5 9 6 SCIENCE PROGRESS 



magnetic, its order of magnitude being io T gauss. If we took 

 a crevasse in a ferro-magnetic crystal whose width of gap was 

 large compared with molecular dimensions, the force in the 

 gap, when the external field is zero, would be 4 tt I, where / 

 is the spontaneous intensity of magnetisation. This is small 

 compared with the local value of the molecular field and is 

 totally inadequate to account for the ultimate tensile strength. 

 The difference is attributed to the localisation of the molecular 

 field, and, while the principle of continuity of magnetic induc- 

 tion is obeyed, the intense local field is still capable of modifying 

 the structure of an immediately neighbouring molecule and 

 of setting up an internal stress equal to the ultimate tensile 

 strength of the material as observed mechanically. If we take 

 an isolated crystal, the stresses between the layers of atoms 

 will be different in different directions, and hence the effort 

 required to separate the layers of molecules or atoms will be 

 different according to the directions chosen. In this way, 

 the positions of the planes of cleavage are found. On this 

 view, the crystalline symmetry is characteristic of the residual 

 valency forces due to the nucleus and nuclear electrons, the 

 symmetry of which is reflected in the crystalline symmetry. 

 It is necessary to point out here that this statement is not 

 inconsistent with the atomic space lattice disclosed by X-ray 

 analysis. The diffracting centres are very small, diffuse regions 

 surrounding the nucleus, and containing numbers of nuclear 

 electrons. The true valency electrons are roaming over the 

 atomic " surface," and are in general small in number com- 

 pared with the nuclear electrons. As the diffracting power is 

 probably proportional to the number of electrons, we should 

 expect that the photographic effect, observed in X-ray analysis, 

 would be characteristic of the nuclear group and that the 

 two valency electrons which determine the special relation of 

 two different atoms would not be discoverable by such means. 

 Therefore it is regarded as probable that the molecule, at least 

 in compounds, still exists as a definite unit within the crys- 

 talline structure (fig. 3). Furthermore, in the liquid state, the 

 atoms of a compound molecule bear a special relation to one 

 another, determined by definite valencies, and it is not likely 

 that, when crystallisation sets in, the residual forces of two such 

 stable systems will be capable of controlling the specially 

 strong valency forces in such a way that a given atom becomes 

 associated equally with all its neighbours. Unless we adopt 

 this view, and regard certain atoms as specially bound together 

 to form a molecule, just as they do in the liquid state, we must 

 admit that a single valency can be subdivided, for, on the 

 alternative view, each atom is related to its neighbour in pre- 

 cisely the same way. 



