1883.] a collimating eye-piece in spectroscopy. 337 



prism in the position of minimum deviation, the definition is not 

 dependent on the exact parallelism of the incident beam. This 

 position of telescope and collimator is obviously attainable, when 

 a reflecting grating is used, only when they are either identical or 

 placed one above the other in the same vertical plane. We have 

 not met with any precise account of the Princeton instrument. 

 A collimating eye-piece with a plate of glass, or quartz, as reflector, 

 while allowing the collimator and telescope to be absolutely iden- 

 tical, has the disadvantage of much loss of light, and we preferred 

 to make the axes of collimator and telescope slightly inclined to 

 one another in the same vertical plane, but at such a small angle 

 that one object-glass and tube serves for both. We use in fact 

 only the upper half of the slit for the admission of light, and 

 reflect the return beam by a right-angled prism placed close to 

 the slit just below the centre. The edges of this prism are of 

 course short and its upper bounding plane passes through the 

 axis of the tube of the telescope. The length of the sides is 

 about | inch in order to give a sufficiently large field of view. 

 When the source of light is large enough to fill the whole of the 

 upper half of the slit the prism intercepts part of the light which 

 would otherwise fall on the lower part of the object-glass, but no 

 practical inconvenience results from that. There is no loss of 

 definition because there is lateral symmetry in the incidence of 

 the light on the object-glass. A stop with a small hole is placed 

 immediately behind the reflecting face (the hypotenuse) of the 

 prism to intercept stray light from the edges of the slit which 

 might otherwise reach the eye. 



The reflexions from the surfaces of the object-glass we have 

 found less embarrassing than other observers seem to have found 

 them. One reason is that the tube is lined throughout with black 

 velvet. We have long since found the necessity of this when ob- 

 serving either extremity of the spectrum, as ordinary blacking 

 and stops will not prevent a good deal of stray light entering the 

 eye. Another reason is that we use a single plano-convex quartz 

 lens as object-glass, so that there are but two surfaces to reflect 

 any light. The plane surface produces a well-defined image of 

 the slit which can be very well seen when the e} 7 e-glass is re- 

 moved. It is however a distant object and cannot be seen through 

 the eye-glass at all. It is quite intercepted by a very small patch 

 on the centre of the object-glass, but this is not really needed, 

 for when a strong light is used the slit can be made narrow and 

 the reflected light reduced. The curved surface of the object- 

 glass disperses the light it reflects more than the plane surface, 

 so that more of it is absorbed by the sides of the tube. 



It is easy to throw a strong light into the slit in such a 

 direction as to give very inconvenient reflexions, but the proper 



