How Screens Work, and the History of the Glass We Stare At

Look closely at the white of this page. If you are reading on a phone or a laptop, there is no white there at all. There are millions of tiny red, green and blue lights, each switched to a carefully chosen brightness, and your eye is too coarse to see them apart. A phone screen of 2,532 by 1,170 pixels holds nearly 3 million pixels and almost 9 million of these coloured lights. If you are reading on an e-ink reader, the trick is completely different: there are no lights at all, just tiny charged grains of white and black pigment pushed up or down by electric fields.

This piece follows the screen from its beginnings in 1897 to the foldable phones on sale in October 2026: who invented what, how each kind of screen actually works, what resolution and refresh rate really mean, and what the evidence says about blue light and sleep. Two interactives let you zoom into a pixel and test the refresh rate of the screen you are holding.

Part 1 — The cathode-ray tube: from watching electricity to television

The first screen was not built to show pictures. In 1897 Ferdinand Braun, a physicist at the University of Strasbourg, published a paper with a very unglamorous title: a method for demonstrating and studying the time course of varying electric currents. His device was a glass tube with the air pumped out. At one end, a high voltage pulled a stream of electrons (then called cathode rays) off a metal plate. The stream passed through a small hole and struck a coating at the far end that glowed where it landed. A magnet coil carrying the current he wanted to study pushed the beam sideways, so the glowing spot traced the current's shape. That is the first oscilloscope, and the "Braun tube", as Germans still call it, is the ancestor of every television made for the next century.

The idea of sending moving pictures by wire was older than the tube, and for a while it went down a different path. Mechanical television scanned a scene with a spinning disc punched with holes. On 26 January 1926 the Scottish inventor John Logie Baird showed moving television images this way to members of the Royal Institution in his London workshop, and the BBC later broadcast his system. But a spinning disc could not scan fast or finely enough. In 1908 the British engineer Alan Campbell-Swinton had argued in a letter to Nature that cathode-ray tubes could do the whole job electronically, at both ends. Getting it to work took two decades and a famous rivalry.

In 1927, in a laboratory in San Francisco, a 21-year-old Utah farm boy named Philo Farnsworth transmitted the image of a single straight line with an all-electronic camera tube he called the image dissector. He said he had got the idea of scanning a picture line by line at 14, from the rows of a ploughed field. Meanwhile Vladimir Zworykin, who had studied in St Petersburg under the television pioneer Boris Rosing, was developing electronic cameras and picture tubes, first at Westinghouse and then at RCA under David Sarnoff. RCA claimed Zworykin's 1923 patent application came first. The US Patent Office disagreed and awarded Farnsworth priority for his invention, and in 1939 RCA agreed to pay him $1 million in licence fees.

Television had no single inventor. It had a tube from 1897, an idea from 1908, and a decade of patent lawyers.

Mechanical television lost the race in Britain in 1937. The BBC had been alternating Baird's system with an all-electronic one from EMI-Marconi since November 1936, and it dropped Baird's in February 1937. From then on, for about seventy years, a television meant a heavy glass tube. So how did the tube make a picture?

A cathode-ray tube (CRT) never shows a whole picture at once. At any instant, only one tiny spot is lit. At the narrow back of the tube, a heated cathode boils off electrons, and a high voltage accelerates them into a tight beam. Coils around the tube's neck make magnetic fields that bend the beam, sweeping it across the screen from left to right in a line, then dropping it a little and sweeping again. Where the electrons hit the coating of phosphor on the inside of the glass, it glows. Varying the beam's strength as it goes paints light and dark. This line-by-line pattern is called a raster, and the word has outlived the tube.

The American standard, set in 1941, used 525 lines, 30 times a second. To stop the picture flickering, each frame was drawn in two passes, first the odd lines and then the even ones, giving 60 half-pictures, or fields, a second. This trick is called interlacing. When colour arrived, the frame rate had to be nudged down by 0.1%, to 29.97, to fit the colour signal in, and it stayed there.

Do the arithmetic and the speed is startling. At 525 lines and 29.97 frames a second, the beam draws about 15,734 lines every second, so each line takes about 63.6 millionths of a second. On a TV half a metre wide, the glowing spot crosses the screen at well over 7 kilometres a second. You never see a dot. You see a picture, because the phosphor keeps glowing for a moment after the beam passes, and your visual system blends what remains into a steady image.

