One of These Knots Can Kill You

It seems every year we lose climbers to a simple user error that can occur when joining two ropes together for a rappel. Compounding the issue is some media outlets refer to the knots (or bends) in question with various misnomers that create further confusion within the climbing community. There needs to be more widespread standardization of the options available for joining two ropes together and it starts with referring to them with the correct nomenclature.

The first thing to understand is any knot used to join two ropes together is technically referred to as a “bend knot” or simply “bend”. While there are quite a few bends that are appropriate for joining two ropes together it is the Offset Overhand Bend that has largely gained popularity for multiple reasons.

Offset Overhand Bend (OOB)

Flat Overhand Bend
Flat Overhand Bend

Like any option this one comes with advantages and considerations.


  1. Fast and easy to tie, especially with thick winter gloves on
  2. Low profile so less likely to get caught in cracks and on features while pulling the ropes
  3. Easy to untie even after a dozen high angle rappels
  4. Pull tests indicate a properly tied Offset Overhand Bend will not capsize unless loads exceed at least 1400 pounds, far more than any climber can generate on a rappel.


  1. Like any knot this one needs to be “dressed and stressed” to be safe. After forming the knot tighten all 4 strands separately.
  2. Leave 12 inches (30cms of tail). This is more than sufficient in the unlikely event of the knot capsizing. There has been at least one fatality when a cautious climber left 3+ foot tails and then threaded a tail through their belay device ending in catastrophic failure of the system. Twelve inches is sufficient.
  3. Use ropes of similar diameter. UIAA recommends within 3mm of each other, which with today’s modern (often skinny) ropes is usually easy to stay within. If using a thin tag line consider adding either an offset overhand tied with the thinner rope over the thicker rope and cinched tight to the offset overhand bend. The idea here is it will help prevent the knot from capsizing but in reality should not be needed. The practice of tying a 2nd flat overhand bend a few inches down from the first seems to negate most of the advantages (fast to tie and less likely to get stuck while pulling) so this author feels that practice is not needed.

I’ve made it this far without calling this bend knot by its more common name. I’ve decided to leave the common name out. It serves no educational purpose and its use should be considered archaic in nature.

Now we get into the two options that really exacerbate this issue. One of the two knots below can kill you.

Reverse Traced Figure of Eight Bend (Flemish Bend)

Reverse Traced Figure of Eight Bend
Reverse Traced Figure of Eight Bend

Like any option this one comes with advantages and considerations.


  1. Low profile so less likely to get caught in cracks and on features while pulling the ropes
  2. Super strong. If you look closely you realize this is the same option we use to tie into our harnesses. We can not generate enough force to get this knot to fail.


  1. Like any knot this one needs to be “dressed and stressed” to be safe. After forming the knot tighten all 4 strands separately.
  2. Leave 6 inches (15cms of tail). Since this knot can not capsize by design it is logical to follow the same guidelines as using the knot to tie into a harness. Six inches of tail on a dressed and stressed knot is sufficient.
  3. Adding “back-up” knots to both tails greatly increases the likely hood of a stuck rope and is completely unnecessary given the strength of the main knot.
  4. Can be very difficult to untie after heavy load, especially with gloves on and slick new ropes.
  5. While tecnhically called the “Flemish Bend” adding the METHOD one uses to create it (reverse traced) to the common name will help differientate between the two.

Flat Figure of Eight Bend

Figure of Eight Bend aka FATAL MISTAKE!

The above knot has a proven track record of killing climbers. It routinely capsizes and fails at loads easily generated in rappelling. So why is it still being used after years of accidents showing it’s not sufficient?

New Climber Perception

I’ve had hundreds of new clients look at the OOB (first knot pictured) and say “That’s it?” Having confidence that such a simple and quick to tie bend could be sufficient for joining two ropes together and committing our full body weight out over the abyss is not so easily won. Let’s be honest, its simplicity and small volume make people nervous regardless of its more than adequate strength.

So what does the new climber do? Well if one twist around the ropes forms a Flat Overhand Bend then two twists around the ropes must be safer right? That mindset creates the deadly Figure of Eight Bend pictured above… and kills people.


The common name I’ve heard used to describe both the Reverse Traced Figure of Eight (Flemish Bend) and the Figure of Eight Bend is the “Flat Eight”.

This needs to stop. One is a strong and suitable bend for rappelling. The other keeps killing people. We should not use the term “Flat Eight” as most climbers, especially new climbers, do not easily see the difference between the two. They tie a Figure of Eight Bend, thinking it is a “Flat Eight”.


In order to reduce or eliminate the amount of fatalities the confusion these options create, we, the climbing community at large, especially the widely read climbing magazines, need to step up and standardize our definitions of these knots and not use mis-leading negative sounding catchy names to describe safe practices. Using misnomers can encourage a new climber to switch to a different, and potentially much more dangerous option for joining two ropes together.

If every guide, mentor, instructor, and tenured climber starts referring to these knots with the correct terminology we will see a reduction in unnecessary loss of life.

UPDATE 11/26/17: An member linked this research paper on this topic recently. It is the best piece of research I’ve seen on this topic yet and am sharing it here for anyone who wants to do some more reading:

Password “thankyou”

Please share if the above content was useful!

