144 



NATURE 



[March 31, 192 1 



his direction, is illustrated in Fig. 2. It con- 

 sists of a horizontal telescope pointing to the 

 west and provided with two similar objectives, 

 A and B, in front of which are placed right-angled 

 prisms, C and D, which reflect the light from two 

 stars into the telescope. The prism D is used only 

 for observing the Pole-star, and can be turned 

 about two perpendicular axes by rods E and F. 



Fig. 2. — The Meridian Photometer. 



The prism C can be turned around the axis of the 

 telescope, and its position read by a circle, G, so 

 that a star of any given declination can be ob- 

 served on the meridian ; there is also a slight 

 adjustment for enabling it to be viewed for about 

 one-quarter of an hour before or after meridian 

 passage. A double-image prism, K, made of Ice- 

 land spar compensated by glass, is placed near 



NO. 2683, VOL. 107] 



the focus of the objectives, and divides each pencil 

 of light into two ; the angles of the spar and glass 

 prism are so adjusted that the two central pencils 

 (one ordinary and one extraordinary pencil) are 

 made to coincide and to pass nearly through the 

 centre of the eyepiece L. In this way errors which 

 might result from having two emergent pupils or 

 from the pencils passing through different parts 

 of the eyepiece are avoided. In front of the eye- 

 piece is placed a Nicol, M, and an eye-stop, N, 

 cuts off the two outside pencils. A graduated 

 circle, O, is attached to the eyepiece and Nicol, 

 and the four positions of the Nicol are observed 

 in which the two images are equal in brightness. 

 Since the beams from the two stars are polarised 

 at right angles, if I is the angle counted from the 

 position where the image of Polaris disappears, 

 then the ratio of the brightness of the star under 

 comparison to that of Polaris is tan^I. In taking 

 the observations, readings are obtained with the 

 image of Polaris first on one side and then on the 

 other side of the star. This photometer is accurate 

 in use, and has the advantage over the Zollner 

 type that similar images are compared. It has 

 several disadvantages ; the two stars are compared 

 through different object glasses which cannot in 

 general be interchanged. Stars of low declination 

 are compared with a star at a very different alti- 

 tude, so that appreciable errors may occur on 

 account of the variations to which the transparency 

 of the atmosphere is liable ; with the Zollner 

 photometer, on the other hand, a star can always 

 be compared with another of about the same alti- 

 tude. The optical combination also does not per- 

 mit of very good images, and there is no pro- 

 vision for matching the colours of the two images. 

 It is also limited in its application to stars near 

 the meridian. At the time the Harvard observa- 

 tions were made the variation in brightness of 

 Polaris had not been discovered. After its dis- 

 covery, the variation was detected in the residuals, 

 although its total range is quite small. 



In another type of photometer which has been 

 greatly used, a neutral wedge of uniformly gradu- 

 ated absorption is employed, and the reading is 

 taken of the position of the wedge when the star 

 under observation just becomes invisible. Owing 

 to the strain on the observer's eyes caused by 

 these observations, which are liable to give rise 

 to personal errors of variable amount, and to the 

 impossibility of obtaining an absolutely neutral- 

 tinted wedge, this type of photometer does not 

 give results of the same order of accuracy as the 

 two described above. 



Although the theory of the determination of 

 visual magnitudes is very simple, there are many 

 possible causes of error, mainly of a physiological 

 nature, arising from the necessary use of the 

 human eye. Most of these are more important 

 when very faint stars or stars differing much in 

 colour or brightness are observed, though in the 

 Zollner photometer difference in colour can be 

 compensated to a certain extent. Errors arising 

 from the observation of stars near the threshold 



