ON OUR KNOWLEDGE OF SPECTRUM ANALYSIS. 279 



had the curious idea of examining whether the rate of propagation of a 

 ray of light in space depended on the amplitude of vibration. The wave- 

 length was measured by means of interference fringes of long difference 

 of path (Newton's rings), and he was incidentally led to inquire into the 

 possibility of a displacement of the centre of the sodium lines, when 

 larger quantities of sodium were introduced into a Bnnsen flame ; in 

 which case it is known the lines are seen to widen. Miiller found such 

 a displacement in the same direction as Zollner, who refers to his 

 experiments as corroborating his own results. But Miilkr also found the 

 effect he was looking for ; that is to say, he found the centre of lines to 

 shift when the light was weakened after it had left the flame. Now this 

 latter part of the investigation has been subjected to a very careful 

 examination by Lippich,' who could discover no such eSeet. As there nn- 

 donbte'dly was a source of en-or in Miiller's experiments, which has not 

 yet been pointed out, we must defer our judgment also on his other 

 results. When he therefore finds a very slight shifting of the hydrogen 

 lines, due to an alteration of the power of the spark, we cannot put his 

 observations on an equal level with the subsequent negative I'esults of the 

 Greenwich observers. 



Speaking of reversals in the voltaic arc, Mr. Lockyer- adds the 

 following note : ' The absorption-line does not always occupy the exact 

 centre of the bright band. This point is occupying my attention, as it 

 raises a very interesting question connected with molecular vibrations.' 

 In the j)lates accompanying the paper we find one at least of the 

 aluminium lines between H and K slightly more expanded on the less 

 refrangible side. 



Mr. Lockyer ^ has referred to the same question in a recent paper, 

 and mentions two examples in the silver sjiectrnm. In one case (4210'0) 

 the line seemed to be much more widened on the more refrangible side of 

 the absorption line ; in the other case (4054'3), it was more widened in 

 the opposite direction. The rubidium line (4202) is also given as more 

 expanded on the less refrangible side. 



Profs. Liveing andDewar'* mention that the magnesium lines (4703) 

 (4354) widen more on the less refrangible side. 



The lines of the lower temperature spectrum of oxygen widen con- 

 siderably on the less refrangible side. This is very marked in the case of 

 the more refrangible of the two green lines, and of the red line. The centre 

 of the former line shifts through 2Xth metres.^ The less refrangible of 

 the two green lines widens more symmetrically. According to Ciamician,^ 

 most of the mercury lines show this one-sided widening, and with some 

 of them it is so marked that they seem exclusively expanded towards one 

 side only. 



I have only come across two cases in which lines seemed to be more 

 widened on the most refrangible side, and neither of them seems to be 

 established beyond doubt. The first is the one in the spectrum of silver men- 

 tioned by Lockyer and quoted above. But Profs. Liveing and Dewar did 

 not notice the reversal of the line in question, the wave-length of which they 

 give as 4208. A new line, however, came out at 4211-3; that is on the 

 less refrangible side. It is possible that the dark space between the two 



' Wie?i. Bcr. (2) Ixxii. p. :{.j.5 (1875). ■• Proo. Roy. Soe. xxviii. p. 'MM (1879). 

 * Phil. Tram, clxiv. p. 805 (1874). » Wien. Ber. (2) Ixxviii. p. 886 (1878). 



' Proc. Roy. Soc. xxviii. p. 428 (1879). ' Natitrc, xvii. p. 148 (1877). 