125 thoughts on “One of These Knots Can Kill You

  1. Well stated, Dave. Since 1982, when I was first introduced to the use of the flat overhand for tying rappel ropes together by French guide instructors in Chamonix, at a meeting of the UIAA Safety Commission, supported by the results of pull tests done in their test laboratory, I have used nothing else for the purpose.

  2. Generally much appreciated this article, but two comments: first, I think if you’re going to advocate properly dressed knots (which, you should), then your pictures should be of properly dressed knots. Second, it may be worthwhile to point out that the safe figure-8 bend is called a Flemish Bend (ABOK #1411).

    • Agreed I could have dressed the 2nd two knots better. When I have time I’ll try to replace the photos. I added the info RE: Flemish Bend… I think using the method of how it is tied in common communication would help differentiate the Flemish Bend from the more dangerous looking twin.

      • There is a problem with your article. We did some test (in labs) and in both case, the knots are capsizing. At a little bit lower forces with the figure height but after the first capsizing all the other flips are higher than the overhand knot. In both cases, if you have min. 30cm and your knot is tight, the rope will rupture before the knot will open. In both cases the knots can capsize 3 to 4 times but the rope will rupture at the end. The main problems, is the knot not tight enough and to lose to the end of the rope. 30cm is the minimum. I can send you the graphics if you send me an address.

  3. hi. thanks for the article. can you add a big red x in the image of the figure of eight. i read the whole article and appreciate your words and for a non-English audience this might help. cheers.

    • Thanks Laura! That is an a great idea. My intention was to get a skull & crossbones on that image but I was anxious to publish as soon as I had the text finished. I will probably upgrade all three images in the near future and will add something visual to the image of the dangerous figure of eight bend!

      • Related: the bad figure eight is the one that shows up in the preview on Facebook. Another reason to add the red x.

        Thanks for publishing this – it’s very clear and helpful!

  4. I think a video of the figure eight bend in failure mode would be very enlightening and complimentary to your write up.

    • Thanks Hawaiigirllovey! I have looked those results over in depth and while the author is a bit anti-flat overhand the numbers help prove the point that a properly dressed flat overhand is more than sufficient.

  5. Obviously the EDK and flemish bends are safe and the flat figure eight is unsafe. Nobody is disputing that. I’d like to talk about the terminology.

    So a typology of relevant knots.

    We have two flat knots:
    flat overhand (aka EDK)
    flat figure eight (you call it the figure of eight bend)

    and two retraced knots:
    water knot (could be called a retraced overhand)
    retraced figure of eight (flemish bend)

    “flat” indicates the tails are next to each other whereas retraced indicates they’re going opposite directions.

    Thinking of it this way (distinguishing flat knots from retraced knots) makes this all pretty clear. You say that we should stop using the term “flat figure eight” because that term has been used to describe the flemish bend. I’ve never seen anyone refer to the flemish bend as a flat figure eight. If someone used the term “flat figure eight” to refer to the flemish bend, they are simply wrong and they need to be corrected.

    But I would encourage people to use the terms flat overhand (to refer to the safe know) and flat figure of eight (to refer to the unsafe knot) and understand what “flat” means to emphasize the similarities between these knots.

    • Thanks for the comments Michael! I believe some people think “flat” means the knot lays flat on the rock this is less likely to get stuck. The retraced 8 looks very “flat” when tied, which might be why some confuse the two. I’m obviously making assumptions about why I think we keep seeing these accidents.

      • > “But I would encourage people to use the terms *flat* …”
        NOOOOO, please, use the appropriate term “OFFSET” –the knots are offset from the axis of tension. (This apt name comes from Clyde Soles’s book _Outdoor Knots_.)
        And “offset water knot” is a further improvement, as “overhand” knots cover too many, while “water knot” is pretty narrow in denotation.

      • “flat” is apt for the common ILLUSTRATION of the fig.8/flemish bend (and eye knot) : i.e., the (lazy) artist makes a pure, 2-D tracing of the initially drawn (single strand) fig.8. THIS is an orientation that will not survive loading –the outer strand will want to move one way or another. (And some illustrate this and then advise to properly dress the knot but w/o a hint of what that dressing might be!) –such as _On Rope_.
        Look at this page’s images : the fig.8 is Asymmetric (mis-dressed, IMO) in both of its versions. The offset fig.8 (aka “EDK-8”) has the orange-ish tail swing out of orientation for the final tuck –it should finish up on the LEFT side of the green. The flemish bend is farther off, having the green rope run father on both ends,
        (Now, when folks test such knots, do you think that they tie them the same, or with some other orientation? –it’s usually not indicated, not appreciated.)

    • > ‘ “flat” indicates the tails are next to each other whereas retraced indicates they’re going opposite directions.’

      No, “flat” was to suggest that the knot’s loaded ends will press flat against a surface when hauling the line down; but, again, for this the term “offset” is much better.
      Having tails adjacent is no assurance of the desired quality (the butterfly lacks it, e.g.), and lacking that is no assurance of NOT having it (doing the offset tying on an EDK, just take the choking strand a half-turn farther –yes, to form a fig.8– and you will need to tuck it out the same loop but in opposite direction to that of the overhand’d end. (This knot seems maybe a bit better, maybe not; but go a full turn –forming a fig.9– and then tuck out in same direction, you get a good, sure *choke* to prevent capsizing!)

