The Width the Moon Keeps
Why does the moon look so much bigger when it sits on the horizon? It doesn't grow. Its angular size is the same all over the sky — and at the horizon it's actually about 1.6% smaller, because you're sitting roughly one Earth-radius farther from it. The atmosphere isn't a lens; near the horizon it flattens the disc, not swells it. The enormous horizon moon is real — but the size is added by you.
A full moon climbs over the rooftops looking vast — a coppery dinner-plate you feel you could touch. Hours later, overhead, it's an ordinary button. Everyone has seen it; most explanations you'll find say the air near the horizon magnifies it, like a lens. That explanation is wrong, and you can prove it wrong without leaving this page.
The moon's apparent width is set by one number: how far your eye is from it. Run that number and the horizon moon comes out smaller, not bigger — you stand an Earth-radius nearer to it overhead. So the swelling can't be in the light. It's in the seeing. Below: measure the discs, run the geometry, and hold a coin to the sky.
1 · Two moons, one width
Here is the horizon moon (over a skyline) and the high moon (in empty sky). They are drawn to the same pixel diameter — guaranteed by the code, not by eye. Turn on the calipers to check. Whether the left one looks bigger to you is the illusion itself; on a small bright screen it fires weakly, far weaker than against a real night sky.
Switch to the bare backdrop and the two discs snap to obviously equal — the context was the whole trick. The glass of the sky did nothing; the skyline did. Which is the first clue to what's really going on, at the bottom of the page.
2 · Run the geometry
Slide the moon from the horizon to straight overhead. The page recomputes, live, the distance from your eye to the moon's centre and the angle the disc subtends — from published constants (Earth radius 6371 km, mean moon distance 384,400 km, moon radius 1737.4 km). Watch the horizon value come out the smaller one.
Two facts fall out. One: from horizon (0°) to zenith (90°) the disc grows about 1.6% — the horizon moon is the smaller one, the exact opposite of what you see. Two: drag the orbital distance and the disc changes by a full ~12% between perigee and apogee — eight times the horizon effect — and almost nobody ever notices that. The eye is a poor meter for absolute size; it needs something to compare against.
3 · Hold a coin to the sky
The oldest check, and you can do it tonight. The moon is about half a degree wide. Work out how big an object, held at arm's length, just covers it — then cover the giant horizon moon with the same object you'd use overhead. It fits both. (Arm length varies, so set yours.)
It works out to a pea, the head of a match, the width of a pencil — well under a centimetre. Your own pinky-nail (about a centimetre across) blots the moon out completely, horizon or zenith. People expect to need something far larger, which is the illusion leaking into the guess.
The check
The illusion is real; the magnification is fictional, and the page proves it from numbers, not assertion. Recomputed live above and offline in research/moon-illusion/verify.mjs (15/15 PASS):
- Eye→moon distance is the law of cosines in the triangle (Earth-centre, you, moon):
zenith
= d − R = 378,029 km, horizon= √(d²−R²) = 384,347 km. The horizon moon is 6,318 km farther. - So angular diameter shrinks at the horizon:
0.5267° → 0.5180°, a 1.64% decrease — opposite in sign to the illusion. - The real orbital swing is
~12%(perigee 33.5′ vs apogee 29.9′) and goes unnoticed — the eye has no absolute scale. - The covering object at 700 mm is
~6.4 mmwide, changing by 0.1 mm across the whole sky. - Atmospheric refraction (~34′ at the horizon, standard atmosphere) lifts and flattens the disc into an oval — it never enlarges it.
The on-screen moons are drawn to identical pixel diameters by construction. The two-moons demo proves the discs are equal, not that your eyes must fire the illusion on a small screen.
What's still genuinely unknown
If the light doesn't change, the moon illusion is a fact about perception — and after two thousand years, why it happens is still not settled. The leading accounts compete; none explains every result. We state them as open, not decided:
- Apparent-distance / flattened sky-dome theory
- The sky reads as a flattened dome, so the horizon seems farther; size constancy (Emmert's law) then scales the same retinal image up. Its flaw: the size–distance paradox — asked directly, people call the bigger horizon moon closer, contradicting the premise. (Ptolemy; Kaufman & Rock, Science, 1962.)
- Angular-size / relative-size contrast
- Terrain and distant objects beside the horizon moon make it loom large by comparison; overhead it floats in empty sky and shrinks. (Restle, Science, 1970.)
- Oculomotor micropsia
- Gazing into empty sky relaxes the eyes toward a resting focus, shrinking the high moon.
Ross & Plug's survey (The Mystery of the Moon Illusion, Oxford, 2002) concludes no single theory fits all the data. The geometry is closed; the psychology is open — and we don't pretend otherwise.