Count Rumford made water boil without fire by drilling a cannon, and that experiment helped kill the idea that heat was a liquid. At the end of the 18th century, most scientists believed heat was a substance. They called it caloric: an invisible fluid that flowed into objects when you heated them and drained out when they cooled. It explained a lot. Metal swells in a flame because caloric pushes its particles apart. It shrinks in the cold because the caloric leaves. Benjamin Thompson, an American-born statesman better known as Count Rumford, spent years trying to break that idea. His first test was simple. If heat is a fluid, it should weigh something. He sealed water, alcohol and mercury in three matched bottles, balanced them perfectly in a warm room, then moved them into the cold. The water froze, losing a great deal of heat. The bottles stayed perfectly balanced. His second test went after another rule of caloric: the colder something gets, the denser it becomes. He chilled water to freezing and touched the surface with a slightly warmer metal point. Within about 20 seconds, a thermometer below began to rise. The warmer water had sunk. Near freezing, water does the opposite of what the theory demanded. The third test came from his job. In Munich he supervised the boring of cannon for the Bavarian army, and he noticed how hot the metal got. If friction squeezes caloric out of metal, he reasoned, the supply should eventually run low. So he surrounded a cannon with a jacket of water and kept the drill turning. After about two and a half hours, the water boiled. The heat never weakened. And the metal shavings, which should have been drained of caloric, held heat just as well as the original metal. Rumford's conclusion was blunt: anything that can keep supplying heat without limit cannot be a material substance. What he didn't have was a better theory. That took another 50 years, until James Prescott Joule measured exactly how much heat a given amount of work produces. Helmholtz and Clausius then built it into the principle we still use today: heat is energy. The film ends with a warning to anyone who feels too sure. By the late 1800s, scientists were certain that Newton's laws and the conservation of mass and energy were absolute. Then came the electron, and Einstein, and the certainty cracked. Every theory is waiting for its Rumford.

50 years of studies couldn't explain why widening a pipe past a small angle made it work worse. Between 1900 and about 1950, engineers measured it again and again. A diffuser is just a passage that spreads out, slowing the flow and raising the pressure. Open the angle a little, and the pressure rise grows. Open it more, and it drops. Everyone had the numbers. Nobody had the picture. So a team stopped calculating and started looking. The tool was a wire 2 mils thick, stretched across a water channel. Run a DC current through it and electrolysis strips hydrogen out of the water. Tiny bubbles peel off and ride the flow. Small enough to follow it almost exactly. Insulate parts of the wire, and the bubbles come off in bands. Pulse it, and they come off as squares. Those squares turned out to be the key. In a flat, 2-dimensional flow of water, each square keeps the same area forever. So when a square stretches long, the water there is fast. When it squeezes short and wide, the water is slow. The flow draws its own speed map. Put an airfoil in the channel at 0 lift, and a line of bubbles splits around it and meets again at the back. Same speed above and below. Tilt it for lift, and the top half of the line pulls ahead of the bottom half. Then the film shows why this work is harder than it looks. There are 4 different lines you can draw in a flow. The path of 1 drop. The trail of every drop that passed 1 point. A line of drops marked at the same instant. And the streamline, the line the math uses, which no dye or bubble can ever show directly. In steady flow, 3 of them are the same line. You see 1, you know the other. Swap the steady flow for a plate swinging back and forth, and they split apart. Two drops enter at the exact same point, a moment apart. Their paths cross almost at right angles. One goes around the top of the plate. The other goes around the bottom. None of the bubble trails match either path. Neither matches the streamline. The water looks like it's telling you something. It's telling you something else. For 50 years, the diffuser problem lived in exactly that gap. The math came later. First, someone had to see the water.

A physicist's experiment explains why telephone wires sing in the wind. The answer starts with a column of smoke. Smoke rises because it is lighter than the air around it. Near the source, the plume is smooth and steady. A little higher, waves appear. Higher still, it breaks into chaos. That moment, when a calm flow stops being calm, is called a flow instability. It is the first step toward turbulence, and it happens almost everywhere you look. To study it up close, the physicist built a wind tunnel. Room air entered at one end, passed over a shallow pool of glycerin and water, and got sucked out the other end by vacuum cleaners. He could control the air speed exactly, and he could poke the liquid to see what happened. With no wind, a poke died out quickly. With a light wind, it still died out. Then, past a certain speed, the same poke grew into waves. A little faster, and waves started appearing on their own, out of nothing. Every flow has a speed like that. Below it, disturbances fade. Above it, they feed. He then added a wave maker running at 5 cycles per second and slowly turned up the wind, recording wave height with a sensor and a pen recorder. Then he did it again at another frequency, and another. When he plotted the points, a curve appeared. Outside it, every wave died. Inside it, waves were amplified. His conclusion was that a flow behaves like an amplifier. It takes random noise as input, such as a gust, a vibration, or a stray ripple. It draws power from the moving air. And it boosts only a narrow band of frequencies. Once the gain on that band passes 1, the flow goes unstable. Wind pushes hardest on waves it outruns. Pressure drops over the crests and rises in the troughs, lifting them higher. Gravity and surface tension try to flatten them back down. The longest waves move almost as fast as the wind, so they feel too little of it to grow. The shortest ones get crushed by surface tension. Only the middle band wins. The same pattern shows up in very different places. Heat a thin layer of fluid from below, and past a critical point it splits into hexagonal cells, with hot fluid rising in the center and sinking at the edges. Tilt the pan, and the cells grow larger where the fluid is deeper. Clouds sometimes trace that same pattern across the sky. Spin a cylinder inside another cylinder, and the smooth flow between them suddenly stacks into ring-shaped vortices. Spin it faster, and a second instability appears. Faster still, and you get turbulence. Drag a cylinder through water slowly, and the flow behind it stays steady. Speed it up past the critical point, and vortices start peeling off one side, then the other, in an alternating trail called a Kármán vortex street. Just below that point, the calm is so fragile that 1 accidental nudge sets it off. A wire in the wind is that cylinder. Every vortex that peels off tugs it sideways, back and forth, hundreds of times per second. That's the sound you hear. Rivers, flames, weather, and smoke from a chimney all work the same way. Calm is a balance, and every flow has a speed at which the balance tips.

