5()2 Effect of Errors in llalincj on a J )ijf radio n- G rating . 



" ohosts/'' one on either side of the main line at an an^idar 

 dist;ince from it of about 4^ 30^' (that is, about three times 

 the distance between the D lines in the first order). If the 

 grating is examined with the eyepiece removed, it is seen to 

 be crossed by a series of strongly marked dark bands, spaced at 

 regular intervals, parallel to the lines ruled on the surface. At 

 first it was assumed that these bands were coincident with 

 periodic errors in ruling, being formed in the manner ex- 

 plained in the former paper. But when the grating was 

 tested by the method jnst described, it was found that the 

 bands in question entirely disappeared when the slit was 

 adjusted so as to allow no light but that from the main line to 

 pass. To render the bands visible it was necessary to allow 

 the light from at least one of the ghosts to pass the slit 

 together with that from the main line. So that the effect 

 must be due to interference between the lio-ht from the o;liosts 

 and that from the main line. 



The reproductions (figs. 4 & 5, PI. XXIV.) are enlargements 

 from photographs showing the interference-bands due, in the 

 first case to the main line and two ghosts, in the second case 

 to the main line and one ghost. The latter is the well-known 

 case of FresneFs experiment, and the reproduction (fig. 5) 

 shows the series of alternately light and dark bands spaced at 

 approximately equal intervals, similar to that which would be 

 obtained with the bi-prism. The former case presents more 

 novelty, as the interference from three light-sources has not, so 

 far as I am aware, been observed previously. The main line 

 is naturally much brighter than the ghosts; so that the inter- 

 ference-bands maybe regarded as arising from two equal sources 

 with a source of double the intensity midway between them. 



It is evident that the system of bands obtained in a plane 

 parallel to the plane of the sources will be symmetrical with 

 regard to a central line. It can be shown by elementary 

 methods that at some distance from this central line the 

 bands will be practically identical with those obtained from 

 a pair of sources. The centre line itself may be either 

 bright or dark, according to its distance from the plane of 

 the sources. A careful examination of fig. 4 will show a fine 

 dark line in the centre of the symmetrical system of bands. 



If it were desirable, it would be possible to produce cases of 

 interference of still greater complexity by ruling gratings with 

 definite periodic errors so as to give a number of ^ghosts." 



In conclusion, I must express my great indebtedness to 

 Lord Blythswood for his continued interest and encourage- 

 ment in this work. 

 Blvthswood Laboratory, 

 Renfrew, N.B. 



