Choosing books for a library like mine in New York is a fulltime job. The head of acquisitions at the Society Library, Steven McGuirl, reads Publishers Weekly, Library Journal, The Times Literary Supplement, The New Yorker, The New York Review of Books, the London Review of Books, The London Times, and The New York Times to decide which fiction should be ordered. Fiction accounts for fully a quarter of the forty-eight hundred books the library acquires each year. There are standing orders for certain novelists—Martin Amis, Zadie Smith, Toni Morrison, for example. Some popular writers merit standing orders for more than one copy.

But first novels and collections of stories present a problem. McGuirl and his two assistants try to guess what the members of the library will want to read. Of course, they respond to members' requests. If a book is requested by three people, the staff orders it. There's also a committee of members that meets monthly to recommend books for purchase. The committee checks on the librarians' lists and suggests titles they've missed. The whole enterprise balances enthusiasm and skepticism.

They want a full collection but don't want to be saddled with books nobody reads.

Boosted by reviews, prizes, large sales, word of mouth, or personal recommendations, a novel may make its way onto the library shelf, but even then it is not guaranteed a chance of being read by future generations. Libraries are constantly getting rid of books they have acquired. They have to, or they would run out of space. The polite word for this is "deaccession," the usual word, "weeding." I asked a friend who works for a small public library how they choose books to get rid of. Is there a formula? Who makes the decision, a person or a committee? She told me that there was a formula based on the recommendations of the industry-standard CREW manual.

CREW stands for Continuous Review Evaluation and Weeding, and the manual uses "crew" as a transitive verb, so one can talk about a library's "crewing" its collection. It means weeding but doesn't sound so harsh. At the heart of the CREW method is a formula consisting of three factors—the number of years since the last copyright, the number of years since the book was last checked out, and a collection of six negative factors given the acronym MUSTIE, to help decide if a book has outlived its usefulness. M. Is it Misleading or inaccurate? Is its information, as so quickly happens with medical and legal texts or travel books, for example, outdated? U. Is it Ugly? Worn beyond repair? S. Has it been Superseded by a new edition or a better account of the subject? T. Is it Trivial, of no discernible literary or scientific merit? I. Is it Irrelevant to the needs and interests of the community the library serves? E. Can it be found Elsewhere, through interlibrary loan or on the Web?

Obviously, not all the MUSTIE factors are relevant in evaluating fiction, notably Misleading and Superseded. Nor is the copyright date important. For nonfiction, the CREW formula might be 8/3/MUSTIE, which would mean "Consider a book for elimination if it is eight years since the copyright date and three years since it has been checked out and if one or more of the MUSTIE factors obtains." But for fiction the formula is often X/2/MUSTIE, meaning the copyright date doesn't matter, but consider a book for elimination if it hasn't been checked out in two years and if it is TUIE—Trivial, Ugly, Irrelevant, or Elsewhere.

Clearly, the CREW formula is not in the same class as E = mc2. A lot of subjectivity is required to decide if a book is Trivial or Irrelevant, even if a book is irretrievably Ugly. And if Elsewhere includes the Internet, the formula raises the whole question of whether it's worth keeping paper copies of older books in your local library, as almost all of them, if Google has its way, will be available online. The writer of the CREW manual emphasizes that the librarian's judgment must be constantly engaged in making these decisions. The CREW formula is a guideline, she insists. "It is important to remember that guidelines are not intended to act as a substitute for professional judgment calls and common sense." Discard works no longer in demand, she recommends, especially second and third copies of past bestsellers. "Retain works of durable demand and/or high literary merit."

It is not clear to me that the author of the CREW manual has any idea of how hard it is to determine "high literary merit" as opposed to "durable demand." My friend said they had done a recent fiction "weed" in which all the books that met the CREW requirements were then reviewed by a librarian to see if the book had a local connection or local interest and also to make sure that such classics as one of Trollope's Palliser novels wouldn't be thrown away, even if no one had checked it out in two years. That's fine for the Palliser novels, but what about, let us say, Rhoda Lerman? Etienne Leroux? Sigrid Undset, the Nobel Prize–winning novelist whose work my mother's generation of women revered, who is now largely unknown? Or even Lawrence Durrell—whose Alexandria Quartet I once bought at a library book sale, being discarded from the collection?

Weeding, even in the garden, has become a remarkably controversial subject. There's a powerful school of thought, more philosophical than botanical, going back to Emerson, that regards a weed as a misunderstood plant. Many people believe that all green life is holy and should not be inhibited.

Weeding, to such people, is akin to eugenics and murder. Some people feel the same about books: no book should be removed from a library.

They are all worth preserving. Thus, there is a tone of defensiveness whenever librarians discuss their weeding procedures. This is from the New York Society Library's 2010 Annual Report:

Circulation count is never used as the sole criterion for deaccessioning a book. The Head of Acquisitions then reviews the list weighing a number of factors: other library holdings, books by the same author in the collection, price and availability via second-hand booksellers . . . age of the material, citation in bibliographies, book condition, whether subject coverage by other books in the collection renders the book in question obsolete or redundant, as well as other variables such as illustrators, binding, donor bookplates, and so forth. It is important to keep in mind that many books have research potential precisely because they are out of date and provide a window into the cultural attitudes of a previous time. It is an involved process, and it can take much longer to select a book for withdrawal than it can to select it for purchase.

