All:
Great comments here. I've had discussions on truth in historical fiction in numerous forums, the latest at a book signing event for my 8th-century BC novel. It really generated some interesting Q&A from the audience.
The guideline I've always used is, if I can ascertain imperically to my emotional and intellectual satisfaction that something
didn't happen, I won't use it. For example, I wanted to use an eclipse (preferrably solar) as a celestial omen in my latest novel. I was able to determine through NASA's Web site that such an eclipse did not happen at the time of interest in my story. Therefore, I could not, in clear conscience, create one. Would a reader have gone to the trouble of plugging my ephemeral formula into a spreadsheet and research the historical accuracy of the event? Not likely. But for me, it was over the line, a matter of integrity--and all I had to have was one question at a book event asking, "Did that really happen?" to get red-faced in the admission that it did not. Then I'm copping out and getting sloppy in facilitating my story. (Happy ending: there
was a penumbral lunar eclipse visible from the location at that time, and I used it.

)
Feasibility in unverifiable circumstances is another story and carries a lighter shade of gray, I think. The thing I don't want a reader to walk away from one of my stories with is the assumption that a contrived/fictional scenario was a true one. You can't always help that--and,
arguably, readers who appeal to fictional stories alone as proofs of historical truth (which happens) deserve what they get. To help avoid such mistakes, as an example, I compile a character list at the beginning of my books (with maps and glossary, if necessary) and identify which are historical and which are fictional. Maybe helps a little; who knows?
Any other purists out there...?
Thanks again for the discussion. This was fun.