Twenty-seven short points, in ordinary words. Every number here is the same as the technical version. Only the vocabulary is simpler.
The claim: if Earth were a ball, pilots would have to push the nose down constantly to follow the curve, and instruments would show it. They never do. So the Earth is flat.
Start by agreeing: pilots never dip the nose to follow a curve. True. Nobody has ever been trained to. Keep reading, because that is exactly what a round Earth predicts.
Flying around a curved planet means that “down,” the line from your plane straight to the center of the Earth, slowly changes direction as you go. Distance is what sets the angle. Fly from the North Pole to the equator, one quarter of the way around the globe, and that line swings a full 90 degrees from where it started. Speed only sets how quickly you cover the distance, so the rate of change is your speed divided by the size of the planet. At airliner cruise, about 560 miles per hour, that works out to roughly eight degrees per hour, a full lap in almost two days. Concorde, at 1,350 miles per hour, turned about twenty degrees per hour. Even at Mach 3, better than 2,000 miles per hour, you would turn about thirty degrees per hour. Set that against a wall clock: the minute hand sweeps 360 degrees every hour, and nobody can watch a minute hand move. The fastest jet ever flown turns eleven times slower than that minute hand, and an ordinary airliner forty-four times slower. As for the instruments: none of them was ever designed to show a human a change that small, and the points below walk through why, one instrument at a time.
Air pressure drops as you climb, so the atmosphere is stacked in layers like an onion. The autopilot holds one layer, and every layer wraps around the planet. Staying “level” means following the curve automatically. No nose-down needed, ever.
In the United States, below 18,000 feet, pilots set their altimeter to local weather so it reads real height above the ground. Above 18,000 feet everyone switches to one standard setting: 29.92 inches of mercury, which is 1013.25 hectopascals, the pressure of a standard day at sea level. Other countries make the switch at their own altitude, but every country makes it. Now every altimeter counts feet up from the same starting layer, the invisible level in the atmosphere where the pressure equals exactly that number. Two planes told to fly 1,000 feet apart really are 1,000 feet apart, because both measured from the same layer. And that layer is made of air wrapped around the planet, so it curves with the planet.
In cruising flight, holding a steady speed at a steady altitude, an airliner’s nose sits about 2.5 degrees above level, because a wing has to meet the air at a slight angle to make lift, like your hand tilted out a car window. Where the nose points and where the plane goes are two different things. Following the curve asks the plane for 0.00028 degrees of adjustment in the shortest slice of time any recorder stores, an eighth of a second — roughly nine thousand times smaller than the 2.5-degree pitch attitude the aircraft is trimmed to. The autopilot already spends the whole flight making small pitch corrections for gusts, thermals, and fuel burning off, and the curve’s correction is buried inside that routine workload, far below anything a person could pick out.
Pilots plan descents with a simple rule: three nautical miles of distance for every thousand feet of height to lose. From cruise at 35,000 feet that means starting down about 105 nautical miles out and riding a steady 3-degree slope, losing about 318 feet with every nautical mile. A rule that simple feels flat. But look at what the numbers in it are measured from. Start down 105 nautical miles out and both ends of that descent, the 35,000 feet at the top and the zero at the runway, are measured from the same curved ocean surface. “Altitude” means height above sea level, and sea level is not a flat floor. It is the surface of the ocean, wrapped all the way around the planet. So the 35,000 feet you are subtracting is not a height above a flat floor. It is a height above a floor that bends away beneath you the whole way down. And the 3-degree slope is measured against level where you are, which tips as you go. A straight ramp drawn against a curved floor is not straight in space; it bends with the floor. That is why the arithmetic never needs a curvature term. The curve is not missing from the rule. It is inside the word “altitude,” in every line of it.
If the rule of three really assumed a flat Earth, a descent from cruise would leave you about 9,700 feet too high at the runway, over a mile and a half up when you expected to land. Pilots arrive on target every day. The curve is built into what “altitude” means.
Most big runways project a radio beam called the glide slope, a ramp in the sky rising from the runway at about 3 degrees, which aircraft ride down to land. The experiment is straightforward. Fly toward the runway at a fixed altitude, note how far out you cross the ramp, and compare that distance against what a flat Earth predicts and what a spherical Earth predicts. The reasoning is sound. On a sphere, the surface you are following falls away from the runway’s tangent plane as you approach, so you meet the rising ramp a little sooner than flat geometry says you should.
