Mechanism of Radiation. 447 



We have, however, seen that each " line " of the original 

 line-spectrum ought to he regarded as the superposition of 

 several lines. The rotation of the atom destroys the spherical 

 symmetry, and therefore separates these lines, each one being, 

 at the same time, spread out into a band. 



The resulting " band " will therefore be a more complex 

 structure than we have hitherto supposed it to be. It may, 

 for instance, spread to both sides of the original line, and 

 may show several maxima and minima of intensity. But in 

 every case the widening effect ought to remain always 

 similar to itself, the amount being always proportional to w 2 , 

 the mean value of w 2 . 



Now if the rotation is governed by a subsidiary tempera- 

 ture, w 2 will be proportional to the temperature, and there- 

 fore to the density (§ 6), so that the shift and the broadening 

 of any line will each be directly proportional to the pressure 

 in a pure gas, and roughly proportional to the total pressure 

 in a mixed gas*. 



In any single spectrum-series, not only the periodicity but 

 also the form of the normal function depends solely upon the 

 single parameter n. From this it follows that the amount 

 of shift in a single spectrum-series will depend only on n. 

 There will therefore be a similarity between the shift of the 

 different lines of the same series, such as does not exist 

 between those of different series. 



The broadening effect just found must be added to that of 

 § 8. The broadening due to each effect, and therefore also 

 the resultant broadening, will be proportional to the pressure. 



§ 35. In the case of complete molecules the rotation will 

 not be governed by a subsidiary temperature, so that the effect 

 of rotation will be much greater than in the case of dissociated 

 atoms. There need therefore be no difficulty in accounting 

 for the continuous spectrum emitted by a solid or a heated gas. 



The Structure of Molecules. 



§ 36. Let us now pass from the consideration of the phe- 

 nomena of spectroscopy to the consideration of the other 

 properties of matter which are implied in our provisional view 

 of the structure of an atom. 



If an atom consists of n negative and n+s positive ions, 

 the total charge on the atom will be se, and s may, by a slight 

 extension of tne usual terminology of chemistry, be referred 

 to as the valency of the atom. 



* A Summary of Experimental Knowledge is given by W. J. 

 Humphreys, Astrophys. Journal, vi. p. 225. 



