The strongest arguments against a level earth
Ten objections that would break the flat reading if any of them holds. Each is set out in its own terms, at full strength, and then measured against what the flat side can actually say in reply.
Steel-manning means stating an opponent's case better than they do, which is the only way to find out what your own position can carry. Each objection below is given the strongest form I can construct, and the mainstream account is described as its practitioners do.
Each entry ends with a verdict, and some go against the flat reading. Where that happens I say so rather than reaching for a rescue.
Full answer — it accounts for the observation as well as the globe does. Partial answer — it explains part and leaves part unexplained, where most of these land. No answer at all — nothing survives the measurement.
1. Gravity and the Cavendish-type experiments
At full strength. Masses attract one another with a force falling off as the square of distance; the local downward direction points toward the centre of mass. Henry Cavendish's 1797–98 experiment measured the attraction between lead spheres with a torsion balance and derived from it a value for a density for the earth; later instruments repeated it, and the constant of proportionality remains the least precisely known of the fundamental constants. Surveyors also measure the deflection of the vertical: plumb lines a hundred kilometres apart do not hang parallel, and the convergence matches prediction.
The flat reply is that density and buoyancy explain falling without a pull toward a centre, since objects sink or rise according to what they displace: a genuine alternative description of a dropped object. It does not answer the torsion balance, which uses inert lead balls and no falling; nor the convergence of plumb lines; nor free-fall equivalence tests. Verdict: partial answer on falling, no answer to the balance and deflection measurements.
2. Circumnavigation and shipping routes
At full strength. Magellan's expedition left Spain westward in 1519 and one ship returned in 1522 from the east, its surviving accounts showing the crew's calendar a day behind the shore. Since then the traffic has been continuous: container ships on fixed rotations, canal records at Suez and Panama, crews who begin and end tours at the same port. A westward course returning from the opposite direction is ordinary commerce now, not an expedition.
A closed journey proves nothing about shape by itself: you can walk a circle on a table. So the version that says you would fall off an edge is no argument at all, and the flat reply answers it fully. What it does not touch is the distances, schedules and star altitudes logged along the way, treated under objection 8. Verdict: full answer to the round trip; partial and map-dependent on the distances.
3. Time zones and the 24-hour sun in Antarctica
At full strength. Local time advances one hour for every 15° of longitude, which is what a rotating earth gives. And on the globe model, poleward of about 66.5°S the December solstice sun never sets: it circles the horizon for twenty-four hours a day for months, as polar expeditions have reported at widely separated longitudes.
Time zones discriminate nothing: a circling spotlight sweeps 15° per hour as readily as a spinning ball does, so that half of the objection should be dropped. The 24-hour sun is harder. The flat reply widens the sun's circling path in southern summer so it stands above all Antarctic latitudes at once, but that is a programme rather than a mechanism, and it must keep the sun's apparent size and angular speed constant year round. Verdict: full answer on time zones; no answer yet on the Antarctic midnight sun.
4. Satellite imagery and GPS behaviour
At full strength. GNSS satellites orbit at roughly 20,000 km, broadcast time-coded signals from atomic clocks, and a receiver hearing four or more solves for latitude, longitude, altitude and time; the solution works mid-ocean and in the south at the same accuracy. Amateurs receive weather-satellite pictures directly with inexpensive radio gear, at pass times published by the operators, and dish pointing data places the same satellite on opposite bearings in the two hemispheres.
The flat reply is that the receiver is really listening to a terrestrial network. Ground-based radio navigation predates GNSS and is a real precedent, though its weakness was coverage and accuracy far from the northern hemisphere — an honest comparison, since GNSS works there at full accuracy. The reply does not explain amateur reception of satellite pictures at predictable pass times, the altitude solution, or the published pointing data, where the same satellite sits on opposite bearings from opposite hemispheres. Verdict: partial on the idea of a ground network; no answer at all to pass predictions and the altitude solution.
5. The Coriolis effect
At full strength. On a rotating earth, moving air and water are deflected to the right in the northern hemisphere and to the left in the south, in proportion to the sine of the latitude. Gunnery and ballistic tables carry the term, ocean circulation is described with it, cyclones rotate opposite ways in the two hemispheres, and cyclones very rarely form within a few degrees of the equator, where the term vanishes.
The bathtub-drain version should be retired: a basin is far too small, and residual motion dominates. The strong version is the sign rule and the latitude factor. The flat reply is that winds and currents are local and need no rotating globe, which describes patterns but not why the sign flips across the equator, nor why inertial navigation corrects for a latitude-dependent earth-rate term. Verdict: partial on describing circulation; no answer to the sign rule or the navigation term.
6. The Foucault pendulum
At full strength. In 1851 Léon Foucault hung a pendulum in the Panthéon in Paris and showed that its plane of swing turned relative to the floor beneath it, at a rate of twenty-four hours divided by the sine of the latitude. Repeats in many cities match that relation.