Richard Feynman described an experiment where simply watching changes the result. In 1964, he stood in front of an audience at Cornell and tried to explain the strangest thing in physics using nothing but a wall with 2 holes in it. He started with a warning. Our intuition comes from everyday objects, such as balls, rocks, and water. That world is made of huge numbers of particles moving slowly, so there is no reason the deeper laws of nature should feel reasonable to us. Most of them don't. Even "at the same time" turns out to be a matter of opinion. 2 observers can disagree about which of 2 events came first, and both can be right. Light had already broken intuition several times. Newton's era saw it as a rain of tiny particles. Later experiments showed that it spreads and ripples like a wave. Then in the 20th century it started arriving in lumps again, as photons. Electrons went the opposite direction. They were discovered as particles, and then Davisson and Germer caught them diffracting like waves. Feynman's conclusion was that neither word fits. Electrons aren't particles and they aren't waves. They behave in a way that nothing in your everyday life does. To show it, he set up an imaginary experiment. First, fire bullets at a wall with 2 holes, and catch them on a backstop. With 1 hole open, they pile up behind that hole. With the other hole open, they pile up behind that one. With both open, you get the 2 piles added together. Each bullet arrives in 1 lump and goes through 1 hole. Now send water waves through the same wall. The waves pass through both holes at once and spread out on the other side. Where the ripples meet crest to crest, they grow. Where a crest meets a trough, they cancel. The result is a striped interference pattern, and it isn't the sum of the 2 holes. Then fire electrons. They arrive at the detector 1 click at a time, in single lumps, like bullets. But as the clicks build up, they form stripes, like waves. Close 1 hole and the pattern changes everywhere, even in places far behind the hole that's still open. Each electron seems to have been affected by both holes, and you can't say which one it went through. So you try to find out. You put a light beside the holes and watch for a flash as each electron passes. You see each electron go through 1 hole or the other, every time. And the stripes disappear. The pattern turns into 2 plain piles, exactly like bullets. Feynman insisted this isn't a problem of clumsy equipment. Any device that can tell you which hole an electron used disturbs it enough to erase the interference. That is the uncertainty principle, and it is built into nature. The math behind it is strange too. Nature doesn't add the probabilities for the 2 holes. It adds something called amplitudes and squares the result, and amplitudes can cancel out. All it gives you is a probability. You can never predict where the next electron will land, only the pattern that thousands of them will eventually draw. He didn't treat this as a gap that future physics will close. This is how the world works, and our imagination has to stretch to fit it. The universe behaves one way when you look, and another way when you don't.

A physicist ran the experiment to its limit and warned that below a certain point, time might stop existing. He got there by trying to define time on camera, failing, and then measuring it anyway. He started by pointing out the trap. Look up "time" in a dictionary and the definition uses words like before, after, and succession, which already assume you know what time is. A watch is easy to define. Time isn't. No one has managed it. So he did the next best thing and asked how short a moment we can actually notice. A film is a stack of still frames shown 24 times per second, and motion looks smooth. Drop it to 12 frames, then 6, and the motion turns jerky. Somewhere between 0.01 and 0.1 seconds, our brains stop seeing a flow and start seeing steps. Everything faster than that is invisible to us. A drop of milk hits a cup and is simply there, already splashed. A high-speed camera shooting thousands of frames per second reveals what happened in between: a crown of liquid rising and collapsing. Flip it around, photograph a rose once an hour, play it back quickly, and the flower opens almost violently. Both clips contained a clock. That, he said, is all a time measurement really is: you compare the start and end of an event with the position of a clock's hands. The hard part is building a clock you can trust. For centuries, the best one was the sky itself, with days, moon phases, and seasons. But the planets tug on each other, and even the solar system doesn't keep perfect time. Galileo found a better one in a church. He watched a lamp swing and timed it against his own pulse. Wide swing or narrow, each one took the same time. Only the length of the rope mattered. In the studio, the physicist hung bowling balls from a ceiling about 25 feet high, burned through their threads, and let them swing side by side. Big or small, they kept the same beat. Pendulum clocks ruled for centuries, but they have weaknesses. Warm weather stretches the pendulum, and the clock drifts. Carry it up a mountain, gravity weakens, and it drifts again. After World War 2, physicists started counting with atoms instead. A cesium clock steers itself by the vibration of cesium atoms flying through a tube. At the time there were only a couple of dozen in the world. When 2 of them were compared over 3 days, they disagreed by a few parts in 100 billion. Run 2 such clocks for 3,000 years, and they would drift apart by about 1 second. At MIT, Harold Edgerton chased the other end of the scale. His strobe flashed for about a millionth of a second, freezing a bullet in flight twice, 100 microseconds apart. An oscilloscope showed that an ordinary camera flashbulb takes about 1/10 of a second to burn out. Electronics can push measurement down to around 10⁻¹⁰ seconds. Beyond that, measuring turns into reasoning. A particle racing near light speed leaves a track a fraction of a millimeter long, which means it crossed that distance in about 10⁻¹⁵ seconds. Light crosses an atom in about 10⁻¹⁸ seconds and a nucleus in about 10⁻²³. No one can observe those intervals. We can only imagine them, and he admitted that no one knows what they mean. Cut gold in half enough times and you reach a single atom, and past that it stops being gold. Time might work the same way. Past some point, it might stop being time. We can measure a second to 1 part in 100 billion, and we still can't say what a second is.

