Ether Wind1

Dayton C. Miller.

In establishing the wave theory of light it was necessary to assume the existence of an all-pervading medium in which waves could arise and propagate; this hypothetical medium was called the “ether.” The ether was endowed with those properties that were necessary for explaining the observed phenomena. Many physicists attempted to discover the existence of the ether by direct experiment. The most important of the proposed experiments of this kind was put forward by A. A. Michelson in 1881 and was based on the conception that the ether, as a whole, is at rest and that light waves propagate in free ether in all directions and always with one and the same velocity with respect to the ether. It was also assumed that the earth, in its motion around the sun, passes freely through the ether, the latter remaining perfectly motionless in space. The experiment proposed by Michelson was intended to reveal the relative motion of the earth and the ether, which is often called the “ether wind” (ether drift). This experiment is based on the proposition that the apparent velocity of light changes according to whether the observer is moving together with the earth along the line of propagation of the light or perpendicular to that line. The velocity of light is 300,000 kilometers per second, while the velocity of the earth is one ten-thousandth of the velocity of light and is equal to thirty kilometers per second. If the orbital velocity of the earth acts directly in the present case, then the two apparent velocities should differ by 30 kilometers per second, or by one ten-thousandth. However, there are no methods for measuring the velocity of light under such simple conditions. In all methods it is necessary for a ray of light to travel to a distant point and return back to the initial point, whereby the positive effect of the earth’s motion on the forward-going ray should be

neutralized on the return beam. However, for a moving observer it was shown that the neutralization would not be complete; the apparent velocity of the beam traveling in one direction and the other along the direction of the earth’s motion must differ from the apparent velocity of the beam traveling perpendicularly in one direction and the other by the ratio of the square of the earth’s velocity to the square of the velocity of light, i.e. $(0.0001)$ or $0.00000001$.

The negligible differences of velocities caused by the ether wind can be detected by a remarkable instrument, the so-called interferometer, invented by Michelson. In this apparatus a beam of light is split into two by means of a so-called “semi-silvered” mirror, in which the layer of silver is so thin that approximately half the beam passes through it, while the other half is reflected in the ordinary way. Thus these two beams travel in directions perpendicular to one another. At the end of the path of each of these beams there is a mirror, from which they are reflected and return to the starting point. If the paths of both beams are optically equal, i.e. if the same number of light waves fits along them, then, when reunited, they are in the same phases. If, however, one path is longer than the other by half a wavelength, then the waves combine in opposite phases, so that the crest of one of them coincides with the trough of the other. As a result, the so-called “interference fringes” are obtained, the observation of which makes it possible to detect insignificant changes in the speed of light along one path and the other.

In 1887 Michelson, in collaboration with Edward W. Morley, made a substantial improvement in the method and apparatus and used the interferometer with the aim of discovering whether the motion of the earth in the ether produces the effects predicted by the theory of influence. Unfortunately, we do not know in what absolute direction the earth is moving in space, and therefore it is impossible to set the interferometer exactly in that direction. For this reason the entire apparatus was mounted on a foundation that floated in mercury, so that the instrument could be rotated through all azimuths of the horizontal plane of observation. The rotation of the earth about its axis causes the plane of the interferometer to move over the surface of a cone whose axis is the earth’s axis, and thus gives the instrument different orientations. What is accessible to observation, however, is only that component of the actual “ether wind” which lies in the horizontal plane of the interferometer. Therefore the apparent azimuth and magnitude of the ether wind must vary depending on the time of observation. A wind perpendicular to the plane of the interferometer should produce no effect at all; it is quite possible that such conditions occur at certain times of the year.

I cannot now set forth the details of this experiment. The observations consisted in examining, with the aid of a telescope, the system of inter-

fringes formed by the two beams of light. When the interferometer is rotated about its axis, the ether wind should cause oscillations of the entire system of fringes, which move first in one direction, then in the other, in such a way that the period of this oscillation is half the time of one revolution of the interferometer about its vertical axis. For a relative motion of the earth and the ether equal to the earth’s orbital velocity, i.e. 30 kilometers per second, the displacement in the original experiment of Michelson and Morley should have amounted to 0.4 fringe.

In November 1887 Michelson and Morley published the conclusions drawn from observations made in July of that year. They formulated these conclusions as follows: “if only the motion of the earth in its orbit is taken into account..., then observation shows that the relative motion of the earth and the ether is probably less than one-sixth of the earth’s orbital velocity and certainly less than one-quarter of it” (i.e. less than seven and a half kilometers per second). This result, which was interpreted by many as a zero result, or, as is often said, a negative result, aroused in many serious doubts as to the validity of the hypothesis of the luminiferous ether. However, in Michelson and Morley’s report there is an important “addition,” which begins with the following remark: “but it does not appear impossible that even at a moderate height above sea level, for example, on the summit of an isolated mountain, the relative motion may be appreciable for an apparatus like that used in our experiments.”

