258 Doublets and 'Triplets in the Spectrum. 



to the author) that now the three constituents of the triplet 

 partially overlapping one another, the maxima are conspicuous 

 and the inner part appears dark by contrast. Thus there 

 would be no reversal in our case. This really seems to be so, 

 as is confirmed by the following experiment. The above- 

 mentioned hazy dark line being visible a nicol was placed in 

 the beam so as to get rid of light whose plane of polarization 

 is vertical ; now only the bright line which emits horizontal 

 vibrations remains visible, but without the slightest trace of a 

 reversal. For this observation it is of course very desirable 

 to use a bright grating. I had no opportunity of obtaining a 

 measurement of the doublet to be seen along the lines of force. 



18. A few words may perhaps be said here concerning 

 Prof. Michelson's paper in the July number of the Phil. 

 Mag. Michelson applies his beautiful method of the inter- 

 ferometer to our subject, and there can certainly be but one 

 opinion as to its particular advantages in such cases. Some 

 of his results seem at first sight at variance with mine. Only 

 in one case is there perfect agreement between Michelson's 

 results and mine, viz. when the light is emitted along the 

 lines of force. We both get a doublet in this case. Michel- 

 son adds that a broadening is inappreciable. Evidently he 

 means a broadening of the components, which I did not 

 advocate, though it exists in a slight degree in some cases. 

 I referred in my first paper only to a broadening, because I 

 had not yet resolved the doublet or the triplet. It is of course 

 a proof of the superiority of Michelson's method that with 

 relatively weak magnetic forces he could recognize the 

 duplicity, whereas for me it was not so ejtey. 



Looking across the lines of force I have observed a triplet, 

 whereas Michelson also for this case finds a doublet, though 

 less clearly marked than in the other case. This certainly at 

 first sight seems a great difference. I venture, however, to 

 give the following explanation : — 



Let us assume, not only that the visibility-curve is practically 

 the same as that due to a doublet, but also that it is un- 

 doubtedly proved that a triplet cannot give a practically 

 identical curve in some cases, of which I for myself am 

 not sure at present. Granting that we must assent to a 

 doublet the question arises whether there cannot be assigned 

 another cause for the difference. I think there can. Michelson 

 in making his experiments was yet imacquainted with the 

 particular states of polarization in the triplet. I think that a 

 sufficient reason for the difference mentioned may be found in 

 the perpendicular polarizations of the outer lines and the 

 central one of the triplet (§ 13) or of the central band and the 

 outer edges of triplet a (§ 6). 



