TRANSACTIONS OF SECTION B. 665 



different elements is far from being placed on a satisfactory basis, and in spite of 

 the researches of Lecoq de Boisbaudran, Ditte, Troost and Hautefeuille, Ciamician 

 and others, it cannot be said that as yet definite proof has been given in support 

 of the theory that a causal connection is to be found between the emission 

 spectra of the several elements belonging to allied groups and their atomic weights 

 or other chemical or physical properties. In certain of the single elements, how- 

 ever, the connection between the spectra and the molecular constitution can be 

 traced. In the case of sulphur, for example, three distinct spectra are known. The 

 first of these, a continuous one, is exhibited at temperatures below 500°, when, 

 as we know from Dumas' experiments, the density of the vapour is three times 

 the normal, showing that at this temperature the molecule consists of six atoms. 

 The second spectrum is seen when the temperature is raised to above 1,000°, when, 

 as Deville and Troost have shown, the vapour reaches its normal density, and 

 the molecule of sulphur, as with most other gases, contains two atoms, and this 

 is a band spectrum, or one characterised by channelled spaces. Together with this 

 band spectrum, and especially round the negative pole, a spectrum of bright lines 

 is observed. This latter is doubtless due to the vibrations of the single atoms of the 

 dissociated molecule, the existence of traces of a band spectrum demonstrating the 

 fact that in some parts of the discharge the tension of dissociation is insufficient to 

 prevent the reunion of the atoms to form the molecule. 



To this instance of the light thrown on molecular relations by changes in the 

 spectra, others may be added. Thus the low-temperature spectrum of channelled 

 spaces, mapped by Schuster and myself, in the case of potassium, corresponds to 

 the molecule of two atoms and to the vapour-density of seventy-nine, as observed 

 by Dewar and Dittmar. Again, both oxygen and nitrogen exhibit two, if not 

 three, distinct spectra : of these the line spectrum seen at the highest temperatures 

 corresponds to the atom ; the band spectrum seen at intermediate temperatures 

 represents the molecule of two atoms ; whilst that observed at a still lower point 

 would, as in the case of sulphur, indicate the existence of a more complicated 

 molecule, known to us in one instance as ozone. 



That this explanation of the cause of these different spectra of an element is 

 the true one, can be verified in a remarkable way. Contrary to the general rule 

 amongst those elements which can readily be volatdised, and with which, therefore, 

 low-temperature spectra can be studied, mercury exhibits but one spectrum, and 

 that one of bright lines, or, according to the preceding theory, a spectrum of atoms. 

 So that, judging from spectroscopic evidence, we infer that the atoms of mercury 

 do not unite to form a molecule, and we should predict that the vapour-density of 

 mercury is only half its atomic weight. Such we know, from chemical evidence, is 

 really the case, the molecule of mercury being identical in weight with its atom. 



The cases of cadmium and iodine require further elucidation. The molecule of 

 gaseous cadmium, like that of mercury, consists of one atom ; probably, therefore, 

 the cadmium spectrum is also distinguished by one set of lines. Again, the 

 molecule of iodine at 1,200° separates, as we know from Victor Meyer's researches, 

 into single atoms. Here spectrum analysis may come again to our aid ; but as 

 Schuster remarks, in his report on the spectra of the non-metallic elements, a more 

 extensive series of experiments than those already made by Ciamician is required 

 before any definite opinion as to the connection of the different iodine spectra with 

 the molecular condition of the gas can be expressed. 



It is not to be wondered at that these relations are only exhibited in the case 

 of a few elements. For most of the metals the vapour-density remains, and pro- 

 bably will remain, an unknown quantity, and therefore the connection between any 

 observed changes in the spectra and the molecular weights must also remain un- 

 known. The remarkable changes which the emission spectrum of a single element — 

 iron, for instance — exhibits have been the subject of much discussion, experimental 

 and otherwise. Of these, the phenomenon of long and short lines is one of the 

 most striking, and the explanation that the long lines are those of low temperature 

 appears to meet the facts satisfactorily, although the effect of dilution, that is, a 

 reduction of the quantity of material undergoing volatilisation, is, remarkably 

 enough, the same as that of diminution of temperature. Thus it is possible, by the 



