of the Centres of Emission of D Lines of Sodium. 1019 



penetrate the vapour. Precisely similar phenomena were 

 subsequently detected by photography in the case of mercury 

 vapour at room temperature in a bulb of quartz excited by 

 the mercury line 2536 *. 



In the paper on the resonance of sodium vapour, it was 

 suggested that an experiment of great interest would be to 

 excite the vapour by the light of one sodium line only, and 

 examine the resonance-light with a spectroscope : in this 

 way it would be possible to determine whether the two 

 centres of emission could be separately excited. 



The experiment appeared, however, to be a difficult one 

 to carry out, and no attempt was made at the time. 



Recent improvements by one of us f in the method of 

 carrying out the experiment enable a much brighter reso- 

 nance to be obtained, and make it possible to extend obser- 

 vations over a period of ten or fifteen hours with a single 

 bulb, whereas with the original apparatus the experiment 

 was over in three or four minutes. These improvements 

 have made it possible to carry out at last the suggested 

 experiment on the separate excitation of the centres of emis- 

 sion. The device employed for the removal of D^ or D 2 

 from the exciting beam has been recently described by one 

 of us t and is an improvement of a method used in an earlier 

 investigation of the anomalous dispersion of sodium vapour. 



It is a polarization method, and may be briefly described 

 as follows : — 



If plane-polarized monochromatic light is passed through 

 a plate of some doubly refracting crystal with its direction 

 of vibration making an angle of 45° with the axis, it .will 

 emerge plane-polarized parallel to the original plane for 

 certain thicknesses of the plate, and plane-polarized at a 

 right angle to this plane for other thicknesses. For inter- 

 mediate thicknesses it will be elliptically or circularly 

 polarized. 



If we employ a plate of quartz 30 mm. thick the emergent 

 waves of D A and D 2 of sodium will be plane-polarized at 

 right angles to each other, and either can be quenched by a 

 nicol suitably oriented. If white light is used, and analysed 

 by a spectroscope, the spectrum will be furrowed by dark 

 bands, the distance between a bright and a dark band being, 

 in the yellow region, 6 Angstrom units, the distance between 

 the D lines. As it is necessary to employ a large condenser 

 and work with very divergent and convergent cones of light, 



* Wood, Phil. Mag. May 1912. 



t Dunoyer, Journal de Physique, Jan. 1914. 



% Wood, Phil. Mag. March 1914. 



