Here’s a ‘Sail’ magazine article that was posted on the Stagecraft mailing list:
http://www.josephchansen.com/ropetest.html
Cheers
Thank you Tom - an interesting article - I hope you noted how the Bowline was tied!
I am now eagerly awaiting Dan’s comments
Gordon
Yeah, it was like a mirror image of a Dutch bowline.
It’s too bad they didn’t test other knots while they access to such nice equipment. But, I understand the cost of destroying rope limits the set of test subjects. It’s about what one would expect from a magazine article.
Imagine doing all that and then not using any repeats! I know line is costly, but they make it right there for crying out loud! An average over, say, fifteen pieces, and then a comparison with five more pieces tied by a different person would have given something useful to look at. It was very telling to me that they did not even comment on the fact that they managed to improve the strength of their own line by splicing in the same way that they normally do - just gives you some notion of the variability of tests don’t it?
I will relate this tale to you as I relate it to others in my classes:
If I stand with one foot in the freezer and one foot on the hotplate, on average I am comfortable!
Interesting but not terribly useful… 
SR
Here are some more tests for you to comment on:
http://oberon.ses.nsw.gov.au/resources/VR%20Load%20Testing%20Wellington.pdf
http://oberon.ses.nsw.gov.au/resources/Load%20Testing.pdf
http://oberon.ses.nsw.gov.au/resources/Equipment_Testing-Kiama_VRPDW.pdf
http://www.xmission.com/~tmoyer/testing/Water_Knot_Testing.pdf
http://www.xmission.com/~tmoyer/testing/High_Strength_Cord.pdf
http://www.climerware.com/knot.shtml
>:( Had a reply nearly finished, the somehow poofed it with careless keystrokes … 
No repeats? They claim to have done a trio of tests per knot per line, which is not nothing.
Other sources sometimes will do five per … , and then feel justified (?) in taking statistics.
Look at the Dave Richards (of/for Cordage Institute, reported to ITRSymposium?) results
for five, and muse over whether another 3-5-10 would affect your thinking of that.
and then a comparison with five more pieces tied by a different person would have given something useful to look at.It's a good point to make: how does testing of [i]this particular tied geometry[/i] bear on the actual usage somewhere else, tied who-knows-how, etc.. Unremarked in the article (typical!) or above, here, is the again testing of a parallel-strand/"trace" knot w/o specifying which of the twin parts was loaded (which parallel end). In one case (what I've called the "strong form"), the loaded part bears against its twin (significantly); in the other, is bears on other parts of the knot. I'm no longer so convinced that "strong" is other than a guessed moniker, with some variance in setting coming into play in either case. But it would be nice to at least recognize the distinction and specify what exactly was tied.
In my lost reply, I opined on the particular form of the Fig.8 shown on the cover page. Know what,
I found that opinion hitting walls! I had thought that it was obviously a “weak form”, with that ruptured
end (pointing leftwards) having wrapped back around & down, and so bearing away from the twin
upper parallel part. But how is it then that the flowering end there (clearly CUT), which would be
an END (unloaded), be sooooo short–one wouldn’t do this, methinks, on any tying. Also, the tightness
of the lowest wrapping part seems to indicate loading–as in the strong form (bearing against the twin),
and how can one SPart be one way, the other the opposite way?!
.:. I’m coming to conjecture: THEY ACTUALLY TIED & TESTED A >>>LOOPKNOT<<< !! --and did so
thinking it okay as an indication of the bend!? Well, maybe not: I was thinking also that the CUT
came in order to remove the knot from the pin, but we see a splice not on a pin, and maybe the pin
can be looped over, with a closed eye, and then secured? (It sure would be awkward–unthinkable–
to have to splice eyes around … .)
.:. But then the knot certainly appears to be imbalanced/asymmetric, on apparent loading–of the
outer turn at bottom, but corresonding inner-part’s turn where ruptured. And in that particular case,
I find the point of rupture to have apparently come at nearly tangential contact–I can’t believe this.
.:. So, again, we don’t know. Hard to grow understanding this way, but some more testing burned … .
In some of the Australian test reports cited above, one has a little better handle on the form of the
Fig.8 loopknot, which appears to be in “weak” form, the unloaded twin being away (eyewards) and
quite untensioned (though at times I’m wondering if it might’ve been a strong form and the loaded
turn just pulled up close … ).
It was very telling to me that they did not even comment on the fact that they managed to improve the strength of their own line by splicing in the same way that they normally do - just gives you some notion of the variability of tests don't it?Well, it raises the question about the relevance of the [i]provided-by-vendor[/i] tensile basis for taking efficiency %-ages, rather than getting this too by testing. And I'm also surprised at the relatively low figure for the polyamide rope!?
If I stand with one foot in the freezer and one foot on the hotplate, on average I am comfortable!But your standard deviation is large--something they didn't give, or even a range to accompany the averages reported. Frankly, I'd much prefer that testers did away with the "percentage lost": if one if figuring values for a rigged structure, the direct method takes the efficiencies times material strengths (56% of 8_250#, say), and not an additional computation to subtract ... . Most knot strength data comes in this way, and it's an annoyance when it comes otherwise (and can lead to some hasty mistakes where knot noted--esp. when values are in the 40-60 range, now seeing 60% (and more!) being for what is LOST!).
“Yeah, it was like a mirror image of a Dutch bowline.”
Huh? A mirror doesn’t change what the knot is. (It might have some influence in laid
rope, but those effects are not well studied & reported.) --reminds me of Pieter van de
Griend’s apparent distinguishing “Picketline H.” from “Groundline H.” in a recent PAB
“Knot News” article; that leaves me really wondering … !
But I too remarked at that photo, but now wonder how much stock we can put in it?
Afterall, the Bwl ought to be pretty common fare, and tied “right-handed” in Ashley
terms. Dave Richards’s test report remarked at testing the “Cowboy” form and finding
it about the same, in the kernmantle ropes he used. There’ve been rumors of both
equality and difference.
“It’s too bad they didn’t test other knots …”
Yes, or even refer to other testing of the knots they did test (and also, as noted, give
details on actual forms). It’s helpful to see the slippage for the standard RT&2HH,
but they might’ve tried putting on a 3rd HH, or a stopper knot, or some other means
to stemming the slippage, so there’d be something to recommend.
And, btw, is the Grapevine (Dbl.Fish.) a common knot for mariners? --I can’t say
that I’ve seen it used much, ever, come to think of it, around boats. Rather, the
(single) Fisherman’s knot is common, and maybe also a Sheet bend. (It would’ve
been good to see if slippage for these occurred, as it did for the kernmantle in
Richards’s tests.) For that matter, I don’t think that the Fig.8 bend (or LK, either),
is in much evidence in the sailing/marine world–rather, the Ring bend (& Overhand LK).
But how often are lines joined? (In climbing: for abseil (to be untied); for making slings
(not to be untied).)
–dl*
I was only commenting on its appearance.
An article on the topic from the physicist point of view (assume a spherical cow 
http://physicsweb.org/articles/news/11/4/2/1