At the international congress of physicists in Paris in 1900, Lord Kelvin delivered an address in which he considered theories of the ether. He observed that “the only cloud in the clear sky of theory is the zero result of the experiments of Michelson and Morley.” Morley and the author were present at the congress, and in a conversation with Lord Kelvin the latter expressed confidence that the experiment would be repeated with more sensitive instruments. The author, in collaboration with Morley, built an interferometer approximately 4 times more sensitive than the instrument used in the first experiment; the path length in the latter instrument was 220 feet, and approximately 150,000,000 waves were accommodated in it. In this instrument the relative velocity of the earth and the ether, equal to the earth’s orbital velocity, should have manifested itself in a displacement of the interference fringes by 1½ fringes. It is precisely such an instrument that has been used continuously since that time. Its optical parts were entirely new, and in general nothing from the original apparatus was used, with the exception of the mercury reservoir and the wooden float.

A similar apparatus, with a foundation made of wooden beams, was used in 1902 and 1903, but the changes in the wooden frame

because variations in humidity and temperature made precise measurements impossible. Engineer Prof. Neff constructed a new supporting frame distinguished by rigidity and symmetry. This frame, or foundation, was built of steel and installed in the basement room of the physics laboratory of the School of Applied Science in Cleveland; observations with it were made in 1904–5. The results of these observations were published in the Philosophical Magazine for May 1905. These results were formulated as follows: “Therefore we may assert that the experiment shows that the ether near the apparatus does not move together with it; the difference of velocities is less than 3.5 kilometers per second, if only the defects of the experimental arrangement do not annul the effect of entrainment. Some think that this experiment shows only that the ether in certain basement rooms does not move together with them. We therefore wish to place the apparatus on a summit, in order to see whether the effect may perhaps be discovered there.”

It was precisely at this time that Einstein became interested in the question; in 1905 he published a paper entitled “On the Electrodynamics of Moving Bodies.” This paper was the first in a long series of articles by Einstein and others that developed the modern theory of relativity. In the aforementioned first paper Einstein sets forth the principle of the constancy of the speed of light, asserting that for an observer on the moving earth the measured speed of light must be constant, independently of the direction and magnitude of the earth’s velocity of motion. The principal physical foundation of the theory of relativity is the assumption that the experiments with the ether wind gave a definite result. However, the interpretation of this experiment was unacceptable to the author, and further observations were undertaken to resolve the question.

In the autumn of 1905, Morley and Miller moved the interferometer from the basement room of the laboratory to the Euclid Heights near Cleveland, to an elevation of approximately about 300 feet above Lake Erie, to a place free of all obstacles and buildings. Five series of observations were carried out (1905–6), which gave a definite positive effect equal to approximately 1/10 of the expected wind. There was a suspicion that this might have been caused by the influence of temperature; however, there were no direct indications of this. It was proposed to check this suspicion after the summer vacation. But the interferometer had been set up on land belonging to our friend; during our absence in the summer the land was sold, and the new owner demanded that the interferometer be removed immediately.

Prof. Morley withdrew from active work in 1906, and the continuation of the experiments passed into the hands of the author of these lines. It seemed desirable that further observations be made at a greater altitude, but numerous obstacles hindered the continuation of the work. The publication of the results of observations of the solar

THE ETHER WIND

the eclipse of 1919, which was interpreted as a confirmation of the theory of relativity, again aroused interest in the experiments with the ether wind. A generous friend allocated sufficient material funds to cover the considerable expenses connected with repeating the experiments. The observatory on Mount Wilson, near Pasadena in California, at an altitude of 6000 feet, seemed a desirable place for further work. Thanks to the kindness of Mr. Merriam, president of the Carnegie Institution in Washington, and of the directors of the observatory, Hale and Adams, the experiments were resumed by the author in March and April 1921 at the Mount Wilson observatory. The apparatus was essentially the very same as that used by Morley and Miller in 1904, 1905, and 1906. Observations were also made at the end of 1921 and again in 1924 and 1925.

Altogether, about 5,000 separate measurements of the ether wind were made on Mount Wilson at various hours of the day and night. These observations were reduced to 204 different series, each series referring to one hour of time. The observations were made in four different seasons of the year: I. April 15, 1921: 117 series of observations. II. December 8, 1921: 42 series. III. September 5, 1924: 10 series. IV. April 1, 1925: 35 series.

