January 5, 1899] 



NA TURE 



radiation. Now the theory which indicates that a 

 spectral hne should be slit up into a pure triplet by the 

 action of the magnetic field, assumes that the freedom of 

 vibration is the same in all directions, and it is from this 

 that the resolution into triplets occurs. This assump- 

 tion is that which one most naturally makes in a first 

 attack on a problem of this nature, but no one making 

 it would be surprised if the facts did not turn out more 

 complicated than the prediction of such a solution. For 

 example, it is quite possible to conceive a state of affairs 

 in which the magnetic field may constrain all vibrations 

 to take place along the lines of force, in which case the 

 side lines of the triplet would' vanish ; or, on the other 

 hand, vibration in the direction of the lines of force 

 might be impossible, in which case the central line of the 

 triplet would vanish. Indeed, one is somewhat surprised 

 that deviations of this kind from the normal triplet type 

 do not more frequently occur. In fact, when I first 

 examined the spectrum of iron, I hoped to find many 

 deviations of this kind, but failed to detect any very 

 marked difference between the behaviour of iron and 

 other substances. This is not much to be wondered at 

 when it is remembered that iron ceases to be magnetic 

 at a comparatively low temperature, and, therefore, at 

 the temperature of the spark of an induction coil, one 

 should not expect its vapour to behave much differently 

 from that of any other substance. 



However, as already stated, the normal triplet type 

 arises in theory because the orbit of the vibrating 

 electron is supposed free from constraints and perturb- 

 ations — that is, that movement is equally free in all 

 directions. When constraints are imposed, or new forces 

 arise w-hich cause perturbations in the orbit, new fre- 

 quencies will be introduced into the vibrating system. 

 Thus if an electron, or an atom, or a particle describes 

 an ellipse under a central force with frequency X, and if 

 disturbing forces came into play which cause the apse 

 line to rotate with frequency //, then, as Dr. Stoney ' has 

 shown, a spectral line arising from the original vibration 

 of frequency N will become replaced by two others of 

 frei|uencies N 4- « and N - // respectively. Again, if the 

 disturbing forces cause a precessional motion of the plane 

 of the orbit round a fixed line with frequency )i, the original 

 vibration of frequency X becomes replaced by three 

 others of frequencies .\ + n, X, and N - ;; respectively, 

 and similar phenomena arise when other periodic dis- 

 turbances occur in the orbital motion. We are prepared, 

 therefore, to find that each line of the normal triplet may 

 become itself a doublet or a triplet. - 



The disturbing forces arising from the action of the mag- 

 netic field should increase with the strength of the field, 

 so that if the distance between components of the doublet v, 

 or C or I) or E (Fig. 3), which takes the place of the central 

 line of the normal triplet,should increase with the magnetic 

 field, as it is found to do Isy experiment. In fact, if the 

 distance between the side lines of the normal triplet 

 A.\' be written in the form (/j = k^W, where H is the 

 strength of the field, and X', a quantity depending on the 

 wave-length and other constants involved in the produc- 

 tion of the particular line in question, then the distance 

 between the components of the modified central com- 

 ponent n, c, &c., may be written in the form d., = k.M.. 

 Thus as the field increases in strength the whole system 

 of lines into which any given spectral line becomes 

 resolved, separate laterally from each other propor- 

 tionately, as it were, according to a given scale Similar 

 remarks, of course, apply to systems like dd' and EE'. 



Now in any particular case, such as Bb' for example, 

 if the distance between the pair of lines b' is 



1 Dr. G. ). Stoney, Trans. Roy. Dub. Sue, Vol. iy .Series 3, p. 565, 

 iSljj. This is a very important p.iper when consideretl in connection with 

 the above-mentioned magnetic perturljalions of the spectral lines. 



- These matters are treated in further detail in a paper by the present 

 writer to appear in the forthcoming number of the Philosopldcat Magazine. 



NO. 



1523, VOL. 59] 



while the distance between the pair B is 



r/, = /!'.,H 

 there is apparently no reason why L\ should be greater 

 than, or less than k.^. Whether k^ is greater than or less 

 tha" '^'c, must be determined by the action of the magnetic 

 field on the system which produces the particular spectral 

 line in question. Accordingly we are prepared to find 

 that ill some lines the components of the central line, as at 

 n, shall be much closer together than the side components 

 at k', while in others, as at c, E and F, the distance d.. is 

 nearly ecjual to, or may be even greater than the distance 

 I-/, between the side lines. Thus, once the production of 

 a quartet of the type BB' is explained, all the other 

 modifications become intelligible. The case in which 

 the components F are wider apart than the side lines f' 

 (so that the centre, as it were, encloses the sides) is merely 

 the same phenomenon (only more accentuated; as that 

 shown at BB' where the separation A is less than (Z,. 

 This point is mentioned here specially because in some 

 cases the separation d., is actually greater than d.^. and it 

 seems to be regarded as a difficulty of a much higher 

 order than that in which occurs in the ordinary quartet,, 

 where d.^ is less than c/,. 



Lines of the former type fi', \\z. that in which d., is 

 greater than «'„ seem to have been first observed by 

 MM. Henri Becquerel and H. Deslandres (see Coiiiptes 

 rendus, t. 126, p. 997, April 4, 1S98) in the spectrum of 

 iron, and subsequently Messrs. J. S. .Ames, R. F. Earhart 

 and H. M. Reese announced that they had observed the 

 form C.g' (Fig. 3) in the spectrum of iron. In this type 

 the side lines c' coincide, or are not sensibly separated, 

 while the components of the central part G are well 

 separated.' (see Astro. PJiys. Joiiriia!, vol. viii. p. 48, 

 June 1898). The form in which this observation was- 

 described was calculated to startle, if not confound, 

 the most firm believer in theory. It was said that 

 these lines exhibited reversed polarisatiou — that is, 

 that the polarisation of the centre is that which should 

 occur in the sides, and vice ','crsa. Stated in this way 

 it is rather calculated to take one's breath away, but 

 when stated as in the foregoing, it loses all special 

 significance, viz. that it is merely a case of d., being 

 greater than </[, that is k., > /(■,, or a quartet in which the 

 distance between the horizontally vibrating constituents 

 is greater than the distance between the vertically 

 vibrating constituents. Stated in this way it falls into 

 line with the other phenomena, and is reduced to the 

 explanation of the doubling of any one individual member 

 of the normal triplet. 



Other similar modifications have been observed by 

 MM. Becquerel and Deslandres, who appear to hax'e 

 examined the spectrum of iron very thoroughly as weH 

 as the bands of carbon and cyanogen. These bands 

 they found to be unaffected by a magnetic field strong 

 enough to sensibly split up the air lines. 



Investigations demanding special attention are those 

 of Prof A. A. Michelson, both on account of his repu- 

 tation as an original investigator and by reason of the 

 nature of the apparatus which he employed. Working 

 with his interferometer. Prof Michelson concluded some 

 years ago {P/a'l. Mag:, vol. xxxiv. p. 280, 1892) that the 

 spectral lines themselves instead of being, as ordinarily 

 supposed, narrow bands of approximately uniform 

 illumination from edge to edge, are on the contrary ir> 

 most cases really complexes, some of them being close 

 triplets, and so on. This structure has never yet been 

 observed by means of any ordinary form of spectroscope.. 

 and accordingly it has been suggested that it does not 



t I have not yet observed this type, nor do my photographs verify the* 

 conclusion of Messrs. Ames, Earhart and Reese, regarding the lines men- 

 tioned by them as belonging to this type. (This is further referred to ii» 

 the forthcoming number of the PhiL Mag.") 



