 And I thought the elephants are a good way to start this conference because they're probably the biggest animals. I think we'll start with elephants and then we'll get to maybe flies and sea elegans. For them, their movement is we're studying their movement of their trunk, which is in this picture here weighs 100 kilograms, but they can use their trunk to pick up very fragile items. Like that's actually a tortilla chip. We eat them in the United States. They fracture very easily, but the elephant can pick it up without breaking it. I'll show you how. The elephant can also pick up other items of various numbers and sizes. They're very, very hungry. The elephants, they eat 200 kilograms per day, so they have to basically pick up and eat 100 grams every minute, which means about two bananas every single minute. So they often pick up many items at once. There's a recent paper talking about how elephants view the world. And people think that it's not very much through vision, they're nearsighted. So there's a recent PNAS paper by Josh Plotnik showing that they can actually determine the number of food items by smell. So they can actually put their nose and determine which to eat by how strongly it smells. And they can count the number of seeds roughly by smell. So in the middle of the presentation, I'll show you a model and some experiments on animal sniffing rates and why animals sniff, why we don't take long inhales and why we, you know, at least if you're an animal, go inhale and exhale. The last part of the talk, I thought we'd have a little fun, even though it's not part of the official title we'll talk about, does anyone know what that animal is? Do we have anyone from Australia? Wombats. Wow. Yeah. Wombats, good. And that's their feces. So they make cubic feces. And we actually, we're working with an Austrian scientist. We have wombats, intestines, and feces in our lab. And we have some ideas for why a soft intestine can make edges and squares. So hopefully we'll have time to get to that before coffee break, but we're gonna have brownies or something that's some other chocolate item. So a lot of this work is done by my graduate students and postdocs. And this is done over the last two years at Atlanta Zoo. We have a zookeeper that basically keeps the students safe because an elephant at any time can really grab somebody and really just break their neck. I'll show us some dissected elephant trunks that's done by Joy Reidenberg, who's on this show dissecting nature's giants. I met her on Twitter, and she found the only elephant trunk of the United States that was able for us to dissect. In the second part of the talk, I'll talk a little about this device we built that mimics the sniffing rates of animals. And we use it to basically get third place in this competition to distinguish different types of cheese. So IEEE has a competition every year to basically make a machine olfaction device. And this is the contest. Those are the cubes of cheese that are being put into the device. And that's the judges eating the cheese after the contest is over. So it's a very sustainable competition. So everything is eaten afterward. And this is my collaborator, Scott Carver, in Australia. And my students, Miles, I'm for sure, who worked on wombats. I turn on a sound here, let's see. So that is about 50 chunks of rutabaga, a potato-like vegetable. And the elephant kind of, this is a force platform. They see the food, and they reach out. And they're also very good house cleaners. They make sure they don't leave anything in the wake. So they make sure to grab each part. This is atypical of what they'll do. But this is an example of them using the entire trunk as an implement. To do this, they have to know exactly how much frictional force to squeeze in order to keep it jammed together and acting like a granular solid. That's a line of mucus. They leave this sort of mucus. They have runny noses all the time. I don't understand why that is. But they always leave that behind. So to do that task, elephants do a lot of sensation. But right now, no one really knows how elephants sense their environments. The skin of their trunk is still thicker than the heel of your foot. But the trunk, people don't notice, it's covered in hairs. The trunk is about a meter and a half long. Most of the hairs are around 10 centimeters long. So those are the hairs on the base in the trunk. They're very thick and wiry. So they have about the same length as whiskers, cats and sea lions and these other animals that sense the world through deformation of these whiskers. So we think that they basically have a tactile function. The whiskers on their trunk are much higher density and they're much shorter. And we think those are really specialized in to be able to measure where these objects are in space and to perform the tasks that I'm going to show you. Let's turn off the sands right now. And pull this and unfold it. So these are some dissections we've done with the elephant of two regions, the same location. Simply one is dorsal, that means the top of the trunk, and one's ventral on the bottom. So you can see this one is basically has these, in order for the trunk to extend to reach these objects, it can extend 20% of its length. And to do that, it has this sort of telescoping skin that allows it to sort of reach without having high strain to the skin. The ventral surfaces has a lot more wrinkles that are similar to our wrinkles. I mean, not like these folds, but these wrinkles. And it's interesting that the dorsal and ventral surfaces are very different. This is one of the hairs that I just showed you. It's actuated. So what we're doing is it has about a centimeter where it can be actuated by muscles. And these hairs are actually embedded inside the folds. So as the trunk opens up, it can extend and sort of release more hairs so it can sort of sense this environment. We have not actually seen the elephant do this with the hairs yet. But it seems like they're actuated, so it is a possibility. So the first kind of way out of the elephants pick up objects, I want to show you this video here. Oops. So if you listen carefully, that's the sound you hear when you go to a Chinese noodle restaurant. Maybe not Italian noodle. I think Italians maybe think that's probably kind of gross. But that's the sound of the elephant generating what we calculated is around almost 70 miles per hour of airflow to pick up these objects and suck it into the trunk. They don't always do that kind of motion. For example, if the objects are fewer, then instead of providing large energy to provide suction, they'll basically just push them together and pick them up individually, as you would expect. So we can create this regime diagram for when they actually turn on the suction power. When the objects are many in number and small in size, the elephants have no that the pressure force they can generate can actually pick them up. And so they'll actually turn on the suction power for a few seconds. When the objects are large enough, they know that basically these things are too heavy to suck up and they basically just pick them up individually. Here's an example of how organized they are when they're picking up objects. They really try to go in vertical columns to sort of get as many objects they can into this tip of this trunk before they pick it up. So somehow they have this sort of coordinate system of basically going vertically and horizontally away from their body. This idea of using suction was originally observed by