How a 19th Century Physics Experiment Trapped Light and Led to the Internet

- In 1887, a young physics teacher attached a thin glass rod to the end of an arrow and loaded it into his homemade crossbow. He torched the rod's end until a molten glass bead. Roughly the size of a pin's head, it was barely hanging on. And then he fired the arrow. (air whooshing) (suspenseful music) Bright sunlight streaming into the room glinted off a tiny glass fiber. At its thinnest, it was less than a tenth of the diameter of a human hair.

Now legally we cannot shoot a crossbow indoors. Luckily I am pretty quick. (Henry chuckles) (suspenseful music) That is cool. Look at that. Look how thin we got it. The physics teacher was using these fibers to measure very small forces, but in the decades that followed, other researchers found a different use for them: sending light signals. (quiet dramatic music) Glass fibers can carry light even through bends. Look at it. It's like that's 90 degrees. And during the 1960s, the telecoms industry was desperate for a better way to carry more information. As one American engineer said,

"It should be possible to transmit "all of the conversations going on "anywhere in the world simultaneously "over a single thread of light." But these glass fibers were terrible. The light faded so much that you couldn't decode a signal after just one kilometer. No one thought they'd be practical. Except for one man. (dramatic theatrical music) His own boss called his dream "a wild goose chase," but if he was right, if he could prove that fiber optics were in fact viable, and if someone could then figure out how to make them

at scale, well, eventually they could connect continents. How much is in this warehouse? - Oh, there's millions of kilometers here that we're pushing through here every month. - It's approximately three times as hot as the surface of the sun. - That's crazy. And one day, it could build the internet. (dramatic music fading) All that was unimaginable in the 1960s, but even then telecom companies wanted a better communication system, just for phone calls.

Between 1960 and 1973, the number of daily calls doubled in the U.S. from 285 million to 576 million. In that same time period, NASA sent people to the moon for a total cost of $26 billion. While AT&T had to spend nearly triple that amount just to install new telephone infrastructure. They just couldn't keep up with the demand. At the time, phone calls were mostly transmitted over copper wires. Now imagine I'm trying to phone my buddy Gregor. Now, when I pick up, a constant current immediately flows through the wire to supply power to the phone. When I speak, my voice vibrates a tiny diaphragm in the mouthpiece that modulates the current.

The louder I am, the bigger the current swing, and my voice's pitch changes how fast the current oscillates. At Gregor's end, that current reaches a speaker that turns it back into sound. Now, this works for one call on one line, but phone companies aren't going to install wires between every single house. One cable needs to carry multiple calls. So phone companies figure out a neat little trick. Even though humans can hear frequencies up to 20 kilohertz, you really only need to hear the ones between 300 and 3400 hertz to understand what someone's saying. So phone companies drop frequencies above four kilohertz or anything outside the voice band.

But if everyone used that zero to four kilohertz band on the same line, well, then all the conversations would be on top of each other and you wouldn't be able to hear the person that you're talking to. Luckily, we can shift those electrical signals, so my conversation can use zero to four kilohertz while another person's can use four to eight and so on all the way down the same copper wire. At the other end, a filter separates out the different frequency bands, similar to how a radio can tune into one station. Then the signal is shifted back down to the zero to four kilohertz range before it hits the speaker, so it sounds like a normal voice. But as phone companies started

using higher and higher frequencies, they started running into problems. That oscillating current carrying the signal generates additional fields, a magnetic field shown here in red, and little loops of current called eddy currents, which are shown here in blue. Now, because of their shape, those loops resist changes to the current at the center of the wire, and they reinforce changes along the surface, but eddy currents get stronger the faster the signal changes. And if you want more calls, you need more bands, which use higher and higher frequencies. So soon the eddy currents are strong enough to cancel out the oscillation in the center of the wire. But if the signal can only travel along the wire's surface,

it's limited to a smaller cross section. So resistance goes up and higher frequency signals can't travel very far before they're absorbed. For a while, telephone companies were able to work around the high loss by boosting signals every few kilometers. But past 65 megahertz, well, the high frequencies died off in less than two kilometers, so they were too impractical to use. And this trade-off between signal power and data rate was actually discovered back in 1948. The Shannon-Hartley theorem described the maximum capacity of a communication channel based on the signal power, noise, and bandwidth, or the difference between the highest and lowest frequencies.

