ARTICLES 59i 



Now during crystallisation the rigidity which arises is due 

 to the forces which the molecules, in virtue of the contained 

 rotating electrons, exert on one another. These forces must 

 be large in order to produce such a large cohesive effect. In 

 spite of the zero magnetic moment of the diamagnetic mole- 

 cule, taken as a whole, the local force, at a point in between a 

 pair of molecules, may be very large, comparable in fact (see 

 Part 2, p. 143) with the local force within a ferro-magnetic 

 medium, and may be capable of distorting the molecules to 

 such an extent as to give rise to the observed change of specific 

 susceptibility on crystallisation. 



The Local Molecular Field of a Diamagnetic Substance. — 

 From the extent of this change of susceptibility per gram on 

 crystallisation, it is possible to derive a value of the intensity 

 of the local force, which is operative in the formation of a 

 crystalline space lattice, and to which the rigidity is due. The 

 usual method of defining the force at an internal point of a 

 material medium is to take a cavity, whose dimensions are 

 small in comparison with ordinary lengths (e.g. 1 cm.), and yet 

 large compared with molecular dimensions. A convenient 

 designation of the size of such a cavity is contained in the 

 phrase " physically small." In molecular theory, the sub- 

 division of the medium into elements is not valid beyond the 

 limits of physical smallness, and only in media which are 

 absolutely continuous may the elements be pushed to the 

 limits of mathematical smallness. The case of a fluid com- 

 posed of discrete molecules has been worked out by Sir Joseph 

 Larmor, who found that the part contributed to the force at 

 an internal point, by the molecules immediately surrounding 

 the point, was separable into two parts, a purely local part 

 and a part due to the rest of the medium. The latter part 

 is determinate, as it is derivable from the potential due to the 

 combined volume and surface density distributions of Poisson, 

 while the former part, on account of rapid motions and irregular 

 distributions of the molecular axes, is shown to be negligibly 

 small. When, however, we come to the case of crystalline 

 media, the molecules, which lie in the immediate neighbour- 

 hood of the internal point, contribute an unknown amount to 

 the local force, for in this case there is no averaging out on 

 account of random motions of the molecules immediately sur- 

 rounding the point. As we do not know the law of force which 

 holds at such close range, or the relative positions of the atoms 

 in the crystalline medium, we can only deduce the value of the 

 local forces indirectly from experimental data. 



Up to the present we have made no supposition as to the 

 nature of the local molecular force. It may be electrostatic, 

 electromagnetic, or indeed of any other nature, for a force of 