      • WRT the book “On Rope” (OR), the second page in Chapter 3 clearly explains what is meant by “dressing” and “setting” knots. Granted, a better index or a glossary would be appreciated, but for a book that was essentially put together by a pair of volunteers and only sparsely updated since it is a pretty decent reference. (Oh, and yes the Rock Climbing related material needs work…which is probably expected of a book written and maintained by cavers. 😉 As a caver I feel perfectly safe saying so.)

  6. What about the double fisherman ?
    I usually use it because it is compact, not too hard to undo, and seem structurally sound although I don’t have data to backup this statement.

    • Its a good knot but it jams in cracks, plants, etc when pulling it down. Hence the need for less strong (strong enough) but “flatter” knots.

    • Was taught to do double fisherman also. Super compact and easy to untie, and never has any issues with slippage, stuckage or using them in icy conditions.

    • Hi Will. The double fishermans was probably the most popular option before the OHB gained popularity. It is unquestionably strong when properly tied, but is quite bulkier than the OHB and definitely more time consuming to tie and untie.

      • It can be an absolute bear to untie if it has been heavily weighted for a long period of time or multiple times, especially in the winter. I have sometimes been tempted to bring the whole frozen mess home and untying it after it thawed out.

      • Damned “time- (and brain?) consuming” knots!

        Better to just use the time tested “slip-n-die knot” (TM)

        Unties much easier, especially on rappel. 🙂

    • Thanks for the link, Steve

      While the article implies that the failure was due to the use of the flat figure 8, it’s not clear how that conclusion was reached, since no one actually witnessed the knot being tied.

      Also, I found the statement at the end of the article particularly interesting:

      “both men who recovered Kautz’s body commented on the slickness of the ropes. Buzzard wrote: “The dry coating on the ropes seemed to be extremely slippery. I tied them together with an EDK and pulled on it with my hands only and could make it walk and slip far more than I have seen any rope slip before.” Perhaps silicone-treated ropes, such as these, are easier to capsize than less slick cords? More testing needs to be done. With any knot used to tie ropes together, leave at least two feet of tail and always tie a backup.”

      //end of quotes

      Personally, I have always used the triple fisherman and don’t intend to change that.

      Maybe I’m just old fashioned, but I really can’t understand the argument being made by some of the comments here about the ease/difficulty of untying . It should be obvious, but the very fact that some knots (eg, the double or better yet — triple — fisherman’s knot) are more difficult to untie after they have been weighted indicates something about their likelihood of coming apart during a rappel. I will leave it as an exercise for the reader to figure out just what that might be. 🙂

      Safety should be the primary concern, trumping the time to untie the knot or whether it can be tied wearing gloves or whether it might get “caught up” a little more readily or cause a little more rope drag during retrieval.

      For the vast majority of climbing situations, these considerations are little more than inconveniences that can certainly be “lived with”. The same can not be said for a knot failure during rappel.

      • Thanks for the comments Lars! The fact is the research proves a FOH tied correctly is 100% safe to use for rappelling, and while the AMGA doesn’t exactly “endorse” a particular knot I don’t know a single tenured guide who isn’t 100% comfortable with its use.

        The triple fishermans is probably 300% safe to use (not counting potential for stuck ropes). It is unquestionably strong! Stuck ropes, by the way, I have noticed tend to create rescues but not fatalities. But they happen.

    • There’s a LOT wrong with this video’s cursory presentation of knots. Mainly, though, to this page’s topic, the HMPE (Dyneema) cordage is WAY too difficult to tie. Note that THIS testing finds the very knot being warned against here to be much stronger –it holds to rupture– than the EDK (which capsizes into nothingness) ! Sadly, they don’t test an EDK-backed EDK (which would be nice to see done, even in HMPE rope).

  7. Great article. Are their other knots being used for this purpose? I know of one but I may not have the name right. I think it’s a half hitch or a fishermans knot. It’s often used on prussic cord to connect the 2 ends to make a loop or circle. May be good to add it and let people know if they should or should not use it.

    • Hi David, thanks for the comment! There are lots of knots that can be used for this purpose. My focus of the article was to draw attention to the confusion within this family of knots and not list every option out there. The Double Fisherman’s that you mention is a very popular suitable alternative, with a few downsides. 1) Takes longer to tie, especially with thick winter gloves on 2) Harder to untie after significant load 3) Bulky and more prone to getting caught when pulling ropes. It is super strong though! While I can’t speak for every AMGA guide out there I can say with confidence that the vast majority of professional climbing guides use the Flat Overhand Bend almost exclusively due to the advantages I list in the article!

    • Probably sarcastic American climbers when it was introduced in the late 70’s early 80’s. Since it was adapted first in Europe and we thought it looked less than sufficient someone started coining the term. Jeff Lea might be able to add more details!

      • This explanation I think is on target. And then comes the irony of Americans going to what they presume must be better but which is (usually –but not in one of tests of that French video!) actually worse : the offset fig.8! (Via the rapid dispersal of misinformation, the Net has now muddied the waters.)

  8. Useful and accurate article. Apart from being safe, the flat overhand rolls over edges safely when retrieving rope.

  9. I agree that the quantity of knots that are taught and the nonsense that gets talked about them has gotten old over time. An artilcle like yours shows the advantage of a consensus : a core set of knots to teach, plus a few examples of what to avoid.

