218 On the Origin of Spectra. 



that the carriers o£ positive electricity in a vacuum-tube are 

 often more complex than the original gas or vapour through 

 which the discharge is passing. Thus in the case of oxygen 

 gas, carriers o£ positive electricity with " electric atomic 

 weights " 8, 16, 32, 48, and 96 have been discovered. The 

 first of these corresponds to the atom of oxygen with two 

 positive charges, ++ ; the second to the atom with one 

 one positive charge, + ; the third to the oxygen molecule 

 (0 2 ) + ; the fourth to the molecule of ozone (0 3 ) + ; and the fifth 

 to (0 6 ) + . The vibrations of each of these positive ions may 

 give rise to luminous radiation, or this may come from neutral 

 systems formed by collisions with electrons. The vibrations 

 of these neutral systems would, on Wien's theory, give rise 

 to spectrum lines. On the other hand, we should expect that 

 if the vibrations of the carriers of greater " electric atomic 

 weight" give rise to luminous radiation, its spectrum would 

 consist of bands, and it is well known that bands, especially 

 in the green, are the characteristic feature of the spectrum of 

 oxygen at low pressures. At higher pressures the bands are 

 not so marked, which is the result to be expected, for the more 

 frequent collisions would tend to prevent the formation of 

 complicated molecules. 



In the case of mercury vapour Professor Thomson has 

 detected positive carriers with electric atomic weights of 800, 

 200, 100, and 66, corresponding to (Hg 4 ) + , Hg + , Hg ++ , and 

 Hg +++ , respectively. As in the case of oxygen, it seems 

 probable that it is the vibrations of large atomic aggregates 

 such as (Hg 4 ) which are the origin of the band spectrum 

 observed by Eder and Valenta *, Huff f, and others. If Wien's 

 view is correct, it is the carriers of smaller mass, whick, by 

 their collisions with free electrons, form the systems pro- 

 ducing the various spectrum lines, and the different line 

 spectra which the author has observed are probably due to 

 the different neutral systems formed. More energy would be 

 required for the formation of a positive ion with three 

 charges than for the formation of an ion with two charges 

 or with unit charge, and so the neutral systems containing 

 positive ions with the smallest electric atomic weights would 

 be the last to be formed as the energy of the discharge is 

 increased. Since these systems contain the largest number 

 of oscillating electrons, they are capable of producing the 

 greatest number of spectrum lines; so that we should expect 

 the spectrum with most lines to be formed when the discharge 



* Eder and Valenta, Denkschrift. Wien Akad. lxi. p. 401 (1894). 

 t Huff, Astrophys. Journ. xii. p. 103 (1900). 



