Bib - Character Encoding
What is this?
MARC records may be stored with their characters encoded in either MARC-8 or Unicode/UTF-8. The choice is stored in the record as LDR/09 and all characters in the record should use the bytes for each character as specified by that encoding scheme. For many records, the bytes will be the same, but they will differ for characters with diacritics (like accents) and characters in non-Roman alphabets.
When is this trouble?
Some library catalogs will accept records with either encoding, but some will only accept one or the other. Some catalogs will check that the encoding is the one they expect as you're importing a record; others will not, and will just try to interpret each record as the encoding they expect. For many records there is no difference (so it may not be caught immediately) but any characters with diacritics or non-Roman characters will be garbled. The results can look terrible in stored MARC records and in the online catalog:
How to Find
If you are in Alma, there is an analysis to search for this issue in the "Looking for Trouble" folder. Go to the "Bibliographic" folder and look for the "Bad Diacritics" analysis. This analysis searches common descriptive fields (title, author, etc) for common indicators of bad character encoding, such as <U+fffd> and �.