Interactive companion — This proof has two interactive companions. They render the sphere and the disk side by side as the same cosmos in different coordinates: The Witness Disk (flat-earth view from above) · The Three Witnesses (sphere view).
Introduction
One of the longest-running debates in cosmology pits two pictures of the Earth against each other: a sphere viewed from the outside, or a flat disk with celestial bodies in their circuits above it. The argument usually proceeds as if these were rival empirical claims, each with its own evidence to be weighed.
This framing is too crude. There are not two empirical claims here. There are at least two distinct philosophical claims bundled under the label flat earth, and the mathematics treats them very differently.
Claim A: The flat-disk representation is a coordinate description of the same underlying geometry as the sphere. The “ice ring” is the South Pole stretched into a circumference. The “circuits” of celestial bodies are orbits expressed in azimuthal coordinates. The firmament cosmology is correct, and it is mathematically equivalent to the spherical model expressed in different language.
Claim B: The flat-disk representation describes a geometrically distinct reality. The Euclidean flat metric — distances measured with a ruler on the page — is the true metric of Earth’s surface. The sphere is wrong about the underlying geometry, not merely about the labels.
The purpose of this post is to prove Claim A rigorously. Every observation made on Earth — horizon dropoff, lunar eclipse shadow, the elevation of Polaris with latitude, the existence of two celestial poles, GPS triangulation, antipodal day and night, southern hemisphere flight times — can be reproduced exactly on a flat disk equipped with the correct projected metric. The proof is not approximate. It is mathematically guaranteed by a theorem of differential geometry.
Claim B is a separate question, and a much harder one. I will deal with it in a later post, where the question becomes whether a flat earth can be made consistent with observation using only the literal Euclidean metric and the kind of new physics that would have to come with it. The answer there is different, and the bottlenecks are real.
The disk-with-firmament and the sphere are not in conflict. They are two coordinate systems describing the same geometry.
The Mathematical Setup
Let me state the result precisely. We have two surfaces:
- The sphere S² of radius R, parametrized by latitude φ ∈ [−π/2, π/2] and longitude θ ∈ [0, 2π).
- The disk D of radius πR, parametrized by polar coordinates (r, ψ) where r ∈ [0, πR] and ψ ∈ [0, 2π).
We define the azimuthal equidistant projection Φ from sphere to disk:
r = R(π/2 − φ), ψ = θ
This map has two essential properties.
It is a bijection. Every point on the sphere has exactly one image on the disk, and every point on the disk has exactly one preimage on the sphere. The North Pole (φ = π/2) maps to the disk’s center (r = 0). The equator (φ = 0) maps to the disk’s middle circle (r = πR/2). The South Pole (φ = −π/2) maps to the disk’s entire outer boundary (r = πR) — a single point unrolled into a ring of zero metric width.
It carries a unique metric. The sphere’s intrinsic metric is ds² = R² dφ² + R² cos²(φ) dθ². Substituting r = R(π/2 − φ), so dφ = −dr/R and cos(φ) = sin(r/R), we obtain on the disk:
ds² = dr² + R² sin²(r/R) dψ²
This is not the Euclidean flat metric of the page. The Euclidean polar metric would be ds² = dr² + r² dψ². The projected metric agrees with the Euclidean metric only near the center, where sin(r/R) ≈ r/R, and increasingly diverges as r approaches πR. At the equator (r = πR/2), the ratio is sin(π/2) / (π/2) ≈ 0.637 — a thirty-seven percent discrepancy with Euclidean.
This is the load-bearing fact. The disk that reproduces sphere observations is not metrically flat. It is the sphere drawn on a page, with the curvature traveling with the points.
Why This Is Sufficient
The reason this construction proves Claim A is a theorem of differential geometry: physical observations on a Riemannian manifold are invariant under diffeomorphism. Said plainly: if you have a smooth, invertible map between two manifolds that carries the metric correctly, then every geometric measurement — distances, angles, areas, geodesics, parallel transport, all of it — gives the same numerical result in either coordinate system.
The map Φ is smooth, invertible, and carries the metric. Therefore every observation about distances, angles, paths, and times on Earth must give the same numerical result whether computed on the sphere or on the disk with the projected metric.
This is not an approximation. It is not a coincidence. It is a structural feature of how geometry works. Let me show this concretely for several observations.
