CONTEMPORARY ADVANCES IN PHYSICS 245 



that of all their magnetic moments, are zero. Now suppose that a 

 field is suddenly applied, parallel to the axis of the cylinder. The 

 substance is suddenly magnetized; this signifies that the resultant 

 of the magnetic moments, and hence that of the angular momenta, 

 are no longer zero. Let / stand (as heretofore) for the former re- 

 sultant, and P for the latter. Noav it is desirable to remember that 

 each atom consists of a nucleus and an electron-family; that the 

 electron-family possesses the magnetic moment and is oriented in the 

 field (it strictly is what I have hitherto referred to as "the atom"); 

 that the nuclei of the atoms in the cylinder are relatively non-magnetic 

 but contain nearly all of the mass of the cylinder.^ At the moment of 

 magnetization, the ensemble of the electron-families acquires a net 

 angular momentum P. Now angular momentum being one of these 

 things (like energy and linear momentum) of which the total in Nature 

 does not vary, an equal and opposite amount, — P, must appear 

 somewhere or other. It appears in the mass of the cylinder, presum- 

 ably because of some interaction between the electron-families and their 

 nuclei. The cylinder makes a sharp turn at the instant of magneti- 

 zation, twisting the suspension from which it hangs through an angle 

 from which (and from the rigidity of the suspension) the value of — P 

 can be found. This effect and its converse (an unmagnetized cylinder 

 may be magnetized by sharply twisting it) are known as the "gyro- 

 magnetic effects." They are delicate and difficult to produce, a 

 fortiori to measure; yet of late years experimenters have succeeded in 

 measuring P together with /, and therefore learning the value of the 

 ratio I/P — first for the ferromagnetic metals and then for some of 

 their compounds and alloys, and lately for certain paramagnetic salts, 

 the work on these last being done at the very low temperatures where 

 alone they can be strongly magnetized. 



This ratio I/P — its reciprocal is called the "gyromagnetic ratio"— 

 is a rare sort of thing: it is a quantity of which the numerical value, 

 measured on pieces of bulk matter, is appropriate also to the elemen- 

 tary particles. If the substance is made up of identical elementary 

 magnets of magnetic moment /jl and angular momentum p, then I/P 

 is n/p. Since /z and p are knowable frora spectra, so also is their ratio. 

 Its lowest possible value (from theory) is e/lmc, in which e, m, and c 

 have their usual meanings;^ this would always occur if the electrons 

 had no spins; actually it occurs if the electron-family of the atom is so 



^ Most nuclei possess magnetic moments, which, however, are so excessively small 

 that they can be detected only by experiments of extreme delicacy. 



^ Charge (in E.S.U.) and mass of the electron, and speed of light in vacuo. For the 

 theory underlying these statements, c.f. I.e. pp. 285-300. Often the ratio of the 

 experimental value of n/p to the quantity e/2mc is called an "experimental g-value," 

 the ratio of the theoretical value to e/2mc being conventionally denoted by g. 



h 