Darwin, except he and a few of these Japanese investigators later showed that they don't use just suction, but they can actually use blowing. So what this elephant is doing is it's here, it's trying to collect these leaves that are just slightly out of reach. So it blows against the ground in order to push these leaves into a pile. So the elephants have this idea of reflection, of using the fluid as sort of an arm to reflect and to grab these objects. So that's trying to grab this tiny, maybe 10-calorie piece of twig and it's almost going to fall on its face because it's on three legs. And it blows against the wall in order to do this. Remember, they've got to eat 100 grams every minute. So for them, that kind of behavior is kind of worthwhile. So this is actually a high-speed video of elephants grabbing this tortilla chip. As I said, the elephants are kind of near-sighted. So they never grab the food directly. This one always misses it, just hits the force platform slightly to the left. And then. When you see it, you decide to go home. It's not really well-known, but they, sorry? Just bullpup that. Like, what's their vision compared to human vision? They tend to feel around for things. I think they can see things far, far away. But for the food items, they're really using touch and olfaction. I mean, in terms of categorizing human vision, I don't think those tests have really been done. But that's a good idea to try to do. So let's play this again. So they can at least see where the force platform is. And you can see it's sort of feeling around looking for this small bump that's going to be this tortilla chip. This is not the first time it's eaten this, but it's one of the first. And it feels this bump and it moves it around. It can actually push it around until it gets to the right position. And what it's looking for is to provide most of the surface area on top of the item and then turn on suction so it can pick it up. So using suction is really the only way I can pick up this object, because this is a millimeter thick. And it really can't get the lip of its trunk underneath. And this is what I thought would happen if we gave them this task. This is what happened if we cover the top of this lid. And they break this chip into pieces. They actually get kind of angry, but they go back and make sure to get the remaining pieces. So this is something they want to avoid doing, breaking it. They really want to be gentle in order to pick it up. So when we did this experiment, the first thing I did was try to go home and see if I could pick up items with my nose, because I thought it's a really good way to save time. So I went home and tried to pick up a grape and a cherry, and I was unsuccessful. And then I went to get tortilla chips. And then I practiced a couple of times. And the thing is, the key is if you want to pick up something by suction, you have to be very close. And the reason for that is because when you provide this pressure force, you're bringing air velocity from your nostrils. But if you look at larger and larger hemispheres away from your nose, the flow rates through those larger and larger hemispheres decreases very rapidly. So you have to be very close in order to your airflow to provide the suction pressure to pick up these objects. So from here, I won't be able to do it. Oops. But if I get close enough, then I can actually pick it up. And I think that's what the elephant's doing. They're getting close enough they can actually apply their suction pressure. Now, the elephant has two things that we do not have. One, its lungs are 30 times bigger than ours. So we can apply that large velocity for three or four seconds, enough to go up around and find the object. And the other thing is that the elephant's nostrils are three times wider than ours. And that allows them to use the same pressure force to generate a much higher flow rate. So you can calculate, basically, using arguments of, you know the pressure the lungs are applying. You know the radius of the nostrils. And you look at, basically, a distance away. You can calculate the radius that you can be away from the tip of the trunk that you can actually generate a force to pick up a tortilla chip of a low amount of weight. And so now we have the task of actually measuring the elephant's pressure force, which hadn't been done before. And so we did this test to measure the pressure it takes to pick up water. So these experiments are done very far away. But this is actually a very large amount of water. It drinks about six liters in two seconds. So six liters is like, you know, they're about this big. And that's about the rate of 15 shower heads. So 15 showers running in reverse. And this little experiment on the side here, you can see. You can see that's. Let's turn this off for a second. This is basically this idea of I mentioned that if you're very far away from the tip of the trunk, you won't have a very high velocity. So these are chia seeds. Everything you give elephants has to be edible. So we basically, and you can see close to the trunk, the velocities are very high. And very far away, it's low. And that's why I could only pick up the food when my nose was within a centimeter. So the elephant trunk has about six liters of capacity. That's how much it can hold before the food is ejected into the mouth. And it can augment that, we think, by basically expanding its nostrils. So this is a picture of a cross-section, not of that same elephant. This is an elephant that died a long time ago. Well, you can see there's these radial muscles going outward. Longer than two no muscles that allow this thing to stretch and muscles that are basically allowed to twist as well. And we think the radial muscles are responsible for expanding these nostrils. From ultrasound, we think it can expand up to 12 or 16%, which can increase the volume that this thing can hold. So given the elephant's pressure of, we measured to be about 10 kilopascals, you can estimate that they're basically siphoning air at around 70 meters per second, or around 160 miles an hour. That's about half as high as the high-speed train in France and a bit faster than a human sneeze because it's sustained over a longer period of time. And I mentioned before that they can do it. They can pick up objects because their lungs are so big and their nostrils are much wider, which allows them a bigger, basically allows them to be farther away from the objects before they actually pick them up. So they don't always pick up tortilla chips. Sometimes they'll pick up small grains, cereals, and things like that. And there they have to be, again, they wanna do as few trips as possible. So what they do is use the ability of these materials to jam. So if you were to make a pizza or something with flour on the table and you had to pick up flour, you would do the same thing. You would squeeze the flour together in order to pick it up. And they're amazingly clean about it. I mean, that's a lot of flour that was, there's 40,000 grains of cereal dust. And they do something comparable when there's basically different numbers of particles. They'll jam them together. These are the force measurements, the normal forces on this plate. And from this, we learned that the minimum force they can really apply is about 10 newtons. So when they're touching the plate, that's sort of the most gentle they can be. And that's pretty much the breaking force of the tortilla chip. 10 newtons or one kilogram of weight is how much when those tortilla chips will break. We also