We're getting ready to load item data from a file into our MARC records, and I proposed that the following fields be added to each MARC record: 945 b biblio.biblionumber c biblioitems.biblioitemnumber 946 a biblioitems.biblioitemnumber b biblioitems.itemtype c biblioitems.dewey and for each copy or, in the case of multivolume works, each volume, a 949 field with the following subfields: 949 a items.itemnumber b items.biblionumber c items.multivolumepart d items.biblioitemnumber e items.barcode f items.dateaccessioned g items.homebranch h items.price i items.datelast seen j items.issues k items.itemsnotes l items.holdingbranch .... .... My question is, I came upon this scheme only because it seemed to pass the Check MARC test within the MARC Related section of version 1.9.0, but is this the optimal scheme to use? I noticed that Paul Poulain has proposed that the 995 tag be used for the item data in conjunction with Biblio.pm and I can certainly make this change. Am I correct in assuming that there will be one such field for each volume and copy of a work (i.e., that the 949s or 995s will be repeatable)? What about the information I have chosen to go into 945 and 946 fields? Is this appropriate? We have also tried to get some guidance from the Library of Congress MARC specifications pages as to whether we should either 1) be creating separate MARC holdings records and/or 2) modifying the leaders of our MARC records to indicate that they contain holdings tags. I personally don't think that we have to concern ourselves with either of these specifications, but I just want to be sure. Larry Currie -- Lawrence W. Currie User Services Librarian California Academy of Sciences Golden Gate Park San Francisco, CA 94118 lcurrie@calacademy.org (415) 750-7108 (415) 750-7106 fax http://www.calacademy.org/research/library/