I'm currently quite interested in visual languages of all types. I'm adopting more of a social orientation to these things (i.e., technology as a creation of groups) rather than an individual orientation (i.e., technology as the brain storm of one great individual). As I've noted before, my main interest is with the early machine books such as the manuscripts of Leonardo da Vinci and the theatrum machninarum of individual like Besson, Ramelli, and Leupold. This period is particularly interesting since it represents a great transition in technological communication spanning from the craft work of the middle ages, through manuscripts, to printed books. During the same period we witness the rise of royal privilege (for both printing and invention) and the rise of a literate consumer body.
While my analysis is pretty interesting, it's also pretty far off. I can't just use inter-library loan to get a copy of Leonardo's Codex Madrid! Instead, I've decided to ILL a collection of graphical works that have become a core cornerstone for one particular occupational community, namely the 10 editions of Architectural Graphic Standards.
For any analysis of AGS I should probably know how popular these books were. These numbers seem pretty hard to find. I've learned that learning the *real* sales numbers for any book is a tricky task. Instead, I've just crunched out some of the numbers from the library collections represented in WorldCat:
1932 1st ed. :: 80**
1936 2nd ed. :: 88
1941 3rd ed. :: 248
1951 4th ed. :: 166
1956 5th ed. :: 450
1970 6th ed. :: 746
1981 7th ed. :: 1052
1988 8th ed. :: 1030 student ed. :: 106 (1994)
1994 9th ed. :: 926
2000 10th ed. :: 787 student ed. :: 99
** 1932/1989 Facsim ed. 138
** 1932/1998 Facsim ed. 98
** 1930/1990 Facsim ed. 10
To get total numbers I could probably contact the publisher... but this seems like an awfully complicated procedure. I'm quite encouraged by the potential of Nielsen Bookscan, a service which captures point-of-sale data for the publishing industry. Their statistics are apparently quite good since they started collecting data in about 2001. There are bound to be some gaps (e.g., Walmart doesn't report) but the tool is certainly better than the current status quo: nothing.
For the purpose of studying book history, Bookscan could be used for some pretty interesting studies. I can think of a few good quantitative questions: do book sales correlate with citations in the journal literature (Bookscan and Web of Science); do book sales correlate with citations in books (Bookscan and A9); do book sales correlate with library collections (Bookscan and WorldCat); are there differences based on format (i.e., paperback), LC code, or DDC code; do book sales correlate with Amazon ranking (already attempted in a seemingly non-rigourous fashion); what's the effect of best seller lists on book sales (Alan Sorenson has a manuscript on this topic); etc.
To effect these analyses, there are a number of other questions that should also be addressed: how many titles have been published each year; how many titles have been added to library collections each year; what are the cull rates; etc.