noticed that basically when the food items were smaller, they applied much higher forces up to 30 newtons. And the reason for that is because of the way these materials behave when they're jamming. And I'll explain that in a second. But one thing we noticed is that the trunks will have different shapes when they're picking up different numbers of materials. For example, when the objects are very few, the trunk will be straight, as you kind of expect. But when there's 40,000 of these very small objects, what they'll do is they'll generate a kink and that's almost 10 centimeters tall. So if you measure the force on this platform, this provides about 20, the weight of this part here provides about 20% of this force. So what we think that they're doing is they're just relaxing part of the trunk so they can use self-weight, sort of part of the weight of the trunk to apply forces in order to jam these materials. And this idea of generating these kinks is not new. Octopuses, which were studied by Benny Hochner in Israel, also generate kinks. And this is a common way to basically pick up objects when you have many, many degrees of freedom. Here, when the object, for an octopus, they have tentacles along the entire arm because they can't tell where they're gonna touch it. They can basically stick to the object anywhere along here. That sends a neurological signal to the head and other signals sent forward. And right in the middle, you basically perform this elbow, this kink, and that allows the octopus to bring it bring it to its mouth. So perhaps something similar is happening with the elephants. Now the reason they have to generate a kink of a larger vertical distance for the smaller objects has to do with the fact that each of these small objects has a small probability of failing. So that idea is kind of demonstrated here. If you basically are trying to hold a series of granular materials, they each have a small probability of failing and you've got to squeeze owner to prevent them from failing. But as you increase the number of objects, the force goes up that you need to go up with basically the exponent of the number of these objects. So basically once you have five or eight, you really have to squeeze very, very hard in order to pick them up. So that's basically what the elephant's doing. It's squeezing in order to prevent them from picking up. And I think it has a sheer force sensor to detect when they're slipping. Because I don't think they can actually calculate the amount of force they need. They just figure out when it's slipping and then it provides just that amount of force. So these are some experiments of elephant's weightlifting. That is actually 20 kilograms. It doesn't look like it's very heavy. And we give them a reward every time they lift the weight. They actually have broken this weightlifting setup. We taught them to lift weights, but we never taught them to put them down gently. So this apparatus is broken. This is 95, about 40 kilograms, 40 kilogram weight. So one thing we observe from this is that they have a couple of principles to lift heavier weights. One is that as they lift heavier weights, their trunk gets more vertical. So the section that interacts with the beam, that's the heaviest weight here, becomes more vertical and so they can basically act as a better actuator. And that's partly done automatically because the trunk is, you can imagine, is an elastic cantilever and it basically bends more sharply with heavier weights. And so they basically make it more vertical as the weights get heavier. And the amount of curvature is, as you would expect for, people have actually tried to estimate the material properties of the elephant trunk. It's about one megapascal. And the amount of curvature that you see is consistent with those previous measurements. The other thing the elephant trunk does is that it increases its surface area of grip. So it can go from basically having a lip contact for very light weights to wrapping the entire trunk around to almost a cycle and a little bit more in order to increase the contact area. The top contact area really doesn't help it lift but I think maybe it helps the stability. This weight really doesn't go vertically up and down but they naturally kind of do this in order to pick up these heavier weights and reduce the stress that they have on the trunk. So even though the weights get heavier because they have higher contact, the contact stresses here don't really change very much. So now I wanna talk a little bit about how these animals sniff and how we built this device to win the sniffing competition. This device is named Gromit for this cartoon where Gromit and his friend go to the moon in order to see if it's made out of cheese. This is the device and it was built in about two or three weeks by a really great engineering grad student. So we wanted to make it basically give behavior that's comparable to a dog. So we wanted the airflow velocity to be around 1.5 meters per second which that's what speed is of a dog sniffing. We wanted the frequency device between zero and 10 hertz because basically the sensor that we use is actually not very strong. Our sensor could only measure around six hertz so we wanted to be in the range of the sensor. And we basically have a system of bellows that moves air in and out so that when airflow moves across the sensor it has this oscillatory pattern. And one thing we made sure is that there's very little dead space so that most of the airflow that goes across the sensor is also not just lost in some stationary part of the sensor. So the sensor for this competition was standardized. I think it's about $10. They're called mini oxide sensors. And they work by having basically a layer of tin oxide and what the cheese or other molecules does is basically just grabs oxygen molecules and that changed the voltage reading on the sensor. And this is what you can see on the sensor in real time if you breathe on it or something like that. Originally there's no reading and then the sensor can change in basically resistance or current. I think I'll report current in the later part of this talk. And then as you take the object away you basically have this signal-backed equilibrium. So currently most machine learning, machine oxide, machine olfaction devices they basically just use two signals. They use basically the amount that it decreases here and the amount that it increases. And as you can see because the process of diffusion is slow these take on the scale of 10 to 20 seconds. That's not very effective. If I wanted to figure out I wanna eat something I don't wanna wait 20 seconds to sort of see this. So that's where sniffing comes in because you can get information on a time scale not of diffusion but of basically of the order of convective motion. For our contest we basically grouped for example Gouda and Manchego into different categories based on the signals that we got from this one this one and the oscillatory signal here. And with machine learning you can group them and basically when you have a new cheese you can identify if it falls near close to the Gouda or if it calls close to the Manchego category. And that method seemed to work pretty well. So these are some schlin videos of a dog actually sniffing. Now one thing is sniffing is a behavior that's really evolved in animals. It happens at about eight times per second for a dog and they breathe about one time per second. So I mean if you try to sniff, I'll show you a plot if you try to sniff at the appropriate