fact, necessarily putting a limit to the accuracy of all prediction, until some other unchanging and convenient measure of time shall be found to replace the "day" and "second."
The question at once presents itself, How can the constancy of the day be tested? The lunar motions furnish grounds of suspicion, but nothing more; since it is at least as likely that the mathematical theory is minutely incorrect or incomplete as that the day is sensibly variable. Up to the present time the most effective tests suggested are from the transits of Mercury and from the eclipses of Jupiter's satellites. On the whole, the result of Professor Newcomb's elaborate and exhaustive investigation of all the observed transits, together with all the available eclipses and occupations of stars, tends rather to establish the sensible constancy of the day, and to make it pretty certain (to use his own language) that "inequalities in the lunar motions, "not accounted for by the theory of gravitation, really exist, and in "such a way that the mean motion of the moon between 1800 and 1875 "was really less (i. e., slower) than between 1720 and 1800." Until lately, the observations of Jupiter's satellites have not been made with sufficient accuracy to be of any use in settling so delicate a question; but at present the observation of their eclipses is being carried on at Cambridge, Massachusetts, and elsewhere, by methods that promise a great increase of accuracy over anything preceding. Of course, no speedy solution of the problem is possible through such observations, and their result will not be so free from mathematical complications as desirable—complications arising from the mutual action of the satellites and the ellipsoidal form of the planet. On account of its freedom from all sensible disturbances, the remote and lonely satellite of Neptune may possibly some time contribute useful data to the problem.
We have not time, and it lies outside my present scope, to discuss whether, and, if so, how, it may be possible to find units of time and length which shall be independent of the earth's conditions and dimensions, free from all local considerations, cosmical, and as applicable in the planetary system of the remotest star as in our own. At present we can postpone its consideration; but the time must unquestionably come when the accuracy of scientific observation will be so far increased that the irregularities of the earth's rotation, produced by the causes alluded to a few minutes ago, will protrude and become intolerable. Then a new unit of time wall have to be found for scientific purposes, founded, perhaps, as has been already suggested by many physicists, upon the vibrations or motion of light, or upon some other physical action which pervades the universe.
Another problem of terrestrial astronomy relates to the constancy of the position of the earth's axis in the globe. Just as displacements of matter upon the surface or in the interior of the earth would produce changes in the time of rotation, so also would they cause corre-
