top of page

How NOT to Break Stuff

The Ultimate Guide to Back Yard Science

HowNOT2 Course

Breaking down how to break stuff

What I do at HowNOT2 isn't rocket science, it's backyard science and it's accessible to almost everyone with enough stoke and creativity. However, it's important to do it super good enough to find out useful data AND not break the things you are not trying to break. I've been break-testing things since 2017 and it was just to find out if my sewn loops were safe and I discovered how much there was to discover. Thousands of break tests later of a dozen different setups and I'm still learning how little I know, but what I have learned I will write out here in this course to help you in you are interested in exploring real answers in "real world" scenarios.

What's the Difference?

Backyard Science vs "Real" Science

Science, in the climbing industry, is deified as it is all behind closed doors. The few clips of break tests that find their way online are polished versions often times without context. This isn't some conspiracy as they are investing a ton of time and money into making better products for us and don't want us to know their "finger-licking recipe". However, people assume it's flawless testing by people who are flawlessly planning these tests with unlimited access to machines that will put the gear to the ultimate test. As you read that, you might say "of course nothing is perfect" but that is only because it sounds like such a crazy assumption when spelled out. None-the-less, when you grab that rope and carabiner you feel warm and fuzzy from that marketing label. "Oh, but there are standards they meet," you say. Yea, we'll get to that.

Backyard science is looked down upon by 42.3574% of engineers because it doesn't fit in the computer programs that do the math for them. A joke like that divides engineers by 2. Get it? The main things that set the types of sciences apart are sample sizes, variables, and thoroughness.

Sample Sizes

If you pull on something the same way 3 times, you will learn something. BUT, if you pulled on it 30 times, you'd get enough data to see the outliers, or the extreme results on the bell curve. It's not that unlikely that you can roll a 6 on dice 3x in a row, we've all done it, but you can't say that you get that 100% of the time. Roll dice 1000x and you'd start to match more closely what probability math tells us we should be seeing. From my experience, metal products without welding are fairly consistent and soft goods vary A LOT in results. I'd feel confident in the data, breaking 5 carabiners but (good) companies are likely to break hundreds and then some from every batch. A rope in a knot on the other hand needs at least 10, if not 30, if not 100, tests to really find out the true range as the outliers are usually far from average. Manufacturers rarely bother with knots and just use big diverters (drums) to eliminate the variables of knots.


The biggest difference between the sciences is the variables. Real science wants to eliminate every variable possible to isolate the question to be answered. To test deet's affect on ropes, real science would apply it exactly the same way, to the same rope, in the same spot on the pre-made rope samples, in the same temperature, and pull them all at 100mm per second on a diverter. Backyard science might just pour a bottle on a small old rope laying around and then cut it up, tie knots, and pull on it with a truck at different speeds on a hot day starting in the cool morning and ending in the blistering heat. Backyard science is seen as sloppy.


When can you make a conclusion? Back to the deet example, real science would carefully add variables like length of time on rope. They would do 10 control samples of no deet, 10 samples with it on for 10 minutes, 10 samples for 24 hours and 10 samples with it on for 1 month. Did you see what happened there!?!?! One variable - length of time exposed to deet - just created 40 samples! When can you make a conclusion that deet does not harm ropes? After pulling on 3 samples with a truck with all those variables mentioned? What if nothing breaks lower??? Call me biased, but that is a great starting point to find out if you are onto something. Once you see deet does affect a rope, then you can put on the lab coat and do those 40 samples to learn the specifics.


You don't need to be missing a front tooth to be doing backyard science and you don't need a lab coat to do real science. If you want to just explore and see what happens, great. If you are trying to determine if if it safe to rig something abnormal, you may want to be ..............

STEP 1: What's your goal?

What are you trying to accomplish needs to be answered BEFORE you plan your tests because there are a lot of ways to skin the cat.

Testing what breaks first

Want to see if your knot breaks before your ascender desheathes the rope? Want to see if a super 8 knot is redundant if one eye breaks? Want to compare to knots to see if one consistently beats another? These are the easiest kind of tests to do because it doesn't require a load cell to tell you the force you generated. You just need to know what happens.

Testing real world scenarios

Instructions tell you everything you can do with a product, not everything you CAN'T do. Want to know what happens if you put a sling directly on a hanger? Does a Purcell prusik PAS slip when drop tested? Want to use a micro-traxion on your alpine adventure with a rope that isn't normally used with it? These things can be done with or without a load cell, depending on if you need to know the actual force. The difficulty in testing it can be simple from just dropping a foot on some gear (backed up to something else) all the way to requiring a hydraulic pull tester.

Testing product limitations

If you made some soft shackles or whoopies and want to test when they break, you need something that can pull hard and a load cell that can take a beating. A lot of climbing gear breaks in the 20kn range and you can do A LOT under 30kN but some slings, carabiners and ropes in loops can get above that.

