Character sequence "ï¿½" being inserted into messages by Outlook 2007
It looks like the encoding discrepancy actually was the issue. I had been retesting with old messages that I thought hadn't been corrupted. In fact, it appears that they were already corrupted. Saeed Sepehr's suggestion to test with web mail led to creation of fresh messages and replies in both directions, which were symptom free.
So setting all of the encoding to UTF-8 on both systems was what solved the problem. And now it's verified.
I discovered that another Thunderbird user posted a similar question on the Mozilla support forum in early December. People experiencing the issue had encoding set to Western (ISO-8859-1). Setting it to UTF-8 was at least part of the solution.
An additional recommendation that seemed to help some users (or at least didn't hurt):
Edit | Preferences | Display | Formatting | Fonts Advanced | Text Encoding |
Checkmark the box for
When possible, use the default text encoding in replies. Warning: only do this in combination with setting encoding to UTF-8. Doing just this with encoding set to Western (ISO-8859-1) will exacerbate the problem.
Note that the menu paths have moved around a bit over the different versions, so you might need to find your way to the settings mentioned here and in the question depending on your version of Thunderbird.
It seems you send emails as HTML format but the source software inserts BOM (byte order mark) characters in the body of email. Maybe you can not see them because it's mostly happens for non-English input types because of different length of bits on each characters.
I'm not using Thunderbird but you can check whether there is any option marked as "without BOM" in the encoding view so check it.
Another option which I think it works is using UTF-16 in sender client (Thunderbird) and the same in encoding view on the destination (Outlook).