Roger Graham · Independent Researcher
In May 2026, seismologists attached 37 sensors to the Great Pyramid and found the whole six-million-tonne structure gently vibrating - about 2.3 times per second, everywhere, all the time.1 A year earlier, a disputed radar scan claimed eight shafts plunge 648 metres beneath Khafre, the pyramid next door.3 One line of arithmetic connects those two numbers, and as far as we can find, nobody has ever written it down. This page explains the machine that would exist if both are real - and exactly how to find out.
The measured one. Every large object has a note it prefers to vibrate at - flick a wine glass, it rings at its pitch; skyscrapers sway at theirs. Egyptian and Japanese seismologists measured the Great Pyramid's - the pyramid of Khufu, classically called Cheops: the entire structure responds at 2.0–2.6 vibrations per second, averaging 2.3 Hz, remarkably uniform from the deepest chamber to the outer stones.1 Too slow to hear (human hearing starts around 20 Hz), but real and continuous - the pyramid is always trembling faintly, jostled by wind, traffic and distant seismic noise.
The team used HVSR - a standard geotechnical technique that compares horizontal to vertical shaking in ambient noise to reveal a structure's natural frequency. Key detail: the ground beside the pyramid vibrates at ~0.6 Hz, very different from the structure's 2.3 Hz. That mismatch is why the paper exists - it explains how the pyramid shrugged off 4,500 years of earthquakes without resonating with the shaking soil. The authors are seismic engineers; nothing in their paper concerns lost technology.
ELGabry et al., Scientific Reports 16:14032 (2026): 37 HVSR points; 76% of structural measurements fall in 2.0–2.6 Hz; Subterranean Chamber ~1.3 Hz (bedrock response); relative amplification rises with elevation to ~4× at King's Chamber height (48.7 m), then drops to ~3× in the relieving chambers above; foundation vulnerability index Kg = 8.2. Authors explicitly flag HVSR's limits and propose operational modal analysis and FEM as follow-ups.1
The claimed one. In March 2025, an Italian team announced that satellite radar reveals eight vertical shafts under the Pyramid of Khafre (classically Chephren - the middle pyramid), descending 648 m - wrapped in spiral structures, feeding two enormous chambers, with water reported below.3 The claim is unverified, was never peer-reviewed, and radar specialists say the method cannot physically see that deep.5 Egypt's antiquities authority rejected it outright.4 This page does not treat it as true. It treats it as a premise worth stress-testing, because of what follows.
The connection. Sound travels through water at about 1,482 m/s. Divide by 648 m and you get 2.29 - within 1% of the pyramid's measured hum. A water-filled shaft of exactly that depth is, on one reading, an organ pipe tuned to the pyramid's own note.
c/L is not a fundamental. 2.287 Hz is the second harmonic of a closed-closed 648 m column (f₁ = 1.14 Hz); the physically natural closed-open configuration gives 1.72 / 2.86 Hz - missing the band by ~25%. Correcting sound speed for temperature at depth (~41 °C at the bottom via geothermal gradient) and 64 bar of pressure gives c ≈ 1,521 m/s and moves the mode to 2.35 Hz - still in band. The match survives corrections; its meaning is what's weak. Try it yourself in section 03.
Grant the premise fully - shafts real, water-filled, the system alive - and a four-stage machine assembles itself. No electronics, no moving parts. Water, stone, geometry and timing.
A 648 m water column carries pressure waves up and down. If the timing of each round trip matches the next push, the column resonates - like pushing a swing at the right moment. At that depth an ideal free water column's round trip takes ~0.44 s: a ~2.3 Hz rhythm. Real confined shafts run slower - wall compliance and even traces of gas cut the wave speed sharply.
Low-frequency vibration crosses rock with very little intrinsic loss - competent limestone is a good cable for slow vibrations over the 487 m from Khafre to Khufu. How much actually arrives depends on wave type and geology, and has never been measured.
Khufu's own natural note is ~2.3 Hz - so any arriving vibration at that frequency gets amplified, the way a singer's matched note makes a wine glass ring. The measured response grows with height - a relative spectral factor of ~4 by the King's Chamber.
The human body mechanically responds strongest at 4–8 Hz. A person lying on granite in the chamber is coupled straight into the vibration - the one payload the physics doesn't rule out.
Run the numbers on the granted machine and its menu collapses fast. Moving stone blocks: impossible - the "push" of sound at any realistic level is short of a block's weight by a factor of ten million, and a 149 m airborne wavelength can't be focused. Generating electricity from the granite's quartz: dead - the crystal grains point in random directions and cancel out. Broadcasting to the world: no - at these frequencies the signal fades within a couple hundred kilometres. Audible sound: none - rock passes only ~0.015% of its vibration into air. What's left is exactly one function: a deliberately shaped low-frequency vibration environment whose only plausible target is a person inside it. Not a power plant. An instrument you stand inside.
