I looked through the logs. At 12:46am pacific time, we received anInteresting. Looking at the View Source in both groups, the header fields have byte-identical content, which makes sense I guess given that they came in on the same connection. Identical ID means that Roadrunner thought they were the same message, which isn't a surprise (that would still be true if I had done something like a BCC).
But the order of the fields is radically different, which is weird.
Were there any other messages in that connection? My prior message to beta, #1096, went out in the same batch from Eudora to Roadrunner, and its timestamp is only one second earlier in your server's Received from Roadrunner, strongly implying that it might have been in the same connection.
Is it possible there was a glitch during the connection between Roadrunner and your server, causing Roadrunner to think the first attempt at the owlperch delivery failed, but your server accepted it and somehow mistakenly applied the destination address of the prior message in the connection?
Nightmare scenario I'm sure. This makes me all the more suspicious of those "duplicate message" bounces I received. I think those may represent three more instances of the same glitch, but those were caught because the same group was the destination.