SOME CONSEQUENCES OF GRAHAM'S WORK 591 



to adopt is a compromise and a combination of two forms of 

 classification — the one involving the arrangement of elements 

 in the order of their atomic weights, the other their arrangement 

 in families. The latter mode is of special importance because, 

 the record being incomplete, we need some means of appor- 

 tioning blanks— this is afforded by the existence of families of 

 elements, a fact of entirely crucial importance. 



As to the method followed, suppose a surface to be formed 

 by wrapping a band spirally around a cylinder and that the 

 cylindrical surface is divided into sections by lines drawn 

 vertically parallel with the axis of the cylinder ; if the surface 

 be cut along one of the vertical lines and spread out, a sheet 

 or table is obtained. In such a table a series of compartments 

 in succession is provided ; each of these will serve for an 

 element. 



Writing the symbols of the elements in succession in the 

 order of the atomic weights— 



— a series of diverse elements is obtained but a point is soon 

 reached at which an element occurs belonging to a family 

 already represented in the list and it is necessary to begin 

 a new period : 



Soon the process must be repeated but this time the period 

 must be extended several places. 



Thus far and in the next line elements in each vertical series 

 belong to the same family. 



It is impossible to continue the arrangement on this prin- 

 ciple. In the fifth line, a whole series of elements is encountered 

 (the rare earth elements) all belonging to the same family : 

 if written across the table, these do not find a place in the 

 families occupying the space above. To get over this difficulty, 

 I have suggested a form of classification more nearly re- 

 sembling that followed in the case of the compounds of carbon 

 and involving the recognition of series of homologous elements. 