Energy isn't the problem (sustaining a 51×10⁶ kg column at 1 mm amplitude, Q = 100, needs ~770 W) and neither is coupling (intrinsic attenuation at 2.3 Hz over 487 m: αd ≈ 0.01–0.02; geometric spreading ~−28 dB). The problem is the drive: no identified source at Giza - natural or human - has been shown to supply a persistent narrowband 2.3 Hz drive. Microseisms peak at 0.05–0.3 Hz, cultural noise is broadband, and documented water self-oscillation (geysers, water hammer) cycles in seconds to minutes. A column that locks onto its own acoustic mode continuously would be new physics. Also note: the pyramid's 2.3 Hz peak is fully explained by broadband ambient noise filtered through the structure's own resonance - no external tone required.
An organ pipe's pitch is set by its length. Same for a water column. Drag the shaft length and watch where its resonant notes land against the pyramid's measured band (green). The claimed depth is 648 m.
Notice which marker lands in the band at 648 m with both ends hard: the second harmonic (n = 2), not the fundamental. And notice that flipping to the physically more natural closed-open configuration (a shaft opening into a chamber) empties the band entirely. The match requires choosing the right harmonic of the less likely boundary condition - and any length from roughly 585–761 m puts some mode in the band. With free choice of mode and boundary, matches are cheap. That's the numerology trap this page is built to avoid. And one deeper layer: everything this simulator computes assumes free-water sound speed in a rigid, gas-free tube. A real confined shaft's wave speed depends on wall compliance, connectivity and entrained gas - tiny gas fractions slash it dramatically - so the alleged system isn't characterized well enough to be assigned a natural frequency at all.
Here's the twist that deflates the mystery - and then rebuilds a stranger one. A structure's natural note comes from its height and stiffness: taller means deeper, like longer guitar strings. Run the standard formula on the pyramid itself:
With completely ordinary masonry stiffness (~1,280 m/s - mid-range for jointed limestone blocks), the pyramid's own height predicts its measured hum. No shafts, no water, no machine. This is the strongest argument that the whole 648 m match is coincidence: the number being "explained" never needed explaining. But now press the Menkaure button. The smallest pyramid (Mycerinus) lands near 4.6–5 Hz - almost exactly one octave above the twins. Two matched bells and one tuned an octave up: that's not evidence of design (it falls straight out of their heights), but it's exactly what you'd build if you were designing a three-note instrument. Whether Menkaure really sounds that note is measurable - and nobody has measured it.
Fixed-free shear beam: f₁ = Vs/4H. Khufu H = 139 m (current) → Vs = 1,279 m/s hits 2.30 Hz exactly; plausible jointed-masonry range is 1,000–3,000 m/s, so the measured value implies nothing exotic. Crucial corollary: fixed-free structures overtone at odd multiples only (2.3 → 6.9 → 11.5 Hz). Khufu cannot produce 4.6 Hz from its own mode family - if the "harmonic ladder" 2.3 / 4.6 / 6.9 Hz exists at Giza, the even step can only come from Menkaure. The ladder requires all three pyramids playing together.
Lay every relevant frequency band on one axis and the hypothesis's whole appeal becomes visible - three independent physical systems potentially converging on the narrow band where the human body itself mechanically responds.
Everyone points microphones at the King's Chamber. The physics points somewhere else. At 2–9 Hz the vibration wavelength in masonry is 140–565 m - vastly bigger than any 10 m room, so no low-frequency standing-wave pattern can exist inside a chamber - there is no wave-based spot to seek out. There are no magic spots within rooms; only elevation matters. And the King's Chamber's own air can't resonate below 16 Hz - it's acoustically mute in this band. The one interior space whose air reaches into single digits is the Grand Gallery: a 47 m sloped corbelled duct with computed air modes near 3.7 / 7.3 / 11 Hz. If the building is an instrument, the Gallery is the pipe and the King's Chamber is just the driven mass. And the Gallery holds one more scale coincidence worth testing: its 27 paired bench slots repeat every ~1.73 m, the half-wavelength of ~99–100 Hz sound - the register of the human voice, and of the chamber-resonance folklore. In thirty years of pyramid-acoustics attention, the Gallery has never once been properly measured.
