April 30, 1926]
analysis for the purpose of determining the azimuth and magnitude of the drift. In this work all the original observations have been used, without any omissions and without the assignment of weights; furthermore, there are no corrections of any kind to be applied to the observed values. The results of the analyses are finally charted in such a way as to show the variation in the azimuth of the drift throughout the day of twenty-four hours for each epoch, and the variation in magnitude is similarly charted. The observations of 1925 thus provide six curves, three showing the variation in azimuth for the different epochs and three showing the variation in magnitude. The curves are shown in Figs. 1 and 2. The dots,
connected by the light lines, represent single observations, each being the average of the readings from twenty turns of the interferometer during an interval of about fifteen minutes. The heavy line represents an arbitrary average of the single observations for the one epoch. In Fig. 1 the base line represents the twenty-four hours of the local civil day; a position on this line corresponds to a direction of motion to the north, while a point above the line indicates an easterly azimuth and one below the line, a westerly azimuth. In Fig. 2 the base again represents the hours of the civil day, while the magnitude of the ether drift, that is, the velocity of relative motion, throughout the day, is charted in kilometers per sec
ond. It is at once evident that there is something real in the observations; each curve has a definite and a characteristic form; certainly, the results are not zero, neither are they due to accidental errors of observation. The azimuth of the observed effect, Fig.
1, varies in a periodic manner throughout the twenty-four hours of the day, the average being about 45° west of north, with the time of greatest westerly deviation varying with the time of year. Fig. 2 shows that the magnitude of the effect also varies periodically, with its maximum of about ten kilometers per second occurring at different times of day at different times of year.
It has been impossible to specify any effects of temperature, radiant heat, magnetism, gravitation or any other cause, which can produce the systematic variations indicated for the different epochs. The presumption was then made that the effects may be due to the motion of the earth and of the whole solar system through the ether, that is, to a real ether drift. Various graphic and numerical solutions were
[Vol. LXIH, No. 1635
made for determining the apex and velocity of such a motion. These trial solutions were checked by means of a mechanical parallelogram apparatus, and finally by a partial least-squares solution. It was found that a direction towards a point in the constellation Draco, having the right ascension of 262° (17% hours) and the declination of + 65°, when projected on the plane of the interferometer at all times of the day for the three epochs of observations, would have an azimuth which would vary as shown by the smooth, heavy-line curves of Fig. 3. The azimuth would, how-
Civil Tibie-Hours .. „
in magnitude of the observed effect, being the averages from Fig. 2.
The curves so far considered have been plotted with respect to local civil time for Mount Wilson. If the
ever, vary equally to the east and to the west of north; that is, the curve should be partly above and partly below the horizontal base line of the figure. As drawn, the curves have been arbitrarily displaced downward (westward) to match the broken line curves which show the actual results of observation, taken from Fig. 1.
If the motion has a direction towards the constellation Draco with a velocity of ten kilometers per second, which remains constant throughout the year, its projection on the plane of the interferometer would vary in magnitude throughout the day, for the three epochs of observations, as shown by the smooth curves in Fig. 4. The broken-line curves show the variation
direction and magnitude of the motion are constant throughout the year the curves of the variation are more appropriately plotted with respect to sidereal time; in Fig. 5 the curves are so plotted, the heavy line representing the averages of all observations for 1925. There is a remarkable agreement of the curves for the different times of year when plotted against sidereal time; the figure shows that the concordance of the curves for the direction of motion is better than for the magnitude. In Fig. 6 the final averages of Fig. 5 are shown by the broken-line curves, whili the computed effects are shown by the smooth curves. For the azimuth the curves are drawn to a scale of displacement twice that of the preceding figures, the better to bring out the remarkable agreement between the curves.
As far as the observed quantities entering into these two curves are concerned, they are quite independent of each other; and each gives values of the