ADDRESSES 57 
the fact that the number of spiral nebulz greatly exceeds that 
of any other form. In announcing their great preponderance, 
Keeler remarked that they must be due to some cause of com- 
mon occurrence in the heavens. In view of ,the multitude 
of stars, the diversity of their courses and the variety of 
their velocities, it is difficult to assign an occurrence of a 
highly potential kind that is likely to be more frequent than 
the close approach of stars to one another. But sufficient fre- 
quency is the least significant part of the coincidence in this 
case, for, to meet the requirements, the occurrences must not 
only have had sufficient frequency, but have had such a nature 
as to give these peculiar results, (1) a spiral form implying at 
once radial and tangential action; (2) a pulsatory effect, shown 
in streams of knots and diffuse matter, and (3) a double out- 
streaming from the center, taking the form of two arms issu- 
ing from opposite sides and branching diffusely in their distal 
parts. The heavens do not present a more distinctive de- 
ployment than the spiral nebula, and yet it is one of singular 
frequency, the dominant species of the nebular type. An ec- 
centric collision is no doubt competent to produce a certain 
class of highly dispersed spiral nebulz, as already remarked, 
but it is doubtful whether it is competent to produce all grades 
and types of spiral nebulze, or even the majority of the forms, 
while the probability of its occurrence is relatively small. Even 
so far as it is competent and adequate, it appeals to essentially 
the same principles of action as those that actuate dispersion 
in cases of approach. It thus appears to be only the last term 
in a long series, the less catastrophic members of which con- 
stitute the majority of actual cases. Eccentric collisions seem 
too rare events to give origin to so dominant and so varied a 
group as the spiral nebule. They are therefore regarded as 
one source of spiral nebulae but not an adequate source. Dis- 
turbing approach may have a numerical competency from one 
million to ten million times as great as actual collision, and yet 
it is not clear that it has any surplusage of competency to 
keep up the supply when the evanescent nature of the smaller 
nebulz is taken into account. This is on the assumption 
that the present supply of nebulz is to be maintained by the 
present galaxy of stars in their normal inter-movements. If 
one great cloud of stars is now passing through another, as 
