182 Michel son and Mori ey — Feasibility of Establishing 



The plane-parallel glasses are pressed by springs against three 

 brass studs which are adjusted by riling away the studs to 

 ■within a few waves of the desired length, and till the angle 

 between the planes was less than half a second. The former 

 operation was effected by the instrument described at the meet- 

 ing of the National Academy in New Haven, which for want 

 of a better term may be called the " interferential comparer." 

 And the latter by the " refractometer " also previously described. 



The relation between these intermediate bars which, commenc- 

 ing with the shortest (about half a millimeter), shall be desig- 

 nated A B CD E F G H I, was then determined with a mean 

 error of a fifth of a wave, as described below, by the " com- 

 parer," and the last one compared directly with a standard 

 meter. Thus the absolute length of A was found to be 0*4891 

 mm. t 0*0001 mm. This was divided by the wave-length of 

 sodium light taken from Rowland's tables and 2 A was thus 

 to contain 1659*1 t *2 of the longer sodium waves ( ^=5896*08 ). 

 A was then placed in the refractometer and the correct fraction 

 found to be *01 instead of *10 ; so 2^L=1659*0i; D . With this 

 corrected value of the length of A, B was found to contain 

 3314 22 waves, and, when corrected by the refractometer, 

 331427 waves. 



Similarly C was found to contain 6617*84 waves, and when 

 corrected, 6617*90. D contained 13237*00, corrected, 13236*99 ; 

 and ^contained 26472*08, corrected, 26472*18. 



The same process was repeated for other kinds of light to 

 diminish the chances of error. Besides yellow sodium and red 

 lithium, the three bright radiations of mercury in a vacuum 

 tube were used ; namely, a moderately bright double line in 

 the yellow, a very brilliant line in the green, and a fainter one 

 in the violet. 



The approximate wave-lengths of these were found by direct 

 comparison with the solar spectrum and Rowland's maps as 

 follow : 



Na, 5896-08 



Na 2 5890-1 



Li .... 6707*8 



(Y, 5769-8 



J T, 5790*6 



Hgj 



[V 4358*5 



I G 5460-6 



In the case of sodium it is somewhat difficult to obtain good 

 results because the two series of interferences overlap, and espe- 

 cially so because they are of unequally varying intensities. 

 "When the flame is bright the intensities are nearly equal — but 

 then the light is not sufficiently pure. When the light is faint 

 Na 2 is nearly twice as bright as Na v Hence to find the fraction 



