801.95/DIL

In October 1995, when the dog days had stretched into a sleepy autumn distinctly unconducive to lucid thought and scholarly application, I fetched up at a provincial university in the south-east of England, hoping to complete, in a year's exile from my home town's distractions, a project that had already, over three years, taxed my intellectual and financial resources to the point of exhaustion. The initial signs were good: the campus architecture was sufficiently grim to suggest hours of austere concentration, long evenings spent brooding a mile or so above a darkening cathedral town. More encouraging still was the library: squat and ugly, like all the best libraries seem to be; its modest collection would at least, I thought, limit the ragged scope of my research.

I wasn't prepared, though, for the giddy horror of my first visit to the library. I went looking for a few beloved books, just to orient myself, and discovered, with a sense of sickening vertigo, that the library's classification system made no sense to me. Its numbers were all wrong; shelves were inscribed with gnomic combinations of letters where there ought to have been familiar three-digit markers. The discovery hit me hard; in the following weeks, as my planned research started to unravel around me, I imagined it was all the fault of this freakish new code (it turned out to be the same system used by the Library of Congress). I began to fantasize about libraries I knew well, where the books (in my stricken, homesick imagination) leapt smartly into my hands from just the right, precisely numbered, place. I began to dream, of all things, about the Dewey Decimal Classification (DDC).

Things got worse, and instead of working I read Jorge Luis Borges, blind Director of the National Library in Buenos Aires (in my mind's eye I saw him quietly shelving his own books at 868.998 in accordance with the DDC). Borges was obsessed by the arbitrariness of efforts to classify knowledge, conjuring up, in his story 'The Library of Babel' (1941), the image of an endless and cyclical library - a horrifying vision of every actual and possible book, haphazardly shelved to infinity. Borges' narrator imagines that somewhere in this maze there must be one book that condenses the chaos into a system, a catalogue that would transform the babble of knowledge into something ordered and serene. But 'the catalogue of catalogues' exists only as an intolerable fantasy; Borges' story suggests that libraries are places of both lucidity and lunacy.

The classification system that I recalled so fondly from my early student days was a more modest invention than Borges' imagined single-volume key to the library's universe, though in some ways just as peculiar. Melvil Dewey created the DDC while a student and librarian at Amherst College, Massachusetts, in the mid-1870s. He published it in 1876 at the inaugural meeting of the American Library Association. As if to suggest from the start that the apparently arcane and tedious business of library classification was intimately linked to the demands of commerce and the glamour of industry and exploration (as well as a sense of American cultural authority), the ALA first convened at the United States Centennial Exhibition in Philadelphia. In fact, Dewey's system seems to have owed something to the catalogue of an earlier exhibition: the sober discipline of librarianship borrowing from the splendour of Victorian progress.

There is still something specifically Victorian about the DDC: it depends on the notion that knowledge can be slotted into a strict hierarchy, a massive tree-like structure in which the fruit of each and every discipline can find its discrete position and unique number. Dewey devised ten main classes: 000 Generalities; 100 Philosophy, paranormal phenomena, psychology; 200 Religion; 300 Social sciences; 400 Language; 500 Natural sciences and mathematics; 600 Technology (applied sciences); 700 The arts; 800 Literature (Belles-lettres) and rhetoric; 900 Geography, history and auxiliary disciplines. Each class in turn branches off into nine standard subdivisions, carving subjects up into theory, reference, history, biography and so on. The whole complex numerical edifice was jealously guarded by Dewey in subsequent decades; he instituted the principle of 'integrity of numbers': once the system was elaborated, he claimed, it would admit of 'expansion without limit'. The latest revision records a further 100 divisions, each with 1000 smaller sections (and that's before you get to the 'complete and exact headings').

The DDC was an immediate success; by the end of the 19th century it dominated the field of municipal library classification (the upstart Library of Congress Classification would eventually win out in universities, to my dismay, a century later). But its contradictions quickly became clear. Like all classificatory systems, the DDC is caught between the organization of knowledge and the ordering of books (what librarians call its 'mark and park' function). This led to unwieldy expansions as the DDC tried to keep up with the growth of knowledge in the early 20th century, culminating, with the legendary 'standard' (15th) edition, in a drastic reduction of categories - all authors' names, for example, were removed (only Shakespeare survives, at 822.3). The DDC came clean as 'a system of classifying books, not a detailed outline of knowledge', losing something of its fantastic Borgesian ambition: the dream of a perfect map of human endeavour inherited from scientists such as Francis Bacon in the 17th century and from the great encyclopaedists of the Enlightenment.

But the system still has its surprises. Shelved at 025.4'31 - dc21, the 21st edition of The Dewey Decimal Classification and Relative Index (1996) is a distinctly strange volume. A detailed introduction and massive index frame the core of the work: the ten classificatory 'schedules' and their numerous, minute subdivisions. It is here that the whole project of classification starts to look slightly mad, with its weird suggestion that if you could grasp the whole structure, you would never need to read the corresponding books at all. Of course, it's impossible: what's missing from this vast anatomy of knowledge is time; it turns the infinite labour of study into an impossibly detailed deep-space snapshot, a vision of a distantly beautiful constellation that doesn't really exist. It is a kind of labyrinthine Gothic fiction, full of obscure instructions ('class here ...') and gloomy dead ends ('do not use ...').

Whatever their artifice, such systems start to seem natural; I never got used to being without the DDC, and much later, when my own endless plans and fantasized structures finally coalesced into something printed, bound, examined and forgotten, there was a slight pang of regret that it would end up shelved under some inelegant number, when all along I'd dreamt of my own little offshoot of Melvil Dewey's imagination. 801.95/DIL would have done nicely.