Measurement of the Laboratory's Absolute Velocity
| Scientific Paper | |
|---|---|
| Title | Measurement of the Laboratory's Absolute Velocity |
| Read in full | Link to paper |
| Author(s) | Stefan Marinov |
| Keywords | Measurement, velocity, absolute |
| Published | 1980 |
| Journal | General Relativity and Gravitation |
| Volume | 12 |
| Number | 1 |
| No. of pages | 10 |
| Pages | 57-66 |
Read the full paper here
Abstract
The report is given on a local measurement of the absolute velocity of a laboratory. This is the resultant velocity due to all types of motion in which the laboratory takes part (about the Earth's axis, about the Sun, about the galactic center, about the center of the cluster of galaxies).
Overview
This is Stefan Marinov's report of his interferometric "coupled-mirrors" experiment, received by General Relativity and Gravitation in June 1979 and published in 1980 — one of the very few claims of a positive ether-drift measurement to appear in a mainstream relativity journal. Marinov reports that his apparatus in Sofia registered an absolute velocity of the Earth of 279 ± 20 km/s in July 1975 and 327 ± 20 km/s in January 1976, giving a Solar-system absolute velocity of 303 ± 20 km/s toward an apex at right ascension 14h17m, declination −23°.
The design idea is to measure the Sagnac Effect along a straight line rather than around a closed loop. Marinov notes that Harress (1912), Sagnac (1913) and Michelson, Gale and Pearson (1925) established a direction dependence of light speed on a rotating disk and on the spinning Earth, but always for closed paths, where the effect goes as the angular rotational velocity. His variant is proportional to the linear rotational velocity, and so, he argues, registers the laboratory's translational absolute velocity — the galactic and supergalactic components — not merely its diurnal rotation. Making this work requires establishing "Newtonian time synchronization" between two spatially separated points without using light, which he achieves with a rotating axle. His conclusion is explicit: the experiments "impel the scientific community to definitely reject the principle of relativity as not adequate to physical reality and restore the aether model of light propagation."
The experiment
The apparatus
A shaft of length d carries at each end a disk of radius R with a mirror (RM1, RM2) fixed on the rim. Light from a source S1 (or S2) is split at a semi-transparent mirror SM1; the transmitted beam runs the length of the apparatus, reflects off the far rotating mirror and returns, while the reflected beam bounces off the near rotating mirror. The two recombine at an observer O1. A mirror-image arrangement feeds a second observer O2 travelling in the opposite sense, which Marinov calls "direct" and "opposite." Slits (initially shutters gated by the shaft itself, opening for about 10−6 s) admit light only when the rotating mirrors are perpendicular to the incident beams.
If the light takes time d/(c ∓ v) to cross, the far mirror has turned by an extra angle. Writing δ for the angle between the radii of the mirror at rest and at speed c, and α for the additional angle when the light speed is c ∓ v,
- δ ± α = dΩ/(c ∓ v) (1),
from which, for v « c, α = Ωdv/c2. The path difference between the presence and absence of an "aether wind" is then
- Δ = 2αR = 2dRΩv/c2 = 2dvrv/c2 (2),
with vr = RΩ the rim speed. In wavelengths, with Ω = 2πN,
- z = Δ/λ = 4πdRNv/λc2 (3).
Detection
Rather than counting fringes, Marinov makes the two rotating mirrors exactly parallel so that each photodetector is illuminated uniformly, and puts two photoresistors in opposite arms of a Wheatstone bridge; because the changes in the two interference patterns are exactly opposite, the bridge reads their difference. He emphasises that this interferometric variant, unlike his earlier "deviative" coupled-mirrors experiment of 1973, is insensitive to small drifts in the rotation rate. With the sensitivity greatest at half-maximum illumination (φ = π/2), ΔW/W = πΔ/λ, giving the working formula
- v = (λc2/4π2dRN)·(ΔW/W) (5).
The procedure is to null the bridge with the illumination set to its average value, then rotate the whole platform from a position perpendicular to the absolute velocity to one parallel with it, and transfer resistance ΔW between arms to restore the null.
Numbers and results
The apparatus parameters are d = 140 cm, R = 40.0 cm, N = 120 rev/s, λ = 633 nm (He–Ne laser). A resistance change of δW = 8×10−4 W was the smallest discernible against galvanometer fluctuations, giving a resolution δv = 17 km/s, rounded up to 20 km/s for safety. Marinov states plainly that "the experiment was not performed in vacuum" and that "the room was not temperature controlled," asserting that reasonable thermal and density disturbances of the air cannot introduce errors larger than the accepted one. The whole platform rotates in the horizontal plane and a measurement takes a couple of seconds.
The method for extracting the apex is to find, over a whole day, the moment at which the bridge balances with the axis east–west — when the absolute velocity lies in the meridian plane — then turn the axis north–south and measure. Two such readings 12 hours apart give va = v sin(δ − φ) and vb = v sin(δ + φ) for laboratory latitude φ and apex declination δ, and equations (8) invert these for v and δ. On 12 July 1975 in Sofia he registered va = −260 ± 20 km/s and vb = +80 ± 20 km/s, giving v = 279 ± 20 km/s, δ = −26° ± 4°, α = 14h23m. Six months later, on 11 January 1976, va = −293, vb = +121 km/s, giving v = 327 ± 20 km/s, δ = −21° ± 4°, α = 14h11m. The mean of the two epochs is offered as the Sun's absolute velocity, equation (13).