A physicist ran a simple experiment in the 1960s and showed that almost every bubble you have ever seen is secondhand. Each one was pinched off from an older bubble hiding in a crack. His whole film was built on one idea: surfaces exert forces. Stretch a soap film across a wire frame and lay a loop of thread on it. The film pulls equally in every direction, so the thread snaps into a perfect circular arc. Pull the thread down and you are doing real work, dragging more molecules up into the surface. A surface is not just the place where a liquid stops. It is a stretched skin that costs energy to make. That skin decides shapes. When 3 soap films meet, they always meet at exactly 120 degrees, because that is the only way their pulls cancel. 4 films can meet for an instant, and then the junction collapses. It also decides pressure. A curved surface squeezes whatever is inside it, and the smaller the curve, the harder the squeeze. That's why a brand-new bubble is nearly impossible to make. At zero size, the math demands infinite pressure. Water with every trace of gas removed can hold more than 5,000 pounds per square inch of tension before it tears. So the bubbles in your boiling pot are not born. They are pinched off from tiny vapor pockets in scratches the water never fully filled. In a carefully cleaned glass beaker, water heats past boiling and nothing happens. Lower a glass rod with a trapped air pocket into it, and bubbles start pouring out of that 1 spot. The pocket never fully empties. A sliver of vapor always remains, waiting to seed the next one. Then the film gets stranger. When surface tension is uneven, the surface itself starts to move. Swirl a glass of wine and watch the thin film on the side. Alcohol evaporates from it faster than from the liquid below, so the film's surface tension rises. The stronger surface pulls liquid upward against gravity, and it keeps pulling as long as alcohol keeps evaporating. The liquid piles up at the top until it is too heavy and runs back down. Those streaks have a name: wine tears. Heat does the same thing. Warm liquid has weaker surface tension than cold liquid. Put a gas bubble in a liquid that is hot on one side and cold on the other, and the cold side pulls surface around from the hot side. That moving surface drags a sheath of liquid with it and jets it out the back. The bubble swims toward the heat. It absorbs energy where surface is created and releases it where surface is destroyed. Physically, it is a tiny self-propelled heat engine. On a heated wire in acetone and water, these bubbles don't rise. They parade along the wire, each one shooting a jet of hot liquid behind it. Their surfaces move at about 1,000 bubble diameters per second. You can see a version of this in your kitchen. Put a homemade ice cube, the kind with bubbles in it, on a metal surface. It starts turning in its own meltwater, stops, and turns again. The bubbles underneath are pushing it. We usually think of a surface as the place where something ends. Most of the time, it is where things actually happen.

An MIT physicist put on a safari outfit and shot a monkey in front of his students. The monkey was a stuffed doll named Robert. He was wearing a bulletproof vest. The physicist was Walter Lewin, and the shot was the final act of a lecture about one of the oldest puzzles in physics. A hunter aims straight at a monkey hanging from a branch. The instant the gun fires, the monkey lets go and drops. Where should the hunter have aimed? Most people say below the monkey, to catch it as it falls. Others say above it, because the bullet will drop too. Both are wrong. The hunter should aim straight at the monkey, because the bullet hits it every time. Before the hunt, Lewin did something less dramatic and more impressive. He used physics to predict exactly where a metal ball would land before he fired it. First he needed the launch speed. He fired the ball straight up beside a stick marked at 3 meters, and the class watched where it peaked. It went about 7 centimeters past the mark. From that single height he calculated the speed, and then he did something most demos skip: he estimated his own error at about 5%. Then he made his predictions. At 45 degrees, the ball should land 6.14 meters away, give or take 31 centimeters. At 30 degrees and at 60 degrees, the ball should land at the same spot, 5.31 meters away. Those 2 shots follow completely different paths. The steep one climbs high and hangs in the air. The shallow one skims low and fast. Both should land in the same place. He fired 3 times, and all 3 landed inside the predicted range. The reason behind the 45-degree rule is surprisingly simple. Double the launch speed, and the ball stays in the air twice as long while also moving sideways twice as fast. That is 4 times the distance. In a vacuum, 45 degrees is the perfect balance between going up and going out. Then came Robert. The doll hung about 3 meters up, held by an electromagnet. The moment the gun fired, the current cut off and Robert dropped. Lewin aimed straight at him, counted down from 3, and pulled the trigger. The ball hit the monkey in midair. It works because gravity doesn't care what it is pulling. Without gravity, the ball would fly in a straight line to where the monkey was hanging. With gravity, the ball falls below that line by exactly the same distance the monkey falls in the same time. They drop together, so they meet. The speed of the ball doesn't matter. As long as it reaches the monkey before hitting the floor, it connects. Seen from a falling elevator, there is no curve at all. Both objects fall at the same rate, so the ball flies in a straight line to its target. "Poor monkey," Lewin said. "See you Friday." Everything falls at the same rate, and that is exactly why the hunter can't miss.

