One century later: Remarks on the Barnett experiment Alexander L. Kholmetskiia) Department of Physics, Belarus State University, 4, F. Skorina Avenue, 220080 Minsk, Belarus 共Received 16 May 2002; accepted 10 January 2003兲 The Barnett experiment 共the rotation of either a solenoid or a short-circuited cylindrical capacitor about a common axis兲 was performed at the beginning of the 20th century to choose between the Hertz, Maxwell, or Lorentz electrodynamics theories. The results obtained by Barnett led him to conclude that both the electrodynamics of Hertz and Maxwell should be rejected, while Lorentz was solely correct. We analyze this experiment from a modern point of view and show that the results can be explained within the framework of Maxwell electrodynamics and relativity theory. However, the explanation still has some physical difficulties. © 2003 American Association of Physics Teachers. 关DOI: 10.1119/1.1557300兴 I. INTRODUCTION The experiment by Barnett1,2 was carried out at the beginning of the 20th century to resolve the problem of unipolar induction. At that time it was an exciting problem closely related to the question: do the lines of magnetic induction move as if they were rigidly attached to the magnet, or do the lines remain fixed in space under the rotation of magnet? It was expected that experimental resolution of this problem 共formulated in the archaic language of the beginning of the 20th century兲 would select either the Hertz, Maxwell, or Lorentz electrodynamic theory. The Hertz theory was associated with a fully entrained ether, the Lorentz theory was associated with nonentrained ether, while Maxwell’s electrodynamics was based on special relativity, independent of any ether model. It seemed that Barnett devised the most interesting way of experimentally resolving the problem, and provided the most convincing results in comparison with other workers. He concluded that the lines of magnetic induction remain fixed in space under the rotation of a magnet, in agreement only with Lorentz electrodynamics and Lorentz ether theory. Later it was recognized that Barnett’s experiment does not contradict the special theory of relativity: Einstein’s relativity principle is valid for inertial motion solely, and in general, it does not imply a relativity of rotational motion. During the 20th century the problems of classical electrodynamics with rotational motion and related problems of unipolar induction were extensively investigated theoretically and experimentally.3–14 Nevertheless, Barnett’s experiment still has no consistent modern explanation. II. DESCRIPTION OF THE EXPERIMENT BY BARNETT A simplified schematic of the experiment is shown in Figs. 1 and 2. There is a solenoid with a current I and a cylindrical ជ capacitor inside the solenoid, where a magnetic induction B is parallel to the axis z. For an elongated solenoid the value of Bជ can be taken to be constant in the inner region far from the boundaries of the solenoid. The capacitor is shortcircuited by a conducting wire A. Both the solenoid and the capacitor are mounted coaxially and can rotate about the common axis z. An electrical connection between facings of the capacitor breaks during rotation, and after the current goes to zero and the rotation stops, the charge on the capaci558 Am. J. Phys. 71 共6兲, June 2003 http://ojps.aip.org/ajp/ tor is measured. As the capacitor rotates, it acquires a charge Q due to the Lorentz force applied to the wire A moving in ជ . Barnett did not perform the experiment the magnetic field B with a rotating capacitor 关Fig. 2共a兲兴, because similar experiments had been already carried out by Faraday and many others. For the rotating capacitor of Fig. 2共a兲 the charge Q can be easy calculated. The goal of Barnett’s experiment is to measure the charge on the capacitor after the rotation of the solenoid 关Fig. 2共b兲兴 and to compare it with the calculated value of Q. The inner armature of the capacitor was a brass tube 