Dynamical Theories of Gravitation. 155 



show such a pressure-effect produced by atomic motion, where 

 elastic rings are converted into the hexagonal forms of crystals. 

 Cohesion (as now is generally supposed) is only gravitation * 

 by contact. From the above considerations, it becomes easy 

 to understand that elastic crystal forms can sometimes change 

 into wow-crystalline forms ; so some crystallizable metals, such 

 as iron and zinc, lose all crystalline structure by rolling and 

 hammering, and become ductile. Crystalline sulphur can, by 

 mere warming, pass over into a sort of indiarubber sulphur. 

 It is evident in fact that the irregular arrangement of the 

 elastic molecules of a substance favours the gliding of the 

 molecules over each other ; while, on the other hand, the 

 regularly arranged molecules which are in contact at their 

 boundaries (corresponding to crystalline structure) cannot be 

 displaced at all without separating entirely : take for instance 

 crystallized cast-iron and some other metals. 



The freedom allowed to the molecules to arrange themselves 

 in the case of solutions may be favourable for the produc- 

 tion of crystalline structure, while rolling, hammering, &c, 

 manifestly forces the molecules to aggregate in an irregular 

 manner. 



When elastic molecules of very open structure cross each 

 other irregularly in all directions, or are arranged in parallel 

 layers (as produced by the rolling of a metal), then it be- 

 comes obvious that a subsequent displacement of the molecules, 

 as by a tensile-stress for example, does not necessarily produce 

 actual severance ; but the atomic streams — from the very 

 nature of this cause — can easily produce contact in fresh 

 places, and so a gliding of molecules over each other is 

 possible, without separation. So a bar of malleable iron 

 gradually lengthens itself under a tensile-stress. On the 

 other hand, because crystalline structure prevents all gliding 

 of the molecules, it becomes impossible in this case for the 

 atomic streams to cause contact in fresh places. The attempt 

 to stretch a cast-iron bar, then, means separation of the mole- 

 cules. I will not pursue these considerations more at length 

 here : they may well be thought out into greater detail. 



It may just be remarked, in passing, how elastic rings, 

 fig. 2, can at first repel each other, merely on account of their 

 elasticity of form ; and how in iig. 3, if the molecules are 

 made to approach closer by force, this has as a consequence 



* A fact observed "by Prof. Dewar may be favourable to this view. 

 The cohesion of metallic wires does not diminish (but rather the con- 

 trary) by a cold of —180° C. that of liquid oxygen. Now cooling of 

 the metal could manifestly have no influence on the atomic streams, 

 which are independent, 



