the Sheet bend has no nipping loop as the bowline.
The loop or coil of the Sheet bend most certainly "nips" the U shape of the rope.
The distal part of the U shape helps anchor that coil so that it can provide nip.
It is more or less, resp., of opposite-side or same-side (recommended)
sheet bends, in that the mechanics of the "loop" here
are of loading-from-one-direction and not of both ends
(of the loop) being pulled upon (50% & 100%, with friction
leading to equalization at some point, in normal cordage).
One can suggest that the
same-side ("proper") sheet bendis just a tucked version of the
thief knot , to put it in
perspective --of more nearly, in effect, U-2-U workings rather
than U-2-loop. But this is getting awfully picky.
As for security and testing,
some arborist did test the
zeppelin eyeknot in a kernmantle
("static", low-elongation something or other, 8-11mm, IIRC)
rope, and found it stronger than some
fig.8 he also tested
("some" meaning that the exact orientation wasn't obvious).
It held to a high rupture value, so obviously didn't slip.
Meanwhile, we have the Dave Richards testing of various
kernmantle cordage (7mm accessory, 10.2? mm dynamic,
and 12.7mm low-elongation) to show that the
sheet bends(single & double) were weaker and less secure --sometimes
needing stopper knots!-- than bowlines.
Or would you like to claim that a bowline ceases to be a bowline
as soon a rotating, high-friction load causes only the leg belonging to the U shape to be loaded?
Well, this certainly points to issues in defining "knot",
and why I use ' *knot* ' often, to alert one to some
problematic definitions. For, surely, in the loading above,
one has an effectively different physical structure and
should expect associated behavior if so! And I do wonder
if by such effective loadings those trawler hawser
bowlinesget capsized; the capsized forms are undeniable, but
the path TO them is unknown to me, up for inference.
--dl*
====