rare. What seems to me much more likely is that a star may collide with a dark nebula such as we concluded must exist in order to account for the patches empty of stars. To take an illustration from a well-known game, it is extremely unlikely that two golf balls will hit one another in mid-air, though on links like those of St. Andrews they may be struck off in opposite directions. I believe such a collision has happened, but thousands and thousands of balls must have passed more or less near one another without hitting. On the other hand it is pretty easy for a golf ball to hit a furze bush, as all golfers know. Now I think these dark nebulae are like the furze bushes and bunkers on a golf course, which almost seem sometimes to attract the ball out of perversity. Up till a few years ago such an idea would have been mere speculation: but in 1901 we had a splendid "new star" which gave us direct evidence of such an occurrence. Nova Persei, the "new star of the new century," blazed up suddenly, and died down slowly like all "new" stars: but after it had become faint again, a photograph of it showed a nebula surrounding it: and by comparing photographs the nebula was seen to be changing its shape (Fig. 96). At first it was thought that there had been an explosion of some kind and the nebula represented the scattering fragments; but it was presently realized that the rate of scattering was far too great for this explanation; and ultimately the startling truth was realized that we were witnessing some of those light "echoes" to which I referred at the beginning of the lecture. Let us
