Under this heading we have to consider a single illustration--the prediction, and the discovery, in 1846, of the planet Neptune. This event roused great enthusiasm among scientists as well as in the popular mind, afforded proof of the reliability of the Newtonian hypothesis, and demonstrated the precision to which the calculation of celestial motions had attained. Scientific law appeared not merely as a formulation and explanation of observed phenomena but as a means for the discovery of new truths. "Would it not be admirable," wrote Valz to Arago in 1835, "to arrive thus at a knowledge of the existence of a body which cannot be perceived?" The prediction and discovery of Neptune, to which many minds contributed, and which has been described with a show of justice as a movement of the times, arose from the previous discovery of the planet Uranus by Sir William Herschel in 1781. After that event Bode suggested that it was possible other astronomers had observed Uranus before, without recognizing it as a planet. By a study of the star catalogues this conjecture was soon verified. It was found that Flamsteed had made, in 1690, the first observation of the heavenly body now called Uranus. Ultimately it was shown that there were at least seventeen similar observations prior to 1781. It might naturally be supposed that these so-called ancient observations would lead to a ready determination of the planet's orbit, mass, mean distance, longitude with reference to the sun, etc. The contrary, however, seemed to be the case. When Alexis Bouvard, the associate of Laplace, prepared in 1821 tables of Uranus, Jupiter, and Saturn on the principles of the _Mécanique Céleste_, he was unable to fix an orbit for Uranus which would harmonize with the data of ancient and modern observations, that is, those antecedent and subsequent to Herschel's discovery in 1781. If he computed an orbit from the two sets of data combined, the requirements of the earlier observations were fairly well met, but the later observations were not represented with sufficient precision. If on the other hand only the modern data were taken into account, tables could be constructed meeting all the observations subsequent to 1781, but failing to satisfy those prior to that date. A consistent result could be obtained only by sacrificing the modern or the ancient observations. "I have thought it preferable," says Bouvard, "to abide by the second [alternative], as being that which combines the greater number of probabilities in favor of the truth, and I leave it to the future to make known whether the difficulty of reconciling the two systems result from the inaccuracy of ancient observations, or whether it depend upon some extraneous and unknown influence, which has acted on the planet." It was not till three years after the death of Alexis Bouvard that the extraneous influence, of which he thus gave in 1821 some indication, became fully known.