In reply to GrahamD:
> How can you out perform the simplest possible knot to tie (with or without gloves) which never comes undone and which has the least chance of getting caught ?
!! Spot on !!
But w/some disdain at the offset water knot's usual baggage of "leave tails long", DO something with those tails (or the one, as RGold & I show) and be done with even the hint of doubt --tie the bloody tail off instead of leaving it long!
Yeah, folks like numbers, as though the knot holding a 2 cars is better than that holding 1.5 for an abseiling task. But numbers are so easy.
As I try to emphasize, for THIS use, each person should be able to do relevant testing (body weight on single strands and so on); don't look to some break tester's numbers.
Your second link links through to some info from some Edelrid tests:
http://www.gudelius.de/spst.htm
> It's got a couple candidate abseil knots I hadn't come across before - a double-T fisherman's (a double fisherman's tied with the loaded ends on the same side) and what they term a triple-T overhand knot (looks like a single fisherman's with a locking knot). In their tests both these out performed a single overhand knot. Shame no test against a double overhand.
Josh's names for knots are not to be replicated, please. Clyde's "offset" works fine here, with the only debatable issue of how one might care to preceive Josh's novelty (something earlier proposed by Austrian Heinz Prohaska & shared by him with Bachmann, to drop some names
. That knot could be seen as an offset (single) fisherman's knot backed up with an overhand, or as the same *guarded* by an overhand --i.e., the first-resisting knot seen as protecting the fisherman's from offset loading. (FYI, one can form the three overhands in any order!) He might have also tried putting the two overhands of the one line adjacent, with the larger rope's overhand on the tails side; that might've resisted the pulling through (which we'll note came only at a load not to be seen in abseiling).
The offset grapevine (aka "dbl.fish.") shows some inefficiency : the added turns in the double overhand structure (or "strangle knot" component) work well to resist being pried apart in the half of this knot that is choking the loaded ends; but that extra turning is superfluous in the other half, which is loaded qua stopper knot, and so might as well be just an overhand. An Australian fellow has favored tying the strangle knot (the "half of a dbl. fish.") with the two ropes and using that as the ARJ knot; I think he'd tested it.
But back to the top : there IS a simple solution, with a simple assurance via a tie-off.
QED?!
*kN*