    But…. even though I am a fan of the flat overhand, I would still be sad to see it lose its nickname of “EDK”. No end of hilarity there! Ahhhh well…. I guess we can move on. 🙂

  10. Yeah, somebody should point out that the EDK is not less prone to getting stuck than other bends because it has a lower profile (it doesn’t), but because it rolls around edges that would catch a Flemish bend or a double-fisherman’s. This is also why the double-EDK is effective, safer (or at least, feels safer), and not prone to getting stuck.

    Any flat bend should have this property (single strand diamond knot, anyone?), with the possible exception of the butterfly (which, is also known as the strait bend, after the Juan de Fuca strait, but it not actually in ABOK [derived from the Lineman’s Loop, ABOK #1053).

    I would also be saddened to see the EDK name die out in lieu of the proper OHB. Climbers have always had a morbid sense of humor, and long may it be so.

    • The lineman’s loop structure used qua bend (which doesn’t need yet a 3rd name (but Brion Toss likes the humor in “straight bend”)) is not offset, though its tails do exit perpendicular to the tensioned line. (And why use this –except because “butterfly” is a known entity– rather than symmetric like knots, Ashley’s bend (#1452) & #1408 (also #1425) ?)
      “EDK” is here to stay; and “EDK-8” has been used for differentiation of the less stable knot.

    • #1 can dispense (even!) with “the same direction” –one further discrimination unnecessary to success (I think).
      I’ll echo what I wrote on your site:
      I’ll presume that “double overhand” is a borrowed name (it has already knot sense, for single strand) for an “EDK-backed EDK”. This knot was introduced some 10-15? years back, IIRC, by some on-line presence. While at the time I dismissed it as needlessly bulky, I realized that the tying guidelines for a more efficient knot included discrimination which if got wrong would … be worse. And so I humbly and more wisely believe now that the EDk-backed-EDK (an offset water knot repeated 🙂 makes great sense, and should work no matter (1) getting ropes ill-dressed, (2) not setting the two knots well, (3) getting the orientation of thick + thin ropes (if that’s your case) backwards, and (4) needing to tie under duress (storm threat, late-climbing fatigue, dim lit, high altitude) !! “idiot proof” to some degree, and using the same knotting that most should be well familiar with.

      .:. Do NOT “leave long tails” : rather, DO something WITH them (tie the 2nd knot) !

      • A FOB-backed-FOB is a great option. Most of your points are addressed by simply dressing and stressing a single FOB correctly, but it’s always wise to stack the odds in your favor. We could call it Stacked FOB’s? Sorry I’m done using that other acronym 😉 I’ll take my opinion one step further and maybe suggest stacked FOB’s would be a good thing to teach newer climbers for the inherent “idiot proofing” you mention, but experienced climbers and guides are probably best served with a single well tied FOB.

  11. Hmm. No discussion of the sheet bend, best for ropes of different diameter. And double/triple fishermans? And as far as getting ropes stuck, if youre careful, and use good technique, your rope shouldn’t get stuck. And then, sometimes, it will, regardless of the knot or how you pulled it. Had a 10.5 mm rope get caught in a 10mm pinch when the free end of the rope caught on the manufactureres taped end.

    • There are dozens of knots appropriate for joining to ropes together to rappel. I decided o focus on the most common one used by professional guides (FOB) and the two that are commonly mistaken for each other and lead to fatalities.

    • It’s a myth that the sheet bend is best for diff.-sized ropes. In fact, the EDK is arguably better here (than with like-sized ropes) in the relative sizes for climbers (6 + 9.5, e.g.), if one orients it so that the thinner line is what makes the *choke* at the entry of the knot, as it will be more impeded in rolling out around the thicker line. BUT, that is one more “discrimination” –something that can be got wrong–, as I note above/elsewhere here (versus the EDK-backed-EDK).

  12. You wrote about the The Figure of Eight being deadly and having a prove. Track of killing climbers.

    Can you explain why it fails?

    • Why? I don’t know the physics involved but there are multiple tests that show it “inverts” at loads less than body weight. Be sure you are thinking about the right “Figure of Eight”…. The Flemish Bend (pic #2 above) is safe. The last photo fails under low loads. Some tests have already been linked in other comments to support that.

    • There are a LOAD of differences in how such knots can be tied –and I’ve noted how the two 8 versions presented here are different and asymmetric– and the quick answer is that this knot tends to have a less tight *choke* at the entry point of the loaded ends. Beyond that, it has been noted that on flyping (capsizing) it will consume much more rope –and do so in a sort of shock vs. the more even “rolling” of the offset water knot– and so more quickly head towards its tails & being untied.
      (“Leave long tails” isn’t reassuring advice, IMO; as I note elswewere, DO something with that long material : knot it (again)! )

    • There are dozens of knots appropriate for joining to ropes together to rappel. I decided to focus on the most common one used by professional guides (FOB) and the two that are commonly mistaken for each other and lead to fatalities.

