Articles in English

Anchor extension tests

Tuomas Pöysti 2021

The traditional wisdom warns against anchors that extend in case of one of the points failing. Of course every single real life anchor will change it’s shape when the load transfers from the original set of points to the new, reduced set, and this will inevitably result in some kind of extension – the anchor getting longer. Usually these warnings are associated with self equalizing anchors like the (sliding/magic) x or quad, which have certain ‘slack’ in case of failure.

Most of us are clever enough to see that extension is not a problem if there is dynamic rope between the anchor and both the climber AND belayer. This is the case in top roping (belayed from ground) and lead climbing. Lead climbing can be seen as constant development of a non-equalized anchor with an incremental potential extension! We all know, though, that the infamous zipper effect (cascading failure of protection points) almost never realizes, since the dynamic climbing rope both clips the impact force and effectively absorbs energy.

But what if there’s no dynamic rope between the anchor and the belayer? This is the case while belaying the second from a hanging stance. If the belayer is attached to the anchor with a static tether, like a dyneema PAS, is it possible for them to generate an impact that breaks the remaining anchor or their back, in case of one point failing and the anchor extending?

Put in a measurable way: how large impact forces are generated by a human body that is suspended on an anchor while it extends? (As we know, in this kind of situation, a human body is significantly different to solid steel mass.)

The test setup

In my “daughter’s room” lab, I constructed a setup for an empirical study. I attached a Rock Exotica enForcer to an anchor point A and another point at about the same horizontal level. Point B was releasable using a Tylaska T5 snap shackle.

In this case, the left, failing side was somewhat higher than the point that takes the final load.

In every anchor, regardless of the anchor length, the width was set so that point A was horizontally 20 cm from the master point. I suspended myself on the master point, clipping the carabiner directly to the belay loop of Ocun Quattro Weebee harness. My weight with gear was 82 kg.

I released the snap shackle and recorded the maximum force from the load cell, which sampled at 500 Hz.

The non-limited anchors

First set of tests was about the actual (sliding) x, one without limiter knots, as in the picture above. The materials are listed by their nominal lengths, which of course are not same as their real lengths. That’s why I measured the actual extension, that is, the difference in height of the master point before and after failure.

  1. Ocun 60 cm nylon sling, extension 29.5 cm
  2. Singing Rock 80 cm nylon sling, extension 51 cm
  3. Wild country 60 cm dyneema sling, extension 30.5 cm
  4. 80 cm Petzl dyneema sling*, extension 40 cm

The 80 cm dyneema sling was a 120 cm one shortened using an overhand knot. I was aware that this is not in any way comparable with a knotless one. The main reason for not going full 120 cm was mostly because there’s no room for that in my setup, and secondly because I was a bit shy with the drops.

The knotted dyneema.

I repeated each test three times. I was able to loosen the knot in the dyneema one each time. The knot was constantly on the left, failing side (B), and the stitches were on the right side (A) for each material.

The limited anchors

At least in my experience, most climbers use limiter knots in their x’s (or would use, if they used x’s to begin with). I included some cases with knots on them. This time I carefully rigged points A and B to the same height.

The materials were:

  1. Black Diamond 120 cm nylon sling, 21 cm between knots
  2. Petzl 120 cm HMPE sling, 21 cm between knots
  3. Black Diamond 120 cm nylon sling, 35 cm between knots
  4. Petzl 120 cm HMPE sling, 35 cm between knots

(Petzl PUR’Anneau, Black Diamond nylon runner). I used eight knots for HMPE and overhands for nylon.

Petzl PUR’Anneau with limiter knots 21 cm apart

The results

The non-limited materials (mean maximum force, average deviation)

  1. 60 cm nylon: 2.9 kN, 0.18kN
  2. 80 cm nylon: 4.0 kN, 0.2 kN
  3. 60 cm dyneema: 3.6 kN, 0.21 kN
  4. 80 cm* dyneema: 3.3 kN, 0.15 kN

The ones with limiter knots:

  1. 120 cm nylon, 21 cm: 1.73 kN, 0.06 kN
  2. 120 cm HMPE, 21 cm: 1.79 kN, 0.1 kN
  3. 120 cm nylon, 35 cm: 2.09 kN, 0.05 kN
  4. 120 cm HMPE, 35 cm: 3.22 kN, 0.12 kN


I think there’s not much new compared to my earlier cowstail as a fall arrester tests.

  1. When the fall distance is short (less than one meter), fall factors and tether materials are less important than in case of longer falls (and thus greater kinetic energies)
  2. The peak forces are far from able to rip solid pieces or cause injuries.

Petzl has published a study that showed similar or larger peak forces during lead falls. Of course this does not mean we shouldn’t avoid any excessive loads when possible. Test results like this should never be interpreted as “OK then, extension has zero effect on anything”.

I find this kind of data useful when considering options and compromises between them. Even recreational rock climbers should be able to exploit data based safety assessment. Such assessment needs comparable, factual criteria. For example, if one believes that a 50 cm”fall factor 1″ fall or comparable anchor extension is able to break a bolt, and thus their probability should be universally zero, what compromises might be made to achieve that (imaginary) goal? Is there not a better use for the mental energy and concentration?

And of course, if an aid climber completely stops worrying about impact loads, the results might be catastrophic. I tend to think, though, that the default mindset should involve less worrying about extension and focus on simplicity and redundance.

There is one more point I’d like to include. This probably isn’t new to most of us, but the knots really seem to do decent job as energy absorbers. As seen in the “80cm dyneema” case, an overhand knot makes a dyneema sling way more dynamic than a knotless nylon one.

The following image shows the difference in the final, extended lenghts of the “80 cm” knotless anchors. The dyneema sling was originally tied as close to the length of the nylon one as possible.

The final lengths of the knotted and knotless slings, that were originally equal. Bare in mind that there might also be a difference in how the materials set around the carabiners.

It is common and justified rule of thumb that a knotted sling should be derated by 50% (in strength). But if it’s not the sling we should worry about, but a sketchy piece on a technical aid route, then a knot might make things better. It should be less about strength and more about probability of load exceeding it. To assess these probabilities, we need a lot of studies like this and better!