multiplication of the examples available for observation,
and some 30 have already been discovered.

Dr Alexander Roberts, of Lovedale in Cape Colony, truly remarks that
the study of Algol variables "brings us to the very threshold of the
question of stellar evolution." ("Proc. Roy. Soc. Edinburgh", XXIV. Part
II. (1902), page 73.) It is on this account that I propose to explain
in some detail the conclusion to which he and some other observers have
been led.

Although these variable stars are mere points of light, it has been
proved by means of the spectroscope that the law of gravitation holds
good in the remotest regions of stellar space, and further it seems now
to have become possible even to examine the shapes of stars by indirect
methods, and thus to begin the study of their evolution. The chain of
reasoning which I shall explain must of necessity be open to criticism,
yet the explanation of the facts by the theory is so perfect that it is
not easy to resist the conviction that we are travelling along the path
of truth.

The brightness of a star is specified by what is called its "magnitude."
The average brightness of all the stars which can just be seen with
the naked eye defines the sixth magnitude. A star which only gives
two-fifths as much light is said to be of the seventh magnitude; while
one which gives 2 1/2 times as much light is of the fifth magnitude,
and successive multiplications or divisions by 2 1/2 define the lower or
higher magnitudes. Negative magnitudes have clearly to be contemplated;
thus Sirius is of magnitude minus 1.4, and the sun is of magnitude minus
26.

The definition of magnitude is also extended to fractions; for example,
the lights given by two candles which are placed at 100 feet and 100
feet 6 inches from the observer differ in brightness by one-hundredth of
a magnitude.

A great deal of thought has been devoted to the measurement of the
brightness of stars, but I will only describe one of the methods used,
that of the great astronomer Argelander. In the neighbourhood of the
star under observation some half dozen standard stars are selected of
known invariable magnitudes, some being brighter and some fainter than
the star to be measured; so that these stars afford a visible scale of
brightness. Suppose we number them in order of increasing brightness
from 1 to 6; then the observer estimates that on a given night his star
falls between stars 2 and 3, on the next night, say b