In the first place, if an observer measures the velocity of light, he must always get the same result, no matter how fast he and his apparatus are moving, or in what direction (so long as the motion is uniform and rectilinear). This sounds harmless; but let us go back to the Michelson-Morley experiment where the light came back in exactly the same time from the two mirrors. If the observer supposes himself to be at rest, he will say that the distances $O M$ and $O N$ were equal. But if he fancies that the whole universe is moving in the direction $O M$, he will conclude that $M$ is nearer to $O$ than $N$ is--for if they were equidistant, the round-trip would take longer in the first case, as we have proved. If once more he fancies that the universe is moving in the direction $O N$, he will conclude that N is nearer to $O$ than $M$ is. His answer to the question which of the two distances, $O M$ or $O N$, is the greater will therefore depend on his assumption about the motion of the universe as a whole. Similar complications arise in the measurement of time. Suppose that we have two observers, $A$ and $B$, provided with clocks which run with perfect uniformity, and mirrors to reflect light signals to one another. At noon exactly by his clock, $A$ sends a flash of light towards $B$. $B$ sees it come in at 12:01 by his clock. The flash reflected from $B$'s mirror reaches $A$ at 12:02 by $A$'s clock. They communicate these observations to one another. If $A$ and $B$ regard themselves as being at rest, they will agree that the light took as long to go out as it did to come back, and therefore that it reached $B$ at just 12:01 by $A$'s clock, and that the two clocks are synchronized. But they may, if they please, suppose that they (and the whole universe) are moving in the direction from $A$ towards $B$, with half the speed of light. They will then say that the light had a "stern chase" to reach $B$, and took three times as long to go out as to come back. This means that it got to $B$ at 1 1/2 minutes past noon by $A$'s clock, and that $B$'s clock is slow compared with $A$'s. If they should assume that they were moving with the same speed in the opposite direction, they would conclude that $B$'s clock is half a minute fast. Hence their answer to the question whether two events at different places happen at the same time, or at different times, will depend on their assumption about the motion of the universe as a whole.