rate for your body mass you will probably faint or go unconscious because you're just basically not gonna get any oxygen to your brain. So it's really, you really have to sniff at very high frequency in order for this to happen. If you look at the scaling which we did across different mammals for elephants will sniff at about 4.5 liters. So this is based on the scaling for the dogs we extrapolated that to different mammals. Elephant needs about 4.5 liters per sniff. A dog needs about 30 milliliters and a rat will need something that's about the size of an eyedropper drop. So they'll need different volumes and accordingly they also have different frequencies. So we actually went to the Atlanta Zoo and measured the frequencies of these animals sniffing and compiled it with data that was on mice and through different breeds of dogs. And across the range of mammals they have basically five orders of magnitude in body mass. Their sniffing frequency decays between two times per second which is that of an elephant which is something we could manage to about 15 times per second which is that of a mouse. If we were to sniff we would fall about here we would need to sniff about 10 times per second or it would be consistent with these animals. So why do bigger animals sniff slower? The primary reason for that is the differing volumes of air. Essentially your lungs can produce a force that has a constant pressure. Most muscles in the body are composed of the same muscle fibers and so the force per unit area is the same. For the elephants it was 10 kilopascals for sniffing water, smelling water. It's the same for mosquitoes and the same for humans. So that lung force is applied to the cross-sectional area. It depends on where you put your control volume but if we consider our control volume to be in the trachea that's the radius of the trachea here. And the inertial force of the air goes as the mass of the air which is the length of the trachea times the radius squared, times its acceleration which is the length of trachea times the frequency squared. So for you to pull this larger volume of air in the body your frequency is gonna go as one over the body length. In other words the bigger you are the slower you're gonna sniff the fewer times per second. And again it's because you've got a larger volume but your radius to volume ratio doesn't go up favorably. So the black points here is the frequencies that these animals sniff at measured from experiments. That theoretical model that I presented a second ago that's given by this blue line where we extrapolate the parameters for the trachea radius and the lung pressure for these animals from data on these animals. You can see they kind of have comparable trends we don't really have the right exponent we have negative one third it's really negative one fifth, one sixth. And breathing is all the way down here. So breathing also has comparable trends too because when you breathe you also have to breathe larger volumes but you only have so much lung pressure. But it occurs eight times slower than sniffing. So did you, someone have a question? Okay. So why do animals sniff? There's no theory for why animals sniff but if you look in the cardiovascular literature there's been a lot of interest in oscillatory flows. This is in the early 1960s, 1950s. He defined as dimensionless group which is essentially a Reynolds number, a ratio of inertia to viscous forces when you have oscillation. So instead of a velocity what you have is the diameter of your tube times the frequency divided by your kinematic viscosity of your blood. And interestingly enough for blood the number for humans is between two and 14. In other words inertia affects dominate. And for animals if you look at their sniffing worms the number it's a comparable. It's between one, it's between three and 14. Maybe there's some reasons why it's a totally different fluid, a thousand times denser but they have comparable worms the numbers. So what does that physically mean if you're bringing fluid in and out in a oscillatory fashion? Well these are an experiment we've done. You can do this probably at home. You take a humidifier and you can bring it into a tube that's a square cross section and transparent walls and you can sort of see, you can use PIV to look at basically the motion of the air. And one of the things that you see when this experiment like this is that the motion of the air depends on where you are in the cross section. That's a student of this experiment. In particular if you're near the walls you will decay, you will delay the transition longer because you have higher viscous forces holding you in place. So the worms the flows have been calculated, these are closed form solutions for basically circular channels. And this is basically if you look at what's the velocity field. Basically a single cross section how fast is the fluid going as a function of position. And as I said if you're near the wall you'll be going slower than if you're in the center. And so I'm gonna show you this movie of basically a stepping through different parts of the cycle. But I want you to keep your eye on what happens near the wall because all smelling that's true for elephants. So elephants have more olfactory neurons olfactory genes than any other animal for elephants and dogs. All of us have the sensors near the wall. So the whole issue in the physical problem of sniffing is getting the molecules in the bulk fluid my nice cheesy tortilla molecules all the way to the wall. And for that I've got to give it enough time for the molecules to diffuse. Because the fluid is going basically axially. There's no convection motion near the wall. So basically that is given what I need is time. In particular if the molecules are diffusing with a constant diffusion constant D. And basically if I'm moving at a velocity U the distance I have a distance X away from the wall where the amount of time I have is given by the size of my sensor D divided by U. And all those molecules will have a chance of actually diffusing to my sensor. Molecules farther away from that are just not close enough to basically diffuse. So let's step through. Right now what this is is high wormsly number. So we're starting at high wormsly number where I'm sorry I always get this mess. This is low wormsly number where the walls actually have a large effect. Okay, that's shown by the velocity field which decays very quickly towards the wall. And this is basically the input and output inhale and exhale and you can see the velocity vectors are going back and forth. But close enough the wall the speed is slow enough that the molecules have enough time to diffuse to the sensor. Now this is basically increasing the wormsly number. So I'm increasing the inertia of the flow so that basically it looks like a plug. And this is actually disadvantageous for sniffing because you can see the velocity near the wall is getting higher. That means my molecules will spend even less time near the sensor and I basically gonna get even fewer and fewer collection. Now this is probably the highest wormsly numbers at all where basically it hardly sees the wall and it's basically just sort of going like a plug. And so that's what animals are experiencing all the way from mice. This would be an elephant. The highest worm is the number of about 14 where it really is getting very, very few of the molecules. So here are a couple of different velocity profiles that you can numerically calculate based on the closed form solution. And all I