Here C-u g shows us the actual bytes in the *Article* buffer,
which you will recognize are UTF-8. Whereas the same bytes become
different bytes in the *Summary* buffer. We are talking about cases when
a raw UTF-8 Subject has arrived.