MAGNETIC RESONANCE. II 403 



Next suppos( that what is plotted against v is not /jl but |ju|, the abso- 

 lute valu( of th( permeability. The portions of the n-vs-v curve which were 

 below the horizontal axis now appear inverted and above the horizontal 

 axis. The curve has an upward-pointing peak reaching to infinity at 

 vo , anc' a downward-pointing peak touching th( axis with its tip at Vi . 



Such is the general aspect of the curve of Fig. 4, pertaining to a Heusler 

 alloy. There are superficial differences: the curve of Fig. 4 is plotted 

 against H for constant frequency, and the scale along the axis of ordi- 

 nates is logarithmic. The reader can easily make allowance for these. 

 There is also a fundamental difference : the curve reveals the presence of 

 damping or relaxation, which broadens the peaks and prevents \^l\ 

 from rising to infinity or dropping quite to zero. The continuous curve 

 is derived from a theory which involves a specific assumption about the 

 damping; one sees that it agrees well with the data excepting in a re- 

 gion around the minimum. Curves such as these are likely to be in- 

 fluenced by anisotropy in the ferromagnetic substance, which reversely 

 can be evaluated from the curves. 



How about the values of g for ferromagnetic substances? The Heusler 

 alloy to which Fig. 4 pertains has a value of g which, so far as the accuracy 

 of the experiment permits us to judge, may be identical with the ideal 

 value (the most probable value is however 2.01). This is an exception 

 and not the rule. The range of values is rather wide, though apparently 

 not so wide as in the strongly paramagnetic salts. Most of them lie 

 between 2.22 (for cobalt) and 2.01 (for the Heusler alloy aforesaid); 

 but there are instances of values still higher, including one of 3.75 for 

 manganese arsenide. There is also at least one value lower than 2.00; 

 it is presented by gadolinium, a very interesting element. Below its 

 Curie point at 16° absolute, gadolinium shows a resonance-peak of which 

 the breadth interferes with a precise location of its top; the value of g 

 is given as 1.95 to 1.96. Above the Curie point, gadolinium is para- 

 magnetic, but the peak persists and is sharper; the value of g is 1.95 

 d= 0.03. I remind the reader that when the experiment is such that 

 formula (21) must be used, a g'-value implies an assumption about the 

 value of M the magnetization of the substance at saturation. 



I must not close this topic without alluding to something which there 

 is not space to expound. Experiments on the ''gyromagnetic effect" — 

 something which has a much longer history than ferromagnetic reso- 

 nance — lead to values of a quantity which has also been denoted by g. 

 Until a few years ago it was supposed that this quantity must be the 

 same as the g of these pages; but experiment has ruled otherwise, and 

 theory has been successful in at least suggesting a reason. The g of these 



