228 SIR DAVID BREWSTER ON THE BANDS FORMED BY THE 



When the combined gratings have the same number of divisions, such as 1 000 

 and 1000, the bands are seen upon all the spectra, and sometimes very faintly on 

 the luminous disc, but when one of the gratings has twice the number of divisions 

 as the other, such as 2000 and 1000, the bands appear as already mentioned, 

 only on the 2d, 4th, 6th, &c, spectra. In such combinations, the 1st, 3d, 5th, 

 7th, &c, spectra of the 1000 grating have no corresponding spectra in the 2000 

 grating, with which they can interfere, whereas, when the divisions in both are the 

 same, all the spectra of the one are superposed upon all the spectra of the other 

 and, therefore, bands are produced upon each of them. 



In like manner, if the number of divisions in the one grating is to those on 

 the other as n to 1, n being a whole number, the bands will appear only on the 

 spectra n, 2n, 3n, 4w, &c. 



When a grating of 1000 is placed above one of 2000, I have observed faint 

 bands upon the spectra, 1, 3, 5, &c, of the 1000 grating, though none of the 

 spectra of the 2000 grating could interfere with them. These bands are more 

 numerous than those between which they lie, and are doubtless produced by the 

 interference of spectra reflected from the plane surfaces of the glass plates with 

 those seen by transmitted light. 



When the gratings of 1000 and 2000 are placed at a small angle, as in Plate 

 XXIII., fig. 1, the grooves being parallel to AM, and the light incident perpendicu- 

 larly, the bands on the left-hand spectra are parallel and rectilineal, and highly 

 purple and green, as in fig. 2. 



By turning the gratings round AM as an axis, in the direction from D to B, 

 the bands descend from m, as in fig. 3, till they become parallel vertical lines, in- 

 creasing in number and less coloured, as in fig. 4, the number of bands on the 

 second left-hand spectrum being double those on the first. 



When the rotation is in the opposite direction from B to D, fig. 1, the bands 

 rise from n, fig. 5, till they become parallel and vertical as before. 



The opposite effects take place when the gratings are placed as in fig. 6, AM 

 and CS being coincident, and when we observe the spectra on the right hand of 

 the luminous disc. The bands now descend and ascend from the same points m, 

 n, now on the outer side of the spectra. 



When the two edges, AM, CS of the gratings are not parallel, as they are in 

 fig. 1, but inclined at a small angle, AMSC, fig. 6, then if, when the fringes are 

 parallel at a perpendicular incidence, we turn the gratings round AM as an axis 

 from B to D, the fringes descend from m, becoming smaller and smaller, till they 

 are parallel and vertical, but when the gratings revolve from D to B, the fringes 

 become larger and larger, less numerous, and more coloured, till they are finally 

 parallel to AM, the fringes being twice as numerous on the second spectrum as 

 on the first. 



When the grooves are perpendicular to AM, as in fig. 7, the bands are faint 



