![[images/NiCgcUTL9_Q_thumbnail.jpg]] **Creator:** Veritasium · **Presenter:** Henry van Dyck · **Published:** 2026-10-12 · **Length:** 34:32 · [Watch on YouTube](https://www.youtube.com/watch?v=NiCgcUTL9_Q) > *Why making glass fibers to send signals was almost impossible.* (from the video description) *Summary written from yt-dlp English captions (manual track) and the creator's chapter list. Timestamps follow the captions.* ## 1) Detailed outline ### [0:00] What are fiber optics? *(chapter)* ![[images/NiCgcUTL9_Q_00m20s.jpg]] - **1887:** a young physics teacher (C. V. Boys) attaches a glass rod to a crossbow arrow, melts its tip, and fires it, pulling a fiber under a tenth the width of a hair. Henry recreates it ("legally we cannot shoot a crossbow indoors"). - Boys used the fibers to measure tiny forces. Later researchers used them to carry light, even around 90° bends. - In the 1960s, telecoms wanted far more capacity. One engineer dreamed of carrying "all of the conversations going on anywhere in the world simultaneously over a single thread of light." But early fibers lost so much light that a signal was unreadable after about 1 km. - One man's boss called his dream "a wild goose chase." Teasers: a Prysmian warehouse moving "millions of kilometers" a month, and a plasma torch about 3× hotter than the Sun's surface. ### [2:02] Why copper phone lines had limited capacity *(chapter)* ![[images/NiCgcUTL9_Q_02m40s.jpg]] - US daily calls doubled from 1960 to 1973 (285M to 576M). Apollo cost about $26B. AT&T spent nearly triple that on new phone infrastructure and still couldn't keep up. - **How an analog call works:** a DC current powers the phone. Your voice moves a diaphragm that modulates the current (amplitude = loudness, oscillation rate = pitch), and a speaker at the other end turns it back into sound. - **Frequency-division multiplexing:** speech only needs about 300–3,400 Hz, so each call gets a 0–4 kHz slot shifted up (4–8 kHz, and so on). Filters separate the calls at the far end, like tuning a radio. ### [4:00] Skin effect and the Shannon-Hartley ceiling ![[images/NiCgcUTL9_Q_04m30s.jpg]] - Higher frequencies create magnetic fields and **eddy currents** that cancel current in the wire's center, so the signal rides only the surface (**skin effect**). Resistance goes up and loss rises quickly. - Repeaters every few km helped, but above about **65 MHz** signals died within 2 km. - **Shannon-Hartley theorem (1948):** channel capacity depends on signal power, noise, and bandwidth. Bandwidth is the main lever, and copper had maxed it out. - *(The description notes the skin-effect explanation is simplified.)* ### [5:13] Trying to go from copper to glass *(chapter)* - Visible light runs at hundreds of THz, a "fabulous El Dorado" of bandwidth (Bell Labs). - Problem 1, no strong light source, was solved by **Ted Maiman's laser (1960)**, the "death ray" that was supposedly "a solution looking for a problem." - Problem 2: **glass is not as clear as it looks.** ### [6:30] The impossible number: 1,000 dB/km ![[images/NiCgcUTL9_Q_07m00s.jpg]] - Loss is measured in decibels (logarithmic, based on signal power). A Bell Labs boss set a goal of **under 20 dB/km**, comparable to copper. The best fibers measured **1,000 dB/km**. - **Leaky sand-truck analogy:** 100 kg to 1 kg over 1 km is 20 dB (1% left). 1,000 dB is like starting with the mass of the observable universe and ending with less than one atom. It needed a **10^98** improvement, so Bell Labs gave up. - **Charles Kao**, at an English telecoms lab (STL), kept asking: *why is glass so lossy, and is that intrinsic?* His boss called it "a wild goose chase." ### [8:15] Why is glass so lossy? Light trapping and total internal reflection *(chapter)* ![[images/NiCgcUTL9_Q_09m20s.jpg]] - Demo: a green laser in a falling stream of **propylene glycol** bounces along inside it. - **Refraction / Snell's law:** light changes speed between materials and bends. Going from high to low index, past the **critical angle**, you get **total internal reflection**. A bigger index difference traps light more strongly. - Propylene glycol (n ≈ 1.43) traps light better than water (1.33). - Touching a bare fiber lets light leak out, so fibers use a **high-index core inside a lower-index cladding**. Now they can be coated, cabled, and bent. Leakage out the sides was *not* the cause of the loss. ### [11:29] Scattering and absorption *(chapter)* ![[images/NiCgcUTL9_Q_11m30s.jpg]] - Glass is **amorphous** silica (SiO₄ tetrahedra sharing corners) with no long-range order. Density variations "frozen in" during cooling cause **Rayleigh scattering**. Slow cooling reduces it. - Using **Robert Maurer's** (Corning) scattering data, Kao estimated about **1 dB/km** of scattering for glass fixed at 1,000 °C. So scattering wasn't the problem. **Absorption** was. - **Band gaps:** silicon (~1.12 eV) absorbs visible light (1.63–3.26 eV) and looks opaque. Silica's gap is ~9 eV, so it looks clear. ### [14:30] Impurities: iron and the green edge of window glass ![[images/NiCgcUTL9_Q_14m30s.jpg]] - Raw materials (sand, soda, lime) bring trace metals. **Iron ions** strongly absorb violet/blue and infrared, which is why the edge of a window pane looks green. - Experts estimated that removing the metal ions would bring loss down to **about 20 dB/km**. - **1966:** Kao's team publishes. The problem is "difficult but not impossible." Nobody believes it, and he had no measurement. ### [16:00] Sponsor: Saily eSIM *(chapter: How can you remove impurities in glass?)