Any Way You Fall

Drill a straight tunnel clean through the Earth, step in, and let go — no air, no friction, nothing but gravity. You reach the far side in about forty-two minutes. Drill a different tunnel, one that misses the centre entirely, and drop again. Same forty-two minutes. Every straight tunnel keeps the same clock. Drag the mouths of the tunnel below and see.

Companion film · 2:36 · every straight tunnel keeps one clock. Five stones drop from a single mouth down five different chords and — in a uniform Earth — arrive together (the verifier sample grid has a spread of 2.98 × 10⁻¹³ min); switch to a realistic PREM Earth and the tie breaks — the diametric tunnel, the longest of the five, is now the fastest at 38.18 min. The fall is a live Runge–Kutta integration of the same physics the instrument below runs (research/gravity-tunnel/verify.mjs, 8/8); every on-screen number is recomputed from it and gated offline by film-facts.mjs. The soundtrack is synthesised from the fall’s own clock — one unison pitch while every chord shares a period, a beating spread once they don’t. Build: research/gravity-tunnel/film/.

Interactive Earth cross-section and gravity tunnel

Surface distance
—
Transit (one way)
—
Clock
0:00
uniform density

Gravity felt on the way down: in a uniform Earth it climbs straight from zero at the centre to full strength at the surface.

The forty-two minutes is not a coincidence and it is not a heavy calculation once you see it. Inside a uniform ball of rock, only the mass below your feet pulls on you; the shell above cancels out (Newton proved this). That inner mass shrinks as you fall, so gravity weakens in exact proportion to how deep you are — pulling hardest at the surface, vanishing at the very centre. A force that pulls you back in proportion to your distance from the middle is the definition of simple harmonic motion: the same law as a mass on a spring, or a pendulum in a small swing. You don't fly past the centre and escape; you slow, stop exactly at the far mouth, and would fall back forever. And the period of that oscillation depends only on the planet — not on where you started, not on how long the tunnel is.

That last clause is the strange one, and it is worth dwelling on. A tunnel from pole to pole is nearly 12,700 km long. A tunnel between two cities a short hop apart is a stubby chord near the surface. The stone in the long tunnel has vastly farther to travel — but it also falls through a much steeper drop and reaches a much higher speed at the middle, and the two effects cancel exactly. Set the spread to anything you like above and read the transit time: it does not move. This is a cousin of the tautochrone — the curve on which a bead takes the same time to reach the bottom no matter where you release it. Here the equal-time property lives not in a special curve but in gravity's own shape inside a uniform world.

The real Earth is faster — and breaks the spell

Switch the model above to Realistic. The Earth is not uniform: it has a small, ferociously dense iron core. Because so much mass is packed near the middle, gravity does not fade away as you descend through the mantle — it actually holds near 9.8 m/s² almost the whole way down, and only collapses once you enter the core. Stronger pull for most of the trip means a quicker fall: the pole-to-pole tunnel now takes about 38 minutes, not 42. And the beautiful equal-time coincidence quietly dies — a shallow tunnel and a deep one no longer share a clock. The tidy answer was a gift of the uniform assumption; the real planet keeps its own, messier time. Both numbers below are integrated live in your browser from the same physics, and both are re-checked offline against the standard Earth model.

The check

The interactive readouts are computed live in your browser as you drag. The table below gives results computed offline in research/gravity-tunnel/verify.mjs (8/8 passing). The offline run integrates the equation of motion by Runge–Kutta and, for the realistic case, reads density from the PREM model (Dziewonski & Anderson, 1981), checking first that its density profile returns Earth's true mass and surface gravity before trusting any fall time.

Uniform Earth, diametric fall (analytic π·√(R³/GM))42.17 min
Uniform Earth, spread of transit over all tunnels d = 0…0.95R2.98×10⁻¹³ min
PREM density → integrated Earth mass5.973×10²⁴ kg
PREM → mean density / surface gravity5514 kg/m³ · 9.82 m/s²
Realistic (PREM) diametric fall — matches Klotz 2015 (38.2 min)38.18 min
Realistic Earth, transit range over tunnels d = 0…0.9R38.18 → 41.51 min

What's idealised, stated plainly. A real tunnel would face severe engineering difficulties: the mantle is semi-solid rock, the outer core is liquid, and the inner core is solid, with temperature and pressure increasing with depth. Any actual shaft would also need to cope with the planet's rotation (the Coriolis acceleration vanishes for motion parallel to the rotation axis, including along offset parallel tunnels; other forces can still press a stone against the wall). Air resistance is neglected; with air, a capsule released from rest loses energy to drag and cannot reach the far mouth without added energy. The point is not an engineering proposal but a clean question with a clean, checkable answer. The PREM figure assumes a non-rotating tunnel and treats the density profile as exact.

Corrections

Corrected 2026-10-14: The page called Earth's interior molten. The mantle is semi-solid rock, the outer core is liquid, and the inner core is solid; found by the assay line, 2026-10-14. USGS.

Corrected 2026-10-14: The page and film said only a pole-to-pole tunnel escapes Coriolis deflection. Motion parallel to Earth's rotation axis has no Coriolis acceleration, including in offset parallel tunnels; other forces can still press a stone against a wall. The film source is corrected; the existing MP4 still contains the old wording; found by the assay line, 2026-10-14. Simonič, Eq. (1).

Corrected 2026-10-14: The page said air resistance merely made the trip take longer. With drag, a capsule released from rest cannot reach the far mouth without added energy; found by the assay line, 2026-10-14. Concannon and Giordano, sections II.A and II.B.

Corrected 2026-10-14: The Realistic readout showed 16666.7 min at 40° and 150°. It now shows 41.8 min and 38.5 min, respectively; found by the assay line, 2026-10-14.

The claims record 129 claims re-read against their sources, 14 October 2026

Written 2026-07-07. Claims re-read against their sources on 2026-10-14: 129 checked, 116 confirmed, 12 wrong, 0 unverifiable, 1 true when written and overtaken since (dated, not corrected). By the assay line (research/assay-line/cited-1014b/), Codex fixer directed by claude-assaying-boyle-1014; source record by Codex, checked by Codex.

This was the directed repair pass for any-way-you-fall at 389e81e2f5: the fixer re-read the evidence for all 13 triage items, reproduced the browser failures, made the supported source and page changes, and carried forward the 116 CONFIRMED claims re-read by the Codex checker in research/assay-line/cited-1014b/any-way-you-fall/N2-codex.md, with their original sources and notes. The 12 in wrong count eight WRONG and four MINOR lines. The film source is corrected, but its existing MP4 still contains X11 because this pass prohibits rendering or re-encoding. This record does not claim the MP4 was repaired.

This record is long (129 claim lines and 13 findings), so it is not part of the page until you open it. It loads here now if your browser runs scripts; otherwise, or to keep it, read the whole record as plain text.

The assay office: what a record is, and every layer re-read so far.