EVOLUTION OF THE UNIVERSE. 599 
accord with expectation ; and such statistical investigations as those in 
R. A. Fisher’s recent volume are a triumphant vindication of the potency 
of natural selection. 
It is worth while recalling these facts when we consider the picture of 
the evolution of the universe which has emerged from recent work. I 
personally doubt whether the time has yet come for an astrophysicist of 
genius to write a book which shall in its own sphere rival the Origin of 
Species. We can point to a few new facts and to considerably more new 
(and occasionally discordant) theories. Out of them there has emerged 
the present picture, immensely exciting, but by no means certain. 
What of it can we regard as certain? First of all, there is no reason 
to doubt the existence of island universes. Such form that vast, fairly 
regular distribution of spiral nebulae through the depths of space which 
is revealed by photographs taken in the great telescopes. We can say 
with fair certainty that our own galactic universe is either a single spiral 
system or an aggregate of several such, each analogous to millions of others 
with which space is strewn. 
Secondly, I would say that it is fairly certain that our space is finite, 
though unbounded. Infinite space is simply a scandal to human thought ; 
and, though we must not assume that the universe was made that man 
might understand it, the alternatives to finite space are incredible. We 
cannot accept the idea of island universes succeeding one another 
indefinitely as we pass in imagination through the depths of space. Such 
a distribution does not accord with a Euclidean-Newtonian gravitational 
scheme, for it would lead to infinite gravitational potentials. Neither 
can we with equanimity think of a vast finite group of universes forming 
a sort of island in empty space. Ultimately such a group ought, one would 
surmise, to aggregate into a single mass. But in Riemannian spherical 
space we can have a finite and uniform distribution of universes, inasmuch 
as such space is unbounded so that every point is related to the whole as 
is every other point. Finally, there is no fact of observation to set against 
the belief that space has a very small positive curvature. 
Thirdly, I think we must accept as highly probable Jeans’s hypothesis 
that in the stats matter is actually destroyed as protons and electrons 
unite to form radiation. To this conclusion we are driven by failure to 
find any satisfactory alternative explanation of the vast output of energy 
by the stars. If, however, their lives were not to be measured by so long 
a period of time as millions of millions of years, the necessity for assuming 
the actual destruction of matter would not arise. 
Consider now some of the uncertainties and difficulties which belong 
to the present scheme. First of all, of course, there is the insoluble 
difficulty of infinite time. No man of science will postulate a supernatural 
intervention, a stirring of the uniformly distributed matter filling space 
with which in imagination the present scheme begins. Yet, in default of 
such a beginning, we must imagine an infinite regress, a never-ending 
sequence of alternate periods of world-building and world-destruction, the 
rise and fall of universes without end. In comparison with such a past 
the future is perhaps less perplexing, though it is not very satisfying, 
because the second law of thermodynamics seems to necessitate an end 
when all energy will have so run down that nothing happens anywhere 
