Ann sometimes fiddles with a system called OCLC FirstSearch, to which I don't have access. She says,
OCLC records what libraries have acquired what books, so you can get not only the total figure, but a list of every library in the database that has the book. I'm not an expert, so I'm not sure how comprehensive the OCLC is in terms of small libraries, but certainly many public libraries are listed.In checking the records on Jay Winik's April 1865, she notices that "Mr. Winik's book has sold a very substantial 1,844 copies to libraries, so the 2,000 overall figure you give must be wildly underestimated since most of these acquisitions occur during the first two years."
I'm willing to admit that many of these calculations may be way off, although I should say that the 2,000 number was intended to show sales for the year 2003 only, rather than total sales. For instance, I see multiple copies of Winik in Borders and B&N, which suggests that his big library numbers supplement commercial sales.
How to explain this distortion? Ann says "Libraries mostly acquire their books through the distributor Baker & Taylor, not Ingrams."
B&T and Ingrams own the wholesale market, between them, but only Ingram's sales are accessible. Thus, although "Library figures ... are a good way of assessing sales, better than Ingrams for a serious hardcover book destined for libraries," we don't have any rule of thumb for extrapolating the data into estimates. (I assume Ann means assessing sales in terms of "getting a sense of".) And we know that libraries favoring B&T skews Ingrams-based projections.
Thus, although as Ann says, "The x6 or even x8 rule is, I believe, quite arbitrary," it seems to me to be all we have for the kind of pseudoscientific wild guessing that conveys the essence of blog informatics. I know the rule lowballs a lot of sales, especially through my use of 6 as a multiplier (rather than 7 or 8) . I have had a guilty feeling about not disclaiming the sales estimates more and will double my efforts in that department in the future.
That said, I like the "rule" for being used by publishing pros and I have confidence that relative position within the Ingram sales list may generally point to relative position in the overall marketplace, moreso for trade books than for university press imprints.
Meanwhile, do take heed that these estimates I give cannot, ultimately, be relied upon and that they should never be bandied about as if they were the actual sales figures from a royalty statement.
I'll look at a raft of Grant book sales tomorrow with a higher number of cautionaries than previously.