June 11, 1920] 



SCIENCE 



581 



PRELIMINARY RESULTS OF ANALYSIS 

 OF LIGHT DEFLECTIONS OB- 

 SERVED DURING SOLAR 

 ECLIPSE OF MAY 29, igigi 

 1. Table 1 summarizes the available obser- 

 vational data for deriving the amount of 

 deflection of a light ray grazing the sun's 

 limb as observed on the earth. The sources of 



tion. If on the other hand the observational 

 results are weighted inversely as the squares 

 of the probable errors, than the weighted 

 mean results, especially IV. (1".76), are found 

 to be in close agreement with Einstein's value, 

 though the probable error (zt 0".2) is still 

 somewhat large. 



2. The weighted mean value IV. depends 



TABLE I 



Summary of All Observations Concerning Deflection of Light at Sun's Limi 



UemarTcs: No. 1 was derived from Dr. Campbell's statement (see Science, March 26, 1920, page 310) 

 that the mean of their results "came out at 0".08 or 0".15, according to which of Einstein's hypotheses 

 was adopted"; the probable error of one star position is given as 0".5, but the probable error of the 

 mean result is not stated. Nos. 2, 3 and 4 are given in Monthly Notices, B.A.S., Vol. LXX., p. 415, Feb- 

 ruary, 1920. (See Science, March 26, 1920, p. 308.) 



the data are given in the remarks below the 

 table. ISTo. 2 has been rejected by the British 

 astronomers because of the difiuseness of the 

 star-images on the photographic plates ob- 

 tained with the astrographic object glass of 

 the Greenwich Observatory used in conjunc- 

 tion with a 16-inch coelostat, the figure of 

 which apparently changed appreciably during 

 the plate-exposures. It will be observed that 

 the indiscriminate mean results, I. and II., 

 would indicate a value about midway between 

 that (0".87) computed on the basis of the 

 Newtonian Mechanics and that (1".'74) com- 

 puted according to Einstein's law of gravita- 



1 K6sum6 of papers presented before the Ameri- 

 can Philosophical Society at Philadelphia (Feb- 

 ruary 6 and April 24), the American Physical So- 

 ciety (February 28 and April 24), and Bureau of 

 Standards at Washington (May 7, 1920). For a 

 general account of observations concerning the solar 

 eclipse of May 29, 1919, and the Einstein effect, the 

 reader may be referred to the author 's ' ' R6sum6, ' ' 

 published in Science, March 26, 1920, pp. 301-312. 



chiefly upon Crommelin's result (N"o. 3), ob- 

 tained at Sobral, Brazil, during the solar 

 eclipse of May 29, 1919, from 7 photographic 

 plates, using a 4-inch lens of 19-foot focus 

 and an 8-inch ccelostat, and from similar 

 check-plates obtained at the same station be- 

 fore sunrise between July 12 to 18, 1919. 

 These observations appear to be the best ones 

 for undertaking a critical analysis of the 

 results with the view to ascertaining, if pos- 

 sible, whether any other effect has been 

 measured than that accredited to the sun's 

 gravitational action. The following results of 

 a preliminary analysis, as made by the De- 

 partment of Terrestrial Magnetism at Wash- 

 ington, are based partly upon data already 

 published in the British journals and partly 

 upon those very courteously supplied by the 

 Astronomer Eoyal, Sir Frank Dyson, to whom 

 we desire to return our appreciative thanks. 

 The chief purpose of our investigation was to 

 ascertain the possible bearing of the geo- 

 physical observations, made by the two chief 



