How many book copies do libraries buy?
I’m curious how libraries decide how many copies to buy of a particular book. While I presume for standalone books they estimate based on buzz, reviews, and an author’s track record, I’m particularly curious how they decide for the volumes in a series.
Case in point: I’m currently in the midst of a six book series. The library has two copies of each of the first three volumes, but only one copy of each of the last three books. So I got sucked into the series with the easy availability of the first three, but upon finishing them and checking on the fourth I discovered that there’s a backlog: three holds on the fourth book, and two holds on the fifth (I didn’t even check the sixth).
A cynical interpretation of this phenomenon would posit that it’s a way to drive readers to pay authors, since if you’re midway into a series and don’t want to wait and see what happens next you’re more likely to buy a copy than wait for one at the library. Sure enough, rather than wait I bought and immediately downloaded the Kindle version of the fourth book and resumed reading. I should note that if this is a partial explanation for the fewer copies of books later in series, I actually don’t mind. I often try out new series by reading the first book or two from the library, and then buy later volumes because I feel like I should be paying the author to reward them for their work.
I suspect that the real reason for more copies of books earlier in a series and fewer copies for later books, however, is simply that fewer people finish a series than start it. Given attrition of readers for any given series, at a certain point the library probably just needs fewer copies.
If that’s true, an interesting follow-up question would be whether that’s a planned behavior (e.g., whether libraries expect to have to purchase fewer copies of books later in a series), or whether it’s an emergent behavior (e.g., they buy based on demand, starting out with one copy of each book and only adding more if sufficient demand arises).
So those are the questions. Anybody got data (or a dataset suitable for analysis)?