The Behavior of Light from Extraterrestial Sources (English Translation)
| Scientific Paper | |
|---|---|
| Title | The Behavior of Light from Extraterrestial Sources (English Translation) |
| Read in full | Link to paper |
| Author(s) | Walter Rella, Rudolf K A Tomaschek |
| Keywords | light, ether |
| Published | 1924 |
| No. of pages | 15 |
Read the full paper here
Abstract
The problem addressed by the present work refers to a sequence of ideas, which have been worked out recently by Lenard, namely, whether it might be possible to demonstrate an absolute reference system, the primordial ether, using Michelson's interference experiment (M.I.). The negative outcome of Michelson's experiment in the usual setting, namely, with terrestrial light, has shown, according to Lenard's interpretation, that the ether carrying terrestrial light waves should be considered as taking part in the movement of the earth. This, however, needs not be the case with extraterrestrial light as Lenard has shown. It could be that light quanta from extraterrestrial sources running with the primordial ether were still linked on the earth to the primordial ether, which should lead to an at least partially positive outcome of the M.I. with extraterrestrial light. A negative outcome should permit elucidating the behaviour of light quanta entering the ether of different property. In the following I will report on experiments conducted with extraterrestrial light.
Originally published as "Über das Verhalten des Lichtes außerirdischer Lichtquellen," Annalen der Physik, V73, pp. 105-126 (1924).
Overview
This is Walter Rella's English translation of Rudolf Tomaschek's 1924 Annalen der Physik paper, an experimental report from the Radiological Institute of the University of Heidelberg written under the direction of Philipp Lenard. It is a real interferometer campaign, carried out at the Königstuhl observatory above Heidelberg between March and July 1923 and submitted on 30 July 1923, and it belongs to the small class of Michelson-type experiments performed with starlight rather than a laboratory lamp.
The motivating idea is Lenard's two-tier aether. On Lenard's reading, the null result of the ordinary Michelson-Morley Experiment shows that the ether carrying terrestrial light is dragged along with the Earth. But light quanta arriving from outside — from the Sun, Moon, planets and fixed stars — might still be coupled to the undragged "primordial ether" (Uräther), in which case an interferometer fed with extraterrestrial light should show a fringe shift where one fed with terrestrial light shows none. Tomaschek set out to test exactly this, and to use whichever answer came out to learn how a light quantum behaves when it crosses from one ether into another. The result is null: no differential shift is found for sunlight, moonlight, Jupiter, Sirius, Arcturus or Vega.
The experiment
Design
Tomaschek began from the Morley–Miller layout with a 12 m light path folded by eight mirrors, and abandoned it: the available starlight was simply too faint. He adopted instead a differential design. Rather than looking for an absolute fringe shift, the apparatus alternately admits extraterrestrial light and a terrestrial reference lamp and looks only for a difference between the two fringe positions. The null result of the ordinary Michelson experiment is thereby presupposed, and only the differential behaviour is measured. This is a considerable gain in sensitivity for the specific question at issue, since it makes the measurement immune to anything that affects both beams equally.
Rotation of the apparatus is replaced by the rotation of the Earth: the two arms are fixed exactly east–west and south–north, and observations are simply taken at different times of day and night.
The interferometer
Three pillars — two sandstone, one hard clinker — were founded on bedrock in the cellar of the observatory's eastern building, 1.5 m below the exterior soil level, 8.6 m apart, 80 cm high with a 60 cm square section. P1 and P2 carry 45 mm silvered glass mirrors on brass plates adjustable by three micrometer screws each; P3 carries the unsilvered separating and compensating plates, 10 × 5 × 1.5 cm. One slide has a micrometer screw of 1 mm per revolution, adjustable in practice to a few wavelengths of light; the other is adjusted without any mechanical touch at all, by two electromagnets that flex the iron support minutely north or south under the control of a resistor — a deliberate measure to avoid vibration and air disturbance.
Fringes are read through a 6 cm, 40× telescope with a reticule of two parallel threads, set slightly less than half a fringe apart so that a dark fringe can be captured between them; one turn of the 60-division micrometer barrel corresponds to one fringe width. Fringe width was set to about 4 mm at the mirror, giving four or five fringes on each side of the centre, this being the best compromise against the curvature that wider fringes showed from imperfect optics.
Fighting the disturbances
The account of the noise sources is the most impressive part of the paper. Air currents made the fringes so restless that the entire beam path had to be enclosed in iron stove-pipes — open-ended, but sufficient. The observer had to enter and then wait 10–15 minutes behind a curtain "until the air flows associated with the human body became damped." The pillars were rock-anchored to about 2.5 m, and even so footsteps were detectable at 50 m; sunshine and wind raised the fringe restlessness measurably.
