866 



SCIENCE. 



[N. S. Vol. XVII. No. 439. 



is 3/800 or 1/266 of the total length of the 

 spectrum, and hence, in passing over it, the 

 wave-length varies by approximately 1/266 of 

 itself, and yet the eye could detect no differ- 

 ence. This certainly seems to indicate that 

 the minute change of 1/100,000 of the wave- 

 length, above referred to, is not at all able to 

 alter or control the chromatic sensation ex- 

 perienced by the eye. The spectrum used in 

 this experiment was a comparatively pure 

 one, since in the yellow band, one centimeter 

 wide, admitted through the opening in the 

 screen, the bordering colors, orange and green, 

 could clearly be seen at the edges. 



But it might be remarked that Dr. Kirsch- 

 mann's case is a purely hypothetical one, and 

 that my arguments are quite as theoretical, 

 and so it will be interesting to see just how 

 far we have been able to actually go towards 

 obtaining a perfectly monochromatic light. 



As is well known, the spectrum given by a 

 grating, with proper adjustments, is normal. 

 The distance between two portions of the 

 spectral image is proportional to the difference 

 of the wave-lengths proper to these portions. 

 Also, the spectrum of a sodium flame (given, 

 for instance, by burning ordinary salt in a 

 Bunsen flame) consists of two bright lines 

 separated by a narrow space. The mean 

 wave-length of the light forming one of these 

 lines is approximately 1/1,000 greater than 

 that forming the other. 



When working with a plane reflexion 

 grating, containing 14,400 lines ruled on a 

 space two inches wide, it was observed that 

 the interval between the two sodium lines was 

 about ten times as wide as either line. Now, 

 as we pass across the interval from one line 

 to the other there is a variation of 1/1,000 in 

 the wave-length, and hence in passing from 

 one side to the other of a bright line the 

 change in wave-length is noi greater than 

 1/10,000. 



But the narrowing of the spectral lines is 

 directly proportional to the total number of 

 rulings on the grating, while the interval be- 

 tween two lines varies directly as the close- 

 ness of the rulings; and in some gratings 

 used by Professor Rowland there were as 

 many as 120,000 lines on a space of six inches. 



i. e., 20,000 to an inch. Here the rulings are 

 eight times as numerous as in the former 

 case, and so the bright lines of the spectrum 

 should be only one eighth as broad; also, the 

 rulings are three times as close, and so the 

 interval between the sodium lines should be 

 increased threefold. Thus a grating like this 

 should give a spectrum in which the interval 

 between the sodium lines is over 200 timea 

 the width of either bright line. If such were 

 the case, we could conclude that the wave- 

 length of the light from each sodium line did 

 not vary as much as 1 in 200,000. 



Rowland,* however, remarks that there is 

 a limit to the applicability of this line of 

 reasoning, and that the width of a spectral 

 line given by a grating depends not only on 

 the width of the slit and the number of rul- 

 ings on the grating, but also on the true phys- 

 ical width of the line. But it is quite certain 

 that at least one half the high resolving power 

 above referred to was reached by him in his 

 experiments (i. e., that he really obtained a 

 spectrum in which the width of either sodium 

 line was but 1/100 of the interval between 

 them), since he reports having actually pho- 

 tographed some lines in which the variation 

 in wave length was not more than 1/80,000. 



But the best test for the homogeneity of 

 any light is to determine what is the greatest 

 difference of path which two portions of it 

 may have and still give interference fringes. 

 With white light this is very small. In the 

 ordinary apparatus for observing Newton's 

 rings only eight or nine rings can be seen 

 with white light, and for the ninth ring the 

 difference of path is about 1/200 of a milli- 

 meter, a very minute quantity. If sodium 

 light be used, many more rings can be seen; 

 and indeed interference has been observed + 

 with it with a difference of path of 200,000 

 waves, or over ten centimeters. 



By using the green line (•'^5461), ob- 

 tained on decomposing by a prism the light 



* Philosophical Magazine, 5 ser., Vol. 16, p. 209. 

 1883. 



f By A. A. Michelson and E. W. Morley. See 

 address before the American Association for the 

 Advancement of Science, Cleveland meeting, 1888, 

 by A. A. Michelson. 



