244 CARNEGIE INSTITUTION OF WASHINGTON. 



The illustration above is taken from regular radial- velocity work with 

 three prisms, and maj'^ be considered representative. The remedy, 

 both effective and simple, is an arbitrary adjustment of the wave- 

 lengths (separately for each spectral class) so that in the mean for a 

 large number of plates every line or group of Hnes will yield the same 

 radial velocity as every other hue or group of hnes. This adjustment 

 would bring the composite curve in the figure into coincidence with the 

 horizontal straight hne representing the mean. Nor is this procedure 

 in reahty more arbitrarj-, except for class G, than the present methods 

 for adopting the wave-lengths to represent zero radial velocity in each 

 spectral class. 



Perhaps the most serious consequence of the neglect of this adjust- 

 ment is the practically invariable introduction of systematic difference 

 between the brighter and the fainter stars. It is the experience of 

 observers that on spectrograms of the fainter stars the portion toward 

 the violet is very much underexposed, so that only about the half of 

 the plate toward the longer wave-lengths is available for measurement. 

 Thus, whatever systematic difference may exist between the entire 

 spectrogram and the half toward the longer wave-lengths will also 

 appear, with nearly full effect, as a systematic difference between the 

 brighter and the fainter stars. The great advantage gained from the 

 adjustment is that the radial velocities from all spectrograms of any 

 one star or of any one spectral class will be on a uniform basis, which 

 basis can be accurately defined, and can thus be employed with con- 

 fidence in various statistical investigations. 



Dr. Albrecht has pubUshed an article on ''Anomalous dispersion in 

 the Sun" (Astrophysical Journal, volume xli. No. 5, June 1915), a 

 summary of which is given. According to the theory of anomalous 

 dispersion as developed by Julius, Fraunhofer lines which are separated 

 from each other by only very short distances (about 0.5 a or less) 

 should produce mutual displacements of the Hnes, the violet and red 

 components of a pair being displaced in opposite directions, the former 

 somewhat more than the latter. Rowland's ' ' Preluninary table of solar 

 spectrum wave-lengths" forms a suitable basis with which to make a 

 test for this effect. In the test made (which is necessarily prehminary) , 

 the Fe Hnes alone were employed, the chief reasons for thus restricting 

 the data being the following: (a) for these hnes well-determined 

 laboratory wave-lengths are available for the necessary comparisons; 

 (6) as spectrum hnes are affected selectively by pressure and as it is 

 desirable to eliminate the unequal pressure effects, the data were 

 further restricted to those Fe hnes which had been classified according 

 to pressure-shifts. 



The principal results derived are : Lines with close companions toward 

 the red in the solar spectrum are shifted toward the violet, and hne 

 with close companions toward the violet are shifted toward the red. 



