CHRISTIAN HUYGHENS 57 



Let S be the sun, B C D E the yearly orbit of the earth, J Jupiter 

 and G H the orbit of his nearest satelHte, for this one because of its 

 short period is better suited to this investigation than any one of the 

 other three. Suppose G to be the point where the satelHte enters, and 

 H where it leaves, Jupiter's shadow. 



Suppose that when the earth is at B, the satellite is seen to emerge 

 [at G], at some time before the last quarter. Were the earth to re- 

 main stationary there, 421^ hours would elapse before the next 

 emergence would take place, for this much time is taken by the satellite 

 in making one revolution in its orbit and returning to opposition to the 

 sun. For example, if the earth remained at B during 30 revolutions, 

 then after 30 times 42I/2 hours, the satellite would again be seen to 

 emerge. If in the meantime the earth has moved to C, farther from 

 Jupiter, it is clear that if light requires time for its passage, the 

 emergence of the satellite will be seen later when the earth is at C than 

 when at B. For we must add to the 30 times 42I/2 hours, the time 

 occupied by light in passing over the difference between the distances 

 [of the earth from Jupiter] G B and G C, i. e., M C. So in the 

 other quarter, when the earth travels from D to E, approaching 

 Jupiter, the eclipses will occur earlier when the earth is at E than 

 when at D. 



Now by many observations of these eclipses throughout ten years, 

 it is shown that these inequalities are actually of some moment, 

 amounting to as much as ten minutes or more : whence it is argued 

 that in traversing the whole diameter of the earth's orbit, K L, double 

 the distance from the earth to the sun, light takes about 22 minutes. 



< 



LjIlibrary 



