A coordinate system from the body itself

Abstract

Every global coordinate system in daily use carries at least one decision that came from the society that built it rather than from the planet it describes. Longitude begins at Greenwich because British charts dominated shipping in 1884, not because anything on the Earth marks that line. Height is often referred to a local sea level that no two nations agree on. This paper works through a different question. If an observer, or an instrument, arrived at a planet with no access to its history, its treaties, or its units, what coordinate system could be built from the body alone?

The answer is more complete than expected. A rotating mass hands over its own polar axis, its own zero of height, and, provided it is not a perfect sphere of revolution, its own prime meridian. Each of these is derived from first principles below, checked against the systems that geodesy and astronomy already use, and marked as either a rediscovery of known practice or something that appears to be unclaimed. The failures are reported as carefully as the successes, including one elegant construction that does not survive the mathematics. The paper is not a proposal to renumber the Earth. It is an account of how much of our coordinate apparatus is physics and how much is habit. The habit is larger, and the physics more reachable, than routine use of a map would suggest.

1. Introduction: a test with no politics in it

A map of a planet needs a small number of agreed starting points. It needs to know which way is north, where longitude begins, and where height is measured from. On Earth these three questions have historical answers rather than physical ones. North is taken from the spin axis, which is defensible, but the prime meridian and the sea-level datum are inheritances. The 1884 International Meridian Conference in Washington chose the meridian through the transit instrument at Greenwich by a vote of 22 to 1, with France and Brazil abstaining. A separate motion, that the initial meridian should have a character of absolute neutrality and should cut no great continent, was put to the same conference and defeated 21 to 3. The delegates knew the choice was arbitrary. They chose continuity of existing charts over neutrality, and Greenwich won largely because about 72 per cent of shipping tonnage already used it.

This is a reasonable way to run a planet that already has navies and treaties. It is a poor way to understand what a coordinate system is. To separate the physics from the habit, this paper adopts a single acceptance test. A reference is legitimate only if a probe arriving cold at an unknown body, knowing none of its history, could recover that reference from the body itself. Anything that fails the test is a fingerprint of a particular civilisation and is set aside. The test is deliberately strict, and it turns out to be productive: it accepts some references immediately, forces a real choice on others, and, in one important case, tells the observer honestly that the planet has no answer to give.

2. What the sphere forbids

Three results bound everything that follows, and it is cheaper to state them at the start than to rediscover them by disappointment.

The first is Gauss's Theorema Egregium. A sphere has intrinsic curvature and a flat sheet does not, so no projection of a sphere onto a plane can preserve both area and angle at once. A flat map must distort one or the other, everywhere or in patches. This is a property of the surface, not a limitation of cartographic skill, and it means the honest response to mapping a planet is to keep the sphere and subdivide it rather than to flatten it.

The second concerns any attempt to tile a sphere with a mostly hexagonal grid. Euler's formula, that vertices minus edges plus faces equals two, forces exactly twelve pentagons into any tiling of the sphere by hexagons and pentagons meeting three to a vertex. The count of hexagons is free; the twelve pentagons are not. More generally, Descartes showed that the total angular defect of any polyhedron shaped like a sphere is always 720 degrees. A designer never removes that defect. The only freedom is how to parcel it out. An icosahedral scheme spreads it across twelve gentle vertices; a cube concentrates it in eight sharper corners. Either way the debt is paid in full.

The third is the hairy ball theorem: a continuous tangent vector field on a sphere must vanish somewhere. In coordinate terms this is why a two-number grid on a sphere always has singular points, the poles of latitude and longitude being the familiar example. The distinction worth holding is this. The theorem forbids a smooth non-vanishing two-dimensional coordinate field. It does not forbid labelling every point of the sphere with its position in the surrounding three-dimensional space. An embedding has no singularities. The pole problem is a property of insisting on two numbers, not a property of the sphere.

3. A natural placement, and why it fails

Before building the frame it is worth resolving how points should be distributed on a sphere at all, because the obvious first answer is wrong in an interesting way. The latitude and longitude grid crowds points badly at the poles and wastes them there while starving the equator. Nature offers an apparent alternative in the arrangement of seeds on a sunflower head and spines on a pufferfish, both of which place new units at the golden angle of about 137.5 degrees from the last. Laid onto a sphere as a spiral, this produces coverage that stays even at any number of points, because the golden ratio is the most irrational number and the arrangement therefore never falls into a repeating pattern that would clump.

