So... as far as I can tell the apostrophes-to-question-marks behaviour has something to do with the way Mailman processes MIME-encoded messages.

The sample message that Dylan sent to the Servers list (Innilgard_ Event Formerly Known as Spring Crown Tournament, 11-12 September.eml) has

Content-Type: text/html; charset="utf-8"
Content-Transfer-Encoding: base64

I've verified that running the message content through base64 -i -d produces some HTML that displays the apostrophe correctly in a text editor using UTF-8. (It does, however, fail to decode without the -i, but this could be due to way I cut the text out of the message.)

After passing through Mailman, the relevant headers become just

Content-Type: text/html;charset=utf-8

with the text no longer MIME-encoded and with a question mark replacing any non-ASCII characters.

Messages sent directly from my e-mail clients have

Content-Type: text/html; charset=utf-8
Content-Transfer-Encoding: quoted-printable

which passes through Mailman unchanged. Roundcube sidesteps the whole issue by changing Æ to Æ and the apostrophe to ’ in the HTML version of the message.

I haven't found where Mailman handles MIME-encoded messages and Selene has just gone down again so that's it for today.


------ Original Message ------
From "Ælfred se leof via Selenetest" <>
To "nico--- via Selenetest" <>
Date 12/02/2023 3:14:19 PM
Subject [Selenetest] Re: Testing apostrophes

... and testing from a different e-mail client: Ploughman’s Lunch

------ Original Message ------
From "Ælfred se leof via Selenetest" <>
Date 12/02/2023 3:09:54 PM
Subject [Selenetest] Testing apostrophes

Testing: Ploughman’s Lunch