Now this is not the sensitivity which you calculated theoretically. I should like to ask how you found this sensitivity. I do not ask this for myself, because I know how you have done it, but for the sake of the audience, because I think the method you applied is so very beautiful. Then I suggest also that you tell us if you could detect the orbital effect on the assumption of a drag of 19/20.
Dr. R. J. Kennedy: Answering first the second question, I think that the effect due to the orbital motion of the earth should be observable with my apparatus.
As to the first question, I thought that the method of determining 31 was rather crude. A weight of 5-6 kg on the slab on which the apparatus was mounted produced a shift of one fringe. I determined the minimum weight (about 10 g.) which produced an effect just observable. The ratio of the two weights gives, then, hl/\.
I might explain also that I got rid of the surplus scattered light, using a different method from that suggested by Professor Michelson for his new device. I used polarized light, impinging under the proper angle on the glass plate (Brewster’s angle) so that no light at all was reflected. [See description in Kennedy’s report. Michelson exclaims: “Very nice indeed.”] The method I used is not my own invention. It has been suggested somewhere in Comptes rendus (1911), if I remember rightly.
The shift of azimuth (50° to the west) in Miller’s experiment seems to indicate that some spurious effect is present, dependent only on the position of the apparatus relative to the meridian, which shifts the azimuth of the whole effect to the west. The result must then be considered as a superposition of spurious effect and ether drift. This explanation would probably require a magnitude for the effect due to ether drift smaller than anything that could have been observed with the devices used. It might also explain, as I think, the difference between the results obtained by Mr. Miller and myself.
Piccard’s experiment does not seem to be of great value. As far as I can make out, he worked just at a time of the day when hardly any effect was to be expected.
Lorentz: I do not think that Kennedy’s last remark is quite right. Piccard really ascended at the time when the constellation of Hercules rose above the horizon.
Kennedy: Piccard ascended twice. Once, when the sidereal time
CONFERENCE ON MICHELSON-MORLEY EXPERIMENT 397
was right, his observations were spoiled by temperature effects. His errors were thirty times greater than the effect he was looking for. The second time he got rid of his errors, but there was no effect to be expected at the sidereal time chosen for this observation.
Miller: I agree with Hedrick that the theory of the instrument used for the experiments should be thoroughly studied. The theory of Lorentz is exact ; but it is general, and does not take into account the special conditions of the apparatus used. What actually happens to the fringes is dependent on the adjustment of the mirrors. When I became interested in the experiment in 1900, there existed no really adequate theory of the instrument. A theoretical study of the apparatus was then undertaken by W. M. Hicks, which was published in the Philosophical Magazine for January, 1902. We [Miller and Morley] thought it necessary to take up the question again, as Hicks had suggested that there was an additional term in the expression for the effect which had not previously been considered. This term represents an effect of appreciable magnitude, which is periodic in each full turn of the interferometer, while the ether-drift effect is periodic in each half-turn. In the Philosophical Magazine for May, 1905, we gave a review of the theory, showing that Hicks’s calculations did not affect the conclusions previously drawn. The full-period effect is actually present in the experiments of 1887, as well as in all those that have followed. In Comptes rendus, 168, 837, 1919, Righi began a series of articles, setting forth the theory in detail. He thought that our conclusions were not justified by the theory. It seems to me that Righi’s theory is correct in the abstract; but it does not deal with the actual things happening in the interferometer, as Hicks’s theory does. The question needs still further investigation, as suggested by Professor Hedrick. Hicks’s theory takes into account the fact that in practice the image c (Fig. 23) of mirror a with regard to a is slightly oblique to mirror b. This is necessarily true when straight-line fringes of finite width are obtained. Righi’s calculations are based on the assumption that b and c are exactly parallel, which would produce fringes of infinite width; thus, his criticism does not apply to the actual case. When b and c are oblique to each other, an actual ether drift will produce the additional effect predicted by Hicks, which is periodic in a full turn of the apparatus. Hicks has calculated its magnitude, showing that it