Breaking the GM Degeneracy

J. Rogers, SE Ohio

A Deep-Space Oscillating Test Mass Experiment to Determine

the Gravitational Constant G to Nine Significant Figures

Abstract

The gravitational constant G remains the least precisely known fundamental constant in physics, determined to only approximately five significant figures after three centuries of effort. We identify the structural cause: G and the mass M of any gravitating body large enough to produce a measurable field are observationally inseparable. Every astronomical measurement yields only the product GM. Cavendish-style laboratory experiments have failed to converge across 300 years. We propose a conceptually simple resolution: construct a toroidal mass of precisely known composition in deep space, far from competing gravitational sources, bore a hole through its center, and drop a test mass through the hole. The test mass oscillates indefinitely under gravity alone, with period determined entirely by the known surface mass density of the toroid. An LED-based laser interferometer simultaneously tracks oscillation period T, instantaneous acceleration a(t), and distance r(t) from the center hole continuously throughout each oscillation, providing three independent and overdetermined routes to G from a single data stream. Because the mass is known from construction rather than inferred from gravity, the GM degeneracy is broken for the first time. The test mass is repeatedly lifted, settled, and released, accumulating thousands of independent period measurements per run over a mission lifetime exceeding one year on station. Tidal contamination from the Sun at 3 AU is verified by calculation to be 1.1 × 10⁸ times smaller than the measurement signal, and galactic tidal forces are 6.7 × 10⁷ times smaller. The deep space environment is not merely quiet — the rest of the universe has gone effectively silent. All required technologies are flight-proven. The mission requires no new physics and no new engineering principles.

1. The Problem: GM is What Nature Exposes

Newton's law of gravitation is conventionally written as:

F = G M m / r²

and general relativity encodes gravitational geometry through terms of the form GM/c²r. In both frameworks, G and M appear as a product. This is not a mathematical convenience — it reflects a deep operational fact: no measurement that relies on gravity to establish the behavior of a gravitating body can separate G from M. The observable is always GM.

NASA's operational practice makes this explicit. Planetary ephemerides, spacecraft navigation, and GPS relativistic corrections all use the gravitational parameter mu = GM, known for Earth to approximately nine significant figures. The individual values of G and M are not used. They cannot be, because the mass of any astronomically significant body is inferred from its gravitational behavior, making any G determination from astronomical sources tautological. If we knew the mass of the Earth independently, we would know G to identical precision. We do not, and the circularity is exact and complete.

G is not philosophically uncertain. It has an exact value. Nature knows it precisely. The uncertainty is entirely on the measurement side, and the measurement side has a specific, identifiable structural problem that has gone unresolved for three centuries.

2. The Failure of Laboratory Methods

The Cavendish torsion balance, introduced in 1798, was designed to escape this circularity by using laboratory-scale masses whose weight could be determined independently through mechanical means. After 300 years of refinement across dozens of independent experiments at leading metrology institutes worldwide, the results do not converge.

Recent high-precision determinations of G disagree with each other by 40 to 50 parts per million, while individual experiments claim uncertainties of 10 to 20 parts per million. The discrepancy between results exceeds the claimed precision by a factor of two to five. CODATA periodically widens the accepted uncertainty interval to accommodate the spread.

The fundamental difficulty is the signal-to-noise environment. Gravity is the weakest force. Laboratory-scale masses produce gravitational forces at or below the level of seismic noise, thermal expansion, electrostatic coupling, and the gravitational influence of nearby structures. No experimental design has succeeded in isolating the gravitational signal cleanly from this noise floor at the required precision. Three hundred years is sufficient to conclude this is not a solvable engineering problem in the terrestrial environment.

3. The Consequence: A Blurred Axis

The inability to determine G precisely propagates directly into every dimensionless ratio that crosses the gravitational-electromagnetic interface. The Planck mass is defined as:

m_P = sqrt(hbar * c / G)

and inherits the full uncertainty of G. The fine structure constant alpha is known to twelve significant figures. The ratio m_p/m_P — proton mass to Planck mass — is a fundamental dimensionless number that any unified theory must address. It is known to only five significant figures, not because of any difficulty on the electromagnetic side, but entirely because G limits the determination of m_P.