Observation 1: The Horizon Dropoff
On the sphere of radius R, an observer at height h above the surface sees the horizon at angular dip α below the local horizontal, where:
cos(α) = R / (R + h), or for small h, α ≈ √(2h/R)
For Earth (R ≈ 6,378 km) and an observer at h = 2 m (eye level standing on a beach), α ≈ 0.025° and the horizon distance is d ≈ √(2Rh) ≈ 5.05 km. A six-foot observer sees the horizon about 3.1 miles away.
Now on the disk with the projected metric: an observer at position (r, ψ) is elevated to height h. The local geometry around the observer is identical to the sphere because the metric is the sphere’s metric. The horizon calculation uses the same formula and gives the same answer. The disk reproduces the horizon dropoff exactly — at the same rate, at the same distance, with the same √h scaling — because the disk is the sphere in different coordinates.
Observation 2: Two Celestial Poles
On the sphere, Earth’s rotation axis intersects the surface at two antipodal points: the North and South Poles. Observers near the North Pole see stars circulating around Polaris near the celestial north pole. Observers near the South Pole see stars circulating around Sigma Octantis near the celestial south pole. The two poles are diametrically opposite.
On the disk, the rotation axis corresponds to the line through the disk’s center perpendicular to the disk plane. The North Pole at the center has the celestial north pole directly above; observers there see stars circle a fixed point overhead. The South Pole, unrolled around the outer boundary, has its celestial counterpart at the projective infinity of the disk plane — observers anywhere on the outer ring see stars circling a fixed point in the same projective direction.
The two celestial poles still exist. They are just labeled differently in disk coordinates. The stars visible from each location are exactly what the sphere predicts: Crux, the Magellanic Clouds, and Sigma Octantis become visible as you approach the disk’s outer regions, exactly as the sphere predicts they should from southern latitudes.
Observation 3: Lunar Eclipse Shadow
The shape of Earth’s shadow on the Moon during a lunar eclipse is always a circular arc. This is a 3D-shape claim: Earth’s silhouette, viewed from the Moon’s direction, must be a circle.
On the sphere model, this is automatic. A sphere is the unique 3D solid whose silhouette is circular from every viewing angle. On the disk model with the projected metric, the disk is geometrically a sphere — it just looks flat when drawn on the page. Its silhouette in physical 3D space, seen from any angle, is a circle. The same Earth casts the same shadow. The shadow’s curvature gives Earth’s diameter as 2R, matching all other measurements.
Observation 4: Antipodal Geography and Flight Times
The flight from Sydney to Santiago takes about twelve hours at standard cruise speed, implying a distance of roughly 7,000 miles. On the sphere, the great-circle distance between Sydney (33.9° S, 151.2° E) and Santiago (33.4° S, 70.7° W) is about 7,060 miles, computed via the spherical law of cosines.
On the disk with the projected metric, the geodesic between the same two points — computed using ds² = dr² + R² sin²(r/R) dψ² — is the same 7,060 miles. The path looks curved when drawn on the flat page. It does not look like a straight line from Sydney to Santiago on the disk. But the metric makes the curved-looking path the shortest one, with the correct length.
If you were to use the Euclidean metric on the disk instead (the Claim B move), the same two points would be roughly 15,000 miles apart. The plane would need to fly twice as fast as physics permits. Flight times do not lie, and they only close with the projected metric. This is one of the first places Claim B begins to fail and Claim A continues to succeed.
Observation 5: GPS Triangulation
The Global Positioning System triangulates a receiver’s location by measuring time-of-flight from at least four satellites. The math assumes a specific Earth geometry — the WGS84 spheroid, with equatorial radius 6,378,137 m and polar radius 6,356,752 m. Errors in geometry produce errors in position.
The fact that GPS works to meter precision globally is direct evidence that the geometry used in the math is correct. On the disk with the projected metric, the same coordinates and times produce the same position fix, because the geometry is the same. On the disk with the Euclidean metric, the position fix would be wrong by miles. Receivers would never have agreed with reality. They do agree, to within meters, billions of times per day.
The General Theorem
The proof of Claim A is now formally stated:
Let M and N be two smooth Riemannian manifolds, and let Φ: M → N be a diffeomorphism that carries the metric of M to the metric of N. Then every geometric observation on M has an exact counterpart on N giving the same numerical value.
Apply this to M = the sphere, N = the disk, and Φ = the azimuthal equidistant projection. The conclusion follows: every observation made on Earth as a sphere has an exact counterpart on the disk with the projected metric. The two are mathematically the same object.
