---
title: 'This Material Sorts Itself'
source: 'https://youtube.com/watch?v=Z-2pTzSZtLU'
video_id: 'Z-2pTzSZtLU'
date: 2026-10-31
duration_sec: 1116
channel: 'Steve Mould'
---

# This Material Sorts Itself

> Source: [This Material Sorts Itself](https://youtube.com/watch?v=Z-2pTzSZtLU)

## Summary

This video explores the fascinating phenomenon of granular segregation, where a rotating drum filled with glass beads of different sizes spontaneously separates into striking patterns, seemingly defying entropy. The creator explains the underlying physics, including the Brazil nut effect, shear-induced size sorting, and the role of entropy, while also demonstrating similar behaviors in other systems.

### Key Points

- **Mysterious Pattern Formation** [00:02] — Glass beads of the same density form distinct patterns when rotated in a drum, despite expectations of mixing, leading to questions about entropy.
- **Key Variable: Size Difference** [00:44] — The unmixing occurs because the beads are different sizes; large beads end up at the bottom, contrary to the typical Brazil nut effect where large particles rise.
- **Petal Formation Conditions** [01:39] — Petal patterns only form when the drum is exactly half full; more or less fill reduces or eliminates the pattern, and square/triangle drums produce radial patterns regardless of fill level.
- **Shear-Induced Segregation** [02:11] — In a non-continuous avalanche model, large grains migrate to the top due to shear, where lower beads experience more force, causing upward pinching and downward falling of smaller beads.
- **Ratchet Mechanism** [04:50] — Large beads are pinched upwards but cannot fall back down because smaller beads fill the gaps, acting like a ratchet, leading to size segregation.
- **Finger Formation in Avalanches** [05:04] — Avalanches form fingers due to large beads piling up at the front, creating walls that are eroded by focused flows, a phenomenon also observed in real landslides and on the moon.
- **Petal Formation Mechanism** [07:16] — Petals form due to perturbations in the bead distribution that are amplified by the stacking of large beads, which have higher friction, leading to a nonlinear feedback loop.
- **Size Islands in Shaken Beads** [10:17] — When shaking a container of beads, adding smaller beads causes larger beads to clump together due to net forces from collisions, forming 'islands' surrounded by smaller beads.
- **Microscopic Analogy** [11:29] — Similar clumping occurs at the microscopic scale with colloidal particles, where smaller particles cause larger ones to aggregate and sediment, driven by entropy increase of the small particles.
- **Pentagonal Drum Anomaly** [14:06] — In a pentagonal drum, patterns never settle into rotational symmetry, which is explained by the fill level where the avalanche meets corners simultaneously, canceling mixing.

### Conclusion

The video concludes that granular segregation is a complex phenomenon driven by shear forces, friction, and entropy, with practical implications for industries like pharmaceuticals and even for understanding landslides and microscopic particle behavior.

