[Web4lib] The missing Google books

Crocker, Daniel crockerd at neco.edu
Fri Nov 17 12:45:55 EST 2006

Jim, That's an interesting theory, but doesn't hold true for more than
90% of the missing volumes affecting our collection. Most of the books
we were providing links to were single volume works.

What I've heard back from Google in response to my several inquiries has
mostly indicated that the problem wasn't understood - the first response
after the perfunctory "Thanks, we're looking into it" was about how to
read Google's books when the "Download full book" button isn't

Interestingly, the one book that I thought wasn't affected, Howe's
Muscles of the Eye (
http://books.google.com/books?id=f3qRlNQhRSkC&printsec=toc ) is a
multivolume work, and volume 2 is now gone. 

This means that every full title we'd established links to has
evaporated or half-evaporated. My enthusiasm for Google Books has done
much the same. If you're correct that it is the result of
over-aggressive database maintenance, perhaps this will be fixed at some

Daniel Crocker
Electronic Resources Librarian
New England College of Optometry - Library
424 Beacon St.
Boston, MA 02115
crockerd at neco.edu
(617) 236-6269 
-----Original Message-----
From: web4lib-bounces at webjunction.org
[mailto:web4lib-bounces at webjunction.org] On Behalf Of Jim Campbell
Sent: Friday, November 17, 2006 11:42 AM
To: web4lib at webjunction.org
Subject: [Web4lib] The missing Google books

Searching in Google Book Search seems to be working better today - at
least I'm no longer seeing anomalies in author searching - but the
missing books noted here a few weeks ago still seem to be missing. I've
got a theory that explains some of them and wonder if anyone else has
noticed patterns in what's gone.

The theory is that they're deduping their database.  What I noticed was
missing was volumes in sets. Looking again this morning I see pretty
consistently that in a given multi-volume set, say Ruskin's Works, only
one volume remains.  Since Google Book doesn't seem to track volume
numbers in their metadata, they're probably deduping based on
author/title/date/publisher so Ruskin Works 1889 ends up with one volume
left.  This would also affect journals that publish more than one volume
in a year.

- Jim Campbell
Campbell at Virginia.edu

Web4lib mailing list
Web4lib at webjunction.org

More information about the Web4lib mailing list