      The Double Fishermans is of course a common and suitable knot with a few disadvantages over the FOB

      1) Longer time to tie

      2) Harder to tie with gloves on

      3) Harder to untie after repeated loads, especially on new skinny ropes

      4) Larger profile and “lip” that makes it more likely to snag on retrieval

  13. Well written article David. I’d like to hear your input on the “alpine” butterfly bend, a knot with a low profile, easy to tie, very easy to untie after repeated loading, and no potential to capsize seeing as it is designed to take opposing loads. I personally use an EDK but wonder why the butterfly bend is not brought into this discussion.

    • Hi Alex, thanks for the comment! I wanted to keep the focus of the article to what I feel are the ones getting the most current press and suffering from the most confusion. The alpine butterfly is a very strong knot but here are my reservations against it:

      1) Harder to tie with gloves on

      2) Harder to teach & inspect

      3) Creates a “lip” that is more prone to catching, unlike the series of “flat” knots like the Flat Overhand Bend….

      And you personally use the FOB…. there is no such thing as EDK in my book 😉

    • Could you specify what knot you are referring to? In common practice I would say there are at least three “popular” knots:

      1) Flat Overhand Bend
      2) Flemish Bend
      3) Double Fishermans

  14. Just a thought: This is the first time I’ve ever heard of a knot capsizing… surely a knot fails, tightens, loosens or comes apart. What does ‘capsizing’ mean in the context of knot failure? I reckon it would be good to avoid jargon in such an important post – and in general. Capsizing is a nautical term meaning a vessel overturning in the water. How does that translate to knot failure?

    • Hi Peter, thanks for the comment! Capsizing is the same as “flipping” right? The way the figure of eight bend (and the flat overhand bend with enough force) usually fail in testing is they “flip over” repeatedly until they flip off the end tails and ultimately fail. Capsizing, and “flipping” conveys that the rope did not break at the knot, but inverted to the point of failure. I hope this helps! I’ll think about editing the post if it comes up again.

    • “Capsizing” means that the knot changes its geometry, in a somewhat dramatic way. The carrick bend, e.g., is typically presented to be formed in an open, lattice form, and then capsized into the final form. In the case of offset knots (not “flat” things!), the loaded strands in coming into the knot together at one point will have a strong force on prying the knot apart, which will often result in capsizing. E.g., Ashley’s #782 (IIRC) which is a “2-strand lanyard knot” (something intended to be mostly decorative in the out-&-back run of strands in some lanyard), could be loaded qua offset end-2-end knot; but if heavily loaded it can capsize into “Ashley’s Bend” (#1452) –a quite secure knot. (However, beware : it can also capsize in a partial/imbalanced way, and spill the knot –one end pulling out!)

    • Wrt the flat overhand, I’ve used 10.5 w/ either 8 or 7.7mm static no problem. Multiple raps all day long, and with some partners about 280lbs, no issue in rolling nor getting the knot stuck on a pull in alpine or a wall. Properly tied, dressed, and set, the flat overhand is simple, safe, and effective.

    • Hi Andrew! I haven’t seen any hard data to respond to your question with absolute authority but other than agreeing with MarkN I’ll say the often referred to “bible of climbing” Freedom of the Hills, and I believe the UIAA (the people who certify our ropes) recommend joining ropes be within 3mm in diameter…. This point has become more moot in the last 10 years IMO since none of us climb on 11mm ropes anymore. I would wager that a 10mm lead rope and 6mm tag line connected with a FOH will still hold 1400lbs in pull testing…. No question.

      • More specifically I am concerned with whether an FOB tied with two 8 mm ropes is more likely to collapse than an FOB tied with two 10 mm ropes.

  15. Its also worth noting that, if I remember correctly, that when an EDK inverts (capsizes, flips, etc…) it actually increases the force it would take for it to invert again. Almost making it impossible for it to roll more than once. So if your tails are long enough it physically should not be able to fail on account of the knot. I wish I could find the report I read online about it…

    • Don’t believe this! (And I think that some of the videos linked to here will show a different story –some have load-cell readings running along with multiple (more than once) flypings. And, i.p., the capsizing appears to often cross strands, which is not a good thing. .:. It’s better to do something with “long tails” than just leave them long on the dubious belief that the knot can roll but not “too much”! Prevent the rolling from the start.
      (The number of factors that can influence the behavior of these offset knots under load are many and no test has come close to checking them.)

  16. How about situations where the two ropes are of different diameters? I’m probably always going to use double fisherman knot – old school, and looks like it will work as I’m staring at it before lowering onto it…

    • Addressed in article. 3mm or less difference should not matter with FOB according to UIAA/AMGA and every test I’ve looked at. Nothing inherently wrong with “old school” though you miss out on new school benefits.

    • As I commented elsewhere here, ropes with diff. dia. can actually be better for the stability of the offset water knot : tie it so that the thinner (or if roughly equal, the more flexible) rope is the one making the choke of the loaded strands –that would be the GREEN rope in the image at the top of this page. (And a better knot comes by taking this rope’s strand an extra turn around the ends before tucking it out, forming a fig.9 thus.)
      The thinner rope will have a harder time being pried out around the thicker, than vice versa. (Which, yes, implies that getting this discrimination wrong will result in a more easy-to-roll knot.)

  17. I think you should put a side by side comparison circling the issues with small text.

    That would help people more. Good article though

  18. For many years I used the fisherman’s to join two ropes, but of course it has the disadvantage of getting caught easily. Then I learned about the overhand and first I wasn’t very convinced. I felt like it could capsize easily and get too tight with the weight. One day decided to try it and know it is the one I always use.

