September 22, 192 1] 



NATURE 



125 



heliometer can be used to measure angular dis- 

 tances of several minutes with an accuracy second 

 only to that of the modern photographic tele- 

 scope. The other determinations of parallax were 

 by Struve, who observed o Lyrae relative to faint 

 stars in its neighbourhood by means of a position 

 micrometer as used for double stars, and by 

 Henderson, who used meridian observations in 

 both co-ordinates of a Centauri. 



During the next fifty years a number of ob- 

 servers made parallax determinations by these 

 three methods, but although they showed great 

 skill in their work and prosecuted it with the 

 greatest assiduity, it cannot be said that many 

 trustworthy results were obtained. The errors to 

 which even the best results are liable are shown by 

 the following seven determinations of the parallax 

 of Procyon made by Elkin with the Yale helio- 

 meter : — 



o-2r7±ooi8 

 0-461 ±003 5 

 0367 ±00 18 

 0-366 ±0023 



0-503 ± 0-049 

 0-294 ±0019 

 0-228 ±0-020 



Soon after the first application of photography 

 to astronomy it was found that star places could 

 be determined with great accuracy from photo- 

 graphic plates. It was only natural that attempts 

 should be made to apply the new method to paral- 

 lax determination. The initial results showed no 

 greater accuracy than those obtained visually, but 

 gradually the difficulties have been overcome, and 

 a remarkable degree of accuracy attained. It 

 might have been supposed that the development of 

 a photographic plate would lead to a distortion 

 sufficiently great to vitiate the results, but appar- 

 ently this is not the case. In fact, the distortion 

 is less than o-ooi millimetre and can be ignored. 

 The difficulty is to eliminate systematic errors in 

 the apparent centres of the star images on the 

 plates so as to get the photographs to faithfullv 

 represent the heavens at the different epochs. 

 Provision must be made for the automatic elimina- 

 tion of every imaginable source of systematic error, 

 since every preconceived source of possible error 

 has turned out to be a reality. The most im- 

 portant precautions to be taken were pointed out 

 about twenty years ago by Kapteyn in the first 

 of the Groningen Publications. These include the 

 taking of all the photographs under as nearly as 

 possible the same instrumental conditions— the 

 telescope should always be on the same side of 

 the pier, and as nearly as possible in the meri- 

 dian. These precautions are now obvious, as not 

 only is the objecti\ e liable to behave differently 

 in different positions, but the atmospheric effect 

 might var>' differently for the different stars, as 

 they are not all of the same colour. A more seri- 

 ous source of error, called the "guiding error," 

 was pointed out by Kapteyn. For the brighter 

 stars an impression is produced on the photo- 

 graphic plate more quickly than for the fainter 

 stars, so that if during the exposure an error in 

 guiding allows all the stars to be slightly dis- 

 NO. 2708, VOL. 108] 



placed for a short time, the brighter stars wilt 

 show an elongated image with a displaced centre, 

 while the fainter stars will show round undisplaced 

 images. For this reason Kapteyn urged the im- 

 portance of good guiding. But it is impossible 

 to guide sufficiently well, and the difficulty was 

 satisfactorily surmounted only when Schlesinger 

 introduced the occulting shutter. This is a sector 

 which is made to rotate rapidly in front of the star 

 the parallax of which is to be determined. By 

 reducing the opening in the sector, the time during 

 which the " parallax star " is exposed is reduced 

 relatively to the other stars. It is usual for the 

 comparison stars to be of the loth or 'iith 

 magnitude, while the "parallax stars" are gener- 

 allv considerably brighter. It is possible by means 

 of the rotating sector to cut down the brightness 

 bv five magnitudes. For the very bright stars this 

 is not enough, and some observers have used two 

 rotating sectors to give the required reduction. 

 Another method is to place a screen in front of 

 the brighter stars. It is now generally recognised 

 that it is most important to have the parallax 

 star and the comparison stars forming images of 

 nearly equal size and density. At the same time, 

 every care is made to have the guiding as accurate 

 as possible. In this connection the exposure 

 should be as short as will produce readily measur- 

 able images — two or three minutes with photo- 

 graphic refractors with an aperture of 20 or 

 30 in. 



Kaptevn's plan was to photograph the region 

 under consideration at three different epochs on the 

 same plate, which was stored between the ex- 

 posures and developed only after the third epoch. 

 The times of exposure were chosen so as to give 

 maximum parallactic displacement in one direction 

 at the first and third epochs, and maximum paral- 

 lactic displacement in the opposite direction at the 

 middle epoch. For example, a region might be 

 given three exposures in May of one year, six 

 in the following September, and three in the 

 next May. The telescope would be moved 

 slightly between the exposures, and for each star 

 there would be twelve images on the plate. As 

 all the images of each star would lie close to- 

 gether, it would only be necessarv to measure 

 verv small distances on the plate, while by a sym- 

 metrical arrangement of the images any possible 

 distortion of the film would be eliminated. This 

 method is ideal, and was applied to some extent, 

 but on account of bad weather interfering with 

 the exposures it has practically been abandoned. 

 The photographs at the different epochs are now- 

 taken on separate plates, and this method allows 

 of greatest weight being given to those exposures 

 made under the best atmospheric conditions. 



The difficulties to be overcome having been fully 

 realised, and the necessary precautions having 

 been devised, a large scheme for the determina- 

 tion of parallaxes has been undertaken. In this 

 work the .\lleghenv. Dearborn, Greenwich, 

 McCormick, Mount Wilson, Sproul, and Verkes 

 Observatories take part. The following table 