14.9 cm long and an external diameter of 3.97 cm. The outer armature was a brass tube 28.0 cm long and an internal diameter of 6.67 cm. The length of the solenoid exceeded the length of the outer armature. During the rotation of the solenoid about the z axis at 20 rev/s, a connection between the armatures is broken by a special key system after which the field is reduced to zero. Then a sensitive electrometer was used to measure the charge on the capacitor. The results of this experiment and other modified Barnett experiments showed that the capacitor acquired a charge that was not more than 1.4% of the calculated Q within the precision of the experiment. Hence, he concluded that the relativity principle is violated, and only Lorentz electrodynamics is able to explain the experimental results. III. MODERN EXPLANATION OF BARNETT EXPERIMENT In the following calculations we assume for simplicity that aⰆr, where a is the length of the wire A in the laboratory frame, and r is the modulus of the radius vector of any point on the wire A. We will carry out all calculations to order r/c 共in Barnett’s experiment this ratio was about 10⫺8 ); we will use units such that c⫽1 共c is the speed of light in vacuum兲. To this order we first consider the experiment in Fig. 2共a兲. If the capacitor rotates with frequency in the ជ ⫻rជ , laboratory frame, the wire A has a linear velocity uជ ⬇ ជ which is orthogonal to the magnetic field B at any moment. Hence, the conduction electrons in A are subject to the Lorentz force F⫽euB 共1兲 in the radial direction 共e is the charge of electron兲. This force charges the capacitor to the potential difference © 2003 American Association of Physics Teachers 558 It is interesting to analyze these experiments from the viewpoint of an observer in an accelerated frame, attached to either the capacitor 关Fig. 2共a兲, case 1: Q⫽0] or to the solenoid 关Fig. 2共b兲, case 2: Q⫽0]. 共The various cases discussed in the following are summarized in Table I.兲 Under rotation about the axis z at the constant angular frequency , the relation between the spatial coordinates of the laboratory 共rest兲 frame (x ⬘ ,y ⬘ ,z ⬘ ) and the spatial coordinates of the rotating frame 共x, y, z兲 has the form x ⬘ ⫽x cos共 t 兲 ⫺y sin共 t 兲 , Fig. 1. Simplified schematic of the Barnett experiment: 共a兲 top view and 共b兲 side view. y ⬘ ⫽x sin共 t 兲 ⫹y cos共 t 兲 , 共3兲 z ⬘ ⫽z. The nonzero metric coefficients in such a rotating frame are R0 R0 ⌬V⫽auB ln ⫽ raB ln . r0 r0 The measured charge is Q⫽C⌬V⫽C raB ln R0 . r0 共2兲 Now let us calculate the charge on the capacitor for the experiment in Fig. 2共b兲. Because of the rotation of the soleជ and magnetic Bជ field change inside. noid, the electric field E The magnetic field is not relevant because the capacitor is at rest in the laboratory frame. To determine the electric field Eជ inside the solenoid in the laboratory frame, we note that a moving conductor with nonzero current acquires a nonzero charge density . A physical reason is the different scale contraction for the systems of negatively and positively charged particles in a conductor, because when I⫽0, they have different circular speeds. 共Formally, can be found by the transformation of the four-vector of the current density.兲 At the same time, due to the symmetry of Barnett’s experimental apparatus to rotations and translations with respect to the axis z, the charge should be distributed homogeneously on the surface of the solenoid. In the region far from the boundaries of the solenoid, the electric field inside is equal to zero for any value of the charge density according to electrostatics. Hence, we conclude that inside the solenoid the electric field is equal to zero. Therefore, there are no electrical forces acting on the electrons in wire A, and the capacitor does not acquire any charge. This reasoning explains the result of Barnett’s experiment. The results of the experiment might imply that the concept of relative motion is not applicable, in general, to accelerated motion. g 00⫽1⫺ 2 r 2 , g 11⫽g 22⫽g 33⫽⫺1, g 01⫽g 10⫽ y, g 02⫽g 20⫽⫺ x. To determine the electric and magnetic fields in the rotating frame, it is necessary to apply a corresponding transformation to the electromagnetic field tensor F ik ⫽ x ⬘l x ⬘m F⬘ , x i x k lm 559 Am. J. Phys., Vol. 71, No. 6, June 2003 共5兲 taking into account its form in the laboratory frame with a solenoid at rest: ⬘ ⫽ F lm 冦 0 0 0 0 0 0 ⫺B 0 0 B 0 0 0 0 0 0 冧 共6兲 . By substituting Eq. 