TRANSACTIONS OF SECTION A. 347 



temperatures. Applying the same process to the case of electrolytes which give 

 three Ions on dissociation, the following equations are obtained corresponding to 

 (a) and (6). 



a? 



— , = Const. 

 (1 - «)V* 



s 



a? 



= Const. 



(1 - a)V* 



which should hold for dilute solutions as well as Rudolphi's and van't Hofi'a do for 

 binary electrolytes. 



3. Anomalous Dispersion and Solar Phenomena. 

 By Professor P. V. Bevan, M.A., Sc.D. 



If light from a non-uniform source, such as an arc light, be sent through a 

 tube containing non-homogeneous vapour of a metal which can show anomalous 

 dispersion, and then an image of the source be focused on the slit of a spectro- 

 scope, an apparent double reversal of certain lines may appear. This pheno- 

 menon was observed by the author with potassium vapour first and with other 

 alkali metals later. The explanation of the phenomenon is simple. If the 

 brightest parts of the image formed without the tube of metallic vapour be 

 just above or below the slit of the spectroscope, then when the metal vapour is 

 in the track of the train some of the light of wave length very near that of an 

 absorption line of the metal may be deviated by the metallic vapour sufficiently 

 to fall on the slit. In the most simple case when the vapour acts as a prism 

 with refracting edge horizontal and directed upwards, light of wave length just 

 greater than that of an absorption line is diverted downwards, light of wave 

 length less than that of the absorption line is deviated upwards. When, then, 

 we have the images of the two poles of the arc, one just above and one just 

 below the slit, we may have light from those poles on each side of the absorption 

 line getting into the spectroscope giving bright lines on a darker background, 

 which is formed by the light which has come from elsewhere through the vapour 

 and showing a dark broadened absorption band. This appears to be a double 

 reversal of the line — as a dark line in the position of the actual absorption line 

 always appears between the two bright lines. Various modifications of this 

 appearance can be obtained and similar results to those described by Julius. 

 It seemed as if this phenomenon might have a bearing on solar phenomena and 

 might give some help in the interpretation of spectro-heliographs. The essence 

 of the phenomenon, is that a non-uniform source of light should be employed — • 

 it is clear that with a point source of light no such double reversal could occur, 

 and that with a uniformly bright source no such phenomenon could occur with 

 vapour acting as a single prism. It is also clear that with any number of masses 

 of vapour of identical nature and a uniform extended source behind them no 

 double reversals could occur. Suppose, however, a uniform plane of luminous 

 matter represents the photosphere and that this is observed normally. In 

 front of this suppose a mass of dense vapour of sodium, for example. This 

 would give broadened absorption lines in light received on a spectroscope. If 

 now above this there is another mass of sodium vapour of less density than the 

 first — this will be transparent to light near the centre of the absorption lines 

 which is absorbed by the first mass. But for this light the refraction index of 

 the vapour may be greater or less than 1, so that light which has come from 

 the parts of the photosphere outside the part shielded by the first mass of 

 vapour may be deviated by the second mass so that it emerges from this nor- 

 mally to the photosphere. We shall have then, when the image of this part of 

 the sun is thrown on the slit of the spectroscope, in addition to the broadened 

 absorption bands, bright light very near the centre of the absorption band and 

 on each side of it. This will appear, therefore, as a double reversal. The con- 

 ditions for such a phenomenon are simple and, in fact, seem very likely to 

 occur ; the effect would be due to numbers of these masses of vapour, and the 

 conditions necessary are simply that the masses of vapour near the photosphere 

 are of greater density than those further away; which is what we should 



