The eclipses of Jupiter's satellites had been observed for many years, and the times of their occurrence had been recorded. At length it was perceived that a certain order reigned among the eclipses of these bodies, as among all other astronomical phenomena. When once the laws according to which the eclipses recurred had been perceived, the usual consequence followed. It became possible to foretell the time at which the eclipses would occur in future. Predictions were accordingly made, and it was found that they were approximately verified. Further improvements in the calculations were then perfected, and it was sought to predict the times with still greater accuracy. But when it came to naming the actual minute at which the eclipse should occur, expectations were not always realised. Sometimes the eclipse was five or ten minutes too soon. Sometimes it was five or ten minutes too late. Discrepancies of this kind always demand attention. It is, indeed, by the right use of them that discoveries are often made, and one of the most interesting examples is that now before us. The irregularity in the occurrence of the eclipses was at length perceived to observe certain rules. It was noticed that when the earth was near to Jupiter the eclipse generally occurred before the predicted time; while when the earth happened to be at the side of its orbit away from Jupiter, the eclipse occurred after the predicted time. Once this was proved, the great discovery was quickly made by Roemer, a Danish astronomer, in 1675. When the satellite enters the shadow, its light gradually decreases until it disappears. It is the last ray of light from the eclipsed satellite that gives the time of the eclipse; but that ray of light has to travel from the satellite to the earth, and enter our telescope, before we can note the occurrence. It used to be thought that light travelled instantaneously, so that the moment the eclipse occurred was assumed to be the moment when the eclipse was seen in the telescope. This was now perceived to be incorrect. It was found that light took time to travel. When the earth was comparatively near Jupiter the light had only a short journey, the intelligence of the eclipse arrived quickly, and the eclipse was seen sooner than the calculations indicated. When the earth occupied a position far from Jupiter, the light had a longer journey, and took more than the average time, so that the eclipse was later than the prediction. This simple explanation removed the difficulty attending the predictions of the eclipses of the satellites. But the discovery had a significance far more momentous. We learned from it that light had a measurable velocity, which, according to recent researches, amounts to 186,300 miles per second.