In the early 1920s, the art historian Aby Warburg arranged the books in his private library in Hamburg according to a principle he never fully wrote down but that his colleagues understood well enough to preserve after his death. He called it the Gesetz der guten Nachbarschaft — the law of the good neighbor.
The idea was simple and, to anyone trained in library science, mildly heretical. Warburg did not organize his books by author, by title, by date, or by subject classification. He organized them by affinity — by the principle that the book you are looking for is rarely the one you actually need. The one you need is the book next to it, the one you were not looking for, the one that would never have appeared in a subject catalog or a search query because it belongs to a different discipline entirely.
The Library
Warburg’s collection eventually grew to some sixty thousand volumes. When he died in 1929, his associates Fritz Saxl and Gertrud Bing moved the entire library from Hamburg to London, where it became the Warburg Institute — part of the University of London, still operating, still organized on roughly the same principle.
What made the library unusual was not its size but its architecture of thought. A visitor looking for a book on Renaissance painting might find it shelved next to a treatise on astrology, a volume of classical mythology, and a pamphlet on theatrical costume. The adjacency was not accidental. Warburg believed that the movement of images and ideas across cultures — what he called the Nachleben der Antike, the afterlife of antiquity — could not be understood within the boundaries of any single discipline. His library was built to make those connections visible. You could not browse one shelf without encountering the shelf next to it, and the encounter was the point.
What Adjacency Reveals
The conventional library is organized for retrieval. You know what you want; the system helps you find it. Warburg’s library was organized for discovery. You thought you knew what you wanted; the library showed you what you had not yet considered.
This is a distinction worth sitting with. Most systems for organizing knowledge — card catalogs, databases, search engines, recommendation algorithms — assume the user has a question. They are optimized for answering it. Warburg’s principle starts from a different assumption: that the most productive encounter with knowledge is the one you did not plan. The question you arrived with is less important than the question that forms while you are looking.
Anyone who has spent time in a physical library has experienced this. You go to the stacks for one book and leave with three, two of which you had never heard of and one of which turns out to be more useful than the book you came for. The experience is so common that it has no name — it is simply what happens when books are physically near each other and a reader is moving through them.
Warburg gave it a name. He recognized that the phenomenon was not an accident of shelf space but a property of knowledge itself. Ideas that are formally unrelated often share deep structural connections. A book about Renaissance astrology and a book about Greek mythology are filed in different departments by every library in the world, but they are about the same underlying question: how did ancient symbolic systems survive into modernity? Warburg’s shelf put them together because the question linked them, even though no classification scheme recognized the link.
The Problem with Classification
Every classification system is a theory about what matters. The Dewey Decimal System encodes a nineteenth-century American Protestant worldview. The Library of Congress Classification reflects the structure of the U.S. Congress’s research needs. Neither is neutral. Both impose a topology on knowledge that makes certain connections easy to see and others invisible.
Warburg’s arrangement was also a theory, but it was a different kind of theory. It did not claim to be universal or systematic. It was personal — the product of one scholar’s lifetime of reading and thinking about how images, symbols, and ideas migrate across cultures and centuries. The shelves reflected not a classification of knowledge but a path through knowledge, the specific path Warburg had walked.
This is what makes the principle interesting beyond the history of libraries. Every serious reader’s bookshelf is a path through knowledge. The arrangement — whether deliberate or accumulated over years — records a set of adjacencies that no algorithm would generate. A shelf that puts Marcus Aurelius next to a book on cognitive behavioral therapy is not making a classification error. It is recording a connection that the reader found real, a bridge between Stoic philosophy and modern psychology that the reader crossed and found sturdy enough to keep both books.
The Pair You Did Not Expect
The most valuable thing a bookshelf can show you is the neighbor you did not expect. Not the obvious pairing — the business book next to the business book, the novel next to the novel — but the one that crosses a boundary. A shelf that pairs a book on evolutionary biology with a history of Renaissance banking is telling you something that neither book says on its own. The reader who kept both found a connection between them, and that connection — replicated across enough shelves, across enough readers who have never met — becomes a signal.
Warburg understood this a century before anyone had the tools to measure it. His library was an argument, expressed in the arrangement of sixty thousand volumes, that knowledge does not live in categories. It lives in the spaces between them. The good neighbor is not the book that belongs to the same subject. It is the book that belongs to a neighboring question — close enough to be relevant, far enough to be surprising.
The principle has not aged. If anything, it has become more urgent. In a world where book discovery is dominated by algorithms that optimize for similarity — “readers who liked this also liked” — the encounter with the unlike, the adjacent, the unexpectedly relevant is harder to come by. Search engines are retrieval machines. They answer the question you typed. They do not show you the book next to the one you searched for, because there is no “next to” in a database.
A bookshelf has a “next to.” Every bookshelf does. And the aggregate evidence of thousands of bookshelves — what readers chose to keep in the same room, what books ended up side by side not because an algorithm placed them but because a human mind found them connected — is a dataset that Warburg would have recognized immediately.
At SPINES, we call the study of these adjacencies Shelfology, and at its core is the book pair — two titles that recur together across unrelated libraries, shelved in proximity by readers who arrived at the same pairing independently. The pair is the smallest unit of Warburg’s principle made measurable: not a classification, not a recommendation, but evidence that two books belong to the same neighboring question, observed across enough shelves to be more than coincidence.
Simply put — we read between the spines.