共6兲 into Eq. 共5兲 and taking into account Eq. 共3兲, we obtain the nonzero components of F in the rotating frame to the assumed accuracy of the calculations: F 01⫽⫺F 10⫽⫺Bx , F 02⫽⫺F 20⫽⫺By , 共7兲 F 12⫽⫺F 21⫽⫺B. The nonzero components of the electric and magnetic fields are found according to 共see, for example, Ref. 15兲 冑␥ E 1 ⫽⫺F 01g 11 11 g 00 , E 2⫽ F 02 冑g 00 , B 3 ⫽ 共 g 11F 12⫹g 01F 02兲 冑␥ 11, 共8兲 where ␥ is the metric tensor of three-dimensional space, related to the metric tensor of four-dimensional space–time g by the relation ␥ ␣ ⫽⫺g ␣ ⫹ Fig. 2. 共a兲 Cylindrical capacitor rotates around the axis z at a constant angular frequency ; 共b兲 solenoid rotates around the same axis at the angular frequency ⫺. 共4兲 g 0␣g 0 g 00 共 ␣ ,  ⫽1,...,3 兲 . 共9兲 Now let us consider the experiment in Fig. 2共a兲 共case 1, Q⫽0 for a laboratory observer兲 from the viewpoint of an observer rotating with the capacitor. 共We designate this case CR.兲 For this observer the wire A does not move, and the magnetic field is not relevant. Hence, the capacitor can only be charged due to the action of the electric field on the conduction electrons in the wire A. If we use Eq. 共4兲 and Eqs. 共7兲–共9兲, we obtain E x ⫽⫺Bx and E y ⫽⫺By to the given order of approximation. Thus, the electric field vector is in Alexander L. Kholmetskii 559 Table I. Summary of Barnett’s experiment. Conditions Place of observer Charge Case 1. Faraday’s experiment 共rotating capacitor兲 Case 2. Barnett’s experiment 共rotating solenoid兲 Case CR. Faraday’s experiment Observer in laboratory frame Q⫽0 ជ ). Capacitor is charged by the force e(uជ ÃB Observer in laboratory frame Q⫽0 Capacitor is uncharged, because E⫽0 inside the solenoid for a uniform charge distribution over the surface of the solenoid. Observer attached to rotating capacitor Q⫽0 Case SR. Barnett’s experiment Observer attached to rotating solenoid Q⫽0 Case CRG. Faraday’s experiment in a gravitation field Observer attached to rotating capacitor; it is equivalent to case 2 according to local Lorentz invariance Observer attached to rotating solenoid; it is equivalent to case 1 according to local Lorentz invariance Observer in laboratory frame Q⫽0 ជ . The electric field E results Capacitor is charged by the force eE from the nonuniform charge distribution over the surface of the solenoid. Capacitor is uncharged, because the resultant Lorentz force ជ ⫹e(uជ ÃBជ )⫽0 inside the solenoid. Here Eជ ,Bជ ⫽0. The eE ជ is induced by the nonuniform compensating electric field E charge distribution over the surface of the solenoid. Capacitor should be uncharged as in the equivalent case 2. Q⫽0 Capacitor should be charged as in the equivalent case 1. Q⫽0 Observer in laboratory frame Q⫽0 Capacitor should be uncharged as in case CRG. Hence, the resultant ជ ⫹e(uជ ÃBជ )⫽0 inside the solenoid. Lorentz force eE ជ ,Bជ ⫽0. At the same time, the gravitational field with the Here E axial symmetry creates a uniform charge distribution over the surface of the solenoid and, according to the laws of electrostatics, ជ should be equal to zero inside the solenoid. the field E Capacitor should be charged as in case SRG. A single possible force for its charge is eEជ . At the same time, the gravitational field with axial symmetry creates a uniform charge distribution over the surface of the solenoid and, according to electrostatics, ជ should be equal to zero inside the solenoid. the field E Case SRG. Barnett’s experiment in a gravitation field Case CG. Faraday’s experiment in a gravitation field Case SG. Barnett’s experiment in a gravitation field the radial direction, and its magnitude is E⫽uB. Hence, we obtain the same expression, Eq. 