## Transcript

glass beads become this pattern of separated glass beads? And what about these beads? Where did this pattern come from? They're all the same density, by the way, so you can't just shake it to get them to separate. And I certainly
didn't do it one by one with a pair of tweezers. Instead, I used this rotating drum, which you might expect would mix the beads like this. But when I use these beads, look what happens. This continuous avalanche at the leading edge
somehow deposits the beads into increasingly well-defined groups forming all sorts of striking shapes. But how weird is that? We're mixing something and it's becoming unmixed. Does that mean entropies going down? But
anyway, why does this unmixing happen in the first place? What's different about the setup on the right compared to the setup on the left? Well, the key is that the beads on the right are different sizes. In this case, the large beads are
blue and the small beads are white. This is a total pain for the pharmaceutical industry, by the way. Like, when you're mixing the ingredients for pills, you don't want half the batch to be 100% pure medication and the other half to be
100% filler. This is art, Mr. White. Actually, it's basic physics, but thank you, Jesse. So, what's going on? Well, you might be thinking, I bet this is the Brazilnut effect. But with the Brazil nut effect, the large nuts end up at the
top, but here the large beads end up at the bottom. And even stranger, you get the bottom. And even stranger, you get these radial stripes where the smaller beads migrate outwards a little in places leading to the formation of
petals. So where do they come from? Well, interestingly, you only get petal formation when the drum is exactly half full of glass beads. Any more or any less. and the petal pattern goes away or at least it's drastically reduced. And
in square drums and triangle drums, you get these radial patterns regardless of how full they are. So where do they come from? Well, the key thing is this kind of continuous avalanche that you get when you rotate the drum because it
turns out something quite surprising is happening in there. To see what's going on, we built this non-ontinuous avalanche in here. There's a mixture of large black grains and small white grains. And the surface is coated with
turquoise grains just to add roughness. And look, when the avalanche stops, you always end up with a line of black grains at the leading edge. So why did they separate out? Looking at the avalanche in motion, you can only really
see the top surface that's moving quite fast. But you also know that everything at the bottom of the avalanche must be stationary because of that high friction turquoise surface. And so there must be a gradual increase in speed from the
bottom surface of the avalanche to the top surface of the avalanche. In other words, that whole layer of moving beads is being sheared like this. To see what happens when you shear a load of glass beads of different sizes, we built this
model that shears a small packet of glass beads back and forth. This is a time-lapse where the timing between each frame of the video is very slightly longer than a full cycle of this box. So, it looks like it's tilting very
slowly side to side, but you can see how the large grains eventually migrate to the top. And in that way, it's a lot like the Brazil nut effect, but the mechanism is different. Any one grain in this mixture is being squeezed from
various directions by its neighbors. The force it feels from the lower down beads force it feels from the lower down beads will be greater than the higher up beads because the lower down beads are being crushed by more weight from above. It's
only a small difference, but it's enough that beads tend to get pinched upwards more than they get pinched downwards. This pinching upwards leaves holes that This pinching upwards leaves holes that other beads can fall into. So, you get a
other beads can fall into. So, you get a general mixing of beads via pinching upwards and falling downwards. But when you add smaller beads into the mix, you add smaller beads into the mix, well, they tend to fall into those holes
that are left behind by the pinching up motion before they get big enough for the large beads to be able to fall back down into. It's a bit like a ratchet. The large beads are pinched upwards, but they can't fall back down again because
the smaller beads get in the way. So over time, the large beads migrate to the top and the small beads fall to the bottom. In this model, the beads are sheared one way and then the other. But in this avalanche, the beads are
continually sheared in the same direction. That means that when the large beads migrate to the top, they continue to be sheared away ahead of the eventually stops, you find that large
Actually, a slight tangent here. You might have noticed that the avalanche forms these fingers. And in fact, landslides on Earth do the same thing. That's because even before the avalanche has come to a stop, you have large beads
reaching the front, spilling onto the ground, and piling up in front of the moving beads behind it. Because we've chosen particularly rough beads here with high friction, you get a kind of wall forming up the front. The wall
isn't perfectly even because of random fluctuations. So, some parts are easier to push through than others, and so we should expect to see grains pouring should expect to see grains pouring through in those places. This focused
flow erodess the wall underneath, which reinforces the flow in that location. This is bad for real landslides because they can reach populated areas that are really far away from the initial slide. Amazingly, it also happens on the moon.
This is what it looks like when loose material slides down inside a crater, for example. But anyway, the point is you can see how that would lead to large beads being deposited at the bottom of this rotating drum and the small beads
