In reply to jwa:
> (In reply to ian Ll-J)
> [...]
>
> I think everyone needs to go back and read what John Long says he did and then copy what he did at home at ground level. He basically stuck the rope through his harness and then climbed.
What he said he did doesn't really make sense : i.e., forming the double-turn of the main line AND THEN reeving the tail through the harness (which would have to be done with one hand, as those double turns need to be held to BE ... !) ?? So, I think that, yes, he essentially did not tie <anything>, but he's got an odd way of expressing that.
> Same as what happened to Lynn Hill except she'd tied a figure of eight then poked the loose end through her harness.
No, according to Lynn's book, she simply reeved the line into her harness and then was distracted from finishing her tie-in (aka starting & finishing her knot), and she intended to tie a bowline. [_Climbing Free_, p.4]
> Doesn't matter what type of knot you use, if you don't actually tie it to your harness in any way you're going to be in trouble.
Which points to the solution to these "failed bowline" problems being checking that a knot is tied, less than checking WHICH knot is tied. Though it must be emphasized --i.p., to these old "sailor" types who never backed up the bowline and think that that's fine (ditto for the caver)-- that IN SOME ROPES the bowline (without further securing, by extension or additional knot) WILL LOOSEN and can come untied. Your half-century of tied bowlines in other cordage in other circumstances does not guarantee future performance in different conditions, especially in different cordage. (The knot --even the "double" bowline-- will not even hold, in (bare, 12-strand) HMPE line : the rope just feeds out of the knot, collapsing the eye (on slow-pull loading)!!)
The "water bowline" --i.e., a bowlinesque tucking of the tail through a clove hitch base-- has merit in that the base structure, the clove hitch, resists loosening. Taking the tail through the harness a 2nd pass and reeving it up through this knot (in any of a variety of ways) will give redundancy, and the strangle knot (aka "half a dbl.fish.") finish can be put on right outside of the bowline's collar, out of the way, more easily tied & inspected as noted above.
There is a similar tact that can be taken with a fig.8 base knot, reeving the tail in-&-out of that in a simple "U" geometry similar to the "rabbit"'s path in the bowline, and then a 2nd pass made through the harness leading again to a simple finish; and this leaves the fig.8 much easier to untie, I think.
The end-bound dbl. bowline was designed to provide a bowline knot that could be made tight when unloaded (the problem being the loosening of the knot when slack), but in such a way that it didn't become overly tight on loading. The "end-binding" of wrapping the tail once around the coils of the dbl. knot achieves this, pretty well --one does need to set it pretty tightly, though! (Watch the knot when it's loaded and you'll see the loaded parts shrink in diameter and leave some gap in this end's binding, which should be filled up on relaxation.) It makes the knot a little harder to untie, but it's still pretty easy. (First, pull some main line through the collar; then use the tail & eye legs to pry open the coils.)
*kN*