共1兲, for the force charging the capacitor, F⫽eE⫽euB. It is interesting to find the origin of this electric field. The reason for the creation of the electric field inside of the solenoid is the nonzero charge density on its surface. Its value can be found using the transformation of the fourvector of current density j i⫽ 冉 , u␣ 冑⫺g 冑⫺g 冊 from the laboratory to the rotating frame, applying the transformation law j i⫽ x ⬘k j⬘ xi k for the transformation 共3兲. 共Here g⫽det g.) We omit the simple calculations and present the final result to the accuracy of the calculations. The charge density at any fixed point on the solenoid is equal to ⫽ j R cos ␣, where j is the current density of the solenoid in the laboratory frame, R is the radius of the solenoid, and ␣ is the angle between the axis of the wire A and the radius-vector in the rotating frame. We conclude that is distributed nonuniformly over the surface of the solenoid. 560 Am. J. Phys., Vol. 71, No. 6, June 2003 Comment According to electrostatics, such a distribution of charge creates a constant electric field in the inner volume of the solenoid, which is collinear to the axis of the wire A. The calculated electric field is E⫽uB, which coincides with the value obtained by the direct transformation of the electric and magnetic fields to the rotating frame. Thus, we conclude that for the case CR, the electric field, which is responsible for the charge of the capacitor, results from the nonuniform distribution of charge over the surface of the solenoid. Now let us consider the experiment in Fig. 2共b兲 共rotating solenoid in a laboratory frame, case 2兲, and explain the result (Q⫽0) for an observer attached to the solenoid 共case SR兲. In its reference frame the capacitor rotates with the angular velocity and both the electric and magnetic fields can act on the wire A. We use Eqs. 共4兲, 共7兲, and 共8兲 and obtain the nonzero components of these fields, E x ⫽⫺ xB, E y ⫽⫺ yB, and B z ⫽B, to the desired accuracy. In such a case ជ lies in the radial direction, and the forces eEជ and e(uជ E ជ ) act in opposite directions. The magnitude of the electric ⫻B field vector is uB. Hence, we see that the total Lorentz force per each electron of the wire A is ជ ⫹e 共 uជ ÃBជ 兲 ⫽0. qFជ ⫽eE 共10兲 These considerations mean that for case SR the capacitor remains uncharged. As for the case CR, the electric field Alexander L. Kholmetskii 560 inside of the solenoid is induced by the nonuniform distribution of charge over its surface. The result of Barnett’s experiment is also useful for an analysis of the principle of local Lorentz invariance if we imagine that in this experiment we are able to create a gravitational field that is axially symmetrical with respect to the axis z. Under rotation of the capacitor 共case 1兲, we choose the gravitational potential to be equal to the centripetal acceleration of an observer on the outer surface of the capacitor 共case CRG兲. For rotation of the solenoid 共case 2兲, the gravitational potential is equal to the centripetal acceleration of an observer on the surface of the solenoid 共case SRG兲. If the radii of the solenoid and the outer armature of the capacitor are close to each other, we can assume the same gravitational field in both cases CRG and SRG. Then, according to local Lorentz invariance, cases 2 and CRG, as well as cases 1 and SRG should be equivalent. This equivalence means that for an observer in the laboratory frame, the gravitational field reverses the results of Barnett experiment: under rotation of the capacitor it remains uncharged 共case CG兲, while under rotation of the solenoid the capacitor is charged 共case SG兲. Such a reversal of the Barnett experiment can only be explained by the redistribution of currents and charges in space in the presence of the gravitation field. In particular, we