﻿616 The Earl of Berkeley and Mr. D. E. Thomas on a 



Other measurements made at different times and on 

 slightly different parts of the plate confirmed these results. 



It* we assume, as an approximation, that placing the plate 

 at an angle of! 45° increases the relative retardation in the 

 two beams by about 13 per cent., we get for the plate at 

 right angles to the optic axis a retardation (4'25-'55)/50 = , 074 

 of a band. Hence, for a single transmission, the difference 

 in paths T>(fi— 1) = '037\. 



Obviously this difference may lie in a variation in D or 

 ji — 1 or in both, and we are unable to distinguish; but we 

 would draw attention to the fact that the condition for a 

 perfect echelon spectroscope plate is precisely that D (jj, — 1) — 

 constant, while Michelson's method *, which is generally 

 used for testing the figuring of these plates, only gives in- 

 formation on the product D jjl ; over and above this, the 

 method here described is some 30 per cent, more sensitive. 



As regards sensitiveness, it is most probable that if the 

 cylindrical lens were fitted at its focus with a fine vertical 

 fibre just wide enough not to cover completely the bright 

 space between the black bands in the upper field of view, 

 then the lower bands could be fitted to the upper with extra- 

 ordinary accuracy — this arrangement is the analogue of that 

 used by Dr. C. V. Burton | in his inicro-azimometer, where 

 single settings were made with a probable error of 1/800 

 part of the width of the central bright band of a diffraction 

 pattern. 



The Practical Application of the Method. 



Echelon plates are usually obtained by cutting up a large 

 plate, which has been figured under the Michelson test, into 

 strips, and these are then cut into the requisite lengths. It 

 is obvious that, short of building a very large Rayleigh 

 interferometer, the initial plate cannot be placed in our 

 instrument. We would suggest, however, that if the above- 

 mentioned strips were made somewhat wider than necessary 

 they could be examined separately by our method and then 

 refigured. 



There are two ways by which such an examination may be 

 made ; both involve a comparison with a standard plate. 



(1) By fitting the interferometer with a system of mirrors 

 or prisms so as to separate the two beams widely enough 

 (as outlined in fig. 3), we can get an "optical contour" 

 of the strip by the direct comparison of its different parts 

 with the standard. 



* Astrophysical Journal, vol. viii. 1898, p. 37. 

 t Phil. Mag. vol. xxiii. 1912, p. 385. 



