316 BELL SYSTEM TECHNICAL JOURNAL 



show how valueless is the mere statement that " the early portion of the 

 magnetization-curve is very nearly straight." 



Values of the coefficient a, which is frequently called initial suscepti- 

 bility — the corresponding value of (1 + 47ra) is called initial perme- 

 ability and denoted by ^to — are rather often determined; it is an 

 important constant of each material. Some pairs of values of a and 

 b are quoted by Ewing, others by Bidwell, and some others were 

 determined by the pupils of Weiss. According to one of these latter 

 (Renger) the values for very pure freshly-annealed iron at room- 

 temperature are: a = 49.9, b = 108. Tempered steel however 

 yielded values of 2.23 for a, and 0.032 for b; from which, and from a 

 mass of other observations on metals hardened by stretching, one 

 sees that the effect of hardening is to lower a a great deal and b a 

 great deal more, so that the curve slopes less sharply upward and 

 does not begin to bend appreciably for a much longer way. I cannot 

 quote all of the relevant data; but it is worth remembering that 

 Rayleigh made measurements so delicate that he was able to follow 

 the curve (for unannealed iron) all the way from .04 to .00004 gauss. 

 Over this range his magnetometer reported no variation in the ratio 

 of / to H. For nickel the detectable upward curvature commences 

 at a much higher fieldstrength — five gauss, according to Ewing. 



The alloys of iron and nickel, containing more than 30 per cent of 

 the latter element, develop extraordinary magnetic properties when 

 they are submitted to certain heat-treatments,* as G. W. Elmen 

 discovered towards 1915. For these, the first segment of the magneti- 

 zation-curve shrinks to a small fraction of the length it has for iron; 

 the two-term formula 



I = all -\- bW~ 



becomes visibly inadequate at 0.02 gauss, as the curve sweeps upward 

 into its rapidly-rising stage. The value of a for some of these "perm- 

 alloys " is as great as 8000, the value of b as great as 4000. 



As the value of H is increased the later terms in the power-series 

 for / bulk larger, and eventually the first segment of the curve passes 

 over into what I have called the second. In this second section the 

 ratio oi I to H rapidly rises, and attains the enormous values which 

 form one of the distinguishing marks of ferromagnetic substances, 

 and are responsible for much of their utility in the world of engineering. 

 Plotted against //, the ratio of / to // appears as a curve with a high 



* For samples of a certain specified shape and size, this is the heat treatment 

 which was recommended: "They are first heated at about 900° C. for an hour and 

 allowed to cool slowly, being protected from oxidation throughout these processes. 

 They are then reheated to 600° C, quickly removed from the furnace, and laid 

 upon a copper plate which is at room-temperature." 



