What is new on Empirical Earth

What has changed on Empirical Earth, newest first. New claims answered, entries updated, and improvements to the site. The reference is maintained, and challenges that hold up get added here with a sourced response.

Two things worth knowing before you scroll. Everything here is linkable. Every one of the 200 entries, 223 glossary terms and 44 equations has its own address, so you can send someone the single claim you are arguing about instead of the whole site. Hover any entry and a copy-link button appears. And you can build your own list. The claims index lets you tick the claims you want, save them under a name, and export them as a cue sheet you can read off, paste into a chat, or print.

September 2026

New Baumgartner at 39 kilometers. On 14 October 2012 Felix Baumgartner stepped off a capsule at 38,969.4 meters and afterwards said he could see the curvature of the Earth. The new page treats that sentence as the weakest evidence he left. He was also clear that he could not see the whole Earth, and that to see it as completely round you would have to be much further out, and the page says so before anyone else does. What decides the question is not the view but the dip: the angle the horizon sits below level, which is 6.3 degrees at that altitude and grows with height in a way a single formula predicts. On a flat plane the dip is zero at every altitude, because a plane has no curve to fall away from you. The page also works through the lens objection, which is partly right and fails for three checkable reasons, and through the evidence that is not a photograph at all: a supersonic fall through air thin enough to permit it, a balloon that swelled as the pressure dropped away beneath it, and a sonic boom that reached the ground. Alan Eustace repeated the jump higher in 2014, and the dip came out a fifth of a degree larger, as the formula requires.
Updated The objection that the reception came after moonset, answered. The claim is that Baysinger recorded after the Moon had already set over Louisville, which would end the story. The tape dates itself. The Courier-Journal reported that the recording caught the entire text of President Nixon’s message, and NASA’s own Manned Spacecraft Center newsletter places that call at about 10:45 in the evening, Central time, which is a quarter to midnight in Louisville. The Moon set at nineteen minutes past midnight. A recording made after moonset cannot contain a call that happened thirty-four minutes before it. The page now carries the whole night in order, notes the three ways a wrong answer gets produced, and names the check that would overturn it: Baysinger’s own transcript against the Apollo air-to-ground transcript, line by line, timestamps read off both.
Updated The Moon set at 00:19, and the moonwalk ran until 01:11. The Baysinger page said the recording ends when the Moon set over Louisville. It now says when. From Okolona on the night of 20 July 1969 the Moon stood 17.7 degrees up when the hatch opened, 14.7 when Armstrong stepped off, and touched the horizon at 00:19 EDT, fifty-two minutes before the hatch closed again. Of the two and a half hours the astronauts spent outside, only the first eighty-three minutes were ever available to an antenna in Kentucky, and the thirty-five minutes the newspaper reports fits inside that window. It is the hardest test on the page because the constraint is not his: anyone copying the television broadcast had the whole moonwalk available, and nothing would make a fabricated tape stop at a quarter past midnight.
New Two skies, one rotation: a figure for the southern sky entry. Both panels are drawn from catalogue right ascension and declination rather than sketched. An observer at forty degrees north facing north, and one at forty degrees south facing south, at the same instant, with the star trails swept over six hours of sidereal time. The celestial pole sits forty degrees above the horizon in each case, because pole altitude equals the observer’s latitude. Polaris is drawn where it actually is, three quarters of a degree off the true pole rather than on it. On the right-hand side of each pole the northern stars climb and the southern stars descend, which is the whole argument: one Earth turning one way, seen from opposite sides of it. A disc has one center and can produce one rotation.
Fixed The newer pages were nearly invisible to search engines. Google crawled the Gleason map page in September and declined to index it. The reason turned out to be measurable: the pages listed in the footer strip are linked from eighteen or nineteen other pages, and every page added since that strip was written was linked from three or four. One page, the plain-language aviation piece, was linked from nothing at all. The strip has been extended to cover the whole site, and the worst cases have gone from three inbound links to eleven or twelve. Nothing about the pages changed; they were simply hard to reach, and a page that is hard to reach is one a search engine reasonably decides is not important.
New Deep dives — every long-form page in one list. The site has grown seventeen long-form pages alongside the main reference, and until now the only way to find them was a nav menu or an accident. They are now indexed on one page, grouped by what they do: the measurements you can take yourself, the cases followed all the way down, the aviation and radio material, and the instruments you can print. Each entry carries its own title, its own description and its length, so you know what you are opening. The page is generated from the pages it lists, which means it cannot go out of date: add a page and it appears there on the next build.
