Friday, April 25, 2014

Eric White, "A Census of Print Runs for Fifteenth-Century Books" (2012): the missing graphs

Recently I came across Eric White's census of all known incuncable print runs, similar to that of Uwe Neddermeyer but more cautious in the kinds of sources he accepts. In short, White includes only historically documented sources and excludes broadsides, and his introduction explains why even documented print runs must be interpreted carefully.

Instead of treating them carefully, I've plugged his numbers into a spreadsheet to generate some quick graphs in order to get a look at the distribution of print runs documented by White.

In his introduction, White is concerned about the distorting effect of one outsize print run on the average, but the easiest way to deal with that is to provide a median instead: Where the arithmetic mean is 593, the median is 500.

But the median may also not be the right figure. Here's the overall distribution, and it seems that we're still dealing with a bimodal distribution. (In all of the graphs - but not the calculations - I've exclude the one edition, ib01102000, with a print run of 5000, in order to keep the column widths sensible, and I've used a bin size of 135 in order to keep all those Sweynheym and Pannartz 275-copy editions together).

Fig. 1: Distribution of fifteenth-century print runs

Just as we saw when looking at Neddermeyer's list, most print runs fall into the range of 271-405 and several more between 406-540, while there's a secondary peak around 946-1080 (nearly all the recorded print runs in this group are right at 1000).

But how much of this is due only to Sweynheym and Pannartz? If we exclude their records, we still find a bimodal distribution, although the two peaks are more equal in size.
 
Fig. 2:  Distribution of fifteenth-century print runs, excluding Sweynheym and Pannartz

One interesting observation is that the bimodal nature of the distribution disappears nearly entirely if we look only at the folios. This doesn't mean there were no folio editions with large print runs, but that they were rather uncommon in comparison to the number of editions with print runs in the range of 271-540. For folios, the average print run is 535, while the median is 400.

 Fig. 3:  Distribution of fifteenth-century print runs, folios only

The majority of the bimodal character of the distribution instead comes from the smaller formats, the quartos and octavos. The mean for quartos and octavos is 715, while the median is 568. It would be more accurate, however, to say that the major mode is around 300, and the minor mode around 1000.
Fig. 4:  Distribution of fifteenth-century print runs, quartos and octavos only

While there were many quarto and octavo editions with print runs in the range of 271-540, attempting print runs of double that size was something that relatively more printers appeared willing to attempt in the smaller formats.

1 comment:

  1. I am delighted to see this interesting offshoot of my ongoing toil with edition sizes. If anyone needs the bibliographic citation for a particular print run, I have those, not quite fully edited, and hope to make them available via CERL soon.
    Eric White

    ReplyDelete