388 



BELL SYSTEM TECHNICAL JOURNAL 



In Fig. 5 is shown the result of plotting Brem/BrVs. {L/D)ylHc(T/Toy 

 for the same set of bars. This combination of variables was arrived 

 at purely by a cut and try method, but as shown, a surprisingly good 

 correlation is obtained. 



It should be mentioned in connection with Fig. 4 that it was found 

 that the data relating to some of the bars did not fit the curve satis- 

 factorily if Brem/Br was plotted against L^HdD, but in each case 



4 5 6 7 8 



LENGTH OF BAR IN INCHES 



Fig. 4 — Curves of remanence vs. length for straight bar magnets of different cross- 

 sections and materials. 



where this was so the bar had been quenched from above or below the 

 optimum hardening temperature. It was found, however, that the 

 data from these bars could be correlated with those from the properly 

 hardened bars if Brem/Br were plotted against {LID)^Hc{TITqY in 

 which T is the actual, and Tq is the optimum hardening temperature 

 on the absolute scale. The ratio TJTq is, of course, equal to unity 

 for bars properly hardened. The definition of the optimum hardening 

 temperature will be given later. 