Updated Mercury cannot eclipse anything, and its own chart says why. The diagrams offered in support of the claim print the number that refutes it. Mercury’s orbit is tilted about seven degrees to the plane of Earth’s, so it sits well off the Earth-Sun line most of the year, crosses that plane at only two points, and can transit the Sun only in May or November, about thirteen times a century. The last was November 2019 and the next is November 2032. The entry also names the pattern those charts use: the inclination figures printed on them are correct, and they are heliocentric values measured against Earth’s orbit, relabeled as a geocentric diagram. The tell is the missing retrograde loops.
Updated Davis-Besse across Lake Erie, worked through. A nuclear plant photographed from a Michigan beach twenty-six miles across open water is offered as proof the lake is flat. The site carries two structures of known and different heights, which makes it a far better experiment than a single tower, because one atmosphere has to explain both and the difference between them cancels most of the uncertainty. In the photographs the cooling tower narrows going downward and stops at the waterline, so the widest part of the building is missing, and the smaller structure beside it is cut off at the same line. Two buildings of different heights amputated at a shared height is a horizon.
Updated Three people, one Sun: the worked example is now a real one. The page previously used a constructed trio. It now uses the observation as it was actually run, from Victor Harbor in South Australia, Phoenix and Bristol, Virginia, at one agreed instant in November 2025. Two of the three readings match the sky to within a degree. The third, in the southern hemisphere, is out by nearly twelve degrees, and the error is largest for the observer furthest from the middle of the flat map, which is where a projection has to stretch to fit.
Updated Geocentric is not flat. The diagram usually offered in support of geocentrism is Peter Apian’s woodcut of 1524, the one with Earth ringed by the spheres of the Moon, Mercury, Venus, the Sun, Mars, Jupiter and Saturn. The Earth at the center of it is a globe, drawn with land and sea. Ptolemy, whose system that diagram sets out, gave a chapter of the Almagest to arguing that the Earth is sensibly spherical and computed every table in the book on that basis. Aristotle said the same, and so did the scholastics who followed. Geocentric and flat are different claims, and the historical record holds no flat-Earth geocentric cosmology to appeal to: what those astronomers disagreed about was which body moves, not what shape the ground is. Producing a Ptolemaic chart for a flat Earth grants the point in dispute, because the shape is drawn at the middle of the exhibit.

August 2026

New Three people, one Sun. A new page on the cheapest decisive experiment available. Three observers far apart agree on one instant, face true north, and each writes down two numbers: the Sun’s bearing and its altitude. Six numbers in total, no equipment beyond a compass and a protractor. Two observations can be fitted by either model, because two measurements and two unknowns leave nothing over. Three over-determines the problem, and an over-determined problem is one a wrong model cannot satisfy. A globe reproduces all three readings from one Sun position. The flat disc, handed the same six numbers, must place its lamp at two or three different heights at once, because height equals ground distance times the tangent of the altitude, and every observer returns a different answer. The worked example gives 2,870 km and 8,060 km for the same moment, and a third observer for whom the Sun has set at all, which a lamp above a plane cannot produce.
New The “conceptual” flat-Earth simulator, audited in its own words. A new page on the open-source browser model that renders the sky on a flat vault and a globe and reports that both fit. The audit is of what the model carries: its own documentation says it holds no Earth radius, no astronomical units and no kilometers, so it computes directions for one observer and nothing else. Its sky data is imported from spherical-Earth ephemerides, and its eclipse demonstrations replay a globe-derived catalogue rather than predicting from a flat model. The sharpest part is a file it ships for another purpose: a real flight track carrying airspeed in knots alongside a central angle. Speed times time divided by angle is a radius, and it returns the Earth’s to within about two percent.
Updated Six accuracy corrections. The gravitational-wave entry conflated a bound with a measurement: the detectors are 3,002 km apart, which light crosses in 10.0 milliseconds, and the 2015 signal arrived 6.9 milliseconds apart, inside that bound rather than equal to it. A claim that no tropical cyclone forms within five degrees of the equator now names the storm that did, Typhoon Vamei in 2001. The lunar module’s engine is no longer described as having no oxygen to burn, since it carried its own oxidiser. Three Apollo minerals are now “never before identified on Earth” rather than non-existent here, because all three have since been found. The nautical mile is defined as exactly 1,852 meters, with the arcminute as its historical basis. And the orbital photography of the landing sites is credited to the mission that took it.
Fixed Two arithmetic errors on the math page. A worked radio-horizon example gave 58.7 miles where its own formula returns 52.3. And a geostationary orbit was described with its radius in kilometers and its altitude in miles inside one sentence, so the two numbers did not reconcile. Both are corrected, and a new automated check now recomputes every stated calculation on every build and refuses to pass if the checker itself stops working.
