102 A. A. MICHELSON 



was decided to supplement the work by observing on a N.-S. line 

 which was accordingly laid out in the same manner as the E.-W. 

 line. 



The final calibration of the microscopes gave the following 



results : 



West 17 .52 turns = I mm 



East 17.60 " =imm 

 North 17.00 " =imm 

 South 17.60 " = I mm 



The mean for the E.-W. Kne is 17. 56, and for the N.-S. line, 17.30. 

 Accordingly the factor by which the calculated difference in level 

 in mm should be multipUed to reduce to micrometer divisions is 



For the E.-W. line, 285 

 For the N.-S. line, 289 



The factor actually used in the computation was 293, and accord- 

 ingly these should be diminished by 2.8 per cent for the E.-W. 

 line, and by i .4 per cent for the N.-S. line. 



A continuous series of observations was conducted on both lines, 

 beginning September 27 and ending November 29. The observa- 

 tions were made by setting the cross-hair of the micrometer on the 

 pointer and taking a number of readings, the mean of which gave 

 the fiducial reading to be subtracted from the readings of the 

 reflected image. This difference was subtracted from a similar 

 difference taken at the other end of the pipe, and the final differ- 

 ences from hour to hour (plus a constant) gave the observed curves. 

 These readings were taken every hour from 6 a.m. until midnight, 

 and every two hours from midnight until 6 a.m. The order of the 

 observations was usually south end, north end, east end, west end. 

 The time between readings at south end and north end, or between 

 east end and west end, was about four minutes, and the mean 

 between the two was taken to represent the time of observation. 

 The mean time of the two observations gives the same result as 

 though the observations had been simultaneous. 



Occasionally the reflected image would be obscured by floating 

 particles, and in clearing these away the value of the constant 

 would be altered. The new constant was accordingly chosen so 

 that the succeeding observations continued with the smallest dis- 

 continuity in the curve. 



