Sagnac Effect

The intuition — a race around a spinning track

Imagine a circular racetrack with one runner sprinting clockwise and an identical runner sprinting counter-clockwise, both leaving the same start line at the same instant and moving at the same speed. On a still track they return to the start together. Now spin the track. The start line moves. The runner going the same way as the spin has to chase a finish line that is fleeing from them, so they cover slightly more ground; the runner going against the spin meets a finish line rushing toward them and covers slightly less. They arrive at different times.

Replace the runners with light. Split a single beam in two, send the halves in opposite directions around a closed loop of mirrors or fiber, and recombine them. On a stationary loop they recombine in step. Spin the loop and the co-rotating beam takes a longer path than the counter-rotating beam — they come back out of phase, and the interference pattern shifts. That phase shift, proportional to how fast the loop spins, is the Sagnac effect.

The remarkable part: the light's own speed is unchanged in either direction — that is the postulate of relativity. What changes is the path length each beam must traverse before the moving detector catches it. The effect needs no outside landmark; it reads absolute rotation directly.

How it works — chasing a moving detector

Consider a circular loop of radius r rotating with angular velocity Ω. Both beams start at a beam-splitter that is itself carried around with the loop. In the lab frame, light travels at c both ways, but the beam-splitter has moved by the time each beam comes around.

The co-rotating beam must travel the full circumference plus the extra arc the splitter advanced during transit; the counter-rotating beam travels the circumference minus the arc the splitter advanced toward it. To first order the two transit times are:

t₊ = 2πr / (c − Ωr)        (co-rotating, longer)
t₋ = 2πr / (c + Ωr)        (counter-rotating, shorter)

The difference, keeping terms to first order in Ω (since Ωr ≪ c always), is:

Δt = t₊ − t₋ = 4πr²Ω / c² = 4AΩ / c²

where A = πr² is the area enclosed by the loop. This is the deep result: the time difference depends on the enclosed area, not the radius or perimeter separately, and the same Δt = 4AΩ/c² holds for any planar loop shape — triangle, square, circle.

The governing equations

Light of wavelength λ has period T = λ/c, so the optical phase difference between the recombining beams is:

ΔΦ = 2π · Δt / T = 2πc·Δt / λ = 8πAΩ / (λc)

This is the canonical Sagnac phase shift. Each full fringe of shift corresponds to ΔΦ = 2π. Notice what is absent: there is no refractive index n in the result. Even though light moves slower inside glass fiber, the medium drags partially with the rotation (Fresnel drag) in just the way that cancels n from the leading term. That is why a fiber-optic gyro and a vacuum loop of the same area read the same rotation.

For an active ring laser gyroscope, the readout is even cleaner. The ring is a laser resonator; the two counter-propagating modes must each fit a whole number of wavelengths around the cavity, so rotation forces them to slightly different frequencies. The beat frequency between them is:

Δf = 4AΩ / (λP)            (P = perimeter of the cavity)

The combination 4A/(λP) is the scale factor — a fixed geometric constant. Count beat cycles and you integrate rotation angle directly: a ring laser gyro is, in effect, a digital angle counter.

How big is the effect? Worked numbers

The factor c² in the denominator makes the Sagnac shift tiny — which is exactly why it took until 1913 to measure and why modern gyros are marvels of engineering. Take a HeNe ring laser (λ = 633 nm) and work an example.

Scenario	Area A	Rotation Ω	Result
Earth's spin, 1 m² loop	1 m²	7.29×10⁻⁵ rad/s	Δt ≈ 3.2×10⁻²¹ s
Same, as phase shift (633 nm)	1 m²	7.29×10⁻⁵ rad/s	ΔΦ ≈ 9×10⁻⁶ rad
RLG beat freq (P = 4 m, A = 1 m²)	1 m²	7.29×10⁻⁵ rad/s	Δf ≈ 115 Hz
Aircraft turn, navigation FOG	50 m² effective	0.1 rad/s (≈6°/s)	ΔΦ ≈ 0.66 rad ≈ 0.1 fringe
G ring, Wettzell (16 m²)	16 m²	Earth rate (×sin 49° latitude)	Δf ≈ 348 Hz, resolves 10⁻⁹ of Ω⊕

The lesson of the table: a bare 1 m² loop gives a phase shift of only ~9 microradians for Earth's rotation — barely measurable. The trick to practical gyros is multiplying area. A fiber-optic gyro winds the same fiber into N turns; the effective area becomes N·A. Wind 1 km of fiber into a 0.1 m radius coil (≈1600 turns) and you multiply the single-loop sensitivity ~1600-fold without enlarging the device.

RLG vs FOG — two ways to read the shift

Property	Ring Laser Gyro (RLG)	Fiber-Optic Gyro (FOG)
Type	Active — the ring is the laser	Passive — external laser + fiber coil
Readout	Beat frequency Δf (frequency)	Accumulated phase ΔΦ
Sensitivity boost	Large mirrored cavity	Many fiber turns: A → N·A
Failure mode	Lock-in dead band at low Ω	Polarization & temperature drift
Workaround	Mechanical dither (~400 Hz)	Phase modulation, PM fiber
Moving parts	Dither motor (tiny)	None
Typical use	Airliner inertial reference (Boeing 757+)	Missiles, drones, satellites, robotics
Best bias stability	<0.001 °/hr (navigation grade)	0.0001–0.01 °/hr

Lock-in — the gyro's blind spot

A ring laser gyro has one nasty failure: at very low rotation rates the two counter-propagating laser beams couple to each other through microscopic backscatter off the cavity mirrors and pull to a common frequency. The beat note collapses to zero and the gyro reads zero rotation even though the platform is slowly turning. This dead band is called lock-in, and it typically spans a few hundredths of a degree per second.

