It is, of course, true that the immediately prior condition of a sun or a planet was that of matter in an extremely attenuated or dissociated state. This is essential to the nebular, as well as to the meteoric, hypothesis. But it is not with the immediately prior condition that we are at present concerned, but with the primitive, or pre-nebular, condition. Take, for example, the case of the solar nebula, out of which our sun and planets were formed. Was this nebulous mass formed from matter in a state of extreme tenuity, scattered through space and collected together by gravity? Or did it result from two solid globes shattered to pieces by collision, which were then converted into the nebulous condition by the heat generated from the collision? It is no doubt true that the analogies of nature would, at first sight, be apt to lead us to the conclusion that the former theory was the more likely of the two, as the larger is generally made by aggregation from the smaller. But a little consideration will show that, in the present case, the weight of this analogy is more apparent than real. The impact theory does not rest upon a purely hypothetical basis. The cause to which it appeals has a real existence. The point of uncertainty is whether the cause actually produces the effect which is attributed to it. We know from observation that there are stellar masses, some of them probably larger than our sun, moving through space with enormous velocities in all directions.[77] According to the ordinary laws of chance, collision at times would be an inevitable result, and when such an event did take place the destruction of the colliding bodies, and their consequent transformation into a nebulous mass, would, at least in many cases, be a _necessary_ result. In fact, we have, in the case of these vast stellar masses, what we know occurs among the invisible molecules of a gas. So far as mere analogy is concerned, the impact theory is just about as probable as the other. [Footnote 77: The dark stellar masses which escape observation may be as numerous as those that are visible.] From what has been stated it would follow that in most cases the stellar masses have been formed out of the destruction of pre-existing masses, like the geological formations out of the destruction of prior formations.