wanna say is that you can actually calculate the number of molecules that diffuse and hit the wall by estimating because the velocity fields for the closed form solution are known. You can figure out what distance X away from your sensor that you have a velocity of sufficient low speed that you actually get diffusion of the wall and you integrate across the entire cycle. Not just for a single part of the cycle but for all the velocity fields across the cycle. And so you can estimate the number of molecules that could land. And that's given by this curve, the black curve here. And the data here is actually from our sniffing device. So I'll walk you through how we get this data. This is the, that mini oxide sensor and before I put the cheese in the experiment, originally you have no change in current. As soon as the cheese is introduced, I start my sniffing and you can see there's some information getting to the sensor. And so what I'm interested in is not basically the ascent or the descent because those occur at too slow a time scale. What I care about if I'm an animal is how high these amplitudes are. Because that's basically what I'm gonna tell if this is something I wanna eat or not. And that data's plotted on this side. So basically if you sniff at very, very high frequency, issue is the amplitude basically is getting is not very high. As you saw from the high frequency data, the high Wormsley number data, the speed near the walls is very high. So there's very little time for the molecules to diffuse to the sensor. And I'm basically getting a very little signal. And there's a limit. If you sniff too fast, you actually won't be able to get signal at all. That's sort of the noise limit of your sensor. So animals really wanna be as high on this, I think as possible because what I think animals wanna do is they wanna get information as quickly as possible. If you sniff too slowly on the other hand, then you're basically getting into this regime where you're just waiting for diffusion to happen and you're not using the capabilities of your sniffer. So this is a little kind of open debate because I'm not sure what triggers an animal to decide if something is food or not. Because as I mentioned, if you sniff very quickly, you get very low signal per sniff. But what we did is we calculated the number of particles that land over a set duration of time. Even though you're sniffing more quickly, you basically have more cycles. So those higher frequencies actually, you get more, the integral of all those particles that land is higher. So they actually get more particles over a set distance. But basically they have less, but each amplitude is lower per cycle. So it's not clear if animals are detecting it based on basically a single sniff or maybe they need multiple sniffs in order to get the total sum of particles attracting. So Ring Cardan and a few other speakers are here tomorrow talking about olfaction. So I thought I'd talk, and it has nothing to do with elephants, but we've applied some similar methods to look at how moth antenna are designed. And I think we'll probably see some of this tomorrow as well. But this is actually what, so these moths that fly around at night, they're purported to find female moths, the males at least, at 12 kilometers distance away. And they do so through a tentacle that's inherently hierarchical. So it has this initial base stem. Which you can see visibly. And then they have these things called branches that leak out. And on those branches, they have individual hairs. So that's the physical picture I want you to take away before I show you these images. This is taken with a con focal microscope where we basically added slices of this moth antenna together. The olfactory sort of sensors are on, I think they're on the individual hairs. So they're sort of all over here. And what we did was we measured the geometry of 50 different types of moths. And what we focused on was the angle of this branch with respect to the stalk. And there's a lot of parameters you could look at. For example, this is that moth antenna originally. And if you calculate the surface area of all these hairs, what you get is the old life, you add the length of all these hair, you get the antennas effectively like a foot long. 12 centimeters long. And so basically having this greater surface area really helps add to the amount of particles this thing potentially capture. It's hierarchical also in the sense that the diameter scales. So basically, as you go up in hierarchy, the width of that stalk is 10 times wider than the width of the branch, which is 10 times wider than the width of each of these hairs. And that seems to be true across different moths, similarly for the length and similarly for the spacing. So there's some rules that are guiding sort of why these are scalings are the way they are. I won't really talk about that today. I already showed that. So we did this experiment where we tried to mimic the Reynolds number of fluids striking the moth antenna. This is actually a small dental piece of dental material where we built a wind tunnel and we basically hit it with fluid. And this was based on this measurement on moths that we found was surprising that most moths when they stick their antenna into the flow, they don't stick it basically 90 degrees. And like if I want to capture as many air molecules impossible, you would think you would be like this, like a scarecrow. So you expose the most area and so most of this thing would hit it. But the moths tend to have their angles at 45 degrees with respect to the flow. And so we did this experiment where we had this humidifier sending fluid, these small drops and these are the, with ultraviolet light, you can see what has been adhered to the antenna after about a minute. And you can see as you expect, if I put my antenna into the flow, I only get stuff adhering to the very tip. All the juicy olfactory molecules, the cheese, the tortilla are just flying past and nothing's adhering. And at 90 degrees, I get adhering, but the surprising thing is that at 45 degrees, you get the most particles that land. And that's surprising because it has a lower surface area of contact, so there must be some fluid dynamite effect that's causing more of these particles to land. This, these particles are about 10 microns, so they're orders and orders of magnitude bigger than those of the affection. But we think maybe some, this might have an effect of why moths are doing this. You can do the same thing with smoke. If you blow smoke past, you know, a wooden dowel. And 90 degrees, you basically get this kind of effect where the smoke will expand due to the fact that it's slowing down when it hits this thing and has to go around it. And then in the back, you have a wider wake. But for any oblique angles, you get this effect we call the lingering effect where smoke basically has to, will travel down the rod for some length before it passes. And that's really enhanced. This is about 45, this is, I think, 45 degrees. You get a very long, long swath. So you can imagine if you have an olfactory plume and it hits basically these rods that are 45 or at oblique angles, you get some benefits. Here's a schematic of that same process. And physically what that is doing, if you have more time near your antenna or rod, it's similar to what I showed with the Wormsley number on the elephants. And that basically if you have to delay your, you know, the change of trajectory, so that you're spending more time around this rod than you would