Numerical computation confirms the scaling precisely. Across the current uncertainty range of G — approximately 15 parts per million — the natural unit scale derived from hbar, c, and G shifts by 7.5 parts per million, following the theoretical scaling X proportional to G^(-1/2) exactly. The internal self-consistency within any fixed value of G is maintained to floating-point precision (~10^-16). The geometry is exact. The uncertainty is entirely in our measurement access to G.

G is a unit conversion factor — the bridge between our independently-defined kilogram and the natural geometry of gravity. Like c, it would be 1 by construction if our mass unit had been defined gravitationally. The constants G and c are not deep truths about nature. They are artifacts of defining length, time, and mass independently using different physical processes. The scandal is not that G is uncertain. The scandal is that we have treated a units alignment problem as a fundamental mystery for three centuries.

Any theory proposing an exact relationship between m_p/m_P and alpha cannot be tested better than five significant figures. The relationship may be exact, numerically sitting in plain sight, and we cannot resolve it. This is structurally identical to how epicyclic astronomy concealed elliptical orbits for over a millennium. Epicycles made accurate predictions. The system was internally consistent. And buried inside that predictive success was exact geometry the representation could not expose. GM is our epicycle. The exact dimensionless ratios connecting gravity to electromagnetism are hidden inside a perfectly predictive framework that cannot expose them.

4. The Proposed Experiment

4.1 Core Concept

The GM degeneracy is broken by one structural change: use a gravitating body whose mass is known from construction, not from its gravitational behavior. A body assembled from measured components in a zero-gravity environment has a mass determined by the sum of its parts, each weighed through the metrological chain anchored to the SI kilogram, independently of gravity. This mass M is known before the gravitational measurement begins.

A flat toroidal disk — a large washer — with a hole bored through its center axis is the chosen geometry. A test mass m is dropped through the hole. It oscillates back and forth through the disk under gravity alone. The period of this oscillation depends only on the surface mass density sigma of the disk, which is known from construction. The measurement of G reduces to a measurement of oscillation period, acceleration, and distance — all tracked simultaneously by a single LED interferometer.

4.2 Physics of the Toroidal Oscillator

For a uniform infinite plane of surface mass density sigma, the gravitational acceleration is constant on both sides:

g = 2 * pi * G * sigma

independent of distance from the surface. A test mass released from rest at height h above the disk undergoes simple harmonic motion with period:

T = 2 * pi * sqrt(h / (2 * pi * G * sigma))

The critical feature is that period is independent of amplitude. As the oscillation slowly damps, T remains constant. Every cycle from first to last gives the same measurement of G. The experiment does not degrade as energy dissipates. The LED interferometer simultaneously tracks three independent observables throughout every oscillation:

— T: oscillation period, from precise timing of successive zero crossings at the disk plane

— a(t): instantaneous gravitational acceleration, from the second time derivative of interferometric position

— r(t): distance from the center of the hole at every instant, providing continuous geometry verification and confirming the mass distribution model

These three observables are overdetermined for a single unknown G. Their mutual consistency throughout each oscillation provides direct internal systematic error estimation with no additional apparatus. Any unmodeled perturbation shows up as inconsistency between the three routes to G before it corrupts the result.

4.3 Washer Construction: Batteries as Known Mass

The toroidal disk is constructed from cylindrical battery cells packed into a toroidal form and encased in a precision metal shell. The batteries serve a dual purpose: they are the power source for the mission and they constitute the primary known mass. No dead-weight ballast is carried. Every kilogram of structural mass is simultaneously delivering power.

The metal shell provides a uniform, precisely characterized outer geometry. Shell thickness is known. Battery cell geometry and individual mass are measured before assembly. Total mass M is the sum of precisely accounted components, all measured on the ground before launch. The surface mass density sigma = M / pi(R_outer^2 - R_inner^2) is known to the precision of the pre-launch mass accounting.

Battery discharge does not meaningfully change the mass. The chemical energy released is accounted for by E = mc^2 at a level far below the measurement floor. Mass M is effectively constant throughout the mission lifetime, and any slow drift is trackable from power consumption telemetry.

