8 6". R. Williams 



did not shift due to that cause. The temperature was taken at 

 the beginning of the series of photographs, and is indicated on 

 the plates. 



At times, a shift of the bands did occur, but this was due to 

 a yielding of the cork strip between the plates or other mechani- 

 cal effects. It was found that if the cell was allowed to stand 

 before using, no difficulty was encountered from this source, for 

 an eye-piece was attached to the side of the camera box, whose 

 focus was on the focal curve of the grating, and through this 

 the bands were watched, to see that they did not shift with respect 

 to the Fraunhofer lines. 



Mitchell 1 has shown that the astigmatism of a grating depends 

 upon the length of the illuminated portion of the slit, the length 

 of the ruled lines, and the angle which the incident light makes 

 with the axis of the grating. On account of this property, com- 

 parison work is not possible as in the prismatic spectrum. In 

 photographing the two sets of bands, comparison work was done, 

 but other conditions were introduced that do not occur in ordi- 

 nary comparison work. The light falling upon a grating was 

 slightly divergent, and the dividing strip in the cell cut out a 

 narrow section of light across the middle of the ruled portion of 

 the grating. The light above this dark strip came from one 

 film, and that below from the other. The length of ruling that 

 was producing the spectrum for one set of bands was therefore 

 less than half the actual length of ruling of the grating, and as 

 a result the interference bands, from one film, extended about 

 half the width of the spectrum. 



If D is the thickness, /* the index, A the wave-length, and JV 

 the order of interference, the distribution of the bands through- 

 out the spectrum is represented by the equations, 



A 

 2D 



(2) A r =— - where /u.=i, as for a vacuum. 



A 



1 Johns Hopkins Univ. Circulars, June, 1898. 



136 