Ships passing the fjords of Norway would suddenly crawl, engines at full power, on a perfectly calm sea. A physics experiment in a tank of salt water shows what was holding them back. Sailors called it dead water. A vessel would be making its usual speed along the Norwegian coast, reach the mouth of a fjord, and slow down as if something had grabbed the hull. Nothing was visible. No current, no wind, no waves. The engines kept burning fuel, and the ship barely moved. The Swedish oceanographer Ekman worked out the cause. Fresh water from melting snow and rivers pours out of the fjords and spreads over the heavier salt water of the ocean. It doesn't mix. It forms a separate layer, a lid of fresh water sitting on top of the sea. A ship moving through that top layer makes waves, but not on the surface. It makes them underneath, on the hidden boundary where fresh water meets salt. Those internal waves can be enormous, and they swallow the energy of the engines. From the deck, the sea looks flat. Below the keel, the ship is dragging a mountain of water. At higher speeds the effect fades and the ship breaks free. In the lab, the experiment is almost absurdly simple. Put a lighter liquid on top of a slightly heavier one and push a small obstacle through. The top surface barely ripples. The boundary between the layers heaves in huge, slow waves. They are slow because gravity barely acts on them. Ordinary waves are pulled back by the full force of gravity. Internal waves are pulled back only by the tiny density difference between the layers. If 2 liquids differ in density by 1%, the restoring force is roughly 1% of gravity. The waves move as if they were on a planet with almost no pull. Real oceans and the atmosphere don't have 2 neat layers. Their density changes gradually with height. Push a parcel of that fluid downward and it finds itself surrounded by heavier fluid, so it gets shoved back up. It overshoots, sinks again, and keeps oscillating. In a fluid with uniform density, these waves are impossible. To test this on a real landscape, researchers built a model of Owens Valley, east of the Sierra Nevada in California, and filled the tank with salt water layered to match the atmosphere. Pilots and balloon crews had already mapped the air over that valley on a day with a strong mountain wave. Tuned to the same conditions, the tank produced the same 3 waves over the model valley. The same physics explains smog. On a calm morning or early evening, the lowest air is cool and heavy under a warmer layer above. Smoke leaving a chimney can't rise through it. It flattens into a thin sheet and hangs there, trapping everything we breathe close to the ground. The ocean does it too. Waste released from a pipe on the seabed rises only until it reaches water of its own density, then spreads sideways at that level and stays there, hidden in the middle of the sea. The surface is usually the calmest part of the water. Everything that matters is happening underneath.

Two machine guns firing at the same wall can never cancel each other out, but two beams of light can, and that single fact started a 300-year fight over what light actually is. In a 1950s television film, Harvard physicist Edwin C. Kemble laid out the case like a trial. Since the seventeenth century there had been two answers. Isaac Newton said light was a stream of tiny particles, fired from the source like bullets. Christiaan Huygens said it was a train of waves, spreading out the way sound spreads from a tuning fork. Newton's best argument was the shadow. Water waves and sound waves bend around corners. Light casts sharp edges. So light, he concluded, could not be a wave. It took two centuries to see the flaw. In a ripple tank, waves passing through a gap do spread past its edges, but the shorter the wave, the tighter the beam stays. Light waves are so short that their spreading is almost invisible. Almost, but not quite. Look closely and every shadow is a little soft. Newton's evidence ended up as proof for the other side. Then came the decisive test. Cut two very narrow slits close together and shine light through them. Each slit alone makes a plain patch of light on a screen. Together, the patch is crossed by dark stripes. In those stripes, light plus light makes darkness. Bullets can't do that. Waves do it all the time. Two vibrating fingers in a pool of water send out rings that cross like the aisles of an amphitheater. Where crest meets crest, the water heaves. Where crest meets trough, it lies still. From the spacing of those stripes, scientists measured the wavelength of green light: about one fifty-thousandth of an inch. The same method later measured everything from radio waves a mile long to gamma rays shorter than a billionth of an inch. All of them, it turned out, travel through a vacuum at the same 186,000 miles a second. The wave theory had won. Then the short waves started behaving badly. A poker heated gently glows dull red. To make it glow blue, you need far more heat. Ultraviolet takes a violent electrical spark. X-rays take a beam of high-speed electrons slammed into a block of heavy metal. Short waves also tear molecules of air apart and burn living tissue, in ways long waves never do. Nothing in a smooth, spreading wave explains why shortness should mean violence. The cloud chamber showed what was happening. Long X-rays left a faint scatter of ions. Shorter X-rays left something else: sharp tracks shooting off at angles from the beam, exactly like the tracks of electrons. They were electrons, knocked clean out of air molecules. Make the X-rays brighter, and those electrons came out no faster. Make the wavelength shorter, and they came out with more energy, as much as you liked. A wave spread thin across space could never concentrate that much energy on a single electron. Only a packet could. Kemble called them explosive shells. Physicists called them photons. Light spreads out like a wave and lands like a bullet. Kemble's comparison was Dr. Jekyll and Mr. Hyde. Then the paradox spread. Fire X-rays through powdered crystal, and they draw rings on film by interference. Fire electrons through the same crystal, and they draw the same rings. Matter itself behaves as if waves guide it. An atom, Kemble said, makes sense only as a system of particles and a system of standing waves at once, like a bowed violin string. That double nature became quantum theory, and he ended on the screen his audience was staring at. The picture was being painted by a beam of electrons, guided by laws no one can picture. We built television out of something we still can't imagine.

