Abstract
The history of earlier attempts to detect the motion of the Earth in the “world ether” by optical and electrical methods can be found in many books and articles related to the theory of relativity. There is no need to recall it once again here. An account of the most recent works in the same direction, dating to 1925, has already appeared in the pages of our journal. The present review sets out, chiefly, works concerning the “ether wind” published in 1926.
Full Text
NEW SEARCHES FOR THE “ETHER WIND”
S. I. Vavilov.
The history of the old attempts to detect the motion of the earth in the “world ether” by optical and electrical methods can be found in many books and articles connected with the theory of relativity. There is no need to recall it here once again. An account of the most recent works in the same direction, relating to 1925, has already appeared in the pages of our journal1. The present review sets forth, chiefly, works concerning the “ether wind” published in 1926.
§ 1. Dayton Miller’s New Experiments
For more than 20 years D. Miller has been engaged in repeating the Michelson experiment. In 1925 there appeared the first, comparatively detailed description of the results of these many years of experiments with an interferometer2. For details we refer the reader to the translation of Miller’s article in our journal3. In Fig. 1 a photograph is reproduced of Miller’s apparatus on Mount Wilson in the form in which the apparatus was in 1921. The interferometer floats in mercury. On the walls of the vessel with mercury are marked the azimuths according to which the readings are made. The results of Miller’s first series of experiments are seen most clearly from the following graphs. In Fig. 2 (p. 244) are presented the observations in Cleveland at a small elevation of 90 m. Along the abscissa axes are plotted the azimuths of the interferometer, and along the ordinates the displacements of the interference fringes. The maximum marked height of the ordinates corresponds to 0.1 of a fringe. As is seen from the figure, deviations exceeding 0.01 of a fringe were not observed. On the contrary, at the height of Mount Wilson (1800 m) the interference fringes (Fig. 3) sometimes shifted by a very appreciable amount (about 0.2). On the basis of these experiments Miller drew the following conclusion: “There exists a definite displacement of the interference fringes, such as would be caused by the relative motion of the earth and the ether on Mount Wilson
with a speed of approximately 10 km/sec., i.e., about one third of the Earth’s orbital velocity. In comparing this result with the earlier observations in Cleveland, the thought suggests itself of a partial entrainment of the ether, decreasing with altitude. It is thought that a reconsideration of the Cleveland observations from this point of view should show that they are in agreement with such assumptions, and lead to the conclusion that the Michelson–Morley experiment should not yield
Fig. 1. General view of the Michelson–Miller interferometer.
a null result in the exact sense of the word and, in all probability, never did yield such a result.
At the beginning of 1926 Miller published the results of a new large series of observations with the interferometer on Mount Wilson1. At the same time, he gives a substantially different interpretation of the earlier observations published in 1925. Above all, the dependence of the effect on altitude, which had been so clearly illustrated,
earlier than Figs. 2 and 3. Miller now writes: “Experiments with the ether wind have never been carried out at sea level, or anywhere at all except Mount Wilson, with the completeness sufficient for accurate measurements of the effects. New data show that the ether wind at Mount Wilson does not differ significantly in magnitude from the wind in Cleveland, and that at sea level it will probably have approximately the same magnitude.” Miller gives no explanation, in the cited communication, of the sharp difference between the diagrams in Figs. 2 and 3. Concerning the interpretation of the previously published data, Miller says the following: “Extensive calculations were made in order to reconcile the observed effects with the accepted theories of the ether and the presumed motions of the earth in space. At certain seasons of the year the observations were repeated in order to test the proposed hypotheses one after another. At the end of 1924, when no solution seemed possible, a complete calculation was made of the effects expected at that time for each month of the year. The calculation showed that the effect should be maximal around April and, moreover, that in the course of 24 hours it should make a complete circuit around the horizon. To test these predictions, observations were made in March and April 1925. The effect found proved to be equal to the largest of those previously found, but it did not pass successively around the whole horizon, i.e., it did not turn through 90° every 6 hours and did not pass into the opposite direction after 12 hours. Instead, the direction of the effect oscillated back and forth within an angle of about 60°, having on average a northwestern direction...
Fig. 2. Experiments in Cleveland.
Fig. 3. Experiments on Mount Wilson.
The circumstance that the direction and magnitude of the observed ether wind are independent of local time and constant with respect to sidereal time shows that the effect of the earth’s orbital motion is not perceptible in the observations. In the indicated observations of 1925 no effect of the orbital motion was discovered, in complete agreement with the results obtained by Michelson and Morley in 1887 and by Morley and Miller in 1905.