The even coverage is real and is documented. Gonzalez showed that such a lattice estimates areas on the sphere with about 40 per cent lower error than a latitude and longitude grid of the same size, and the meteorological work of Swinbank and Purser built global model grids on the same idea. A refinement, placing the endpoints slightly off the poles rather than on them, improves the spacing further.

The reason it is useless as a map is decisive, and it took the mathematics to see it clearly. A map needs hierarchy. A location's name should refine as you zoom, so that a longer name sits inside the region named by a shorter one. This requires that fine cells nest exactly inside coarse cells. A spiral has no cells and no nesting. Worse, adding points does not merely fail to nest, it reorganises the neighbour relations of the points already placed, and it does so in jumps. A study of the Ising model on Fibonacci spheres found that the fraction of six-neighbour sites rises with the point count and then collapses abruptly at critical counts, because a new spiral arm forms, in the same way a sunflower switches from thirty-four arms to fifty-five. A recent computer graphics paper states the underlying problem plainly, that there is at present no way to add points to such a set while preserving its properties.

The spiral therefore sits at one end of a genuine trade. It gives even coverage and continuous resolution but cannot nest. The lesson worth keeping is smaller and survives the failure. A sunflower keeps an empty disc at its centre. The centre is where growth is measured from, not a place a seed is put. Applied to a sphere, the reference points should be voids rather than occupied cells. The latitude and longitude grid does the opposite and drives every meridian through the pole, so the one genuinely singular point on the planet becomes the busiest intersection on the map. Keep the singular points clear: that inversion is the one durable result of the spiral work.

4. The reference frame from the body

The frame is built by running each candidate reference through the acceptance test of Section 1. A probe measures each quantity from the body alone or the quantity is rejected.

4.1 Origin

The centre of mass is fixed by the body's own gravity and is found by watching how anything falls toward it. No choice enters. It is the origin.

4.2 Axis and its sign

Every planet spins, because a collapsing cloud with any angular momentum must spin faster as it shrinks. The spin supplies an axis, and with it two poles and a unique equator. The axis is a line, not yet a direction, so calling one pole positive needs a rule. The spin supplies its own rule through the right-hand convention: curl the fingers in the direction of rotation and the thumb gives the positive pole. This is the angular momentum vector, and it depends on nothing but the body turning. It works for a planet with no star and no neighbours. The alternative used by the International Astronomical Union for the major planets, which defines north as the pole lying on the north side of the solar system's invariable plane, fails the cold-probe test because it borrows from the rest of the system. The right-hand rule borrows from nothing, so it is preferred here.

4.3 Vertical

Mass produces a gravity field, the field has equipotential surfaces, and one such surface is chosen as the zero of height. This is already the modern standard on Earth. In 2015 the International Association of Geodesy defined the global vertical datum not as a sea level but as a fixed value of gravity potential, W0 equal to 62,636,853.4 square metres per second squared, and defined heights as differences of potential from that value, the geopotential number. Metric heights follow by dividing that potential difference by a chosen value of gravity, which is why orthometric, normal, and dynamic heights differ only in the divisor. The physically fundamental quantity is the potential, and the metre-valued height is derived. Earth's choice of which equipotential to call zero was tied to mean sea level, which a dry planet does not have. The cold-probe version of the rule is to take the equipotential that encloses the body's own volume, the level a global ocean would settle to if one existed. Up is toward lower potential.

4.4 Meridian

The prime meridian is the hard case, and the acceptance test earns its keep here by giving different answers for different bodies. A perfect sphere of revolution is symmetric about its axis, so nothing on it distinguishes one meridian from another, and no natural zero of longitude exists. Real bodies are not perfect. A solid planet is slightly triaxial, its equator a faint ellipse, so its axis of least equatorial moment of inertia is a real and findable direction that a probe recovers from the gravity field. For the Earth this axis lies along the 14.93 degrees west and 165.07 degrees east diameter, determined from the degree-two gravity coefficients by Liu and Chao. A tidally locked body gives an even stronger signal, the permanent bulge that points at whatever it orbits, which is why the Moon's meridian is set by the mean sub-Earth point. A fast-spinning fluid planet that has relaxed into a true surface of revolution has neither triaxiality nor lock, and it genuinely has no natural meridian. Jupiter's official longitude is defined from the rotation of its magnetic field for exactly this reason. A body with neither shape asymmetry nor field has no prime meridian at all, and this is not a gap in method but the same symmetry that the hairy ball theorem expresses.

