Calculation shows that the deviation of light by the moon or planets would be too small to detect. But for a ray which had passed near the sun, the deflection comes out 1.7'', which the modern astronomer regards as a large quantity, easy to measure. Observations to test this can be made only at a total eclipse, when we can photograph stars near the sun, on a nearly dark sky. A very fine chance came in May, 1919, and two English expeditions were sent to Brazil and the African coast. These photographs were measured with extreme care, and they show that the stars actually appear to be shifted, in almost exactly the way predicted by Einstein's theory. Another consequence of "general relativity" is that Newton's law of gravitation needs a minute correction. This is so small that there is but a single case in which it can be tested. On Newton's theory, the line joining the sun to the nearest point upon a planet's orbit (its perihelion) should remain fixed in direction, (barring certain effects of the attraction of the other planets, which can be allowed for). On Einstein's theory it should move slowly forward. It has been known for years that the perihelion of Mercury was actually moving forward, and all explanations had failed. But Einstein's theory not only predicts the direction of the motion, but exactly the observed amount. Einstein also predicts that the lines of any element in the solar spectrum should be slightly shifted towards the red, as compared with those produced in our laboratories. Different observers have investigated this, and so far they disagree. The trouble is that there are several other influences which may shift the lines, such as pressure in the sun's atmosphere, motion of currents on the sun's surface, etc., and it is very hard to disentangle this Gordian knot. At present, the results of these observations can neither be counted for or against the theory, while those in the other two cases are decisively favorable. The mathematical expression of this general relativity is intricate and difficult. Mathematicians--who are used to conceptions which are unfamiliar, if not incomprehensible, to most of us--find that these expressions may be described (to the trained student) in terms of space of four dimensions and of the non-Euclidean geometry. We therefore hear such phrases as "time as a sort of fourth dimension," "curvature of space" and others. But these are simply attempts--not altogether successful--to put mathematical relationships into ordinary language, instead of algebraic equations. More important to the general reader are the physical bearings of the new theory, and these are far easier to understand. Various assumptions which we may make about the motion of the universe as a whole, though they do not influence the observed facts of nature, will lead us to different ways of interpreting our observations as measurements of space and time.