Maech 26, 1920] 



SCIENCE 



309 



one. Though the result from 56 Tauri is dis- 

 cordant, it still is about midway between the 

 two computed curves (Causes a and h). It 

 should be noted also that the probable error of 

 observation, as shown by the size of the circle 

 around each star, is largest for 56 Tauri, so 



Fig. 2. 



that no whoMy safe inference as to cause of 

 its departure from the Einstein value may be 

 made. 



In view of the recognized difficulties of the 

 observations and the conditions undter which 

 they had to be made, and recalling, further- 



more, that the preparations and securing of 

 the requisite instrumental equipments were 

 undertaken during the stress of the great war, 

 every one will surely agree that the Astronomer 

 Royal of England and the British observers 

 are heartily to be congratulated upon the 

 splendid results of their labors. 



ANALYSIS OF OBSERVED LIGHT DEFLECTIONS 



21. In conclusion an analysis was sketched 

 of the observed light deflections and some evi- 

 dences were pointed out showing that while the 

 simple law (1) was followed to the greater ex- 

 tent, the effects in addition to varying in- 

 versely as the distance from the sun's center 

 also apparently depended in some measure 

 upon the heliographic latitude, 0, of the star. 

 As a consequence the observed effects are not 

 Strictly radial, the departures from radiality 

 occurring in a strikingly systematic manner, 

 and not in the accidental manner that would 

 be the case if the non-radial effects were at- 

 tributable wholly to errors of observations. 

 When such trigonometric functions are added 

 to law (1) as would arise from forces similar 

 in effect to centrifugal ones, the additional 

 effects are largely accounted for. This pos- 

 sible additional cause, whatever it turns out 

 to be, is designated as e. In complete allow- 

 ance for differential atmospheric refraction 

 effects in the earth's atmosphere may also be 

 the cause of non-radial effects. Resolving 

 the observed actual deflections into two com- 

 ponents, radial (along radius vector) and the 

 other non-radial (perpendicular to radius 

 vector), preliminary computations were made 

 with the aid of the expanded law. 



« = -° +i{9, 



(2) 



A value resulted for a^ agreeing better 

 with the Einstein value of 1".74, than the 

 value 1".9S stated in paragraph 18. A future 

 paper will give further account of this interest- 

 ing matter.^^ I must not fail to record here 



IS The possibility of non-radial effects arising 

 from cause e was announced at the meeting of the 

 Amerioan Philosopihjieal Soeieity, Philadelphia, Feb- 



