PHYSICS: A. A. MICHELSON 
211 
It was found, however, unless the two optical paths of the interferometers 
were equal, which would involve the presence of a second lens equal to the 
one to be corrected, that the circular interference bands are extremely small 
and difficult to observe. 
The following simple, fairly direct method, obviates all these difficulties and 
has given excellent results. 
A silt in the focus of the mirror or lens to be tested is illuminated by light from 
a Nernst glower, concentrated by means of a microscope objective and a total 
reflection prism. The light returns immediately above the prism forming an 
image of the slit which is viewed through a microscope with a T V inch objective. 
A series of screens (an adjustable double slit would be much better) with two 
rectangular apertures are placed in succession in front of the lens or mirror 
to be tested; one of the apertures being central and the other at varying dis- 
tances from, the center. 
The resulting diffraction figure will be a series of bands parallel with the slit, 
and the position of the central band (achromatic in white light) will remain 
constant if the adjustment is right and the mirror perfect. 
The error in light waves will be half the observed error in fractions of the 
fringe-width. 
The lens or mirror is rotated through the entire circumference, at intervals 
of 45 degrees or less, and the same operation repeated; and the results plotted 
on the corresponding chart, which gives accordingly the error in light-waves 
at every selected point of the surface. 
This process applied to a 5-inch achromatic lens showed errors so small that 
artificial errors were introduced by placing in the path of the pencil a plane 
parallel plate which had been made roughly cylindrical by retouching locally. 
The errors were then measured as described, and amounted to about seven- 
tenths of a light wave at the greatest. The corrector plate was again retouched 
by local polishing, and after a half dozen trials (time occupied being about 
six hours) the errors were reduced to the order of one- or two-hundredths of a 
light- wave; and the resulting image (which was badly astigmatic) was rendered 
practically perfect. 
It is clear that such a process may be applied to even the largest astronomical 
mirrors or lenses, both in the original figuring and in the final correction; 
further, this final correction may be made upon the auxiliary plate, thus incur- 
ring no danger to the objective. 
With evident modification the method applies to the correction of prisms 
and gratings. In the latter, however, since the light must be nearly homogene- 
ous, there may not be sufficient intensity to observe the interference fringes 
under the high magnification required. 
It may therefore be of advantage to apply the interferometer (replacing 
one of the mirrors by the grating) or even more simply, by observing the inter- 
ference of the light reflected from a plane surface with that diffracted from the 
grating. 
