192 PEOCEEDINGS OF THE AMERICAN ACADEMY. 



intervals in which these changes complete themselves. The perform- 

 ance of this instrument is as yet far from satisfactory. If it could be 

 made perfect, we should have an ideal method for permeability deter- 

 minations, for we could then increase the magnetizing field as slowly 

 as we please, reading off the corresponding magnetic inductions for 

 any desired values of the field. It is probable that the oscillograph 

 methods are at present much more to be preferred, as they can be 

 made to record accurately the slow and long-continued changes of 

 magnetic induction through large masses of iron. 



A very good method to use is the " step-by-step " magnetization, 

 where ballistic throws are produced in a Thomson galvanometer, or in 

 a D'Arsonval galvanometer when we use proper precautions to secure 

 the proportionality of throws to the flux changes. These changes in 

 magnetic induction through a secondary coil wound around the iron 

 specimen to be tested are most conveniently obtained by sudden de- 

 creases (or increases) in the resistance of the primary circuit, consisting 

 usually of a storage battery and the magnetizing solenoid. By this 

 arrangement it is not difficult to obtain cyclic hysteresis curves. It 

 has been shown ^^ that the maximal induction B (ox I) which is 

 reached varies with the number of steps taken, the difference being 

 most marked in the region of greatest permeability. As the num- 

 ber of steps is increased continually in different experiments, the B 

 vs. H curves move nearer the ^'-axis, but soon approach the limiting 

 curve for a slow continuous change of H', which, as we saw before, 

 is the one curve that, after the proper back-shearing, will give values 

 for the permeability (and susceptibility) conformable to the ideal 

 definition. Lastly in order of accordance with the ideal definition of 

 susceptibility comes the " reversal " method of measuring ballistic in- 

 duction throws, which is entirely contrary to a slow magnetization, 

 but which is often the most convenient of all the methods to use, and 

 which gives the most self-consistent determinations ; that is, repeated 

 magnetizations will give almost identical results. Both the " step-by- 

 step " and the " reversal " methods of measuring magnetic induction 

 may give results depending on the particular experimental conditions 

 employed, unless one takes proper precautions. Thus the time-constant 

 L/R of the primary circuit should be only one or two per cent of the 

 time it takes the galvanometer-needle to reach its greatest deflection, 

 which time will be the quarter-period of the needle suspension system. 

 It should be noted that when there is a great bulk of iron in the mag- 



10 F. Rucker, Diss. Halle, 1905, 106 pp. 20 plates ; Elektr. ZS. 26, 904-905, 979 

 (1905). 