A CRT is a single point of light moving at several kilometres a second, and your brain turns it into a face.

Colour needed three phosphors (red, green and blue) and three electron beams. The problem was making sure the "red" beam only ever hit red phosphor. RCA's solution was the shadow mask: a thin metal sheet just behind the glass, punched with a fine grid of tiny holes. Each beam came in at a slightly different angle, so through any hole it could only reach its own colour of dot. The US approved a colour standard in December 1953, and RCA's CT-100 of 1954 was the first colour set made in quantity. In 1968 Sony launched the Trinitron, which replaced the mask of holes with fine vertical wires and phosphor stripes. Trinitron sets were known for their brightness, and the vertical stripes of red, green and blue on modern LCDs echo them.

CRTs reigned for decades. Worldwide sales of CRT computer monitors peaked at about 90 million in 2000, and CRT televisions at about 130 million in 2005. Then they collapsed, and in 2012 Samsung's display arm, one of the last big makers, stopped making them. The tubes were heavy, deep, power-hungry and full of leaded glass, and once flat screens became cheap, nothing could save them.

Part 2 — Going flat: plasma and the liquid-crystal display

Plasma came first to big screens. In 1964 at the University of Illinois, Donald Bitzer, Gene Slottow and a graduate student, Robert Willson, built a flat display for PLATO, a pioneering computer system for teaching. Each pixel of a plasma screen is a tiny sealed cell of gas. A voltage turns the gas into a glowing plasma that gives off ultraviolet light, which makes a coating of red, green or blue phosphor glow, as in a fluorescent tube. Plasma TVs became the first big, flat televisions in the late 1990s. They had deep blacks and fast motion, but they were heavy, ran hot and used a lot of power. Panasonic stopped making plasma panels in March 2014, and LG and Samsung stopped the same year.

The winner was a stranger technology, and its story starts with a carrot. In 1888 the Austrian botanist Friedrich Reinitzer, studying a compound extracted from carrots, noticed that it seemed to melt twice: first into a cloudy liquid, then into a clear one. He had found a liquid crystal, a substance whose rod-shaped molecules can flow like a liquid yet line up together like a crystal. For eighty years it was a laboratory curiosity.

In the 1960s a team at RCA's laboratories in Princeton led by George Heilmeier found that an electric field could change how liquid crystals scattered light, and in 1968 they showed the first liquid-crystal display. It was not yet practical. The breakthrough was the twisted nematic cell, and it came with a dispute. Martin Schadt and Wolfgang Helfrich, at the Swiss drug company Hoffmann-La Roche, filed a patent in December 1970 and published in February 1971. The American James Fergason filed a similar US patent in 1971. The dispute was settled out of court, with all sides sharing royalties. Who sold the first LCD calculator is argued over too. Sharp's EL-805 of 1973 is usually given the credit, though calculators with an earlier type of LCD were sold in the US in 1972.

An LCD does not make any light. It is a stack of filters in front of a lamp, and each pixel is a tiny shutter. Read the stack from the back, where the light starts:

Partial voltages give partial twists, which give shades in between. To control millions of shutters separately, each subpixel needs its own tiny switch, a thin-film transistor. In 1973 a Westinghouse team led by T. Peter Brody showed a liquid-crystal display driven by thin-film transistors, and in 1975 Brody named the idea the "active matrix". It took the industry, led by Japan, Korea and Taiwan, until the 2000s to make large panels cheaply. In the last three months of 2007, LCD televisions outsold CRTs worldwide for the first time.

The LCD's weakness is built in. The backlight is always on, and a closed shutter still leaks a little, so blacks look dark grey in a dark room. Much of the LCD's later history, from LED backlights that dim in zones behind dark areas to "quantum dot" layers that purify the colours, is a fight against that leak.

Part 3 — Pixels: colour, sharpness and smooth motion

Your eye has three kinds of colour-sensing cone cells, most sensitive to long ("red"), medium ("green") and short ("blue") wavelengths. As Light & Color explains, your brain builds colour by comparing their three signals. That is the loophole every screen exploits. A screen does not need to produce the actual wavelength of yellow light. It only needs to excite your red and green cones in the same proportions that yellow light would, and you see yellow. Two physically different lights that look the same are called metamers. Every colour on a screen is one.