normally. More time on the rod means the envelope that this rod is absorbing from the wind is increasing so that you have more potential particles that can land by diffusion. That basic idea, so unfortunately when we, we have this model that we consider the diffusive effects of particles landing on this thing. And for particles of the size of moths, we basically don't get any, we don't get any effect. But for particles of the size of the experiments of time microns, we do get, you know, extra benefits where we basically measure numerically the amount of distance that they travel along the rod and the amount of extra time they're gonna be spending. All right. Now it's for the fun part of the talk. Looking to decrease this volume here. So yeah, every couple of years I try to do a project that's just for fun. This one I was giving a, we were giving a talk on defecation on this universal law of, so I won this thing called the Ignobel Prize for Research Makes People Laugh and Think. And we're giving this talk on defecation and someone asked me at the end of the talk as if my theory could account for cubic feces. And so I never heard about these animals before but these live in Tasmania. They are marsupials, they have pouches. Their pouches actually point backwards so that when they dig they don't get dirt into their, you know, into their babies. But unfortunately that also means they poop directly on their, on their, basically. Juveniles, heads, but that's the way evolution works. Just good enough. So this is actually the actual feces. We got shipped from Australia. We play games of chance in my lab. You can see the feces does have, you know, eight sides. I wouldn't bet a million dollars using this pair of dice that we made with Wombat feces, but it is surprising like this is a 3D scan of the feces. It really does have flat edges. So if you look in Australian folklore, they have a lot of theories for why Wombats have cubic feces because people walk around and they'll find cubic feces. And so basically they rely on these three ideas of how cubes are built in, you know, our made world. So if you have a pair of dice that was probably made by injection molding where you send in a hot liquid which is in a liquid state and it forms the shape of its mold because liquids fill the shape of their container and then it solidifies once it cools down. We don't think that's the case from the dissections I'm gonna show you in a second. The other option is, this is great because we're in Italy, it's like pasta making or extrusion. If you wanna make a pasta, you send soft dough through, you know, very hard dyes. And so the Australian folklore said that Wombats have square anuses so that it would come out like a square and we quickly showed that that was not the case. They're circular just like ours. And the last method is basically the process of like basically just drilling or basically removing parts of it so that you get a cube. And neither of these work. So the Wombats seems to be like a third method. And cubes indeed are pretty rare. So this is, for years people have used different animal poop sizes and shapes to identify different animals. In general animals will defecate about one one hundredth their body mass. For me that's like 1.5 pounds per day. And that's also true for the Wombat. But what's special about the Wombat is the dryness of the feces. So biologists often ask me why in the world would cubic feces evolve? My current theory is this, is that the Wombats generally do not like each other. They spend time mostly underground and they only count each other through territorial markings which leave on the outskirts of their territory. And they like to make these markings laid on the tallest point that they can climb with their cubic squat bodies. Which for them is a rock or a stump or a log. So every time you find their poop it will be basically on these, this is a tall point for a Wombat on these outcroppings here. And I think if over evolutionary time as they were doing this more and more if the cube, if the feces grew more and more square they wouldn't roll down the rocks and it would act like it's a better marker. Because they are the only animals, well kangaroos would be a close second that have edges on their poop. And you can see it doesn't do a pretty good job of staying on top of that rock. Not so much for these, but that's the theory that I have right now. So this is the actual intestine that was shipped to us from Australia. It cost about $1,000 to ship this thing to the United States. It's liquid in the stomach and the end it becomes solid. In the beginning it's really shapeless. It has no length width or height. You can see it's very wide. But as it gets to end, in the end it gets more and more similar length width and height properties. This is a pretty gross image right before a coffee hour. But I thought this is pretty telling. So we didn't actually harm any wombat in the study. They're naturally hit by cars and we have a collaborator who, he studies the wombat genetics so he needs to squeeze a fresh wombat heart into a jar to collect their blood to get the genes. But before he does that, he cuts it open and he sends us the intestines. So that's the actual poor wombat and that's the intestines. It eats grass and the wombat is very drought tolerant. So you can see the feces, I mean it's somewhat dry, but it gets really, really dry as it gets to the end. So this intestines is about five meters long. That long length helps accommodate basically the amount of dryness it's gonna get. So our feces through information is about 70% water. A wombat's feces is about 30%. So it's about twice as dry and that's what's normal for them. And you can see originally, this is probably similar to what you see for a horse or a donkey, but then after it gets 70% level of dryness, here it's getting to 30% level of dryness, you can see the characteristic edges and square cross sections. That's really just what came out, that's as fresh as it gets. And that's associated also with the change in color. So I was at a geology conference trying to understand why, how they do this. Oh, Nick Raffich was there. We were at, I was talking to these geologists and then they were telling me about this place. There's places like this all over the world. One of them is Giant Cosway Ireland, where you have natural geometric structures forming in rocks due to the particular cooling conditions of that rock. This is, here you get hexagons and this is a cross section of those rocks as you go all the way down, five meters down you get continual, these shapes forming naturally in these conditions. And so this was studied for a long time by geologists and recently there was a paper by Mahadevan in PNAS about repeating these same structures, not in rock but in cornstarch. So this is actually cornstarch with, this is a pretty, it's a nice experiment where they have water, bath, cornstarch and they have a heat lamp that heats the cornstarch at different rates. And what they find is that the geologists have found that there's a certain pecle number, a dimensionless group that relates the convection force, the force of heat leaving the rock to the diffusion of heat across the rock. And these structures are formed only when the pecle number is a small but constant quantity of, at least Mahadevan found it was 0.15 for cornstarch, 0.3 for lava, at least for the Great Cosway in Ireland. And basically the pecle number is the ratio of the velocity of the fluid leaving here, I'm gonna consider, this is a