  19. When caving we use a double fisherman’s knot to join ropes together, ISOLATED in an alpine butterfly knot, Doing this allows the double fisherman’s to be tied and untied in an UNLOADED state, because the Alpine butterfly takes any directional loading.

    An added bonus of doing this is that you Alpine butterfly can be used to clip into, plus with the addition of a screwgate carabiner allows the whole rig to be used as a safe pull down rig!

  20. Thanks for the article, David. I’m totally new to climbing, so I’m quite confused by knot 2 and 3. As far as the pictures go, I cannot tell the difference between them – they both look exactly the same to me! That’s very concerning! Any way to tell them apart?

    • Hi Ramona, thanks for the comment and welcome to the wonderful world of climbing! I do not think you are alone in not seeing the difference between #2 and #3 right off the bat. The first thing to take notice of is the “tails” of the two options. #2 is tied by creating the “8” pre-knot in one rope, the same 8 we make when getting ready to tie into a harness before we thread the rope through our tie in point on the harness. It’s finished by threading the 2nd rope in the reverse direction through the formed “8”. The easiest thing to notice on the end result is the two tails (short ends) come out opposite sides of the finished knot. #3 is tied by holding both ropes simultaneously, forming a “twist and a half” then pushing the two tails through the formed loop. It is a very similar motion to tying #1 (my preferred choice), but there is an extra half twist. The easiest thing to notice on the end result is the tails exit the knot on the same side (like #1). The grave difference here is the structure of #2 greatly weakens the knot when loaded against it’s main axis allowing it to “flip” easily. I hope this explanation helps. This post has generated a great deal of conversation and while I hoped it would help clear up some confusion (and I think it has) it’s also created some, so I plan to post a video next week showing the methods each of these are tied. If a picture is worth a thousand words how much is a video worth?

  21. I remain fascinated by the epistemology of the discussion of rappel knots, in particular the distinction between qualitative and quantitative knowledge. The consensus seems to hold that a “properly tied” FOB will bear body weight safely. However, as near as I can tell there is not an accepted way to quantitatively determine if an FOB is “properly tied” which mean this is a qualitative determination. This means that if I rappel on an FOB then I am fully entrusting an asset that I can quantify (my life, I have exactly one) to a process that I cannot quantify (properly tying an FOB). Applying a qualitative process and expecting quantitative outcomes is just bad science, especially when my life is on the line.

    • Hi Andrew, thanks for your comment! There has been some research done on what kind of loads cause a FOB to roll when tied “properly” or loosely. The general consensus, IMO, is that “properly tied” means; 1) 12 in. or 30cm of tail 2) Dressed & Stressed. Keep strands parallel throughout (no crossover on any strand)… if there is a crossed strand the bend is not properly dressed. Tighten each of the 4 stands exiting the bend individually. If the above is followed I think the determination has been made that it is properly tied.

      • The UK’s Association of Mountaineering Instructors researched this in depth about 5 years ago and came up with similar results, published in our members’ magazine at the time. It is annoying to see what we simply call a double overhand commonly referred to in the USA as the “euro death knot”! You can see another piece about abseil knots on Andy Kirkpatrick’s blog: ” The ultimate abseil knot’

  22. Hey Dave, thanks for the article.
    Can you put in a photo of what you mean when tying a flat overhand bend with a tag line… i.e. ropes of different diameter.
    Also, what is the thinnest diameter tag line safe to use with a 9mm main rope?

    • Hi Rich, I’d like to leave the original article as is (made a few small technical edits) but I am happy to elaborate on what I tried to describe. In fact after some searching I found a decent article with the image we need (kind of). Check out the 3rd image down at this link. In practice I tie the simple overhand “around” the fatter tail, but the concept is the same. The more I research the more I feel this is an un-necessary step with a properly tied FOB, and while I have read in Freedom of the Hills that the UIAA recommends this option with two ropes within 3mm of each other I’ve seen other info saying greater differences are not to concerning (some might argue even less likely to roll). I don’t have hard testing to back up these impressions, but am happy to share them. So, a 9mm single rope with a 6mm static tag line IMO is an awesome set-up and 100% secure with the FOB. Tag lines less than 6mm present other issues so I wouldn’t go thinner than 6mm anyways.

  23. Thanks to all for the great links in this tread… I think it is clear that if either of the knots ( overhand or figure 8 in the «less likely to catch an edge configuration» ) if either are not well dressed AND do not have long enough tails… BOTH BECOME DANGEROUS.

    It is even scarier to see these knots being used to link up webbing and cordelettes… with very short tails of course…. 1 to 2 inches!!!

    I’ve seen several climbers with prussik loops closed up with the FOB ( David’s acronym for the overhand in the less likely to catch an edge configuration)…

    And found a few rap stations in Central Gully/Huntington »Ravine/Mt Washington with webbing tied with the FOB…

    I think there was one incident reported lately (EDK failure!!!) when the cordelette linked with a FOB with short tails failed when the climber loaded it to bail… failed not suprisingly!!!

