that the chance of collisions is not to be neglected, and this idea
finds a certain degree of confirmation in the appearance of
``temporary stars'' which have more than once blazed out in, or close
by, globular star-clusters.

This leads up to the notable fact, first established by Professor
Bailey a few years ago, that such clusters are populous with variable
stars. Omega Centauri and the Hercules cluster are especially
remarkable in this respect. The variables found in them are all of
short period and the changes of light show a noteworthy tendency to
uniformity. The first thought is that these phenomena must be due to
collisions among the crowded stars, but, if so, the encounters cannot
be between the stars themselves, but probably between stars and meteor
swarms revolving around them. Such periodic collisions might go on for
ages without the meteors being exhausted by incorporation with the
stars. This explanation appears all the more probable because one
would naturally expect that flocks of meteors would abound in a close
aggregation of stars. It is also consistent with Perrine's discovery
-- that the globular star clusters are powdered with minute stars
strewn thickly among the brighter ones.

In speaking of Professor Comstock's extraordinary theory of the Milky
Way, the fact was mentioned that, broadly speaking, the nebulæ are
less numerous in the galactic belt than in the comparatively open
spaces on either side of it, but that they are, nevertheless, abundant
in the broader half of the Milky Way which he designates as the front
of the gigantic ``plough'' supposed to be forcing its way through the
enveloping chaos. In and around the Sagittarius region the
intermingling of nebulæ and galactic star clouds and clusters is
particularly remarkable. That there is a causal connection no