Thanks for finding ...
Indeed, it was the top (rather, um, *wavery*) knot.
Note that this can be tied by inserting a bight tip
into the nipping loop, then "backflipping* it into
final, collaring position --there are 4 orientations
(or more) to this, and I think the one shown might
be best (and can be loaded on either end).
This particular bowline was presented decades
ago by Pieter van de Griend & John Smith in Knotting Matters
(#18, is it?).
--dl*
====