In 1923, a Paris PhD student claimed that every particle of matter is also a wave, and his examiners almost refused him the degree for it. The student was Louis de Broglie. Physicists had only just accepted that light, a wave, also arrives in particles called photons. De Broglie turned the idea around. If light can act like particles, he argued, then electrons, atoms, every piece of matter might act like waves. His examiners called it foolish. He got his doctorate only because Albert Einstein happened to be visiting Paris and told them it wasn't such a bad idea. Maybe it was even true. De Broglie had also written down what that wavelength should be: Planck's constant divided by the particle's momentum. Put real numbers in and the answer is tiny. An electron pushed by 100 volts should have a wavelength of about 10⁻¹⁰ of a meter, roughly the width of a single atom. To catch a wave that small, you need something to scatter it, with gaps the same size. A ruled grating splits light into beams because its grooves are spaced close to light's wavelength. For electrons, the grooves would have to be one atom apart. No one could rule lines that fine. They didn't have to. The surface of a crystal is already a grating, atoms lined up in perfect rows with exactly the right spacing. A Bell Labs film from the early 1960s brought viewers to Lester Germer, the man who first saw it happen. He turned off the lights and fired a beam of electrons at a crystal. The electrons didn't scatter randomly the way bullets or paint droplets would. They bounced back in separate beams and lit up a neat, symmetrical pattern of spots on a fluorescent screen. Then he turned a dial. At 40 volts the pattern was large. At 47 volts it shrank. Back to 40, it grew again. Faster electrons meant a shorter wavelength, and the pattern contracted exactly as much as de Broglie's formula said it should. Thirty-five years earlier, Germer and Clinton Davisson had no screen. Their tube was a sealed metal can with a small collecting box inside that swung freely under gravity. To measure at a new angle, they tipped the whole apparatus and read the current, point by point. At about 53 degrees, the current spiked. The strange part is that they weren't testing de Broglie at all. They were running the experiments before they had even heard of his theory. When they finally read it, they checked their data against it, and it fit. At almost the same time in Cambridge, G. P. Thomson fired electrons through a thin gold foil and got a set of rings on film. Fire X-rays through the same foil, and the rings land at the same spacing. A beam of light and a beam of particles were obeying the same rules. Then came the rest of matter. In 1929, beams of helium atoms bounced off a crystal in diffracted beams. Later, neutrons fired at ordinary table salt drew nearly the same pattern as X-rays. By the time of the film, Germer had turned his own discovery upside down. He no longer bounced electrons off crystals to prove they were waves. He trusted that they were, and used them to map the atoms on a crystal's surface. An idea that nearly cost a student his degree became an ordinary tool on a lab bench.

7 millionths of a centimeter at sea level. 80 meters at 100 miles up. That is how far an air molecule travels before it hits another one. The gap is over 1 billion times. And a 1960s film showed that above a certain height, air stops behaving like air. A vacuum chamber is pumped down to 10^-5 mm of mercury. A small oven boils sodium metal. A sodium vapor lamp throws a thin sheet of light across the chamber. Every glowing point is a sodium atom. The atoms leave the oven in straight lines. Almost none collide. A second hole cuts them into a beam with razor-sharp edges. Molecules flying like light rays. Then argon is let in. At first, nothing changes. More argon. The top of the beam fades. More. The beam gets shorter and shorter. Then it's gone. The atoms are still there. They're just knocked off course before they get far. What decides the flow is 1 ratio: how far a molecule flies, compared to the size of the object. The Knudsen number. Above 1, molecules act alone. Below 1, they act like a fluid. A returning spacecraft crosses the whole range on the way down. So the film builds one. A small heated washer stands in a jet of sodium and argon, with a tiny hole through its center. Thin gas first. Molecules that hit the washer bounce back upstream. The rest fly straight through the hole and form a new beam behind it. Add argon. Incoming molecules start hitting the ones bouncing back. The beam behind the hole disappears. A haze builds in front of the washer. Its edge sharpens into a bow shock. The flow is now at about Mach 9. Then the reverse. The washer is removed. A jet fires at a pressure ratio of 100 to 1, forming barrel shocks and a Mach disk, the pattern seen in rocket exhaust. The argon is slowly cut. The shocks blur. They melt into 1 hazy spot. Then even that is gone. Jet molecules fly past background molecules without touching them. One last test. A polished aluminum oxide surface is placed in the beam. A mirror would bounce it back as a clean ray. It scatters in every direction. Shock waves need crowds. Up there, every molecule flies alone.

A physicist's 1880s experiment explains why a thinner liquid can be harder to push through a pipe. A 1969 film from the National Committee for Fluid Mechanics Films showed it happening. Two tanks held glycerin and water mixed to different thicknesses. A pump drove the mixture through a long pipe at a fixed rate, and a gauge showed how hard it had to push. Thinner liquid, less pressure. Then the valve turned one notch further, to an even thinner mix, and the gauge jumped up. Thicker again, the pressure fell. Thinner again, it rose. The answer was at the end of the pipe. The thicker mixture poured out in a clean, glassy stream. The thinner one came out with blurred edges, its surface writhing in slow motion. The flow had turned turbulent, and turbulence brings drag with it. Reynolds found the rule that decides it: a single number built from the pipe's width, the speed of the flow, and the liquid's viscosity. Below about 2,000, pipe flow stays smooth. Above it, it usually breaks down. With a gently flared entrance and great care, the flow can stay smooth up to nearly 100,000. Turbulence doesn't simply switch on. It grows from small disturbances the flow can no longer damp out. A thread of dye shows the difference. In smooth flow it slides down the pipe as one unbroken line. Past the threshold, it seems to explode across the whole tube in an instant. That churning is why turbulence mixes so well. Two liquids layered in a jar would take about a week to blend on their own. Stirred into turbulence, they blend in under a minute. Large swirls stretch every blob into thin filaments, smaller swirls fold those, and so on down, until diffusion can finish the job almost instantly. Physicists call it the energy cascade. At large scale, a turbulent flow barely cares how thick the liquid is. Only the smallest swirls give it away. The film used that to catch Hollywood: studios often burn a scale model instead of a full set, and the big billows of smoke look real. Look at the fine detail, the narrator said, and you can tell which fire is fake. Turbulence can also be smothered. Warm water flowing over cold barely mixes, because every swirl has to lift heavy fluid upward. Flip the layers and the channel boils. Above a city, the same arrangement is called an inversion, and it traps the smog. The film ended on something still unsolved. The smallest turbulent motions don't fluctuate evenly. They arrive in sudden bursts, separated by stretches of calm. In 1969, no one could explain why. It mixes rivers, fools film studios, and churns on the surface of the Sun, and no one can fully explain it.