Thus, in Miller’s new interpretation, the displacement of the interference fringes observed by him, first, scarcely depends on altitude and, second, cannot be the result of the annual motion of the Earth. The negative result of the Michelson experiments, undertaken with the aim of detecting the annual motion, was confirmed with even greater precision than before.
Let us now turn to the effect observed by Miller. Miller describes the apparatus and procedure of the observations as follows: “The observations must be made in darkness; in daytime the room with the interferometer is darkened by screens of black paper. The observations must be performed at a temperature exactly the same as outside. The observer has to walk around a circle of about 3 m in diameter, directing his eye to the moving eyepiece of a telescope connected with the interferometer, which floats in mercury and is constantly rotated about its axis, making approximately one revolution per minute. The observer must in no way touch the interferometer, and at the same time he must not lose sight of the interference fringes, visible only through the small aperture of the telescope eyepiece, about 6 mm in diameter. The observer makes 16 readings of the fringe position during 1 revolution, at instants indicated by an electric bell. This procedure must continue uninterruptedly in a series of observations usually lasting 15–20 minutes. The observations are then repeated one after another for several hours during the working period.” The position of the interference fringes is estimated in tenths of the width of a single fringe. A relative motion of 30 km per second should cause a displacement of 1.1 fringes. Temperature and other influences, if such exist, may affect the absolute magnitude of the observed displacement, but can alter only very little the direction of the maximum displacement. Therefore Miller considers that determinations of the azimuth of the effect are more accurate than determinations of its magnitude.
In 1925 three series of observations were made, in April, August, and September. The total number of observed revolutions of the interferometer was 4,400; the number of individual readings exceeds 100,000. A group of 8 consecutive readings gives the value of the magnitude and direction of the “ether wind.” Thus, in all, 12,500 data were obtained. The results of the measurements are very clearly compared by Miller in Figs. 4 and 5. The three parts of Figs. 4 and 5 correspond to the three epochs: April, August, and September. In Fig. 4 are shown the positions of the azimuths of the maximum “wind” at various hours of the day (by local time). The thin line connecting the black points represents an “individual observation.” By this Miller means the average of 20 data obtained from 20 revolutions of the interferometer over about 15 minutes. The heavy line connecting the circled points represents the variation of the mean value of the azimuths obtained from all the “individual measure-
“...of this epoch. The zero of the ordinate axis corresponds to north; above the axis the eastern directions are plotted, below it—the western ones. From the figure the presence of a preferential north-western direction of the effect is quite clear. In Fig. 5, as before, local time is plotted along the abscissa axis. The ordinates represent displacements of the fringes, expressed in km/sec (1 km is equivalent to approximately 0.04 of the fringe width). The meaning of the thin and heavy lines is the same as in the preceding drawing. The maximum displacements are equivalent to approximately 10 km. In Figs. 4 and 5 all the experimental content of Miller’s observational results is concentrated.
Fig. 4. Changes of the azimuth of the maximum displacement of the fringes at different hours.
As was already said above, Miller’s observations do not reveal the annual motion of the earth. The systematic displacement found for the interference lines must be caused by some other reason. Miller believes that the former interference experiments as well, beginning with the experiments of 1887, gave the same systematic effect. “In all these observations,” he writes1, “there was a definite positive result, appearing as a periodic displacement of the interference fringes; but for the two precomputed times of day the phases of the positive periods differed in such a way that, when the two series of readings were combined, they neutralized one another, and the remaining very small result was quite correctly attributed to the orbital component of the existing ether wind. It is this positive effect, which was then eliminated, that is now being investigated... Every possible disturbing cause that could be suspected was exhaustively studied. Among other circumstances there were studied: diur-
...daily and annual variations of temperature, meteorological conditions, radiant heat, magnetism, magnetostriction, gravitational anomalies, gyrostatic action, the influence of methods of illumination, the influence of transparent or dark covers of the light path, the speed and direction of rotation, deficiencies of equilibrium in rotating parts, the position of the observer, and other conditions. It turned out that none of these disturbing circumstances produced the observed effects... It is very important that the effect is always observed, in each individual measurement, and not only in the average result.” Miller finds that the observed effect can be explained if one assumes that the solar system is moving toward a certain point in the constellation Draco (right ascension \(262^\circ\) and declination \(+65^\circ\)) almost perpendicular to the plane of the ecliptic. The interferometer detects a displacement of the fringes equivalent, approximately, to \(10\ \mathrm{km/sec}\); on the other hand, the annual motion of the Earth at \(30\ \mathrm{km/sec}\) is not detected. It must be supposed that the hypothetical motion of the solar system exceeds \(200\ \mathrm{km/sec}\), with only about \(1/20\) of this motion manifesting itself in optical phenomena in the form of an ether wind. If this assumption is made, then the azimuth of the maximum displacement will change as shown for three epochs in Fig. 6, in the form of smooth curved lines. The broken lines depict the observed azimuths. The theoretical curve, however, must change equally both in the western and in the eastern
Fig. 5. Change in the magnitude of the maximum displacement at different hours.
