Cross-loading — this word may give a chill in the spine of climbers. It is a real terror, be it with a karabiner or knot. In the industry they adopt the simple and very straightforward solution for this. Just use super-strong metal links, a.k.a., absolute bombproof steel maillon rapides, wherever in suspect.
In climbing, whereas the same approach is indeed recommended in some cases like a group activity of abseiling or bottom-roping, the weight, bulkiness and awkwardness in use of steel maillons put off most, understandably and justifiably in many cases. Instead, climbers usually use either a climbing-rated (aluminium) karabiner or knot to bear with cross-loading. We climbers know it is not ideal, but we are also somehow confident they are strong enough for our purposes.
But how confident are you? Recently, a video of an experiment of cross-loading to a few different popular knots used in climbing has been circulated across SNSs, as I have noticed. The point raised in the video is good, as climbers should be aware of the strength and risk of each knot in that respect.
Unfortunately, the experiment presented in the video is a rubbish (and that is why I am not giving a link here; it is a waste of time to watch). To make an experiment even remotely credible, the following is essential:
- Each trial must be repeated multiple times, preferably hundreds of times.
- Any figure must be presented with the associated error, preferably both systematic and statistical errors with the confidence level quoted. Admittedly the former is hard to pin down by its own nature; however an experiment without the latter practically says nothing.
- All the knots must be in the same conditions. In particular, each knot must be well set, unless deliberately experimenting the weakness of unset knots.
- The conditions in each trial are preferably identical. Understandably, it is not easy and actually costs a lot to achieve. Nevertheless, to abandon the effort from the beginning is unacceptable.
Related to the point 2, if numerical figures are presented with an unrealistic precision, such as 1.23456789, as opposed to 1.2 (or something, depending on the accuracy the experiment has achieved), it is usually a good indication that the author knows little or nothing about the scientific experiment, principle in engineering, and statistical handling of the data, and hence you can dismiss the result fairly safely with the fact alone. It is an easy and fairly reliable litmus test.
The video has none of those points listed and also failed in the litmus test. Consequently I have no hesitation to dismiss the video as any meaningful reference.
Instead, here is my summary of the issue to my best knowledge.
The best knot against cross-loading is alpine butterfly (also known as lineman's loop). That is a part of the reasons alpine butterfly is the best knot by a good margin to tie a climber(s) in the middle of a rope, when climbing as a group of three or more.
Unfortunately it is not suitable for a tie-in point to a harness, simply because to tie alpine-butterfly in any way but on the bight is very awkward. Indeed, alpine butterfly is probably the most notorious knot among the knots climbers use in terms of remembering how to tie even on the bight. I guess very few can tie it at the end of a rope in a rethreading way.
On the other hand, one of the worst knots I know against cross-loading is bowline. You can experiment it by just pulling the knot on the loop by your hands for different directions. The video below “Bowline knot and cross-loading” by Masa Sakano is a simple demonstration.
In addition to the vulnerability to cross-loading, it is a well-known fact the strength of Bowline is heavily marginalised without a stopper knot, whereas it makes no to little difference in figure-of-eight.
The bog-standard way of tying a stopper knot in Bowline is inside the loop (I mean, in the loop side, be it right-handed or left-handed, the latter of which is also known as cowboy bowline, Dutch marine bowline and winter bowline). Unfortunately, any stopper knot can be undone in a course of a day or during a sustained use or even in a single use if it is not set well. Especially, if the stopper knot is rubbed by something, like repeated cross-loading, it can happen very easily. Look at the demonstration video above, and although no stopper knot is employed in the knot used in the video, you can guess what would happen to a stopper knot in such a situation.
Yosemite Bowline or Edwards Bowline can improve the vulnerability. The difference must be significant in principle; however I don't have a quantitative answer of how much. Note that they are not foolproof, either, if better than the standard Bowline (A video "Yosemite Bowline not safe for climbing" by "Yosemite Bowline" — seemingly the one-time account — raises one of the points). Also, I do not think Double Bowline is any safer than the standard Bowline in terms of this vulnerability, judging from the mechanics.
[2016-04-28, added] Note: In fact, the difference between the standard right-hand and left-hand bowlines is significant in the vulnerability to cross-loading (or sometimes referred to as ring-loading, namely,
pulling the eye apart and essentially treating the knot qua end-2-end jointi).
The left-hand Bowline is better because the configuration under the ring-load is the correct Lapp Knot/Bend (or Lap Knot),
whereas that with the right-hand one is anti-Lapp Knot, which is weak, as explained in "The Lapp Knot and its Evil Impostor" by David Delaney (see also the comment by Dan_Lehman for the application of Lapp Bend to Bowline). I understand it is the same reason why Left-Hand Sheet Bend (Ashley #67) is inferior to Righ-Hand Sheet Bend (Ashley #66, #1 etc).
The tie-in point to a harness (i.e., rope loop) is often used in belaying in climbing. Some go so far to claim that is the only acceptable way to belay (like Hill and Johnson 2000), while another (Andy Kirkpatrick; "Rope Loop Belaying?") says he only uses the harness belay loop in practice. Both the methods have pros and cons; and therefore each different situation determines which is better and/or safer. In other words, there is no right or wrong for a general use, and use of the tie-in point is certainly an accepted and maybe the most suitable way in many climbing situations.
However, you should be aware that it is always the cross-loading to the tied loop of the knot, when it is weighted. And the repeated cross-loading always weakens the knot.
When climbers tie a rope to a harness, many of them do and should check it to perfection — partner-check is always highly recommended. However, once they have tied in, climbers tend not to check the knot again during a series of continuous climbing like multi-pitch climbing or alpine climbing, just assuming the knot is fine and safe.
For that reason, bowline knot as a tie-in point has a serious risk even with a stopper knot, or even for Yosemite bowline, during a course of a day.
The best use of the bowline knot is to tie an end of the rope directly to an anchor, such as, a boulder and tree trunk. It is far easier and quicker to tie, and is more easily adjustable, than figure of eight for that purpose. There is no cross-loading in those cases. As long as it is not cross-loaded and the pressure is on in the right direction, bowline is strong enough and a good knot.
Bowline has been a favourite knot by sailors, long before
a karabiner was invented, as referred to as
a real sailor's knot by Ashley (#2551, pp.410),
and with a good reason. I presume sailors use bowline
in that way in most of the times.
In conclusion, while I do not disregard bowline entirely from climbing — far from it, as I do use it often — climbers should be aware of its risk and limitation and serious about them.