February 17, 192 1] 



NATURE 



787 



mirrors. The possibility had to be faced that the 

 mirrors might suffer some distortion from the 

 sun's heat. Apart from this, two of the mirrors 

 did to some extent spoil the sharpness of the 

 stellar images. 



Thref series of photographs were taken. Prof. 

 Eddington and Mr. Cottingham at Principe had 

 very cloudy weather, but obtained photographs 

 showing a few stars. The check field obtained on 

 two nights shortly after the eclipse gave images 

 similar to those on the plates taken during the 

 eclipse. The temperature during the eclipse was 

 77° F., and 76-5° F. when the check fields were 

 taken. There was no reason to suppose any 

 change of scale, and on this assumption the plates 

 when measured gave for the deflection at the sun's 

 limb i-6i"±o-3o". 



Dr. Crommelin and Mr. Davidson in Brazil 

 were favoured with ideal weather conditions. 

 They found, however, that the images on the 

 eclipse plate differed from those taken the pre- 

 vious night, and from those taken on the same 

 field of stars two months later. This is attributed 

 to the effect of the sun's heat in distorting the 

 coelostat mirror. If it is assumed that the .scale 

 has changed, then the Einstein deflection from 

 the series of plates is o-go" ; if it is assumed 

 that no real change of focus occurred, but merely 

 a blurring of the images, the result is 1-56" ; little 

 weight is, however, attached to this series of 

 photographs. 



With a smaller lens of 4-in. aperture and ig-ft. 

 focus the same observers were extremely suc- 

 cessful. The images taken during the eclipse 

 are in sharp focus and exactly similar to 

 those on the comparison field. . The result 

 of measurement gave i-g8''±o-i2" for the 

 deflection at the limb ; seven stars were photo- 

 graphed, and the individual displacements, both in 

 right ascension and in declination, were in good 

 accordance with the law i-75''(fl/r). It has been 

 pointed out by Prof. H. N. Russell that the photo- 

 graphs show a difference of scale of one part in 

 12,000 in the horizontal and vertical directions 

 of the ca-lostat mirror, and that if allowance be 

 made for this the results for individual stars will 

 be in still closer accordance. 



The result of the eclipse, particularly of the 

 second series of photographs at .Sobral, is a close 

 verification of Einstein's predicted displacements 

 i-75''(a/r) radial from the sun's centre. 



We proceed to consider the objections raised 



against this result. It has been suggested that 

 allowance was not made correctly for the ordinary 

 terrestrial refraction. The method adopted was 

 that usually employed in astronomical photo- 

 graphy, the second order terms being omitted, as 

 they in no case amount to more than 002". The 

 possibility of distortion of the film of the photo- 

 graphs, owing to the presence of the corona, has 

 also been suggested. Examination of the plate in 

 Phil. Trans., vol. ccxx., will show how different the 

 conditions are from those cases in which distortion 

 of the film has been observed. The possible dis- 

 placements on the eclipse photographs could not 

 amount to 005" for any of the stars, and would 

 be in the wrong direction. 



Prof. Anderson made the interesting suggestion 

 that a possible error might arise from the fall of 

 temperature in the air as the moon's shadow 

 moves over the place of observation, thus causing 

 an exceptional refraction. It has been shown by 

 Prof. Eddington and Sir Arthur Schuster 

 (Nature, vol. civ., pp. 372, 468) that this effect 

 would in all cases be very much smaller than the 

 quantities in question. Further, it happens that 

 at Sobral, owing to cloud in the early morning, 

 which checked the normal daily rise of tempiera- 

 ture, the fall during the eclipse was not more than 

 2° or 30 F. 



The possibility that the observed displacements 

 may be due to refraction by gaseous matter sur- 

 rounding the sun has received a good deal of con- 

 sideration. If such an atmosphere is controlled 

 entirely by the sun's gravitation, an impossibly 

 high density is required. Prof. Newall has in- 

 vestigated the consequences of assuming that 

 gravitation is partly balanced near the sun by 

 electrical forces and radiation pressure assumed 

 to vary as the inverse square of the distance from 

 the sun's centre. To obtain agreement with the 

 eclipse observations, he finds it necessary that 

 0-999 of the weight of the gas should be thus 

 balanced. The difficulties of such a supposition 

 have been pointed out by Prof. Lindemann (Oh- 

 servatory, April, 1920). Further, Dr. Crommelin ' 

 has directed attention to the fact that several 

 comets have approached nearer the sun than paths 

 of the rays of some of the stars photographed at 

 the eclipse, yet their motion has not been retarded, 

 or their substance entirely vaporised, although 

 they were for two hours at this small distance 

 from the sun and moving with a velocity of 

 300 kilometres a second. 



Relativity and the Motion of Mercury's Perihelion. 

 By Dr. A. C. D. Crommelin. 



THE effect of P^instcin's law in changing the 

 position of Mercury's perihelion is some- 

 times looked on as something mysterious and re- 

 condite; but in reality anything that disturbs the 

 law of inverse squares in the smallest degree is 

 bound to produce such a shift; and as in all the 

 NO. 2677^ VOL. 106] 



.systems known to us such disturbing factors 

 exist, the line of apses is invariably in motion. 



It is easy to show that when the central force 

 falb off more rapidly than the ratio of inverse 

 squares, the apse-line advances ; for we may con- 

 sider the inverse square law as holding, with the 



