240 CARNEGIE INSTITUTION OF WASHINGTON. 



van Maaiien in 625 spectra, representing observations on 20 days. 

 Altogether 46 days, distributed over the interval of 110 days for which 

 plates are available, have been finished. From observations on 37 

 days, covering an interval of about two and one-half months, the 

 following preliminary results have been derived by Mr. Scares: 



Per:od = 3ls'51=tOJ62 



Inclination = 5 ?2±0?5 



Magnetic pole on central meridian, 1914, June 25.07 =t0'?43. 



INVESTIGATIONS ON THE RELATIVE POSITIONS OF SOLAR AND TERRESTRIAL LINES. 



Mr, St. John has continued his extensive researches on the relative 

 positions of solar and terrestrial lines, in order to determine their 

 bearing upon pressure, motions, and the possible influence of anomalous 

 dispersion and the Einstein effect in the solar atmosphere. Independ- 

 ent sets of observations are being made with several different spectro- 

 graphs, and a further check will be afforded by the use of interference 

 methods. Mr. St. John is also continuing his work on the Evershed 

 effect in sun-spots and carrying on an investigation of the solar rotation 

 in order to detect any possible change in its period. 



ERRORS IN THE MEASUREMENT OF CLOSELY GROUPED LINES. 



The data relative to differences between solar and arc lines have 

 been greatly increased during the year by Mr. St. John, but the major 

 effort has been an investigation of the accuracy of measurement and of 

 the possible mutual influence of closely adjacent solar lines. Personal 

 equation and its elimination have been given much attention. The 

 most important result is the evidence of systematic errors for close 

 pairs of lines in the Rowland tables. The measurement of solar lines 

 near the limit of spectrographic resolution is a matter of extreme diffi- 

 culty and liable to system.atic error. Separations have been deter- 

 mined, as far as possible, upon each of five series of spectrograms with 

 dispersions varying from 0.23 to 1.8 A per millimeter. For incomplete 

 separation, those obtained from registering microphotometer curves 

 are smaller than those found by filar-micrometer settings. Near the 

 hmit of resolution, microphotometer curves, settings on the edges of 

 the doublet and of lines similar to its components, and micrometer 

 measurements have been employed. The first appears more reliable 

 and yields smaller values than the third. Separations equal to the 

 theoretical spectrographic resolution, though ser\dng to detect duplicity, 

 are not sufficient for micrometer measurements to the third decimal 

 place of angstrom units. 



Filar-micrometer measurements of the separation of close doublets 

 vary with the width of the slit, the precision of the focal settings, and 

 the density and definition of the spectrograms, those of the finest 

 definition yielding the lowest values for doublets near the limit of res- 

 olution. Whatever decreases the intensity of the common region 

 relatively to that of the continuous spectrum produces a tendency on 



