I may have discovered (invented?) a new way to tie the alpine butterfly knot. Please check out this youtube link:
The closest I have been able to find to “my” version of the ABK is in Knotting Matters 8: a version of tying the ABK by Rob Chisnall, which is similar, but slightly more complicated.
I’d love your collective feedback
cheers
andy
PS. At the suggestion of Aplineer, I’ve re-worked the video slightly, to clarify the method of tying:
I’ll just say that I’m fairly proficient with knots and I managed to mess up the method for nearly the first dozen attempts. It’s easy to miss the little twist of the left hand as the right hand is moving around. Then again, maybe it’s just too late in the evening. 
I notice that if you accidentally reverse the loop twist of the very first step, you end up with an evil imposter knot for the butterfly loop, namely the half-hitch loop. In light of this, I would recommend more conventional methods.
There’s a more ergonomic and visually simpler way (actually two ways) to tie the ABK on a hanging line and also ensure that one cannot tie roo’s favourite knot, his so-called evil imposter.
Thanks for all the feedback.
One question for the group, is this indeed a new method of tying the ABK? Has anyone seen it before?
As far as merit goes, let me reply
Roo:
Perhaps in the light of morning you may find this method easier? I don’t think a video is the easiest way to learn a new knot, but it’s certainly the easiest and quickest way to get the idea out to this board. I’ll work up a traditional storyboard instructions for tying, and post it as well. As far as tying it wrong goes… well, you can tye most knots wrong and get disastrous results, tying it correctly is the trick. Of course if a knot very easy to tie wrong, and difficult to tell if you have, then there is some risk, and the knot in question may not be very useful for that reason. Is that your assessment of this method of the ABK Roo? You did identify the “trick” with this method: beginning with the left hand pointed thumb down is the key.
on a lighter note Roo: I’m perversely pleased that this version gave a little challenge to such a knotting expert
Xarax:
Perhaps you could clarify what you mean by the knot being oriented the “wrong” way. A primary benefit of the ABK as a mid line loop is that it may safely be loaded in either direction. Do you mean that this method doesn’t accurately form the ABK? If so, I disagree. I’ve compared this method with traditional methods, and haven’t been able to find any structural difference in the finished loop knot. Maybe you’re seeing something I am not, in which case I would appreciate your clarification.
Alpineer:
The two methods you mention of tying the ABK: would those be the hand wrapping, and twisting methods? Or are there two other better ways that I don’t know of? I find that which method one ties the ABK is usually a matter of which way one was first taught. There are drawbacks to each of the two “traditional” methods of tying the ABK. Two criticisms of the hand wrap method are: 1. it is difficult to tie a large loop with this method (a large loop being a useful technique for certain rope access techniques) 2. if the rope should come under unexpected load while tying the hand wrap method, you’re hand will likely be injured, which is why most search and rescue personnel avoid the hand wrap technique. For the twisting method: This method requires multiple grasp and release moments to form the twist, and to pass is through the doubled loops… which I find awkward (at best) and often lead to the attempted tying falling apart, or too errors in the tying. Also, if tying this method of the ABK at height, with a lot of rope weight below you, the challenges of the twisting method are magnified.
It’s likely that “my” method has drawbacks as well, some of which you three have already picked apart to some degree. I do think it’s benefits address the drawbacks of the two traditional methods of tying. The more I use is, especially while working at height, the more useful I have found it to be.
Thanks again for all the input, please keep it coming. And I would be especially thankful if anyone can tell me if they have encountered this method previously.
thanks
andy
Xarax : are you asserting that because of the asymmetry of the ABK that loading the loop from one direction is inherently weaker than from the other? I have never seen any data to suggest that, and if you have, please share with me where I may find it.
It seems to me this criticism of the ABK, based on asymmetry, would apply to any butterfly knot, regardless of what method is used to tie it. Would you agree with that? Or, do you have some special criticism for this(possibly) new method of tying the ABK.
Thanks
I think other methods do a better job at either forcing the tyer to make the correct move or if incorrect moves are made, they result in obvious failure or even accidental success.
