Tis indeed a crazy body of work for someone to take on,
especially when our excellent rope testing team are all volunteers.
One might say that even the testing done just now
is a considerable lot --and I think you'd agree. But
if only one knot were tested and the report out read
"double-eye knots ...", readers should rightly want
to know
which ... ; and my point is that they should
be no less attentive as to details of how knots were
tied/loaded.
Loading of either end has been extensively tested in the past
with at least one study finding
Really?! I'm aware of just two reports that indicate
that the aspect was even noticed : long back, the
Ontario Rock Climbers' Assoc. Safety(?) Manual
asserted that one way (loaded strand bearing upon
its twin) was 6-10 %pt.s stronger; and the Lyon
report --which I surmise is your referent-- came out
as you noted (bearing upon, pulling away from, and
didn't matter (O , 9 , 8 , resp.). The ORCA reference
was for the
fig.8, btw.. Dave Merchant's testing led
him to make some similar claim, though he has yet
another, somewhat irregular orientation of the knot
(still symmetric).
But, for the most part, such aspects are not noted,
and I'll wager not noticed. (And the "with at least one..."
suggests this lack of results : otherwise, we must ask
was there any consensus!)
In the case of the (single eye)
bowline, tying it with
the tail "inside" --i.e., the usually presented, "right" way--
yields a knot that is vulnerable to failure (slipping out)
on ring-loading; tying it with the tail outside (sometimes
mis-referred to as a "Dutch Naval" or "Cowboy" or, by
Ashley, "Left-handed" --to be taken pejoratively-- bowline)
significantly prevents such slippage. (The effective knot,
in such loading, can be seen as incorrectly / correctly tied
Lapp bends.) Though I just saw that if the usually loaded
("active") part is also loaded, both versions can fail
--this, by manual loading of 8mm, so way less than
any fall force (but this is for a particular positioning
of the knot : active line
beside its continuation
into the eye, but not were it beside the tail's side
of the eye).
This consideration of loading applies directly to the current
BotB issue --either the loaded eye feeds into the collar
from one direction, or the other : the vulernability might
exist in both, but be greater in one vs. the other. If one
group tests it one way and the other the other way, we
should expect different results; we should though understand
why the results differ, and not shrug it off as some vagaries
of testing or ropes (though ropes could well have differences!).
In bowline-vs-fig.8 debates, one can see sometimes
people asserting how even if the last tuck of the fig.8
is undone, it is still secure; do they consider that the
different loadings of the knot imply differences on
what knot results from undoing the last tuck?
(In terms of Ashley, it's roughly #1057 & #1058,
though the latter is ambiguously illustrated and not
quite the arrangement that will commonly result).
-SirK*