Marinov then sets this beside Wilkinson and Corey's 1978 figure from the Cosmic Microwave Background anisotropy — v = 320 ± 80 km/s, δ = −21° ± 21°, α = 12h ± 1h — and declares it "beyond doubt" that the two are the same physical quantity.
Note added in proof
An appended note answers the objection Marinov says was raised at every lecture he gave: is the registered effect merely a non-inertial rotational effect? He replies with an "Archimedean" argument that any uniform velocity can be regarded as rotation about a sufficiently distant point, so that all motion is non-inertial and the distinction cannot save the relativity principle. He records that Prof. P. Bergmann wrote to him, "I affirm that your 'coupled-mirrors' experiment must give a null result, and the effects registered by you are due to side causes," and that Marinov offered $500 if Bergmann would publish that opinion; he heard no more. He also names Prokhovnik's published criticism that a "twist" in the rotating axle would annihilate the effect — Marinov's term is the "Lorentz twist" — and answers that his experiments "undoubtedly show that such a hypothetical 'Lorentz twist' does not exist."
Assessment
The design is genuinely ingenious, and its central idea is the right one for the problem it sets. Any experiment that sends light out and back along the same path measures only the round-trip speed, which is why Michelson–Morley is a second-order null test; to get at a first-order, one-way anisotropy you must establish simultaneity at the two ends by some non-optical means. Marinov's rotating axle is a serious attempt at exactly that, and the differential Wheatstone-bridge readout — with the two counter-propagating channels changing in opposite senses — is a sound way to reject common-mode drift. The paper is also unusually forthright about its own conditions, stating outright that there was no vacuum and no temperature control.
The algebra checks. Equation (1) expands to δ = dΩ/c and α = Ωdv/c2; equation (2) follows; equation (3) follows from Ω = 2πN. And the quoted resolution is right: λc2/(4π2dRN) with λ = 633 nm, d = 1.40 m, R = 0.400 m, N = 120 s−1 is 2.15×107 m/s, and multiplying by δW/W = 8×10−4 gives 17.2 km/s, exactly as stated.
Putting those same numbers back into the paper's own equation (3), however, shows how small the thing being measured is. For v = 300 km/s the predicted path difference is z = 4.4×10−3 wavelengths — an optical path difference of 2.8 nm — and the claimed 17 km/s resolution corresponds to detecting 2.5×10−4 of a fringe, about 0.16 nm of path. This is the crux, and two quantitative consequences follow that the paper does not address.
First, the air. A 2.8 nm path difference over a 1.4 m arm is a fractional optical-path change of 2×10−9. With dn/dT for air near 9×10−7 per kelvin, a differential temperature of roughly 2 millikelvin between the two beam paths reproduces the entire claimed signal. Marinov asserts that "it is easy to calculate" that thermal and density disturbances cannot exceed his error bar, but does not show the calculation, and in an uncontrolled room containing a 140 cm shaft spinning at 120 rev/s — a substantial stirrer and heat source — millikelvin uniformity between arms is not plausible. Rotating the whole platform, which is the operation that generates the signal, is precisely the operation most likely to change the thermal geometry.
Second, the shaft. The angle the measurement rests on is α = Ωdv/c2 = 3.5×10−9 radians for v = 300 km/s — 1.4 nm of rim displacement at R = 40 cm. Any orientation-dependent torsion, bearing play or gravitational sag in a 1.4 m shaft that changes the relative phase of the two disks by three and a half nanoradians as the platform turns will produce the full effect. This is exactly Prokhovnik's objection, and Marinov's answer — that his experiments "undoubtedly show" the twist does not exist — is an assertion, not a measurement. No independent monitoring of the shaft's torsional phase is reported. Given a mechanical requirement at the nanoradian level, the burden is the other way.
The agreement with the Cosmic Microwave Background dipole, which is the paper's strongest rhetorical card, does not survive better data. Against Wilkinson and Corey's 1978 value, with its ±21° declination error and ±1h in right ascension, Marinov's apex looks compatible. The dipole is now known to far higher precision: COBE, WMAP and Planck give 369.8 km/s toward RA 11h11m, declination −6.9°, with sub-percent uncertainty. Marinov's apex at RA 14h17m, declination −23° is about 48° away on the sky — more than ten times his own stated uncertainties of ±20m and ±4° — and his 303 km/s is 67 km/s low. Two quantities that were "beyond doubt the same physical quantity" in 1980 are now demonstrably different.
Nor was the result reproduced. Marinov's own 1973 "deviative" run gave veq = 130 ± 100 km/s where the 1975 apparatus implies 251 km/s, a discrepancy he describes as leaving him "even surprised that our very imperfect deviative 'coupled-mirrors' experiment led to such relatively good results" — a generous reading of a factor of two. Meanwhile modern one-way and isotropy tests, including rotating optical-resonator experiments, bound any anisotropy in the speed of light at the 10−17 level, far below what Marinov reports.
Finally, the "note added in proof" argues past its objectors rather than answering them. The question Bergmann and Prokhovnik raised was concrete and instrumental: does the apparatus's own rotation, through shaft torsion or otherwise, generate the signal? Marinov answers with a philosophical claim that all motion is ultimately rotational, which leaves the instrumental question untouched. Read charitably, this paper is a careful and honest description of an experiment operating three orders of magnitude beyond the stability its own construction can plausibly guarantee, whose headline agreement with the microwave dipole was a coincidence of large error bars. Its historical interest is real; its measurement has not stood.