So each pixel is three subpixels, red, green and blue, and mixing is additive: light adds to light, and all three at full strength make white. (Paint is the opposite, as Color, Explained shows: each pigment subtracts light, which is why printers use cyan, magenta and yellow.) In the usual system each subpixel gets a number from 0 to 255, so there are 256 × 256 × 256 = 16,777,216 possible colours.

Try mid grey. Every subpixel gets the code 128, halfway to 255, yet each one shines at only about 22% of full power. Screen codes do not map to light in a straight line. The standard sRGB curve was designed to suit the way old CRTs responded to voltage and the way our eyes respond to light, and it puts more of the 256 steps among the darker shades. It takes a code of about 188, not 128, to light a subpixel to half power.

Try yellow, then blue. Yellow's two lit subpixels look far brighter than blue's single one, and not only because there are two of them. Our eyes are much more sensitive to green light than to blue. In the standard formula for brightness, green contributes about 72%, red about 21%, and blue only about 7%. That is why a pure blue screen looks dim, and why blue text on black is hard to read.

You have never seen yellow on a screen. You have seen red and green, side by side, too small to tell apart.

Not every screen uses three equal stripes. Many OLED phones arrange their subpixels in other patterns, with different numbers of each colour, and software adjusts for it. The principle is the same: three primaries, added, too small to resolve. Which raises the next question: how small is small enough?

Resolution is just a count of pixels: 1920 × 1080 ("Full HD") or 3840 × 2160 ("4K"). On its own it says little about sharpness, because the same 4K picture can cover a 27-inch monitor or a 65-inch television. What matters is pixel density, in pixels per inch (ppi), and how far away you are.

A person with normal, "20/20" vision can make out detail about one minute of arc across: a sixtieth of a degree, about the width of a coin seen from 80 metres. When Apple launched the iPhone 4 in June 2010, it called its 326 ppi screen a "Retina display", claiming the eye could not tell individual pixels apart at a normal distance. The arithmetic roughly agrees: at 30 cm, one arcminute is 0.087 mm, which needs about 290 ppi. Some people see finer than 20/20, so the claim is about typical eyes, not all eyes.

Computed from each screen's diagonal and pixel count, and the distance at which one pixel spans one arcminute. Beyond that distance, extra pixels add little that a 20/20 eye can see.

The table explains a quiet truth of the TV business. From a sofa three metres away, a 55-inch Full HD set is already at the limit of normal vision, so most of the extra detail of 4K, let alone 8K, is invisible unless you sit closer or buy a bigger screen.

Refresh rate is how many times a second the screen draws a new picture, measured in hertz (Hz). For decades the standard was 60 Hz, inherited from television. In June 2017 Apple's iPad Pro introduced "ProMotion", refreshing up to 120 times a second, and many phones now do the same. Since 2021 some phones change their rate as they go, from 120 Hz when you scroll to as low as 10 Hz for a still page, which saves power.

Why does a higher rate look smoother? Most LCD and OLED screens are "sample-and-hold": each picture is held still until the next one replaces it. When your eye follows a moving object, the object stays put for a whole frame while your eye keeps moving, so it smears across your retina. Display engineers found in 2005 that this holding, not the slowness of the liquid crystal, caused about 70% of LCD motion blur. The cure is more frames, or brief black frames in between. Test it on your own screen:

Part 4 — Beyond the backlight: OLED, microLED, e-ink and touch

In 1987 Ching Tang and Steven Van Slyke, chemists at Eastman Kodak in Rochester, New York, published a short paper on "organic electroluminescent diodes": thin films of carbon-based molecules that glowed when a small voltage pushed current through them. Glowing organic materials had been seen in France in the 1950s, but Tang and Van Slyke's two-layer design was the first that was efficient enough and needed a low enough voltage to be practical. It is the ancestor of the OLED screen.

In an OLED, every subpixel is its own tiny lamp. There is no backlight and no shutter. A black pixel is simply switched off, so blacks are truly black, and each pixel can turn on and off almost instantly. The screen can be thin and even flexible. The costs are real too. Organic materials wear out with use, blue ones fastest, so a logo or toolbar shown for thousands of hours can leave a faint permanent ghost: burn-in. And making large OLED panels is still expensive.