being left at the top, but it doesn't explain the petals. My first thought was that it's maybe related to the banding you see in these sandscape toys. Looking at this time lapse, it seems like sand is deposited in bursts and that grains
of different colors fall at different rates. So that for each burst, you get a layer of dark fast falling grains with a layer of white, slow falling grains on top. You can even see there's some kind of bubble mechanism that leads to these
bursts of deposition in the first place. And actually, you can get something similar in the drum if you run it really slowly. See here, it's so slow that the slowly. See here, it's so slow that the avalanche periodically stops and starts.
And each new slip is like a new burst of deposition in the sandscape toy. So you get these tight stripes forming. But we get petals forming even when the avalanche is continuous. To see why, let's first run the drum really fast
because that smooths everything out and gets rid of the petals. Okay, so we'll slow it down again. But notice that the boundary between the two regions isn't perfectly smooth. Like in this strip here, there's a bit more of the blue
beads than the white beads compared to nearby strips. This strip is about to tumble down the avalanche. And as always, the larger blue beads rise to the top where everything is moving faster, so they end up at the front. But
crucially, larger beads have more friction than smaller beads when they're because that's a whole other thing. But it means that when the blue beads first hit the bottom, there's so much friction that the blue beads coming in behind
aren't as easily able to flow over and instead stack up. What that means is this short wide supply of blue beads becomes a long narrow deposit of blue
beads. So what was a small pertubation is now a slightly larger pertubation. Now, when the blue beads are stacking up, you can see that the top of the stack is moving upwards as the stack grows. And that's happening at a certain
speed, which means the stack takes a certain amount of time to reach its peak. And in that time, the drum has turned this much. Now, what if there were regular pertubations to begin with that were spaced this exact amount?
Well, then this stacking effect would always be reinforcing existing pertubations. In reality, you don't start off with these perfectly spaced pertubations. You get a whole jumble. So, what I'm really saying is that the
So, what I'm really saying is that the system amplifies pertubations of a certain size. I'm oversimplifying a bit here. And there are other mechanisms here. And there are other mechanisms that might be equally or more important,
but I guess that's the leading description of what's going on. One interesting detail actually is that it's nonlinear. In other words, the pattern nonlinear. In other words, the pattern size that gets amplified changes as the
pattern grows, which means you often get these nice transitions where small petals get consumed by larger neighboring ones. You can see that really nicely here in this time lapse. This explanation is paraphrased from one
given by a group at the University of Manchester. Actually, a big thank you to group who really helped out with this video. They even sent us a load of mixes video. They even sent us a load of mixes of different colors and sizes of grains
huge help. By the way, the only reason it works when the drum is exactly half full is because when it's over half full, any slice of beads that slide down are then chopped up on the next go round. So, there's no reinforcement
mechanism. The same is true if it's less than half full. But when the drum is exactly half full, these slices stay together each time they go around with just a bit of interaction with its neighbors. I want to show you one more
really cool thing where glass beads separate into size islands. When you shake them, we're vibrating the container, which causes the beads to fly around randomly. But watch what happens when we add some smaller beads.
Incredibly, the larger beads start to clump together. To see why, think about collisions between the small beads and the large beads. When a large bead is surrounded by small beads, it's being knocked from all directions, which
averages out over time to zero net force. But as they drift around, they will occasionally come into close contact with other large beads. And there's no room between them for the smaller beads. So now this bead is
getting a net force in this direction and this bead is getting a net force in this direction. In other words, once two large beads come together, they tend not to come apart again because there's a net force pushing them together. And so
over time, the large beads form these islands surrounded by a sea of the smaller beads. It's cool, isn't it? But incredibly, something similar happens at the microscopic scale. If you add tiny solid particles to a liquid that are so
small they experience brownie in motion, then they never form a sediment. They stay dispersed in the liquid. But if you introduce even smaller particles, then they act like the pink beads in this model, causing the larger particles to
clump together and fall to the bottom. This really does look like it breaks the second law of thermodynamics because surely separate groups of particles are more ordered. But look, if we track where the center point of a small bead
is allowed to go, but we introduce large beads, then the center of a small bead can't get any closer than this. So, not only is a small bead excluded from the
volume taken up by the large beads, there's an additional exclusion zone around the large beads. But look, when two large beads come together, their exclusion zones overlap a little bit, which means this volume here is only