    I teach and use the overhand with an second overhand configuration …and if an name is needed…suggest you name it the URK….Ultimate Rappel Knot… or the KTLRRTILLTCAE…knot to link rappel ropes that is less likely to catch an edge… 😉

    • Thanks for commenting Charles! I may add a reference link section to the original post soon. Your comments regarding the use of the FOB in cordelettes caught my eye. While I’ve never thought to do that with webbing, I, and many other guides, have started connecting our cordelettes with FOB over the more traditional double or triple fisherman’s for a few reasons. 1) It’s strong enough, especially when you are using the cord in a 3 point pre-equalized anchor. 2) Being able to actually untie your cordelette has a lot of advantages in creative anchor construction, especially if you need to cut sections off for v-thread anchors while ice climbing, or to back-up tat. That said I would never have 1-2 inch tails. Can you find a link or reference for this? “I think there was one incident reported lately (EDK failure!!!) when the cordelette linked with a FOB with short tails failed when the climber loaded it to bail… failed not suprisingly!!!” In reality being able to untie my 20 foot cordelette makes it easier to construct redundant bail anchors than having a welded double fisherman’s. Finally, I really like the idea of the URK… it solves a lot of concern without a lot of hassle. Stay tuned for my followup post hitting at 9AM EST tomorrow!

      • I’ve seen that on a few videos… FOB on one leg of a 3 legged equalized anchor… I imagine you make sure the FOB is on the longuest leg (the one that will get let least load in theory).

        I guest its «ok» since you have 2 other legs in the system… do you have any references/lab testing regarding this usage of the FOB on a anchor cordelette?

        I’ll stay away from this for now (I’ll wait for some testing to come out)… I dont like the idea of recommending redundant anchors if I’m not sure the FOB leg will hold if the other legs fail!!! But I fully understand the advantages in the ease to untie your long cordelette.

        Even with longer tails I dont think it is a good idea to use a FOB to link prussik loops or single legged anchors…what do you think?

        Here is the link to the edk failure article… it was not the overhand configuration but the figure 8 configuration… in this usage I thing the overhand would of simply taken a bit more time to fail.( overhand rolls vs the figure 8 capsizes) .. only one leg in the system!!!

        I give out liinks to my clients went I give out leading on trad courses… I’m adding your link/article to the lot…thanks again.

      • I love the conversation this article has sparked! I do not have any references/lab testing regarding FOB on anchor cordelettes. I think the variables would be too many to get easy to interpret data (strong leg vrs. weak leg) etc.

        As for linking prussik loops, no, I don’t use the FOB, mainly because the double fisherman’s provides a better spot to manipulate the hitch, especially if you tie your prussik hitch with knot “inside” the hitch so to speak.

        As for single legged anchors, I hadn’t thought a lot about this but pondering it now I would say I would discourage using an FOB linked cordellete in a single leg (example strong single tree anchor) via girth-hitch or loop because the obvious weak point will be the FOB. In practice with my own cordelette anytime I use a single point anchor I make the material redundant by doubling and tying off, so the FOB is no longer a real concern (two independent loops at this point).

        I’ve seen that link, thanks for posting it here, for sharing your opinion, and for sharing my link to your future clients! If you see me at the Moat Brewery post climb let me buy you a beer!

  24. I’ve thought it out…and by simply adding a clove hitch on the leg with the FOB….it solves my concerns….I’ll add this way to use ones cordelette… (advantage; total lenght of your cordelettte is easy to have).

    I can compare lab test used for the equalette.


    • > “I’ve thought it out … clove hitch
      > FOB on one leg of a 3 legged equalized anchor…
      Think better :: there is no reason to have the end-2-end knot IN THE LEGS –those can be *pure*–; put the knot in the clip-in loops, and ignore it. Here, an offset water knot (EDK) will be loaded if at all (unclipped, i.e.) qua stopper knot, pulled into the cordelette’s combining knot. You don’t need THREE eyes to clip-in to, two are fine, and let the third one hold the end-2-end knot (or NOTHING at all (unknotted tails) ). Consider that a cordelette is expected to work if any leg fails (nut pops out, say); same thing for the clip-in loops.

      • Thanks…good other option…my way is a bit closer to what we normally do…when building an equalette or when using a cord with a fisherman knot on it… putting a clove hitch to «isolate» the fisherman so its not in the way when tying the power point.

        I think we all agree that we rather not load the FOB.

        Always good to have options…thanks.

        ?for David… when using FOB on a cord…do you leave 12” and more of tails…+ a second «stopper» FOB….as recommended for the Ultimate Abseil Knot on rappel ropes.

        Most of the videos I see showing the FOB on a cord in a anchor….had a single FOB and short tails.!!!!

        It might be where some of the confusion starts regarding the infamous EDK.

    • Great convo KnudeNoggin & Charles. I think both options solve any concerns about over loading an FOB in a cordelette arranged anchor. To answer Charles question my FOB tails are less than 12″, probably close to 6″, and here’s why. Proportions. The theoretical purpose of the long tails is to give some space for the knot to capsize and form again, hopefully not failing a 2nd & 3rd time until it ultimately fails. I am working on the premise that a properly tied single FOB needs 1400lbs to begin this failure process. Since I have confidence it will not roll under the forces I will subject it to I see no need for 12″ of tail in 6mm cord, especially when it is only one component of a 3 point anchor. Now for rappelling, there isn’t two other legs to fall back on, and the rope is thicker, so I leave the recommended 12 inch tails, and just yesterday I rapped 4 times on the “Ultimate Abseil Knot” .. it gave my partner more confidence and I like it a lot, though will probably only add that extra step in certain situations.

