ARCHITECTURE OF MOLECULES 223 



attractive force of an atom, but it is equally convenient to sup- 

 pose the whole space filled as it would be if the spheres were 

 elastic and compressed till there were no interstices between them. 



It is interesting to note that some 85 per cent of the crystalline 

 elements seem to be in general agreement with these assumptions, 

 50 per cent being cubic and 35 per cent hexagonal, while the remain- 

 ing 15 per cent which present a divergence still await explanation. 



If now a case is considered in which the atoms are not alike but, 

 as in common salt, are of two kinds, it is evident that throughout 

 the crystal structure they must be ranged in regular order alter- 

 nately, so that each sodium atom is related directly to six chlorine 

 atoms, and each chlorine atom is similarly situated with respect 

 to six sodium atoms. The whole mass might be partitioned into 

 equal and similar spaces, each containing an atom of sodium 

 and an atom of chlorine or the conventional molecule of common 

 salt. Such division would not necessarily correspond to any 

 physical division into molecules, for the selection of partners by 

 the sodium and chlorine atoms respectively does not necessarily 

 occur till the compound is melted or dissolved, when the sodium 

 may become associated with any one of the six chlorine atoms 

 adjacent to it, and the chlorine similarly may pair with any one 

 of the sodium atoms. Each component atom then has a separate 

 existence in space, but a further question arises as to the relative 

 spaces occupied by atoms of different chemical characters and 

 their relations to the crystalline forms assumed by the solid 

 compounds in which they exist. This has been answered by the 

 entirely new conception of valency introduced by Mr. W. Barlow 

 and Professor W. J. Pope in 1906. 



In the development of the ideas briefly described in the fore- 

 going paragraphs they found that in a crystalline substance each 

 component atom appropriates a portion of space which is 

 approximately proportional to its fundamental valency. Thus the 

 volume of the sphere of influence of carbon is nearly four times 

 that of the hydrogen sphere. This is illustrated by the models 

 shown in figures 63, 64, and 65, which represent the molecules 

 of methane and benzene respectively. 



The volumes of other univalent elements are not exactly, 

 though very nearly, the same as that of hydrogen. 



The fundamental valency of an element must also be regarded 

 as of a different nature from other valencies exhibited by the 

 same element. This has already been indicated by the researches 



