Chapter 7

Updated 2026-07-02

"A library is a growing organism." — S. R. Ranganathan, The Five Laws of Library Science, Fifth Law (1931)

He was not supposed to become a librarian at all. Shiyali Ramamrita Ranganathan was a mathematics teacher, twice over by degree, in colonial Madras — born in 1892 in the small town of Shiyali, the first grandchild on both sides of his family, raised largely by his maternal grandfather after his father, a farmer, died when Ranganathan was six. He taught mathematics for years, wrote papers on its history, and applied for the newly created post of University Librarian at the University of Madras in 1924 for the least romantic reason imaginable: it paid better than his professorship. He had no training in libraries whatsoever. In the days before his interview, by his own account, he read up on the subject in the Encyclopædia Britannica.¹

He got the job, and he hated it. Cataloging bored him; the work felt like clerical filing with none of the intellectual stimulation of mathematics, and within months he was asking for his old teaching post back. The university, instead, sent him to London to actually learn the profession he'd talked his way into, at the School of Librarianship at University College London. He toured more than a hundred libraries while he was there, watching what worked and, more often, what didn't. And on one of those London days, away from any library at all, he wandered into the toy department at Selfridge's department store and stopped to watch a salesman demonstrate a Meccano set — a box of perforated metal strips, nuts, and bolts that children used to bolt together model cranes, bridges, cars, whatever they could imagine, then take apart and rebuild as something else entirely.² The same handful of parts, recombined, became a different object every time. Nothing about the parts changed. Only the combination did.

Ranganathan would later describe the moment as the one where his whole approach to classification clicked into place, and the comparison historians of the field have reached for ever since is almost too neat: Newton had his apple, Kekulé his dreaming snake biting its own tail, and Ranganathan, appropriately for a man whose new science would be built and rebuilt out of standard parts, had a toy.

---

What he saw in the Meccano set was a way out of the trap Dewey, Wilkins, and Otlet had all built for themselves without realizing it: one shelf, one number, one true place for a thing — and no good answer for the thing that's two things at once.

A book on Dewey's shelves, or in Wilkins's vocabulary, or in Otlet's thousand slots, gets exactly one address, chosen in advance by whoever built the system. If you need to find "the treatment of lung tuberculosis with x-rays, in India, in 1950," the old systems make you decide, once and forever, which single fact about that book matters most — is it shelved under medicine, or under India, or under the year? Ranganathan's answer, worked out over the following decade and published in 1933 as Colon Classification, was to stop forcing the choice.³ Instead of one number, he broke a subject down into up to five kinds of facts about it, each one optional, each one combinable with the others like Meccano strips: what the thing fundamentally is (its Personality — here, medicine, or more narrowly, lungs); what it's made of or concerns (its Matter — tuberculosis); what's being done to it (its Energy — treatment by x-ray); where (its Space — India); and when (its Time — 1950).⁴ Personality, Matter, Energy, Space, Time — PMEST, in the abbreviation every library-school student since has had to memorize — joined together with colons, which is where the system gets its name: a book doesn't get one number anymore, it gets a sentence built out of facets, assembled fresh for whatever it actually is, the way a Meccano model gets built fresh out of the same strips and bolts every time.

The gain is easiest to see in a case the old systems simply couldn't handle gracefully. A pipe organ is, by the physical mechanism that makes its sound, a wind instrument — air forced through pipes, no different in principle from a flute, just larger and more elaborate. It is also, by the interface a musician actually touches, a keyboard instrument, sitting in every practical sense next to the piano and the harpsichord. A strict hierarchical tree, the kind Dewey or Wilkins would have built, has to pick one branch and hang the organ from it, and whichever branch wins, the tree is now lying about half of what an organ is. A faceted system doesn't have to choose. It just tags the organ with both facts — sound-production: wind; interface: keyboard — and lets a search or a shelf or a catalog card combine them however the question at hand requires. Two facets, not two competing parents fighting over an only child. It is the same move that Ranganathan's own University of Madras had no working answer for when it came to a book that was, with equal seriousness, history and geography, or medicine and a place and a year. Facets don't resolve the tension between two true things. They just stop pretending there has to be only one.

