We find trailing colons on some site's domain names as if there were to be a port number but none is specified. We note how this might happen, where we see it has happened, and note places where this causes trouble.
how
A lineup specified as site/slug will accept a trailing colon and treat it as a remote site. try-it 
http://fed.wiki/fed.wiki:/welcome-visitors
If forked, this "remote" address shows up in the Journal where it behaves as an alias for the unadorned site name.
where
We've seen this appear Hitchhiker sites.
"art.hitchhiker.academy:", "david.pod.fab.fish:", "david.self.fish:", "david.vision.hitchhiker.academy:", "david.vision.towel.fish:", "david.wigtown.club:", "feast.fab.fish:", "graham.pod.fab.fish:", "yvette.vision.towel.fish:",
trouble
We've seen alias site names fool federation search into indexing the same content twice. There are places where we explicitly remove the www prefix.
We have seen operators configure site to respond to IP addresses as an alias. This waist is on them.
The stray colon can arise for any site because it depends only on the habits of some author. This also tricks the Roster plugin into thinking a site name is free text legend rather than an aliased site name.
Roster reported by federation search's activity/