The novelist Nicholson Baker is perhaps the most vocal critic of deaccessioning in libraries, first in 1996, in a much-discussed exposé of the San Francisco Public Library (SFPL) published in The New Yorker, and then in a book called Double Fold. Concerned initially about the disappearance of card catalogs and the way that enthusiasm for information technologies was damaging the traditional culture of libraries, he sued the San Francisco Public Library for access to their old card catalog, which had been replaced by an electronic one as the library moved into a new building. What he discovered went far beyond the issue of paper versus electronic cataloging. He found that there were many books noted in the card catalog that no longer existed in the library. Between the time the SFPL left its old main building and the time it moved into its high-tech, built-for-the-future, and supposedly more roomy new one with its electronic card catalog, somewhere between a hundred thousand and a quarter of a million books were removed from the collection.

This was weeding on a scale—and in a time frame—that suggested reckless destruction more than considered selection. Many books that existed in no other copies, many books arguably with historic value, had been simply thrown away and buried in landfill. Partly this had been done because the new library, while boasting great architectural flourishes and lots of architectural space, did not have enough shelf space. Partly it had been done because the current librarian had a view of what books belonged in the collection that differed from that of previous librarians. He saw the library as serving the general reader, as opposed to researchers and literary professionals, arguing that with the Berkeley and Stanford university libraries nearby, there was no lack of research libraries in the San Francisco Bay Area. He conceived of the SFPL rather as a library for a current urban population and therefore saw an opportunity to pare its collections radically.

A weed is something you don't want growing in your garden—more formally, "a plant that interferes with management objectives for a given area of land at a given point in time." Every garden represents someone's "management objective," and so does every library. The definition reflects the careful relativism and systems orientation of our time. Management objectives might change, and then the weed would be very welcome.

But skeptics of library weeding, like Baker, are keenly aware of the difference between gardens and libraries: once you've weeded out a book, it isn't going to grow back again.

Still, Baker may have ignored the extent to which a library must articulate and fulfill its own objectives. Viewing libraries as repositories, he overestimated their preservation function and underestimated their need to serve a specific community. I find myself sympathetic, if unequally, to both parties in this dispute, wholeheartedly to Baker's book-loving bellows of rage at the destruction of precious objects but also to the librarian's desire to create an institution that serves its community. That this battle about the form and function of libraries is not over was made clear when the New York Public Library's announcement of new construction caused protests from people (like me) who fear sweeping changes to libraries. (Editor's note: The New York Public Library recent announced that it was abandoning its renovation plan.)

The New York Society Library is in the fortunate position of not having to worry as much about weeding as many other libraries. They consider the forty-eight hundred books a year they acquire a magic number. If it were three or four times greater, they would have to weed their collection much more severely. As it is, they can concentrate on finding ways to make more space while keeping the collection—at any rate the fiction collection—fairly stable. Last year, McGuirl moved the collections of O. Henry prize stories and Best American Short Stories to closed storage in the basement while leaving the last few years' volumes on the open stacks. That freed up a lot of room. Eliminating duplicate copies of books that were popular in the past but are no longer read much frees up more room. Inch by inch, space has to be found for the new. Regarding it as a collection with a special character—to record the reading tastes of New Yorkers over the years—the NYSL librarians are reluctant to deaccession unique copies of fiction.

Elsewhere, the process is more complex and potentially contentious—for example at a university library like mine, where professors are ready at every turn to watchdog and yelp. The Wesleyan University library is engaged in a three-year project to weed out sixty thousand volumes. They are out of space, and weeding on the small scale, as the Society Library does, only frees space that is immediately filled. To achieve the goal of weeding sixty thousand volumes, the library has undertaken to consider ninety thousand, or 6 percent of all the books in the library. Lists of books that meet the initial weeding criteria are available to students and faculty, members of which can champion any volume they care to. The scale of the operation is stupefying. I looked at the list for the Library of Congress category PR—English literature—and there were nine thousand entries. This means that nine thousand books, published before 1990, had been checked out only two times or less since 1996 and not at all since 2003. To my deep sadness, I recognized titles on the list. They were works of literary criticism that had been written by friends of mine when we were young and now were considered at the end of their useful life, just like their authors. Their removal from the library was like an actual death, a kind of death I had never imagined.

People who feel strongly about retaining books in libraries have a simple way to combat the removal of treasured volumes. Since every system of elimination is based, no matter what they say, on circulation counts, the number of years that have elapsed since a book was last checked out, or the number of times it has been checked out overall, if you feel strongly about a book, you should go to every library you have access to and check out the volume you care about. Take it home awhile. Read it or don't. Keep it beside you as you read the same book on a Kindle, Nook, or iPad. Let it breathe the air of your home, and then take it back to the library, knowing you have fought the guerrilla war for physical books. This was the spirit in which I checked out the third book in Etienne Leroux's Welgevonden trilogy with no intention of reading it.