interferometer in terms of the invariable length of
light-waves. A different form of interferometer has more recently
enabled him to measure the minute tides within the solid body of the
earth--not the great tides of the ocean, but the slight deformations
of the earth's body, which is as rigid as steel, that are caused by
the varying attractions of the sun and moon. Finally, to mention
only one more case, it was the Michelson-Morley experiment, made
years ago with still another form of interferometer, that yielded
the basic idea from which the theory of relativity was developed
by Lorentz and Einstein.

[Illustration: Fig. 22. Photograph of the hydrogen atmosphere of
the sun (Ellerman).

Made with the spectroheliograph, showing the immense vortices,
or whirling storms like tornadoes, that centre in sun-spots. The
comparative size of the earth is shown by the white circle traced
on the largest sun-spot.]

The history of the method of measuring star diameters is a very
curious one, showing how the most promising opportunities for scientific
progress may lie unused for decades. The fundamental principle
of the device was first suggested by the great French physicist
Fizeau in 1868. In 1874 the theory was developed by the French
astronomer Stephan, who observed interference fringes given by a
large number of stars, and rightly concluded that their angular
diameters must be much smaller than 0.158 of a second of arc, the
smallest measurable with his instrument. In 1890 Michelson, unaware
of the earlier work, published in the _Philosophical Magazine_ a
complete description of an interferometer capable of determining
with surprising accuracy the distance between the components of
double stars so close together that no telescope can separate them.
He also showed how the same principle could be applied to the
measurement of star diameters if a sufficiently large interferometer
could be built for this purpose, and developed the theory much
more completely than Stephan had done. A year later he measured
the diameters of Jupiter's satellites by this means at the Lick
Observatory. But nearly thirty years elapsed before the next step
was taken. Two causes have doubtless contributed to this delay. Both
theory and experiment have demonstrated the extreme sensitiveness
of the "interference fringes," on the observation of which the
method depends, and it was generally supposed by astronomers that
disturbances in the earth's atmosphe