The best way to increase capacity is to increase the bandwidth. But they'd already reached the limit for copper. The telecoms industry didn't have another alternative, (dramatic music) but then they saw the light. Visible light has frequencies in the hundreds of terahertz. As one researcher for Bell Labs put it, "the possibility of extremely high frequencies "and accordingly vastly wider bands of frequencies "appeared as a fabulous El Dorado, "always beckoning us onward." But if you're going to use light, there are two problems. First, there wasn't a strong enough light source to send a signal.

A light bulb couldn't produce enough power. There just weren't any options. - And along in 1960, Ted Maiman invented the laser. - [Reporter] A science fiction weapon that melts steel, a death ray that could be more deadly than a nuclear bomb. It's a beam of light. This is called a laser. - [Henry] There's a running joke that the laser was "a solution looking for a problem," but that's not really true.

- [Donald] It rejuvenated the interest in possible optical communications. - As one Bell Labs engineer said, "Think of all this bandwidth!" But they couldn't figure out the second problem. Glass is not as clear as it looks. Only a small amount of light actually makes it through. In fact, only a year after the laser's invention, a Bell Labs scientist started looking into fiber optics. His boss told him to check its loss, or how much a light signal weakens as it travels through the glass. Now, signal loss is measured with decibels, similar to how you measure audio volume.

Both are on a logarithmic scale, but while volume uses sound pressure, loss uses signal power. His boss said it would only be worthwhile if the loss was less than 20 decibels a kilometer, so it was comparable to copper. But when he checked the loss for the best fibers, a thousand decibels a kilometer. To understand just how bad that was, imagine a leaky truck carrying sand over one kilometer. If the truck starts with 100 kilograms, and it ends up with one, well, that represents a loss of 20 decibels per kilometer. That's only 1% of the signal left. It's what was happening with copper, bad but still manageable.

A loss of a thousand decibels a kilometer is like starting with the mass of the observable universe and ending up with less than the mass of a single atom. To be practical, these fibers would need to improve by a factor of 10 the 98th. It sounded impossible. It would have to be one of the most absurd engineering improvements in history. So Bell Labs gave up. And fiber optics might have died right there if it wasn't for one man. Charles Kao. In 1960, he started working for an English telecoms lab that was looking for a way to communicate with higher frequencies, but not with fiber optics.

Kao's boss said that it was "a wild goose chase" and the rest of the lab agreed. But Kao kept coming back to it, asking, "Why is glass so lossy "and can we remove those factors "or are they intrinsic to the material?" If it was the latter, then fiber would be dead in the water. But if not, then maybe it had a chance. So the question was: Why was the loss so high? Well, you might think it's because light is exiting through the sides of this fiber, but glass is really good at trapping light, even through bends.

Here's how. I've got a bucket of propylene glycol here and a green laser. When I remove the tape, (gentle dramatic music) the laser bounces one, two, three times. And there are two things working together here that actually trap the light. See, light travels slower in some materials than others, and it travels as a wave shown here as a line of crests with the beam always pointing perpendicular to that line. When the wave meets a boundary at an angle, one end of the crest crosses into the new material first, and it changes speed before the rest. If light travels slower in that new material, then that end drags behind and the beam bends or refracts away from the boundary.

How much it bends depends on the angle it hits the boundary and on how much each material slows light down, which is described by their index of refraction. This relationship is defined by Snell's Law. Now if light is moving from a material with a higher index of refraction into one with a lower index, like propylene glycol to air, well then one part of the wave crest speeds up and the beam bends toward the boundary. But as you increase the angle of the light, the refracted beam gets closer and closer to the boundary until it's running right along it. That's the critical angle.

A little bit more and all the light reflects back into the glycol. That's total internal reflection, and you get the exact critical angle by setting the angle of refraction to 90 degrees and then rearranging Snell's Law. - The key of it is the bigger the difference in index of refraction, the more powerful this effect is. If I have a really subtle difference between these two, then the amount of bending that takes place will be smaller. - You can do the same demo with water, but propylene glycol works a bit better because it has a higher refractive index, 1.43 compared with 1.33.