Now size that difference, because everything turns on it. Ten nautical miles out is the edge of the beam’s certified range, so work there. Over ten miles a curved surface drops about 88 feet below a straight line — the same arithmetic as point 20, at a shorter distance. Allow for the atmosphere bending the beam slightly downward, which it always does, and the signal shrinks to about 76 feet. Now convert that into what the experiment actually measures: along a 3-degree ramp, 88 feet of height is about 1,700 feet of range, roughly a quarter of a nautical mile. That is the whole thing being hunted. Good test design. Wrong tool, as the next point shows.
Four separate things at that ten-mile mark are larger than the 88-foot signal.
The test was fair. The tool cannot see the answer. Neither side wins that one, and saying so is the honest reading.
Cockpits hold several generations of it, and all of them are still flying. Most small planes today carry the oldest kind: a spinning gyroscope, a heavy wheel that resists being tipped the way a spinning top resists being pushed over, with marks every five degrees and a needle about one degree thick. Airliners compute attitude from laser gyros, and on the glass screen the underlying number is rounded before it is ever drawn. And small planes that have been upgraded to glass screens, plus newer ones built that way, use MEMS chips, the same family of motion sensors as in a phone, with no spinning wheel at all: their software works out “level” from gravity and GPS, and since the bare chips would wander off level within minutes, the code re-levels them constantly. Different hardware, same answer on the screen: where is the nose, compared to local level. None of them was built to read thousandths of a degree, and none of them is a curvature detector.
A spinning gyro left alone keeps its axis pointed at a fixed spot among the stars. In a flying aircraft, two separate things carry the horizon away from that fixed axis. The Earth rotates underneath at up to 15 degrees per hour, and the aircraft rides around the curve at about 8 more at cruise. Left uncorrected, the little horizon in the case would fall out of agreement with the real one inside the first hour. Three things follow from that, and each stands on its own.
So the slow tipping the claim wants to see on the dial is precisely the tipping those vanes exist to cancel.
Both widths are drawn to the same scale. The green line is not thin for effect: at this size it is the true width of the movement the claim asks a recorder to show.
In the smallest time slice any flight recorder stores, an eighth of a second, the plane moves 0.00028 degrees around the curve of the Earth, and tips by that same 0.00028 degrees to stay level with it. And the recorder cannot write down just any number. Like a ruler with marks only so close together, it stores the nearest mark on its scale, and by regulation those marks sit 0.176 degrees apart, 600 times wider than the curve’s motion in that slice. The rules allow slower sampling too, as slow as one sample per second, and at that end the mark is about 80 times wider. Either way, a change smaller than one mark does not move the recorded number at all. The painted marks on the old instruments are wider still, 18,000 times the curve’s motion. It is a grain of rice measured with a truck scale.
Measured against the stars, the aircraft rotates about 8.1 degrees every hour, so a ten-hour flight really does turn it about 81 degrees, nearly a quarter turn. Nobody denies it and nobody hides it. The question is which instrument could see it, and that turns on the next point.
Walk around the whole planet carrying a spirit level. A quarter of the way around, you have tipped 90 degrees compared to the stars, yet the bubble never moved, because “level” tipped with you at every step. The attitude indicator is that bubble. It measures you against local level, and against local level the curve’s contribution is zero. Forever. A gyro with a million bits of resolution would read the same steady nose-up, because that is what the geometry says it should read. That is not a malfunction. That is the definition doing its job.
Deep in an airliner’s navigation system sit ring-laser gyroscopes, and nothing in one of them looks outside the aircraft. Each is a sealed loop of mirrors with laser light running both ways around it at once. Turn the loop and the two directions no longer fit the same whole number of wavelengths, so they come out at slightly different frequencies and beat against each other, a few beats every second. Count the beats and you have the rotation. Because the light never consulted the horizon, what comes out is rotation measured against non-rotating space itself, the frame the stars sit still in, and not against local level.
So the curve shows up right there in the raw output, every flight, sitting beside the Earth’s own spin: the spin contributes about 15 degrees per hour to that signal, and the curve-following contributes 8.1 more. The curve term is more than half the size of the Earth-rotation term. It is no whisper.