The standard cure is mechanical dithering: the entire gyro block is rotationally oscillated back and forth a few hundred times a second through a fraction of a degree. The instantaneous rotation rate then almost never lingers in the dead band, and the known dither motion is subtracted out in software. This is why a navigation-grade RLG, despite being all-solid-state optics, emits a faint mechanical hum — that is the dither. Fiber-optic gyros, being passive, have no lasing modes to lock and so sidestep the problem entirely, which is one reason they have taken over the low-cost and low-rate market.

Where the Sagnac effect shows up

• Inertial navigation. Three orthogonal ring laser or fiber gyros plus three accelerometers form an inertial measurement unit. Integrate rotation to track heading; integrate acceleration to track position — no GPS, no external signals. This guides airliners, submarines, ballistic missiles, and Mars rovers through GPS-denied environments.
• Spacecraft attitude control. Star trackers fix orientation occasionally; gyros bridge the gaps with sub-arcsecond rate sensing for pointing telescopes and antennas.
• Geodesy and seismology. Giant ring lasers (the G ring at Wettzell, GINGER in Italy) measure length-of-day variations and rotational ground motion from earthquakes that translation seismometers miss.
• Tests of relativity. The Michelson–Gale–Pearson experiment (1925) used a 2 km optical loop to detect Earth's rotation, confirming the predicted fringe shift. Proposals like GINGER aim to measure the general-relativistic frame-dragging (Lense–Thirring) of Earth's spin.
• Consumer and automotive. Low-cost FOGs and their MEMS cousins stabilize camera gimbals, drones, autonomous-vehicle dead reckoning, and antenna pointing on moving platforms.
• Fiber sensing. The same Sagnac loop topology underpins distributed acoustic and strain sensors that listen along pipelines and perimeter fences.

JavaScript — computing the Sagnac shift

const c = 2.99792458e8; // speed of light, m/s

// Sagnac transit-time difference for a planar loop of area A (m^2)
// rotating at Omega (rad/s). Result in seconds.
function sagnacDeltaT(area, omega) {
  return 4 * area * omega / (c * c);
}

// Optical phase shift (radians) at wavelength lambda (m).
function sagnacPhase(area, omega, lambda) {
  return 8 * Math.PI * area * omega / (lambda * c);
}

// Ring-laser-gyro beat frequency (Hz) for cavity of area A, perimeter P.
function rlgBeatFreq(area, perimeter, omega, lambda) {
  return 4 * area * omega / (lambda * perimeter);
}

// Fiber-optic gyro: N turns multiply effective area.
function fogPhase(loopArea, turns, omega, lambda) {
  return sagnacPhase(loopArea * turns, omega, lambda);
}

const EARTH_RATE = 7.292115e-5; // rad/s
const LAMBDA = 633e-9;          // HeNe wavelength

// 1 m^2 loop sensing Earth's rotation
console.log('Δt  =', sagnacDeltaT(1, EARTH_RATE).toExponential(2), 's');     // ~3.25e-21 s
console.log('ΔΦ  =', sagnacPhase(1, EARTH_RATE, LAMBDA).toExponential(2), 'rad'); // ~9.7e-6 rad

// Square ring laser, side 2 m: A = 4 m^2, P = 8 m
console.log('Δf  =', rlgBeatFreq(4, 8, EARTH_RATE, LAMBDA).toFixed(1), 'Hz'); // ~230 Hz

// Navigation FOG: 0.0079 m^2 coil, 1600 turns, modest 0.1 rad/s turn
console.log('FOG ΔΦ =', fogPhase(0.0079, 1600, 0.1, LAMBDA).toFixed(2), 'rad');

Misconceptions and edge cases

• "It violates the constancy of the speed of light." No. Light travels at c in both directions. The asymmetry is in the path length the beams must cover before the moving recombiner catches them, not in the speed.
• "It's purely classical." Not quite. The rotating frame is non-inertial. The leading-order ΔΦ = 8πAΩ/(λc) comes out the same from classical EM, special relativity, and general relativity, but it is not a naïve Galilean result.
• "Longer fiber means more signal." Only if that length encloses more area. A 1 km fiber stretched in a straight line and folded back encloses almost no area and gives almost nothing; the same fiber coiled into N turns gives N×A. Area, not length, is king.
• "It depends on the glass's refractive index." To leading order it does not — Fresnel drag cancels n. This is the famous index-independence that makes FOGs work in glass at all.
• "It measures velocity through space." No — only rotation. A Sagnac loop in uniform straight-line motion (an inertial frame) reads exactly zero, consistent with relativity. It is blind to constant velocity but exquisitely sensitive to spin.
• "Lock-in is a fundamental limit." It is a practical artifact of mirror backscatter in active rings and is engineered around with dither; passive fiber gyros do not suffer it at all.