Once more, let us suppose that $A$ and $B$, with their clocks and mirrors, are in relative motion, with half the speed of light, and pass one another at noon by both clocks. At 12:02 by $A$'s clock, he sends a flash of light, which reaches $B$ at 12:04 by his clock, is reflected, and gets back to $A$'s clock at 12:06. They signal these results to each other, and sit down to work them out. $A$ thinks that he is at rest, and $B$ moving. He therefore concludes that the light had the same distance to go out as to return to him and took two seconds each way, reaching $B$ at 12:04 by $A$'s clock, and that the two clocks, which agreed then, as well as at noon, are running at the same rate. $B$, on the contrary, thinks that he is at rest and $A$ in motion. He then concludes that $A$ was much nearer when he sent out the flash than when he got it back, and that the light had three times as far to travel on the return journey. This means that it was 12:03 by $A$'s clock at the instant when the light reached $B$ and $B$'s clock read 12:04. Hence $A$'s clock is running slow, compared with $B$'s. Hence the answer to the question whether two intervals of time, measured by observers who are in motion relative to one another, are of the same or of different durations, depends upon their assumptions about the motion of the universe as a whole. Now we must remember that one assumption about the motion of the universe as a whole is exactly as good--or bad--as another. No possible experiment can distinguish between them. Hence on the Principle of Relativity, we have left no absolute measurement of time or space. Whether two distances in different directions are to be called equal or not--whether two events in different places are to be called simultaneous or not--depends on our arbitrary choice of such an assumption, or "frame of reference." All the various schemes of measurement corresponding to these assumptions will, when applied to any imaginable experiment, predict exactly the same phenomena. But, in certain important cases, these predictions differ from those of the old familiar theory, and, every time that such experiments have been tried, the result has agreed with the new theory, and not with the old. We are therefore driven to accept the theory of relativity, strange as it is, as being more nearly "true to nature" than our older ideas. Fortunately, the difference between the results of the two become important only when we assume that the whole visible universe is moving together much faster than any of its parts are moving relatively to one another. Unless we make such an unwarranted assumption, the differences are so small that it takes the most ingenious and precise experiments to reveal them. THE GENERALIZATION Not content with all this, Einstein proceeded, a few years ago, to develop a "general" theory of relativity, which includes the effects of gravitation.