There are drawbacks to each of the two "traditional" methods of tying the ABK. Two criticisms of the hand wrap method are: 1. it is difficult to tie a large loop with this method (a large loop being a useful technique for certain rope access techniques) 2. if the rope should come under unexpected load while tying the hand wrap method, you're hand will likely be injured, which is why most search and rescue personnel avoid the hand wrap technique. For the twisting method: This method requires multiple grasp and release moments to form the twist, and to pass is through the doubled loops... which I find awkward (at best) and often lead to the attempted tying falling apart, or too errors in the tying. Also, if tying this method of the ABK at height, with a lot of rope weight below you, the challenges of the twisting method are magnified.
It’s possible you may be thinking of another hand wrap method, but I’ve never had any problem with controlling loop length with the two standard methods. It’s just a matter of leaving the structure open at the right time as you pull material.
The right time for the coil method is at the initial pull to the right. The right time for the twist-fold method is represented by the down-pointing gray arrows in the second step.
I do wonder about people who are worried about the line suddenly seeing severe load while they’re tying. Trying to reduce hand exposure to such a scenario via the method of tying seems to me like only looking down the barrel of a loaded gun for a second rather than five seconds. I’d much prefer to eliminate that risk. I do notice that in the method you present, your left hand is in the rope for a long time, relatively speaking.
Hi All,
Thanks for sharing your video Andy, Sometimes it is the ideas of another that can instigate sparks of creativity
and I dare say, even private appreciation.
Here I like to share my method to tie this beautiful butterfly loop with you. https://www.youtube.com/watch?v=MRnoevwqYto.
謝謝 alan lee.
Hello all,
I am still curious if any of you have seen or read this method of tying the ABK before.
now, to reply to some specific points:
Yes, because, when loaded from the one end, the Standing Part s first curve is almost two times as wide as when loaded from the other end. At their X-shape crossing ( clearly visible when the knot is viewed from one of the two "flat" sides ), the one segment goes "over" the other, and that makes the curve which incorporates the "over" segment ( and encircles the "under" segment ) wider than the curve which incorporates the "under" segment. At this first curve the knot bears 100% of the load, and, presumably, a wide, gentle curve is always stronger/more evenly loaded than a narrow, sharp one, therefore the first curve is the weakest link of the knot s structure. ( However, the point of rupture is, most of the times, different from the point of maximum strain... )
Xarax, I follow you’re thesis that there is a Stronger and a Weaker loading side of any ABK. First, it looks to me that the stronger side, with the wider internal knot bend, is pointing up (to the standing end) in my video. Therefore, the stronger side is taking a load should you load the loop of the knot. Am I missing some detail that you are seeing?
Simply because there are no data, for any knot, anywhere ! :)
The CMC rescue guide states that the alpine butterfly has a efficiency of 75% when the loop is loaded, and 57% when pulled end to end (as a butterfly bend would, or the ABK used to isolate a damaged section of rope). I’ve seen similar efficiencies posted in other resources as well. In reading dozens of books, attending multiple professional trainings, and working with many experienced riggers over the years, I’ve never encountered the idea that there is a right and wrong way to load a ABK. I would say that conventional wisdom is that they are both equal, or that the difference is so negligible as to be disregarded.
Of course, conventional wisdom could be wrong. In the multiple sources I can find stating the efficiency of the ABK, none of them specifically mention testing the left hand and right hand loading of the loop. I had considered this due to the fact that there is no meaningful difference, rather than inattention on the part of the testers. You, yourself, mention the important point that a rope with almost always break just outside the body of the knot (a fact widely and generally noted). Given that fact, it is mysterious how the internal structure of the knot affects the strength.
I’ve read multiple theories of what makes knots weak, or strong. And while there does seem to be some agreement, there are also contradictions (as I’m sure you’re aware) and I believe that the only reliable way of measuring knot efficiency is through break testing. I have access to a break testing rig, and when I have time, I intend to test your thesis that the left and right hand loading of the ABK, have a significant difference, I’ll let you know.
In the mean time, I’ll stick with conventional wisdom.
on to Roo:
I think other methods do a better job at either forcing the tyer to make the correct move
I agree with you about this, although I do think the point is worth discussion. I think that this “new” method of tying the ABK would be useful to “expert” users who tye this knot regularly, know it well, can reliably inspect a tied version to insure it’s tied correctly. It has advantages over the two traditional methods, in my opinion. At the very least, it’s useful for impressing your peers over beers after the gig 
I don’t think I would teach this method to someone who didn’t already know the ABK. The “coil” method is probably best for that.