4.4 LED Interferometer

The measurement instrument is an LED-based Michelson interferometer. A narrow-bandwidth LED with a bandpass filter provides sufficient coherence length for path differences involved in tracking oscillation amplitudes of order one meter. Power consumption is in the milliwatt range. No laser cooling, no frequency stabilization, and no moving optical components are required.

At 3 AU from the Sun the sky is dark. There is no solar background to filter against. Starlight provides negligible interferometric noise. The deep space environment is here an advantage: the LED signal completely dominates the detector with no competing background illumination.

The test mass carries a retroreflector. The interferometer arm length changes as the test mass oscillates, producing fringes that encode position continuously. Zero crossings at the disk plane are timed to atomic clock precision. The combination of T, a(t), and r(t) as continuous functions throughout each oscillation delivers G from three independent routes simultaneously from a single passive optical system.

4.5 Operational Procedure

The test mass is lifted to a specified height above the hole center and held. The LED interferometer confirms the test mass is at rest — zero velocity, stable position — before release. The actuator releases. The test mass falls through the hole, decelerates on the far side, returns, and oscillates. The interferometer tracks every cycle continuously.

The oscillation runs until amplitude has decayed to near the noise floor or lateral drift approaches the hole wall. The actuator catches the test mass, lifts it back to the starting height, and waits for the interferometer to confirm stillness. A new run begins. This cycle repeats throughout the on-station mission phase.

Each run provides thousands of independent period measurements. Each lift-settle-release cycle is independently characterized. Run-to-run consistency is a direct systematic error check. The experiment is not a one-shot measurement — it is the same experiment performed thousands of times with the same apparatus in a zero-noise environment, accumulating statistics continuously for over a year.

4.6 Mission Architecture

The washer payload is delivered to the target location by a conventional booster. On arrival the booster releases the washer and fires away to a minimum separation distance of 100 kilometers. At this separation the booster's gravitational influence on the test mass oscillation is negligible, and the booster acts as a radio relay — receiving low-power data from the washer and forwarding it to Earth at full deep space communication power. The booster requires no further maneuvering and recedes on a diverging trajectory.

Following separation, the washer is left to settle. Vibration from separation, thermal distortion as the system reaches radiative equilibrium, and any residual rotation are all monitored by the LED interferometer and allowed to damp naturally. This settling phase may take weeks to months. Formal measurement runs do not begin until the interferometer confirms the system is at rest to the required precision. The settling period characterizes the system in detail and verifies the mass distribution model against the measured gravitational field geometry before any G determination is attempted.

The target location is at or beyond 3 AU from the Sun. The solar tidal gradient across the experimental apparatus at this distance is verified by calculation (Section 5) to be more than 100 million times smaller than the measurement signal. The precise location is determined by a standard orbital mechanics trade between mission delta-V and acceptable tidal contamination. Minimum on-station measurement duration is one year.

4.7 Power Budget

The LED interferometer operates in the milliwatt range. The test mass actuator draws power only during brief lift and release operations. The onboard computer handles data logging and interferometer control at low clock rates. The radio transmitter to the booster relay at 100 kilometers requires trivial power at that range. Total payload power budget is estimated at 20 to 30 watts continuous. The toroidal battery mass, sized for the known mass requirement of the experiment, provides this power for the required mission lifetime with margin.

5. Tidal Contamination Analysis

The primary concern for any deep space gravitational experiment is contamination from external tidal forces. We calculate the tidal acceleration from all significant sources across the 1-meter scale of the apparatus and compare to the measurement signal.

5.1 Signal Strength

For a toroidal mass M = 5000 kg with a test mass at distance r = 1 meter from the center hole, the gravitational acceleration constituting the measurement signal is:

a_signal = G*M / r^2 = (6.674e-11)(5000) / (1)^2 = 3.34e-7 m/s^2

This is the reference against which all tidal contaminations are compared.

5.2 Solar Tidal Force

The tidal acceleration from the Sun across an apparatus of length delta_r is:

a_tidal = 2 * G * M_sun / R^3 * delta_r

where R is the heliocentric distance. At 1 AU this gives 7.93 × 10⁻¹⁴ m/s² across 1 meter — already 4.2 million times smaller than the signal. The tidal force scales as 1/R³, so at 3 AU it drops by a further factor of 27:

a_tidal,Sun (3 AU) = 2.94e-15 m/s^2

This is 113 million times smaller than the measurement signal. As a fraction of the signal it represents 8.8 parts per billion — well below the 1 part per billion threshold required for nine significant figures in G.

