MOUNT WILSON SOLAR OBSERVATORY. 253 



for different groups of lines, must be measured and eliminated, and 

 their effect on pressure displacements and other spectroscopic phe- 

 nomena must be determined. Only after the completion of much work 

 of this kind can we advantageously resume the task of interpreting the 

 displacements of the Fraunhofer lines at the center and limb of the sun. 



But the increased refinement now demanded is not limited to a 

 knowledge of line displacements in the arc and other sources ; indeed, 

 these very shifts can not be determined with sufficient accuracy by 

 existing methods. IMr. St. John and Mr. Babcock have materially 

 reduced the errors arising in the comparison of lines from two sources 

 by making both the exposures simultaneously, but the difficulty remains 

 of securing precisely similar illumination of the grating in both cases. 

 Doubtless this inequahty of illumination is one of the causes of the 

 surprising discrepancy between the results of different determinations 

 of the solar rotation. Mr. Adams's conclusion that the rotational 

 velocity is a function of the level in the solar atmosphere, though fully 

 supported by the measures of Mr. St. John and Miss Ware, and appar- 

 ently placed beyond doubt by recent work with the Koch micropho- 

 tometer (see p. 262), has not been confirmed by other observers in 

 England, Canada, and the United States. The detection of differences 

 in velocity for fines of different intensity probably depends mainly 

 upon the Unear dispersion of the spectrograph and the diameter of the 

 solar image employed, but errors in the absolute velocity are fikely 

 to arise from imperfect illumination of the grating. In any case, the 

 use of a purely differential method, involving only a single exposure 

 and the illumination of the grating from a single source, is highly 

 desirable in all of the investigations mentioned. 



In the study of the general magnetic field of the sun, displacements 

 of solar lines as small as 0.0001 a can be measured with the aid of a 

 compound quarter-wave plate, mounted above a Nicol prism over the 

 slit of the spectrograph. The photograph of the spectnun is divided 

 into narrow strips (1 or 2 mm. wide) corresponding to the successive 

 strips of the quarter-wave plate, the principal sections of which make 

 angles of +45° and —45° with the axis of the nicol. Thus the ellip- 

 tically polarized edges of the lines affected by the field are cut oft", in 

 the odd strips on the violet side, in the even strips on the red side. 

 The resulting displacements of the lines in adjoining strips are meas- 

 ured with a parallel plate micrometer, in which the lines are brought 

 into coincidence by inclining a narrow glass plate through which a 

 single strip is seen. 



The advantage of this device is that the spectra to be compared are 

 obtained in a single exposure, with the same illumination of the grating. 

 The superposition upon the solar or arc spectrum of an iodine absorp- 

 tion spectrum for comparison purposes would accomplish a similar 

 result (for some regions), were it not for the multipHcity of the iodine 