I do wonder about people who are worried about the line suddenly seeing severe load while they're tying
While working at height, any reduction in risk, however slight, has merit. It is a subtle (but worthy) distinction between having three wraps of rope around your palm for 10 seconds vs. having a loosely formed loop around your fingers for two seconds. As far as sizing of the loop: it is possible to form a 5 foot long loop in the ABK for a Y hang anchor, or rope re-direct, it’s just very awkward. The twisting method is superior in this application, but with that much rope to pass, it can be clumsy and awkward to complete the “fold” of the knot. I believe “my method” can be better in this (admittedly very specific, and somewhat limited) application.
finally to Alan:
Sometimes it is the ideas of another that can instigate sparks of creativity and I dare say, even private appreciation.
It brought me great joy to watch your video of your method of tying the ABK. The loop appears almost as if by magic, and has a wonderful element of theatricality to it. I’m greatly pleased to have been able to connect with a fellow knot aficionado in China!
Thanks to all of you for taking the time to watch the video I posted, and to share your thoughts with me.
andy
Simply because there are no data, for any knot, anywhere ! :)
Xarax,
I thought this link might interest you
http://l-36.com/rope_articles.php
page 21 of the first link in the list has some interesting observations regarding the tightness of a bend internal to a knot, and how it affects knot strength.
cheers
andy
Hi knot rigger,
A while back I studied the problem of tying the ABK on vertical lines under tension. However, I’m rethinking my initial post. Your method may be as/more effective than my methods at managing the weight of the rope below the knot to-be-tied. I’ll have more to say on this, but I must go out in the sun. Later.
alpineer
hello again
Alineer:
A while back I studied the problem of tying the ABK on vertical lines under tension
I would be greatly interested in seeing, or hearing from you about, your studies of this issue, and what conclusions you once drew, as well as any new thoughts you have about the method I started this tread with. Thanks
Xarax:
1. It "states" :), but it is not referring to a single ( = even ONE ) experiment which corroborates that conjecture. Such "statements" should not be considered more seriously than any conspiracy theory - and we all know very well that the internet has multiplied the production of conspiracy theories on everything, it had not eliminated them !
I trust the guys at CMC, and in my opinion, you would do well to trust them as well. Here is a link to a link to their testing:
http://caves.org/section/vertical/nh/59/Rescue%20Knot%20Efficiency%20Revisited.pdf
When I have a little more time, I’ll be happy to look up some other testing data for you, if you like.
3. ...dozens...many...multiple... never encountered... but you have just said it : "there is a Stronger and a Weaker side in the loading of any ABK ".
you actually misquote my earlier post there: here it is unredacted:
Xarax, I follow you're thesis that there is a Stronger and a Weaker loading side of any ABK.
you’re assertion that there is a weaker and stronger side to the ABK is a thesis which may or not be true. as we both agree, the only way to know for sure it to do some break testing
I do not expect that you would like to BET on this ( because I do... :) )
I would be happy to make a gentlemanly wager with you. Lets refer to you’re “stronger side” as the right handed side of the loop, and the other side as the left handed. I wager my copy of Jeff Jepson’s Knots at Work against whatever you think is fair, that the difference in efficiencies between the left and right handed sides of the ABK is less that 5%.
now I just need to find some line I don’t mind breaking
cheers
andy
Xarax
Are you backing away from our wager now?
I think, perhaps, we’ve just had a little miscommunication, so let me clarify:
My wager is that the two sides of the ABK are essentially the same strength, and your position (as I understand it) is that the two strengths are different.
If you win, you win my copy of Knots at Work, and if I win, I get something of equal knot nerdy value from you.
I’ll do the break testing, record the data, and provide you (and anyone else) the results. I know you only trust numbers, not people, but in this instance you’ll have to trust me to do the testing, and be honest about the results
So what’s the threshold between same strength and different strength? I propose that a 5% difference in breaking load between the sides, or less, should be judged the “same” strength, and anything above 5% different. So, if the lefthand side breaks at 66% efficiency (33% strength reduction) and the righthand side at 72%, then you win! We agree that 6% is a meaningful difference (for the purpose of the wager).
