One can almost see how it comes about. A decrease in excentricity means an increase in mean distance of the earth from the sun. This means to the moon a less solar perturbation. Now one effect of the solar perturbation is to keep the moon's orbit extra large: if the size of its orbit diminishes, its velocity must increase, according to Kepler's third law. Laplace calculated the amount of acceleration so resulting, and found it ten seconds a century; very nearly what observation required; for, though I have quoted observation as demanding twelve seconds per century, the facts were not then so distinctly and definitely ascertained. This calculation for a long time seemed thoroughly satisfactory, but it is not the last word on the subject. Quite lately an error has been found in the working, which diminishes the theoretical gravitation-acceleration to six seconds a century instead of ten, thus making it insufficient to agree exactly with fact. The theory of gravitation leaves an outstanding error. (The point is now almost thoroughly understood, and we shall return to it in Lecture XVIII). But another question arises out of this discussion. I have spoken of the excentricity of the earth's orbit as decreasing. Was it always decreasing? and if so, how far back was it so excentric that at perihelion the earth passed quite near the sun? If it ever did thus pass near the sun, the inference is manifest--the earth must at one time have been thrown off, or been separated off, from the sun. If a projectile could be fired so fast that it described an orbit round the earth--and the speed of fire to attain this lies between five and seven miles a second (not less than the one, nor more than the other)--it would ever afterwards pass through its point of projection as one point of its elliptic orbit; and its periodic return through that point would be the sign of its origin. Similarly, if a satellite does _not_ come near its central orb, and can be shown never to have been near it, the natural inference is that it has _not_ been born from it, but has originated in some other way. The question which presented itself in connexion with the variable ellipticity of the earth's orbit was the following:--Had it always been decreasing, so that once it was excentric enough just to graze the sun at perihelion as a projected body would do?