Reed’s library uses the Library of Congress system: books are shelved under a one- or two-letter general prefix, then by a three-digit number, with edition and date and so forth as a tiebreaker. For example, I have here E99.H2B74 1999 (as in most North American books, it’s given on the copyright page). Maybe this isn't as elegant as the absolutely one-dimensional Dewey system, but I think it’s denser, more conductive to sanity, and, given practice, as clear.

The library here where my grandmother lives, anyway, uses the Dewey system. Hm, I think, stepping inside and shooting my ample cuffs, Herewith I shall find and peruse Edward Tufte’s outstanding trilogy on practical graphic design, viz. Envisioning Information, , and of course . And indeed, the computer finds each book – Envisioning Information at 001.4226, in the misc reference section, at 302.23, under media studies, and at 760, as printing and graphic design.

Huh? These books were written by one author in one decade, and if you’ve seen one of them, you can’t see either of the others without wanting to shelve them together. Each has on its back something like if you liked this book, please try its two companion volumes, ... – really, because the first two have been reprinted since the last came out [double-check]. I guess I had some kind of tacit supposition that competant librarians, and programs acting as their agents, had a new book subfunction that asks where the author’s previous books went. Sheesh. I’ll look for the encyclopedia under machinery for Abacus–Anywhere, botany for Amaranth–Axon, and paleontology for Aztec–Boycott.

Henceforth, this will be known as the Tufte lemma of—

Loyd’s Dewey Misc Theorem:

The proportion of a library’s interesting books shelved between Dewey 000.0000 and 003.0000 approaches 1/3.