MOUNT WILSON OBSERVATORY. 227 



The displacement deduced from the relativity hypothesis is 0.011 a. 



Owing to the importance of the question in the theory of generalized 

 relativity, as well as in the interpretation of solar and stellar spectro- 

 graphic observations, a further and more comprehensive investigation 

 is in progress. The behavior of the lines in the cyanogen band is being 

 studied under decreasing pressure and the characteristics of their 

 reversals in furnace spectra are under investigation. The complete 

 achromatism of the image given by the Snow telescope furnishes 

 exceptional conditions for studying the relative behavior of these 

 lines at the center and Umb of the sun. 



An extensive investigation has been begun of lines whose behavior 

 in the solar spectrum is exceptional, depending upon level, intensity, 

 wave-length, or molecular weight. An accumulation of such data 

 offers a promising means of reaching a more definitive conclusion upon 

 the general question of displacement of solar lines as well as to the 

 presence or absence of the Einstein displacement. The bearing upon 

 these questions of the indirect study of the solar spectrum by means of 

 observations of sunlight reflected from Venus is referred to in the report 

 of that investigation. 



RELATIVE WAVE-LENGTHS OF SKYLIGHT AND SUNLIGHT REFLECTED 



FROM VENUS. 



Evershed finds that wave-lengths in the spectrum of Venus are 

 always shorter than in the spectrum of direct sunlight. His observa- 

 tions show that the difference is very small when Venus and the earth 

 are on the same side of the sun, but that the wave-lengths for Venus 

 appear to decrease as the angle Venus-sun-earth increases. This 

 observation is directly opposed to an explantion of the solar shift by 

 the Einstein theory, anomalous dispersion, pressure effects, or con- 

 vection currents in the reversing layer. 



Spectrograms were secured at Mount Wilson by Messrs. St. John 

 and Nicholson when Venus was east of the sun in 1919 and when west 

 of the sun in 1919-20. Thirty lines in the region of X4500 have been 

 compared with the same lines in the spectrum of the sky. The obser- 

 vations show a slight shift to the violet for the spectrum of Venus, 

 -0.0026 ± 0.0004a for the first series and -0.0006=^0.0008 a for the 

 second. The mean difference, Venus minus sky, is —0. 0017=^=0. 0004 a. 

 This difference is near the limit of measurement, as indicated by the 

 discrepancy between the two series for Venus and by the fact that 

 mean wave-length of 30 lines on 23 sky-plates of the second series is 

 0.0008 A shorter than on 18 plates of the first series. 



Tests of superimposed spectra showed that skylight with 5 per cent 

 the intensity of Venus was sufficient to produce a shift of the order 

 observed. When the angle Venus-sun-earth is large and Venus appar- 

 ently close to the sun the plates must be taken at a very low altitude. 

 On 18 nights, when Venus was near greatest elongation, both high 



