the clock nearly 5 seconds wrong! Many people set their watches by a big clock striking without thinking of this: perhaps 5 seconds is not very important to them. But now suppose a man had two houses, his dwelling and his office, let us say, one three miles away from the city clock, and the other only two miles away; and suppose his watch and the city clock were both keeping perfect time. Nevertheless when he was at home he would always think his watch was 15 seconds too fast, and when at his office he would think it was only 10 seconds too fast. The first time he noticed the difference, he might think his watch had gone wrong; but if he went on noticing he would soon find out that the watch was all right and that the discrepancy must be due to something else; and he would perhaps find out the real reason, namely, that the extra mile between house and office made the difference of 5 seconds because sound took that time to travel one mile.
In this kind of way it was found that light takes time to travel. For the striking clock we substitute an eclipse of one of Jupiter's satellites as seen from our Earth. For the dwelling and the office we substitute the two positions of the Earth, on opposite sides of the Sun, one far from the eclipse, the other nearer. The great Danish astronomer Roemer calculated times for the eclipses (which we may regard as his watch); and when he compared them with the observed times (which we may regard as the clock) he got a regular difference according as he was on the hither side (office) or the further side (dwelling) of the Sun, and he reasoned that light must take time to travel. Compared with sound, it travels fearfully