A Cessna carries no such instrument and does not need one. Its panel could never separate an eight-degree-per-hour signal from its own drift. That costs the argument nothing, because both panels are doing exactly what a round Earth predicts. The airliner’s gyro is built to measure rotation in inertial space, and it measures the curve every day. The Cessna’s attitude indicator is built to show the nose against local level, and the curve adds nothing to that reading, because local level turns with the aircraft. Neither instrument is failing. They are answering two different questions, and both answers came out the way the geometry says they should.
The navigation computer keeps a running answer to one question: which way is down, right now? Its gyros do not measure turns against the ground. They measure rotation against non-rotating space, so their total always contains two turns that come from the planet rather than from flying the aircraft: the Earth spinning underneath, about 15 degrees per hour, and the aircraft riding around the curve, about 8 more. The computer calculates both and subtracts them, many times each second. The curve term is nothing exotic. It is the aircraft’s own speed divided by the size of the Earth, the same arithmetic as point 1. Skip that subtraction for an hour and the computer’s “down” tilts 8 degrees off the real one. Then the motion sensors, which cannot tell gravity from acceleration, read a slice of gravity as fake speed, fake speed piles up into fake miles, and the computed position runs away. The math the claim says does not exist is arithmetic the box performs continuously, and you can name the exact term: speed over radius.
The computer’s sense of “down” is tuned to behave like a pendulum with a string about 4,000 statute miles long, one reaching from the aircraft all the way to the center of the planet. Why that length? A pendulum that long would always point at the center of the Earth no matter how the plane accelerated or turned, so software tuned to imitate it cannot be fooled by the plane’s own maneuvers. And the 84.4 minutes is that pendulum’s swing time, there and back. A pendulum swings more slowly the longer its string: a grandfather clock’s one meter of string takes two seconds to go across and back, and 4,000 miles of that string would take 84.4 minutes. Engineers call the choice Schuler tuning, and building it requires one number set in the hardware before the plane ever flies: the radius of the Earth. Nothing measures that number en route. Wrong-sized planet, lost airplane. Every ocean crossing is a test of it, passed daily.
First, know the shape of the file. A black box records like a spreadsheet: one column for each measurement, a fresh row written up to several times a second. The rules require 88 columns on modern airliners: speeds, altitudes, headings, accelerations, control positions, engine numbers, and yes, pitch. Pitch is recorded around the clock. But pitch is measured against local level, so the curve never enters that column at all (point 13). Even if it somehow did, the recorder’s steps are far too coarse to hold it (point 11). And the pitch column is a noisy place: light turbulence swings it half a degree, thousands of times bigger than the curve step.
In other columns of that same file, and in the navigation system feeding it: the rotation corrections, the position solution, the gap between pressure altitude and GPS height. Anyone with the data can pull it. It was never in the pitch column, for the same reason your bathroom scale does not record your height: an instrument writes down the one thing it measures, and pitch measures the nose against local level.
Press the point and most people are not picturing a data column. They are picturing a plane flying dead straight while the ground curves away below, and asking why nothing shows that.
Fly perfectly straight for 55 nautical miles and the ground drops 2,600 feet below you. After 250 nautical miles, 55,000 feet, more than one and a half times your cruise altitude. Across an ocean, the surface ends up about 1,400 nautical miles down, far outside the atmosphere, though not in orbit: orbit is a matter of sideways speed, not height. You would notice within the first hour. The picture is right that something huge would show. The picture is wrong about what a plane does.
“Flight level 350” does not mean 35,000 feet of rope-measured height. It means the altimeter reads 35,000 with 29.92 inches of mercury dialed in, which marks a layer of air at one particular pressure, and the layer hugs the planet like a contour line on a hiking map. The plane is not fighting to stay level over a falling horizon. It is riding a curved surface that does the following for free.
Websites like FlightAware display the altitude the plane broadcasts in its ADS-B message, and that message does not report every foot of change. It reports blocks: 25 feet each at best, 100 feet on older equipment. The line on your screen only moves when the plane crosses from one block into the next, so a plane drifting a dozen feet up and down draws the same flat line as a plane glued to its altitude. And where the graph looks smooth between points, the smoothness was drawn by the website, not flown by the plane. A flat line at that resolution is exactly what level flight looks like, on a round Earth or any other.
