310 C. Bancs — Resolution of Interference Fringes. 



line of the effective slit is tb, the spectrum corresponding to 

 the top point of the slit will be v V, so that all the bands have 

 been displaced toward the violet. Since the internal points of 

 the slit correspond to the intermediate spectra between vr and 

 v'r', the spectrum will thus contain curved black bands with 

 the tops toward the violet and their bottom toward the red. 

 Precisely the opposite will be the case at t"' V" for a negative 

 value of a?, i. <?., on the other side of the center of interference 

 fringes. If the slit moves, the general motion of bands will 

 remain as already explained, only there must now be succes- 

 sive changes of form. 



Thus if the slit takes the position t'U (some distance apart 

 from top to bottom of spectrum) there can be but one black 

 band in the spectrum, running from the red at the bottom to 

 the violet at the top and being, therefore, nearly horizontal. 

 At t"b" a single band must run from the violet at the bottom 

 to the red at the top, i. e., with reversed slope, so that clearly 

 with the projected slit symmetrical to x = the bands if 

 appearing (as they do in multiple in a proper position of the 

 bi-prism j5, fig. 1, close to the spectroscope S'G), must be quite 

 horizontal in the middle and curved upwards, or the reverse, 

 at both ends. 



Following fig. 3 for a given obliquity and length of slit, all 

 conditions may be easily computed. It is clear, moreover, 

 that the bands can never be closed curves, but are limited to 

 arcs cut off by the band of spectrum from a series of concen- 

 tric closed curves. In this respect they differ from the 

 elliptic interferences, which they in many respects recall ; but 

 the elliptics are essentially closed curves moving as a whole in 

 the same direction. 



Brown University, Providence, R. I. 