A ball that bounces like rubber on a lab bench is actually a liquid, and left alone for 45 minutes it melts into a puddle. That ball opens Rheological Behavior of Fluids, a film from the National Committee for Fluid Mechanics Films. Water and glycerin follow a simple rule: when they aren't moving, they push equally in every direction, and when they flow, their resistance depends only on how fast they're being sheared at that instant. Physicists call these Newtonian fluids. The ball breaks that rule. It's made of long polymer molecules, and it behaves as if it remembers. Hit it fast, and it springs back like a solid. Give it time, and it forgets its shape. The film calls this a fading memory, and it shows it with a cylinder dipped into a polymer solution. Twist it a full turn and let go at once, and it swings back almost halfway on its own. Twist it, hold it for a few seconds, and then let go, and it barely comes back at all. Another experiment makes the memory measurable. A weight turns a cylinder inside a ring of liquid. In an ordinary fluid, the cylinder starts at a steady speed and stops the instant the weight is released. In a polymer solution, it speeds up slowly, and when the weight is released, it turns backward. Nothing is pushing it. Part of the energy that drove it was stored in the liquid and given back. Some liquids refuse to flow at all until they're pushed hard enough. A clay suspension sits in an open tube with no stopper and doesn't drip. A small weight on top does nothing. A bigger one makes it creep, and when the weight comes off, it stops. Mayonnaise works the same way: it holds its shape on a plate and spreads easily on bread. Others change their thickness as they move. Double the pressure on water in a tube and twice as much comes out. Double it on a polymer solution and more than twice as much pours out, because the faster it flows, the thinner it gets. Some suspensions do the opposite and thicken under strain. Then the narrator brought in one of his wife's favorite recipes. Stir glycerin with a spinning beater and it gets flung outward. Stir cake batter and it climbs up the shaft. The long molecules in the batter are stretched around the rotating rod, and as they pull tight they squeeze the liquid inward and up. The same effect, measured between spinning plates, produces such high pressure near the center that engineers used it to design a pump for molten plastic. It also explains why a polymer swells as it leaves a nozzle. Push one through a small hole and the stream comes out wider than the opening. Factories that extrude plastic have to cut their dies smaller than the part they actually want. Water forgets everything the moment you stop stirring. Some liquids remember what you did to them.

Around 200 BC, a Greek astronomer measured the size of the Earth with nothing but the shadow of an obelisk and the distance between 2 towns. In a 1957 television lecture, the Harvard physicist Philippe Le Corbeiller used that story to answer a question most people never think to ask. Science says the Sun is 93 million miles away and that an atomic nucleus is so small it would take several million million of them to span an inch. Nobody has walked to the Sun. Nobody has held a nucleus. So where do the numbers come from? He started by taking the drama out of big numbers. Sixty million sounds enormous, but sixty million inches is about a thousand miles, an easy drive. A number is only large next to a unit, and our units were built to fit us. The foot began as the length of a king's foot. The pound is something a man can lift. The second is a beat of the clock. Every measurement in science, however vast, ends at a dial read by a human eye and a knob turned by a human hand. The trick is to get from that dial to the thing you can't reach. Eratosthenes did it first. He knew that at noon on the summer solstice, the Sun stood exactly overhead in Syene, in southern Egypt. The obelisk there cast no shadow at all. On the same day, farther north in Alexandria, the obelisk did cast one. From its length and the obelisk's height, he worked out that the Sun was about 7 degrees from straight overhead. The Sun is so far away that its rays reach both cities in parallel lines. That means the 7 degrees in the sky is the same 7 degrees of the Earth's curve between the 2 towns. And 7 degrees is about one fiftieth of a full circle. So the Earth's circumference had to be 50 times the road from Syene to Alexandria. Surveyors put that road at 5,000 stadia. No one today knows exactly how long a stadium was, but his answer lands close to the modern one: a radius of about 4,000 miles. A shadow on the ground had measured a planet. The Sun was much harder. Repeat the same method and the angle shrinks to about 8 seconds of arc, far too small to measure well. Astronomers did not know the Sun's distance even roughly until the 1930s, when they used a small asteroid called Eros as a stepping stone. The answer was 93 million miles. You can write that in a notebook. You can never picture it. The small runs the same way. A micrometer screw measures a ball bearing to a fraction of a thousandth of an inch. A hairline inside a microscope, multiplied by the lens maker's careful design, gives a red blood cell at about one three-thousandth of an inch. Light itself stops at around one ten-thousandth of an inch, and for years that seemed to be the floor. Then electrons replaced light, magnets replaced glass lenses, and a virus showed up on film at one millionth of an inch. Le Corbeiller then pointed at the screen his audience was watching. They weren't seeing him either. A beam of electrons had scanned his image, and another beam was redrawing it in their living rooms. Only a purist, he said, would insist they weren't looking at him. Science sees the Earth, the Sun and the virus the same way you saw him that night. Between the Milky Way and the electron stands a person reading a dial, and the dial is enough.