There is a subtlety that must be handled honestly, because skipping it would smuggle in a choice as arbitrary as Greenwich. The moment of inertia tensor is symmetric, so it gives the meridian as a line with two ends 180 degrees apart and cannot say which end is zero. The ellipse is an even shape: rotate the planet 180 degrees about its axis and it maps onto itself, so its two ends are true twins. To choose an end, a quantity is needed that flips under that same half turn, an odd asymmetry rather than an even one. The body carries one at the most basic level. Its centre of mass, already the origin, is where the mass balances. Its centre of figure, the centroid of the actual surface, sits somewhere else, because no planet is built evenly. The offset between the two is a genuine direction rather than a line. On Mars the offset is about three kilometres, the whole northern-lowland and southern-highland dichotomy. On the Moon it is about two kilometres toward the far side. The rule then reads: zero longitude is the end of the least-inertia axis lying toward the body's figure offset.

This tiebreaker fades in step with the meridian it resolves, which is the reassuring behaviour. A strongly lopsided body names its zero loudly. A nearly symmetric one names it faintly. A perfectly symmetric one has no offset and no triaxiality, so it has neither a meridian nor a need for one. A single physical quantity, the body's lopsidedness, answers both whether a zero longitude exists and which end it is. The two questions were never separate.

5. A grid mounted on the frame

The frame fixes reference points but does not by itself address locations. For that a scaffold is needed, and the honest scaffold is the one that respects the trade found in Section 3. Squares nest, hexagons and spirals do not: four squares tile one larger square exactly, forever. A cube inscribed in the sphere, its faces projected outward and each face repeatedly quartered, gives a clean hierarchy. This is the cubed sphere, and its indexed form is the coordinate system that Google's S2 library uses and that the Finite-Volume Cubed-Sphere core uses inside operational weather models. The construction was reached here independently before the prior art was checked, which is worth stating plainly as a mark of its naturalness rather than its novelty.

The cube pays the Descartes debt at its eight corners, where three faces meet, rather than at the twelve vertices of an icosahedron. Fewer defects, but sharper: cell areas in such a scheme vary by roughly a factor of two, worst at the corners. The gain is that the cube's symmetry does something the earlier ambiguity needed. Seat the cube with a face centre on each pole and it has four-fold symmetry about the axis, so a 90 degree turn maps it onto itself. The natural meridian is a line with a 180 degree ambiguity, and 180 degrees is one of the cube's own symmetry turns, so the two ends of the meridian give the identical cube. The information the planet refused to provide is exactly the information the cube does not need. Only an even-order polar axis works, and the four-fold face-centre choice is the most symmetric of those. It places the calmest cells of the grid on the poles, which are the points Section 3 said to keep clear.

With face centres on the poles the eight corner defects fall in two rings at plus and minus 35.26 degrees latitude, and the natural meridian fixes their longitude. Computing where they land on the real Earth gives a pleasing but imperfect result. Six of the eight fall in open ocean. The seventh grazes the Chinese coast near the Yellow Sea, and the eighth falls on land in the Argentine pampas. There is no freedom to do better, because the lean axis fixes the longitude. A free choice of orientation, as in Buckminster Fuller's Dymaxion map, can drop all defects into water, but only by abandoning any physical anchor. Anchoring to the body costs Argentina.

6. Time, and why the frame is never finished

Every reference so far drifts, and a frame that ignores when it was measured is a snapshot pretending to be a law. The drifts sort into two families, and conflating them is the error to avoid. One family is motion relative to the stars. The spin axis precesses, tracing a wide cone against the sky once every 25,772 years, about 50 arcseconds a year, which is why Polaris is only a temporary pole star. The other family is motion relative to the ground. The pole wanders over the crust by a few metres on a roughly 433-day wobble, the lean meridian slowly reorients as mass moves, and the length of the day itself grows by a couple of milliseconds a century.

That split points to the solution. The frame is built in two parts bound at an instant. A body frame carries the zeros derived above, fixed to the crust and stated at an epoch. A sky frame is anchored to the most distant matter available, the quasars, which is the only reference in the entire construction that belongs to no planet. A probe at any body in any system sees the same quasars. The modern realisation of this is the International Celestial Reference Frame, built from about 4,500 radio sources with a few hundred defining the axes. Astronomy adopted it in place of the moving equinox precisely because the old zero drifted.

