48 THE INTERFEROMETRY OF 



The ellipses move as a whole with M, without changing form appreciably, 

 throughout the spectrum; but they move very slowly, quite differently in this 

 respect from the round ellipses in displacement interferometry, which are 

 extremely sensitive to displacement of M. In the present work it may take 

 5 or 10 cm. at M to pass the ellipses quite through the spectrum. They are 

 strong and fine in spite of the film gratings used. 



In the endeavor to explain these phenomena one may notice that the main 

 features have already been accounted for. As to details, since the gratings 

 are films which may act from both sides, explanations are hazardous. I do 

 not believe, however, that the films (cemented with balsam on glass plate) 

 had any other discrepant effect here than to make straight lines sinuous. The 

 character of the phenomena, as a whole, is trustworthy. 



In the case of the duplicated fringes (fig. 27, a, b, and the strands) seen 

 with a fine slit, the danger is perhaps greatest. But it appears to me that the 

 coarse lines in figure 27 are vestiges of the ellipses of figure 29, due to a wide 

 slit. These are superimposed on special fringes resulting from the diffraction 

 of the narrow slit. It is difficult to conjecture any other cause of duplication. 



The shift of ellipses through the spectrum follows as before from figure 25. 

 Their occurrence in case of a wide slit might be associated with the equation 



or with their independence of wave-length. They would therefore result 

 merely from the obliquity introduced by dispersion. But the presence of the 

 glass-plate compensator militates against this. In the peculiar behavior of 

 the compensators when added or rotated around a vertical axis, the dispersion 

 of the glass itself comes prominently into play , for the effect of introducing 

 a corresponding air-path is negligible in its effect on form. Thus the removal 

 of a 3 mm. compensating plate may leave the fringes almost too small to be 

 seen, whereas the displacement of M over 3 cm. produces but little change 

 of form. 



Finally, the ellipses are developed in arc or contour from left to right, for 

 instance, when the slit is widened ; and they vanish from right to left as the 

 slit is more and more nearly closed. The last lines for a closing slit make a 

 narrow grid of fringes quite straight and strong. 



22. The same. Prismatic adjustment. The 60 prism has certain ad- 

 vantages in experiments like the present, particularly when non-reversed 

 spectra are to be obtained. Figure 30 is a device of this kind, in which P is 

 the separating prism and P' the collecting prism, the beam of white light L 

 from a collimator entering the flat face normally on the front side and issuing 

 normally on the rear side at c and c' . M and N are opaque mirrors parallel 

 to each other, G a direct-vision prism-grating. The telescope is at T. The 

 reflection may be either internal, as in the strong lines of figure 30, or it may 

 be external on silvered faces of the prisms p and p', the appurtenances being 

 shown in dotted lines. In this case the separated rays a, a', b, b' are collected 



