light entering the rotating interferometer along the axis of rotation. A further arrangement of mirrors, rather complicated in practice, was tried, for observing the fringes from a stationary telescope; the necessity for frequent adjustment of the fringes in the field of view made this method impracticable. Experiments were made with devices for the photographic registration of the positions of the fringes, both from the fixed observing station and by means of a motion picture camera carried on the interferometer. Even with an arc light as the source, there was not sufficient illumination to produce a satisfactory photographic record without slowing the rotation of the apparatus more than is consistent with our method of procedure, and the necessity for frequent adjustment of the fringes also made this method unsuitable. After abandoning the photographic method, an astronomical telescope having an objective of 13 centimeters aperture and a focal length of 190 centimeters was mounted on the interferometer. The object-glass is attached to the steel base near the halfsilvered diagonal glass and the eyepiece is supported on the end of the arm, there being no tube for the telescope. With a magnification of fifty diameters, the fringes are observable on a large scale and with ample illumination, so that direct reading with the eye was very satisfactory; this arrangement has been used in all subsequent observations.
Trials were made with various sources of light; with electric arc and incandescent lamps, the mercury arc, acetylene lamp and also with sunlight. The interchange between sunlight and laboratory sources in no way altered the results. The final choice for the stationary source placed outside of the interferometer room (or house, on the mountain) was a large acetylene lamp of the kind commonly used for automobile headlights. This arrangement was used in Cleveland in 1924 and at Mount Wilson in September, 1924, and in April, 1925. The use of a stationary light source with the light brought to the interferometer in the axis of rotation required very careful adjustment of the several mirrors involved in order to avoid a periodic displacement of fringes due to non-axial alignment. Careful trials were made which showed that it was better to place the source on the interferometer outside of the
cover and near the axis; thus the relation of source to the instrument remained constant. When this method was adopted, the small acetylene lamp, such as was used in the earlier experiments, was employed. This method of illumination has been used exclusively since April 9, 1925.
An extended series of experiments was made to determine the influence of inequality of temperature in the interferometer room and of radiant heat falling on the interferometer. Several electric heaters were used, of the type having a heated coil near the focus of a concave reflector. Inequalities in the temperature of the room caused a slow but steady drifting of the fringe system to one side but caused no periodic displacement. Even when two of the heaters, placed at a distance of three feet from the interferometer as it rotated, were adjusted to throw the heat directly on the uncovered steel frame, there was no periodic effect that was measurable. When the heaters were directed to the air in the light-path which had a covering of glass, a periodic effect could be obtained only when the glass was partly covered with opaque material in a very nonsymmetrical manner, as when one arm of the interferometer was completely protected by a covering of corrugated paper-board while the other arms were unprotected. These experiments proved that under the conditions of actual observation, the periodic displacements could not possibly be produced by temperature effects.
The Mount Wilson Experiments, 1924
Upon the conclusion of the experiments just described, in July, 1924, the interferometer was taken again to Mount Wilson. In 1921 the apparatus had been located on the very edge of a deep canyon; it was feared that the air currents up and down the face of the canyon might produce a disturbance and also that the unsymmetrical distribution of the rock of the mountain itself might be undesirable. In August, 1924, a new site was chosen on a very slightly rounded knoll, removed from the canyons. The interferometer house, Fig. 15, was erected with its orientation, as regards the ridge of the roof and the location of the door, changed by 90° from
Fig. 15. Ether-drift house at Mount Wilson in 1924-1926.
that of 1921. The house was about twenty-two feet square, and there were canvas windows all around as before; but instead of the corrugated iron for the sides, beaver-board was used, as this material is less absorbent of heat from the sun. Large pieces of canvas were placed over the entire house and at the end, to protect the house from the direct rays of the sun, greatly facilitating the making of observations throughout the period of daylight. The interferometer, Fig. 16, had the improved mirror mountings, protection from heat, improved light-source, large viewing telescope and other refinements which had been developed in the laboratory tests at Cleveland in 1923 and 1924.
This series of observations, of September, 1924, at Mount Wilson, was undertaken in a wholly unprejudiced but very confident state of mind. The extended laboratory tests had involved every suggested source of instrumental and external disturbance and had proved that none of these was operative in the experiment. The method of observing was so developed that there was perfect confidence in the readings. It was
Fig. 16. The ether-drift interferometer as used at Mount Wilson in 1924-1926.
felt that if any of the suspected disturbing causes had been responsible for the previously observed effects, now these were removed, the result would be a true null effect. Such a conclusion would have been accepted with entire satisfaction; and indeed it was almost expected. On the other hand, if the observations continued to give the positive effect, it would certainly have to be considered as real.
Ten sets of readings, consisting of 136 turns of the interferometer, were made on September 4, 5 and 6, 1924. These observations all show a positive periodic displacement of the interference fringes, as of an ether-drift, of the same magnitude, about ten kilometers per second, as had been obtained in previous trials. Part of these observations were made with the glass case over the light-path covered with corrugated paper board, which had been found in the Cleveland experiments to exclude all effects of radiant heat; the results were not altered in any way by this covering. The effects were shown to be real and systematic, beyond any further question.
In spite of the long continued efforts, it had so far been impossible to account for the effects observed in the interferometer as being due to terrestrial causes or to experimental errors. Very extended calculations were made in the effort to reconcile the observed effects with the accepted theories of the ether and of the presumed motions of the earth in space. The observations had been repeated at various epochs to test one after another of the hypotheses which had been suggested. At the end of the year 1924 when a solution seemed impossible, a complete calculation was made, for all hours of the day and for twenty-four epochs during the year, of the then expected effects due to the orbital motion and the apparent motion towards Hercules. This indicated that the effect to be expected had its greatest magnitude in April and that the minimum in April should be two and a half times as great as the effect at the time of the observations which had been made in September and that the maximum effect in April should be four and a half times as great. Furthermore, the effect in September would be directed to the northward at all times of the day while in April the azimuth of the effect would move progressively all around the horizon, the maximum value being attained
