Lord Rayleigh's Investigations in Optics, 269 



rise at is much less. Probably this arrangement is about as 

 efficient as any. 



I have endeavoured to test these conclusions experimentally 

 with the spectroscope, using the double soda-line. The hori- 

 zontal aperture of a single-prism instrument was narrowed by 

 gradually advancing cardboard screens until there was scarcely 

 any appearance of resolution. The interior rays were blocked 

 out with vertical wires or needles, adjusted until they occupied 

 the desired positions when seen through the telescope with 

 eyepiece removed. With the arrangements either of fig. 2 or 

 of fig. 4 a very decided improvement on the full aperture was 

 observed ; but there was no distinct difference between these 

 two arrangements themselves. Indeed, no such difference was 

 to be expected, since the brightness of the first lateral band 

 has no bad effect on the combined images, as appears from the 

 curve E'BEF (fig. 2). Under other circumstances the in- 

 fluence of the bright lateral band might be more unfavourable. 



In powerful spectroscopes the beam is often rendered un- 

 symmetrical in brightness by absorption. In such cases an im- 

 provement would probably be effected by stopping some of the 

 rays on the preponderating side, for which purpose a sloping 

 screen might be used giving a variable vertical aperture. It 

 should be noticed, however, that it is only when the vertical 

 aperture is constant that the image of a luminous line is im- 

 mediately deducible from that of a luminous point. 



§ 3. Optical Poiver of Spectroscopes. 



As the power of a telescope is measured by the closeness 

 of the double stars which it can resolve, so the power of a 

 spectroscope ought to be measured by the closeness of the 

 closest double lines in the spectrum which it is competent to 

 resolve. In this sense it is possible for one instrument to be 

 more powerful than a second in one part of the spectrum, 

 while in another part of the spectrum the second instrument 

 is more powerful than the first. The most striking cases of 

 this inversion occur when one instrument is a diffraction-spec- 

 troscope and the other a dispersion-spectroscope. If the in- 

 struments are of equal power in the yellow region of the 

 spectrum, the former will be the more powerful in the red, and 

 the latter will be the more powerful in the green. In the 

 present section I suppose the material and the workmanship 

 to be perfect, and omit from consideration the effects of un- 

 symmerrical absorption. Loss of light by reflection or by 

 uniform absorption has no effect on resolving-power. After- 

 wards I propose to examine the effect of some of the errors 

 most likely to occur in practice. 



So far as relates to the diffraction-speetroscojoe, the problem 



