696 SCIENCE PROGRESS 



more stable cubic form. The present writers have ascertained 

 that in the case of several dimorphous organic compounds the 

 two forms of these are also very closely related (p. 710). 



It is also apparent from a study of the two forms of silver 

 iodide, that the homogeneous partitioning of the hexagonal and 

 cubic assemblages of spheres simulating these two forms can 

 be so carried out as to give exactly the same aggregate or unit 

 in both cases. The dimorphism of the compound is therefore 

 to be attributed to the existence of alternative methods of close- 

 packing units of this form. Polymorphism in general may be 

 defined as the existence of alternative methods of homogene- 

 ously close-packing the same set of spheres of atomic influence 

 under the condition that the homogeneous partitioning of the 

 assemblages shall yield the same form of unit or molecule 

 identical in composition, constitution and configuration in all 

 the cases thus related. This definition gives us for the first 

 time a clear idea of the nature of polymorphism and is of 

 considerable importance. 



Another exceptionally interesting point arising out of this 

 investigation has reference to the abnormal coefficient of ex- 

 pansion of silver iodide. It is well known that hexagonal silver 

 iodide has a negative coefficient of cubic expansion, a large 

 negative coefficient of linear expansion in the direction of the 

 principal axis and a small and positive coefficient in directions 

 at right angles to this axis. Supposing that in the hexa- 

 gonal silver iodide assemblage the silver and iodine spheres 

 of influence which are slightly unequal have different relative 

 thermal coefficients of expansion, such that as the temperature 

 rises the volumes of the spheres of the two kinds become more 

 nearly equal, heating will cause a closer approximation to the 

 ideal closest-packed assemblage and will actually produce a 

 contraction of the mass as a whole corresponding to the 

 observed negative coefficient of cubic expansion of the crystals. 



There seems, in fact, to be no end to the fruitfulness of the 

 new theory nor to the diversity of problems which it helps to 

 elucidate. The study of the geometrical properties of assem- 

 blages representing the halogen compounds of the alkali metals 

 potassium, rubidium and caesium has led to many interesting 

 results. For instance, assemblages have been derived to repre- 

 sent the crystalline forms of these substances which show 

 peculiarities corresponding to the observed properties of the 



