As the title suggests, this is the place to discuss the theory behind designing and setting up a knot test rig.
I have offered up 3 different configurations to start discussion...
In the first image ('onesided') - the knot specimen and rope will s-t-r-e-t-c-h ...
For any variety of reasons,
I'm willing to go with Derek's dismissal of concern
re this #1 set-up being biased; at the very least,
it might be so at a so-small/negligible level as to
be totally irrelevant. Maybe in a drop test, not so!?
THIS set-up (#1) is most
material,
time,
and machine/force-generating
efficient. For S # of specimens (each w/2 knots)
you run S # of tests
and get 2xS number of results :
i.e., breaks of S knots,
and survival of those various breaking forces
by the other S knots.
IMO, this is best.
It might not be the charm of statistical purists,
but I think that the commonly seen test-FIVE
specimens regimen falls also well shy of giving
much of a high confidence statistical record,
and as I've previously said, having a biased-LOW
strength value is probably to the better.
(And, e.g., if one is comparing knot-vs-knot
--relative rather than absolute strengths--,
the bias will be there for all candidates,
just as arbitrarily multiplying the values
by some constant could be.)
The idea behind using duplicate knots --one at each end termination--
is thought to be originated by Dan Lehman.
Dan Lehman's idea was to have a 'survivor' knot specimen
(one always breaks, leaving a 'survivor' knot to examine).
And to have a more sure, biased-lower strength mean.
Indeed, I'm very happy that my few tested
eye knots in 5/16" 12-strand HMPE were with
surviving knots (all knots having been *marked*
with embedded gold,pink, white threads so as
to gauge locations in the SPart & tails)!!
In one test (per knot), what was I going to do
with a single (even two) break value? But with
a single survivor & a broken one, well, I can
believe that there could be variance even here,
in terms of the *where-it-broke* evidence,
still, getting SOMEthing to examine,
irrespective of the value,
IMO is a considerable value.
--and i.p. I so much would NOT have preferred
the same knot, even if 2per-specimen, to have
been tested, so as to make a set of values for
statistical manipulation.
--dl*
====