M.-P. VOL. I.] CRA WFORD CONSTANT OF REFRA CTION. 121 



the two gives the correction o".oisin Z. D.; so that for 

 these observations the flexure correction has been considered 

 zero. The mean of the values of one revolution of the 

 declination micrometer, determined at the same time, is 

 48". 05. The value adopted, as noted before, is 48". 10. 



For the computation of the preliminary refractions 

 (called r' in the reductions) the Pulkowa tables have been 

 used. The reductions for the barometer, for the attached, 

 and for the external thermometers were taken from Vol. I, 

 " Publications of the Lick Observatory." 



The graduation errors of the i divisions of the fixed 

 circle have been determined by Aetronomer Tucker. His 

 results are given in Vol. IV, " Publications of the Lick 

 Observatory." He says there, in part: "The probable 

 error of a reading upon four divisions of the fixed circle 

 due to graduation may be adopted as o".i5. * * " * 

 There is some evidence of periodic character in the errors, 

 and it may be assumed, in absence of further data, that 

 the probable error due to errors of graduation is not 

 diminished by reading upon two adjoining divisions under 

 each microscope. * * * The largest error measured is 

 o".7 for the mean of four divisions." 



The errors are not sufficiently systematic to warrant 

 interpolating for undetermined divisions, so that no correc- 

 tion for division error has been applied. 



Three nadirs were observed every night. The changes 

 during a night were usually very small. The following 

 table gives the means of the three determinations on the 

 several nights: 