  25. > “putting a clove hitch to «isolate» the fisherman so its not in the way when tying the power point.”

    Still, there’s good reason to ask Why tie the ends together AT ALL –and a grapevine bend consumes plenty of cord?!
    Bring the (uknotted) tails of the cordelette out through the big joining knot (typically an overhand) and just leave them. Optionally, THEN tie them, with an overhand knot (stopper). It amounts to two of the trio of legs being essentially eye knots with a tail, and the third is the odd one where all four ends see loading (both sides of the leg, of course, and both continuations of these strands will be in one or other of the TWO (not 3) clip-in eyes (the would-be 3rd being now two unknotted tails –or overhand knotted but unclipped). So, your tying of the main knot won’t need to accommodate bringing knotted ends through IT.
    It’s worth noting that the *shared*, multi-strand legs-joining overhand is just that and not simply an eye knot of 4 ends, but it should be amply secure, esp. in nylon (as distinct from hi-mod-core’d slick stuff)

  26. Simply because 99.9 % of the time… a cord loop (linked with a double fishermans) work’s for me.

    Used to carry a untied piece of cord.(in the late 70’s).. even took the time to regularly loosen my fishermans to make sure I could untie them in case of need)…dont anymore.

    If someones asks how to deal with untied piece of cord…gives me more options to share…thanks.

  27. Hi, gary d here, would a double fishermans knot backed up with half hitches ,work as well ? Allthough more time consuming,i thought this particular knot was bomb proof,

    • Hi Gary, that option is incredibly strong, but, like you say, is more time consuming. It also uses more rope, and is difficult to untie after heavy load. Much of what I am trying to convey is that a properly tied Flat Overhand Bend has been proven to be plenty strong and has the advantages I mention in the post. Simply put, in my opinion, a double fishermans, especially with backup half hitches, is serious overkill for normal recreational rappelling loads (rescue loads are different).

  28. Hi David, thanks for the article. One suggestion: replace the (!) with the stronger Death’s Head icon used by gear manufacturers. You could use Petzl’s (for example) and provide a credit in the text.

  29. Hi David, thanks for the article. One suggestion: replace the (!) icon used for Flat Figure 8 Bend with the stronger Death’s Head icon used by gear manufacturers. You could use Petzl’s (for example) and provide a credit in the text.

    • Hi Tim. Thanks for commenting and for providing that link. I was a bit surprised to see it. I have reached out to Petzl for confirmation. There is abundant testing showing the knot they have pictured failing at far less loads than 12Kn so I think there might be an error here. I’ll update my response once I hear back in regards to their data.

  30. David, I haven’t read all comments (yet) to see if this has been covered but, the term “flat” needs to be deleted, as it doesn’t well describe a particular property of a knot. However, the term “offset” describes an intrinsic property, i.e. that the bulk of the knot actually sits outside the axis of tension of the joined ropes. The term “flat” also leads to confusion which doesn’t help differentiate the between the two types of bends.

  31. There is absolutely nothing wrong with the offset figure 8 bend. It is actually stronger than the offset overhand bend also referred to as the EDK or flat overhand bend, both names which by the way need to be deleted from the climbers lexicon. I’m not sure why or how the offset figure 8 fell into the category of death knot. Test it yourself and you will see. There are two video links posted in the comments that demonstrate/prove this and not a word from anyone!

    • Greg, thank you for your comment. I could not find the video links you are referring to. Could you reply with a direct link to the video you mention?

    • Thanks for linking to these. I believe both of these videos clearly demonstrate the “8-version” of this bend fails at lower loads than than the Offset Overhand Bend. On this video reference 1:58 where the knot aggressively capsizes at just 6kN. The OOB earlier in the video resisted capsizing until at least 9kN. IMO there is no reason to use the option that capsizes or inverts at lower loads than the OOB or misnamed “Flat” Overhand Bend.

    • I’m not sure what the units of force displayed here are but again at 3:59 in the the bend inverts at a much lower force than the OOB inverts earlier in the video. I’m not sure how you see either of these two videos as demonstrating greater strength in the 8-version of this bend. To me they clearly show the opposite.

      • Knot strength is irrelevant in this case; the capsize load(s) are much more interesting. Offset 8 capsizes at a significantly lower load when not-well-tied than the Offset Overhand. Hence the 8 is lower-performing and should not be used, as a matter of best practice.

  32. Wow! Overall excellent comments! Only able to read through half so far. Sorry I’m just reading this almost two years after publication. (David-
    Likelihood, not likely hood.) Very interesting proposal to link better knot name communication to safety for the proverbial novice. I like and support the idea but IMHO, knot names will always be variable, localized, casual, inconsistent… multiple names for same knot, same name for multiple knots. The key is understanding knots- applications, advantages, limitations, comparisons, failure mechanisms… How do we motivate the novice to get proper training and the latest and greatest technical information for their safety? That is the right question.

    • They are tying a second OOB on top of the first one, not a true “double OOB”. This second knot likely increases security while negating one of the OOBs best traits “low profile, quick, unlikely to get stuck”. A single well tied OOB has proven to be more than strong enough for rappelling.

Leave a Reply