5.3 Galactic Tidal Force

The local galactic tidal acceleration is known from stellar dynamics and pulsar timing studies to be approximately 5 × 10⁻¹⁵ m/s² per meter of apparatus length. For the 1-meter scale of this experiment:

a_tidal,Galaxy = 5.00e-15 m/s^2

This is 67 million times smaller than the signal — comparable to the solar tidal and equally negligible.

5.4 Planetary Tidal Forces

Jupiter, the most massive planet, presents a tidal acceleration across 1 meter of approximately 1.18 × 10⁻¹⁸ m/s² at a conservative minimum separation of 4 AU. This is 280 billion times smaller than the signal and requires no further consideration.

5.5 Summary

Table 1 summarizes all tidal contamination sources. The worst-case external contamination — the galactic tidal force — is 67 million times smaller than the measurement signal. At 3 AU, the rest of the universe has gone effectively silent. This is not a marginal improvement over the terrestrial environment. It is a qualitative change in what measurement is possible.

Source	Tidal Acceleration (m/s²)	Ratio to Signal	Orders of Magnitude Below Signal
Toroid 5000 kg at 1 m (Signal)	3.34 × 10⁻⁷	1 (reference)	—
Sun at 1 AU across 1 m	7.93 × 10⁻¹⁴	4.2 × 10⁶ × smaller	6.6
Sun at 3 AU across 1 m	2.94 × 10⁻¹⁵	1.1 × 10⁸ × smaller	8.1
Milky Way galaxy across 1 m	5.00 × 10⁻¹⁵	6.7 × 10⁷ × smaller	7.8
Jupiter at 4 AU separation across 1 m	1.18 × 10⁻¹⁸	2.8 × 10¹¹ × smaller	11.4

Table 1. Tidal contamination at 3 AU compared to measurement signal (M = 5000 kg, r = 1 m, delta_r = 1 m).

The solar tidal force at 3 AU, as a fraction of signal, is 8.8 parts per billion. For reference, nine significant figures of precision in G requires controlling systematics to 1 part per billion. The solar tidal is below this threshold by a factor of nearly 9. For experiments requiring fewer than nine figures of precision, 3 AU provides ample margin. For the full nine-figure target, an orbit at 4 AU reduces solar tidal contamination by a further factor of 2.4, comfortably below the threshold.

6. Statistical Power of Repeated Oscillation

The fundamental advantage of this experiment over every previous G determination is statistical accumulation. A single period measurement T has some uncertainty epsilon from timing precision and environmental noise. After N independent cycles, the uncertainty on the mean period is epsilon / sqrt(N).

In a one-year on-station mission with oscillation periods of order minutes, the number of measurable cycles is of order tens of thousands per run and millions across the full mission. The statistical reduction factor sqrt(N) is of order 1000. Random errors that would limit a single measurement to five significant figures are beaten down to nine or more by accumulation alone.

No ground-based experiment has ever had this. Cavendish apparatus yields one measurement per configuration. Resets are slow and noisy. N never gets large. The noise floor never drops because statistics never accumulate. Here N is limited only by mission lifetime. The experiment improves continuously as long as the apparatus operates.

The three simultaneous observables — T, a(t), and r(t) — provide independent routes to G from the same data stream. Their mutual consistency serves as a continuous systematic error monitor throughout the mission. Any unmodeled perturbation that would corrupt one observable will show up as inconsistency among all three before it biases the G determination.

7. Scientific Return

A determination of G to nine significant figures immediately propagates precision improvement through every dimensionless ratio in physics that involves gravity. The Planck mass m_P = sqrt(hbar*c/G) becomes known to nine figures. The ratio m_p/m_P — proton mass to Planck mass — sharpens from five to nine significant figures with no additional measurement on the electromagnetic side.

The gap between five and nine significant figures is where proposed exact relationships between m_p/m_P and alpha either are confirmed or are falsified. A correct unified theory would predict this ratio exactly as a function of the fine structure constant and other dimensionless electromagnetic parameters. Such predictions are currently untestable beyond five figures. This experiment makes them testable to nine.

