Legacy Content is about 8 bit Windows and Western encoding and Unicode is not part of this. Unicode is used when the server sends this encoding or you can select it via View > Character Encoding if the wrong choice is made.

Did you check how the server is sending the pages that are shown as a Western (ISO-8859) encoding?

The protocol used to access the files is not HTTP but “file:” (i.e. they are local files). The Firefox page information says it's windows-1252 encoded, but I don't know how this could be infered in a linux system using UTF-8 everywhere. How can I check that my system is sending this information (if it does)?

(The “file” comand output for the documents is “HTML document, UTF-8 Unicode text, with very long lines”, mime type is “text/html” and mime encoding “utf-8”.)

I know I can use the new button “Character Encoding” to change encoding, but I SHOULD NOT have to do it for every file. That it is a real pain in the ass is not even relevent.

If ”legacy content” is not the setting I am looking for, what is the setting to set a default encoding when none is specified?

I am not sure this will help, but Chromium and Lynx are displaying correctly the files.

Bug 1071816 - Support loading BOMless UTF-8 text/plain files from file: URLs

Thanks for pointing me to this bug report. Somehow I managed not to find it.

Too bad there is no workaround in the comments. I guess I'll change my default browser for local files.