EXPERIMENTS WITH THE DISPLACEMENT INTERFEROMETER. 97 



The difference (observed value 2.1 cm.) is due to the necessary reversal of 

 the grating, which requires considerable readjustment. Using the third posi- 

 tion, the observed A7V was more closely 2 .059 cm. But here again two separate 

 pairs of slit images must be placed in coincidence, which calls for a double 

 rotation of the mirror M. 



59. Continued. Film grating not cemented to glass. To test this case 

 for the film grating, a number of fine specimens kindly made for me by Mr. 

 Ives were at hand. The film of these was not attached continuously to the 

 glass by a layer of Canada balsam. Nevertheless the film was perfectly 

 smooth and reflected well. It had about 15,000 lines to the inch, which is 

 rather too great a number for work of this kind, as the ellipses have their long 

 major axes horizontal. Accuracy is enhanced if the major axes are vertical. 

 Furthermore, the spectrum toward the right of the refracted ray is too far 

 distant from the micrometer for the easy manipulation of the latter. Hence 

 the spectra toward the left must be used, if available. 



The refracted slit images are shown in fig. 60, when the ruled face is toward 

 the rear. There are but two lines visible from the opaque mirror M and three 

 from the opaque mirror N, the fainter, yellow and less even line, No. 3, coming 

 from the film. To obtain the ellipses which are here alone sought, lines 2 and 

 3 reflected from the same face are put in full coincidence. 

 The micrometer positions were approximately as follows: 



M at 2.9 cm. Film toward the collimator. Good flat 

 ellipses, but not very strong. Path difference 6. 



M at 2.0 cm. Compensated position. Path difference 

 5 = o. Fine hair-lines, revolving. 



M at i.i cm. Film toward the telescope. Path dif- 

 ference 5. Ellipses resembling the first case. 



The stationary interferences were present, but not 

 objectionable. The spectrum, showing a very bright 

 sodium line, is brilliant. FlG 6o 



A large number of other film gratings of the same kind 

 were tested, but with no further results. Sometimes there are four slit images, 

 at other times six, depending upon the adjustment and shape of the films. The 

 tests were somewhat severe, as the arms of the interferometer were nearly 150 

 cm. long; but it is clear that in order to get the best results the film should be 

 continuous with the glass, being cemented with Canada balsam on the unruled 

 side, while the ruled side is outermost. In this case the plate grating is almost 

 reproduced. 



60. Single plate, film grating. It has been stated that the best results 

 would be obtained with a film grating cemented to plate glass on the smooth 

 side, with the ruled side exposed. Having failed to make one adequately 

 plane myself, I was fortunate in securing a sample through the courtesy of 

 Mr. Ives. The number of lines to the inch, 14,438, was rather in excess of 