I. The very first observations, made in March 1921, gave a positive effect corresponding to a real ether wind, as if it were caused by the relative motion of the earth and the ether with a velocity of about 10 km/sec. However, before publishing this result, it seemed necessary to study all possible causes that might produce an effect similar to the ether wind. These possible causes could be reduced to magnetic deformations of the steel frame of the interferometer and to the influences of radiant heat. In order completely to eliminate the effect of radiant heat, the metallic parts of the interferometer were entirely covered with a layer of cork about one inch thick. Fifty series of observations made under these conditions revealed a periodic displacement of the fringes coinciding with the earlier observations.

II. In the summer of 1921 the steel frame of the interferometer was dismantled. In its place, on a mercury float, a concrete foundation reinforced with brass rods was installed. New supports made of aluminum or brass were made for the optical parts. In this way the apparatus was completely unaffected by magnetic influences, and the possible heating effects were considerably reduced.

In December (4–11), 1921, about 900 separate observations were made, reduced to 42 series. The results with this nonmagnetic interferometer gave a positive effect corresponding to an ether wind of exactly the same magnitude and direction as those obtained in April 1921.

Numerous variations of the conditions of the experiment were tried. Observations were made while rotating the interferometer clockwise and counterclockwise; with rapid rotation (1 revolution in 40 sec.) and with slow rotation (1 revolution in 85 sec.); with a heavy weight placed on the bracket of the tube and then on the bracket of the lamp; with the float raised high above the level of the mercury, so that first one quadrant was loaded and then another quadrant. The assistant who recorded the observations walked around or stood in various parts of the room, far from the apparatus or close to it. None of these variations had any influence on the results of the observations.

Then the entire apparatus was transported back to Cleveland. During 1922 and 1923 a great many tests were carried out under a variety of controllable conditions, and with various modifications in the arrangement of the parts of the apparatus. The mirrors and prisms were arranged in such a way that the light source could be located outside the room in which the experiments were performed, with the light entering the rotating interferometer along the axis of rotation. This method was usually employed in all observations of epochs III and IV. Another arrangement of the mirrors, which in practice proved very complicated, made it possible to observe the fringes with a stationary telescope; the necessity of frequently adjusting the fringes in the field of view made this method impracticable. Experiments were also carried out with photographic registration of the position of the fringes, both from a fixed point of observation and by means of a motion-picture camera moving with the interferometer. However, even an arc lamp as a source did not provide illumination bright enough to obtain satisfactory photographs; moreover, the necessity of frequent adjustment of the fringes made this method unsuitable. After the photographic method was abandoned, an astronomical telescope with an aperture of 5 inches and a focal length of 75 inches was mounted in connection with the interferometer. At a magnification of 50 diameters the fringes were observed on a large scale and with bright illumination, so that direct readings by eye were very satisfactory; this arrangement has since been used continuously. Various light sources were also tried: an electric arc, an incandescent lamp, a mercury arc, an acetylene lamp, and, finally, sunlight. Replacing the laboratory sources with sunlight did not affect the result in the least. In the end, an acetylene lamp of the automobile-headlamp type was chosen as the permanent source; as a rule, it was located outside the room in which the interferometer was situated.

A long series of experiments was carried out in order to study the influence of temperature inequalities in the interferometer room and the influence of radiant heat falling on the interferometer. For this purpose several electric heaters were used, arranged in such a way that

the heating coil was placed at the focus of a concave mirror. Temperature inequalities caused a slow but constant displacement of the system of fringes to one side, but did not cause a periodic displacement. Even when two heaters were placed at a distance of three feet from the interferometer, which was set in rotation, and sent heat directly toward the uncovered steel frame, no measurable periodic effect was observed. When, however, the heaters were directed toward the paths of the light rays through closed glass, a periodic effect was observed, but only when the glass was covered with an opaque material, and moreover in a very unsymmetrical manner, when, for example, one arm of the interferometer was completely protected by cardboard, while the other was not protected. These experiments showed that, under the conditions under which the experiment is actually carried out, the periodic displacement of the fringes cannot be caused by the influence of temperature.

The laboratory investigation showed that the “full-period” effect mentioned in the preliminary report on the observations at Mount Wilson in April 1921 is a necessary geometrical result of the adjustment of the mirrors when fringes of finite width are obtained. At the low temperature that prevailed at Mount Wilson in April 1921, it was necessary to use narrow fringes, and in this case the “full-period” effect is relatively large; as the width of the fringes increases, this effect diminishes and disappears completely only for fringes of infinite width, as is assumed in the simple theory of the experiment.