(liquid splashing) See, once the liquid is low enough for the angle of the stream to be much sharper, the light breaks through the water stream, but it stays trapped in the propylene glycol. And if you use something with an even higher refractive index, like the glass types Kao was investigating, well, then the critical angle decreases even more, which traps the light signals through even sharper turns. Okay, but the light has to move from a high refractive index to a low one. In air that works right up until something touches the glass. If you disturb that boundary, well, light can leak out the side. So about a decade before Kao,

researchers realized that a fiber needed to have two types of glass. The middle, the core, has the higher refractive index, and around it goes a second glass, the cladding, with a lower refractive index. Together, they keep the light trapped. Now the fiber could be touched, coated in plastic, buried in a cable, bent around things, and the light would still travel down the core and come out the other end. (quiet dramatic music) So Kao knew that if the high loss wasn't from light leaking out the sides, well, it had to be from something else.

He spoke with material scientists and glass experts, but no one seemed to know what was really causing it. He kept looking and eventually he landed on two potential sources: scattering and absorption. If you shine light through a perfect crystal that's transparent, there's going to be a little bit of reflection, but almost all of the electromagnetic energy will keep going down in the same direction that it started. But glass isn't a crystalline solid. The basic building block is silica or silicon dioxide, one silicon atom connected to four oxygen atoms in a tetrahedra or pyramid shape. In the glass, the pyramids are connected at the corners, so every oxygen is shared with another silicon.

That's why it's called silicon dioxide. For every two oxygens, there's one silicon atom. This is an amorphous solid. Unlike a crystal, its atoms have no long-range order. Instead, its disordered liquid-like structure locks in as it cools, with the exact arrangement depending on the glass type and cooling speed. For example, if you quench the glass really fast, the structure locks at a higher temperature. The final solid has lots of dense regions, which re-radiate more light than the sparse ones. The light scatters, which weakens the signal.

This is called Rayleigh scattering, and you can't get rid of it completely, but if you cool the glass slowly, those pyramids have more time to settle and pack together evenly. The structure locks in at a lower temperature and light scatters less. But Kao needed to know the minimum scattering. Was it 10 decibels a kilometer, a hundred, a thousand? He needed something concrete. Finally, he stumbled on a paper by Robert Maurer, a scientist from Corning Glass who measured the Rayleigh scattering across several glass types and cooling rates.

Kao used his calculations to estimate that if the glass's structure was locked at a thousand degrees Celsius, well, its scattering loss would be around one decibel a kilometer. He was surprised. He expected scattering loss to increase with higher temperatures. That's what causes more density fluctuations, but a thousand degrees Celsius was pretty hot compared to other glass types, so then scattering couldn't be behind those crazy high losses. Absorption had to be the problem, but Kao didn't know what exactly was absorbing so much light.

Was it the glass itself or something else? Well, absorption works like this. In a single atom, electrons sit at discrete energy levels, and there are usually some empty spaces in the upper ones, but pack atoms together into a solid and those levels smear up into wide bands separated by gaps. Now when light hits the surface, it can excite an electron to a higher band, but only if the photon has at least as much energy as this band gap. If it does, the photon is absorbed and eventually it releases that energy as heat. In something like silicon, that band gap is very small. It's only around 1.12 electron volts.

Visible light is between 1.63 to 3.26. It's got plenty of energy, so it gets absorbed. That's why pure silicon looks opaque to us, but the silicon in glass is connected to oxygen atoms that pull more on the shared electrons. Now it takes a lot more energy to excite an electron. The band gap is much larger, around nine electron volts. Visible light doesn't have enough energy, so it passes right through, and that's why glass looked clear to us. But most glass types aren't pure silica. The raw materials for glass, from sand, soda, lime, and other ingredients, well, they contain impurities like trace metals, and these metals absorbed far more light than silica did.

