier in the long term than books and journals.) But the largest irony of our own Battle of the Book is not that certain forceful advocates preferred computers to books for intrinsic reasons--in truth, they could care less--but that with the cynicism of panic, they latched on to that medium, whatever it might be, which looked cheapest, glitziest, and most "cost effective." If the book had been shown to be more cost-effective than the computer--with hindsight a not untenable position!--such administrators would have been baying for more books and journals. As it was, opportunists soon abounded on all campuses, bringing in their wakes not only blatant hucksterism--of the sort we thought had gone out with the snake-oil salesmen of the previous century--but also all varieties of new, and not-so-new, "philosophies of scientific management" and strategic planning. With the advent of the computer, and its ability to provide minute statistical data on operations, came the hard-headed, clear-eyed, unsentimental, bottom-line purveyors of the MBA world-view, and universities--and their libraries--have not been the same since. Gentlemanly library directors (and they were all men), accustomed to submitting elegantly drafted and wryly understated annual reports, were now being asked to draw up "business plans" for their operations. Efficiency experts, systems analysts, and high-priced consultants were not far behind. In all this, perhaps the greatest harm occurred because the research library was simplistically likened to any other large and complex organization when in fact its distinctive operations, requirements, and, yes, institutional culture--in short, all those factors that had made our great libraries great--were utterly at variance with those of a factory or a corporation.

Many librarians accepted the changes with relief, others were dismayed. With the advantage of two decades of hindsight, one can say that the dismay was perhaps not exaggerated. Of course, libraries have been unusually resistant to change, and with justice: an institution and a profession whose sacred duty it has always been to safeguard and make available the records of our past should be conservative, indeed, should be exceedingly cautious, about change. Even so innocuous an innovation as microfiche or microfilm (pace Baker), as well as the unprecedented position of "microfilm librarian" ("fiche-wives" as my colleague Orest Pelech of the Duke University Library promptly dubbed them), originally seemed alarming to librarians who feared, without justification, as it turned out, that this new format would replace the book. My own first supervisor and mentor in a research library harbored such anxieties. An immensely learned scholar and bibliophile, he once led me on an excursion through the darkest and most remote regions of the stacks. As I followed in his wake, I noticed that he was mumbling something indistinct but fervent. As I came closer, I saw that he was stroking the spines of the shelved books, and I heard him muttering, "I will never let them turn you into microfilm, my darlings!"

Of course, there is no question of a "return to the book" or any other such quasi-Luddite approach. Not only would that be ill-conceived, it would also be superfluous, for the book is thriving as perhaps never before. (According to The Bowker Annual, that bible of the library profession, some 100,405 tides were published in 1999 in the United States alone. And on the authority of Dr. Knud Dorn, of the firm of Otto Harrassowitz Buchhandlung in Wiesbaden, I am told that book-publishing in English has increased enormously in recent years in Europe, probably because of the EU.) This is not parallel to, or in spite of, advances in automation but because of it, for printing technology has been revolutionized like everything else and the robust state of publishing, at least in terms of production, owes much to automation.

At best the computer and the book are symbiotic. What the computer does well--sorting, indexing, linking, retrieving, etc.--no book can equal but what the book can do--proffer knowledge as a vital experience--computers mimic only very imperfectly. Among many possible examples, let me give two which represent for me the opportunities computers offer to research libraries; one is drawn

from my own field of research, the other from my experience as a library consultant.

