1883.] a collimating eye-piece in spectroscopy. 331) 



that the magnification is independent of the object-glass of the 

 telescope in this case. With the 6th order no more can be seen 

 though the lines are wider apart. It will be observed that for 

 a given inclination of the telescope to the normal the dispersion 

 with the collimating eye-piece is just double what it is when 

 there is a separate collimator placed normal to the grating, for 

 in the former case we have n\ = 2a sin <p and in the latter 

 m\ = a sin (j>, so that n = 2-hi. Against this we must set the dis- 

 advantage of the increased number of overlapping spectra, the 

 most troublesome inconvenience attending the use of gratings. 

 The uncorrected lenses which we use for the sake of observing 

 the ultra-violet rays, are an advantage in regard to the over- 

 lapping spectra, because the parts of two spectra which are in 

 the field of view at the same time are not in focus together. 

 Thus if the image of the sun, or other source of light is pro- 

 jected on to the slit with a lens of short focus, the spectrum when 

 in focus forms a narrow but bright band, whereas the overlapping 

 spectrum which is out of focus forms a much broader and con- 

 sequently less bright band. This weakening of the overlapping 

 spectrum is sufficient when the brighter parts of the solar spectrum 

 are under observations, but not sufficient when either extremity 

 of the spectrum' has to be observed against an overlapping bright 

 part. In such cases we have used a coloured glass or liquid to 

 cut off the brightest light. Where the spectrum is more discon- 

 tinuous than that of the sun or electric arc the overlapping spectra 

 are not so frequently in the way. 



2. Facility and accuracy in measurement of wave-lengths. 



The method of measuring wave-lengths commonly adopted, 

 namely, to fix the grating perpendicular to the collimator and read 

 the deviation of the ray in the corresponding orders of spectra on the 

 two sides, leaves little to be desired in point of accuracy, inasmuch 

 as the errors of adjustment affect the readings on the two sides in 

 opposite ways and so are compensated in the result. By the use 

 of the collimating eye-piece equal accuracy is obtained by readings 

 on one side only. This is owing in the first place to the facility 

 and accuracy with which the zero reading is obtained. The 

 collimator being fixed and the grating moved the zero reading is 

 taken by placing the grating so that the image of the slit formed 

 by direct reflexion from the grating may be on the pointer or cross 

 wires in the eye-piece. The readings of the ray to be measured 

 may then be taken in the successive orders of spectra one after 

 another. In the next place there are no instrumental errors of 

 adjustment to be taken into account. It does not matter if the 

 axis of the collimator or the plane of the grating do not pass 

 accurately through the centre of rotation. All that is required is 