III. After the completion of the experiments described, the interferometer was again transferred to Mount Wilson. In 1921 the apparatus had been placed in a deep canyon. I feared that air currents in the canyon might introduce disturbances; likewise, the unsymmetrical distribution of mountain rocks around it seemed undesirable. In August 1924 a new site was selected on a slightly rounded hill, away from the canyons. The enclosure for the interferometer was built so that its orientation—the direction of the ridge of the roof and the placement of the doors—was at \(90^\circ\) to the orientation of 1921. The interferometer was in every detail the same as that used in Cleveland in July 1924. In September—on the 4th, 5th, and 6th—1924, 275 measurements of fringe displacement were made, the measurements being arranged in 10 series. The results of the observations revealed a definite displacement, in contrast to the negligible results obtained in Cleveland. The ether wind corresponding to this displacement, in magnitude and direction, was in complete agreement with that first observed at Mount Wilson. Part of the observations was carried out under conditions such that the paths of the light rays were covered by glass boxes, lined on top with corrugated paper which—as the experiments in Cleveland showed—completely excludes the influence of radiant heat. However, these covers

did not change the result in the least, whence it follows that there are no such disturbances at all.

IV. Observations on Mount Wilson were resumed on March 27, 1925, and continued until April 9. During this interval 1600 measurements were made, arranged in 35 series. The interferometer was the same as in September 1924. During the first half of the measurements the lamp was stationary and placed outside the room for the interferometer; during the second half of the measurements the lamp was placed on the interferometer cover near the axis of rotation, whereby the arrangement was somewhat simplified. This change in the position of the source did not affect the result in the slightest. The assistant occupied various positions inside the building, moving from one quadrant to another, likewise without any influence on the result.

During epoch IV the conditions for observation were exceptionally good. For some time there was fog, which kept the temperature very uniform. Four accurate thermometers were hung on the outside walls of the hut; in many cases the variations of temperature did not exceed \(0.1^\circ\) and were usually less than \(0.4^\circ\). However, even a change of several degrees, which may cause a constant displacement of the interference fringes, cannot alter the periodic displacement either in magnitude or in direction.

The observations in April 1925 gave results wholly identical with those of April 1921, despite the fact that the interferometer had been revised, that a different system of illumination and different methods of observation were used, and, finally, that the interferometer was installed in another place and in a differently oriented house.

The observations, when plotted by means of harmonic analysis (carried out by a mechanical analyzer), directly gave the azimuth and magnitude of the ether wind. No corrections were introduced into the observed quantities; up to the present all reports of the ether wind made on Mount Wilson have included their full values. Not a single observation was omitted, even if it appeared inadequate, and no “weights” were assigned to the observations, since no assumptions were made concerning the expected result. It may further be added that, at the time the readings were made, neither the observer nor the person recording the readings could form the slightest idea as to whether any periodicity existed, still less as to the direction or magnitude of this periodicity.

If these observations are correct, they must give definite indications of the motion of the entire solar system in space, combined with the orbital motion of the earth and its daily rotation about its axis. There must exist a special relation, for the given latitude, between the observed azimuth of the ether wind and the sidereal time of observation. Observations at different sidereal times

must give different azimuths, and observations at one and the same sidereal time must give one and the same azimuth for the given epoch. A preliminary graphical study of the observations shows that these conditions are fulfilled.

It is hardly necessary to say that determining the absolute motion of the solar system from such interferometric observations is extremely complex. Prof. Nassau (J. J. Nassau) and Dr. Strömberg (G. Strömberg) have rendered very substantial assistance in the mathematical analysis and have developed solutions for various parts of the problem, as well as a complete solution of the general problem by the method of least squares. The final numerical computations require several months of continuous work and are currently under way.

The experiments described, carried out on Mount Wilson during the last four years—1921–1925—lead to the conclusion that there exists a definite displacement of the interference fringes, of the sort that would be caused by the relative motion of the Earth and the ether at this observatory with a velocity of approximately \(10 \frac{\mathrm{km}}{\mathrm{sec}}\), i.e., about one third of the Earth’s orbital velocity. When this result is compared with the earlier results obtained in Cleveland, the idea suggests itself of a partial entrainment of the ether, diminishing 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 an assumption, and lead to the conclusion that the Michelson–Morley experiment should not give a null result in the exact sense of the word and, in all probability, never did give such a result. The complete processing of the experimental results, which will be finished in the near future, should provide definite indications concerning the absolute motion of the solar system in space.

1. A report read at the Washington Academy of Sciences. Published in Science and in Proceedings of the National Academy of Sciences of U. S. A. 11, p. 1925. Both texts, which differ somewhat in the completeness of individual parts, have been taken into account in the present translation. ↩