196 PROCEEDINGS OF THE AMERICAN ACADEMY. 



diameter = 0.153 cm., the length varying from 2 to 6 cms., also of 

 diameter == 0.115 cm. and a length originally 33.4 cms. For the shorter 

 specimens he used Gauss's A position, that is, the rod is placed east 

 and west and the magnetometer is placed in the prolongation of the 

 rod's axis; for the longer wires Ewing's method was used, in which 

 the solenoid and wire are placed vertically, with an extra solenoid to 

 compensate for the earth's field, and the magnetometer being placed east 

 or west of one end of the wire. 



Du Bois subjected these data to a very extensive discussion. He 

 developed the proposition that, provided the length of the rod is 

 sufficiently great compared with its diameter, then i\^m^ = constant. 

 This constant he finds fi:om Ewing's curves to be equal to 45, provided 

 m k 100. The reason why this formula cannot possibly hold for 

 short rods is that the theory of Du Bois assumes that the average 

 magnetization intensity / in the whole rod differs but very little from 

 the / within the secondary coil in the middle of the rod ; in other 

 words, that the magnetization is practically uniform. Of course this 

 is never realized for finite rods and ordinary fields ff', but it seems at 

 first sight as if the magnetization in a rod of large m should be fairly 

 uniform. If we follow Du Bois's method, which gave him the necessary 

 data to construct his table of values for N in case of cyHnders, we may 

 measure abscissa-differences, which are proportional to N, for the 

 curves for rods of large in's, and form three or four simultaneous 

 equations, each of which Hnearly contains cc, the abscissa-difference of 

 the normal curve and the / vs. H' curve for the largest m used in the 

 equations. Any two of these equations give x, and we can thus con- 

 struct the normal curve, which gives us immediately all the A"-curves 

 by plotting abscissa-differences as before. Du Bois, from the meagre 

 data at his command, found values for AT for various m's and has col- 

 lected the results in tabular form (see table, page 204) in his book " Die 

 Magnetischen Kreise in Theorie und Praxis " ("The Magnetic Circuit 

 in Theory and Practice," translated by Atkinson). He apparently con- 

 siders the A^-curves to straight lines, as far as practical purposes are 

 concerned, that is A^ is not a function of H (or i) ; at any rate he 

 does not mention giny such variation of A"^. And as to the question 

 whether or not the A^ for a given m and 1 varies with the diameter 

 of the rod, no data were at hand. 



Now there is no reason to believe the AT-curves for cylindrical rods 

 of the same diameter to be straight lines ; and since we know that the 

 building up of magnetization, and perhaps even the final result, is very 

 decidedly modified by the bulk of iron magnetized, it is quite likely 

 that thick massive rods of iron really give different values for AT from 



