146 Prof. E. W. Wood on the Magnetic 



snugly over the porcelain tube which contained the sodium, 

 the tube passing through the pole-pieces of the large electro- 

 magnet, built by Professor Rowland for the study of the 

 Zeeman effect. Sunlight has been used in all of the recent 

 work, as it was found that the Fraunhofer lines formed 

 useful reference -marks in determining the wave-lengths. 

 The rotations observed have ranged from 90° to 1200°, and 

 the errors in the determinations of wave-lengtb do not 

 amount to more than one-sixtieth of the distance between the 

 D lines. As was expected, it was found that errors of con- 

 siderable magnitude, considered from the present point of 

 view, had occurred in the previous work, and it now seems 

 as if data of greater accuracy than those at hand cannot be 

 hoped for. Though I have fully described the arrangement 

 of the apparatus in the earlier paper, it may be well to 

 recapitulate. Sunlight, reflected from a heliostat, is passed 

 in turn through a Nicol prism, the sodium tube, and a second 

 nicol, after which it is focussed on the slit of the spectroscope. 

 The sodium tube must be highly exhausted, for, as I have 

 shown, the presence of even a small amount of any inert gas 

 masks the effect. The action of the gas does not seem to be 

 direct., that is the rotatory power is not affected by the gas, 

 which acts rather by modifying the nature of the absorption 

 of the vapour. Of this more will be said presently. 



The large spectrometer of the University was used, as the 

 instrument is mounted on a pier of masonry, and considerable 

 difficulty was experienced in the earlier work with the con- 

 cave grating on account of vibrations. The objectives of this 

 instrument are eight inches in aperture, and have a focal 

 length of about eight feet. The eyepiece of the instrument 

 was furnished with a filar micrometer, and when working in 

 the spectrum of the second order it was possible to determine 

 wave-lengths to within 1/100 of an Angstrom unit. The 

 slight variations of the density of the vapour, and the breadth 

 of the bright rotation bands upon which the settings were 

 made, made it impossible to get readings much closer than 

 1/10 of a unit in the actual experiment. 



With the nicols crossed, the field of the spectroscope 

 appears dark until the sodium vapour is formed by raising 

 the temperature of the tube. On throwing the current into 

 the magnet, a pair of narrow bright lines appear at each of 

 the D lines. The light is made up of wave-lengths imme- 

 diately bordering the two absorption-lines, for which the 

 plane of polarization has been rotated 90°. As the tempera- 

 ture of the electric oven gradually rises these lines separate 

 and broaden, and we presently see a multitude of finer lines 

 between them, corresponding to rotations of 270°, 510°, 



