Somewhere around 11 PM on a Tuesday — it's always a Tuesday — you will attempt to catalog a book printed in Lyon in 1642 and the software will ask for an ISBN. Not politely. Insistently. A red border. A validation error. A small, pixel-perfect indignity.
Your book has a colophon. A printer's device. A privilège du Roy. It has survived the Fronde, two world wars, and at least one previous owner who annotated it in pencil. What it does not have is a thirteen-digit number starting with 978. This will be treated as a deficiency.
The Tyranny of the Barcode
Gordon Foster, a statistician at Trinity College Dublin, devised the ISBN in 1965 for the British bookseller W.H. Smith, who were drowning in paperwork. The system — nine digits initially, ten from 1970, thirteen from 2007 when it merged with the EAN barcode standard — is genuinely elegant. The check digit uses a modulus-11 algorithm (ISBN-10) or a simpler alternating-weight scheme (ISBN-13). It uniquely identifies an edition, a format, a publisher. It solved a real problem for a real industry at a specific moment in time.
The difficulty is that the moment in time was 1970, and the industry was mass-market publishing. The ISBN was designed for offset-printed paperbacks moving through national distributors to high-street bookshops. It was never intended to be the universal key to all human publishing. It became one anyway, in the way that a temporary workaround always becomes permanent infrastructure — not by design, but by default.
Open any book cataloging application built after 2005. The ISBN field isn't just present; it's primary. The entire architecture often hangs from it like a mobile from a ceiling hook. Scan the barcode, fetch the metadata, populate the fields. It's fast, it's clever, and it fails silently and completely for roughly five centuries of Western printing.
The Identifiers That Actually Matter (And That Most Software Ignores)
If your collection extends before the ISBN era — or into any corner of publishing that the ISBN system doesn't reach — you already navigate a parallel universe of reference numbers, each with its own history, scope, and constituency of devoted users.
ESTC (English Short Title Catalogue) covers approximately 480,000 editions printed in English or in English-speaking territories before 1801. If you handle early English books, ESTC numbers are not optional metadata — they're the language your insurance company speaks. A "Wing S4482" or "STC 22273" will locate a book with more precision than any ISBN, and will be understood by every rare book librarian from the Bodleian to the Folger.
The German equivalent is a trio: VD16, VD17, and VD18 — the Verzeichnis der im deutschen Sprachbereich erschienenen Drucke — cataloging German-language books of the sixteenth, seventeenth, and eighteenth centuries. VD16 alone exceeds 100,000 entries. VD17 is nearing completion at over 300,000. These aren't academic curiosities; they're the infrastructure of the German antiquarian trade. Quote a VD17 number to a dealer in Munich and you will be taken seriously. Quote an ISBN and you will be gently redirected to the contemporary fiction table.
USTC (Universal Short Title Catalogue) aims to catalog every European book printed before 1601 — currently above 400,000 records and growing. GW (Gesamtkatalog der Wiegendrucke) covers incunabula specifically, with the kind of obsessive precision that only German bibliographers seem capable of sustaining across multiple generations.
Then there are the great retrospective bibliographies: Brunet's Manuel du libraire, Graesse's Trésor de livres rares et précieux, Sabin's Dictionary of Books Relating to America — each a monument of nineteenth-century scholarship, each still cited daily in auction catalogs. A "Brunet III, 1247" is a perfectly functional identifier. It just happens to reference a page in a book published in 1864, which presents certain UX challenges for developers raised on REST APIs.
OCLC numbers and Library of Congress Control Numbers predate the ISBN and remain, quietly, the most comprehensive identifiers in existence. If a research library anywhere on earth holds a copy, the book has an OCLC number. This is the closest thing we have to a universal book identifier — and it is, characteristically, the one that most consumer software ignores entirely.
The Invisible Assumption
The problem is not that ISBNs exist. They're genuinely useful for the books they cover. The problem is the assumption, baked so deeply into most software that it's become invisible, that the ISBN is the identifier — the primary key, the sine qua non, the thing without which a book is somehow incomplete.
This assumption has consequences. Try importing a catalog of 500 pre-1970 titles into software built around ISBN lookup. Watch it auto-match your 1742 octavo to a 2019 reprint because they share a title string and the algorithm, finding no ISBN to anchor itself, grabbed the first thing it could. Watch it treat your Elzevir as a gap in the data rather than a book that predates the data.
Try selling on a marketplace that requires an ISBN for listing. There is a cottage industry of collectors who know, with weary precision, that entering "N/A" in the ISBN field of certain platforms triggers a validation error, while "0000000000" passes but may auto-match to a Slovakian phone directory from 1994. These are not edge cases. These are Tuesday evening for anyone who deals in books printed before the Nixon administration.
The ISBN covers roughly 50 years of mainstream publishing comprehensively, another 20 years patchily, and everything before that not at all. For private press books, artist's books, zines, pamphlets, broadsides, ephemera, and the entire tradition of samizdat and clandestine printing — nothing. For books published in countries that adopted the system late (Greece: 1981; China: 1982; parts of Africa: still incomplete) — gaps everywhere.
What Good Software Ought to Do
A serious cataloging system should treat the ISBN as what it is: one reference number among many, useful for a specific and historically narrow subset of the world's books. It should support ESTC, VD, USTC, OCLC, LCCN, and arbitrary bibliography references without requiring a workaround or a custom field. It should let you record that your book is "Hain *4995" and not ask you what kind of barcode that is.
It should also understand that a book without an ISBN is not a book with missing data. It is a book with different data — older, richer, and considerably more interesting.
Your 1642 Lyon quarto will outlast the software you catalog it in. It has outlasted everything else so far. The least we can do is build systems that don't treat it as an exception.
📖 Related in the Wiki: Identifiers, Barcode Scanner
Next in this series: ISBD — the international standard that describes your books better than you do, and why learning to read it will make you a sharper collector.