The weak flat reply appeals to a driven pendulum or stray air currents, which fails against a long, isolated suspension. The strong reply is different and, I think, right: a pendulum's plane stays fixed in inertial space, and a rotating earth of any shape would turn beneath it, so the latitude term demonstrates rotation rather than figure. And a rotating disc produces the same latitude dependence. Verdict: partial answer, and this objection should not be used as a shape argument at all.
7. The Eratosthenes measurement
At full strength. In the third century BCE Eratosthenes compared the noon shadow at Syene, near the Tropic, with the shadow at Alexandria and found about a fiftieth of a circle between them; with a distance usually given as roughly five thousand stadia, he obtained a circumference on the order of 250,000 stadia. The figure is reported by later writers rather than in his own surviving words, and the length of the stadion is uncertain, so any modern conversion carries a margin. The method works because the sun's rays are treated as parallel and the two cities lie on a curved surface.
Two stations can always be fitted by a low sun at a chosen height, so the two-station version is genuinely weak on both sides. The over-determination comes with a third and fourth station at other longitudes, which forces the low-sun model to make the sun's apparent size change with an observer's distance from it — something telescopes and cameras do not show. Verdict: partial against two stations; a full answer against the simple near-sun model once four are used, which is why we recommend four on Try it yourself.
8. Aircraft routes and flight times
At full strength. The shortest path between two points on a sphere is a great circle, which is why transpolar and transoceanic tracks look curved on a Mercator chart. Routes between South America, Australia and southern Africa are flown daily, with published block times, fuel loads and crew rosters built on them.
Two parts of the flat reply hold: a curved track on a flat map is a projection artefact, and southern routes do get strong jet-stream assistance. The second does not shorten distances, though. On the standard azimuthal-equidistant map, the gaps between those southern continents become far longer than the flight times allow, so the flat reading must either move the continents — abandoning the standard map — or leave the distances unexplained. Verdict: partial on route shape; no answer on southern block times while the standard map's distances stand.
9. The 1968 lunar photographs
At full strength. In December 1968 Apollo 8 carried three crew around the Moon, whose crew took the photograph known as Earthrise. Five further crewed landings followed to 1972. Samples went to laboratories in several countries, and retroreflectors left at three sites are still used by observatories that fire laser pulses at them and time the returns, tracking the Earth–Moon distance to within centimetres.
Images can be fabricated and so can samples, which makes the photographic version of this objection weak on both sides. Laser ranging is a different kind of claim: a pulse sent from the ground, detected in return and timed at independent observatories implies a reflector placed on the Moon. The honest limit is that no reader can check it personally, so what remains is a judgement about institutions rather than a geometric test. Verdict: no answer at all to the ranging; the photograph-only version gets a full answer from "images can be faked" and should not be the argument made.
10. Professional consensus, and its limits
At full strength. Geodesy is a working profession. National agencies publish ellipsoidal datums, every GNSS fix is computed on one, and canals, tunnels, pipelines, bridges and shipping lanes are surveyed and built on the same basis. Thousands of practitioners use the curvature daily, and their work is checked by outcomes: a tunnel that meets in the middle, a platform towed to position, a navigation solution that arrives where it should.
Consensus is evidence about what people believe, and most practitioners apply inherited models rather than re-testing them, so as an argument from authority it settles nothing — which is where the flat side's institutional reply has real force. Its strength is that it is partly checkable in a day, by levelling between published benchmarks or watching a receiver's own solution behave. Verdict: partial — strong as indirect evidence, decisive for nobody who has not run a measurement themselves.
The verdicts at a glance
| Objection | What the flat reply can say | Verdict |
|---|---|---|
| Gravity and torsion balances | Density and buoyancy describe falling | Partial |
| Circumnavigation | A closed loop needs no edge | Partial |
| 24-hour Antarctic sun | A widened circling path | No answer yet |
| GNSS pass predictions | A terrestrial network | No answer |
| Coriolis sign rule | Local wind and pressure patterns | Partial |
| Foucault pendulum | Rotation, not shape | Partial |
| Eratosthenes, two stations | A chosen sun height fits both | Partial |
| Southern flight times | Jet streams, not distances | No answer |
| Laser ranging to the Moon | Nothing that survives | No answer |
| Professional consensus | Inherited models, not re-tested | Partial |
The pattern is consistent: the flat reading is strongest where an observation is close-range, optical and repeatable by one person with ordinary gear, and weakest wherever the claim is navigational, orbital or timing-based. Test it yourself with the sightline and shadow measurements on Try it yourself, which come with an arithmetic prediction you can write down in advance.
This page is mostly the mainstream case, stated by its own advocates. The standard account of the earth's figure rests on independent lines of evidence — geodetic levelling and triangulation, GNSS timing geometry, the behaviour of pendulums and gyroscopes, seismic tomography, and orbital imagery — and it can state each one with numbers that match observation. Where an objection is marked a full or partial answer above, that is the mainstream position being taken seriously, not dismissed.
This page presents the flat-earth interpretation as its subject matter. It is a belief-based reading of the observations, not a scientific finding — see About for how we frame claims and for the mainstream sources.