It is, therefore, not uncommon to find spectra of certain orders mtich fainter than 

 others. But iis far as this cause is concerned, it affects alike spectra on botli 

 sides of the incident beam and the explanation of the inequality referred to must 

 he looked for elsewhere. 



It seems natural, of course, to look for the cause in some lack of symmetry 

 (if the tliin reflecting surfaces, in reference to the normal line, produced in the 

 ojieration of cutting the lines in the metal. 



To estahlish the existance of this an examination of the grating was made 

 by means of polarized light. It is well known that if light, polarized in the 

 plane of incidence, be retiected from a metalic surftxce the reflected beam will 

 remain polarized in the same jilane. If the plane of first polarization is not 

 identical with that of incidence the reflected ray is elliptically jiolarized and if 

 examined in the ordinary way it will, in general, present the appearance of being 

 polarized in a plane inclined to that of first polarization. The grating was 

 arranged so as to reflect a beam of light polarized in the plane of incidence and 

 in such a way tliat it could be shifted about in its own ]ilane an<l also turned 

 about a normal axis. It was at once observed that when the light was reflected 

 from the ruled surface of the plate, the ruled lines being nearly parallel to the 

 plane of incidence, a decided change in the plane of polarization took place. 

 When the grating was turned through an ai>gle of 180° about a normal axis the 

 inclination of the plane was on the opposite side. When the lines were at 

 right angles to the plane of incidence no change of polarization could be observed. 



From this it seems j)robable that in the process of cutting the grooves in the 

 metal each narrow reflecting surface had been slightly ti[iped out of the general 

 plane, towards that ]iart of the surface already ruled and this would evidently 

 have the affect of increasing the brightness of the spectra on that side at the 

 expense of those on the other. Another grating was then examined in the same 

 manner, the relative brightness of the spectra on dift'erentsidesnot being known. 

 The change in the plane of polarization was not so marked in this as in the other, 

 but it was very readily distinguished. In this the grating space is twice as 

 wide as in the first and it is evident that the disturbance of the plane of reflection 

 is not likely to lie so great. It was easy to predict, however, on which side of 

 tlie normal the spectra would be faint and, in the hands of an observer who was 

 entirely ignorant of the examination that had been made, this ]irediction was at 

 once verified. The microscopic examination of these metallic gratings is difficult 

 owing to the difficulty of illumination when an objective of sufficiently high 

 power is iised. A good illumination was finally feciued, however, by throwing 

 the light in a plane nearly parallel tu the surface of the grating. When this 

 was done the appearance presented was exactly in accordance with what had been 

 anticipated. When the light was received from tlic side towards wliich the narrow 

 rcilecting surfaces were lipped, as indicated by the pnlarization experiments, 

 these surfi^ces were easily and distinctly visible as bright lines; when the grating 

 was turned in its own plane through 180° from this position it was only with the 