directions. Miller arbitrarily shifts the theoretical curves downward, to the west, in order to bring them into agreement with the observed results. Miller finds an analogous agreement for the change in the magnitude of the effect in comparison with the theoretical curves.
The apex, determined by Miller by optical means, is in general in agreement in direction with the astronomical data.
Fig. 6. Comparison of the observed changes in azimuth with the theoretical curves.
Miller’s communication concludes with the remark that in February 1926 a new series of observations was made (2,000 revolutions of the interferometer). In general the new observations are in full agreement with the former ones, although some differences in the numerical results are possible.
Miller’s results (Figs. 2, 3), as regards the qualitative aspect of the phenomenon of the systematic displacement of the interference fringes, are very convincing. The fluctuations of the readings in “individual observations” (which are in fact averages of 20 observations), however, are so large that the quantitative course of the phenomenon can hardly be regarded as fully established. The phenomenon discovered by Miller is new; it is not predicted by any of the existing theories. On the other hand, Miller’s interferometer is so sensitive that many local influences, difficult to take into account, may prove to be the cause of a systematic displacement of the fringes1. In this connection, above all it is desirable to repeat the interferometric and other experiments under different conditions.
§ 2. The Piccard and Stahel Experiment.
Basing themselves on Miller’s first report, in which it was indicated that the “ether wind” had been distinctly detected only at the altitude of Mount Wilson, Piccard and Stahel1 in Brussels undertook a repetition of Michelson’s experiment, ascending with an interferometer in a balloon. The first ascent took place on the night of June 20 to 21, 1926, in the balloon Helvetia (2,200 m³ of hydrogen).
Michelson’s interferometer with multiple reflection (9 mirrors), with an optical path of 280 cm, was placed in a thermostat. The light source was the mercury line 4358 Å. The interference fringes and a fixed reference point on the last mirror were continuously recorded on a moving photographic film. Azimuths were marked by light signals. The rotation of the instrument was produced by two electric motors, which imparted to the whole balloon a rotational speed of 2–3 revolutions per minute. The first observations were made at an altitude of 2.5 km. In all, 96 revolutions of the interferometer were recorded. An ether wind of 30 km/sec should have manifested itself in the apparatus as a sinusoidal motion of the fringes with a full period corresponding to a half-turn of the balloon and with an amplitude of 0.064 of a fringe width (17 times smaller than in Miller’s case). The photographs obtained were analyzed by a dividing machine. The amplitude found did not exceed 0.0034 of a fringe width ($\sim 7$ km/sec), with a probable error of the same order of magnitude. Visual observations at an altitude of 4.5 km likewise revealed no effect, but the accuracy of these measurements was still lower (0.1 of a fringe width!). Of course, the experiments described, with an insensitive instrument, are not capable either of confirming or of refuting Miller’s experiments.
§ 3. Repetition of the Trouton–Noble Experiment.