New The laser across the water. A new entry on the most popular modern flat-Earth field test: a laser fired across miles of open water that lands on its target with no apparent drop. The instrument is the problem. Divergence is the number nobody reports, and a good pointer spreads about a meter per kilometer, so at ten kilometers the beam is already ten to fifteen meters across, wider than the seven point eight meters of drop it is meant to detect. It will hit the target on a flat Earth or a round one. Aiming adds more slack, since a milliradian of tripod tilt moves the far end ten meters. Refraction bends the beam downward, which removes apparent drop rather than adding it. The same ten kilometers of water, read with a survey level and a row of targets at known heights, returns a sphere of about 6,371 km.
New The Lake Pontchartrain towers, turned into a measurement. The famous photographs of the transmission line are a target row that somebody else already built: twenty-four kilometers of identical, equally spaced towers standing in water. Because they are uniform, each one is its own scale bar, so the fraction of a base cut off by the water can be read from the image and inverted for the radius of the Earth. From a camera five feet up, towers at nine and a half miles with about twenty-four feet of base hidden return a radius near 6,250 km at standard refraction. The entry is equally clear about the limit: one tower and one assumed refraction coefficient carry about twenty percent uncertainty, and the way past that is to fit the drop against distance across a dozen towers, which yields the radius and the refraction together.
New The Gleason map, and the patent nobody reads. A new page walks through Alexander Gleason’s 1892 patent in full. The map flat-Earthers cite most often is a projection of a globe, and it is patented as a time calculator, which the document says in its own words.
New Larry Baysinger, and the arithmetic of a backyard Moon reception. The page on the Louisville broadcast engineer who received Apollo 11 directly from the lunar surface has been rebuilt around new technical work. The transmitter was half a watt, not the twelve commonly quoted, established four independent ways including the transistor on the RCA drawing. At that power the link cannot produce clean speech, only fragments surfacing out of noise, and the surviving tapes measure like fragments surfacing out of noise. The antenna question is now settled too: Baysinger’s own sketch shows a folded-dipole array in front of flat chicken-wire screens, which is a corner reflector rather than the horn the newspaper called it, and that pushes the link budget to its harder branch.
Updated Refraction, in the design standard that uses it. The microwave entry now carries the international recommendation for terrestrial links. The K-factor comes from the refractivity gradient, and a standard atmosphere puts it at four thirds. Towers are sized for two atmospheres rather than one, and building higher is not automatically safer, because over-correcting for one failure worsens another. The point worth keeping: a flat plane has no bulge to clear, so there would be no effective-radius factor to compute at all. The refraction term exists because the curvature it corrects is real.
Updated A sourcing claim removed. The site described its citations as primary sources. A count showed 253 of them point at Wikipedia, which is a tertiary source, so the claim was not true. Rather than leave it standing while the references are upgraded, the wording now says what is actually the case: a source for every figure and result, cited in full so you can check them yourself.
Updated Mercury did not cause the lunar eclipse. A claim circulated after the partial lunar eclipse of 28 August that pinned the shadow on Mercury, which reached superior conjunction the day before. The coincidence is real and the two events fall about thirteen hours apart. But superior conjunction is the moment a planet passes behind the Sun, so Mercury was about 207 million kilometers away, while a lunar eclipse requires the Moon opposite the Sun. The two were roughly 180 degrees apart in the sky. Mercury is also about one 384th of the Moon’s apparent width, which is the same speck we watch crossing the Sun during transits.
Updated The mirror-across-a-lake test, and why it usually decides nothing. Worked through on Lake Tahoe. Along the 35 km length an observer with eyes at 1.7 m needs the far mirror about 60 m above the water; across the 19 km width it is 13 to 16 m. The lake sits in a bowl where tens of meters of elevation are available almost anywhere, so both low over the full length gives no sighting, which the globe predicts, and a little height at either end gives a sighting, which the globe also predicts. Neither outcome decides anything unless the heights and the distance are published first.
Updated Gases are affected by gravity, and the meme answers itself. Added to the entry on gas under gravity. A helium balloon rises only because gravity pulls the heavier air around it down harder, and in orbit, where nothing has weight, a flame burns as a small sphere because hot air rises has switched off. Cold air drains downhill on calm nights, which is why frost strikes valley floors first and why Antarctic katabatic winds reach hurricane force. At the largest scale, Jupiter and the Sun are gas held in spheres by their own gravity, and a gas indifferent to gravity could not form a single star.
