The Shape of Incunable Survival

Jonathan Green, Frank McIntyre, and Paul Needham. “The Shape of Incunable Survival and Statistical Estimation of Lost Editions.” Papers of the Bibliographical Society of America 105.2 (2011): 141-75.

The article I coauthored with Frank McIntyre and Paul Needham is now out in PBSA. The article addresses an old question: How many editions printed in the fifteenth century have disappeared without a trace? A couple of pictures illustrate the problem. If you think of book survival as a coin toss, even with a coin that comes up heads only 3% of the time, then you can plug in reasonable estimates of print runs and come up with something like this:
So you'd expect lots of incunabula to survive in 7-18 copies, relatively few with more or less than that, and not many lost editions at all. If you count up all the copies recorded in the ISTC (using a process like the one I proposed in the third article I ever published), however, the cruel reality looks like this:
In other words, you find thousands and thousands of editions with only one known copy, and a steep drop off to two or more copies, but also a few editions surviving in hundreds of copies (and one, the first Latin edition of Hartmann Schedel's Nuremberg Chronicle, in over a thousand).

So simulating incunable survival as independent coin tosses just won't work. Our approach in this article uses a maximum likelihood estimation for a negative binomial distribution and some reasonable assumptions about how many editions any particular printer might produce, and we find that the number of lost editions is quite a bit higher than most earlier estimates, quite possibly in the range of 40-60%, and 35-50% even if you eliminate broadsides from the analysis.

Between Frank's statistical reasoning and Paul's vast knowledge of early printing, the article proved to be a fascinating project to work on. For German Studies, I think the article serves as a useful caution that what the early modern book world looks like to us is possibly very different from what it looked like at the time. We only have the books that five centuries worth of book owners and librarians have chosen to keep around, which were not the same as the books that people chose to buy and read in the fifteenth century.