246 RELATIVITY, THE GENERAL THEORY
of the existence of Ampere's molecular currents* (explanation of para- and fer-
romagnetism). .. within the limits of error (about 10 per cent) the experiment
yielded in all detail a confirmation of the theory' [E14].** Their experiment, sim-
ple in principle, riddled with complexities in practice, gave the first proof of the
existence of rotation induced by magnetization. Their result was qualitatively
right. However, in the pre-spin days of 1915, any dynamic theory of ferromag-
netism had necessarily to be incorrect. Einstein could not know that his theoretical
prediction was wrong by a factor of about 2. Since Einstein and de Haas claimed
to have found agreement between theory and experiment, their experiment had
also to be wrong by a factor of 2. Their estimate of a 10 per cent experimental
error had to be too optimistic. As we shall see, the alleged agreement between
theory and experiment was largely a theoretical prejudice.
Characteristically, Einstein was unaware of earlier efforts to measure gyro-
magnetic effects until some time after his own work had been completed [E33].
These attempts go back to Maxwell, who remarked in his treatise of 1873 that
'there is as yet no experimental evidence to show whether the electric current is
really a current of a material substance' [Ml]. He proposed several methods for
testing this idea: acceleration of a conductor should generate a currentf; and a
magnet should act like a gyroscope, which is the basic idea of the EdH effect [M2].
In 1861 Maxwell himself attempted to detect such gyroscopic effects, but without
success.
Two other instances of related work prior to 1915 must be mentioned.ff The
theoretical derivation by Einstein in 1915 had already been given in 1907 by
Owen Willans Richardson, who had also tried in vain to observe the rotation by
magnetization, at Palmer Laboratory in Princeton [R1J-H In 1909 Samuel Jack-
son Barnett, then at Tulane University, began the study of the inverse effect, mag-
netization by rotation, now known as the Barnett effect. I shall next outline the
EdH work of 1915 and then state the interesting results obtained by Barnett at
about the same time [B2].
Let us first phrase Ampere's hypothesis in modern language. ^^ The magnetic
moment M of a magnetized body (assumed at rest) is due to circulating 'hidden
* Andre-Marie Ampere had conjectured around 1820 that magnetism can be considered to be caused
by electricity in motion.
**There exists a German [E33], a Dutch [E34], and an English [E35] version of the EdH paper.
Each one of these differs slightly from the other two. The statement on the limits of error in each
paper agrees in substance with what is said in the letter quoted in the text. All three papers appeared
in 1915.
•(•This effect was first observed in 1916 [Tl].
ffA detailed early history of gyromagnetic effects is found in papers by Barnett [Bl].
1[For some years after 1915, the effect was called the Einstein-Richardson effect.HHIn EdH and other early papers, the magnetic moment is defined as cM.