Frequency alignment is necessary, not sufficient - dose makes experience. Published perception thresholds for whole-body vibration depend on frequency, posture and duration, and sit well below 1 mm/s in parts of the relevant band - so whether the system would be felt depends on the measured spectrum, not a single number. What the power budget constrains is strong exposure: input scales with velocity squared, so sustained large amplitudes run past the granted hydraulic budget (6.4 MW per m³/s of flow at 64 bar head - itself conditional on a continuous inlet/outlet, which nothing in the claims provides). Realistic output: modest, potentially perceptible - not overwhelming. In any altered-state stack (darkness, enclosure, ritual, expectation), the machine would be one modest ingredient. The famous coffer contributes nothing at these frequencies: stone-on-stone contact stiffness pins its rigid-body modes into the kHz range; its celebrated "ring" is plate flexure at ~70 Hz–kHz, an unrelated phenomenon. At 2.3 Hz it's a three-tonne granite passenger riding the floor.
The existing conversation - scientific and fringe - runs on about a dozen terms. Tap any card. Plain meaning first; the Engineer depth adds the precise version.
Every object has notes it "wants" to vibrate at. Push it at that rhythm and small pushes build into big motion - a swing, a wine glass, a bridge in wind.
PreciselyAt a natural frequency, energy input phase-matches stored energy; amplitude grows until damping losses equal input. Growth factor at resonance ≈ Q.
The deepest note an object makes is its fundamental; harmonics are the family of higher notes above it, like a bugle's bugle-call notes.
PreciselyMode family depends on boundary conditions: fixed-free structures produce odd multiples only (1×, 3×, 5×). This single fact reshapes the Giza hypothesis - see section 04.
When waves bounce back and forth in a confined space, some spots barely move (nodes) and some swing hard (antinodes) - like the still and wild points of a shaken jump rope.
PreciselyRequires the container to be comparable to the wavelength. At 2.3 Hz in masonry (λ ≈ 565 m), no 10 m chamber can host one - which kills all "sweet spot in the room" claims below ~80 Hz.
What the ends of a pipe or string are doing - clamped, free, open, capped. It decides which notes exist. A pan flute and an open flute of equal length play different notes for exactly this reason.
PreciselyClosed-closed/open-open: f = nc/2L. Closed-open: f = (2n−1)c/4L. The 648 m match exists only under the first family, at n = 2.
The measuring trick behind the 2.3 Hz discovery: record a structure's constant faint trembling and compare horizontal to vertical shaking. The ratio peaks at the structure's natural note. No shaking machine needed.
PreciselyHorizontal-to-Vertical Spectral Ratio (Nakamura's method). Reveals dominant site/structure frequency; does not resolve full mode shapes - which is why the 2026 authors propose operational modal analysis as follow-up.1
Sound too slow for human ears - below about 20 vibrations per second. Elephants and whales use it; storms and surf produce it. You feel strong infrasound in your chest rather than hear it.
PreciselyPerception threshold rises steeply below 20 Hz: ~120 dB SPL needed at 2 Hz. The granted machine's airborne output (~80 dB) is imperceptible; any felt effect must arrive through the floor, not the air.
How long a thing rings. A bell has high Q (rings for ages); a book has low Q (thud). High Q also means a resonator amplifies more but only in a narrower band.
PreciselyQ = 2π × stored energy / energy lost per cycle. Limestone at seismic frequencies: Q ~ 50–200, which is why 2.3 Hz crosses the plateau nearly unattenuated.
The bang in old plumbing when you shut a tap fast - moving water doesn't like stopping, and the pressure spike hammers the pipes. Older Giza theories (Cadman) proposed the pyramid's underground chamber as a deliberate water-hammer pump.
PreciselyJoukowsky surge: ΔP = ρcΔv. Cadman's pulse-pump model cycles at ~0.5 Hz - a different regime from the 2.3 Hz acoustic-resonance reading, and often conflated with it.
Blow across a bottle top and it hums at one note set by the bottle's size. Cavities with necks do this to moving air - possibly including the pyramid's shafts in wind.
Preciselyf = (c/2π)√(A/VL₁). Danley's 1990s claim of persistent chamber infrasound from wind across shaft mouths12 would make the pyramid a passive wind instrument - the cheapest live hypothesis on this page, and testable with one windy-day recording.
Satellite radar that maps the ground in fine detail. The 648 m claim uses a controversial extension: reading tiny surface vibrations and processing them into deep 3-D images - like claiming to "hear" the basement through the roof.
PreciselySynthetic Aperture Radar Doppler tomography (Biondi & Malanga).2 Critics' core objection: X-band (~3 cm) EM waves attenuate within metres in rock; the claimed 648 m requires the unproven micro-motion inversion to work perfectly. Method paper exists for Khufu (2022, peer-reviewed); the Khafre claims were press-released only.3,5
Squeezing water through porous rock generates faint electrical signals - real, measurable geophysics used in oil exploration. A vibrating water-filled shaft system would wrap the plateau in a weak electric "hum" at the same rhythm.
