536 



NA TURE 



[April 6, 1899 



Another important series of facts was soon brought to 

 the front. I'liicker and Hittorf in the year 1S65 an- 

 nounced that "there is a certain number of elementary 

 substances which when differently treated furnish two 

 kinds of spectra of quite a dift'erent character, not having 

 any line or band in comnion." The difference in character 

 to which reference is here made consists in the spectrum 

 produced at the lower temperature being composed of 

 flutings, which are replaced by lines when the higher 

 temperature is reached. 



This was the first blow aimed at the general view — one 

 element one spectrum — to which I have referred above. 

 It was met in two ways. 



Taking the line spectrum as representing the true 

 vibration as the chemical unit, I have already shown that 

 the continuous spectrum was explained as due to its 

 physical environment, the solid or licjuid state. This, 

 then, had not to be considered from the chemical point 

 of view. 



The fluted spectra were boldly ascribed to " im- 

 purities," but not alwajs wisely, for, to get rid of the 

 difficulty presented by the two spectra of hydrogen, 

 two perfectly distinct spectra were ascribed to acetylene. 

 Again the "bell-hypolhesis" was suggested, accoiding 



Fic. 3.— The method of throwing an im.-tge of the light source (in this case 

 on the slit plate of a laboratory spectroscope. 



to which the spectrum did not depend so much upon the 

 substance as upon the way it was made to vibrate. Ac- 

 cording to this view the same chemical "atom" might 

 have a dozen spectra if struck in a dozen different 

 ways. 



But it was answered that this argument proved too 

 much ; and for this reason. Mitscherlich showed in 1864 

 that some bodies known to be chemical compounds 

 when raised to incandescence, give us a spectrum special 

 to the compound ; that is, they have a spectrum of their 

 own ; no lines of either of the constituents are seen. 



I showed later that when the temperature was sufficient 

 to produce decomposition the lines of the elementary 

 bodies, of which the compound was composed, made their 

 appearances according to the temperature employed. 

 And I also showed that precisely the same thing happens 

 with regard to the fluted and line spectra of the same 

 chemical element. We may get the first alone at a low- 

 temperature. We may mcrcase the temperature and 

 dim it slightly, some lines making their appearance ; 

 and next, by employing a very high temperature, we can 

 abolish the fluted spectrum altogether and obtain one 

 with lines only. 



Since then the difference between the two spectra of 



NO. 1536. VOL. 59] 



the same element was no more marked than the 

 difference between the spectrum of a known compound 

 and its constituents after the compound had been 

 broken up by heat ; it was as logical to deny the 

 existence of compound bodies as to deny that more 

 molecular complexities than one were involved in 

 spectral phenomena. 



Attacks like these finally caused the chemists to 

 reconsider their position, and some time later, being 

 under the impression, which has turned out to have no 

 justification, that "inonatomic" elements like mercury 

 have not fluted spectra, they conceded that the fluted 

 spectra might represent the vibration of the " di- 

 atomic '' molecule in the "diatomic" elements. This, 

 of course, was to give up the "bell-hypothesis." 



At the time when the differences of opinion arising 

 from the existence of fluted as well as line spectra in the 

 case of many elements were being discussed, solar 

 observalicns were beginning to bring before us a perfect 

 flood of facts apparently devoid of any law or order. In 

 1866 I tlircw an image of the sun on the slit of a spectro- 

 scope (Fig. I , in order to observe the spectra of its different 

 parts, and in this way the spectra of sun-spots and 

 eventually of prominences were observed. 



In the first method of work adopted in 

 the laboratory the spectroscope was directed 

 to the light source, so that the spectrum was 

 built up of the light, coming from all parts 

 of it without discrimination (Fig. 2). 



In 1S69 1 introduced into laboratory work 

 the method adopted in the case of the sun in 

 the observatory ; that is, an image of each 

 light source experimented on was thrown 

 on to the slit by a lens (Fig. 3}, so that the 

 spectrum of each part of it could be observed, 

 and some of the results obtained by the new- 

 method were the following : 



The spectral lines obtained by using 

 such a light source as the electric arc or 

 spark were of different lengths : some only 

 appeared in the spectrum of the centre of the 

 light source, others extended far into the 

 outer envelopes. This effect was best 

 studied by throwing the image of a hori- 

 zontal arc or spark on a \ ertical slit. The 

 lengths of the lines photographed in the 

 electric arc of many metallic elements 

 were tabulated and published in 1873 ^nd 

 1874 (Figs. 4 and 5). 

 Here then was the first glimpse of the idea that the 

 complete spectrum of a chemical element obtained at the 

 highest temperature might arise from the sunin:ation of 

 two or more different line spectra produced at different 

 degrees of temperature, and therefore bringing us in 

 presence of two or more molecular complexities ; that is, 

 different molecules broken up at different temperatures. 

 So soon as experiments. in the laboratory had given a 

 definite result with regard to the spectrum of a metal in 

 this way, I proceeded to study the sun with a view of 

 determining how that metal behaved in the sun. 



This involved, first, photographs of the solar spectrum 

 with its dark lines, photographic comparisons of these 

 dark lines with the bright lines constituting the spectra 

 of the metallic elements. This enabled us to compare 

 the total light given by each light source with the lighl 

 received from all parts of the sun indiscriminately. 



Next the spectra of different parts of the sun— chromo- 

 sphere and iirominences and spots— were compared with 

 different parts of the light source, the core of the arc, 

 and the centre of the spark, and the outer regions of both. 

 It will be seen that the inquiry now- had a very broad 

 base, and it could be immediately tested in many ways 

 at every stage. 



I candle flame) 