One of the worst impurities in glass is iron. Inside the glass structure, iron atoms form positive ions that strongly absorb violet, bluish, and infrared wavelengths. Of the visible light spectrum, impure glass doesn't absorb green light as much, and you can even see this in a regular glass pane. Looking at it from the edge, light travels down the entire width of the glass, so by the time it reaches your eye, only green light is left. So if you can see this attenuation after just a few centimeters, imagine a glass fiber trying to carry light for kilometers and kilometers. No wonder it had a crazy loss. Kao spoke with more experts who estimated that without those metal ions, the loss would drop to around 20 decibels a kilometer, comparable to copper,

just by removing impurities. (quiet dramatic music) It was all theoretical at this point, but it was enough for Kao. In 1966, his team published a paper on their theories, concluding with "the crucial material problem appears "to be one which is difficult but not impossible." But nobody really paid attention to his paper because it really seemed impossible. A 10 to the 98th times improvement was unprecedented, and Kao didn't actually measure the loss in glass. He didn't have anything concrete, so he needed proof.

Kao was trying to invent a new way to communicate, but even today actually staying connected can be hard. In fact, we just traveled to Spain for the eclipse a few weeks ago, but with so many other projects going on, we still needed to check in and send updates to the rest of the team, not to mention my mom. (laughs) We drove seven hours to Burgos, but the Wi-Fi when we got there was a little touch and go. We needed to have everything set before we arrived on site, so of course we used Saily, today's sponsor. After I downloaded their app and activated my eSIM, all I had to do was choose my data plan and country,

and then I was set before I'd even reached the airport. It's available in over 200 destinations, and they even offer regional plans, which is great if I have an extended shoot in multiple countries. I'm automatically connected right when I land, ready to update my team, access maps, or get transportation to wherever I'm filming. I installed an eSIM once, and I use it almost anywhere in the world. There's no hidden charges, so it's extremely affordable and convenient.

Actually, let me show you. (buzzer ringing) Wait, I can't get SMS messages right now. But with Saily you can now get a U.S. phone number right through the app for voice calls or SMS messages as an optional add-on. There's no contract or physical SIM card needed. So if you've got a trip coming up, and you still don't know where you're going to get your data well, you can scan this QR code for the Saily app and check out with code "Veritasium" so you'll be able to get 15% off your first purchase. Again, just scan this QR code, use the code "Veritasium" and get 15% off, and you can stay connected.

Now, let's go back to Kao. (striking dramatic music) Kao needed to prove that his theoretical low-loss glass was really possible, so he decided to measure the purest glass anyone had ever made, fused silica. Now, the reason glass manufacturers had added all those extra ingredients was to bring sand's melting point down to around 1400 degrees Celsius, but fused silica was pure silicon dioxide melted at over 2,000 degrees Celsius, so it was a lot harder to make, and that meant that Kao couldn't get his hands on much of it. Measuring loss normally takes hundreds of meters of cable, if not kilometers, but he only had pieces less than 30 centimeters long. To measure the loss in something that small,

he and his colleagues invented a new, very precise instrument. Kao spent the next three years working on it, but when he finally measured a sample, he found that fused silica had zero loss. (dramatic music continues) That couldn't be right. In a panic, Kao ran his experiment again. It turned out that the loss was so low that it was within their margin of error, a mere plus or minus four decibels a kilometer. He spent months fine-tuning his equipment, and in 1969, he finally reported that fused silica had a loss of less than five decibels a kilometer.

Now, that got everyone's attention. One scientist at Bell Labs said, "Kao gave everyone a jolt. "That was the first practical measurement "which said, 'Hey, you're not just whistling Dixie.'" Now everyone was interested. Telecom labs around the planet rushed to fund fiber optics groups, but Kao had only worked with small samples. Nobody knew whether kilometers of low-loss fiber were even possible. The problem caught the attention of one Robert Maurer, Corning researcher and author of the scattering paper that Kao had referenced. To make a practical fiber, he saw three major problems to tackle.

First, he needed to decide which glass to use. He could try purifying raw materials for normal glass, or he could try using fused silica, which Kao had already proven to be very pure. Most labs were avoiding fused silica because of how hard it was to work with. But luckily, Corning had a secret technique. Back in the 1930s, another Corning scientist had ran silicon tetrachloride gas through an oxygen flame. In the heat, water vapor from the flame would break silicon tetrachloride into hydrogen chloride gas and silicon dioxide.