When a charged condenser moves relative to the stationary ether, from the point of view of classical theory a torque should arise:
\[ K = \frac{1}{2} C V^2 \left(\frac{v}{c}\right)^2 \sin 2\lambda \sin^2 \mu, \]
if the condenser is suspended so that the plates are in a vertical plane ($C$ is the capacitance of the condenser, $V$ the potential difference, $\frac{v}{c}$ the ratio of the speed of the earth’s motion to the speed of light, $\lambda$ the angle between the direction of motion and the plane of the plates, $\mu$ the angle between the direction of motion and the suspension thread). From the classical point of view
this rotation is equivalent to the expected displacement of the interference fringes in the Michelson experiment. The corresponding experiment was performed with a negative result in 1904 by Trouton and Noble. In 1925 and 1926, in connection with Miller’s first report, the experiment was repeated by Tomaschek1 on the summit of the Jungfrau (3457 m). In the first experiments (September–October 1925) Tomaschek succeeded in increasing the sensitivity of the apparatus approximately 20-fold in comparison with the apparatus of Trouton and Noble. Observations at various times of day revealed no effect exceeding approximately 3 km/sec. In all, about 10,000 observations were made. In the spring of 1926 a second series of observations was carried out, with the sensitivity of the instrument increased still further, by approximately 40 times. This was achieved by increasing the capacitance of the condenser, by a new suspension system, by changing the electrical shielding, etc. No effect exceeding approximately 1/2 km was found. After Tomaschek, the Trouton–Noble experiment was again repeated in America, in Pasadena, by Chase.2 Chase considers Tomaschek’s experiments unconvincing and doomed in advance to a negative result for the following reason. In Tomaschek’s apparatus the condenser was suspended (in the first experiments) on a phosphor-bronze wire, 50 cm long and 0.0015 cm in diameter. This wire served as the lead-in to one system of plates of the condenser. The charge was led to the other half by a thin wire attached to the lower end of the condenser and lowered into a solution of sulfuric acid. Chase believes that the surface tension of the solution must exert frictional forces on the immersed wire which considerably exceed the effect that can be produced in a charged condenser by motion relative to the ether. Experiments with a model, carried out as a check, confirm this supposition.
In Chase’s own experiments the condenser is stretched from above and below by thin wires, through which the charge is also supplied. No effect exceeding 4 km/sec was found at different times of day.
Thus Miller’s effect, if it is explained by the classical “ether wind,” contradicts the results of the experiments described above. In connection with the negative result of his experiments Tomaschek writes: “If the positive effect of Michelson’s interference experiment is confirmed, then this would mean that the result found in my work reveals an entirely new, completely unexpected property of electromagnetic fields associated with matter, i.e. of lines of force associated with charges, on the one hand, and of fields in a light beam (lines of force closed upon themselves), on the other.”
The dependence of the “ether wind” on altitude, the possibility of which Miller indicated in his first communication, prompted a repetition of “ether experiments of the first order,” which, after the classical work of Lorentz, had been filed away in the archives of history. Tomaschek repeated on the Jungfrau Röntgen’s experiment1 (the presumed convection current from a charged capacitor at rest relative to the earth) and the interference experiments of Mascart and Ketteler2 (the presumed difference in the times of passage of light along and against the ether wind in media with a large refractive index). The result proved negative in both cases, with a very high degree of accuracy. Thus, the phenomena “of the first order” behave at altitude in agreement both with the classical theory of a stationary ether and with the theory of relativity.
§ 4. Critique of Miller’s Interpretation of the Experiments.
The experiments set forth in §§ 2 and 3, if they do not refute Miller’s experiments, neither do they confirm them. The existence of a systematic displacement of the fringes in Miller’s experiments may be regarded as indubitable. But the question of possible unaccounted-for local influences still remains open. A comparison of the curves in Figs. 1 and 2 makes such a suspicion particularly legitimate. In any case, a repetition of the experiments in another place and with another apparatus is necessary under the existing circumstances.
Miller’s interpretation of the observed displacements as the result of an “ether wind” directed toward the constellation Draco is presented in § 1. It is assumed that the “wind” is caused by the motion of the solar system in this direction with a speed of more than 200 km/sec. As a result of the partial entrainment of the ether, the speed of such a wind is reduced to only 10 km/sec at the surface of the earth. How reliable is such an interpretation, i.e., how well does it agree with Miller’s own data? Thirring,3 at first basing himself only on Miller’s first communication (where the position of the apex had not yet been stated exactly), compared the curves of Fig. 3 with theoretical curves for all possible apexes and came to the conclusion that “the permissible errors in the individual curves must have the same magnitude as the observed effect itself, if the observations made during a day (the mean values) are to be equated with the theoretically expected picture.” Later,4 after familiarizing himself with Miller’s second communication and with the still unpublished data on the February measurements of 1926, Thirring did not change his opinion.
