Here's another interesting bit that may be of interest. By my count there are 3848 undifferentiated name records that contain non-vernacular 400 fields in them (and not a combination of vernacular and non-vernacular). Should it be found proper automatically to tease undifferentiated records apart, in all but 478 cases all of these non-vernacular 400 fields can be matched up with the texts of 670 fields, and so (in theory) could be sorted out to the appropriate record(s).
As an earlier report noted, almost none of the vernacular 400 fields can be found in 670 fields because they were created from information in bibliographic records and are not "justified" in authority records. One suggestion for dealing with these might be to attempt to find bibliographic records containing vernacular fields whose 245s match titles in the 670s. Any such work would be complicated by the switch in romanization schemes (which I think was not applied to authority 670 fields). Other than that, I don't see much hope for any automatic handling of vernacular 400 fields (either keep 'em all in each split-apart record, or discard 'em all), though I would be happy to hear of other ideas. There are by my count 13,386 records at present with vernacular 400 fields in them.
Gary L. Strawn, Authorities Librarian, etc.
Northwestern University Library, 1970 Campus Drive, Evanston IL 60208-2300
e-mail: [log in to unmask] voice: 847/491-2788 fax: 847/491-8306
Forsan et haec olim meminisse iuvabit. BatchCat version: 2007.22.416