The latter would then start clicking together in a fine glassy silica powder. Then the powder would be pressed into shape and melted to make fused silica. Corning was one of the few companies in the world using this technique called flame hydrolysis to make ultra-pure glass. So they had the supply, but there was a bigger problem with fused silica. It has a refractive index of 1.46, which is the lowest of any glass, but for total internal reflection to work, the core needs to have a higher refractive index than the cladding. So if Maurer wanted to use a fused silica core, well then finding another glass

with a lower refractive index for the cladding, well, that'd be a nightmare. It simply didn't exist. So everyone abandoned fused silica except for Maurer. He said, "I had no reason to think fused silica "was any better, but nobody else was working on it. "If you do something that's different "from what everybody else is doing, "you've got two advantages. "One is you may succeed where they fail, of course, "but even if you fail, you will gather information "that they don't gather."

All the other labs were thinking of using a fused silica core. After all, that's where the light travels, but Maurer decided to use fused silica for the cladding. Then he could easily raise the refractive index of the core by adding some impurities or dopants. - The only glass that we had with a slightly higher index than fused silica was this titanium-doped glass. - [Henry] But that would mean adding titanium right where light needed to travel. Kao worked hard to prove that glass could be pure enough. All the other labs were trying to get rid of impurities, and Maurer wanted to add some back in. But he wasn't completely crazy.

Like silicon, the titanium in glass can easily share electrons with four oxygen atoms, forming the same pyramid shape, so it wouldn't disrupt the overall glass structure too much. (gentle music) Now that he had the composition, Maurer needed to tackle that second problem. How do you make a tiny fiber with that perfect core and cladding cross section? To help, he hired two more scientists, Peter Schultz, a ceramics engineer, and Donald Keck, a physicist. - Each time you find a problem, you put a development team on it and strike it down and make it happen.

- To understand their technique, I went to the Bay Area Glass Institute for some hands-on demos. Here, our instructor, Erin is taking a colored rod of glass and then dipping it in a bowl of clear molten glass. In fiber optics, this large cylinder is called a preform. (gentle music) We have two colors of glass here. We have a color core and a clear outside. If you want your inner rod to, say, be a quarter of the diameter of your outer one, you just set that in the preform, and then all the length along your pole, the cross-section's going to be identical. And this is pretty similar to how hard candies with patterns are made.

This sugar is an amorphous solid, just like glass. So the process is the same, just at a lower temperature. But there is one problem with this method. Resting glass on a surface like this introduces a ton of stress and impurities, but when I try and suspend it in the air, well, it sags. - [Erin] But yeah, from all that movement, it did create stress, so now it's a curved piece of, of glass. - Thankfully, there's another option.

Now, gravity is pulling our glass for us, and the surface section is keeping it together. (light dramatic music) So now we get a much more straight and consistent fiber. That's awesome. (dramatic music continues) - With their drawing technique sorted, the corning team could tackle the final problem. How do you actually make the preform? Well, instead of starting with a core and adding the cladding, Schultz and Keck decided to start from the cladding and work in.

They used the flame hydrolysis technique to make titanium dope silica powder, or soot as they called it. Next, they took a pure fused silica tube and tried spreading soot down the inside. If they could get an even layer, well, then they could melt and collapse the tube into a solid rod, a perfect preform. But the soot collected at one end no matter how they sprayed it. - So one of the technicians grabbed a vacuum cleaner from the corner of the room, and he held it up to the backside of this tube. And it sucked a stream of titania dope silica glass down through the center of the tube, coated it just like we'd hoped it would.

- [Henry] The vacuum did not survive that experiment, but they tinkered with a new setup until they got the gas flow just right. (striking dramatic music) All through the summer of 1969, their furnace burned "hotter than Hades." They drew kilometers of fiber from preform after preform, testing various compositions, but none of it was good enough. Titanium was breaking down in the high heat and absorbing too much light. They ran the finished fibers through a second heat treatment in an oxygen environment before they finally started seeing results.

