Here is a very interesting link to download a test report on fixed eye knots:
Link: https://www.researchgate.net/publication/344689340_Revision_of_Commonly_Used_Loop_Knots_Efficiencies
...
Summary:
[ ] The ongoing speculation about the differences in MBS yield between various geometries of #1047 Figure 8 can finally be put to rest.
There is no statistical difference in MBS yield from one geometric form to another (ie where the SPart is positioned within the knot core).
This is jumping to a conclusion absent good,
well-focused attention :: I submit that we must
SEE the respective knots near crunch time so
to understand their results; actually, we need
to see their pre-loaded, tied-&-just-SET knots.
As I've stated elsewhere, IFFFF there is to be
the advantage to the "I" (my "EXterior") loading,
it would come by virtue of having SET THE KNOT
HARD TO PUT CURVATURE & SUPPORT OF THAT
into the knot --i.p., to have set by hauling hard
upon the TAIL, and upon the eye; to leave the
SPart rather UNset so to have to get all of its
bending from actual loading, other parts now
braced to resist.
And the extent to which this process has effect
might vary per rope --greater in firm, less elastic
rope, less so in other?!
Note that they find no gain from not 3 but FOUR
dia. in SPart's turn --the double-eye knot not being
stronger:
The difference in the efficiency of the standardly loaded double figure eight loop
compared to the referential figure eight loop is statistically indistinguishable at a
significance level 5%. This conclusion is supported by atotal of 44 measurements
performed on the same rope in both geometries O and I. Based on these results,
we can conclude that double figure eight loop does not perform better than a much
simpler figure eight loop.
[p.411]
And here I'll suggest that pretty much however
one ties the "bunny ears (two eyes) fig.8" there
will be a diameter or more gain in material that
the SPart U-turns around at its critical point
--in contrast to having to make the careful
dressing & setting I describe above--,
so this result puts the hard question to my thinking
of 3dia bowlines as being a plus.
Again, though, we must SEE what is going on.
Maybe we'll see how sometimes the extra material,
the extra diameter(s), gets pushed aside, out of
where we thought it would give some good effect.
(Certainly in the case of casually set Fig.8 eye knots
one can see the exterior SPart pushing quickly to a
quite-tight turn.)
BTW, this is not to say that I think that having the
stronger form is of importance --one might prefer
the other (by my assertion of strength, weaker),
for ease of untying; and realize that all are equally
"strong" in the sense that "ropes don't break"
observations made over many decades of actual
usage. But it's a behavior worth knowing/understanding.
--dl*
====