Sony sold the first OLED television, the small XEL-1, in 2007. Apple's first OLED iPhone, the iPhone X, came in September 2017. Newer panels stack two light-emitting layers on top of each other ("tandem OLED") so each layer can work less hard, which raises brightness and slows ageing. On 28 October 2026 Apple announced a MacBook Pro with a tandem OLED screen, refreshing at up to 120 Hz and reaching up to 2,000 nits of brightness.

MicroLED keeps the "every pixel its own lamp" idea but uses tiny inorganic LEDs, the same kind of material as the LEDs in a light bulb. They are brighter and longer-lived than organic ones, without burn-in. The first microLEDs were made by Hongxing Jiang and Jingyu Lin's group in the US in 2000, and in January 2018 Samsung showed a 146-inch microLED wall called The Wall. The problem is manufacturing. A 4K screen needs nearly 25 million separate microscopic LEDs placed precisely, and every faulty one must be found and replaced. Apple reportedly gave up on a microLED watch screen in 2024. For now, microLED lives mostly in giant video walls, plus a few luxury-priced TVs sold since 2021.

Every screen so far makes its own light. In the 1970s Nick Sheridon at Xerox's Palo Alto Research Center wanted the opposite: a display that worked like paper, by reflecting the light around it. His "Gyricon" used tiny balls, black on one side and white on the other, that rotated in an electric field. The idea that won came from MIT's Media Lab. Its team, which founded the company E Ink in 1997 and published the method in Nature in 1998, sealed tiny white particles in dark fluid inside microcapsules. A voltage pulls the white particles to the front, and the spot looks white; the opposite voltage pushes them back, and it looks dark.

Because it reflects light like paper, e-ink is easy to read in bright sunlight, where emitting screens wash out. And once the particles are in place they stay there with no power at all: the screen is bistable. It uses energy only to change the page, which is why an e-reader's battery lasts weeks. Sony's Librié, sold in Japan in 2004, was the first e-book reader with an E Ink screen. Amazon's first Kindle went on sale on 19 November 2007 for $399, advertised as reading "like real paper, even in bright sunlight".

An e-ink screen showing this page right now is using no power to show it.

The price is speed. Moving pigment through fluid takes far longer than switching a liquid crystal or an OLED, so e-ink refreshes slowly, can leave faint ghosts of the last page, and needs a full flash now and then to clear them. Colour e-ink exists, usually by laying colour filters over a black-and-white layer, but its colours are muted. That is why the refresh test in Part 3 cannot work on an e-reader, and why this library, read in an app that also runs on e-ink devices, keeps its text plain and its animations few.

Touch began, like the CRT, as an instrument. In 1965 E. A. Johnson of the Royal Radar Establishment in Malvern, England, published a short paper describing a "touch display" for computers, to be used in air traffic control. In 1972–73 Frank Beck and Bent Stumpe at CERN built transparent capacitive touchscreens for the control room of a new particle accelerator. In the 1970s the American company Elographics developed the cheaper resistive screen, which for decades was the common choice, including for the stylus-driven handheld computers of the 1990s.

Then came 9 January 2007. On stage in San Francisco, Steve Jobs introduced the iPhone, with a 3.5-inch "multi-touch display" that you pinched and swiped with bare fingers and no stylus. It went on sale in the US that June. The iPhone was not the first phone with a capacitive screen: LG's Prada phone went on sale in May 2007. It was not the first touchscreen phone either; IBM's Simon came in the early 1990s. What it did was combine multi-touch glass, software built around fingers and a large screen with almost no buttons. Within a few years nearly every phone looked like it.

Part 5 — Blue light, sleep and your eyes: what the evidence shows

Sensitive-topic note: this is general information, not medical advice. Persistent trouble sleeping, or eye pain or changes in vision, deserve a visit to a doctor.

The worry about screens and sleep has a real biological basis. In 2002 researchers led by David Berson showed that some cells in the retina are not for seeing at all. They contain a light-sensitive pigment called melanopsin, they report to the body clock, and they respond most strongly to blue light, around 480 nanometres. Bright light in the evening, especially blue-rich light, delays the body clock and suppresses melatonin, the hormone that signals night.

The best-known screen study, from Brigham and Women's Hospital and Harvard in 2015, had volunteers read either a printed book or a light-emitting e-reader in the hours before bed, over several evenings, in a laboratory. On the e-reader, they took longer to fall asleep, made less melatonin, had a later body clock and were less alert the next morning. It was a careful study, but a small one, under controlled laboratory conditions rather than ordinary life.