excluding the small beads once, when before it was excluding twice. So every time two large beads come together, the small beads gain a bit of extra room. And by giving the small beads more room to move around, we increase their
entropy. And the brilliant thing is when the large particles clump together like this, the energy doesn't go down. So this separation isn't driven by an energy gradient. Instead, the process is actually driven by the increase in
entropy of the small beads. By the way, the reason this separation of beads doesn't break the second law is because even though the entropy in inverted commas is going down, the second law only applies to proper thermodynamic
entropy where the thermal energy in a system is causing everything to constantly mix. Whereas this mixing of beads isn't proper thermodynamic entropy. It's just a system that shares a lot of the maths. Anyway, that was a
bit of a tangent. I really want to get back to the square and triangle drums because I never explained where these shapes come from. This is a simulation, by the way, that just takes whatever is in this little strip and deposits it
in this little strip and deposits it over here in strict size order. But look what happens on the second rotation between here and here. Most of the beads in the avalanche are green. So you get a big pile of green deposited here. Then
you're sweeping through mostly red. So you get this thick arm of red beads here. And when that comes around again, the effect is even stronger. Just when I thought I understood all this, we printed a pentagonal container. And the
weird thing with this one is, well, the pattern never [clears throat] seems to settle down. It never develops any kind of rotational symmetry. And I have no idea why. Okay, so a quick update on that. I just had a call with Chris
Johnson who did the original research and I think I do understand it now. &gt;&gt; It seemed very suitable for you given that it's it's flows between two perspect sheets which is [laughter] yeah &gt;&gt; nobody's really looked at polygons
beyond a square to any great extent. So, it turns out we got kind of lucky because this refusal of the grains to form any kind of rotational symmetry only happens when the drum is filled up
exactly to this line where it's touching these two corners. And that's because for symmetry to appear, the grains need an opportunity to mix. So it turns out an opportunity to mix. So it turns out that the the mixing all happens from
when the interface hits one of the corners and if it hits two corners at exactly the same time that cancels out. &gt;&gt; To see what Chris means, let's follow one single grain and see where it ends up after a full turn of the drum and
another turn and another turn and so on. Look, if you just let it run, you can see that one grain ends up moving all around all over the place. And if we there's a lot of really good mixing going on. But if I now fill the drum to
this exact level here, watch what happens to those tracer grains. Now, see, at this exact fill level, the particles don't mix. They stay in their lane. So any asymmetric arrangement of the grains that gets locked in at the
beginning can't be undone by grain mixing. Why don't you get grain mixing Well, Chris explained that mixing happens when the avalanche meets a corner up here or a corner down here, but if it meets both corners at the same
time, the effect cancels out. I'm not totally clear why that would be, but one clue is to look at how the center of the avalanche moves. When the drum is filled almost to that perfect level, the center
point does a rapid little jiggle as it turns through the corners. But if the corners line up exactly, the jiggle goes away. But anyway, there you go. You can tidy up your unruly glass beads with a
simple gliding motion. And the same is true for your facial hair, but only if you've got a good quality razor. Specifically, you need to avoid those plastic cartridge razors because with those, there's always a little bit of
give. So, you can never get a consistent shave. There's an easy fix, but manufacturers won't do it. I'll get to why in a minute. Instead, what they do why in a minute. Instead, what they do is they add more blades to try and get
more hairs in a single pass. But the extra blades cause irritation, so they make the blades springier, which means that you press harder, which causes more irritation. So, they add a lubricating strip, and then even more blades. So,
you've got this ridiculous escalation when actually there's one simple solution, which is just to hold a single blade firmly at exactly the right distance and exactly the right angle. Unfortunately, you can't do that with
cartridge razors because they're made of plastic and they have to be made of plastic because they have to be cheap to make. So, the real solution is don't use make. So, the real solution is don't use a cartridge razor at all. instead buy a
precision engineered aluminium handle that holds a single blade. And if you're the sponsor of this video, Hensen shaving. The Hensson AL13 is engineered to hold the blade firmly in place at exactly the right angle and distance so
that you don't have to apply pressure. And the single blade means that you're not causing unnecessary irritation. The handle is more expensive, but you only ever buy it once in your whole life. And if you use promo code Steve Mold at
checkout, you'll get 100 free blades with your purchase of an AL13, which is equivalent to about 3 or 4 years of shaving. So, it ends up being cheaper than a cartridge razor after just a few months, and you get to own something
really nice. If you're interested, go to hensenshaving.com/stevemold. The link is also in the description. So check out Hensson Shaving today. I hope don't forget to hit subscribe. And the algorithm thinks you'll enjoy this video
algorithm thinks you'll enjoy this video next.
