216 Dr. F. Horton on the 



canal-rays formed in various vapours and gases, has come to 

 the conclusion that the carriers responsible for the line spectra 

 are neutral atoms, and that the vibrations producing the 

 luminosity are caused by the recombination of negative 

 electrons with the positive "remainder atoms." This is the 

 view which is, I think, supported by the result of my recent 

 observations of the spectra of mercury under different elec- 

 trical conditions. In the paper already referred to five 

 distinct line spectra are recorded. These, I think, are due to 

 the recombinations of remainder atoms of different "electric 

 atomic weights" with the appropriate number of negative 

 electrons to form neutral systems. These systems are pro- 

 bably not similar in all respects to the atoms of the unionized 

 gas, but are less stable arrangements of positive ions with 

 one or more negative electrons revolving round them — each 

 system being electrically neutral. The periods of these 

 systems would depend upon the arrangement of the electrons 

 around the central positive ion. There would be different 

 equilibrium positions and consequently several possible periods 

 for any one system. These would be affected by a magnetic 

 field in such a way as to account for the Zeeman effect. 



The fact that compounds always give banded spectra would 

 seem to indicate that these are due, not to electrons, but to 

 the vibrations of larger bodies. Since they produce light of 

 the same wave-length, these larger bodies must be under the 

 influence of more intense forces, which suggests that they are 

 not separate atoms, but systems of atomic dimensions vibrating 

 inside the molecule itself. On these views of the origin of 

 line and of band spectra we should expect that a gas or vapour 

 consisting solely of molecules in a normal condition -would 

 give a band spectrum, a conclusion which is borne out bv 

 the fact that the absorption spectra of such gases always 

 consist of bands. In the case of the vapours of the alkali 

 metals, which give line spectra, we have molecules which are 

 not in a "normal" condition, as is shown by the abnormal 

 conductivity of these vapours. 



To obtain the emission spectrum of a gas we have to render 

 it luminous by an electric discharge, and the effect of this is to 

 ionize the molecules and to alter the composition of the gas. 

 Electrons are set free ; positive ions are produced, and by the 

 collisions of these, vibrating systems are formed which radiate 

 the energy of the spectrum lines. This difference in the 

 composition of the gas is clearly shown by the experiments 

 of Ladenburg * and of Pfluger f on the absorption of light 



* Ladenburg 1 , Deutach. Phys. GeselL Verh. x. 14, p. 550 (1908). 

 f Pfluger, Ann. d. Ph$& xxiv. 3, p. 5] 5 (1907). 



