202 Lord Rayleigh on Copying Diffraction-gratmgs, 



At 19}, for example, if we use a green glass, we lose sight of 

 the proper period, and have before us an almost uniform field; 

 but if without making any other change we substitute a red 

 glass for the green one, we see the bands again with great 

 distinctness. At about the greatest distance included in the 

 table the positions of best definition are again in coincidence ; 

 but here there is an important remark to be made. If, using 

 the green glass, we adjust a needle-point to the centre of a 

 bright band, we find, on substituting the red glass, that the 

 needle-point is now in the centre, not of a bright, but of a dark 

 band. The fact is that at every revival of definition the image 

 changes sign, in the photographic sense, from positive to nega- 

 tive, or from negative to positive — a clear proof that the ap- 

 pearance in question is not a mere shadow in any ordinary 

 sense of the term. 



With respect to the numerical values of the distances given 

 in the table, theory indicates that the interval from worst to 

 worst or from best to best definition should be a third pro- 

 portional to the period of the grating d, and the wave-length 

 of the light X, i. e. should be equal to d 2 /\. In the case of 

 red light, the mean interval from worst to worst is 4*8 inches, 

 and from best to best 4*7. The corresponding numbers for 

 green light are 5*5 and 5'3. In the subsequent calculation, I 

 have used the first stated intervals as probably the more 

 correct. 



For the grating employed the actual value of d was '0104 

 inch ; but a small correction is required for the want of paral- 

 lelism of the light. The distance of the source was about 27 

 feet ; so that, as the mean distance behind the grating at which 

 the appearances were observed was 1-g- foot, the above value of 

 d must be increased in the ratio of 28J to 27. Thus for the 

 effective d in centimetres, we get 



57 

 2-54 x ~ A x -0104. 

 54 



Calculating from this and from the observed intervals a by 

 means of the formula \=d 2 /a, we get in centimetres 



Vd>=6-40 x 1(T 5 , X (gl , en) = 5-59 x 1(T 5 . 



Direct determination of the mean wave-lengths of the lights 

 transmitted by the red and green glasses respectively gave 



V-a) = 6 ' 64 X i ^ Veen) = 5'76 X l(f*. 



The true wave-lengths are certainly somewhat greater than 

 those calculated from Talbot's phenomenon ; but the difference 

 is perhaps hardly outside the limits of experimental error. If 



