How quickly you forget about (nearby of your papers)
the offset water knot --which produces the most often
Found-In-The-Wild mid-line eye knot.
(But you might want to examine other in-line eye knots,
such as the fig.8 and so on.)
You should have hit your refresh button before jumping the gun!
If you read more carefully, I stated 'of this type' - meaning of the type formed from inter-linked overhand knots.
To clarify that remark - I then amended the paper to include words to that effect the next day after uploading.
Anyhow, thats the way the paper reads. Have another look to check this for yourself.
...
l00ks like you got careless on your
"tied with bights" variation
As for the alleged 'carelessness' in tying the Zeppelin bend 'with bights' - it doesn't appear to make any difference as to precisely how the tail segments lie with respect to the SParts.
What matters is that you begin with 2 superposed bights with opposite chirality.
The alleged 'carelessness' lies in the eye of the beholder.
p.22 (of 32). Oh, my, that "toggle axis" ...

Bit much of X. coming through here,
but in any case, one can (also) see that
--and esp. where the material deforms more,
or wasn't dressed & set snugly--
the SParts can mimic the bowline's nipping turn.
(YMMV.)
Jumping the gun again...Xarax theory is based on the principle of a hinge - about which each 'half' of the Zeppelin bend pivots.
I substituted 'toggle-axis' for 'hinge' - but the principle is essentially the same.
The tails are crushed together and the SParts pivot about them. Because the SParts are not in direct alignment with respect to each other, there is a slant - which you can see in my photo.
The blue rope is displaced to the right, and the white rope is displaced to the left. When force is applied, it causes the 'toggle-axis' to slant.
I don't believe there is any 'mimicking' of the Bowlines 'nipping loop'. The Zeppelin bend functions differently.
Obviously, your alleged mimicking of a 'nipping loop' cannot be so because both ends would need to be loaded (which they aren't).
A nipping loop is loaded at both ends.
For example, in #1431 Sheet bend there is no nipping loop (because it isn't loaded at both ends).