Defining guidelines for Edition entities is probably going to be the largest style effort, so I’m going to kick it off by considering a single problem of when to create separate Edition entities: print runs. (Hereafter, I’ll use capitalization to distinguish between BookBrainz entities and colloquial concepts of an edition, i.e. Edition vs. edition.) Publishers frequently do multiple print runs of what they consider a single edition; typically, large publishers include a printer’s key in the front matter to distinguish between different print runs.
In my ideal world, we could consider print runs as an unimportant distinction and simply create one Edition per publisher edition. Unfortunately, the real world is not so tidy. In order to help us find a guideline, I’ll present several examples of how copies of books differ between printings for discussion, then follow up later with some of my own thoughts on the problems presented.
Springer publishes a book titled Graph Theory, authored by J. A. Bondy and U. S. R. Murty. My copy contains a note next to the printer’s key that the second printing contains corrections. I have compared the second printing to the first and pagination diverges significantly and exercises differ between the two in addition to whatever other corrections were made. However, Springer continues to describe this book as in its first edition and with no change in ISBN.
My copy of Ursula K. Le Guin’s The Left Hand of Darkness is the 56th printing of the Ace Books edition. The front matter indicates that the first printing of the Ace edition was in 1969, before the development of UPC barcodes or ISBNs, but my copy has both on the cover. Title: The Left Hand of Darkness shows many different printings with the same ISBN and significant variation in covers as well as the printed retail price on the cover. Searching Amazon shows that the ISBN remains in use as the ISBN of a “50th Anniversary Edition”.
I own two copies of The Commitments by Roddy Doyle, both from the Vintage Contemporaries imprint of Random House. The printer’s key in one indicates that it’s the 20th printing. I am unable to decipher the printer’s key in the other, as it just reads “79B86”. Both appear broadly identical, but the “79B86” printing shows a price of U.S. $11, also reflected in the EAN-5 on the cover. The 20th printing shows a price of U.S. $12, also reflected in the EAN-5. They have the same EAN-13 and ISBN, but the barcode area is laid out differently between the two.
My copy of Disco Pigs & Sucking Dublin by Enda Walsh appears broadly identical to other printings. However, it has an ISBN-13 printed on the back cover, while other printings borrowed from the Internet Archive were released before the introduction of the ISBN-13 and show an ISBN-10 on the back cover.
I own two copies of Samuel B. Griffith’s translation of Sun Tzu’s The Art of War, both published by the Oxford University Press. One is the 56th printing, the other is the 57th. I can discern no other difference between the two.
Please do bring up additional examples if you have anything you think isn’t covered by these examples.
To my mind, the Graph Theory example is clear-cut. The contents of the book differ materially between the first and second printings and each should be a different Edition. The next three deal primarily with differences on the cover (though I believe different printings of The Left Hand of Darkness from Ace may also differ in pagination), but it’s my inclination to say that differences in cover art constitute different Editions as well.
I’m not really sure how we should deal with the last example. On the one hand, there is an identifiable difference between one printing and another if the printer’s key is different, while trying to sort out clusters of printings based on cover art and determining which cluster a particular copy belongs to could be quite difficult.
On the other hand, as intimated by some of the examples, a given edition can go to the printers quite a few times, and asking users to sift through 60+ Editions of a book for information could be problematic. It also makes it more difficult to provide complete information. If a dozen printings are identical in all respects save the printer’s key, editors would need to add all of the relevant relationships to each one. We’d likely end up with some Editions with really good information and some with very little, even if the information should be the same. And while some publications will include printing dates, many (probably most) don’t, and I don’t know of any public source for this kind of information, so date fields will go empty.
It seems most sensible to me that we try to find a balancing point between “all printings in one Edition” and “all printings in separate Editions”. Differences in cover art or content beyond changes to the printer’s key seem the easiest, though we could expect problems as non-obvious content differences get missed or editors have difficulty determining whether to create a new Edition. However, it provides a compromise of completeness and ease.
My inclination is to always err on the side of caution; if you have a printing that is different, be it different cover, or just a different/added printing date in the edition notice, we should track it as a separate entity. Right now, we have Editions, so that seems the logical place to store this information, but maybe it’s worth someone considering if there’s a better way to model this.
This is a very interesting topic and really shows the complexities in creating a database like this one.
Maybe we can add fields for reprint number, date, printer’s key (when used, I think that is not popular out of US), … to Editions, for reprints that doesn’t alter any content of the book other than the reprinting-related data?
In your last example, information on the 56th and 57th reprints should be just additional data for the first edition that had that same characteristics.
For Disco Pigs & Sucking Dublin I think it should also be additional data for the first edition with those characteristics, with a note that edition have different ISBN printed. Same for The Commitments.
Left Hand of Darkness have clear different covers, that is a “real” change that makes it deserve its own edition. Same for changes in Graph Theory.
I have the 48th reprint of this one.
On one hand, ISBN matches the one on the database, but the date doesn’t. I dunno if it is an error, or it is a different reprint date. If there was an option to add reprints and dates, I would probably correct the release date to first print, and add the 48th reprint date as additional info in another field.
On the other hand, I know that older prints have a different ISBN, but I can’t know if they had any other differences…