how this computation is made, it will be evident to you that we have thus the bases of accurate computation.
In like manner we can from observations of Jupiter's satellites, compute how far Jupiter draws one of his satellites in one hour, and therefore how far Jupiter would draw a planet at the same distance in one hour; and then by the law of gravitation we can compute (by the proportion of inverse squares of the distances) how far Jupiter will draw a planet at any distance in one hour; and this is to be combined, in computation, with the space through which the sun will draw the planet in one hour. In like manner, by similar observation of Saturn's satellites, and similar reasoning, we can find how far Saturn will draw any planet in one hour, and we can combine this with the space through which the sun would draw it in one hour. Thus we are enabled to compute completely the perturbations which these three planets produce in any other planets; and then comes the critical question. Do the planets' motions, as computed with these disturbances, agree with what we see in actual observation? They do agree most perfectly. Perhaps the best proof which I can give of the care with which astronomers have looked to this matter, is the following: the measures of distances of Jupiter's satellites in use till within the last 16 years had not been made with due accuracy, and. in consequence the perturbations produced by Jupiter had all been computed too small by about 150 part. So great a discordance manifested itself between the computed and the observed motions of some of the planets (especially the small asteroids whose orbits are between those of Mars and Jupiter, more particularly Juno), and also in the motions of one of