model of that wombat intestine, the velocity of the water leaving the inside of the feces, going to the outside, times the length between feces and divided by the diffusion of water that goes laterally. So if you actually measure the wombat intestines five hundred centimeters long, they keep their feces for a hundred hours which is twice as long as us, is about five days. And so the velocity of the scats are about five to 10 centimeters per hour. So you can estimate how quickly they lose, go from a hundred percent saturation to the third percent of the end. And you can get a water velocity about a millimeter per day. That's basically the speed of water leaving the feces. And if you use the pecle number for these different cornstarch and lava geometric structures, you find a crack length and you use the same diffusion constant, you get a crack length of about two centimeters which is half of what's observed in nature. So what I propose is that the wombats because of their particular drying rates and the width of their intestine, they can control that the spacing, the length of their poop pieces are comparable to their intestinal width which are two independent quantities. So we still do not have a good explanation for forms the corners. That's what sets the basic size. Oh yeah. There's actually instances of unit W. Oh, this is not to heating. This is just to where you would put the cracks. The number of cracks you would get is L over W. Oh, okay. And it's going to give you a Q. So. That's what you propose. But that argument is not based on heat. It's basically the number of cracks you would apply to just. So in this. Yeah, and they seem to be somewhat, you know, it's pretty quite consistent. You know, you don't have any long ones. And in fact, I mean, the intestines are not this long inside the body. There's not enough space, but there are segments that are maybe five widths long, which isn't maybe enough to do like a finite beam. I'll have to, we'll have to talk about feces during the coffee break. So we, that gives them, there looks like there are several reasons why they would be of constant length. We still do not know why the corners form. So this is a measurement of the volume in each of these corners. And you can see the more volume is in these corners, the sort of rounder it is. And they really get quite, you know, square, really very high curvature at the very end. So my graduate student is actually a clown balloon artist. He likes to make clown balloons for fun. And so we decided to use that technique to measure the basically, well, I had this hunch that there was some elastic anisotropy or elastic inhomogeneity. And that's because when we hung the intestines vertically, we saw that all the corners of the poops were aligned. That means that inside the intestine, inside the walmart's body, some kind of coordinate system that tells you where this corner is gonna form and it's not gonna be arbitrary. And in fact, when we blew up a balloon inside the intestine, these are surgical markers. You can find that basically the stiffness changes. For a pig, it's more or less uniform, maybe plus or minus 10 or 20%. For a walmart, there's basically a factor of four in the elastic modulus as you go around. I was disappointed to find there were only basically two regions, so that which means that we are not gonna get four corners out of this thing. But we went with that. So we basically try to do these experiments with pantyhose and basically increasing the stiffness of certain regions of pantyhose and squishing it, which is to emulate the effect of drying. We get structures, this is not cubic by any means, but you can see if you have a higher stiffness area that will cause it to be flatter in that zone. As opposed to if you have no stiffness changes at all, as you expect, you get a uniform cross section. This is a crude numerical model where we have particles that are pushing out with a constant radial force and we have this, there's no bending in this membrane but basically resistance of stretching. And at least in the beginning, you can see if you have uniform stiffness, you get a circular poop, but if you don't have uniform stiffness, you can get different kinds of shapes. So with that, thanks for being such a great audience. Some of the work here and my previous work so it was published in this book. So I wrote this book called How to Walk on Water and Climate Walls by Princeton University Press and I went on this international book tour last year. The book is on Amazon, on Audible, read by Seven Hours by an Italian-American actor. It's been translated to Korean, Japanese and Chinese. Today's ICTP, I'll be at Fermilab, then at the Gnobals and Dartmouth, American College of Rheumatology, Caltech, Waterloo, North Carolina Academy of Sciences, Lenoir, Rhine and then back in Toulouse for the next year to talk about my book and about other things. But thanks for being such a great audience. Happy to take any questions. Yeah, the stomach is mostly, but basically as it goes from a liquid state, what's optimal is less fear and maybe there could be something different. Yeah, there was a paper on square drops. They put arrays of hydrophobic and hydrophilic surfaces and then showed that, I don't think they changed the viscosity, but they showed they could make drops, force drops to be square. The poop is actually liquid but it's still composed of very small grass fibers. And I think maybe also the drying process helps arrange these fibers as well. But yeah, this idea of surface area, yeah, cubes doesn't seem like it's good for minimizing surface area. But yeah, that's good. Maybe it would be interesting to try to measure the surface area of the different materials, yeah. Yeah. I have a different posture depending on the weight of the item they have to lift. Yeah. I would like to know whether the elephant assumes his posture already the first time watching the item or if they have to assess, test the weight and then they learn how to assume this festive vertical posture. So they, at the first glance, they decide how to lift the item or just they try the first time. I think the first time we gave them the item, that's definitely the case that they spend a lot, they basically do different attempts. But let me see if I can get to that video. Oh, here it is. Yeah, that's a good question. We didn't expect weights kind of in order, of increasing order. So it is possible maybe they kind of learned to do that. I mean, this is 45 pounds and this is 20 kilograms, which is just the bar itself. So it's possible maybe they see their weight in kind of in preparation. So in nature, they use this behavior to uproot trees. So they'll pick up a tree and eat it. They'll literally poke it up out of the ground. I think they can maybe based on the side of the tree they can get an initial estimate guess of basically how many wraps they're gonna need to pick it up. The object are probably processed by at least in primates, in humans also. And there are also neurons involved in this task in the sense that there are neurons responding. In the motor cortex, there are neurons already responding to the visual property of the object. Oh, and tell us basically how to prepare to lift this thing. Yeah, and so it's possible that like it happens also in monkey that they imagine the weight of the item they have to lift at the first sight that may adjust later depending on the experience. You can try to cheat them. Yeah, I think we had covered up cover the weights maybe. No, to list it all. Oh