A flat Earth with a level plane would draw the same flat trace. That single line fits both stories, and pretending otherwise would be dishonest. What decides it is the machinery underneath the line, and every piece of it is sized to one particular sphere. The navigation platform subtracts ground speed divided by Earth radius, every second, or its position runs away within the hour. The flight computer plans routes as great circles, which is why a flight between two cities at the same latitude keeps changing heading the whole way, something you can watch on any tracker. And the same ADS-B message that carries the pressure altitude carries a second height, measured by GPS against a mathematical globe called WGS-84, one flattened by 21 kilometers between equator and pole. Every receiver decoding that message is decoding a height above a round Earth.
On the face of every altimeter is a small window showing a number near 29.92 inches of mercury, with a knob beside it. It has a name. It is the Kollsman window, after Paul Kollsman, who built the first accurate altimeter working in his attic in 1928. Jimmy Doolittle used one the next year to fly the first blind flight in history, taking off, flying a set course and landing without once seeing outside the cockpit. Now watch what the knob does. Turn it and the altitude needle moves while the plane sits still on the ground. It is not a sensor. It rotates the entire mechanism, changing the zero the instrument counts up from. The altimeter is openly, visibly adjustable, because it does not measure height. It measures pressure, and trusts a human to tell it what to compare against.
The altimeter does not sense height. It is a pressure gauge with its dial relabeled in feet, and it does not know where the ground is. When the needle reads 35,000 feet it is not saying the aircraft is 35,000 feet above anything. It is saying the outside pressure matches what 35,000 feet would be, if sea-level pressure were 29.92 inches of mercury. Weather moves that starting pressure around, so below the transition altitude, 18,000 feet in the United States, pilots dial in the local setting and refresh it about every hundred nautical miles. One inch of mercury is worth about a thousand feet, so ignore that for 150 nautical miles while the pressure falls 0.26 inches of mercury, and the needle reads 260 feet high while the aircraft sits 260 feet lower than it says. That is the whole argument, in the end. The claim that a level altimeter proves a flat Earth needs the altimeter to be measuring height above the ground. It is not, and every pilot knows it is not, because they spend the entire flight correcting it by hand.
The recorder’s smallest pitch step is 600 times bigger than the curve’s motion in one recorded slice at the fastest sampling the rules allow, and about 80 times bigger at the slowest. Turn that into a ruler: if the step were one millimeter, the finest mark on a school ruler, the curve’s share would be a line 1.6 micrometers wide, forty times thinner than a human hair. The cockpit dial is coarser still. Its painted marks sit five degrees apart, and against that spacing the curve’s motion is 18,000 times too small: in one recorded slice the needle tip would creep about half a micrometer, a distance smaller than one wavelength of visible light. Instruments that measure below a light-wave do exist. They fill buildings and detect gravitational waves. A painted bar on a dial is not one of them.
One thing this argument does not do on its own. A steady drift, however small each step, would eventually walk across a mark and show up — 78 seconds would do it. So size alone is the second objection, not the first. The first has not moved: the curve is not in the pitch column at all, because pitch is measured against local level, and there is nothing there to accumulate.
The message an airliner broadcasts has a name. It is ADS-B, Automatic Dependent Surveillance–Broadcast, sent on 1090 megahertz as what the standard calls an Extended Squitter. It is the signal FlightAware and every other tracker in point 22 is reading. Its format is public and it is not classified. It is written down by the Radio Technical Commission for Aeronautics, RTCA, in a document called DO-260B, with the message formats set out in its Appendix N. There are four generations: DO-260, then 260A, then 260B, which is the version the Federal Aviation Administration accepts, then 260C. Every field is listed. Latitude. Longitude. Barometric altitude. Geometric altitude. Ground speed. Heading. Vertical rate. Identity. Pitch is not among them, in any generation.
And that absence is specific. Air traffic control can ask an aircraft’s transponder for a report on how it is turning, and the aircraft answers with its roll angle, its track, its ground speed and how fast its track is changing. So aircraft do send attitude. They send roll, because a controller watching a turn can use it. Pitch is in neither that report nor the broadcast, and never has been. The reason is the one this page started with: pitch does not tell you where an aircraft is going. The same nose-up angle can mean climbing, holding level or descending, depending on speed and weight. A controller needs the flight path, so the message carries vertical rate instead.
So when someone says the curve should be visible in the pitch data from a flight tracker, one question settles it: name the field. There is nothing to name.
Every figure on this page is worked out in full, with sources, in the main reference.