The humbling part is that this rotation cannot be fully computed. The smooth terms are predictable, but the last part is the oceans, the atmosphere, and the core shoving the planet in ways that must be observed rather than derived. Earth's own service publishes orientation parameters continuously for this reason. The connection between the fixed sky and the turning ground is a measurement, taken fresh and never final. A person navigating by the stars at night was doing the same thing, reading tonight's sky because tomorrow's is not knowable in full, and the difference between that and a radio observatory is precision rather than principle.

7. The numbers

7.1 The natural frame against the current system on Earth

Only one quantity changes by much when the natural frame replaces the current one, but it changes a great deal. North does not move, since the spin axis is already in use, and the two agree to within arcseconds. Height shifts by up to a metre or two depending on location, because local sea-level datums disagree with the true geoid by that much. Longitude, though, is renumbered by the full 14.93 degrees between Greenwich and the Earth's own lean axis, which is 1,662 kilometres at the equator.

The scale of that shift is best seen against the offset that geodesy already discusses. The well-known fact that the Greenwich meridian moved refers to the 102.5 metre gap between the brass line at the old transit circle and the modern reference meridian. The natural frame moves the zero by 1,662,000 metres, about sixteen thousand times further. The profession refines a detail measured in metres while the choice of where zero sits at all is arbitrary at the scale of a continent. The natural meridian is also, by accident, close to the neutral ocean line the 1884 conference wanted and rejected: it clips eastern Iceland, runs down the empty eastern Atlantic, and its far side falls through the open western Pacific.

7.2 Extending the method to other planets

The method travels to every body without change, since no part of it depends on a particular planet's history. The specific zeros do not travel, because each body has its own axis, its own lean, and its own datum value. A single grid therefore cannot be laid across two planets. The point is quantified by asking how far each planet's north points away from Earth's north, measured on the sky both share. Nothing lines up. Saturn is the closest cousin at 6.5 degrees, and Uranus points 105 degrees away, its north lying almost in its own orbital plane.

The grids do not share a pole or a meridian, but every pole in the table is quoted in one frame, the celestial reference frame of distant quasars. That shared sky is the only common anchor, and it is what allows any two planetary grids to be related, through the sky rather than directly. This is not a novelty; it is how spacecraft navigation and the International Astronomical Union's planetary coordinate standards already work. The construction here reproduces that architecture from first principles.

8. Relation to existing work

Almost every component of this frame exists somewhere in the literature, and honesty requires saying so plainly. The value of the exercise is the assembly and the acceptance test, not the parts. Gravity potential as the true vertical is the current international standard, adopted in 2015. The golden-angle spiral is a studied object in numerical integration, graphics, and meteorology, and its inability to nest is stated in the recent graphics literature as an open problem. The cubed sphere is in production use in mapping and weather modelling. The abandonment of the moving equinox for a quasar-based celestial frame is the basis of modern astrometry. Defining each planet's frame from its own rotation and lean, referenced to a common celestial frame, is the working practice of the International Astronomical Union's cartographic working group. On all of these the present work aligns with what is already done.

Three things appear to be either unusual or unclaimed. First, the criterion used to place the spiral offset, the ratio of smallest to largest neighbour spacing, does not seem to be treated in the peer-reviewed literature. Second, and more substantially, there appears to be no published proposal to tie the Earth's terrestrial prime meridian to a physical property of the body, such as the least-inertia axis, rather than to an observatory. The strongest available candidate, the equatorial principal axis, has been measured since 1991 but never proposed as a meridian, and the astronomical working group has explicitly declined to move even Mercury's meridian to a dynamical definition, on the stated ground that continuity of the cartographic record outranks physical naturalness. Third, the specific tiebreaker for the 180 degree ambiguity, using the centre-of-figure offset to select the end of the axis, is a natural application of a criterion the same working group already uses for asteroids, but its application to a planetary prime meridian does not appear in the sources consulted.

The strongest objection to the whole exercise is the one the working group itself makes. A coordinate system's value lies largely in the continuity of the record built on it, and a physically purer frame that renumbers every existing coordinate is not obviously worth the disruption. The paper accepts this fully. The claim here is not that the Earth should be renumbered. It is that the difference between the two frames measures how much of the current system is convention, and that on other bodies, where no legacy record exists, the natural frame is the correct default and is close to what is already done.

