>>>>>>> the full butterfly knot
(see attachment)
Where did THAT fly in from ?! --quite impressive!!

Dan Lehman, the figure 8 ring is a really interesting idea, I'll definitely look into that as well. Also, I would love to know why Germans would use dual loops. Their climbing literature is the most authoritative I've found and I really enjoy it. Do they finish their BoB's with a backup knot?
1) The finishing with a "back-up" knot might in some
cases be simply viewed as "tying off an adequately long tail
out of the way" more than some thought that security
needs it. But in some other venues, it might be a good
plan for anticipating a major goof in the main knot
coming for whatever reason, esp. understandable in times
of stress & with maybe limited hands-use & a tricky rope.
2) The additional eye I will surmise is made for simple
redundancy. If the 2nd eye comes unsecured --i.e. the
tail comes out of the knot--, then there is an assuredly
lonnng tail, and one that while still threaded through
the harness should flap about and thereby warn the
climber of that degree of loss of knot integrity.
(When looking over such 2-eye knots, one needs
to remind oneself that there should now be only
25% load per eye leg (!), half of the usual half
of the load --and what might THIS imply ... ?! !)
I will probably end up continuing to use the figure 8 & strangle knot backup - I would climb on any knot, even a half hitch, if it had a strangle knot backup. I like the symmetry of the loop, plus I don't want to befuddle my partners. But I will also keep experimenting with the suggestions you provided.
3) I recall on one knots discussion on RC.com (back when
things were discussed there) where a woman climber
reported that sometimes here
strangle knot --of course
not actually using that name(!)-- came untied (!). !?
You should try the
offset fig.8 eye knot I've described
above, and might even prefer the slightly smaller
offset overhand (EDK) version,
where the 2nd eye's returning leg just gets tucked
out through other eye legs --or whatever looks snug.
(Though in using it qua belay loop, one needs to take
care NOT to use just one --or not the wrong one-- eye!)
As for being happy to tie off with a hitch,
and taking assurance from some secure safety knot,
I would really want to see some tests in which these
were what held :: what I'm thinking of is how much
damage(?) might be done by slippage of the rope
through the harness as the hitch tightened or as
the failed main knot shifted forces onto the safety
--and got some similar slippage of rope through
the harness nylon.
One might use a hitch qua eye knot :: Heinz Prohaska
proposed adopting the
anchor bend as the to-become
eye knot base, as it --like his
Prohgrip/Blake's hitch--
could pretty securely grip just a tucked-through-once
strand, should one have a "Lynn Hill moment" and
have tie-in interrupted and not concluded (at just this
tucked-the-end-once point). The finish would be some
wrap & exit like that for the
bowline.
--dl*
====