April 22, 1922] 



NA TURE 



519 



T T has been known for many years that the radia- 

 tions which an element emits in the state of a 



^^Pluminous gas are not invariable but depend on the 

 presence of other elements^ the manner in which the 

 substance is excited to luminosity, and other circum- 

 stances. It was recognised in some of the earliest 

 investigations that many band spectra were to be 

 associated with compounds and that a spectrum might 

 be due partly to such compounds and partly to un- 

 combined atoms. Thus, for example, if strontium 

 chloride is introduced into the flame of the bunsen 

 burner we find lines associated with the element, 

 bands due to strontium oxide, and also bands due to 

 the chloride, and when strontium bromide is substi- 

 tuted for the chloride the spectrum is the same as 

 regards the lines due to the element and the oxide 

 bands, but bands peculiar to the bromide are found to 

 have replaced those due to the chloride. 



Minute quantities of substances can sometimes be 

 detected by means of these characteristic bands due 

 to compounds, a familiar example being the blue flame 

 which is seen when common salt is thrown onto a 

 coal fire and is due to the copper chloride formed 

 from the chlorine in the common salt, and the minute 

 trace of copper which is present in the coal. A number 

 of different elements are present in most flames, and 

 the reactions which occur are probably very complex. 

 In gases contained in vacuum tubes which are excited 

 to luminosity by electrical discharges it is possible to 

 work with pure substances, and a discussion of the 

 spectra observed is simpler. 



In the case of gases in vacuum tubes the spectrum 

 sometimes consists of bands, and the band spectrum 

 from the negative pole may be different from that seen 

 in the positive column. Thus nitrogen, when excited 

 by uncondensed discharges, shows in the visible regions 

 two band spectra, one known as the positive band 

 spectrum, which appears in the capillary of a vacuum 

 tube of the conventional type, and the negative band 

 spectrum, found in the neighbourhood of the cathode, 

 which constitutes an important part of the spectrum 

 of the aurora. 



Both these band spectra, and indeed all band spectra, 

 are generally attributed to molecules rather than atoms, 

 but if a condensed discharge is passed through nitrogen 

 the spark spectrum associated with the nitrogen atom 

 is obtained, and this is capable of further modification 

 when discharges of great intensity are employed. The 

 action of the condensed discharge is almost certainly 

 due to the greatly increased current density which 

 obtains during the very brief periods while the discharge 

 is passing. Its first effect is to break up the mole- 

 cules into atoms, and the further stages brought about 

 by an increase in the intensity of the discharge are 

 generally supposed to be due to the removal of suc- 

 cessive electrons from the atoms. There are other 

 methods by which the current density can be increased 

 with similar changes in the spectrum ; the effect 

 of an increase in the current density is to increase the 

 number of charged particles in a given volume of the 



• From a discourse delivered at the Royal Institution on Friday, March lo. 

 NO. 2738, VOL. 109] 



Problems in the Variability of Spectra.^ 



By Prof. Thomas R. Merton, F.R.S. 



gas, with the result that a large number of the radiating 

 atoms are subjected to intense electric fields due to 

 neighbouring charged particles. 



Similar results are observed in the spectra associated 

 with carbon. There are at least six spectra due to 

 compounds of carbon with hydrogen, oxygen and 

 nitrogen, and special experimental conditions are neces- 

 sary for the production of some of these spectra. In 

 addition to these band spectra carbon shows line 

 spectra, and with the most intense discharges which 

 can be employed in the laboratory a number of new 

 lines appear which are also found in the spectra of the 

 hottest type of stars, known as the Class 0, or Wolf- 

 Ray et stars. 



All these changes can be reasonably accounted for, 

 but there are a number of other changes which are 

 more difficult to explain. For many reasons the spec- 

 trum of hydrogen is of particular interest, because the 

 atom of hydrogen is the simplest known atom and is 

 supposed to consist of a positive nucleus and a single 

 electron. There are two spectra associated with hydro- 

 gen, one of which, the Balmer series, is found in almost 

 all celestial spectra and also in vacuum tubes in the 

 laboratory unless the most rigorous precautions are 

 taken to exclude all traces of hydrogen. The explana- 

 tion of the origin of this spectrum has been one of the 

 most striking successes of the quantum theory of spectra 

 developed by Bohr and by Sommerfeld. The other 

 spectrum of hydrogen, known as the secondary spec- 

 trum, consists of an enormous number of lines and 

 differs in its mode of production from the Balmer 

 series in that the secondary spectrum is characteristic 

 of pure hydrogen. In the purest hydrogen obtainable 

 the secondary spectrum may be as bright as the 

 Balmer series, but if the smallest trace of impurity is 

 present the Balmer series gains in intensity and the 

 secondary spectrum becomes very much weaker. In 

 a vacuum tube containing water vapour the lines of 

 the Balmer series are extremely intense whilst those 

 of the secondary spectrum are relatively very faint. 

 The investigations of Michelson and Lord Rayleigh, 

 and of Buisson and Fabry have shown that under 

 certain conditions the masses of the atoms or mole- 

 cules from which the spectrum originates may be de- 

 duced from a knowledge of the widths of the spectrum 

 lines, and recent investigations, in which the widths of 

 the lines of the secondary spectrum of hydrogen have 

 been measured to a high degree of precision, have 

 shown that the secondary spectrum is to be referred 

 to the hydrogen molecule. 



The presence of impurities in vacuum tubes con- 

 taining hydrogen not only enhances the lines of the 

 Balmer series but also brings about changes in the 

 relative intensities of the Balmer lines themselves. 

 Some of these changes are very striking, but there are 

 other variations of a more "subtle kind which are only 

 discovered when accurate quantitative measurements 

 are made of the relative intensities of the lines. A most 

 striking effect is observed when a relatively large 

 quantity of helium is admitted to a vacuum tube 

 containing hydrogen. Under these conditions the 



