1881.] Mode of Eliminating Errors of Adjustment, tyc. 471 



The idea is an elegant one, but I apprehend that there wo aid be 

 considerable difficulty in carrying it out. For a spectroscope giving 

 large dispersion is of considerable weight, and the reversal of so heavy 

 an apparatus would be liable to introduce possible errors arising from 

 flexure.* It would be difficult to make sure that such did not exist, at 

 any rate, unless the instrument were constructed with great nicety 

 and firmness, which would add considerably to the cost ; and even 

 then the care and time required for the reversal would help to oblite- 

 rate the observer's memory of what he had seen in the first position of 

 the instrument. 



A method has occurred to me of effecting the reversal without re- 

 versing the spectroscope, but merely giving a lateral push to a little 

 apparatus which need not weigh more than a few grains. 



If the base of an isosceles prism be polished as well as the sides, 

 and a ray of light parallel to the base and in a plane perpendicular to 

 the edge fall on one of the equal sides of the prism so as to emerge 

 from the other, after suffering an intermediate reflection (which will 

 necessarily be total) at the base, its course after refraction will be 

 parallel to its course before incidence ; and there will, moreover, be 

 no lateral displacement, provided the lateral distance of the base from 

 the incident ray be such that the point of reflection is at the middle of 

 the base. 



If the slit of the spectroscope be covered by such a prism, placed 

 close to the slit and facing the collimating lens, to the axis of which 

 its base is parallel, it will not disturb the general course of the light 

 incident on the spectroscope, nor even produce a lateral displacement 

 provided the lateral position be that mentioned above ; but in conse- 

 quence of the reflection there will be a reversal as regards right and 

 left, and any error in the placing of the lights to be compared will 

 thus be detected and eliminated, by comparing the spectra seen with 

 the light from the slit direct or reflected. If the prism be placed 

 quite close to the slit it may be made very minute in section, though 

 it should be long enough to cover the slit, and then the change of 

 focus which it produces will be insignificant. 



There will be no need, however, to make the prism so very minute, 

 nor to place it so close to the slit, provided it be associated with a 

 plate to take its place in the direct observation, and compensate for 

 the change of focus which is produced by its introduction. 



Let ABCD be a section of the prism, let M be the middle point of 

 the base AB, KLMISTO the course of a ray passing as above described, 

 which is supposed to be the axis of the pencil coming through the 



* After the present paper was sent in to the Society, I was informed by Mr. Stone 

 that the spectroscope he had in his mind was a direct- vision one, which could be 

 turned in its socket, the slit and cylindrical lens remaining fixed. To such an 

 instrument the objection as to flexure would not apply. 



VOL. XXXI. 2 L 



