230 SIR DAVID BREWSTER ON THE BANDS FOB MED BY THE 



circles at great angles of incidence, the radius of which increases as the angle of 

 incidence diminishes, so that they become straight lines at a perpendicular in- 

 cidence. The bands are smaller at their upper and lower ends, and those on the 

 second spectrum are, as before, double those on the first, as shown in fig. 9. 



In the spectra on the left hand of the bar of light, the concave side of the 

 circular bands is towards the bar ; and in the spectra on the right hand of the bar 

 of light the convex side of the circular bands is toward the bar. The bands on 

 the right-hand spectra are smaller and more numerous than those on the left- 

 hand spectra ; and yet, by increasing the angle of incidence, the bands on all the 

 spectra increase in size and diminish in number. 



If at any particular incidence we turn the grating in its own plane, the bands 

 cross the spectra at angles increasing with the degree of rotation, and becoming 

 smaller and more numerous. When the end of the grating nearest the eye (A, fig. 

 8) ascends, the fringes, great and small, diminish and become more distinct, and 

 the centres of the circles descend. When the grating is turned in the opposite 

 direction, the centres of the circles ascend. 



In the principal gratings which I possess, when upon thin glass * including 

 those of 1000 and 2000 in an inch, these circular bands are accompanied by 

 another system of circular bands, convex to the luminous bar when seen on the 

 left-hand spectra, and concave to it when seen on the right-hand spectra ; but, 

 what is remarkable, the}' are smaller and more numerous on the first spectrum 

 than on the second, as shown in fig. 10. They are best seen when the principal 

 circular bands cross the spectra obliquely. 



In the preceding experiments with one grating, the grooves of the reflected 

 image are necessarily parallel to those of the real grating, owing to the parallelism 

 of the surfaces of the plate of glass, and therefore they cannot exhibit the result 

 of superposing two systems of grooves inclined to each other. This condition, how- 

 ever, may be obtained by drawing the grooves on the faces of a prism with a small 

 angle, or by placing a fluid prism between an ordinary grating and a plate of thin 

 parallel glass, which would enable us to vary the inclination of the two sets of 

 grooves. A better arrangement, however, is to place the grating AB (fig. 11) 

 upon a polished metallic surface, MN. A ray from the luminous bar at R, incident 

 on AC at r, reaches the eye at E, after reflexion from the steel surface MN, so that 

 the reflected image of the grating, AB, is superposed as it were on the direct image. 



When the grating, AB, of 1000 grooves in an inch is laid upon a steel surface, 

 MN, and the grooves are in the plane of incidence, the paragenic spectra of a 

 luminous bar are covered with bands, not serrated, parallel to the spectra, exhibit- 

 ing all the phenomena already described as seen by reflexion from a single grating. 



* These hands are not seen on a beautiful Munich grating, kindly lent me by Professor Stokes, 

 having 3750 divisions in an inch. As the bands become smaller with the thickness of the glass, 

 their absence in this grating arises doubtless from its great thickness, which is - 158 of an inch, 

 the thickness of the gratings upon which they appear being about 0'04. 