require the appearance of the electric field in the solenoid, which compensates the action of the magnetic force on the wire A under rotation of the capacitor 共case CG兲, and charges the capacitor under rotation of the solenoid 共case SG兲. At the same time, the gravitational field with axial symmetry induces a uniform distribution of the charge on the surface of the solenoid in the Barnett experiment. In such a case, according to the previously mentioned result of electrostatics, the electric field cannot appear inside the solenoid. IV. CONCLUSION We have seen that Barnett’s experiment can be fully explained within the framework of Maxwell’s electrodynamics theory. A consideration of this experiment for the observers attached to either the rotating capacitor or the rotating solenoid also explains the experimental observations: the rotating capacitor acquires a charge, while it remains uncharged under the rotation of the solenoid. According to the principle of local Lorentz invariance, the application of an external gravitational field with an axial symmetry relative to the rotational axis should reverse the result of the experiment: the rotating 561 Am. J. Phys., Vol. 71, No. 6, June 2003 capacitor remains uncharged, while under rotation of the solenoid the capacitor acquires a charge. However, the electric field, which is responsible for such an inversion of the results of Barnett’s experiment, cannot appear inside the solenoid for any uniform distribution of charge over its surface, induced by the axially symmetrical gravitational field. It would be impossible in the near future to realize the Barnett experiment in such a strong gravitation field for an experimental resolution of the calculated contradiction between the requirements of local Lorentz invariance and electrostatic laws. Nevertheless, it would be interesting to do a further analysis of the local Lorentz invariance principle. ACKNOWLEDGMENTS The author warmly thanks Vladimir Onoochin 共Moscow兲 and Oleg Missevitch 共Institute of Nuclear Problems, Belarus State University, Minsk兲 for helpful discussions. a兲 Electronic mail: kholm@bsu.by S. J. Barnett, ‘‘On electromagnetic induction and relative motion,’’ Phys. Rev. 35, 323–336 共1912兲. 2 S. J. Barnett, ‘‘On electromagnetic induction,’’ Electromagnetic Induction 11, 323–326 共1913兲. 3 W. K. H. Panofsky and M. Phillips, Classical Electricity and Magnetism 共Addison–Wesley, New York, 1955兲. 4 R. Feynman, The Feynman Lectures on Physics 共Addison–Wesley, Reading, MA, 1964兲, Vol. II. 5 J. Guala-Valverde and P. Mazzoni, ‘‘The principle of relativity as applied to motional electromagnetic induction,’’ Am. J. Phys. 63, 228 –229 共1995兲. 6 F. Müller, ‘‘Unipolar induction experiments and relativistic electrodynamics,’’ Galilean Electrodynamics 1, 27–31 共1990兲. 7 G. M. Graham and D. G. Lahoz, ‘‘Observation of static electromagnetic angular momentum in vacuo,’’ Nature 共London兲 285, 154 –155 共1991兲. 8 P. Graneau, ‘‘Electromagnetic jet-propulsion in the direction of current flow,’’ Nature 共London兲 295, 311–312 共1992兲. 9 A. E. Robson and J. D. Sethian, ‘‘Railgun recoil, Ampere tension, and the laws of electrodynamics,’’ Am. J. Phys. 60, 1111–1117 共1992兲. 10 J. Guala-Valverde and P. Mazzoni, ‘‘The unipolar Faraday generator revisited,’’ Galilean Electrodynamics 4, 107–108 共1993兲. 11 A. K. T. Assis and D. S. Thober, ‘‘Unipolar induction and Weber’s electrodynamics,’’ in Frontiers of Fundamental Physics, edited by M. Barone and F. Selleri 共Plenum, New York, 1994兲, pp. 409– 414. 12 J. Guala-Valverde, P. Mazzoni, and Ro Blas, ‘‘The quasilinear generator and 兰 B"dS⫽0,’’ Am. J. Phys. 65, 147–148 共1997兲. 13 A. G. Kelly, ‘‘Experiments on unipolar induction,’’ Phys. Essays 12, 372– 382 共1999兲. 14 P. Graneau and N. Graneau, ‘‘Electrodynamic force law controversy,’’ Phys. Rev. E 63, 058601–2 共2001兲. 15 A. A. Logunov, Lectures on the Theory of Relativity 共Moscow State University, Moscow, 1984兲 共in Russian兲. 1 Alexander L. Kholmetskii 561