In ordinary life, the effect of the light alone looks smaller. A 2021 study randomly assigned 167 young adults with iPhones to an hour of phone use before bed with Night Shift (Apple's warmer evening colours) on, with it off, or to no phone at all, for a week. Across the whole group, sleep did not differ between the three conditions. A 2016 review of 20 studies of 125,198 children found that using a phone or tablet at bedtime went with roughly twice the odds of too little sleep. But children who merely had a device in the bedroom and did not use it also slept worse. That points beyond light, to what we do on screens: the messages, the games, the next video, and the bedtime that keeps slipping.

The same review found that bedtime device use went with poorer sleep quality, and with nearly three times the odds of excessive sleepiness in the daytime. But all 20 of the studies it pooled were cross-sectional: surveys that catch children at a single moment. They show that bedtime screens and bad sleep go together, not which causes which. A teenager who cannot sleep may reach for a phone, rather than the other way round. Untangling cause and effect needs studies that follow people over time, or experiments, and those are rarer, smaller and less dramatic in their results.

What actually helps follows from that. An international expert panel in 2022 recommended keeping evening light dim: from three hours before bed, no more than about 10 lux of "melanopic" light at the eye (roughly a dim, warm room), and darkness for sleep. In the daytime, they recommend bright light: at least 250 lux of melanopic light, more than many indoor rooms give. America's CDC suggests switching devices off at least 30 minutes before bed. The ophthalmologists' 20-20-20 rule is for eye strain: every 20 minutes, look at something 20 feet (6 metres) away for 20 seconds. Sleep Is a Skill in this library covers the rest.

Part 6 — The state of screens in 2026, and open questions

The newest frontier is screens that bend. Samsung's first Galaxy Fold, in 2019, weighed 276 grams. In September 2026 Samsung said its Galaxy Z Fold8 weighed 201 grams, calling it the world's lightest Fold. On 9 September 2026 Apple announced its first foldable iPhone, the iPhone Duo, with a 7.6-inch inner screen and a 5.4-inch outer one, starting at $1,999 in the US. Pre-orders opened on 16 October, it reached shops on 23 October, and it uses Apple's ProMotion variable refresh. Apple's own description admits the crease is still there, saying its matte finish "minimizes crease visibility". Folding screens are OLEDs: with no backlight to carry, an OLED can be built on a bendable plastic sheet instead of rigid glass.

Samsung's own weight figures show how quickly the engineering has moved: 253 grams for the Fold5, 239 for the Fold6, 215 for the Fold7 and 201 for the Fold8. Variable refresh rates have spread too. Apple says the new MacBook Pro's screen drops from 120 Hz to as low as 10 Hz when nothing on it is moving, saving battery in the way phones have done since 2021. How many people actually buy foldables is harder to say. Reliable, independent sales figures were not publicly available for this piece, so treat confident claims about their market share with care, whether they come from boosters or sceptics.

Some questions are genuinely open:

Can microLED be made cheaply? If the problem of placing and testing millions of tiny LEDs is solved, it could combine OLED's blacks with LCD's long life. If not, it stays a luxury.
Will foldables become normal? They are now light and thin, but still expensive and still creased. Whether they become the ordinary phone or stay a premium niche is undecided.
Can paper-like screens catch up? Colour e-ink, sold since about 2019 using colour filters, still looks washed out next to a phone, and it is far too slow for video. E Ink, owned since 2009 by the Taiwanese company Prime View International, and its rivals are chasing brighter colour and faster pages. A reflective screen that could show video comfortably would change what a reading device can be, and how much light we put in our eyes at night.
How good is good enough? Phones are already past what most eyes can resolve at reading distance, and TVs past it at sofa distance: from three metres, even a 55-inch Full HD set reaches the limit of a 20/20 eye. Much of what is sold as better resolution now buys little you can see; brightness, contrast and refresh rate may matter more.
What do screens do to sleep and attention? The light question is better understood than the behaviour question. The evidence increasingly suggests that what we do on screens matters more than the colour of their light. Larger studies that follow the same people over years, rather than surveys at one moment, are what would settle it.

The thread running through all of it is the one Braun started: a screen is a way of making light do what we want, one tiny spot at a time. Every generation has found a better spot, from a glowing dot moving at kilometres a second to a pixel that turns itself off.