“Unfortunately, I must say,” writes Tirrинг, “that my opinion concerning the significance of the observed displacements differs from Miller’s opinion to such an extent that I cannot ascribe the effect to any cosmic cause... The asserted good agreement between the assumed ether wind and the observations is explained by the fact that Prof. Miller arbitrarily shifted the theoretical curves, which determine the azimuth of the wind as a function of sidereal time, in order to superpose them on the empirical curves. Such a procedure is permissible in all cases where only the form of the curves is essential. In the present case, however, the absolute magnitudes of the curves have fundamental significance. As Prof. Miller himself quite rightly observes, the projection of a definite direction in space onto the horizontal plane must oscillate equally to the east and to the west in the course of a sidereal day. In fact, in 95% of all observations the effect is observed in the northwestern quadrant. This fact seems to me fatal for the supposition of an ether wind of constant direction toward some point of the celestial sphere. If, in general, the effect is real, it must prove the existence of a northwestern ether wind accompanying the rotation of the earth. The velocity of this wind must be at least equal to 10 km/sec, whereas the velocity of the diurnal motion at the equator is only about 5% of this quantity. The defenders of the ether will find it difficult to agree to a vortex motion of the ether around the earth with a velocity exceeding the velocity of the earth’s diurnal motion by approximately 20 times... I therefore conclude that the effect cannot be ascribed to any cosmic cause and must be produced by local disturbances.” Thus the interpretation of Miller’s effect as the result of the motion of the entire solar system through the ether is open to doubt.
§ 5. Miller’s experiments and the theory of relativity.
The great interest in Miller’s work even outside physical circles is explained by the significance of these works for the theory of relativity. Many, following Silberstein1, understood Miller’s results (especially in their first version) as a refutation of the fundamental postulate of Einstein’s theory. The negative result of the Michelson experiment and others, as is well known, gave an experimental basis for the particular principle of relativity. Miller’s experiments (in their second version) quite fully and with still greater accuracy confirm the impossibility of detecting the annual motion of the earth by optical means. Consequently, even if the cosmic character of Miller’s effect were fully proved, the mathematical framework of the theory of relativity, the Lorentz transformations, remain valid with great
with a degree of accuracy. Further, the general theory of relativity has explained many apparent paradoxes connected with the incorrect application of the special principle (for example, aberration and the “ether wind” accompanying the daily rotation of the earth). The general theory by no means denies the possibility of phenomena equivalent to an “ether wind”: they correspond to the presence of special gravitational fields (in the sense of the theory of relativity). A displacement of interference fringes observed experimentally will contradict Einstein’s theory only if the existing gravitational fields are incapable of explaining it. Before refuting the theory of relativity, one must first give a detailed theory of the Michelson experiment from the standpoint of the general theory of relativity. The capacity and flexibility of this theory have not yet been exhausted, and in Miller’s experiments one must distinguish observations from interpretations (cf. § 4).
Page and Sparrow1 point out that the conclusion that Miller’s experiments detect an “ether wind” is too hasty. What is found experimentally is only a difference in the time of passage of light in two mutually perpendicular directions. One may suppose, as Page and Sparrow do, that the Miller effect is not of second order (proportional to \(\frac{v^2}{c^2}\)), but of first order, and is caused by the fact that the velocity of light is slightly different in two mutually perpendicular directions, i.e. that space is anisotropic. To explain the Miller effect from this point of view it is sufficient to suppose that the velocities differ by only \(16.7\ \mathrm{cm/sec}\). If space is conceived, after Einstein, as spherical with uniformly distributed matter, then anisotropy is excluded. “But,” write the authors just mentioned, “the distribution of matter is not arbitrary, like our space-time coordinates; it is absolute, not relative, and if it is not uniform, then one must expect an asymmetry reflected in the phenomena of nature in the fields of mechanics and optics. If the universe has a ‘size,’ why should it not have a ‘shape’?” Page and Sparrow point to certain solutions of Einstein’s law of gravitation compatible with the Miller effect. Of course, such assumptions are still very conjectural, but they are no less admissible than the “ether” interpretations of the Miller effect. If the Miller effect is not caused by local causes, then the theory of relativity faces a new important problem, but in any case it is premature to speak of a refutation of the theory.
The relation of Einstein’s theory to Miller’s experiments is, to some extent, similar to the relation of the theory of electrons to Ehrenhaft’s experiments on “subelectrons.” In both cases there is
theory, irreproachably and coherently encompassing a large number of facts, and, on the other hand, certain experiments that seem not to agree with the theory. Whether there is hidden in these experiments something new that will require a change or supplementation of the theories, or whether, on the contrary, in the final analysis the theory will absorb the facts—the future will show. “The history of science,” says Lodge1, regarding Miller’s experiments, “has constantly shown that small residual effects may contain the seeds of important discoveries.”