---

The facets were never just a technical fix, either, and Ranganathan never treated them as one. In 1931, two years before Colon Classification reached print, he published a short, plain-spoken manifesto called The Five Laws of Library Science, and the five laws are less a theory of cataloging than a theory of who a library is for. Books are for use — not for preservation behind glass, not for the librarian's convenience, for use. Every reader his book. Every book its reader. Save the time of the reader. A library is a growing organism.⁵ Read against Dewey's tree, the laws sound almost like an argument: a system that locks a thing into one number, decided once by someone else, is a system built around the convenience of the shelf, not the reader standing in front of it. Facets are what those five laws look like once you sit down and actually do the engineering — a classification scheme designed, from its first principle, around the person trying to find something rather than the person who once decided where it ought to go.

---

Ranganathan's faceted idea barely dented Dewey's dominance — Colon Classification itself never spread much beyond a handful of Indian libraries that adopted it directly, and most of the world's books are still, today, shelved on Dewey's hundred-and-fifty-year-old ten-class tree. But the underlying logic — describe a thing by its attributes instead of forcing it onto one branch of one tree — turned out to be everywhere else already, doing useful work under names that have nothing to do with libraries.

Walk into almost any wine shop and the wine is never organized by just one fact. It's filed by region, and you can also ask for it by grape, and by vintage, and by price band, and the same bottle answers to all four questions depending on which one you walked in asking. Nobody designing that shop floor read Ranganathan. They just discovered, by trial and error, that wine doesn't have one true address either, and built colon classification from scratch to solve a problem libraries had been mishandling for a century. A recipe site does the same thing with ingredient, cuisine, cook time, and dietary restriction. An inbox that lets you label a single email both "Work" and "Tax Documents" and "Urgent," all at once, without forcing a choice among them, is running PMEST without the acronym. So is a bookstore's recommendation page that ranks the same novel, simultaneously, as a bestseller in literary fiction, a bestseller in books about grief, and a bestseller in books set in a particular city — three true facts about one object, three shelves, no tree forced to pick a favorite. A handful of librarians in postwar Britain, working under the banner of something called the Classification Research Group, spent the 1950s trying to formalize this insight for the sciences, refining facet theory into something more rigorous than Ranganathan's own first draft.⁶ Almost nobody outside the field has heard of them. Nearly everyone now uses what they built, every time they filter a search by price and brand and color all at once instead of hunting through someone else's idea of the one correct department.

---

And yet even Ranganathan's system, for all the ground it gained back from Dewey's tree, runs into the same wall Wilkins and Dewey and Otlet all hit, just further down the road. Five facets cover an enormous amount of what can be said about a tuberculosis study or a bottle of wine. They do not cover everything. Sooner or later something arrives that resists Personality, Matter, Energy, Space, and Time alike — that isn't quite about anything in particular, or is about so many things at once that naming its facets doesn't help — and the system needs somewhere to put it anyway. Every scheme, faceted or hierarchical, ancient or built last year, eventually grows the same small, embarrassed shelf for the things it can't otherwise place: a miscellaneous category, a general, an other, a folder nobody quite owns. It is the quietest possible confession a system can make, and it is universal. Wilkins didn't have one, and that was his problem. Otlet had one and gave it a number — 179 — and Borges caught him at it. Every well-organized house has a drawer like this too, the one with the spare keys and the dead batteries and the single chopstick, and everyone who owns one knows what it means: not chaos, exactly, just the system admitting where its categories ran out.

That drawer is harmless on a kitchen counter. It is harmless in a library catalog. It stops being harmless the moment the things going into it are people, and the form is asking what race they are, and "miscellaneous" is no longer a shelf nobody minds being on.

---