9. Limitations

The meridian is only weakly defined for the Earth. The lean axis itself is solid, coming from the degree-two gravity field, but the tiebreaker that chooses which end is zero relies on the centre-of-figure offset, and the Earth's offset is small, largely oriented north to south rather than in the equatorial plane, and partly seasonal as water and ice move. The honest reading is that the Earth genuinely under-defines its own meridian, which is why the 1884 committee had a vacuum to fill. On strongly lopsided bodies such as Mars the same rule answers cleanly. The natural frame would also invalidate every existing chart, address, and survey record if adopted on Earth, which is a decisive practical objection to adoption and is not disputed.

10. A consequence for reading the past

If the everyday coordinate system is largely convention, then applying it to artefacts built by people who used a different system can mislead in both directions. It can hide a real pattern, and it can manufacture a false one. The second risk is the one usually underestimated. Given enough monuments, enough measured dimensions, and a free choice of unit, clean numbers are certain to appear by chance, which is why claims that a monument encodes a fundamental constant should be treated as coincidence until they survive a change of unit.

The disciplined version of the idea is mainstream and productive. The builders of the Egyptian pyramids used the reference that was actually available to them with no clock and no instruments, which was the sky. The Great Pyramid is aligned to true north within about three or four arcminutes. The leading explanation, from Spence, is that the builders held a plumb line against two circumpolar stars and waited for the line joining them to hang vertical through the pole. That alignment was true only for a narrow window around 2467 BCE, because precession slid it off at about 27 arcminutes per century, so the small alignment errors across successive reigns date the monuments. Read in an arbitrary modern frame the pyramid is a slightly imperfect square. Read in the sky frame it is a dated instrument.

The ratio of pi to the pyramid's proportions makes the opposite point. The Great Pyramid's perimeter is close to two pi times its height, but the builders almost certainly never used pi. Their slope was measured in a unit called the seked, and because they divided the cubit into seven palms, the slope they chose gives the ratio 22 over 7, which approximates pi to three digits by an accident of base-seven arithmetic. The coincidence lives in the modern observer's tools, not in the monument. The frame built in this paper is useful here less as a discovery than as a filter. It says which ancient claims are even reachable. Latitude from star altitudes, cardinal alignment from circumpolar stars, and a construction date from precession are all reachable. Longitude, the least-inertia meridian, and the planet's precise dimensions are not reachable without a clock or satellite gravimetry, so any claim resting on them is almost certainly an artefact of analysis.

11. Conclusion

A rotating planet supplies more of its own coordinate system than daily use of a map suggests. It gives an origin at its centre of mass, an axis and a sign from its spin, a zero of height from its own gravity field, and, if it is lopsided enough, a prime meridian from the long axis of its equatorial figure with the end chosen by the offset of its shape from its mass. A usable grid can be mounted on that frame with a cubed sphere, whose symmetry absorbs the one ambiguity the planet leaves, and whose unavoidable defects fall mostly in ocean when anchored to the Earth. All of it is bound to a sky of distant quasars and stamped with an epoch, because nothing is perfectly still and the link between sky and ground must be measured rather than computed.

None of the parts is new, and most match what geodesy and astronomy already do. What the exercise shows is the size of the gap between the physics and the habit. On Earth the gap is 1,662 kilometres of longitude, sixteen thousand times larger than the offset the profession usually frets about, and it exists only because a choice made in 1884 was never meant to be physical. On any other body, where there is no chart to preserve and no navy to please, the frame the body defines for itself is the natural default. The Earth is the exception, and it is an exception of history rather than of physics.

Key sources: Gauss, Theorema Egregium (1827); Euler and Descartes on polyhedral defect; Gonzalez, Mathematical Geosciences 42 (2010); Swinbank and Purser, Q. J. R. Meteorol. Soc. 132 (2006); Marques, Bouville and Bouatouch, IEEE TVCG 27 (2021); Pochinok, Molochkov and Chernodub, arXiv:2301.06849 (2023); OGC Topic 21, DGGS; Uber H3 and Google S2; IAG Resolution No. 1 (2015), W0 = 62,636,853.4 m^2/s^2; Liu and Chao, Geophys. J. Int. 106 (1991); Malys et al., J. Geodesy 89 (2015); International Meridian Conference protocols (1884); Charlot et al., ICRF3, A&A 644 (2020); Archinal et al., IAU WGCCRE (2011); Souami and Souchay, A&A 543 (2012); Spence, Nature 408 (2000).