My only thoughts are that there is something fishy going on between Excel and MS-Windows and character sets. *I* never had any problems on my Linux system creating CSV files that ‘just worked’, but I was not using a spreadsheet program, instead I was using a text filtering program (coded in Tcl) to filter from one ASCII format to another ASCII format. Or else I was using a plain text editor. I don’t use MS-Windows for my desktop system (and never have).
I think that somewhere along the way, UTF-8 might be involved and possibly something odd about newlines or item quoting (eg missing quote marks or improper quote marks), but I don’t really know. It very likely also depends on the locale setting on the *server* WordPress is running on vs the locale setting on the desktop computer the CSV file was created on. Most often, a pure, 7-bit ASCII file will have no problems, ever. I know that sometimes that is not a viable option, since book titles, descriptions, and author names might possibly require UTF-8 characters. It is important *never* to use anything but 7-bit ASCII characters for the bar code: eg only the digits 0-9, and ASCII letters A-Z and a-z — note: most bar code scanners only do digits (0-9), so if you are in fact using a bar code scanner, bar codes can only be digit strings. Note: ASCII is generally a proper subset of most UTF-8 character sets, esp. for Latin character sets.