New Nobody floated: what August 12 actually proved. Two predictions were made in advance about the total eclipse of 12 August 2026. One said gravity would switch off for seven seconds. The other gave the minute the Moon’s shadow would touch western Iceland. The page is a record of which one was written down beforehand, which one could have failed, and which one did. Nobody floated, and the shadow arrived on schedule.
New Roll is broadcast. Pitch is not. The “planes never dip the nose” entry now answers the obvious comeback to name the field, which is that aircraft simply do not send attitude data at all. They do. Interrogate a Mode S transponder for its Track and Turn Report and it returns the aircraft’s roll angle, alongside track angle, ground speed and track angle rate. Roll is downlinked; pitch appears in no register used for surveillance, and in no generation of the ADS-B standard. The one axis a controller can use gets sent, and the one the claim demands does not, because pitch measured against local level carries nothing anybody needs.
Updated A drop figure corrected, and the comparison with it. The straight-line drop table said the ground falls 54,000 ft over 250 nautical miles, “the entire cruise altitude, twice over.” The drop is closer to 55,000 ft, and against a cruise of 35,000 ft that is a little over one and a half times, not twice. Corrected in the entry and on both aviation pages. The point the table is making does not depend on the error, which is exactly why it needed fixing.
Updated The eclipse page, corrected and sourced. The countdown and the text beside it disagreed, and both were wrong. Greatest eclipse is 17:45:54 UTC on the clock. Catalogues print 17:47:06, but that figure is in Terrestrial Dynamical Time, which runs about 73 seconds ahead of the clock on your wall. The countdown, the caption and the text now all read the same clock time, and the gap from the hoax’s 14:33 is three hours and thirteen minutes. The page also carries NASA’s answer on the record, that Earth will not lose gravity on 12 August because the total gravitational force is set by the mass of the Earth system; the fact that the account behind the original claim went offline in the first week of January, before the first fact-check was published; and the budget drifting from $89 billion in the first post to $89 million in the retelling.
New Airy’s failure, answered by name. The 1871 water-filled telescope is offered as proof that the Earth stands still. Airy found no change in the aberration angle, which is what a moving Earth predicts: the angle is set by the observer’s own velocity before the light enters the tube, so slowing it inside cannot alter it. What the experiment ruled out was aether-drag. The entry on parallax and aberration now says so under that name, and the glossary defines it.
New The mirror-flash test. Flashing a mirror across a bay and being seen at the far end is offered as a curvature test the globe fails. It is the same sightline as every other long-distance claim, with a brighter target. A flash still has to clear the bulge, so the arithmetic is unchanged: mirror height, eye height, and how much the air is bending that day. Raising the contrast beats haze; it does not lower the horizon.

July 2026

Updated The eclipse page now sets out a test you can run on the day. The hoax names a time: 14:33 UTC. Astronomers name a different one: greatest eclipse at 17:47 UTC, 45 km off western Iceland, totality 2 minutes 18 seconds. Those are three hours and fourteen minutes apart, so on 12 August one of them is simply wrong, and anyone with a clock can see which. The page also now concedes the part the hoax gets right, which is the date, and ends with what would show the page itself to be wrong.
Updated Does refraction ruin the shadow-angle measurements? A widely shared page argues that the coordinated multi-latitude shadow experiments fail because they assume sunlight travels in straight lines, and it supplies a table of refraction shifts: 4.4 meters over 12 miles, 110 meters over 62. Those figures are about right, and they are the right numbers for the wrong geometry. They describe a sightline running horizontally through the densest air for its whole length. A shadow angle uses light coming down through the atmosphere, and near noon it comes down almost vertically, where the correction is a few thousandths of a degree. Worked through for two stations 4,448 km apart it changes the derived circumference by 0.04 per cent — and it makes the Earth come out slightly larger, not flatter.
Updated The Chicago skyline calculation, answered on its own terms. A card circulates with the arithmetic worked out: observer at 200 feet, Willis Tower 53 miles away, 848 feet of it hidden by the curve. The sum is correct and the entry says so first. What is wrong is the assumption underneath it, that air does not bend light. Apply the standard refraction correction every surveyor uses and the hidden height falls to 672 feet; on the cold-water inversion days when these photographs are actually taken it falls to around 430. The photographer says the Earth is round and goes out only when an inversion is forecast, because the view is not there most nights — which is the test: a fixed geometry cannot produce a picture that changes every time you look.
New Interactive: the potato radius. Thirty-three real bodies plotted by diameter, from a 330-metre asteroid to Earth, with the rounded ones on one row and the irregular lumps on the other. The interesting part is that there is no clean line. Between Mimas at 396 km, which is round, and Vesta at 525 km, which is still a lump, both kinds occur, because ice gives way to its own gravity sooner than rock does. Drag the slider to any size and it tells you what is round there and what is not. Earth sits about thirty-two times past the band.