* - The team's trip to Burgos, Spain, for the eclipse. Saily eSIM data plans in 200+ destinations, an optional U.S. number add-on, and code "Veritasium" for 15% off. ### [17:25] Kao measures fused silica ![[images/NiCgcUTL9_Q_17m45s.jpg]] - **Fused silica** is pure SiO₂ melted at over 2,000 °C. Normal glass uses additives to melt around 1,400 °C. Kao only had samples under 30 cm, so his team spent three years building a precise instrument. - First reading: **zero loss**, actually within the ±4 dB/km error margin. **1969:** reported **< 5 dB/km**. A Bell Labs scientist: "Kao gave everyone a jolt… you're not just whistling Dixie." Labs worldwide fund fiber groups. ### [19:00] How they make fiber optic glass: Corning's inverted design *(chapter)* ![[images/NiCgcUTL9_Q_21m40s.jpg]] - **Robert Maurer** (Corning) lists three problems: which glass, how to form core and cladding, and how to make the preform. - Corning's secret was **flame hydrolysis** (from the 1930s): SiCl₄ in an oxygen flame gives HCl and fine silica powder, a source of ultra-pure fused silica. - Fused silica has the **lowest index of any glass (1.46)**, so nothing was available to clad a silica core, and other labs gave up on it. - Maurer's twist: **pure silica cladding, doped core.** Titanium-doped silica raises the core index, and Ti forms the same tetrahedra, so the structure isn't disrupted much. His quote on contrarian research: even if you fail, "you will gather information that they don't gather." - He hires **Peter Schultz** (ceramics) and **Donald Keck** (physics). - **Bay Area Glass Institute** demo with instructor Erin: a colored core dipped in clear glass makes a **preform**, and drawing it keeps the cross-section ratio (like patterned hard candy). Letting gravity draw it, with surface tension holding it together, gives a straight, even fiber. ### [23:00] Building the preform, a vacuum cleaner, and "Whoopie" ![[images/NiCgcUTL9_Q_23m35s.jpg]] - Schultz and Keck work from the cladding inward. They deposit Ti-doped silica **soot** inside a pure silica tube, then collapse it into a solid rod. - The soot piled up at one end until a technician held a **vacuum cleaner** to the tube to pull the flow through. "The vacuum did not survive." - Summer 1969: kilometers of fiber, none good enough. Titanium broke down in the heat, so they added an oxygen heat treatment. ![[images/NiCgcUTL9_Q_24m50s.jpg]] - **July 22, 1970:** a ruined preform still yields a few meters. Laser light comes back off the far end into Keck's eye. **17 dB/km.** His notebook says "Whoopie" and "must remeasure." - **Aug 21, 1970 (Friday afternoon):** alone in the lab with 210 m of fragile fiber ("it would break if you looked at it crossways"): **16.9 dB/km.** "QED." ![[images/NiCgcUTL9_Q_26m00s.jpg]] - Nobody believed it. Maurer hand-carried spools to the phone labs and guarded them so no one snipped samples. ### [26:20] The technology that made the internet possible *(chapter at 26:47)* - A switch to **germanium doping** gives stronger fiber and drops the second heat treatment. The first Ge fiber stays lit around the whole drum ("it never went out"): **4 dB/km**, which Keck calls the first truly practical low-loss fiber. - **1977:** AT&T fiber trial under Chicago. **1979:** a Japanese team reaches **0.2 dB/km**, still the standard. **1988:** first transatlantic fiber cable (TAT-8). - **Digital transmission:** sample the audio, convert to binary, pulse a laser on/off **280 million times a second**, and use **time-division multiplexing** (interleaving bits "like a zipper"). The cable carried **40,000 simultaneous calls**, 10× the copper cable. Then texts, images, video, and the internet. ### [28:20] Inside Prysmian's North Carolina fiber plant ![[images/NiCgcUTL9_Q_28m20s.jpg]] - A 50.477 km reel (about 30 mi). The warehouse moves "millions of kilometers" a month. ![[images/NiCgcUTL9_Q_29m50s.jpg]] - **Preform:** a microwave plasma deposits glass inside a tube over **thousands of passes**, and the composition can change per pass. Heating then collapses the center channel **by surface tension alone** (no vacuum). - **Overcladding:** a plasma torch about 3× hotter than the Sun's surface melts silica grains onto the rod. The core is only **7.2%** of the diameter. ![[images/NiCgcUTL9_Q_30m30s.jpg]] - **Draw tower (seven