PreciselyElectrokinetic coupling at the pore scale. Amplitude is strongly site-dependent (porosity, permeability, water chemistry, saturation) and cannot be predicted generically; the test is coherence between synchronized seismic and electric recordings, which would support wet coupling - its absence would weaken, not refute, a saturated model.
Your body is a stack of springy masses - organs, spine, chest. Shake the floor at 4–8 cycles per second and the body responds most strongly. It's an occupational-safety field, studied for truck drivers and pilots.
PreciselyISO 2631 / Griffin: seated/standing vertical transmissibility peaks ~4–8 Hz (viscera), secondary 8–10 Hz; supine responses sit in a similar low-single-digit band.10 The literature treats prolonged exposure as hazard, not therapy - dose and duration are everything.
Every claim downstream of section 02 stands on these. If any one falls, the machine reading falls with it - the measured facts (the 2.3 Hz hum, the bell physics) survive regardless.
Sole source: a never-peer-reviewed 2025 press announcement using a method mainstream radar physicists say cannot reach that depth.3,5 Egypt's antiquities authority denies any scan was authorized.4
Measured groundwater under the plateau is 4–7 m deep - about 100× shallower than the claimed system; this neither supports nor excludes deeper confined fluids, but provides no direct evidence for water at 648 m.11 The claim's own "water at 1,200 m" is part of the same unverified announcement.
No documented natural process produces a sustained narrowband tone at this frequency. This is the hypothesis's hardest physics gap, independent of archaeology.
Attenuation math permits it; whether meaningful amplitude arrives is unmeasured.
Even if every frequency aligned perfectly, alignment can't distinguish "designed instrument" from "consequence of building three big pyramids from the same stone." Only converging independent predictions - section 08 - can move this.
Ranked by information per dollar. All are non-invasive and legal with ordinary site permission. None has ever been run.
Repeat the 2026 HVSR survey on the smallest pyramid. The bell formula predicts ~4.6–5 Hz from its height alone.
The 2026 team's data already answers two questions their paper didn't ask: does Khufu show its predicted odd overtones (~6.9, 11.5 Hz), and is there energy at 4.6 Hz that its own physics can't produce?
A 0.5–150 Hz survey along the Gallery with simultaneous floor accelerometry - windy day versus still day, testing the passive wind-instrument hypothesis for free. The upper band carries a second, sharper question: the Gallery's 27 paired bench slots sit at 1.72–1.73 m spacing, which is exactly the half-wavelength scale for ~99–100 Hz sound. If the missing inserts once formed a strong periodic array, physics predicts a stopband there - how wide and how deep depends on the inserts' scattering strength, which is exactly what modeling must determine; empty scaffolding sockets predict only a diffuse ripple. Both spacing and length are survey facts - nothing was tuned to hit the number. (The half-wavelength reading of the slots is prior art: M.J. Collins, 202613; the quantitative prediction and this discriminating test design are what's new here.)
Three accelerometers. Closes the 30-year-old "is the coffer specially coupled?" question the entire pyramid-power genre rests on. Prediction from wavelength physics: no difference below ~80 Hz.
~20 seismometers mapping whether the plateau has standing-wave structure as a site, plus electrode pairs hunting a coherent 2.3 Hz seismoelectric residue - the signature that distinguishes wet, water-coupled resonance from dry stone.
Confirmed shafts would establish geometry, not function. The decisive follow-up is hydraulic: excite the water column gently, measure its actual wave speed, damping and reflections directly, and check for a hydraulic mode coherently coupled to a structural mode in the monuments above.
Best current estimate: ~90/10 that the 648-metre match is a numerical coincidence. The measured hum is fully explained by the pyramid's own height and stone. The match requires picking a convenient harmonic of a convenient boundary condition of an unverified depth. And no known process could power the machine continuously.
What keeps the file open rather than closed: the three-pyramid octave structure, the Grand Gallery's uniquely low air modes, and the overlap with the human body's mechanical band are each independently checkable - and every single check is cheap, legal, and unrun. A hypothesis this testable doesn't deserve belief. It deserves measurement.
If the tests come back ordinary - and they probably will - what remains is still remarkable: the tallest building humans would manage for 3,800 years, whose stones happen to ring together as one bell, tuned by nothing but their own weight to the same slow rhythm as the planet's quietest noise. The mountain hums either way. The only question is whether anyone meant it to.
Companion preprint: Graham, R. (2026). A Falsifiable Resonant-Systems Hypothesis for the Giza Pyramid Complex. Independent research preprint, v1.0. DOI: 10.5281/zenodo.21965056 - the full audit with evidence classes, calculations, and pre-registered predictions.
Compiled 16 August 2026 by Roger Graham (Independent Researcher) from a multi-lane adversarial research audit. Amber-tagged items are premises under examination, not findings. The most likely outcome of every test on this page is "ordinary physics" - and that is exactly why they're worth running.