A slow, steady drift of the fringes of about 1/20 of a fringe width per minute persisted, attributed to soil movement from thermal gradients or from the observer's own weight — sometimes reversing before midday when the sun warmed the ground. Tomaschek's answer is the alternating measurement protocol: reference fringes read left and right, then experimental light read left and right, then reference light again, with equal exposure times, so that linear interpolation removes the drift. He then bounds the residual systematic explicitly: one sequence takes at most a minute, a single reading about 8 seconds, so drift within one reading cannot exceed 0.006 fringe widths; an implausible 4-second imbalance between the two intervals would produce an apparent shift of 0.008 fringe widths, well inside the other disturbances. A control experiment with deliberately mismatched colours — very reddish experimental against very greenish reference light — gave shifts of −0.009 ± 0.019 and −0.000 ± 0.010 fringe widths, confirming that colour mismatch does not bias the reading of the central dark fringes.
For starlight he used a faster method still: fix the stellar fringes with the reticule, then switch on the reference light and cut off the star within 1–2 seconds. This makes the comparison nearly independent of air flows and avoids the dazzle of reading the barrel. He states that a shift of 0.04 fringe widths would certainly have been detected this way.
Collecting the light
Sun and Moon light was fed in by a small clock-driven Silbermann heliostat with a single reflection — a blackened-back plane glass plate for the Sun, a silvered mirror for the Moon — and damped by a 2 mm glass filter carrying a dispersed dried soot suspension acting as a veil. The single reflection matters for the interpretation, and the geometry gave the meridian ray path proposed by Vogtherr.
For fixed stars a larger heliostat lent by the firm C. P. Goerz was used — the same instrument taken by the German expedition to the 1914 solar eclipse in Norway. Since it could rotate about only one axis, parallel to the Earth's axis, a second 35 cm mirror rotating about a vertical axis and running on 2 m of north–south rail was required; a light-amplifying telescope of 17 cm aperture then carried the beam some 25 m into the cellar. A small 2 × 2 cm mirror diverted part of the beam to a second telescope beside the heliostat so that an assistant could keep the star centred, in telephone contact with the observer below. Installing all this required excavating a 6 × 2 m trench to cellar floor level.
Results
Every measurement is null within its stated error. Solar readings, each the mean of 20 series of 6 readings over 35–45 minutes, spread from −0.010 ± 0.007 to +0.023 ± 0.006 fringe widths across April, May and July 1923. Lunar readings run from −0.007 ± 0.013 to +0.054 ± 0.014, the largest values flagged with an asterisk as taken before the air currents had settled. Jupiter gave four measurements between −0.011 ± 0.011 and +0.012 ± 0.015, with repeated immediate comparisons on April nights "always a negative result." Sirius, badly served by the weather, gave no measurable shift on 21 March and in 50 comparison observations on 4 April, with nothing exceeding 0.04 fringe widths. Arcturus gave +0.004 ± 0.005, −0.025 ± 0.012 and −0.007 ± 0.009; Vega gave +0.001, −0.003 and −0.004, each ± 0.005.
What was expected
Tomaschek works out the predicted effect carefully. If extraterrestrial light arrives from the primordial ether and retains its velocity relative to it, the expected differential shift is N = (L/λ)(v2/c2) when the relative velocity lies along an arm. He then enumerates four candidate motions.
Earth's daily rotation gives a constant positive shift far too small to detect. Earth's orbital motion is the main term, maximal at noon and midnight at +0.15 fringe widths, with diurnal variation
- N1 = (L/λ)(v2/c2)[1 − sin2t(1 + sin2φ)]
for hour angle t and latitude φ = 49°24′, vanishing at 3:40 pm and 8:20 am and reaching a negative extreme at 6 o'clock of only 0.58 of the noon value. The solar system's motion through the fixed stars, taken with apex RA 270°, Dec +30° and v0 = 20 km/s, would give at most +0.044 and −0.062 fringe widths at stated hours after culmination of the apex. Superposing the second and third terms gives expected noon and midnight deviations of +0.19 and 0.08 fringe widths, with March the best observing month. A fourth term, the motion of the Galaxy relative to other systems, is left out as unknown, though on Courvoisier's estimates it would have implied shifts of order ten whole fringe widths.
The expectation is therefore 0.1–0.2 fringe widths. The observed deviations are at most one eighth of that, already within instrumental error; measured against Courvoisier's galactic velocities they are about one thousandth of the calculated value.
Interpretation
Tomaschek's conclusion is that light quanta from extraterrestrial sources are "no longer in connection with the primordial ether of the universe", but travel at the speed of light relative to the ether of the Earth, taken to be at rest with respect to the Earth's surface. A quantum entering an ether in a different state of motion adopts the light speed relative to that ether in its direction of propagation, while the phenomenon of stellar aberration shows that the perpendicular component of its velocity is not changed. He adds, in a footnote, that if the Lorentz contraction is rejected as unsatisfactory then these experiments show directly that the ether associated with starlight exhibits no relative motion to the Earth.
