All errors are due to my faulty memory
Note: The Shelf list = LC and Dewey classified titles, and also items with an accession number. The shelf list did not include SuDocs, Swank or other classifications used in Government Documents. The shelf list included classified serials (LC, Dewey or accession number) but did not include unclassified titles. All serials were in the Serials Catalog. Some content was severely underdescribed; the problems of these hidden collections were described in a 2006 internal report.
The University of Iowa Libraries started cataloging on RLIN in 1980 (and RLIN remained the preferred source for copy until the 2006 merger of RLG and OCLC). All materials cataloged from that time forward had a MARC record. Titles that needed recataloging (especially serials) from 1980 on were converted at the time of recataloging.
In the 1980s, temporary librarians were hired to do retrospective conversion of the non-serial shelf list, inputting information from cards into RLIN. They may also have searched for copy; the details are fuzzy as it was before my time. After the grant money ran out the project ended (despite having well trained staff that could have completed the project in another couple of years.) This project ended shortly before our June 1988 launch of NOTIS (locally branded OASIS). (And for a real blast from the past, take a look at this 1991 video introducing the catalog http://ir.uiowa.edu/lib_ar/17/)
Just before and after the launch of OASIS, serials recon on the currently received titles was done by deriving records in RLIN and later by students marking up enlarged copies of cards to indicate correct fields and inputting them to OASIS. Note that latest entry unclassified titles were not changed to successive titles. See also Ruth Christ & Selina Lin (1992) Serials Retrospective Conversion, Cataloging & Classification Quarterly, 14:3-4, 51-73, DOI: 10.1300/J104v14n03_05
Serials retrospective conversion continued as time allowed by serials catalogers searching for copy in RLIN (and later OCLC) or creating an original record from the card. In this phase, pieces were used as needed. During the years following the first project, many of the records were input from cards were overlaid with more complete records from RLIN and later OCLC.
Monographs in analyzed serials (i.e. analytics) were also done in house. The serial title was in the shelf list, but the individual titles were not. These titles were listed on contents cards in the serials catalog. This retro was completed as part of our dual linkage project (adding a field to the monographic record to link to the item record and circulation information on the serials record, making those items records linked to two different bibliographic records for display)
In 1999, connected to the barcoding project and migration to Aleph (locally branded InfoHawk), all remaining non-serials in the shelf-list were outsourced to OCLC for retrospective conversion over three years. There were about 2.2 million records in this project (see http://ir.uiowa.edu/lib_ar/10/ for a little bit about the project status in 2001) . Shelf list cards were sent to OCLC for conversion. The project parameters allowed for something like a 95 or 97% correct record match rate, with most of the errors being the edition/printing problems. (A 3% error rare would be more than 66,000 titles). However, the old OCLC title search 3,2,2,1 was used and if only one record was found, that record would be considered a match, even if the titles were completely different. This was not expected to be a high percentage of problems, but even 1% of the problems would yield 6,600 completely incorrect titles.
I don’t think “smart” barcoding instructions specifically looked for these title problems, but problems definitely were noticed. (Smart barcodes is the term when the barcode numbers were assigned to records and the author/title/call number printed on the barcode, and students applied the “smart” barcode to the books). Some of these problem titles were noticed by patrons or library staff and corrected. (Particularly memorable to me was our first encounter with the problem when a colleague excitedly saw a record for an obscure Elizabethan embroidery book and on the way to the shelf realized the call number was odd for the subject and we were all puzzled by the book she returned with.) Correcting these became routine for database maintenance staff (and was regularly done while on the phone with a concerned librarian).
We know some of these completely incorrect matches are still out there. We have gotten corrections from Google because our bibliographic data did not match the book they scanned. Presumably the check of the book and record before materials go to our offsite storage facility will find more of these problems.
Other articles of given a bit of the history of our serials department
- Ruth Christ , Mary Monson & Marjorie Wilhite (1990) Serials in Transition:, The Serials Librarian, 19:1-2, 57-67, DOI: 1300/J123v19n01_06
- Mary H. Monson (1992) The NOTIS Opac and Serials, The Serials Librarian, 22:1-2, 137-163, DOI: 1300/J123v22n01_10
- Mary Monson MA (1987) The Use of RLIN for Serials Cataloging, The Serials Librarian, 12:1-2, 157-168, DOI: 1300/J123v12n01_15