Post by Joe Bernstein Post by Kevrob
ISBNs, Bookland/EAN #s, Library of Congress numbers, etc.
[long riff on ISBNs]
ISBNs haven't always been unique identifiers. They may always have
been intended to be, but I'd want to see evidence of that long-term
intent, because when I worked at Stars - twenty years ago - I saw two
volumes, unrelated but both at our store so probably both speculative
fiction, with the same ISBN. (Of course, given the structure of
ISBNs, they had to have the same publisher.)
Probably a mistake, either in editorial - the publisher assigning
the same ISBN to different books, while the ISBN Agency [a division
of R R Bowker, publishers of "Books In Print,"] had sold them two,
different numbers - or it was a misprint. Misprints could sometimes
have the correct number on the copyright page, and the wrong one on
the book cover, or vice versa.
We had a copy of an algorithm for testing ISBN check digits. In the
80s, before the 13-digit version became common, ISBNs were composed
of a prefix [0-06 was Harper's main one] the body of the number [in
the case of an exemplar from Harper, -456789- and a check digit, which
could range from 0 to 9, and included X. So, 0-06-456789-X would have
to pass the algorithm to be cromulent. Some publishers didn't seem to
check them all, and tried iterating their check digit in order to get
more bang for their buck. Publishers had to pay a fee for each ISBN.
Note the "Buy ISBNs" menu item?
Post by Joe Bernstein
Unfortunately, I don't
remember which books they were, but a Google search on "non-unique
ISBN" gets a few results. These would've been books that we were
stocking as new books, because the only context in which I could've
caught the duplication would be the data entry I did when the store
computerised, so this isn't a matter of cluelessness way back in the
1960s, but in at least the 1980s and probably the 1990s.
I'm not sure what you mean by "Library of Congress numbers",
Used much more in academe than in bookstores.
Post by Joe Bernstein
that some books published outside the US never go to the LoC, and the
LoC doesn't keep everything it gets. Does it number everything it
But if by that phrase you mean Worldcat's OCLC numbers,
Nope, unless there's a connection I'm unaware of. I took my
only "Library Science" course twice a week as a Freshman in High
School, eons ago.
Post by Joe Bernstein
these are profoundly one-to-many, especially if the book is widely
owned in Europe, because several European libraries don't play well
with others. Worldcat deals with this partly by treating each OCLC
number as an "edition" and creating structures that link them. Such
structures seem, from this user's perspective, to have severe
database flaws. Many are asymmetrical - for example, given a set of
OCLC numbers considered "editions" of each other, some may have
"editions" that aren't in that set, while others don't. Additionally,
as I found repeatedly while studying Shirley Goulden last year,
Worldcat is, or has recently been, plagued with enormous, poorly
controlled "edition"-sets around heavily published titles. (Several
of her titles were linked with, for example. other retellings using
the title phrase "Arabian Nights", which is at least understandable.
However, at the time, others were linked to thousand-"edition" sets
dominated by Russian printings of Pushkin - I kid you not, there were
*more than one* such - and others to other unrelated topics. This
problem had been substantially reduced when I looked at her books
again a few months later, after my laptop was stolen, so maybe
Worldcat is on top of it.)
Also, books exist that never get into Worldcat, though I have to go
to the Near East to name clear examples. I should think someone
better acquainted than I with the new e-book world could name plenty
of American examples. Do those all go to the Library of Congress?
I'm fairly certain that incunabula specialists would be shocked to
hear that "books all have unique identifiers".
I probably should have said "trade books from established publishers."
As someone who had to do a lot of special order buying, I am well
familiar with self-publishers and local publishers who sold nearly
all their books direct-to-consumer who balked at making their books
"store ready," forcing us, if we decided to carry them, to create
our own SKUs. Fortunately, the IBID system I worked with allowed
us to do that, as long as we started the string with a letter.
Our used book folks could replace the first number in an ISBN prefix
with a letter [A-Z] and we could have up to 26 discrete versions of the
same used book in inventory, based on condition, with separate price points.
Post by Joe Bernstein
I think there's lots of evidence that assigning unique identifiers to
books, and then linking them appropriately, is actually quite *hard*.
In the limited futzing around with bibliographic database work I've
done myself, it's certainly looked that way. So while at one level
Amazon, being a company, certainly *should* be able to assign
everything it sells (probably including zero incunabula) a unique
identifier like every other company, it clearly has policies that
require the linking of related products, and being, as the subject
line notes, "the world's largest retailer", it's finding that
freaking *hard*, and I see no reason to be surprised.
Actual, historical incunabula were never anything we sold. Rare books
printed before the advent of the ISBN system in the 1960s? We had a
few of those, but the B Dalton I worked at wouldn't have. BDB used
7-digit stock keeping units, though.
The indie I worked for the longest published a few books in the
1930s, when it was in its first decade. Of course, they had no
Non-ISBN items that were booklike that caused problems were things like
chapbooks. Damned hard to shelve, too. A rack with pockets could work,
but as there wasn't a uniform size, that was a challenge. Then there
were books distributed as both books, through the publisher and book
wholesalers, and as magazines, via the local distributor or the
supplementary national distributors we used. If we carried it as a
book, at a better discount than the mag system gave us, the distributor
might still drop some off, and we had to make sure they didn't reach the
racks, and migrate into the book inventory. Things like THE FARMER'S
ALMANAC would be issued both an ISBN and an ISSN:
Mercifully, we sold mags by pricepoint, not by ISSN. We didn't keep
strict inventory records on those, treating them like produce at the
grocery store. [3 @ 5.95 ea, 2 @ 4.95 ea, etc]
Curiously, "blank books" from sources like Workman Press might have
their own ISBN, even though they might not have any text at all!