So all three of those look fine in my desktop clients but rubbish in Roundcube.

As far as I can tell, it’s Mailman itself doing the conversion of non-ASCII characters to hex codes. At least, the HTML-to-text programs I’ve tried work exactly as expected when invoked with UTF-8 input from the command line. Mailman’s documentation is quite sparse on the subject of character sets, but as far as I can see, English lists are assumed to be encoded with us-ascii, and the way to change this is not documented.

For now, I’ve switched off HTML-to-plaintext conversion for this list.

    Ælfred

From: Lochac Masonry Deputy via Selenetest
Sent: Wednesday, 16 March 2022 2:03 PM
To: nico--- via Selenetest
Subject: [Selenetest] Re: Testing HTML-to-plaintext withlanguage.en.charset=utf8

And one last test with charset=iso-8859-1

From: Ælfred se leof via Selenetest

Sent: Wednesday, 16 March 2022 1:53 PM

To: nico--- via Selenetest

Subject: [Selenetest] Re: Testing HTML-to-plaintext with language.en.charset=utf8

And back again with charset=utf-8

------ Original Message ------

From: "Lochac Masonry Deputy via Selenetest" <selenetest@lochac.sca.org>

To: "nico--- via Selenetest" <selenetest@lochac.sca.org>

Sent: 16/03/2022 1:37:01 PM

Subject: [Selenetest] Testing HTML-to-plaintext with

language.en.charset=utf8

>A non-ASCII character: Æ

>

>“Smart” quotes and ‘apostrhophes’