stories):** the tapered tip is heated above 2,000 °C, the "gob" (teardrop) drops, and the fiber is thinned to **125 µm**. The extra floors are for **slow cooling**, because fast cooling leaves stress and adds loss. Then two plastic coatings, and winding at over 50 m/s. ![[images/NiCgcUTL9_Q_31m55s.jpg]] - **Testing:** fiber is stronger than steel wire of the same thickness. Henry can't pull it apart barehanded, but a tight knot snaps it. Fibers get cabled to resist knotting. - More than **8 billion km** of fiber is installed. Teaser: **optical amplifiers** and submarine cables could be a future video. ### [32:50] Outsiders and the lesson ![[images/NiCgcUTL9_Q_33m05s.jpg]] - **Kao, Nobel Prize 2009:** "The information revolution that started in the 1990s could not have happened without optical fibers… when even one gigahertz was only a hope." - Henry's takeaway: Kao was an **electrical engineer, not a materials scientist**, and as an outsider he asked questions no one else did. Maurer's team made the glass "worse" (doped) to make it better. "Who knows what other forward thinkers will dream of?" - Closing: a newsletter Q&A with writer/director **Sulli Yost**. ## 2) Things mentioned **Key numbers** | Milestone | Loss | |---|---| | Best 1960s fibers | ~1,000 dB/km | | Copper / Bell Labs target | ~20 dB/km | | Kao, fused silica (1969) | < 5 dB/km | | Corning Ti-doped (Jul 1970) | 17 dB/km | | Corning 210 m (Aug 1970) | 16.9 dB/km | | Corning Ge-doped (~1972) | 4 dB/km | | Japan (1979) / today | 0.2 dB/km | **Physics and engineering concepts** - Frequency-division multiplexing, time-division multiplexing, skin effect, eddy currents, Shannon-Hartley theorem, decibels - Refraction, Snell's law, critical angle, total internal reflection, core/cladding, refractive index - Amorphous solids, Rayleigh scattering, band gaps, impurity absorption (iron) - Flame hydrolysis, soot deposition, preform, plasma deposition, draw tower, dopants (titanium, germanium), optical amplifiers **Organizations and places** - AT&T / Bell Labs, Standard Telecommunication Laboratories (Kao's English lab), Corning Glass Works, NASA / Apollo - Prysmian (North Carolina plant), Bay Area Glass Institute, Rosendahl Nextrom (animations), Diffraction Limited (fiber kits), Trulala CANDY Chef - Burgos, Spain (eclipse trip), Chicago (1977 trial), TAT-8 transatlantic cable (1988) **Books and references** - Jeff Hecht, *City of Light: The Story of Fiber Optics* (main source) - Kao & Hockham 1966 paper; Kapron, Keck & Maurer 1970 "Radiation Losses in Glass Optical Waveguides" (shown on screen) - References list: ve42.co/FiberRefs **Products / sponsor** - Saily eSIM (sponsor), Snatoms (Derek Muller's magnetic molecular model kit), Veritasium newsletter and Patreon **Demos in the video** - Crossbow fiber pulling, laser in a propylene glycol stream, glassblowing a preform, pulling and knotting coated fiber ## 3) People **Henry van Dyck.** Veritasium presenter, co-writer, and executive producer. He hosts this video and does the demos and factory tour. **Sulli Yost.** Writer, producer, and director of the video. Featured in the newsletter-only Q&A. **Charles K. Kao (1933–2018).** Shanghai-born electrical engineer at Standard Telecommunication Laboratories in England. His 1966 paper with George Hockham argued that glass loss came from impurities and that fiber could beat 20 dB/km. Called the "father of fiber optics." Nobel Prize in Physics 2009. **Robert Maurer (1924–2018).** Corning physicist whose Rayleigh-scattering measurements Kao relied on. He led the Corning team that made the first low-loss fiber, choosing a fused-silica cladding with a doped core. **Donald Keck.** Corning physicist (interviewed in the video). He measured 17 dB/km in July 1970 ("Whoopie") and 16.9 dB/km in August, and later led the germanium-doped work. National Inventors Hall of Fame. **Peter Schultz.** Corning ceramics/glass chemist on the Maurer-Keck-Schultz team. He developed the doped-silica soot and preform processes. **C. V. Boys (Charles Vernon Boys).** British physicist, the "young physics teacher" of 1887 who drew quartz/glass fibers with a crossbow to make very sensitive torsion balances. **Theodore (Ted) Maiman.** Built the first working laser (ruby, 1960) at Hughes Research Labs. **Claude Shannon and Ralph Hartley.** Implied by the Shannon-Hartley theorem: Shannon's 1948 information theory built on Hartley's earlier work. **Sujeet Rao and Jon Fitz.** Prysmian's Operational Excellence Fiber Director and Technical Sales Support Sr. Manager. On-screen guides for the factory tour. **Erin.** Instructor at the Bay Area Glass Institute who demonstrates preform-making. **Other credited experts:** Prof. Vincent Chan (MIT), Sir David Payne (Southampton, erbium-doped fiber amplifier pioneer), Dr. Nikola Alic, Jim Hayes (Fiber Optic Association), Jeff Hecht (author). **Derek Muller**, Veritasium founder and executive producer.