(light dramatic music) A year later, on July 22, 1970, a preform slipped against the furnace wall after Keck only got a few meters of fiber. The preform was ruined, but he decided to test the short fiber anyway after its heat treatment. - When I slid it across and the laser light hit the core of the fiber, all of a sudden I had another beam coming back in my face. The core was brightly lit, and it took a while to realize that the light was going all the way to the other end of the fiber and reflecting off the far end and coming back into my eye.

- [Henry] The loss was 17 decibels a kilometer. - The only word I could come up with for this auspicious moment in history was "Whoopie". So that's what's in my lab notebook, "Whoopie". - [Henry] But the fiber was much shorter than usual, only 29 meters to their typical 200. He added a careful "must remeasure" in his notebook. Team quickly remade the preform, but it took almost a month to get 210 meters of finished fiber. It was finally done by late afternoon of Friday, August 21st.

It was a gorgeous summer day and almost everyone left early, but Keck didn't want to wait. The fiber was so fragile, small temperature changes might break it over the weekend. - It would break if you looked at it crossways. - So alone in the lab, he anxiously measured the loss, 16.9 decibels per kilometer. (dramatic music continues) - I said "QED, quod erat demonstrandum" There was no, (chuckles) no eureka. - [Henry] Maurer published their results that fall. As Schultz described it, no one believed it. And telephone companies all demanded

to see the fibers firsthand, so Maurer hand carried a spool of fiber to all the big telephone laboratories. He would literally guard to make sure that they didn't snip pieces off and keep them. (dramatic music continues) For two years, they refined their methods. And to make the fiber stronger, they switched to a germanium doping. Like titanium it easily swaps with silicon while still raising the refractive index. But unlike titanium, germanium fibers didn't break down in the heat, so they could skip the second heat treatment. When they drew their first germanium fiber,

the end lit up with the light from the furnace. Even when it reached the bottom and wound up around a large drum, the fiber stayed bright. - And we kept winding and winding and thinking the light was going to go out, and it never went out. - The loss was only four decibels per kilometer with excellent strength. Keck considers that the very first truly practical low-loss fiber. By 1977, AT&T trialed a fiber system underneath Chicago.

Two years later, a fiber team from Japan was able to get a loss of just 0.2 decibels a kilometer, the standard up to today. Fiber networks continued to grow with the first transatlantic fiber optic cable installed by 1988. And it transmitted those phone calls differently than in copper. The audio wave was split up or sampled into discrete values, and then converted into binary to get a long string of ones and zeros. Then a laser flashed on for one and off for zero, 280 million times a second.

These pulses of light traveled down the fiber, but that wasn't all. Multiple conversations could now be weaved together, one bit from one conversation, then one from another and so on, slotting it together like a zipper. As long as the receiver knew the timing of each conversation and the sampling rate, well, they could reconstruct the original waves. (dramatic music) So this first intercontinental cable could carry 40,000 simultaneous calls, 10 times more than the previous copper version.

This insane capacity opened up a whole new world, and with digital signals people realized that they weren't limited to phone calls. It was easier than ever to send texts, images, videos. All building up to the internet, and since then demand has only increased. - This reel right here? (light upbeat music) - Yeah. - 50.477 meter kilometers, so 50 kilometers. So it's about 30 miles.

- [Henry] That's crazy. Like how much is in this warehouse? - There's millions of kilometers here that we're pushing through here every month. - [Henry] To see how modern fibers are made, we visited Prysmian's North Carolina factory. It all starts with the preform. Each pass of this microwave source ignites a mixture of gases into a plasma which deposits a thin layer of glass on the inside of the tube.

- And you see when it turns purple, that's when we're making a plasma with that gas. - So, how thick are these deposition legs? Like how many times are you doing this? How thick are you doing-- - Thousands of passes. - [Henry] "Thousands of passes." - Several thousand times it's doing this. Every time we may go back and forth, we could change the composition of that if we wanted to.

- [Henry] Oh, okay. After all the inner layers are done, the tube has a thicker wall, but there's still a gap down the middle. - So we want to collapse that channel upon itself. So what we do is we heat it up. We have a furnace that goes back and forth, and just via surface tension, it collapses upon itself. - Interesting, I would've thought maybe you'd like pull it, or just-- - [Sujeet] And I asked this exact question. I said, "Are you pulling a vacuum in there "or what's making that thing shrink?"

