Articles in English

ABK vs inline eight

Tuomas Pöysti 2021

I have written an intentionally provocative article about inline figure of eight loop in Finnish. The article is not meant to be read as a factual statement, but rather a thought-provoking essay. The main message is that the directional figure of eight is useless.

In this article, I’ll try to provide some facts to back the claim up. Or to undermine it, however the facts are interpreted!

The knots

Let’s have a brief look into the rival knots. The ABoK numbers refer to Ashley Book of Knots, which is the standard knot reference at least then it comes to identifying knots. The names are part of language and thus quite fuzzy, whereas the ABoK reference numbers are absolute.

Inline figure of eight loop

The inline figure of eight loop, ABoK #1058, is for some reason a quite common inline loop knot among rock climbers. The function this knot serves is to provide a loop in the middle of a line so that the line can be tensioned through the knot, and the loop can be loaded as well. This is useful when building anchors or in varying improvised rescue situations.

The inline figure of eight loop is also known as directional (figure of) eight (loop), since the loop can only be tensioned in certain directions. In the case of these pictures, the correct loading is between the left of the rope end and the loop.

Alpine butterfly

The alpine butterfly, ABoK #532, might actually not be at the peak of it’s popularity in the Alps, but where ever rope technicians work. It is introduced in the Irata ICoP, the “bible” of ropework techniques according to the worldwide organization.

The alpine butterfly knot or ABK may be loaded between either end of the rope and the loop. The knot is not symmetric, though, which means the two loading cases are somewhat different in a technical perspective – we’ll see.


I don’t think strength or efficiency is a very important feature of a knot. It certainly is not the most important one, but since we tend to appeal to simple, quantifiable attributes, let’s take care of this one first.

The strength or “efficiency” of a knot means the relative residual strength of the rope after tying the knot. Knots are different in how tight bends the rope is forced to make inside the knot, and how tensioned the rope is at those points. The residual strength is typically expressed in percents.

A good rule of thumb is to assume a 50% reduction for every knot. This means than if a knot happened to have typical residual strengths close or lower to 50%, it would be problematic. On the other hand, if we have two otherwise equal knots, of course we should prefer the one that can be shown to give statistically higher residual strengths, no matter what the absolute figures are.

According to an excellent meta analysis by Tomas Evans (ITRS 2016), the alpine butterfly is generally stronger:

  • Inline Figure 8 (End to End) 48.2% … 58.7%
  • Inline figure 8 (Loop to End) 62.5% … 74.7%
  • Butterfly (End to End) 59.2% … 68.8%
  • Butterfly (Loop to End) 60.7% … 80.6%

How easy they are to untie

Another attribute that might not be the single most important one, but that is quite easy to assess with some level of objectivity is how easy it is to untie the knot after tensioning.

I conducted a little experiment. I tensioned three pieces of each load case using a retired EN 892 rope (a dynamic climbing rope). Each knot was tied into a segment of rope that had not been tensioned during this experiment. The used tension was 2 kN, which was applied using a motorized winch and measured with a Rock Exotica enForcer load cell.

There are two load cases for the inline figure of eight: loop to end and end to end. The alpine butterfly, on the other hand, can be loaded using either end of the rope, and this has an impact on how easy it is to untie afterwards.

Loop to end

In this article, I use the following definitions for ABK loop to end load cases 1 and 2:

Simply put: load case 1 is from the “lower” leg, load case 2 is from the upper leg.

The ABK’s loaded with 2 kN:

It turns out that the ABK load case 1 is significantly harder to untie than the load case 2, and somewhat harder than inline figure of eight. The reason is clearly visible in the following images:

End to end

I tied both knots next to each other and tensioned the line over them to 2 kN, repeating three times.

The knots before end to end loading
The knots after 2 kN end to end loading
Another view that makes it obvious

The ABK is bowline-level easy to untie. It can be done with one hand. The end to end loaded inline figure of eight was the hardest one in the whole study.

Stickiness conclusion

In my probably biased opinion, this is how easy it is to untie the loaded knots:

  • ABK end to end: Very easy
  • Inline figure eight loop to end: quite easy
  • ABK loop to end, case 2: quite easy
  • ABK loop to end, case 1: quite hard
  • Inline figure eight end to end: very hard.

Of course, in real life, both types probably coexist. I guess it does not make things better for the inline figure eight, but this should be actually tested. The current data suggests that the ABK is not worse, at least.

A note on the efficiency figures from Evans’ meta study: In general, there might be an inverse correlation between how hard a knot cinches and how efficient it is strength-wise. There also might be a difference in efficiency between cases 1 and 2 loop to end loading. The ABK loop to end data happens to have bigger deviation than the other load cases. This does not necessarily mean there’s a connection, there, but it would be interesting to further study this!

