6 Mr. W. Sutherland on the 



fields in the atoms of different non-metals are of nearly the 

 same strength, so that the atomic vibrators have not only 

 the same periods but nearly the same lengths r. From (1) 

 also we get that the intensity of electrization throughout the 

 sphere of radius R due to er is equal and opposite to that of 

 the atom due to es. 



With the atoms of metals in their compounds the conditions 

 are different. It is proved in "The El. Or. of Mol. Attr." 

 (loc. cit.) that in the Li family (H 2 0*/10 6 = e*/A10 8 = l-2 

 (2, 3, 4, 5, 6), where h is the mass of the hydrogen atom. 

 It is mentioned there too that the volume B of a gramme- 

 atom of the combined metal is given by 2 + 2*7 (n — l)n 

 where n has successive integral values from 1. But a still 

 simpler relation holds, namely, that B 1 / 3 = , 65 (2, 3, 4, 5, 6), 

 as the following comparison shows : — 



Table I. 



Li. Na. K. Eb. Cs. 



B 2 7-4 18-6 34-4 (56) 



Bi/3 1-26 T95 2-60 3'25 382 



BV3 C alc 1-30 1*95 2-60 325 390 



In this family es is proportional to a, and in the next 

 section a similar result is proved for the Be family. In the 

 atoms of metals of the same family es/a? is not constant as in 

 the non-metals, thus in the Li family it falls for Cs to (2/6) 2 

 or 1/9 of the value for Li. Yet by virtue of (1) r is reduced 

 in the same proportion, the larger atom has the shorter 

 vibrator and the frequency of all the vibrators remains the 

 same. We shall see later that in the characteristically non- 

 metal families where a 3 is proportional to es there is a strong 

 tendency for a 3 in a family to form a series 1, 2, 3, 4, just as 

 we have seen in the characteristically metal families where a 

 is proportional to es there is a tendency for a to form such 

 a series as 2, 3, 4, 5, 6. Electric moment is a controlling- 

 factor in the architecture of the atoms. These conceptions 

 of the structure of the atom and of the chief vibrating 

 mechanism in the atom ought to furnish a definite quanti- 

 tative theory of the nature of dielectric capacity, which will 

 now be investigated. 



One other matter requires consideration before we leave 

 this section, namely, the magnitude of r. In " The Ions of 

 Gases M (Phil. Mag. Sept. 1909) it was stated that this is 

 of the order 0*05 X 10~ 8 , but when worked out with the 

 molecular data used in this paper it is smaller. We have 

 er/W = es/a? = (M. 2 l)>/B. For non-metallic elements in their 