A physicist loads film into a camera with no lens, and the developed sheet comes out completely blank. Cut it into pieces, and every fragment still holds the entire image. That's a hologram. It has almost nothing in common with a photograph, except that both use film. To understand it, you have to start with water. Drop 2 sources of waves into a tank. Where crest meets crest, the water heaves higher. Where crest meets trough, the motion cancels to nothing. If the 2 sources have the same frequency and amplitude, these zones of motion and calm lock into a fixed pattern. Do the same with 2 speakers playing an identical pure tone, and a microphone moving through the room finds the same thing in sound: bands of loud and bands of silence, hanging in the air. Light is a wave too. So when 2 beams of light cross, they should paint bands of light and dark. They don't. An ordinary bulb throws out many wavelengths at once, none of them in step. Even passed through a red filter, the light is too messy to interfere visibly. Then came the laser: light so pure and so monochromatic that it finally behaves like the water in the tank. Split a laser beam in 2. Where the beams cross, place a film in a holder with no lens. Develop it, and it looks blank. Now remove 1 of the beams and shine the other through the film. The missing beam comes back. The film has rebuilt it from nothing but a recorded pattern. Under a microscope, that pattern is a forest of microscopic layers inside the emulsion, acting like a stack of tiny, partially silvered mirrors. Now replace the second beam with a car. Every point on its surface reflects laser light, and each reflection interferes with a clean reference beam aimed straight at the film. The table must be absolutely still. The exposure lasts about 10 seconds, and the slightest vibration smears the pattern and ruins everything. Under magnification, the developed film shows swirls that look nothing like a car. But shine the reference beam through it, and the car is there again. Fully 3-dimensional. Each tiny pattern reconstructs its own point of light, and together they rebuild the whole object. Some holograms can be seen in ordinary white light. Inside them, the mirror layers are spaced to reinforce only 1 wavelength. Red laser light made the image, yet it glows green, because the film shrinks during processing and squeezes the layers closer together. The strangeness keeps going. 1 film can store 2 separate scenes, and you switch channels by tilting it. A lens recorded inside a hologram still focuses light. Laid over the real object, a hologram reveals what no eye can see: a single tap of a finger visibly bending solid steel, and the invisible shimmer of heat rising from a surface. When this film was made, holography stood roughly where photography stood 100 years earlier. A photograph remembers how light fell on a thing. A hologram remembers the light itself.

Centrifugal force says the leaves in a stirred cup of tea should fly to the edge, yet every time they pile up in the center, and the reason is a hidden flow running backward along the bottom. A 1960s fluid mechanics film set out to explain it, and it started with a question about a drain. The tank was a whirlpool. Water entered from the sides, swirled, and left through a hole in the center of the floor. The question was simple: what path does the water take to get out? A long-needled syringe of dye gave the first answer. Released at the surface, the dye just circled. At that rate it might never reach the drain. Released at half depth, it circled again. Then the needle went down to an eighth of an inch above the floor, and the dye shot straight toward the center and vanished down the hole. Almost all the water was going around. Only a thin layer at the bottom was actually going out. The film gave the two motions names. The big circular swirl that fits most of the water is the primary flow. The small departure from it, the thin inward rush along the floor, is the secondary flow. And it is the secondary flow that moves things. To see it clearly, the filmmakers spun a cylinder of water for a couple of hours until all of it turned as one, then stopped the container. The water kept spinning, but near the wall and the floor friction began to slow it. A fine vertical wire released pulses of hydrogen bubbles into the flow. From above, most of the lines curved in circles. Near the bottom, they bent inward. Seen side by side, the pattern was exact. The inward flow lived in precisely the same thin layer where the water had been slowed by the floor. That was the key. To travel in a circle, water needs pressure pushing it inward, and the main swirl sets that pressure for the whole cup. Fast water can resist it. The slow water near the bottom can't. Under the same push, it takes a tighter turn and gets swept toward the center. The tea leaves sit in that slow layer. They ride it inward and stop where it stops. Then the film flipped the drain. The bottom hole was closed and a siphon pulled water from just under the surface. Dye went in near the floor, rushed to the center, and had only one way left to go. "You ever see a tornado?" The same rule held in a channel with a bend. When all the water entered at the same speed, it went around the curve as a single sheet, top to bottom. When some of it entered slower than the rest, the bend pulled the slow water toward the inside of the curve, and a second flow appeared that no one had put there. Curve the path, give the water different speeds, and it will quietly start moving in a direction nobody stirred. Every swirl hides a second flow, and that is the one that carries things away.

1 alpha particle in every 8,000 bounced off a sheet of gold foil only a few thousand atoms thick. Ernest Rutherford already knew his bullets. He had measured the alpha particle's mass, its charge, its speed. He had worked out that it was a helium atom stripped of its 2 electrons. And he had watched it pass straight through gases, liquids, even metal. So he aimed it at gold. In a cloud chamber, the result looks like nothing. Without the foil, the longest tracks reach about 4 divisions on a scale. With the foil, about 3. The particles lose a little energy and keep going. The foil might as well not be there. His colleague Hans Geiger measured it properly. A glass tube 1.5 meters long, pumped empty. Radium at one end. A slit in the middle. A glowing screen at the far end, watched through a microscope, 1 flash at a time. Without foil, the beam was 1° wide with sharp edges. With foil, about half the particles shifted by 1°. A few by 3°. None beyond 5°. A small blur. A tidy result. Then a young graduate student, Ernest Marsden, looked where no one expected anything. Decades later, a film recreated the moment. Polonium source. A transistor detector wired to a loudspeaker, 1 smack per particle. About 500,000 particles a second pour through the beam. The detector swings to the edge. The smacks fade. Past the edge, silence. Then, 25° outside the beam, a click. Another. That's 3. The foil is pulled out. Silence. The foil goes back in. A click almost at once. Marsden's count was 1 in 8,000 thrown back more than 90°. Not nudged. Reversed. Nothing in a smooth sheet of gold could do that. The film shows why with a toy. Steel balls roll down a ramp across a slab of felt, leaving tracks on wax paper. 200 go through. Most roll straight. 6 fly off at wild angles. Under the felt: hard steel pins. Rutherford's guess was the same. Almost all of an atom's mass, packed into a tiny core at the center. Everything else nearly empty, except for a few light electrons. The felt was the atom. The pin was the nucleus. The 1 that bounced back mattered more than the 7,999 that didn't.

