but if it is very long, the slightest tilt will make you lose the end.
Now, if these patches showing no stars are really vacant spaces, they must be like very long corridors or tubes pointed very straight at us: that they are very long, we know, because we have evidence that the stars extend to immense distances. The people in this audience are confined within a definite room—quite a nice, large room, it is true, but still a room
Fig 82.
bounded by solid walls. There are no walls confining the stars: behind the front rows there are others, and behind them others still, and again others, and we never come to walls, so far as we know at present; so that the corridors through the audience of stars must be enormously long. Therefore, if we really see blank spaces at the end of them, they must be pointed dead straight at us, because the very slightest tilt would cause the end to disappear when the corridor or tube is so very, very long.
