414 Prof. Carl Barns on Interferometry 



as with ruled gratings, the ideal being an optical plate. 

 With thin films two sets o£ interferences are liable to be in 

 the field at once, and I have yet to study these features from 

 the practical point of view. If the film is mounted between 

 two identical plates of glass, rigorously linear, vertical and 

 movable interference fringes, as described * by my son and 

 myself, may be obtained. 



2. Cause of Ellipses. — The slit at L (fig. 1) furnishes a 

 divergent pencil of light due (at least) to its diffraction, the 

 rays becoming parallel in a horizontal section after passing 

 the strong lens of the collimator. But the vertical section 

 of the issuing pencil is essentially convergent. Hence, if 

 such a pencil passes the grating the oblique rays relatively 

 to the vertical plane pass through a greater thickness of 

 glass than the horizontal rays. The interference pattern if 

 it occurs is thus subject to a cause for contraction in the 

 former case that is absent in the latter. Hence also the 

 vertical axes of the ellipses are about the same in all orders 

 of spectra. They tend to conform in their vertical symmetry 

 to the regular type of circular ring-shaped figure, as studied 

 by Michelson and his associates, and more recently by 

 Feussner f. 



On the other hand, the obliquity in the horizontal direction 

 which is essential to successive interferences of rays is 

 furnished by the diffraction of the grating itself, as the 

 deviation here increases from violet to red. In other words, 

 the interference which is latent or condensed in the normal 

 white linear range of the slit, is drawn out horizontally and 

 displayed in the successive orders of spectra to right and 

 left of it. The vertical and horizontal symmetry of ellipses 

 thus follows totally different laws, the former of which have 

 been thoroughly studied. The present paper will therefore 

 be devoted to the phenomena in the horizontal direction 

 only. 



At the centre of ellipses the reduced path-difference is 

 zero; but it cannot increase quite at the same rate toward 

 red and violet. Neither does the refractive index of the 

 glass admit of this symmetry. Hence the so-called ellipses 

 are necessarily complicated ovals, but their resemblance to 

 confocal ellipses is nevertheless so close that the term is 

 admissible. This will appear in the data. 



If either mirror or the grating is displaced parallel to 

 itself by the micrometer-screw, the interference figure drifts 



* Phil. Mag. I. c. 



t See Prof. Feussner's excellent summary in Winkelmann's Handbuch 

 der Physik, vol. vi. p. 9o8 et seq. (1906). 



