( 608 ) 



value of (/. Ill the second place, if the layer is so thick that the 

 second term in tiie fornnda lias a cerlain intlnence, tor a iii\en 

 value of \:. die amount of absorption will increase wiUi </. It will 

 tlierefore increase by a cdinpix'ssidii of liie uas, if by this means die 

 coeflicient // takes a larger \aiiic. An elfect of diis kind has really 

 been ol)ser\e(l by AN(iSTK()M ') in his e\|)eriiiieiils on die absoiphon 

 produced by carbonic dio.xyde. 



This result coidd have iieeii pre(Hcted by theory if die idea that 

 the succession of regular \ilirations would be distiirlied by the colli- 

 sions between the molecules had lieeii confirmed ; tiien, by an increase 

 of die density, the time r would becdiuc siiorter and the formula (18) 

 would give a larger \aliif for du' coeflicieul i/. As it is, the vilira- 

 tions must be supjiosed Id 1k' (hslurb(>d by sdiiie odicr cause (§ 14) 

 and we can only infer from An(;sthÖiM's measureiiients that the intln- 

 ence of this cause must depend in some unknown way on tlie density 

 of the gas. 



§ 17. Thus far, we have constaihly assumed in our calculations 

 that the coeflicient i/ is very much larger than unity ; this hypotiiesis 

 has been conlirnied by the values given in § 12 and, to Judge from 

 these numbers, i( would e\eii seem hardly probable dial j; can in 

 any case have a value (Mpial lo, or smaller than I. Yel, there is a 

 |ilieiioiueii(iii w liicl: can only be CNplaineil by asi-ribing to i/ a small 

 value. This is the dissymmetry of the Zekm.\n elfect, which has been 

 predicted by \'oiut's theory '■') and has shown itself in .some experi- 

 ments of ZiiKM.VN''). In so far as we are here concerned with il, it 

 consists ill a small iiiiM|iialily, observable only in weak magnetic 

 fields, of die distances at w liicli die Iwo outer coiii|ioiieii(s of the 

 triplel are situated from die |»lace of llii' original spectral line. 

 Whereas in slrtuig lields die posidon of lliese components is ileter- 

 mined by die eqiiadons ^ = -|- ? and t = — ?, it corresponds to 

 g = and §=I, if iliL' magnetic intensity is very small. 



VoKiT has immediately |)oiuted out diat the dissymmetry can only 

 e.vist, if >j is not very large. Yet, from (he fad that the elfect could 

 scarcely be detected In Zkkm \n. lie coiicliides that the coeflicient must 



'I A.NGSTRÖM, IJber dii' .\ljliaiii<igl<eit ilri' AhT:ür|jliuii iler Gase, besüiiders der 

 Kolileiisaure, voii dor Diclitr, Ann. Pliys., (i (l',)0l), p. Iü3. 



') VoiGT, Uber eine Dissymmetrie der ZEE.M.^N'schen nonnaleii 'fii|>lels, Ann. 

 l^hys., 1 (1900), p. 37G. 



') Zeeihn, Sume observations eoiiceniing an asyniineti-icai cliange of tlicspcctml 

 lines of ii'oii, |-adiatir,g' in :t iiiagnetic field. Tliese Proeeedini^'s, II (190(1), p. -298. 