In Islamic intellectual history, my particular discipline, manuscripts form a core around which entire subsidiary literatures have developed; usually these consist (as in the Rabbinic examples I noted above) of school texts with their ancillary commentaries and glosses, but they may also contain counter-commentaries or even wholly original treatises inspired by a work. In short, a given text may have spawned dozens or even hundreds of subsidiary texts which themselves often span centuries; an author who wrote in the twelfth century may find a glossator today in the madrasahs of Qum or Fez. To amass and compile all of these texts in printed or handwritten or microfilmed form is a daunting job and to collate them is even harder. But through such devices as hypertext, it should be possible not only to convert such writings to electronic form but to index them, so that at the touch of a finger, all the relevant explications on a given theme or problem, from any century whatever, could be summoned to the computer screen. This is in fact happening; the Qur'an is available on CD-ROM together with two of its standard commentaries. Rather than rummaging through several densely printed volumes, one now can call up a definition or an explanation instantaneously. What is possible for Islamic texts is possible for those in other fields: the creation of a stratified palimpsest of related writings which are mutually transparent. Imagine such a simultaneous array of the different manuscripts of Ulysses or King Lear.

For the second example, when I worked as a consultant in Morocco, I found that the quickest and best way of building new library collections from scratch was to acquire materials in an online format. Automation has in effect made it possible for Third World institutions to replicate the laboriously assembled collections of great Western libraries, and by this I mean not only their catalogs but much of their holdings as well. Newspaper files, government documents, scientific databases, huge electronically indexed collections of poetry, fiction, and history in several languages are available and more or less affordable. Here the approval plans of various bookdealers, used as a cost-cutting expedient in the West, proved indispensable for the acquisition of printed books and journals. Such "instant" libraries are springing up in the wealthier countries of the Arab world. Of course, they lack the depth of the rich old collections but at least they have a viable foundation on which to build. Moreover, many such collections are available gratis. More and more libraries are offering the full texts of rare and unusual resources. At McGill University Library, for example, it is now child's play to access and download special collections as disparate as those on the North West fur trade or the history of the treatment of tuberculosis.

Research libraries in their inherent conservatism have always striven for balance and it seems to me that in these, and other ventures, a new equipoise is taking shape. Both the Antients and the Moderns represent extremes and extremes are antithetical both to libraries and to librarians, who have a fine-grained instinct for equilibrium. It behooves librarians, who are the guardians of collective memory, to look back into their own history for exemplars.

As one example, almost a century ago, in 1904, a figure emerged in the world of librarianship who is now forgotten but whose practice and thinking, in my own opinion, provide pointers for the future. Adolf von Harnack was the keeper of the Royal Library in Berlin from 1904 to 1921 and a renowned historian of early Christianity. Like our own two most recent Librarians of Congress, von Harnack came to his director's post not from the ranks of librarians but from the professoriate (and was heartily resented, at least at first, by the profession, as were both Daniel Boorstin and James Billington). This is worth mentioning because von Harnack administered the Royal Library with a scholar's eye; from long training and practice, he knew exactly what a great research collection and its services should be, and he set out to create them. To this end, he not only expanded and improved the holdings of the library but refined and enhanced the positions of the librarians who were responsible for its continuance. As it happens, von Harnack wrote frequently and well on his new position and its entailments. During his fifteen years in the directorship his view of scholarly collections evolved and became quite subtle, particularly so when juxtaposed to our present "approval plan" approach. Here is how von Harnack enunciated his guiding principle of selection in his 1911 work Aus Wissenschaft und Leben:

   Only what is used or at least has a prospect of use should be conserved.
   The librarian's task in large libraries has become for this reason much
   more difficult, but also much more significant. The superfluous is always
   harmful. This is true for large libraries as well. On the other hand,
   however, it is true to say that even the least significant item is valuable
   as a mass phenomenon.