Fix One date, given three different ways. James Bradley’s discovery of stellar aberration appeared in the reference as 1727 in six places, 1728 in one and 1729 in six. All three trace to real events: observations began with Samuel Molyneux in 1725, Bradley continued alone from 1727, he announced it to the Royal Society in 1728, and the letter to Halley was published in 1729. Every mention now gives 1729, the publication, and the entry on parallax carries the fuller history so the date is anchored rather than merely consistent.
Updated Why the rover is always parked when the camera is running. A hoax argument holds that the lunar rover being stationary in almost all the television footage is suspicious. It is a consequence of one piece of hardware. The dish that carried television was aimed by hand and locked while driving, so video could only be sent from a standstill, while voice went out over a small near-omni antenna that worked while bouncing over rocks. Hand-aiming was enough because the Moon keeps one face toward us, so the Earth hangs almost motionless in the lunar sky, a target 1.9 degrees wide. Driving ten kilometers only shifts the aim by a third of a degree; parking on a slope shifts it far more, which is why re-aiming was a checklist item at every stop.
Updated Most entries now open by saying what the claim gets right. A cloud does weigh millions of pounds and it does stay up. Heat does flow from hot to cold. The Van Allen belts are real and the radiation in them is genuinely dangerous. The Orion heat shield really did lose material in a way engineers did not expect, and NASA said so. Freemasons really did hold prominent positions in early spaceflight. Project Echo was real and NASA really did bounce signals off giant balloons. Around 140 entries now begin with the true part, marked in green and stated without a “but”, before any measurement arrives.
Updated Water does find its level. Pilots never dip the nose. Rowbotham really ran his experiment and really saw what he reported. Zooming does bring detail back. Nobody has ever seen gravity. Most intercontinental internet traffic goes through undersea cable rather than satellites. Each of these entries now starts with the true part, stated plainly and without a “but”, before it goes on to the measurements. Conceding what is true costs nothing and is the honest place to start.
Site Six entries were sitting on their own. Treating the reference as a network rather than a list turned up six entries with no connection to any other, all of them recent additions: the two Panama Canal entries, the 1794 Moon observation, the Antarctic Circumpolar Current, the 2018 Antarctic crossing, and the hollow Earth. Each is now linked from the entries a reader would arrive from, and links back out. None of them were wrong, they were just hard to find.
Updated Eleven entries now say what they do not prove. A measurement of rotation is not a measurement of shape, and a laboratory result about gravity would come out the same on any shaped world. Foucault’s pendulum, the gyroscope, the Coriolis effect, star trails, sundials, Cavendish and several others now end by saying plainly what the measurement reaches and what it does not. In most cases a single reading in a single place shows only that the ground turns; what carries the argument to shape is the pattern across latitudes and the reversal below the equator. Stating the limit costs nothing and stops the argument being pushed further than the evidence goes.
Updated Three more entries laid out as chains. The entries on what “level” measures, on the abyssal plains being the flattest places on Earth, and on the three-body problem now walk through their argument a step at a time rather than listing evidence. All three turn on a single word being read the wrong way — level, flattest, solution — which is exactly what this layout is for. Every fact and source is unchanged.
Site A new way of laying out an argument. Most entries here give a list of separate pieces of evidence, any one of which stands on its own. Two entries now use a different shape: a numbered chain where each step follows from the one above it, so a reader who stops halfway is still holding something true. It suits an argument that takes one word apart rather than piling evidence up. The two Panama Canal entries are the first to use it, one on what “level” actually means and one on why two oceans forty-two miles apart have completely different tides.
Updated The 1958 encyclopedia that mentions a dome in Antarctica. Three new points in the cut-quotations entry. The passage is real and describes reconnaissance flights finding a dome about 13,000 feet high near 80 south. That is Dome Argus, and the Australian Antarctic Program gives it as 4,093 m of ice at 80°22′S, sitting on more than 3,000 m of ice above the buried Gamburtsev Mountains. An ice dome is a standard feature class and the plateau has three of them. The reference literature describes Dome A as a prominence whose elevation is visually not noticeable, because four kilometers spread over hundreds is a slope under one degree. It was mapped by airborne radar between 1967 and 1979. The circulating version also gets the source wrong: it is Americana 1958, not Britannica 1968.