Testing to make a product

If you are inspired to make some gear, it needs to be more than super good enough. It needs to be CE certified to be sold in Europe so it needs to meet those standards and the UIAA has its own standards that are often similar but slightly more rigid than CE. You need to buy the standards and follow their instructions on how to test the gear. BUT that is only for the development stage. To get certified, you need to send the gear to a third party for testing which you have to pay for and then pay for the certification itself. Do you want to sell that in blue also? That also has to be tested separately. Hope you weren't trying to be profitable!

RANT: Standards are good for comparing one product to another but in order to do so, they have to eliminate most variables down to a sterile environment. It costs money for them to create the standards and to make sure everyone is following them, it's a service and it costs money, and that's fine. However, you, as the end user of the product, don't get access without paying for the standards, to know how your gear is being tested. The gear YOU trust YOUR life to. Why do we have to pay to find out carabiners are tested with a 10mm pin pulled at 100mm per second, or UIAA falls are 80kg of steel weight at a 1.8 fall factor? HMPE (Dyneema) standards are actually wacked as they pull eye to eye several times and hold it at a high force, then pull to destruction. Why you ask? Because they can get a much higher force than if they pulled a sample normal. You may also be shocked to find out how loose the standards are. They give general guidance but leave plenty of wiggle room. Rumor has it there is very little guidance on how to break cams.

This is the most boring and difficult type of testing. It requires consistency in your tester and thoroughness in your experiment.

STEP 2: What are you breaking?

This matters a LOT on how you break it. Ropes can stretch A LOT. Metal makes stuff go flying. Bolts are a PITA. You need to understand what's involved with the material you are messing with.

Soft stuff

Ropes stretch, recoil is insane. Soft shackles are strong. What is your bend radius

Hard Stuff

Carabiners go boom.


Mobile setups requiring 10,000lbf is not easy. Connectors fail

STEP 3: What kind of test needs to be done?

Once you know what you are trying to break and what level Do you need to slow pull it, drop it, pull on it a million times or rub it a lot?

Slow Pulling

PUlleys vs hydraulics

Drop Testing

Towers vs existing structures/rocks

Cyclic Loading


Abrasion Testing


STEP 4: What force are you expecting?

Stronger gets harder

0kN to 4kN

You can drop test your body and pull with a 3:1 with a friend. Pull direct with a truck.

4kN to 8kN

Pulleys, or drop test on rocks. No human testing at this level. 2:1 a truck.

8kN to 16kN

Hangers get deformed, your system could start breaking, things recoil more. 5:1 with 3 friends

16kN to 30kN

LS3 can handle this but don't damage it. Careful using biners. 5:1 with 8 friends or winch or truck. SIT DOWN when you pull. Drop towers and cyclic loading machines stop here. Instrons stop here too.

30kN to 50kN

Crane scales or legit load cells are required. Shackles on everything. 1.5" rod hydraulic maxes out here.

50kN to 100kN

Big shackles with reducing shackles. 2" rod, 4" bore hydraulics. Don't watch

100kN to 200kN

Very dangerous. You will break the stuff you are not trying to break.

STEP 5: Load Cells

If you are going through all the hassles of breaking something, you probably want to know the force your sample failed at.

Kill A Newton

Don't worry, Issac is already dead. So you can kill-a-newton. Or do a 1000 newton jokes. A kilonewton is a universal way to measure force because if you do lbf or kgf 50% of the audience will bitch everytime.

Resist the Force

A load cell is a fancy resistor. Electricity goes in, the load cell deforms ever so slightly when force is applied to it, changing the voltage when doing so. The change in voltage is measured and assigned a force to that new voltage. Do that 10x and you have "calibrated" it for the computer to know if it reads .09785342v that means you got 12.34kN.

Know how much it hertz

How often it measures the voltage change is the hertz. You can measure the force but if you miss the peak, you might think a fall on a personal anchor only causes 2kN of force to If you are lifting something slowly, the peak force is just the weight of the item so crane scales are often times just 10hz or 10x per second. This works in slow pull machines since your slowly getting to the ultimate force before it breaks.

If you fall on a dynamic rope in a gym setting and use a crane scale, you will miss the peak force as it is only going to last 1/25th of a second. A LineScale 2 was 40hz in fast mode and that is fast enough to capture drop tests with stretchy materials.

Very static ropes need about 100hz and LS3 goes up to 1280hz.

Steel cable needs about 2500hz to pick up the tippy top peak of the force graph.

Get Unplugged

Load cells require wires unless they don't. Wires suck. LS3 is awesome. Crane scales are too.

Crane scales

Ebay baby!

Line Scales vs Enforcers (all in ones)

Kiss it

Load Star Sensor

Wired options

STEP 5: Gear Porn

If you don't film, did it happen? Unless you have a shield, you can't watch it live. Film it so you can review it

Phones vs Cameras

Kiss it

Slow mo myths

120fps is lame, you need lots of light



bottom of page