I have long length of a lightly used 3mm sampson double braid (i’ll look up the exact specs) which I’ll do the testing with. It is used, so it won’t be a perfect measure of breaking efficiency, but it will suffice for a comparison purposes. All the knots will be tied, dressed and set by me, as identical as I can make them. I’ll tye an equal number of samples with a lefthanded loaded ABK, and with a righthanded, as many as I can get from the sample rope (which is something like 100’ long). The breaktesting rig consists of a chain puller connected to an analog dynamometer. I pull each sample until it breaks.
do we have a bet?
cheers
oh yeah, PS. yes the double fisherman bend is stronger than rest! Somewhat surprising. Three things that I find interesting in the CMC report
Firstly one has to be specific in what constitutes the
supposed impossible knot. For I can refute the assertion
if having_one_eye and being symmetric is the goal,
except insofar as you might point to parts of my
“solution” as being (degenerate) unused “eyes” !
–of which there’d be two.
Indeed! Sometimes --rather often, I find-- when one attempts
to illustrate (by pen & paper) a knot, the arranging of its parts,
the fiddling of this going here and that there, … instigates the
“What if …?!” thinking that arrests the recording of the extant
knot and sends one into explorations of what else, of what if … !
(And one might end up with TWO (or more) unrecorded knots
to do, instead of one done.) 
Here I like to share my method to tie this beautiful butterfly loop with you. https://www.youtube.com/watch?v=MRnoevwqYto.謝謝 alan lee.</blockquote>PLEASE NOTE that Alan’s knot has as different geometry
than the OP’s, in which the legs of the eye cross within
the knot body rather than abutting each other as in so
many of the available illustrations and in the OP’s. In part,
this arises from torsion imparted during certain of the tying
methods --i.p., of the “twirly flop” old method. Wright &
Magowan, who introduced this knot to mountaineers,
specifically called for this crossing.As for the question of strength & orientation,
my surmise is that the crossed-legs version is stronger
than the other, and that between which end, it might
not matter much, in both cases. (I disagree with X’s
assertion of curvature in the OP’s case --it seems a
1dia turn either way.)–dl*
Yes, I know… You would had been more successful as a lawyer than as a knot tyer.
Of course, if I had been so naively cunning to dare to present any single eye loop with a second unused, redundant pseudo-eye gawking around, I would had been “keystroking”, and “filling lines” !
We have discussed this issue some time ago - in case you have forgotten it, read :
http://igkt.net/sm/index.php?topic=4425.0
Noope - In the “common” dressing, without crossed-legs and with the two Standing Parts ( as seen from the one of the two “flat” sides ) parallel to each other, we have an X-shaped crossing : at this area, the one segment of the Standing Part engulfs the other. The diameter of each first curve is determined by the paths of both legs of the U-shaped turn, the leg before and the leg after the tip of the “U” : so, it is determined by the direct continuation of the Standing End, before the tip of the first curve - where, indeed, there is no difference between the two curves -, AND by the continuation of the Standing End after the tip of the U-turn, where we see the difference I am talking about. In order to be able to go “around” the segment of the line which passes “under” at the X-crossing, the segment of the line which passes “over” has to follow a wider path, that is, to trace a more gentle curve. The segment of the Standing Part which passes “under” makes an 1 rope diameter turn, while the segment of the Standing Part which passes “over” makes an 1 + 1 = 2 rope diameters turn - because the later has also to encircle the former.
Your quip merits at least two of these xarax.
Well xarax, you’ve naively misunderstood my comment completely! “These” refer to “:)” It was just my way of applauding your sense of humour (or was it pure sarcasm?) in this particular case.
You have used a naive syntax ! A pronoun of a smiley !
Hi knot rigger,
I think your method is well thought out. After a bit of playing around it became apparent that you have addressed what is perhaps the most important consideration regarding a vertically oriented line - that it can be under considerable tension - by virtue of it’s own weight - where the knot is to be located. The method is only fully appreciated when much of the rope’s weight hangs below, where the common “Handwrap” method is a fail. Your method has good flow and feels good in the hand - i.e. ergonomics - once one is familiar with it, which doesn’t take long.
Your method provides an effective means for isolating the weight of the rope without having to relieve the tension by means external to the tying process.
N.B. Any risk(examples?) of body parts getting caught or injured requires management prior to tying.
Your method elegantly avoids unwanted friction during the tying process. Other methods require mitigation.
P.S. It would have been better had the tying process been shown even more slowly in the middle portion of your video, allowing the viewer more time to see the process clearly and also make it easier to stop and study.
P.P.S. Got a name for your method?