SIR DAVID GILL EDDINGTON. 517 



of the long- focus refractor, and he fully shared the modern tendency 

 to depend more and more on photography. But there is one ad- 

 vantage of the heliometer over the photographic refractor, both 

 for solar and stellar parallaxes, on which Gill strongly insisted — 

 the heliometer measures are independent of the color of the object 

 under observation. He maintained, and confirmed by experimental 

 observations, that the skilled observer in making • coincidences of 

 the images matches the colors and not the most intense points of 

 the minute spectrum caused by atmospheric dispersion. This is a 

 refinement obviously impossible in photography, and, for example, 

 it is well known that the doubtful effect of atmospheric dispersion 

 leaves a little uncertainty in the solar parallax deduced from the 

 photographic observations of Eros. 



So early as 1872 Gill had begun to plan a series of determina- 

 tions of stellar parallax with a micrometer attached to his reflector — 

 an investigation which was interrupted by his removal to Dun 

 Echt. On his appointment to the Cape he began to apply his 4- 

 inch heliometer to this work. In this he was joined by Elkin, as 

 a volunteer observer, and they set to work on a program of 9 stars, 

 including Sirius, Canopus, a and ^ Centauri, with some stars of 

 exceptionally large proper motion. The most important outcome 

 of this work was the parallax of a Centauri, 0".75, with a probable 

 error of only a hundredth of a second of arc. The desirability of 

 a larger instrument with some alterations of design soon became 

 apparent, and in 1887 a 7-inch heliometer was constructed at a 

 cost of £2,200. With this. Gill and Finlay, and afterwards De 

 Sitter, measured 17 stars, including 12 of the brightest in the 

 southern sky, in most cases with a probable error as low as ±0".01. 

 These results were of great interest, establishing the remoteness 

 and intense luminosity of some of the brightest stars, such as 

 Canopus and Kigel. Whenever they have been put to the test 

 Gill's values have always been confirmed. Spurious parallaxes are 

 a great bane in stellar investigation, and, at least until recentlj'^, 

 few observers have escaped an occasional bad error; but Gill's 

 l^arallaxes can always be relied on. His general accuracy has been 

 equaled, perhaps a little surpassed, by some modern photographic 

 determinations ; but when we compare the sizes of the instruments — 

 the 40-inch telescope at Yerkes or the 26-inch at Greenwich with 

 his 7-inch heliometer — we must marvel at the precision he could 

 obtain. The following table (given by him) will show the com- 

 parative accuracy of his work. It gives the probable error of the 

 measured position of a parallax star: 



Cambridge refractor (19.3 feet focus), 4 exposures ±0". 04S 



Yerkes refractor (63 feet focus), 3 exposures ±0 .026 



Heliometer, one complete observation, i. e., 16 pointings ±0 . 036 