Section 6 then confronts the difficulty this creates. Terrestrial and extraterrestrial light traverse the two arms in the same time — but starlight is aberrated, so its ray direction (the direction in which energy advances) differs from the direction of propagation of the quanta themselves. Tomaschek considers two ways out. The first is that starlight loses its lateral component at the heliostat mirror, "where the incoming light interacted for the first time intimately with the material atoms" — treating the lateral component not as a mere geometrical artefact of relative motion but as a real property of the quantum. This would explain the null result at once, but requires that an extraterrestrial ray entering along the same apparent direction as a terrestrial one should leave the mirror displaced by the aberration angle, contradicting the usual assumption that "relative" rays obey the ordinary reflection law; he notes no unambiguous test exists and that one is in preparation.
The second, which he develops geometrically in his Fig. 4, is that the lateral component itself obeys the reflection law. Constructing the elementary wave from the incident wave front, he shows that the angles satisfy i = i′ to second order, so that reflected starlight and terrestrial light run in the same direction — but with a different speed along the relative ray, because the lateral component is still present. Since each component of the motion keeps the same magnitude and direction relative to the quantum in both arms, both arms are traversed in equal times and the comparison with terrestrial light comes out null, even though the absolute traversal time differs from the terrestrial case. He gives the round-trip time along one arm in the best case as 2l/c(1 − (v2/c2)sin2θ), with θ the angle between the incoming ray and the Earth's motion.
He also records, citing a 1922 Physical Review result, that repeated experiments with terrestrial light on high mountains appear to show a noticeable transition to primordial ether — a shift of about one tenth of the effect expected from Earth's orbital motion — while maintaining that this does not affect his conclusion about starlight.
Assessment
This is a genuinely careful piece of experimental work, and its value does not depend on whether one accepts Lenard's ether. The differential design is the right instrument for the question asked; the error budget is constructed rather than asserted, with the drift rate measured, the systematic bounded arithmetically, and a colour-mismatch control run explicitly to close an identified loophole; and every result is quoted with an uncertainty. The candid reporting of nuisance effects — footsteps at 50 m, the observer's own body heat, the reversal of the soil drift under morning sun, the asterisked lunar points taken before the air had settled — is exactly what one wants and is rarer than it should be. The prediction is worked out in advance, with an explicit formula and a stated expected magnitude, and the experiment is sensitive to roughly a tenth of it. That is a properly designed null result, and it deserves to be better known than it is.
Its significance, though, is narrower than it is sometimes taken to be, and Tomaschek is honest about this in a way later citers often are not. Because the design is differential and presupposes the null result with terrestrial light, it cannot test for absolute motion at all; it can only test whether starlight behaves differently from lamplight. A null answer therefore refutes Lenard's specific conjecture — that extraterrestrial quanta remain coupled to the undragged primordial ether — and refutes it cleanly. It does not, by itself, discriminate between the earth-dragged ether Tomaschek retains and no ether at all, since both predict exactly the same null.
The second and more serious limitation is one Tomaschek raises himself and cannot resolve: all the light passes through a heliostat before entering the interferometer, and his own first hypothesis is that the decisive change happens at that mirror. If so, the apparatus can say nothing about the state of the starlight before reflection. His footnote that a version without prior reflection would test the point is precisely right, and the experiment appears never to have been done in that form. The single-reflection solar and Sirius runs mitigate but do not remove this, since a single reflection is still a reflection.
Where the paper is weakest is in the theoretical patchwork of Section 6. The two proposed accounts of aberration are mutually exclusive, and Tomaschek's choice between them is governed by which one saves the null result rather than by independent evidence. The first requires a novel reflection law for aberrated light for which he concedes no test exists; the second requires that the lateral component obey the ordinary reflection law while the longitudinal speed is reset to c on entering a new ether — a hybrid rule with two different prescriptions for two components of the same quantum's velocity, adopted because that combination yields equal traversal times in both arms. Nothing derives the rule; it is reverse-engineered from the data. The result is that the ether interpretation survives, but only by acquiring a mechanism specified to whatever it needs to be, whereas the kinematics being described — light propagating at c in the frame of the apparatus regardless of the source's motion, with aberration affecting direction but not speed — is precisely what special relativity says without any additional rule at all. Tomaschek's data are consistent with that reading; his framework is what forces the extra machinery.
Finally, the mountain-top result cited approvingly in Section 6, of about one tenth the orbital effect, sits uneasily beside the paper's own conclusion, and the 10-fringe-width prediction from Courvoisier's galactic velocities — reported as being contradicted by a factor of about a thousand — is a striking illustration of how far the estimates of the day could be from the observations, on either side of the argument.