December 7, 1899] 



NATURE 



13 



Set B or A 



Set A or B 



Principal 



Subordinate 



I 

 Subordinate 



I 

 ( Principal 



Subordinate 

 I I 



Subordinate 



Fluted spectra 

 and we get six depolymerisations. 



The number of lines measured by Runge and Paschen 

 in the spectrum of oxygen at low temperature was 76 ; 

 of these the six series referred to contain 56, leaving 20 

 residual lines. Now if we employ a strong induced 

 current at atmospheric pressure, we practically extin- 

 guish these six series of lines and produce a new spectrum 

 altogether, containing a still greater number of lines : 

 114, according to Neovius. Only one line is common to 

 his table and that of Runge and Paschen. About the 

 series conditioning of these new lines we are at present 

 profoundly ignorant. 



Let us take the simplest course in harmony with the 

 principle of continuity, and suppose that the great 

 number of new lines is due to the breaking up of the 

 molecules of the upper principal series given in the 

 previous table into representatives of a still finer form, as 

 hydrogen, as we know it, is broken up into a finer form 

 at the highest stellar temperatures. 



Have we, on the line of reasoning we are pursuing, 

 any means of estimating the number of finer forms 

 which may be at work to produce the 113 new lines ? 



One possible way — a statistical way — seems open to 

 us. Taking the number of lines already recorded in the 

 spectra roughly between \ 7000 and X 2600 of the follow- 

 ing substances, which give us three series — lithium, 

 sodium, potassium, helium, asterium, hydrogen — we find 

 that the number of lines jn each series and the total 

 numbers are as follows : — 



Totals 



39 



This indicates that in oxygen we are slightly above 



the average with 1-= 28 lines per set. If we take the 



facts for oxygen itself, which give us 56 lines for two sets of 

 three, the 113 lines will givealmost exactly four additional 

 sets of three series, and therefore the possibility of twelve 

 more depolymerisations if this method of simplification 

 is considered. 



Of course we can halve the number of depoly- 

 merisations by assuming that the fluting molecule 

 instead of being depolymerised is broken up into x and 

 y, the bases of the two systems of series. 



Now it is this last crop of new lines alone which are 

 represented in the hottest stars, and no one, I think, 

 will now urge that some kind of simplification which 

 may include depolymerisation has not taken place before 

 they were brought into evidence. 



Our base of 16 then vanishes, and with it the 

 previously considered possible atomic weights of the 

 forms of magnesium and calcium which precede the 

 appearance of oxygen in the hottest stars. We must 

 therefore assume further depolymerisations in the case of 

 these metals beyond those considered in the first 

 instance. 



NO. I 57 I, VOL. 61] 



I now come to another point. How do the above 

 considerations bear upon hydrogen with its atomic 

 weight of I ? Of this hydrogen we know nothing 

 spectroscopically. There is evidence that it is broken 

 up into something which gives the complicated structure 

 spectrum with hundreds of lines not yet sorted into 

 series, again into the one series seen in our laboratories 

 and in the cooler stars, still again into two other forms 

 we cannot get here. 



Let us apply the statistical method we employed in 

 the case of oxygen. 



In the region included in these inquiries the number 

 of hydrogen lines in the three series referred to is 17: 

 Hasselberg has measured 454 lines in the structure 

 spectrum between >X 642 and 441. Now if this spectrum 

 is built up of series similar to those observed at the 

 highest temperatures, we must have more (seeing that 



Hasselberg's work was limited) than ^^^ = ± 27 series 



or 9 sets of 3 each. We deal then altogether with 12 

 depolymerisations. 



But to be on the safe side, let us assume 6 on the 

 ground that the lines in the series may be more numerous, 

 and that some of Hasselberg's lines may be due to 

 flutings. It will be clear that the masses or "atomic 

 weights " we arrive at must be very small. Here is the 

 story : — 



Spectrum. 

 Line spectrum 



Flutedj spectrum i 



Continuous spectrum 



Where existent. Series, &c. 



I Terrestrial Subordinate 



{Principal 

 Subordinate 

 Subordinate 



I Set A ij;T'?^\ - 

 Terrestrial jSuborchnae ... 



V J^ Subordinate 



f Hydrogen weighed 



\^ in the cold ... 



' SetB 

 Terrestrial 



Mass. 



•0019 



•0039 



•0078 



•0156 



•0312 



•0625 



•125 



•25 



•5 



Such a conclusion as this, and therefore the reasoning 

 which has led up to it, must stand or fall according as 

 science knows anything of such masses. 



I shall show subsequently that, thanks to the in- 

 vestigations of Prof. J. J. Thomson, science is beginning 

 to know a great deal of such masses, and the result of 

 this work may therefore favour the view that poly- 

 merisation is a vera causa for molecular complexity, at 

 all events in the cases of elements of low atomic 

 weight. 



Let us then consider the case of those elements the 

 atomic weight of which is greater. In the first stages of 

 evolution, in which we deal with substances of relatively 

 low atomic weight, the stellar evidence supplies us with 

 definite landmarks, and these are definite because the 

 spectra of the hottest stars are not overcrowded with, 

 lines. After we have passed the gaseous and proto- 

 metallic stages, however, we find the spectra full of 

 lines which we see at the temperature of the arc, and 

 metals of relatively high atomic weight and melting 

 point are involved ; the exact sequences are naturally 

 more difficult to follow, and therefore the method of 

 evolution may escape us. 



Kayser and Runge have shown that the melting point 

 has a profound influence on the " series " conditions. 

 Those with the highest melting-points, such as barium 

 and gold, present us with no series. There is generally 

 such a flood of lines that it has been so far impossible to 

 disentangle them ; we have the " structure spectrum " of 

 hydrogen repeated in these metals at arc temperatures- 

 in the so-called " arc spectrum." 



NORM.\N LOCKYER. 

 {To be continued.) 