Site Every claim now has its own page. Linking somebody to a single claim used to mean sending them the whole reference, which is a 3 MB page with 200 entries on it. Each claim now also lives at its own address, about 5 KB, holding the claim, what is measured, what would show it is wrong, and the sources, with a link through to the full entry. They open instantly on a phone, they preview properly when pasted into a chat or a comment, and there are 191 of them.
New Two Panama Canal entries. The canal puts an ocean at each end, 42 miles apart, under a Moon 239,000 miles away that cannot tell them apart. Their tides differ by more than fifteen times in range and about three hours in timing, because what sets a tide is the shape of the basin, not the strength of the pull. And the canal needed two separate definitions of sea level, because mean sea level at the Pacific end sits about eight inches above the Caribbean end. A 1926 survey paper notes the Canal Zone is the only place on Earth where bench marks on the sea level datums of two different oceans can be joined by a short line of precise levels. Both entries say plainly what they do not show: the locks are a water elevator over hills and prove nothing about curvature, and eight inches disproves a flat sea surface without proving a round Earth.
New Entry 34: the lumpy geoid picture. NASA released a geoid visualization on 15 July 2026 and it went round as proof that the Earth is a potato. The image is real and it is NASA’s. The caption printed with it says the heights are exaggerated by a factor of 10,000, and gives the true range: +85 m at Iceland to −106 m over southern India, on a radius of 6,371,000 m. That is 3 parts in 100,000. Shrink the Earth to a billiard ball and the entire range becomes 0.86 micrometers against a manufacturing tolerance of 0.127 mm, so a real billiard ball is a far worse sphere than the Earth is.
New Entry 127: the Antarctic Circumpolar Current. The largest current on Earth flows east around Antarctica and returns to the longitude it started from. It was measured continuously from 2007 to 2011 by 19 inverted echo sounders and 3 current-meter moorings on the seabed of the Drake Passage. Water can only come back if there is something to travel around, and a rim has an inside and an outside but no way around. The entry reports the disagreement in the literature rather than one tidy figure, and says plainly what the current does not show: it measures the shape of the water, not curvature.
Updated Moonbounce Doppler, worked out in full. Five new points in the moonbounce entry and a new equation in the appendix. A signal bounced off the Moon comes back off frequency by twice the transmitted frequency times the range rate over the speed of light, and the range rate is mostly the Earth turning under the antenna: 465.1 m/s at the equator, scaled by the cosine of the station latitude, the cosine of the lunar declination and the sine of the local hour angle. Two stations at different longitudes measure shifts of opposite sign at the same instant, and the zero crossing tracks each operator’s own local meridian. A stationary plane gives every station the same shift.
Updated The Blue Marble now carries the whole camera setting. The entry said “one frame of 70 mm film” and stopped there. It now gives the record: a Hasselblad 500EL with an 80 mm Zeiss Planar at f/2.8, on Kodak Ektachrome SO-368 reversal film, shutter fixed at 1/250 second, focus at infinity, from about 29,000 km. Two new points explain why those numbers matter: an 80 mm lens on that format is the normal lens, so Earth appears about the size the crew saw it, and the frame is one of a near-identical sequence with 22725 and 22726. The physical transparency also carries uniform emulsion grain, edge perforations and the Hasselblad reseau crosses, printed onto the film by the camera.
New Entry 97: the Flat Earth Dome Model. An interactive dome simulator circulates as proof of a working flat Earth. Its author, Walter Bislin, defends the globe and built it to demonstrate the opposite. Every position it draws is computed from NASA JPL ephemeris data in the heliocentric model and then projected onto the disc, and to connect those positions to an observer the app bends light along curves no known physics produces, differently for every observer and only at sea level. He added a warning to the top of the page because of this exact misuse, and lists by name the sites that removed his description and presented the model as their own.
Updated The twenty-year gap in the Michelson story is filled. The entry jumped from 1887 to 1979 and skipped the generation in between. In 1958 Charles Townes, who had just invented the maser, pointed two ammonia-beam masers in opposite directions on a rotating rack and reran Michelson and Morley with it, through 1958 and 1959. Null. He did it again with helium-neon lasers once lasers existed, which they did not in 1958. That is the real answer to the claim that Michelson proved a motionless Earth: the experiment has been rebuilt every time somebody invented a sharper instrument, usually by the person who invented it, with a Nobel waiting for whoever broke it.
Site Every formula card can be copied whole. The two buttons on each equation only ever handed over the formula, and dragging a selection across a card gave mangled text, because the formula is built from superscript tags and the symbol list is a grid. A third button on all 43 cards now copies the entire thing as clean plain text: the formula, the plain-language explanation, every symbol with its meaning, the worked example, the analogy, and a link back to the card. Paste it into a message and the other person gets the whole explanation rather than a line of algebra.
