I have observed this problem on a 1.8d LISTSERV and I am not sure if this
has already been solved in 1.8e:

When the subject of a message contains an umlaut, a mail user agent may
use an encoded-word since 8bit characters are not allowed in mail headers.
A member of a list who has set the subjecthdr option will receive the
subject line with a subject-tag. When he replies to such a message, his
mail user agent may decide to embed the subject-tag into a new
encoded-word. This way, LISTSERV (1.8d) may not be able to detect that the
subject-tag is already present and may duplicate it. Additionally,
the subject-tag will appear in the archive file of the list.

Example: 

Subject:  =?ISO-8859-1?Q?Re=3A_=5BNONUKA-L=5D_L=F6schen_von_AIESEC_Listen?=

This is a subject line with an umlaut "Ö" and a subject-tag [NONUKA-L]
embedded in an encoded-word which comprises the complete field-body of the
subject field. This line appeared in the archive file "nonuka-l.log0409".

Has anybody observed a similar problem on a 1.8e LISTSERV?

Kind regards,

Siegfried