226 



LIGHT. 



prived of the sun's light and becomes invisible. We can also note the moment 

 of its emergence, because then escaping from the edge of the shadow it comes 

 into the sun's light and becomes visible. It was, then, in this manner that Roe- 

 mer proposed to ascertain the motion of the satellite. But in order to obtain 

 this estimate with the greatest possible precision, he proposed to continue his 

 observations for several months. 



Let us, then, suppose that we have observed the time which has elapsed 

 between two successive eclipses, and that this time is, for example, forty-three 

 hours. We ought to expect that the eclipse would recur after the lapse of every 

 successive period of forty-three hours. 



Imagine, then, a table to be computed in which we shall calculate and reg- 

 ister before hand the moment at which every successive eclipse of the satellite 

 for twelve months to come shall occur, and let us conceive that the earth is at A, 

 at the commencement of our observations, we shall then, as Roemer did, ob- 

 serve the moments at which the eclipses occur and compare them with the mo- 

 ments registered in the table. 



Let the earth, be supposed at A, at the commencement of these obser- 

 vations, where it is nearest to Jupiter. When the earth has moved to B, which 

 it will do in about six weeks, it will be found that the occurrence of the eclipse 

 is a little later than the time registered in the table. When the earth arrives 

 at C, which it will do at the end of three months, they will occur still later 

 tlmn the registered time. In fact at C,the eclipses will occur about eight min- 

 utes later than the registered time. At D they will be twelve minutes later, 

 and at E sixteen minutes later. 



By observations such as these Roemer was struck with the fact that his pre- 

 dictions of the eclipses proved in every case to be wrong. It would at first 

 occur to him that this discrepancy might arise from some errors of his obser- 

 vations, but if such were the case, it might be expected that the result would 

 betray that kind of irregularity which is always the character of such errors. 

 Thus it would be expected that the predicted time would sometimes be later, 

 and sometimes earlier than the observed time, and that it would be later and 

 earlier to an irregular extent. On the contrary, it was observed during the six 

 mouths which the earth took to move from A to E, that the observed time was 

 continually later than the predicted time, and moreover, that the interval by 

 which it was later continually and regularly increased. This was an effect, 

 then, too regular and consistent to be supposed to arise from the casual errors 

 of observation ; it must have its origin in some physical cause of a regdal 

 kind. 



The attention of Roemer being thus attracted to the question, he deteimined 

 to pursue the investigation by continuing to observe the eclipses for another 

 half year. Time accordingly rolled on, and the earth transporting the astrono- 

 mer with it, moved from E to F. On arriving at F and comparing the observ- 

 ed with the predicted eclipse, it was found that the observed time was now 

 only twelve minutes later than the predicted time. At the end of the ninth 

 month when the earth arrived at G, the observed time was found to be only 

 eight minutes later ; at H it was only four minutes later, and finally, when the 



