dispute that without some such background the vector would scarcely be intelligible—but the aether is now only a background and not an active participant in the theory.
There is accordingly no reason to transfer to this vague background of aether the properties of a material ocean. Its properties must be determined by experiment, not by analogy. In particular there is no reason to suppose that it can partition out space in a definite way, as a material ocean would do. We have seen in the Prologue that natural geometry depends on laws of matter; therefore it need not apply to the aether. Permanent identity of particles is a property of matter, which Lord Kelvin sought to explain in his vortex-ring hypothesis. This abandoned hypothesis at least teaches us that permanence should not be regarded as axiomatic, but may be the result of elaborate constitution. There need not be anything corresponding to permanent identity in the constituent portions of the aether; we cannot lay our finger at one spot and say "this piece of aether was a few seconds ago over there." Without any continuity of identity of the aether motion through the aether becomes meaningless; and it seems likely that this is the true reason why no experiment ever reveals it.
This modern theory of the relativity of all uniform motion is essentially a return to the original Newtonian view, temporarily disturbed by the introduction of aether problems; for in Newton's dynamics uniform motion of the whole system has not—and no one would expect it to—have any effect. But there are considerable difficulties in the limitation to uniform motion. Newton himself seems to have appreciated the difficulty; but the experimental evidence appeared to him to be against any extension of the principle. Accordingly Newton's laws of mechanics are not of the general type in which it is unnecessary to particularise the observer; they hold only for observers with a special kind of motion which is described as "unaccelerated." The only definition of this epithet that can be given is that an "unaccelerated" observer is one for whom Newton's laws of motion hold. On this theory, the phenomena are not indifferent to an acceleration or non-uniform motion of the whole system. Yet an absolute non-uniform motion through space is just as impossible to imagine as an absolute uniform motion. The partial