New Entry 125: Captain Cook and the 60,000 miles. The claim is that Cook sailed 60,000 miles along an ice wall and never found a way through. The figure is real and the entry concedes it in the first line: it is the total distance his ship logged in three years across the whole southern ocean, including Tahiti, Easter Island and two refits in New Zealand. His own Cape Horn passage gives his pace, 5,000 miles in 36 days. At that rate a lap of the Antarctic Circle takes 71 days on a globe and 489 days on the flat-Earth map, out of a voyage that lasted 1,103 days. And at his furthest south Cook wrote that the ice reached “quite to the Pole, or perhaps joins to some land” — the man cited as proof there is no pole said he thought there was one.
Updated Does it matter which way a gyroscope spins? Three new points in the gyroscope entry. For a free gyro measuring drift the answer is no, and that makes it a control experiment: reverse the wheel and you still read 15 degrees times the sine of your latitude per hour, because the number belongs to the Earth rather than the machine. For a gyrocompass the answer is yes, and the entry concedes it: reverse the rotor and the pendulous weight has to move from above the gyro to below it or the thing will not settle on north. And the sense of the drift reverses at the equator, clockwise north of it and anticlockwise south, which is the same sign flip the ring laser shows using a spinning brass wheel instead of a laser.
Site Diagrams now describe themselves to screen readers. Nine figures carried no text alternative, so anyone using a screen reader met them as blank space. Each now has a written description of what it shows, and the decorative search icon is marked to be skipped rather than announced.
Updated What an optical resonator actually is, and why it is frozen. The word was used without ever being explained. A cavity resonator is a rod of glass with a mirror at each end: the laser only builds up when a whole number of wavelengths fits exactly between them, the way a guitar string only rings at lengths that fit a whole number of half-waves. That makes it a ruler whose tick marks are wavelengths of light. It is cooled to a few degrees above absolute zero because glass expands with temperature by far more than the effect being looked for, and the ruler has to stay the same length while you read it. The rotation entry also now answers the model where the ground is still and the sky turns above it: a ring resonator does not measure the sky, it measures whether the box is turning, so a stationary Earth predicts zero everywhere.
Site 121 more terms now link to their definition. A third of the glossary had definitions written but nothing pointing at them. The first use of a term in each entry that uses it is now a link, across 72 entries. Terms that double as ordinary English words were left alone, and the rhetoric terms in the method primer stay unlinked because they are there to be browsed.
New A page for the rotation argument on its own. Twenty-three claims and answers about whether the Earth turns, gathered into one page you can hand to somebody mid-argument. Grouped by what kind of instrument does the measuring: mechanical, optical and quantum, astronomical, relativistic, then the consequences a spinning planet leaves on itself, then the objections that backfire. Five families of instrument sharing no principle and no century, all returning the same rotation rate and the same axis.
Updated The modern Michelson-Morley paper, answered from its own abstract. A 2003 Physical Review Letters paper by Müller, Herrmann and colleagues gets held up in debates as proof of a motionless Earth. The Michelson entry now names it, and answers it with the sentence printed in its own abstract: the resonators were subject to the Earth’s rotation. The planet turning is what swept the apparatus through every orientation across a year of data. Without it there is no experiment. The entry also explains why the follow-up had to build its own turntable: relying on the Earth’s spin left the direction along the spin axis unconstrained.
New The rotating optical resonator, with copy-and-paste equations. A sealed optical loop that reads the Earth’s spin with no view of the sky. The flat-Earth answer, that the instrument sees some rotation and the cause is assumed, is stated at full strength and then answered by the shape of the signal: it scales with latitude, it reverses when the box is turned over, and three rings at right angles recover one axis. A flat disc turning about its center would give every laboratory the same reading. Two new equations were added to the appendix with plain-text and LaTeX copy buttons.
New The silent aircraft with no moving parts. MIT really did fly a silent, propellerless plane on electric fields in 2018, and the entry concedes it fully. It is also a fixed-wing aeroplane: the wings lift it, the electric drive replaces the propeller by throwing air backward, it does nothing in vacuum, and it weighed 2.45 kg until it hit the gymnasium wall.
New The Exodus Effect and propellantless drives. A NASA electrostatics scientist with a real record says his company’s solid-state device produces thrust from electric fields alone, in vacuum, beyond one g. The entry concedes all of it: the credentials are genuine, the method excludes the obvious explanation, and the work is unproven rather than debunked. It still does not reach the conclusion drawn from it. A thrust quoted in multiples of g is measured against gravity, and so is every helicopter and hot-air balloon.
