310 



SCIENCE 



[N. S. Vol. LI. No. 1317 



the assistance received in the construction of 

 diagrams and in the computational work from 

 members of my staff, viz., W. J. Peters, H. B. 

 Hedriek, C. E. Duwall and C. C. Ennis. 



22. It is, of course, impossible without 

 further analysis to state at present just what 

 portion of the observed effects may be ac- 

 counted for 'by the various causes described in 

 paragraphs 14 and 21. Dr. ISTewall, for ex- 

 ample, see reference in footnote 13, is ready 

 to accept an effect from cause a (the ITewton- 

 Maxwell effect), but prefers considering the 

 possibility of accounting for the greater por- 

 tion of the remaining effect by cause c (Re- 

 fraction in the Solar Atmosphere). 



23. If it should prove to be the case that the 

 observed light deflections are the result of a 

 combination of the causes mentioned, the 

 way may be open to explain the results 

 oMained ly Dr. W. W. Oamphell's eclipse 

 expedition of June 8, 1918, at Goldendale, 

 Washington. Using two 4-inch photographic 

 objectives photographs were taken of the sun 

 and its surroundings, the exposures being 110 

 seconds, 50 stars to the ninth magnitude being 

 recorded. He states his results as follows:^' 



Tihe measurement of photograiphs, 14 inch X 17 

 indh in size, da a diffioul't problem even witli suit- 

 able apparatus: we found it necessary to con- 

 struct a special measuTing machine, and this was 

 mad© in our own shops. Duplicate photographs 

 of the eclipse field were secured at Mount Hamil- 

 ton seven months after the eclipse. As the dif- 

 ference of latitude between Moimt Hamilton and 

 the eclipse station is only a few degrees, no errors 

 were introduced by not obtaining the comparison 

 field at the eclipse station. These were taken at 

 the proper altitude to avoid the chief refraction 

 troubles in the comparison with the eclipse plates, 

 so that second differences of differential refrac- 

 tion alone entered into the comparison. The 

 plaltes were measured right and left. The same 

 scale- divisions were used for corresponding pairs 



ruary 6, 1920, and slides were shown exhibiting 

 the systematic character of these effects. The 

 maitter was gone into more fully at the New York 

 meeting of the American Physical Society, Feb- 

 ruary 28, 1920. 



10 The Observatory, London, Vol. XLII., No. 

 542, August, 1919 (298-300). 



of stars. As far as possible the measures were 

 freed from any known source of error. The cor- 

 rected differences of position were measured along 

 radii from the sun to each star and were arranged 

 in order of distance from sun to star. Dr. Ciurtia 

 was not able to say that there was anything syste- 

 matic about these differences, which Showed no 

 change of the order required by Einstein's second 

 hypothesis. The probable error of one star posi- 

 tion was the order of 0".5, regrettably large when 

 we are dealing with the differences of small quan- 

 tities — *he difference between the expected dis- 

 placements of the nearest and furthest stars only 

 being 0".26. A telescope of great focal length 

 would have been of great help in this work. For 

 the one we used the stars were too faint and in 

 the long exposure required we suffered from the 

 increased extent of coronal structure. Curtis di- 

 vided his stars into inner and outer groups. The 

 differential displacement between the two groups 

 should have been 0".08 or 0".15, according to whiieh 

 of Einstein's hypotheses was adopted. The mean 

 of the results came out at 0".05 and of the right 

 sign. After getting this result Curtis looked over 

 the collection of 40-foot coronal plates. In the 

 1900 eclipse there were six stars fairly bright, but 

 not well distributed. It is useless to take a dupli- 

 cate photograph now owing to uncertainty in the 

 values of the proper motions. Eeference has been 

 made to the Paris plate in the Carte du Ciel, but 

 Curtis was unable to say from the comparison that 

 the innermost star showed a displacement due to 

 the Einstein effect. 



"It is my own opinion," concludes Dr. 

 Camplbell, " that Dr. Curtis' s results preclude 

 the larger Einstein effect, hut not the smaller 

 amount expected according to the original Ein- 

 stein hypothesis." 



24. It will be observed that although Dr. 

 Campbell was not so fortunate as the British 

 astronomers in the matter of bright stars close 

 to the sun, he obtained an effect at more than 

 twice the distance from the sun of the farthest 

 star (56 Tauri), shown in Fig. 2, in the right 

 direction and of about the same amount as 

 that given by cause a (Newton-Maxwell Ef- 

 fect). It is of interest to note here that the 

 farthest star, 56 Tauri, in Fig. 2, also gave a 

 deflection approaching that given by cause a, 

 though since that star gave the largest prob- 

 able error, not much weight is to be attached 



