Makch 26, 1920] 



SCIENCE 



307 



refraction of light depends upon the atmos- 

 pheric conditions (temperature, pressure, hu- 

 midity) and decreases with altitude of the 

 celestial body above the horizon. Adequate 

 correction of the observed deflections because 

 of this known source of bending in the earth's 

 atmosphere had to be made.] 



(d) Terrestrial Refraction Effects. — Distur- 

 bance refraction effects as rays of light from 

 the distant stars passed through the region 

 of the earth's atmosphere affected by the solar 

 eclipse, especially during totality. This cause 

 would give a deflection in the right direction 

 but apparently not of sufficient magnitude to 

 account for the observed effects.^* 



15. The law of decrease in the amount of 

 deflection of light for causes (a) and (6) is 

 a very simple one, namely, inversely as the 

 distance of the ray from the sun's center 

 when it passes through the solar gravitational 

 field. For cause (c) the law may or may not 

 be as simple as that just stated, dex)ending 

 among other things on the variation of the 

 density and distribution of the solar atmos- 

 phere with distance from the sun."^" For our 

 own atmosphere the law of atmospheric re- 

 fraction is a somewhat complicated one. 

 Sufficient has been said to show how intensely 

 interesting a full discussion of the observed 

 deflections of light will prove to be. Even 

 had no deflections been observed a valuable 

 contribution to science would have resulted. 



16. Table I. contains the deflections of light 

 rays observed by the British Astronomical Ex- 



1* This hypothesis was suggested by Dr. J. Sat- 

 terly at the close of the author's lecture at the 

 University of Toronto, December 2, 1919. It hod 

 also occurred to Dr. Alexander Anderson, of the 

 University College, Galway, and has been dis- 

 cussed by him and others (Eddington, Cromelin, 

 Cave, Dines and Schusiter) in various issues of 

 Nature, December 4, 1919-Jamaary 29, 1920. 



15 In the discussion of the author 's paper be- 

 fore the American Academy of Arts and Sciences, 

 January 14, 1920, Dr. E. B. Wilson, of the Massa- 

 chusetts Institute of Technology, suggested that 

 if the density of the solar atmosphere varied in- 

 versely as the square of the distance from the 

 sun's center, a refraction law would result similar 

 to the one for causes (a) and (6). 



pedition, equipped and sent to Sobral, Brazil, 

 under the direction of the Astronomer Royal 

 of England, Sir Frank W. Dyson. Let a be 

 the total deflection of a light ray coming from 

 a star, 8, and passing through the sun's gravi- 

 tational field and finally reaching the observer 

 on the earth. Suppose a.„ be the value of a if 

 the ray grazed the sun's limb, and P, the radius 

 vector or distance from the sun's center to the 

 ray of light passing the sun. (The values of 

 P for the various stars are given in units of the 

 sun's radius in the third column of the table.) 

 Then 



a = ao/p- (1) 



As already stated, according to the Newton- 

 Maxwell law, ct^ = 0".87, and according to the 

 Einstein law, a^ = 1".Y4. As the observed ef- 

 fects appear to agree better with the Einstein 

 law, the comparison is made in the table with 

 those computed from that law. The main tab- 

 ular quantities have already been given in 

 various publications. Detailed data were also 

 courteously furnished by the Astronomer 

 Royal for my lectures; these data gave the re- 

 sults separately for each of the seven stars and 

 for each of the seven plates obtained by the 

 observer. Dr. A. C. D. Crormnelin, using a 4- 

 inch lens of 19-foot focus and an S-inch 

 coelostat. From the detailed data members of 

 my staff computed the probable errors found in 

 the last three columns of the table. From the 

 coordinates furnished we also were able to 

 compute the angle A, which the radius vector, 

 P, to any star made with the declination axis, 

 counting it from the north end in the direc- 

 tion east or west; these values are contained 

 in the fifth column. The computed effects in 

 right ascension and declination were obtained 

 by multiplying the value of a from (1) by 

 sin A and cos A, respectively. From the fourth 

 column it will be seen that the photographic 

 magnitudes of the stars ranged from 4.5 to 6.0. 

 The British astronomers were thus exceed- 

 ingly fortunate in being able to make their ob- 

 servations during a solar eclipse when there 

 was an exceptionally rich field of bright stars, 

 the Hyades, close to the sun. 



17. It will be observed that from the figures 