How easy they are to learn

THIS is an important one, but way harder to study and put into a quantified format. To me, the alpine butterfly has always been easier to get going. I have a trick that I normally use, but sometimes, in an awkward situation, the muscle memory may fail regardless of a thousand repeats of practice. I’m positive, though, that no matter how tired, drunk, sick or beaten I am, I can always pull off the super visual “snowman” version of tying the ABK.

Make a snowman by twisting the rope 360º. The snowman then bends over, dives underground and goes through its own bottom and head. That’s so weird it’s hard to forget.

In the beginning, I quite carelessly defined the inline figure of eight as the ABoK #1058. ABoK #1057 could also be considered an inline figure of eight:

ABoK #1058 and ABoK #1057

In general, practically no literature seems to suggest ABoK #1059 is the “correct” version of inline figure of eight. If, however, we accept it as an inline loop knot, whatever it should be called, it makes tying an inline figure of eight easier. Or to be exact, easier to end up with a feasible knot when one is aiming for an inline figure of eight.

Having two different outcomes, on the other hand, is not desirable when it comes to visual clarity.

Visual clarity

This point is also important but hard to assess. How easy it is to see if the knot is properly tied?

The inline figure of eight does have a clear advantage here, being related to the most popular climbing knot there is, the figure of eight loop (ABoK #1047). Actually, the inline figure of eight is a rethreaded figure of eight loop without the last thread.

The alpine butterfly does have a false version that might not be idiot-proof level easy to spot, but it is not hard, either. I’m not aware of any such false versions of inline figure of eight, if the aforementioned case of ABoK #1059 does not count.

The false “snowman” that leads to the false ABK above

On the other hand, the alpine butterfly is also very widely used as such. By Irata rope technicians, to whom the figure of eight loop (the end of line version!) is also very important in daily use. It seems to be possible to have both two in one’s toolbox and use them safely.


Popularity tends to be an argument for certain practices in climbing. It makes sense to keep things a bit conservative. Especially when it is about the most basic things that are done by large masses of climbers every day. And particularly when cross checking applies to the practice. Cross checking is not possible if the counterparts have different practices.

But every practice must start with a first time. I guess the bowline is less popular as tie-in knot than it used to be, as the figure of eight loop has taken over. A few people use figure of eight ascender as a belay device today. They either retired or learned how to use a plate. We are moving away from attaching the descender backup prusik into the harness leg loop.

These are very basic procedures, whereas an inline loop knot is related to advanced, probably improvised setups like special anchoring or self rescue. These things are not taught on basic courses, and in any case they require a lot of insight and responsibility.

At least it shouldn’t be up to the (advanced course) instructors not being able to acquire a new knot into their teaching catalog.


To me it seems like

  • The alpine butterfly knot is more versatile than the inline figure of eight
  • it is at least not harder to tie or check
  • it is generally easier to untie (or at least not harder)
  • it is stronger

I’d be happy to be convinced that we should teach the inline figure of eight instead of the alpine butterfly, where ever any of them is needed! Or is there a third option?

3 replies on “ABK vs inline eight”


thanks a lot for your article. Interesting to read – I usally use F8 and/or AB for rigging ropes in different situations. My understanding was always that AB is ‘best’ for several situations and your test basically showed that again… just the untie situation of a F8 can be challening enough sometimes 😉

Greetings from an Austrian rope access tech in Turku 🙂

Hallo Snuffi!

Thank you! I learned a couple of things afterwards, too. There is an argument for the IL8 I hadn’t noticed: it is easier to adjust (the size of the loop).

And when I did some “more realistic” tensioning tests, like 1kN loop to end, 1.5kN end to end, 2kN loop to end, 2 kN end to end to a single knot, I realised that it is way easier to untie both of these knots if they are tensioned “evenly”. The effect was stronger with the IL8.

Even: If you first put 2kN to an IL8 end to end, it will take about 20 seconds to untie. But if you then put additional 2kN loop to end, it will take less than 10 s to untie. It kind of balances the knot out instead of making things worse.

But anyway, I have not found a single expert who’d pick the IL8 if they had to stick with just one inline loop knot.

IL8 is as you mentioned in your article a good tool for rock climbers – in my area of work (so rope access) a IL8 can also get bit risky if used as main anchor – but this an experienced tech should see right away …

Thanks for the other test results too, interesting to see – I wouldn’t know for myself where to use a IL8, but who knows now where I have some more details ^^

Comments are closed.