Decembek 5, 1919] 



SCIENCE 



519 



small shift in the places of the stars that 

 might be produced by the sun's attraction. 

 War conditions, however, made it impossible 

 to construct a suitable equatorial mounting, 

 though it is hoped that this may be done 

 before the eclipse of 1922. 



The results, to some extent, but, fortunately, 

 not entirely, justified these apprehensions. 

 The eclipse plates taken with the 13-in. 

 (stopped down to 8 in.) are out of focus. 

 Since the focus was good on photographs taken 

 at night a few hours earlier, and also on the 

 check plates taken before sunrise in July, the 

 explanation appears to be a change of figure 

 of the coelostat mirror, due to the heat of the 

 sun. These plates were compared with the 

 July check plates by using a duplex microm- 

 eter. They show an undoubted gravitational 

 shift, the amotuit at the sun's limb being 0.93'' 

 or 0.99", according to two different methods 

 of treatment. The probable error, as esti- 

 mated by the individual discordances, is about 

 0.3", but there is reason to suspect syste- 

 matic error, owing to the very different char- 

 acter of the star-images on the eclipse and 

 check plates. This instrument supports the 

 Newtonian shift, the amount of which is 0.87" 

 at the limb. There is one mode of treatment 

 by which the result comes out in better accord 

 with those of the other instruments. Making 

 the assumption that the bad focus did not 

 alter the scale, and deducing this from the 

 July plates, the value of the shift becomes 

 1.52". 



The results with the 4-in. lens are much 

 more satisfactory. The star-images are well 

 defined, and their character is the same on 

 the eclipse and check plates. As the duplex 

 micrometer would not fit these plates, a key- 

 plate, on which the film was placed away 

 from the lens, was taken in July, and all the 

 plates in turn were placed in contact with 

 this plate and compared with it. The result- 

 ing shift at the limb is 1.98", with a probable 

 error of 0.12". The values from the separate 

 stars are in good accord, and they support the 

 fact of the shift varying inversely as the 

 distance from the sun's center; they are thus 

 unfavorable to its being due to refraction, 

 as was suggested by Professor Newall at 



the meeting. Moreover, Professor Lindemann 

 pointed out that the comets of 1880 and 1882 

 had traversed this region without giving the 

 slightest evidence of having encountered re- 

 sistance; as their speed was about 300 miles 

 per second, a vivid idea is given of the ex- 

 treme tenuity of any medium that they 

 encountered. 



The Principe expedition was less fortunate 

 in the matter of weather, but a few plates 

 showed five stars. Since no check plates of 

 the eclipse field could be taken there, another 

 field near Arccurus was photographed, and 

 both it and the eclipse plates were compared 

 with plates of the same fields taken at Oxford 

 with the same object-glass. It was, moreover, 

 necessary to assume that the scale of the 

 eclipse plates was the same as that of the 

 check plate. This is justified by the fact 

 that the diurnal variation of temperature in 

 Principe is only some 4° F., and that there 

 had been no bright sunshine on the mirror 

 before totality. The measures indicate a 

 shift at the limb of 1.60", with a probable 

 error of 0.3". 



It will be seen that the mean of this result 

 and that with the 4-in. at Sobral agrees very 

 closely with Einstein's predicted value 1.75". 

 It was generally acknowledged at the meeting 

 that this agreement, combined with the ex- 

 planation of the motion of the perihelion of 

 Mercury, went far to establish his theory as 

 an objective reality. Sir J. J. Thomson, who 

 presided, spoke of the verification as epoch- 

 making; he suggested that it would probably 

 have a bearing on electrical theory, but he 

 regretted the very complicated form in which 

 Einstein expressed his theory, and hoped that 

 it might be possible to put it into a form in 

 which it would be more generally compre- 

 hensible and useful. 



Dr. Silberstein laid great stress on the 

 failure to confirm Einstein's third prediction, 

 that of the displacement of lines in the aim's 

 spectrum towards the red, to the amount of 

 1/20 Angstrom unit; this had not been veri- 

 fied, in spite of the careful search made by 

 Dr. St. John and Mr. Evershed. As the prob- 

 able error of the measures was much less 

 than the quantity predicted, he looked on this 



