e found by Belopolsky in 1896[1447] to oscillate
once in nearly three days, the secondary globe being apparently quite
obscure. Further study of the movements thus betrayed elicited the fact
that the major axis of the eclipse traversed revolves in a period of
2,100 days, as a consequence, most likely, of the flattened shape of the
stars.[1448] Still more unexpected was the simultaneous assignment, by
Campbell and Newall, of a duplex character to Capella.[1449] Here both
components shine, though with a different quality of light, one giving a
pure solar spectrum, the other claiming prismatic affinity with Procyon.
Their mutual circulation is performed in 104 days, and the radius of
their orbit cannot be less, and may be a great deal more, than
51,000,000 miles. Hence the possibility is not excluded that the
star--which has an authentic parallax of 0.08"--may be visually
resolved. Indeed, signs of "elongation" were thought to be perceptible
with the Greenwich 28-inch refractor,[1450] while only round images
could be seen at Lick.[1451] Another noteworthy case is that of Polaris,
found by Campbell to have certainly one, and probably two obscure
attendants.[1452] Through his systematic investigations of stellar
radial velocities with the Mills spectrograph, knowledge in this
department has, since 1897, progressed so rapidly that the spectroscopic
binaries of our acquaintance already number half a hundred, and ten
times as many more doubtless lie within easy range of detection.

Now it is evident that a spectroscopic binary, if the plane of its
motion made a very small angle with the line of sight, would be a
variable star. For, during a few hours of each revolution, some at least
of its light should be cut off by a transit of its dusky companion. Such
"eclipse-stars" are actually found in the heavens.

The best and longest-known member of the group is Algol in the Head of
Medusa, the "Demon-star" of the Arabs.[1453] This remarkable object,
normally above the third magnitude, loses and regains three-fifths of
its light once in 68.8 hours, the change being completed in about twelve
hours. Its definite and limited nature, and punctual recurrence,
suggested to Goodricke of York, by whom the periodicity of the star was
discovered in 1783,[1454] the interposition of a large dark satellite.
But the conditions involved by the explanation were first seriously
investigated by Pickering in 1880.[1455] He found that the phenomena
coul