But they are deceived.

It is obvious that _if one satellite eclipse another, the shadows of both must occupy the same point on Jupiter's body_.

Thus it is the overtaking of one _shadow_ by another on the disc, and not the overtaking of a _satellite_ by a shadow, which determines the occurrence of that as yet unrecorded phenomenon, the eclipse of one satellite by another.[13] The satellites when far from Jupiter seem to lie in a straight line through his centre.

But as a matter of fact they do not in general lie in an exact straight line.

If their orbits could be seen as lines of light, they would appear, in general, as very long ellipses.