The Weight of Light

The Mechanism

Einstein's 1907 *equivalence principle* — the foundational insight that led him from special to general relativity — implies that a photon, having energy E and therefore equivalent mass-energy E/c², must fall in a gravitational field exactly the way an ordinary object does. The frequency of a photon rising out of a gravitational potential must *decrease* (a *redshift*: lower frequency = lower energy, the photon has "spent" energy climbing the gravitational well); the frequency of a photon falling *into* a gravitational potential must *increase* (a *blueshift*: higher frequency = higher energy, the photon has "gained" energy falling). The fractional frequency shift for a photon traversing a height *h* in a gravitational field of strength *g* is exactly *Δf / f = gh / c²*. For typical laboratory geometries this is a stupefyingly small number: across the 22.6-metre height of the Jefferson Physical Laboratory tower at Harvard University, the fractional shift is *Δf / f = (9.8 m/s²)(22.6 m) / (3 × 10⁸ m/s)² = 2.46 × 10⁻¹⁵* — about *one part in 400 trillion*. Einstein had predicted the effect in 1907 and refined the prediction in 1916. For forty-three years, nobody had figured out how to *measure* it in a laboratory. The three "classical tests" of general relativity — the perihelion precession of Mercury (Einstein 1915), the bending of starlight by the Sun (Eddington's 1919 eclipse expedition), and the redshift of solar spectral lines (predicted 1907, ambiguous solar measurements through the 1920s-50s) — had all been astronomical. The redshift in particular was the *least well-confirmed* of the three: the gravitational redshift of solar spectral lines was contaminated by Doppler shifts from solar convection and by pressure broadening in the solar atmosphere, and as late as 1959 the experimental community was openly skeptical that general-relativistic redshift had ever been cleanly observed. The 22.6 metres of the Jefferson tower at Harvard offered, in principle, the perfect controlled laboratory geometry — no Doppler contamination, no atmospheric broadening, just gravity and a vertical distance. The only problem was the *one-part-in-400-trillion* precision requirement. No spectroscopic technique then existing could measure a frequency shift of that magnitude. The breakthrough came from a different field. In 1958 the German physicist *Rudolf Mössbauer* (born Munich, 31 January 1929), then a 28-year-old PhD student at the Max-Planck-Institut für Medizinische Forschung in Heidelberg, discovered the effect that would carry his name. The *Mössbauer effect* is the recoilless emission and absorption of gamma rays by atomic nuclei embedded in a crystal lattice. Ordinarily, when a free atomic nucleus emits a gamma ray, conservation of momentum requires the nucleus to recoil backward, and the recoil kinetic energy is subtracted from the gamma-ray energy — the emitted gamma ray's frequency is therefore shifted slightly downward from the natural transition frequency. The recoil shift, for typical nuclear gamma-ray transitions in the keV range, is about 10⁻⁶ — small in absolute terms but enormous compared to the natural linewidth of the nuclear transition. Mössbauer's discovery: when the emitting nucleus is *embedded in a rigid crystal lattice at low temperature*, the recoil momentum is absorbed by the entire crystal (not by the single nucleus) — and because the crystal's mass is Avogadro's-number times larger than a single nucleus, the recoil energy shift becomes essentially zero. The gamma ray is emitted at *exactly* the natural transition frequency. Similarly, a gamma ray of exactly the natural transition frequency striking a second crystal containing the same nuclei is absorbed *resonantly* — the absorption probability follows a Lorentzian profile centered on the natural frequency with a width equal to the natural nuclear lifetime. The *Mössbauer effect* therefore gives the experimenter access to a gamma-ray frequency standard with a fractional linewidth as low as 10⁻¹³, and a frequency-comparison technique (move the absorber on a calibrated velocity-controlled platform and read the Doppler shift) that can distinguish frequency shifts of one part in 10¹⁴ or better. Mössbauer published the effect in 1958 (*Zeitschrift für Physik* 151: 124-143) and would win the Nobel Prize in Physics in 1961 — at age 32 — for the discovery. The effect was immediately recognised as the right tool to test general-relativistic redshift in a laboratory. At Harvard, *Robert Vivian Pound* (born Ridgeway, Ontario, 16 May 1919; died Belmont, Massachusetts, 12 April 2010) — then a 40-year-old Harvard physics professor and the inventor of nuclear-magnetic-resonance receiver technology (the *Pound–Knight method* used in essentially every NMR spectrometer thereafter) — read Mössbauer's 1958 paper, recognised the precision implications, and proposed to test general-relativistic redshift in the Jefferson Physical Laboratory tower using the Mössbauer effect in *iron-57*. Iron-57 has a 14.4-keV nuclear gamma-ray transition between its first excited state (lifetime 98 ns, natural linewidth ≈ 4.7 × 10⁻⁹ eV, fractional ≈ 3.3 × 10⁻¹³) and its ground state — perfectly matched, in linewidth, to the gravitational-redshift shift across 22.6 metres. Pound recruited his graduate student *Glen A. Rebka Jr.* — a 28-year-old University of California, Berkeley-trained nuclear physicist who had joined Harvard in 1958 — to build the apparatus. The Pound-Rebka apparatus, assembled in 1959 in the basement and top floor of the Jefferson Lab tower, consisted of: a cobalt-57 source (which decays to the iron-57 excited state and emits the 14.4-keV gamma ray) embedded in a thin iron foil, mounted on a calibrated piezoelectric platform that could move the source up and down at controlled velocities of a few millimetres per second (Doppler-shifting the emitted gamma-ray frequency); an iron-57-enriched absorber foil 22.6 metres away in the vertical direction; and a scintillation detector behind the absorber that counted the gamma rays passing through it as a function of source velocity. The resonance curve — the absorption probability versus source velocity — has a sharp dip at the source velocity that exactly compensates for the gravitational frequency shift between source and absorber. Reading the position of that dip on the velocity axis gives the gravitational redshift directly. The experiment was carried out in 1959 and 1960. The published result: Pound, R.V. & Rebka, G.A., "Apparent Weight of Photons," *Physical Review Letters* 4: 337-341 (1 April 1960). Measured fractional frequency shift over the 22.6-metre vertical path: *Δf/f = (2.56 ± 0.25) × 10⁻¹⁵*. Predicted from general relativity: *Δf/f = (2.46) × 10⁻¹⁵*. *Measured / predicted = 1.04 ± 0.10.* The first laboratory observation of the gravitational redshift, the last of the classical tests of general relativity to be cleanly observed, the first time anyone had weighed a photon — and the experiment took, end-to-end, less than two years. Pound and Rebka refined the apparatus in 1964 (Pound, R.V. & Snider, J.L., "Effect of Gravity on Gamma Radiation," *Physical Review* 140: B788-B803, 1965), reducing the systematic-error budget and confirming the predicted shift to 1.00 ± 0.01 — *one percent agreement with general relativity*. The Jefferson Tower experiment remains, in 2026, one of the most-cited tests of general relativity ever published. Gravitational redshift has since been confirmed by orbiting hydrogen masers (Vessot-Levine *Gravity Probe A*, 1976, to four parts in 10⁵), by frequency comparison between optical-lattice clocks at different elevations on Earth's surface (NIST 2010-2020s, to parts in 10¹⁹), and by the routine operation of the Global Positioning System satellites, whose atomic clocks would lose 38 microseconds per day to combined special- and general-relativistic effects if those corrections were not applied — civil time on Earth, in 2026, is being adjusted continuously for the gravitational redshift Robert Pound and Glen Rebka measured in a Harvard tower in 1959.