dO Transactions of the Boyal Society of South Africa. 
of error either unsuspected or undeterminable from the nature of the data 
employed. It is considerations such as these, which demand an incessant 
repetition of observations even in cases where existing knowledge might 
at first sight appear sufficiently secure to serve for all future time. 
I have selected my illustrations largely from the solar system chiefly 
on the grounds that, thanks to the Newtonian laws, it is here that, in 
spite of the immense mathematical difficulties which have had to be faced, 
astronomical prediction has attained its greatest triumphs. I need scarcely 
mention the oft-quoted incident of the discovery of the planet Neptune, 
but cannot refrain at this time from directing your attention to the work 
of Messrs. Cowell and Crommelin in connection with the recent reappear- 
ance of Halley's comet. In spite of an interval of seventy-five years since 
this body was last observed, the comet was picked up on photographic 
plates long before it could be seen even with powerful telescopes almost in 
the exact position assigned by them, and we may confidently anticipate 
that, when their data are supplemented by those that have been and will 
be secured at the present apparition, its position at its next appearance 
will be assigned with an accuracy scarcely inferior to that of the planets 
at the present day. 
I wish now, however, to divert your attention to the stars. In so far 
as these form the fiducial points to which the motions of the planets and 
other members of the solar system are referred, it is essential that the 
positions of a limited number at least should be determined with the 
highest possible accuracy. Any uncertainty in their positions will un- 
doubtedly be reflected in the positions of the planets, and will constitute 
one of those sources of error so liable to increase with time, and render 
efforts at prediction if not entirely nugatory at least partially ineffective. 
The universe of so-called " fixed stars" is not invariable in aspect, 
though its changes for the most part are of so minute a character that 
they can only be surely detected either by the most delicate measurements 
or by their cumulative effect over long intervals of time. It is chiefly 
through a study of these changes that our knowledge of the stellar universe 
has been acquired in the past, and it is largely to similar means that we 
look for an extension of this knowledge. 
Among changes which lend themselves to observation for this purpose 
we may enumerate — 
1. Changes of the intensity of the light of the stars. The origin of 
these changes except in a few instances remains obscure. In certain 
cases, however, notably in the case of variable stars of the Algol type, a 
satisfactory explanation of the observed phenomena has been found in the 
motions of a system, governed by laws similar to those operating in our 
solar system, of which the visible star forms a constituent member. 
2. Changes of position due to orbital motion in binary or multiple 
