474 



SCIENCE 



[N. S. Vol. LIII. No. 1377 



per second should magnetize them as would 

 a field of intensity 



ff = - 7.1 X 10-W gauss. (17) 



Now in two investigations on cold-roUed 

 steel by a method of electromagnetic induc- 

 tion, a third investigation principally on steel, 

 nickel and cobalt by a magnetometer method, 

 and a fourth investigation on steel, soft iron, 

 cobalt, and Heusler alloy by another magnet- 

 ometer method, Mi-s. Barnett and I have found 

 the above theory verified, except that in place 

 of the rnunber Y.l we find smaller num- 

 bers; in the best work, which appears to 

 be free from any serious systematic error, we 

 find instead of Y.l numbers about one half as 

 great and even smaller.^"- This work, however, 

 is still in progress. 



If these results are correct, we seem to be 

 driven to one of two conclusions : Either nega- 

 tive electrons or magnetons with a value of 

 m/e or B for the motions involved different 

 from that ordinarily accepted are responsible 

 for magnetism; or positive electrons or mag- 

 netons, whose rotation we have seen must 

 produce an opposite effect, are also involved. 

 It does not seem impossible that a ring or 

 other type of negative magneton, with B 

 different for the permanent rotary motion 

 from the value obtained from cathode ray ex- 

 periments and otherwise, should be involved; 

 but the presence of positive electron orbits, 

 or rotating positive magnetons, is also pos- 

 sible, especially in view of the known expul- 

 sion of a particles with great velocities from 

 the radioactive substances. Chemical evi- 

 dence is often quoted in favor of the idea 

 that sux)erficial negative electrons alone are 

 responsible for magnetism ; but I do not think 

 this evidence conclusive. 



Not long after our first conclusive experi- 

 ments on magnetization by rotation were 

 communicated to this society, experiments on 

 the converse effect, viz., rotation by magneti- 

 zation, first suggested by Eichardson, were 

 performed by Einstein and deHaas.^^ Mag- 

 si P%s. Eev., 6, 1915, p. 239; 10, 1917, p. 7; 

 Proc. American Phys. Soc. for December, 1920; 

 Proo. PUl. Boo. of Washington, for October 9, 1920. 

 22 Verh. d. D. Phys. Ges., 17, 1915, p. 152. 



netizing a bar of iron, in turning the mag- 

 netons about until they all rotate in the same 

 direction, produces angular momentum in this 

 direction which must be compensated by an 

 angular momentum within the molecules them- 

 selves, or in the bar, or in the magnetizing 

 solenoid. If we assume that the reactions all 

 take place upon the bar we can calculate 

 47r(m/e) from the measurable angular mo- 

 mentum communicated to it when magnetized 

 to a given intensity on the assumption that all 

 the magnetons are alike. The residt published 

 by Einstein and deHaas agreed closely with 

 the value to be expected on the hypothesis that 

 only the common type of negative electron is 

 involved; but the sign of the effect was not 

 determined with certainty tiU much later, and 

 errors in the value of the assixmed intensity of 

 magnetization and in the experiments them- 

 selves undoubtedly exist. The experiments 

 have been repeated with great improvements 

 by Emil Beck,-^ and experiments on the same 

 subject but by a different method had already 

 been made by J. Q. Stewart;^* both these in- 

 vestigations, on the basis of a single kind of 

 electron and on the assumption made above 

 with reference to the seat of the reaction to 

 the electron momentum, give values of 4ir(m/e) 

 for iron and nickel similar to those which Mrs. 

 Barnett and I have found by the method of 

 magnetization by rotation, into the theory of 

 which no assumptions appear to enter except 

 such as can be justified completely. 



If a magneton is sufficiently free it will, as 

 stated above, when rotated about a given axis 

 align itself with its axis completely parallel to 

 the axis of impressed rotation. If in the unit 

 volume there are N magnetons all alike, each 

 with the moment of inertia C and initial 

 angular velocity U about the magnetic axis, 

 and if the effects of collisions and the demag- 

 netizing field are negligible, the intensity of 

 magnetization will be 



'=S<^+°>- 



(18) 



The first term is independent of 12, which is 

 a measure of the intrinsic intensity of rota- 



23 Ann. der Phys. (4), 60, 1919, p. 109. 



^iPhys. Eev. (2), 11, 1918, p. 100. 