This is not a physics claim about which model is right. It is a mathematical claim that, when both are stated precisely, the sphere model and the projected disk model are not two different models. They are one model expressed in two coordinate systems.
The Theological Connection
For those working within a firmament cosmology, this result is not a defeat. It is a synthesis.
The firmament — the rakia of Genesis 1 — fits inside the disk representation as the celestial sphere viewed from within. The “circuits” of the heavenly bodies in Psalm 19, the sun’s circuit “from the end of the heavens to the end of them,” map cleanly to orbital paths in azimuthal coordinates. The “ice wall” at the edge of the disk is the South Pole, geographically distributed around the outer boundary by the projection — the same Antarctica that has been physically reached and continuously inhabited at Amundsen-Scott Station since 1956.
The patriarchal headship cosmology — glory radiating from a center, lights set in their circuits, the world ordered by Spirit rather than gravitational chance — survives intact. The center of the disk corresponds to one cosmological pole. The structure of the heavens above the disk corresponds to the celestial sphere. The theological language of “set in the firmament,” “circuit from end to end,” “lights for signs and seasons” reads cleanly in disk coordinates, just as it reads cleanly in sphere coordinates.
The choice between disk and sphere is not the choice between honoring Genesis and capitulating to modernism. It is the choice of coordinate vocabulary for describing the same underlying creation.
“It is he that sitteth upon the circle of the earth, and the inhabitants thereof are as grasshoppers; that stretcheth out the heavens as a curtain, and spreadeth them out as a tent to dwell in.”
Isaiah 40:22
The same heavens. The same earth. Different language for describing what God built.
What We Have Proven
To restate clearly: Claim A is the proposition that the flat-disk representation of Earth, equipped with the azimuthal-projected spherical metric, is mathematically equivalent to the standard spherical model. Every observable phenomenon — horizon dropoff, eclipse geometry, polar stars, antipodal geography, flight times, GPS triangulation, the path of the sun, the phases of the moon — produces identical numerical predictions on either model.
This equivalence is not approximate, ad hoc, or theory-laden. It is a theorem.
Within Claim A, there is no empirical evidence that could distinguish the two pictures, because there is no difference between them to be detected. The choice is purely a choice of language. A reader committed to a firmament cosmology need not abandon any observation; they need only recognize that their cosmological vocabulary describes the same physical reality that “spherical Earth” describes.
Where Claim B Becomes Necessary
What Claim A does not prove — what it deliberately does not prove — is that the disk is geometrically distinct from the sphere. The disk in Claim A has the projected metric, which is mathematically identical to the sphere’s metric. The two are not two different geometries; they are one geometry in two parametrizations.
For someone who wants the disk to be a metrically distinct reality — flat in the literal Euclidean sense, with ruler-on-page distances corresponding to physical distances on Earth’s surface — Claim A is not enough. They need Claim B: the proposition that a Euclidean-flat disk can be made consistent with observation through additional physics, such as refractive media, distance-distorting fields, atmospheric effects beyond what standard measurement reveals, or new optical structures within the firmament.
Claim B is a much harder mathematical exercise. It cannot rely on coordinate transformations; it must produce new physics. The next post will work through the bottlenecks: where Claim B can be made to fit the data, where it requires ad hoc tuning, where it generates testable predictions that distinguish it from Claim A, and where it ultimately succeeds or fails.
For now, Claim A stands.
The disk and the sphere, in their mathematically rigorous forms, describe the same Earth.
Claim A, Made Operational
The argument above is best handled, not just read. Two companion apps render Claim A directly — the same observer, the same stars, the same horizon, drawn first as a sphere and then as a disk. Drag the figure in either app and the visible hemisphere changes; flip between the two and the picture changes while the data underneath stays identical.
- The Witness Disk (flat-earth view from above) — The same cosmos in azimuthal projection. The North Pole at center, the equator as a circle at half the disk's radius, the South Pole stretched into the outer ice ring. The horizon curve sweeps the disk as you drag.
- The Three Witnesses (sphere view) — Drag a figure around the globe and your horizon sweeps the celestial sphere. The two polar witnesses fix the axis; the universal band of zodiacs declares to every eye.
The apps are cross-linked at the top so you can toggle between projections with a tap. The math underneath is the projected metric ds² = dr² + R²sin²(r/R)dψ² — the equivalence this essay proves, walked through by hand.