Why the Mass Cancels: From Newton's Coincidence to Geodesic Motion — Epoche B2
The Common Belief Everyone learns that a heavy ball and a light ball fall at the same rate. Most textbooks present this as a simple fact about gravity: that is just how it works, and there is nothing more to say. This essay argues that the statement hides one of the deepest puzzles in classical physics — and that its resolution changes what we mean by "falling" at all. Thesis: Newton's Cancellation In Newtonian mechanics, two logically distinct kinds of mass appear, and the easiest way to see that they are distinct is to notice that one of them can be measured with gravity switched off. Push a trolley on level ground: its resistance to being accelerated is its inertial mass $m_i$, defined by $F = m_i a$ for any force whatever — a spring, a magnet, a shove. Now weigh it: what the scale responds to is its gravitational mass $m_g$, the quantity that determines how strongly it couples to a gravitational field. The second is a kind of charge. Electric charge provides the instructive comparison: an electron's charge and its inertial mass are unrelated quantities, and nobody expects the ratio $q/m_i$ to be the same for an electron as for a proton — indeed it is not, differing by a factor of about 1836. Gravitational mass is gravity's charge, and the puzzle is why gravity's charge should be numerically identical to the thing that resists acceleration. Write $\Phi$ for the gravitational potential, the field whose gradient $\nabla\Phi$ gives the gravitational field strength — near the Earth's surface $\nabla\Phi$ has magnitude $9.8\ \mathrm{m\,s^{-2}}$ and points upward, so that $-\nabla\Phi$ points down. The equation of motion is $m_i \ddot{\vec{x}} = -m_g \nabla\Phi$, where $\ddot{\vec{x}}$ is the acceleration, so $$\ddot{\vec{x}} = -\frac{m_g}{m_i}\,\nabla\Phi.$$ Bodies fall universally only if the ratio $m_g/m_i$ is identical for all materials. Nothing in Newton's theory requires this; it is an input, an unexplained coincidence. One might object that the two masses are simply defined to be equal, but the objection collapses on inspection: choosing units so that $m_g/m_i = 1$ for lead fixes the convention once, after which the ratio for wood, or ice, or a lump of uranium is an empirical question with a possible answer other than one. The coincidence has content precisely because it is a claim about every substance, not about a chosen standard. Antithesis: A Coincidence Under Extreme Test Perhaps the ratio is only approximately constant. Physicists quantify possible violations with the Eötvös parameter, $$\eta = \frac{2(a_1 - a_2)}{a_1 + a_2},$$ comparing the free-fall accelerations of two bodies of different materials. The form is chosen so that the numerator is the difference and the denominator twice the mean, making $\eta$ the fractional difference in acceleration — a dimensionless number, zero if the two fall identically, and roughly $\Delta a / a$ when the difference is small. Its name honours Loránd Eötvös, whose torsion-balance experiments published in 1922 first pushed the test below one part in $10^{8}$. The torsion balance remains the laboratory instrument of choice, and its principle is worth stating because it explains how such tiny differences become measurable. Two test bodies of different composition hang from a fibre; if they fall differently, the net sideways acceleration on the pair — supplied by the Earth's rotation, or by the Sun's or the Galaxy's pull — produces a twist. Crucially, the whole apparatus sits on a slowly rotating turntable, so a genuine composition-dependent effect appears as a signal oscillating at a known frequency, while the drifts and thermal noise that would otherwise swamp it do not. Using beryllium and titanium bodies, the Eöt-Wash group reported $\eta = (0.3 \pm 1.8)\times10^{-13}$ in 2008: consistent with zero at the level of about $10^{-13}$. Going further required leaving the ground, because a laboratory drop lasts a second and an orbit lasts for ever. The MICROSCOPE satellite carried two concentric cylindrical test masses, one of a platinum alloy and one of a titanium alloy, and held each centred by electrostatic forces; any difference in the forces required is a difference in the accelerations the two would otherwise have had. In orbit the pair is in continuous free fall, the satellite spins so that a violation would appear at a known frequency, and thrusters compensate the residual atmospheric drag, so the measurement can be integrated over months rather than seconds. The first results (Touboul et al., 2017) gave $\eta = (-1 \pm 9_{\text{stat}} \pm 9_{\text{syst}}) \times 10^{-15}$ — consistent with zero with a combined uncertainty of roughly $1$–$2 \times 10^{-14}$, an order of magnitude beyond the torsion-balance limits; the mission's final analysis in 2022 tightened this to $|\eta| \lesssim 1.5\times10^{-15}$. Two remarks give that figure meaning. First, its scale: had Galileo dropped his two bodies from a height of $55$ m with $\eta = 10^{-15}$, the fractional difference in distance fallen would leave them separated by about $5\times10^{-14}$ m at the bottom — some thousandth of the diameter of a single atom. Second, its reach. The reason different materials are compared, rather than different masses of the same material, is that the composition of a body's mass-energy varies from element to element. A large fraction of any nucleus's mass is not the rest mass of its constituents but binding energy of various kinds, and the proportions differ: an order-of-magnitude estimate of the electrostatic repulsion energy inside a nucleus, using the standard nuclear radius $R \approx 1.2\,A^{1/3}$ femtometres, gives about $4\times10^{-3}$ of the total mass-energy for platinum ($Z = 78$) but only about $2\times10^{-3}$ for titanium ($Z = 22$), because the repulsion grows roughly as $Z^2$ while the mass grows as $A$. If electrostatic energy gravitated even slightly differently from ordinary matter, the two cylinders would differ in $m_g/m_i$ at the level o