rtf encoding changes between documents

Sorry for another technical question about rtf generation, but I am still trying to track down bugs in my compile process.

I have a long document that I am working on that I compile to rtf. The rtf is then processed through Sente, and then examined in Nisus Writer. In this document (the “legacy document”) I find that

the ñ character is encoded as “\u241 ?” or “\u241[CR]?” (where [CR] is a carriage return)
the á character is encoded as “\u225 ?” or “\u225[CR]?”
italics are encoded as "{\f3\i Italicized text}
endash is encoded as "\endash "
paragraphs are indicated with \par

However if I create a new document, paste the text from the legacy document into it, and compile with the same settings (I think) I get

the ñ character is encoded as “'f1”
the á character is encoded as “'e1”
italics are encoded as "\i Italicized text[CR]\i0 "
endash is encoded as “'96”
paragraphs are indicated with “[CR]”

There must be a switch somewhere that controls the “style” of rtf, but I can’t find it. While these two documents do look the same in Nisus, the first triggers bugs in Sente.

Thanks

Special character encoding is handled at the OS X level, not by Scrivener itself. What happens if you paste the legacy text into a TextEdit file, then process the resulting RTF file as you described?

Katherine

Thanks Katherine for your reply.

I am rather befuddled, but as far as I can tell, I must have hit something in Apple’s code that was then changed.

I looked through my backups, and, looking at the same rtf file in my Scrivener document, which I haven’t edited, I see that the word “Tiburón” is encoded consistently as “Tibur’f3n”.

However, files that I compiled to rtf on Sept 24 encode this word as “Tibur\u243[CR]?n” or “Tibur\u243 n”, files I compiled on Sept 10 encode this as “Tibur\u243 '97n”, but files I compiled today, Sept 27, encode this word as “Tibur’f3n”.

Something very strange was going on. As I said, the only reason I cared was that one of the encodings was causing problems for Sente (the Sept 24 version). But if things have settled down, then this is no longer an issue and I can ignore the rtf coding.