In the 1960s, a physics teacher explained the force that holds every atom together with a butter gun and a few slices of toast. He told it as a fable. A restaurant owner wants to butter toast by machine, so he builds a gun that sprays melted butter out in straight lines. Hold one slice close, and it gets buttered all over. Move it twice as far back, and the same spray now covers a square two slices wide and two slices high. Four slices share the butter, and each gets a quarter. Three times as far, and nine slices share it. Each gets a ninth. "Economy treatment," he called it. Anything that spreads out in straight lines without getting lost follows this rule. Light from a small lamp does. Newton guessed that gravity does. And a century after Newton, Charles Coulomb guessed that electric charge does too: twice the distance, a quarter of the force. Before testing it, the teacher showed what that force can do. He filled a tube with water, fitted a piston that barely slid inside, set it on a soft sponge, and struck it hard with a mallet. The water didn't give. Electric forces hold the parts of atoms together, and they also hold atoms apart. That is why water refuses to be squeezed. Then he built his own version of Coulomb's experiment. Two metal balls, one on a beam, a spring to balance the push between them, and a pointer on a scale. To show how sensitive it was, he placed a tenth of a gram on the ball. The pointer settled near ten divisions. Each division was about a thousandth of a newton. He charged both balls and set them one span apart, about fifteen centimeters. The pointer read 23.7. At two spans, the force fell to roughly a quarter. At three spans, it read 2.5. Multiply 2.5 by nine, and you get 22.5. Almost exactly where he started. The small shortfall came from charge slowly leaking away through imperfect insulators. Then he tested the charge itself. He touched one ball with an identical uncharged ball. Because the two were the same size, symmetry demanded the charge split evenly between them. He grounded the spare and measured again. The force dropped from just above 18 to just under 9. He didn't really need to repeat it with the other ball, he said. If the force depends on one charge, it must depend on the other. "How do the balls know which is which, which is agent and which is victim?" He did it anyway. The force halved again. A spring and a ruler can't reach inside an atom. Physicists have to assume the law still holds at that scale, and then check it when they can. When they checked, it held. The rule that decides how thinly butter spreads across a table is the same rule that keeps matter from collapsing into itself.

A shock wave is 1 millionth of an inch thick, and your kitchen sink makes the same thing out of water. A 1960s fluid mechanics film set out to prove it with a water channel, a knife, and a tube of argon. It starts with 3 channels. Water 1 inch deep, 4 inches deep, 9 inches deep. Small waves are released at the same instant. They arrive at speeds in a ratio of 1 : 2 : 3. The deeper the water, the faster the wave. Speed grows with the square root of depth. That one rule explains why every wave at the beach breaks. In a rising wave, the back sits in deeper water than the front. So the back moves faster. And the water behind it is already pushing forward. The back catches the front. The wave steepens. Then it topples. Once toppled, it stops changing shape. It moves at a constant speed, faster than before. Engineers call it a positive surge. Send the camera along at the surge's speed and the wave stands still. Water rushes into it from one side and leaves slower on the other. That frozen wave is a hydraulic jump. Ahead of it, water outruns its own waves. Behind it, water is slower than its waves. Any ripple behind the jump catches up. Any ripple ahead is swept into it. The jump in the film burns about 1/3 of a horsepower in foam and turbulence. Below big dams, jumps in stilling basins burn millions. The water barely warms. It takes a 778-foot drop in head to heat water by 1°F. Now look at your sink. Water shoots from under the tap faster than its own waves. Then, a few inches out, it hits a ring, thickens, and slows. That ring is a hydraulic jump. Swap water depth for air density and the math barely changes. Sound speeds up where air is denser. A strong compression wave steepens the same way. But air can't topple. So the wave squeezes until viscosity and heat stop it. The front locks at a few molecular free paths thick. That is a shock wave. Thunder. A sonic boom. The blast front of an atomic bomb, and the wind that follows it. In the lab, a celluloid sheet holds back high-pressure gas. A knife cuts it. Streamers at the far end snap forward in the wind behind the shock. Then the lights go off. The low side of the tube is filled with argon. The knife cuts again. The shock hits so hard it tears electrons off the atoms. The tube glows. Cameras running at nearly 400,000 frames per second show shocks sliding over wedges, bouncing, spinning off vortices at sharp edges. All of it starts where the sink ring starts. A sonic boom is a wave that tried to break and couldn't.

1 hertz was all it took to push a physics demo to the edge of breaking. The setup was simple. A mass on an air track, a spring on each side. Nudge it, and it swings back and forth about once a second. That's its natural frequency. The rhythm it wants. Then the professor stopped nudging and started forcing. A motor pulled on the spring at a frequency he chose. Not the one the system wanted. At first, the mass resists. It tries to do its own thing. Then it gives in and moves at the driver's rhythm. He compared it to taking someone by the arms and shaking them. They object. Sooner or later, they go with you. First, he drove it slow. Far below 1 hertz. The mass moved with the driver, step for step. Small swings. Then he drove it fast. Far above 1 hertz. The mass almost stood still. What little it moved, it moved backward, 180 degrees out of step with the push. Same force. Same spring. Almost nothing happened. Then he tuned the driver toward 1 hertz. The swings grew. And grew. He had to turn it off. The equation predicts it. The amplitude equals the force divided by the gap between 2 frequencies squared: the one the system wants and the one you force on it. Close the gap to 0 and you divide by 0. The amplitude goes to infinity. In real life, friction always stops it short. How short depends on the damping. Little friction gives a tall, narrow spike. A lot of friction flattens it. Then it gets stranger. Add a 2nd mass and a 3rd spring, and the system has 2 resonant frequencies. 3 masses, 3 resonances. 5 masses, 5. Every extra part adds 1 more hidden frequency it can't ignore. Nobody has to guess where they are. Drive the system, sweep the dial, and watch where it goes wild. A bridge. A building. A wing. Each one is a long chain of masses and springs. Force doesn't break things. Timing does.