- Exactly. - It's really just a glass phenomenon where the molecules are trying to pull together. You heat it up, you'll lower the viscosity of the glass. It just. (exhales) - [Henry] Now they need to add more silica on the outside to get the preform to the right proportions. The core is very small, just 7.2% of the total diameter. - And you have a several thousand C plume of plasma right there.

There's grains of silica that's flowing into that plume and effectively melting it and spraying it onto that rod. - [Henry] Wow. (steam hissing) - That plasma torch is approximately three times as hot as the surface of the sun. - Plasma always blows my mind. This is absurd. Once the preform's done, it goes to the top of the draw tower, seven stories up to start making fiber. (steam hissing) It really looks like some sort of steampunk thing, you know? All these coils and then this hot glass going down.

(gentle dramatic music) The bottom of the preform is tapered to a point and heated to over 2,000 degrees Celsius. It's like there's some heat coming off this. The end of the taper slowly drops down to the floor below. - [Man] We call it a gob, but it looks like a teardrop. - The fiber is a little thick, so it'll be thinned down to the final 125 microns before it passes to the next level. (gentle dramatic music) It's like a little flashlight coming down. (glass clinking) I didn't expect it to be so violent.

(gentle dramatic music) If the drawing, like the pull down to that final diameter happens in such a short amount of. - In a few meters, let's call it. - A few meters. Then why do you need six more floors? - You got to cool the glass before you can apply the coating. And if you cool the glass too quickly, it'll have residual stress and then have optical loss attenuation. (glass clattering) - Once the fibers are cool enough, they are dipped in two protective plastic coatings. At the bottom of the draw tower, the fiber is wound onto massive spools, each capable of carrying hundreds of kilometers at over 50 meters a second.

Okay. Next, the finished fiber is tested for loss and strength. Fiber is stronger than a steel wire of the same thickness. I guess if I try to do it like. - [Sujeet] If you just try to do it like that, it's really hard to break. - First of all, it just slides down my hand. - [Sujeet] And that's a glass fiber with two coatings on it. - Ah, it's just hurting my hand. Oh, my god. Yeah, not easy.

Definitely not easy. Ow. But it's weak to sharp turns. - [Sujeet] Tie it in a knot. - Tie it in a knot. - And watch what happens and see how much you can get before it breaks. (gentle dramatic music) Yeah, because the angle's so great, you can break it. - But after testing, fiber is bundled and wrapped into thicker cables so that it's harder for them to knot and break. These cables will join the more than eight billion kilometers of fiber already installed that carry information around the world, powering the internet and letting you watch this video. But some experts argue that even with fiber,

the internet would not be possible without another key invention. The amplifier, that's what boosts signals across long stretches, and it revolutionized fiber capacity on land and sea. So if you'd like to see a video about submarine cables and fiber amplifiers, well, let us know in the comments. (quiet dramatic music) As for Kao, his dream went further than he ever thought. When he received the Nobel Prize in 2009 for his work, he said, "The information revolution that started "in the 1990s could not have happened "without optical fibers. "It is beyond the dreams of the first serious concept "of optical communication in 1966, "when even one gigahertz was only a hope." For me, one of the things I kept on coming back to

in this story is that Kao was not a material scientist. He was an electrical engineer. And it was only because he was an outsider that he was able to ask the questions that nobody was asking. He was able to push the status quo, but he was not the only one. Maurer's team were also working with a very unintuitive idea that by making something seemingly worse, they actually were making it better. As Kao said, "Who knows what other forward thinkers "will dream of in the future?" Granted, he was talking about bandwidth, but I think it applies anywhere where people are able to question why things are the way they are and whether that's really the limit.

(electric chirping) One last thing. These videos are only possible because of the incredible work done by our team behind the scenes. We have world-class editors, animators, researchers, and writers that make these videos possible. We'd love to shine more of a light on their work, so we organized a Q&A with the writer of today's video, Sulli. She discusses the fascinating things she learned while researching this topic and things that didn't make the cut, but that are really interesting. So if you want to learn more about that, you can sign up for our newsletter in the description

and you can find her Q&A right there. (light gentle music)