RE: ADDENDUM: call number searching

["Jackson, Richard" <rjackson at huntington dot org>]

I haven't had a chance to respond to this thread until now.  The incorrect "filing" of your DVD call numbers is due to the "LC Call Number normalization (with prestamps removed)" being applied to your call numbers (manual #101832), as some have suggested.  This normalization looks for the first occurring pattern of 1-3 letters followed by 1-4 digits and ignores anything preceding that as a prestamp.  Only if it fails to find the pattern anywhere does it resort to character-by-character indexing.  As I understand it, this normalization is not affected by choice of MARC tags or subfielding.

We have the same problem and are still trying to decide what to do about it.  Although much of our collection is classed with LC, we also use a lot of "homegrown" call number schemes for our manuscripts and rare materials collection.  While LC call number normalization is essential for the proper filing order of LC call numbers, it plays havoc with non-LC numbers when they share the same call number index.  For example, our call number "photCL 400 vol1" files as "vol1".

As Bob Duncan pointed out, if you were to search the entire call number (DVD DT 87.45 .N44), you would retrieve the record.  That's because the system applies the same normalization rules to call number searches that it does to the indexing of call numbers.  But a search on "DVD," which logically should bring together all call numbers starting with those letters, finds nothing.  So the result is a situation where the retrievability of a non-LC call number depends not only on the particular pattern of letters and numbers used, but also on how much of the call number the searcher enters.  This also affects what you see in the OPAC when you virtually browse the shelves by clicking a call number in order to view nearby call numbers.

The solution offered by Innovative is to have two call number indexes, one with LC normalization and another with character-by-character normalization.  The two types of call numbers would have to use different MARC tags.  If you do use prestamps with some of your LC call numbers, this may be your best option.  Although we don't use prestamps, we may eventually opt for this, but our Reader Services Dept. is understandably concerned about the confusion having two indexes might cause our readers.

What I would like to see is an option for "LC call number normalization (with NO prestamps)."  This would apply the normalization only when it finds the pattern of 1-3 letters/1-4 numbers at the beginning of the call number.  In other words, it would assume you have no prestamps.  (The enhancement I sent in last year didn't get on the ballot.)  By the way, I don't consider undertaking a reclassification project to be an option, if it's just for the purpose of making call numbers work with the system.

Richard V. Jackson
Catalog Librarian / Database Manager
Huntington Library, Art Collections, and Botanical Gardens
1151 Oxford Rd., San Marino, Calif. 91108
(626) 405-2100 x2384   rjackson at huntington dot org