We see from these illustrations how the corpuscles would arrange 

 themselves in the atom, confining ourselves for the present to the 

 case where the corpuscles are constrained to move in one plane. The 

 corpuscles will arrange themselves in a series of rings, the number of 

 corpuscles in the rings getting greater and greater as the radius of 

 the ring gets greater. By the aid of the sbove table we can readily 

 calculate the way any number of corpuscles will arrange themselves. 

 Let us suppose for example we have 20 corpuscles and try to arrange 

 them so as to have as few rings as possible ; we see from the table 

 that we cannot have more than 12 in the outside ring, for 13 w^ould 

 require 10 inside, and would be impossible with less than 23 cor- 

 puscles : thus 12 will be the number in the outside ring and there 

 are eight left to dispose of ; these cannot form a single ring with no 

 corpuscles inside, as 5 is the greatest number that can do this ; the 8 

 corpuscles will therefore break up into two systems, a ring of 7 with 

 1 inside. You see that when I try the experiment with 20 magnets 

 they arrange themselves in this way. 



If we follow the kind of atoms produced as we gradually increase 

 the number of corpuscles, we find that certain arrangements will recur 

 again and again ; thus take the case of 20 corpuscles ; this consists of 

 tile arrangement 1-7-12, the arrangement for 8 is 1-7 ; the atom of 

 20 corpuscles may be regarded as formed by putting another storey 

 to the atom of 8 corpuscles ; if we go to 37 corpuscles, we find the 

 arrangement is 1-7-12-17, i.e. another storey added to the atom of 

 20, while for 56 we have 1-7-12-17-19, the atom of 37 with another 

 storey added. Thus the possible atoms formed by numbers of cor- 

 puscles from one to infinity could be arranged in classes, in which each 

 member of the class is formed by adding another storey to the preceding 

 member ; the structures of all the atoms in this class have much in 

 common, and we might therefore expect the physical as well as the 

 chemical properties of the atoms to have a general resemblance to 

 each other. This property is, I think, analogous to that indicated by 

 the periodic law in chemistry. We know that if we arrange the 

 elements in the order of their atomic weights, then, as we proceed in 

 the direction of increasing atomic weight, we come across an element, 

 say lithium, with a certain property ; we go on and after passing many 

 elements which do not resemble litliium, we come across another, 

 sodium, having many qualities in common with litliium. Then as 

 we go on, we lose these properties and come across them again when 

 we arrive at potassium ; exactly the kind of recurrence we should get 

 with our model atoms, if we suppose the number of corpuscles in the 

 atom to be determined by its atomic weight. 



TiCt me give another instance of the way the properties of these 