As a consequence of this principle and its inherent tension--there is no facile role or formula--von Harnack argued that the librarian must resist simplistic solutions. When the Director of the University Library in Graz, one Ferdinand Eichler, argued that libraries should concentrate on acquiring only "the best" books, von Harnack argued, with considerable elan, that this was nonsense. The purpose of a research collection, he countered, was not to offer the best or the right ideas, whatever those might be, but all those writings, right or wrong, wise or stupid, that bear on a given subject. The formation of research collections will always therefore be something of an art, in which scholarship, experience, and sheer intuition play essential roles. Von Harnack's conception of the library is at once utopian and down to earth. In a 1923 article he wrote:

   From the time of Leibniz until the end of the 18th century, there existed
   in the domain of science, literary culture and social life a single
   educated type and a "private company" (societe anonyme), to which the most
   varied spirits belonged. It encompassed Bossuet as well as d'Alembert,
   Leibniz as well as Kant, Newton as well as Voltaire--all those, indeed, who
   understood how to present their insights and points of view with wit and
   style; who knew how to listen to an opponent and to present him with a
   well-founded response from their own position. This "company" no longer
   exists, but it does survive in libraries and must continue to survive
   there. An eclecticism open to all sides must reign there, as it did in the
   period of the Enlightenment.

Quaint as such a vision may sound, I believe that it remains valid. A research library must be as capacious as is possible but it must become so by enlightened and balanced principles of selection. Today surely this implies the fullest deployment of all means at our disposal; in other words, not this format or medium as opposed to that, but all pertinent materials, whether they be found on the printed page or in digital form or on bamboo scrolls and cylinder seals. In a fascination with the means, librarians have lately tended to ignore and neglect the ends. But the purpose of a library is not to cut costs or to showcase new gadgetry or to serve as a mausoleum for books fallen into desuetude; rather, it is to make possible that instant of insight when all the facts come together in the shape of new knowledge. It has always struck me as paradoxical that so many tons of concrete, so many miles of shelves and wires and circuitry as even a modest library contains, are required to prompt that most evanescent of human experiences which a new understanding entails. The poet Paul Claudel wrote that for the flight of a single butterfly, "the entire sky is necessary" and so it is as well, I think, in the empyrean of knowledge.

As the motto for the Royal Library, von Harnack took the Latin tag Biblioteca docet, "the library teaches" By this he meant not that the library staff offer evening classes but that the library itself make a genuine experience of knowledge continually possible; as a historian he knew that past, present, and future are not merely sequentially linked but interlaced by myriad crisscrossing strands.

What strand in all this belongs properly to the librarian? For von Harnack--and I think that he was right, however archaic his vision may appear--it is still the book itself:

   the literary document as such, as the bearer of literature and science, the
   book with its general and yet highly particular natural history from
   inception to binding, the book with its provenance and distribution, the
   book as an object to be collected, because one book is no book (ein Buch
   ist kein Buch).

Librarianship, when it is "vivified by love of books, is the sum total of knowledge about the library and the book in themselves (an sich)." All of von Harnack's musings on libraries are tinged with his own religious sentiment. For this reason he can go so far as to term the librarian a "minister of written and printed words" (minister verbi scripti et impressi). And if it is indeed the word, whether it appear on a printed page or a computer screen, that is important, this definition still holds good.

In The Battle of the Books, Jonathan Swift indulged the fancy that each book somehow preserved the fighting spirit of its author and that libraries were really cemeteries haunted by the spirits of authors who continued to do battle and oust each other:

   So, we may say, a restless Spirit haunts every Book, till Dust or Worms
   have seized upon it; which to some, may happen in a few Days, but to
   others, later; And therefore, Books of Controversy, being of all others,
   haunted by the most disorderly Spirits, have always been confined in a
   separate Lodge from the rest; and for fear of mutual violence against each
   other, it was thought Prudent by our Ancestors, to bind them to the Peace
   with strong Iron Chains.

Perhaps, undetected by our grosser senses, some such bellicose and divisive spirits hover over our computer terminals and urge the susceptible to engage in new "battles of the books" from which either the printed page or the monitor will finally emerge triumphant. I hope not. If the past twenty-five years have proved anything, it is that, for the survival of culture, we need all the help we can get, whether in words baked on ancient tablets, set in cold type, or amid the pixels of the scanner and the computer screen.

Eric Ormsby was the Director of Libraries at McGill University from 1986 to 1996. His latest book of poems, Araby, is available from Vehicule Press.