yeah, styrofoam. Yeah. Yeah, yeah. And then it's a... Oh, and see. Oh yeah, they might break the, they might just lift the whole thing. But they've probably done that trick with humans. And we do that when we lift a box and see if we can get them off balance. I think that would be interesting, yeah. Yeah, I think there's also, they're just pushing against gravity too. Like for uplifting trees, there's different kind of like resistance and things like that. Like somehow they can accommodate and adapt to these things. But I think what they want to avoid is sheer stress. They want to avoid basically scratching their trunk. So they really want to maintain a tight grip. And I don't think I've showed it here, but we measure the acceleration. They tend to pull with constant acceleration. So I think maybe in the beginning, they see if it's not working then they can sort of change their grip. But they basically pull with constant force, it seems. Yeah. I think you were first, Jerome. The antenna of North during flight would be 45 degrees. I'm just saying, if it's not working, it goes more. So we actually study the branch. So I kind of use that as just an example. In reality, the antenna is there actually, there's a, I don't think people actually measured it. What we did measure was the angle of the branch to the stalk. And so when fluid, it was just, when I demonstrated it, I just liked to show the entire thing. But we imagine fluids hitting the stalk and then basically run across it. And basically what matters is the angle of this branch relative to the stalk. And that seems to be, when we look at these moths of different datas, this is peaked at 45 degrees, that angle. The moths, they actually, and other animals, they tend to move there and they can actually move, they have a mobile basis so they can move these things around. And it's not clear. I would love them to always keep at 45 degree to verify my model, but that's not, I don't think that's the case. There are probably other things going on, different kinds of turbulence and things like that. But at least the fixed, they cannot change the branch angle. And for some reason, there's a predominance of them liking 45 degrees for at least the branch. Okay. It's flow from, as you said you said, from particle capture, small particles, or the inertia problems related to those, to the molecules. That's a tricky item, I say by any means. Yeah, even, so these particles are 10 microns. They basically will strike by impaction. There's almost no diffusion associated with their motion. Smoke is still too big compared to real chemicals. And so our model predicts basically, you know, that if moths were trying to pick up particles, they should be 45. But the model also predicts if moths are trying to pick up objects that diffuse, that it's not nowhere close. So we basically, what we really need to do is probably do an experiment where we do electrophysiology on a moth and see if we can move the things around. But that's beyond beyond, that's probably with collaborators. Yeah. Oh, we'll get this one first. Oh yeah, the baby elephants. So when they're born, it takes them about a year or two to learn to pick up anything. I mean, it's really, similar to a human baby is born, can't talk, and you can sort of make sounds. You can't do deliberate things. Certainly you can't do, I mean, picking up small tortilla chips and picking up 10 objects at once, that requires like a lot of practice. So they, I think, they drink milk for the first year or two. They're really nursing. And as they learn to practice, they can sort of pick up more things. The trunks also in the very beginning are totally covered in hair. They're very, very fuzzy. So I think that helps them, you know, figure out where they're touching things. Eventually most of those hair fall off and only a few remain. And in terms of suction, the suction is predominantly because the elephant's mouth is like three meters off the ground and they can't get to the water. So it basically has to use the trunk to pull it to the water. And the baby's because they're drinking milk. I think they can sort of drink the milk or they can just go on. I think when a baby elephant goes to drink, what we see is that it is method of put their whole head in the water. So I think that works if you're small but it becomes undignified if you're like an adult elephant so they don't do that anymore. But that's how they get up around it because I think it takes coordination. They've got two sphincters in the trunk because you're bringing water and so I did not try this at home but if you try to bring milk into your nose so you can drink it into your mouth, you have to be very careful that it doesn't get into your lungs because I can cause pneumonia. So they have two sphincters and they measure very carefully when they fill their trunk with about six liters of fluid. And they basically have a little stopping point in between so I do want to show that. I didn't mention that so I do want to mention that. But they have a stopping point in between so I think they're kind of like as a little safety factor here. Basically when it gets to halfway full, they stop. It's possible, you know, this is coming very fast. It's three liters per second so and that, you know, in that one second it's got three liters of volume in the trunk. It stops and then it sucks again. At the same speed, they don't really have a slow suction but I think the coordination to prevent you from drinking the water and getting into your lungs all takes some out practice. And I think they probably, I'm guessing a baby elephant will put little bits at a time before they fill the entire thing. I think you had a question. Yeah, well you just answered it. I was wondering about what stops things from getting into their lungs, particularly like flour or something that we wouldn't want to hear in your question. Yeah, well they do make mistakes. Like those cubes we saw when they suck, they're such in a rush to eat those cubes. Sometimes they will make a mistake and suck a cube all the way into the middle of their trunk and you'll see them basically stop and they'll turn their trunk into a knot and kind of try to push it down and they'll try to sneeze it down. They also make errors in judgment. So they eat only vegetables but when it rains frogs will come out and as they're picking up food to eat they'll pick up a frog and throw it in their mouth and they just spit it out and disgust. So I think there are times they can't actually judge what it is. I would say almost they're working blind when they're kind of picking up objects. Just one more time. So that being again, maybe the 45 degree angle works very well for them and I think that's probably a problem how they're holding but they're not really branched elaborately as in your example. Oh yeah, we had to work really hard to find the 50 moths with branched in her. Yeah, most of them get along well with having the non-plumos, the males the non-plumos. Do those still mate? They still mate. I hope so, yeah. But maybe they're not working as... Oh really? It's a little enigmatic, I believe. Oh, yeah, maybe it's somewhat of a display or something like that, that's pretty interesting. The number of Censilla, they're very, very small moths and they may only have 1,000 Censilla on the antenna or some of these other ones here, you're talking tens of thousands, huge range to think about. Oh, the difference in numbers of Censilla between the smooth and the plumos. I see. Coffee break, brownie time, thank you.