478 Prof. R. W. Wood on the 



helium, for it has a smaller affinity for electrons according 

 to Franck, and exhibits most remarkable electrical properties. 

 Professor Collie showed me a tube about halt' a metre long 

 containing neon at a pressure of an atmosphere and a half, 

 through which the discharge of an induction-coil passed in 

 preference to tm alternative spark-gap in air of less than a 

 centimetre arranged in multiple with the tuhe. In neon, 

 however, the intensity of the emitted light is about the same 

 as in helium at the same pressure, and much more, it is 

 localized in the resonance lines. 



This matter requires a more careful investigation, for I 

 was able to devote not more than the two or three days 

 immediately preceding my departure from London to this 

 part of the investigation. 



Polarization of tlie Light in Resonance Spectra. 



The polarization of the light of fluorescent gases was dis- 

 covered by the author more than three years ago (Phil. Mag. 

 July 1908). It was found to be much stronger in sodium 

 and potassium than in iodine vapour. This subject has been 

 taken up again, in view of its great importance in connexion 

 with the genesis of resonance spectra, and I now find that it 

 is possible to obtain almost as strong polarization with iodine 

 vapour as was formerly observed with sodium. This is 

 probably due to the fact that in the earlier work very small 

 glass bulbs were used, which quite possibly were insufficiently 

 exhausted. In the case of the resonance spectra I never felt 

 quite satisfied with my observations, though I obtained what 

 I considered fair evidence that when the fluorescence v\as 

 excited by polarized monochromatic light, the light was 

 polarized in all of the lines of the resonance spectrum. This 

 has been fully corroborated in the present work, by a method 

 which admits of no doubt. 



The percentage of polarization was first determined in the 

 case of iodine vapour excited by polarized white light, by 

 the same method used in the earlier investigation. 



It was found that the Savart fiinges could be destroyed by 

 two glass plates at an angle of 43°, and by one glass plate 

 at an angle of 56°, and the percentage of polarization comes 

 out as 17 in each case. For sodium vapour I found 20 per 

 cent, of polarization, or in the case of very rare vapour at a 

 low temperature, as high as 30 per cent. 



The polarization was easily observed when the fluorescence 

 was excited by the mercury arc, the large glass bulb, 20 cm. 

 in diameter, used for the demonstration of the fluorescence 

 before a large audience, being found best adapted to the 



