246 BELL SYSTEM TECHNICAL JOURNAL 



organized that the spins neutralize one another. Its highest possible 

 value is twice as great; this occurs if nothing counts excepting the 

 electron-spins, and signifies either that the electrons are free ^ or else 

 that the electron-family of each atom is so organized that there is no 

 net angular momentum due to orbital motion. Intermediate values 

 are possible and signify different types of organization of the electron- 

 family. The values predicted from spectra have been confirmed for 

 a few of the rare-earth atoms in their paramagnetic salts; but usually, 

 as I have already intimated, the observed value of the ratio n/p is about 

 2(e/2mc), though the spectrum says something else. 



It would be pleasant now to add that the magnetic moment of each 

 of these substances, per atom, amounts to some integer multiple of 

 the magnetic moment jue of the spinning electron. We then could 

 say that the integer is the number of "uncompensated" spinning 

 electrons in the atom, implying by the word "uncompensated" in 

 this connection that all the magnetic moments in the electron-famih' 

 of the atom add up vectorially to zero and so do all the angular mo- 

 menta, with the sole exception of those pertaining to these electron- 

 spins. Such is not, however, the case : some of the experimental values 

 are 2.2/Xefor iron, l.ljjLe for cobalt, 0.6/ac for nickel. It seems necessary to 

 assume that in metallic solid iron, some of the atoms present two un- 

 compensated electrons to the orienting field, and others three. Iron in 

 different chemical compounds exhibits different values of magnetic 

 moment, and sometimes the ratio nlp is different from 2(e 2wf), sug- 

 gesting that angular momenta of orbital motion are not quite cancelled 

 out; indeed it now appears that the ratio is slightly but definitely differ- 

 ent from this specific value even in the cases (such as those of the pure 

 ferromagnetic metals and of permalloy) in which at first the measure- 

 ments suggested that it was the same. 



Such observations as these last are problems for the specialists in 

 atomic theory; magnetism offers great numbers of these problems. 

 Another and a complementary way of viewing this situation is, to look 

 on every measurement of a magnetic moment made upon a solid as 

 an item of information about an atom (or a molecule) existing in a 

 condition which is not accessible to spectroscopic research. Spectra 

 indicate the normal state of atoms in freedom; occasional magnetic 

 experiments (like those on gaseous oxygen here cited, or those on mo- 

 lecular beams by the Gerlach-Stern method, which I hope to treat on a 

 later occasion) also refer to free atoms and molecules, and confirm 

 the indications of the spectra, thus sustaining both the methods; but 



" Certain metals, the alkali metals for instance, exhibit a paramagnetism which is 

 entirely due to the "free" or conduction electrons. 



