Memoirs presented to the Cambridge philosophical society on the occasion of the jubilee of Sir George Gabriel Stokes, bart., Hon. LL. D., Hon. SC. D., Lucasian professor.

PROF. MICHELSON, THE ECHELON SPECTROSCOPE. 323
X/nc. For instance, for 20 plates the average error may be a whole wave-length if the
inclination a is -1. As there is always a more or less perfect compensation of the
errors, the number of plates, or the inclination, may be correspondingly greater. Accordingly it may be possible to make use of 50 elements and the plane may be inclined
at an angle of 20~ to 30~. It would be necessary in this case however to use a rather
large objective. Possibly this may be avoided by cutting the surface A to a spherical
curvature, thus forming a sort of concave echelon.
The second process differs from the first only in that each plate is cut independently to the necessary height to give the required retardation. The first approximation
being made, the plates are placed on a plane surface as in Fig. 7 (side view) and Fig. 8
(front view).
FIG. 7.
The projections a and b are then ground and polished until the upper surfaces are
all parallel, and the successive retardations equal. The parallelism as well as the height
is verified by means of the interferometer.
a _ b
FIG. 8.
These processes are, it is freely conceded, rather delicate, but preliminary experiments
have shown that with patience they may be successful.
41-2