I used named entities in HTML docs that I converted via Mobipocket Creator and they didn't work; just showed the code rather than the character. I ended up inserting smart quotes directly in the original HTML document. (And yes, by "smart quotes" I mean angled quotation marks, single and double--that's what they're called in Word.) I guess that's why I didn't think of it when making an XHTML document for ePub.
I did use UTF-8. I wonder why the pages didn't verify, then? The error didn't name them specifically, just said that there was a character it couldn't read. I assumed it was the quotation marks, because when I took them out there was no problem. I used the verifier on the w3.org site, uploading the document. I just used regular Notepad to make the files.
I don't own a reader that can read ePub. Do most of the readers parse named entities properly? If so, I think that might be the way to go.
When I'm coding regular HTML, I just use the straight quotes. But for something I'm going to read on a dedicated reader, it should be as book-like as possible, in my opinion, especially when trying to convert the paper fetishists!
Thanks for the advice.