New The 1794 “star seen in the dark part of the Moon”. A genuine Royal Society paper is passed around as proof that a star was seen through the Moon. The paper says in: a point of light on the Moon’s own unlit portion. Nevil Maskelyne, the Astronomer Royal, published it, published a second witness beside it, added his own remarks, and noted that Aldebaran was being occulted that same evening. Nothing was buried, and the modern occultation record settles the physics.
New Entry 173: Colin O’Brady’s 2018 Antarctic crossing. The claim is that he never made the journey and that no GPS record exists. Both halves get an honest answer. His billing as the first solo unsupported crossing does not survive scrutiny, and there is no downloadable archive of his track. But his position was public and live throughout, a rival skier and a logistics operator logged it separately, and the investigators who criticize him place him on a named haul road at named coordinates. The geometry settles the rest: his start and finish are 82.6° of longitude apart, about 1,400 km through the pole on a globe, and roughly 27,800 km on the flat-Earth map.
Fix Printed sheets now carry their own definitions. A glossary link is worth nothing once a page is printed, so an exported set now ends with a “Terms used” section holding the full definition of every linked term in it. Four units that the site used but had never defined — the micrometer, the nanometer, the decibel and the watt per square meter — are now in the glossary and linked where they appear.
New Send someone your whole set as one link. A set of claims is saved in your own browser, which is no help when you want it on your phone or in somebody else’s chat window. The share link carries the set itself, so opening it anywhere rebuilds the same list, named and ready. Links that have gone stale load the claims they still recognize and drop the rest.
New Build your own talking points. Open the claims index, tick any claims you want, and save them under a name. Reopen the set later and add to it. Export it as plain text to paste anywhere, or as a page that opens formatted in any browser and prints itself on white paper. There is a switch to include each claim’s “what would change my mind” line, and the sets can be backed up to a file so clearing your browser cannot lose them.
Site Link to any single claim, term or formula. You do not have to send anyone the whole site. All 189 entries, 216 glossary terms and 41 equations have their own address, and every entry carries a copy-link button, so the one claim under discussion can be handed over on its own.
Site Every page now lists every other page. A companion-pages directory in the footer of the reference and of each companion page, so arriving on one of them from a search result no longer dead-ends.
Site Share any page in one click. A copy-link, X and Reddit control in the footer of every companion page, filled in from whichever page you are on.
New The Black Swan oil rigs. The viral clip of two California oil platforms filmed whole from miles offshore, added to the “you can see too far” entry. The shot is taken inches above cold water, the setup that bends light downward the most, and the same rigs have been filmed on other days with their bases cut off.
New Clouds that seem to pass behind the Sun. The viral “local Sun” claim, explained as an overexposure illusion you can reproduce with a lamp and tracing paper.
New The Earth did not fly away while Apollo was in transit. Why a moving Earth did not strand the crew: the ship, the Moon and the Earth all share the same orbital speed, so it cancels.
New The Antarctic Treaty is an open-inspection pact. The treaty orders rival inspection of every base and does not ban travel, the opposite of a secrecy pact. A “read the treaty” link was added to Resources.
New Fram2: a private crew orbited directly over both poles. The 2025 first crewed polar orbit, filmed from overhead and independently tracked from the ground.
Updated Blue Marble. Added the April 2026 Artemis II Earth photos, “Hello, World” and “Earthset,” the newest human-taken full-disc images of Earth.
New The path to space. A dozen-milestone timeline of the human climb to orbit and the Moon, from Sputnik to the crewed Artemis II flyby.
New The phone in your pocket already runs on a round, spinning Earth. GPS, its relativity correction, and the whole-globe compass model, none of which has a flat-plane version.
New Run it yourself: five checks in the curvature calculator. Added to the 8-inch-rule entry, tied to Walter Bislin’s Earth-curve tool, showing where the globe and flat models diverge.
Changed Feedback is now Challenge us. A standing invitation to prove an entry wrong with a measurement, a photo, or a source. Well-posed challenges get added here.
Updated Artemis, current through 2026. The Moon-landing entries reflect the crewed Artemis II return of April 2026.
Site Faster, and friendlier on phones. Edge caching and deferred rendering for speed, a simpler five-item menu, a reworked front page, and a mobile layout that turns the wide measurements ledger into readable cards.

Earlier

Launch The reference went live with more than 160 sourced entries across shape and curvature, gravity and the air, rotation, flight and navigation, the sky, radio, space and satellites, the Moon landings, the boomerang proofs, history and the cosmos, and hands-on tests you can run yourself.