Every GM product for solar system bodies simultaneously becomes a precise mass determination. The mass of the Earth, the Moon, Mars, and Jupiter — all known to nine figures immediately by dividing their known GM by the newly precise G. This is a complete remeasurement of solar system masses at no additional observational cost.

8. Cost and Comparison

The cumulative cost of Cavendish-style G determinations over the past century, across major metrology institutes in multiple countries, has been substantial. No convergence has been achieved. Three hundred years of investment has produced not precision improvement but a widening recognition that the terrestrial environment is fundamentally the wrong place to do this experiment.

The proposed mission is less technically complex than many current planetary science missions. It requires no landing, no sample return, no complex in-situ chemistry, no precise pointing at a distant astronomical target. It requires transporting a known mass to deep space, releasing a booster, and operating an LED interferometer and test mass actuator for one or more years. The payload has no moving parts except the test mass actuator. The primary instrument draws milliwatts.

The mission fits within the cost envelope of an ESA Medium-class or NASA Discovery-class science mission. The scientific return — resolving a 300-year measurement failure and opening the gravitational-electromagnetic interface to genuine precision tests — is disproportionate to the engineering investment. A formal feasibility study is the appropriate immediate next step.

9. Conclusion

G has an exact value. The universe does not have error bars. The uncertainty in G is entirely on the measurement side and has a specific identifiable cause: we have never had independent access to the mass of a body large enough to produce a measurable gravitational field. Every previous approach either uses astronomical bodies whose masses are inferred from gravity, or uses laboratory masses too small to overcome the terrestrial noise floor.

The proposed experiment resolves this by construction. A toroidal battery mass of precisely known composition is placed at 3 AU from the Sun. A test mass oscillates through its central hole under gravity alone. An LED interferometer tracks period, acceleration, and distance simultaneously through thousands of oscillations over a mission lifetime exceeding one year. The mass is known before the gravitational measurement begins. The GM degeneracy is broken.

Tidal contamination from all external sources — Sun, galaxy, planets — is verified to be between 67 million and 280 billion times smaller than the measurement signal. The deep space environment does not merely reduce noise. It eliminates it.

The result is G to nine significant figures, the Planck jacobians to nine figures, and every dimensionless ratio at the gravitational-electromagnetic interface sharpened by four to five significant figures. Proposed exact relationships between m_p/m_P and alpha become directly testable as a ratio against kinematics for the first time. No new physics is required. No new engineering principles are required. The only reason this has not been done is that it falls between the institutional mandates of metrology and deep space science. That gap should be closed.

References

[1] CODATA 2018 recommended values of the fundamental physical constants. Rev. Mod. Phys. 93, 025010 (2021).

[2] Cavendish, H. Experiments to determine the density of the Earth. Phil. Trans. R. Soc. London 88, 469-526 (1798).

[3] Gillies, G.T. The Newtonian gravitational constant: recent measurements and related studies. Rep. Prog. Phys. 60, 151 (1997).

[4] Rothleitner, C. & Schlamminger, S. Measurements of the Newtonian constant of gravitation. Rev. Sci. Instrum. 88, 111101 (2017).

[5] Quinn, T. et al. Improved determination of G using two methods. Phys. Rev. Lett. 111, 101102 (2013).

[6] Rosi, G. et al. Precision measurement of the Newtonian gravitational constant using cold atoms. Nature 510, 518-521 (2014).

[7] Armano, M. et al. Sub-Femto-g Free Fall for Space-Based Gravitational Wave Observatories: LISA Pathfinder Results. Phys. Rev. Lett. 116, 231101 (2016).

[8] Folkner, W.M. et al. The planetary and lunar ephemeris DE 430 and DE 431. Interplanet. Netw. Prog. Rep. 196, 1-81 (2014).

[9] Mohr, P.J., Newell, D.B. & Taylor, B.N. CODATA recommended values of the fundamental physical constants: 2014. Rev. Mod. Phys. 88, 035009 (2016).

[10] Duff, M.J. How fundamental are fundamental constants? Contemp. Phys. 56, 35-47 (2015).

[11] Iorio, L. Galactic tidal effects on the Oort Cloud and the outer solar system. MNRAS 443, 2523-2534 (2014).