Follow-up Q14 (A14): Kennard's experiment and Faraday's disk

Date: 2026-10-08. Role: STEP F follow-up investigator. Web time: about 1.4 minutes of the 20-minute cap (timer 1791484406 to 1791484488). I stopped early because the project's saved files (McDonald's 131-page review, Chyba et al. 2025, Barnett 1919, both Kennard papers) already held most of the answer. The web was needed only for the practical-design side and for one modern "relational" experiment.

Question

"How does the Kennard experiment compare and contrast with Faraday's disk (the unipolar/homopolar generator)? The outcome seems similar: the magnetic field does not spin with its source. What are the implications for theory and for practical designs of electrical equipment?"

What was searched

Project files read: claims.md, arbiter.md section 0, primer.md sections 1-4, sites.md, Kennard 1917 (_text pages 1, 4, 8, 11 and the IA OCR), Kennard 1912 (IA OCR), and McDonald, "Is Faraday's Disk Dynamo a Flux-Rule Exception?" (research/files, sections A.17.7-A.17.9 and the homopolar "drum" section).

Web (Chrome is the default; Wikipedia and two PDFs were fetched with curl into research/pages/ and research/files/):

What was found

1. Faraday's own experiments (1832 and 1851)

• 1832, rotating magnet (Arts. 217-227). Faraday took away the copper disk and spun a conducting cylindrical magnet, with a wire sliding on it at the axis and at the middle. A current flowed. McDonald summarises: "Faraday replaced the copper disk and the external magnet by a conducting, rotating magnet, whose self field acting on the 'free' charges in the magnet also produced a current in the circuit" [read: McDonald, research/files/McDonald-Faraday-disk-flux-rule.pdf, sec. A.17.9].
• 1851, Arts. 3090-3118. Per McDonald [read: same, sec. A.17.7]:
• A current flowed when the external wires rotated about the axis of a magnet at rest.
• A current also flowed when the wires were at rest and the magnet rotated.
• "No current was observed when the external wires and the magnet were rotated together."
• In Art. 3118, Faraday "(re)affirmed that magnetic field lines do not rotate with a rotating magnet."
• McDonald's own reading. "It is better to say that the conduction electrons in the rotating part of the circuit experience a v x B Lorentz force". His footnote adds that this "avoids the issue of whether the lines of the magnetic field are rotating or not, since the value of the vector B for a cylindrical magnet is the same whether or not the magnet is rotating" [read: McDonald, sec. A.17.7 and fn. 79].
• The name. "The term unipolar induction for Faraday's homopolar dynamo is due to Weber (1839)" [read: McDonald, sec. A.23].

2. Kennard's own statement of the difference: closed circuits against open circuits

• 1912. Kennard describes Faraday's rotating bar magnet with a sliding-contact galvanometer circuit and sets out "two views as to the seat of this E.M.F.":
• (1) "that it is produced in the wire, which is at rest, as the magnetic force lines rotating with the magnet cut through it";
• (2) "that it is produced in the magnet itself as the latter moves through its own magnetic field."
• He then says: "It is clearly impossible to decide between these two views by any observations on closed circuits; hence, in the present investigation an electrometer and an open circuit were employed" [read: Kennard 1912, research/files/Kennard-1912-PhilMag-23-937-Unipolar-Induction_IA-OCR.txt, journal p.937-938].
• 1917, p.1. "Experimentally, the question can be answered only by observations upon open circuits." On the theory, Kennard writes: "The oldest view, put forward by Faraday and adopted by Lorentz", refers the effect to (1/c)[v x B] "where ... v = velocity relative to the ether (assumed stationary)" [read: _text p.1].
• 1917, p.4. Kennard's own apparatus contained a Faraday disk. Because he could not record the speed, "a Faraday disk A was mounted on the end of the shaft" with a galvanometer, giving "a reading proportional to the product of field-strength and speed" [read: _text p.4]. That is the galvanometer column of his table. The ratio of electrometer to galvanometer readings is therefore, literally, Kennard's open-circuit charge divided by a closed-circuit Faraday-disk current.
• 1917, p.11 (journal p.189). Kennard uses Faraday's law for closed circuits to argue that the charging cannot be "conditioned by rotation relative to the connecting wires and electrometer". He concludes that "the effect is due to an absolute rotation in the same sense in which the operation of a gyrocompass is due to an absolute rotation of the earth" [read: _text p.11].

3. The return-path argument (Faraday paradox)

The Wikipedia "Faraday paradox" article explains why Faraday's closed-circuit observations cannot settle whether the lines rotate. If the field rotated with the magnet and the disk, "a current should still be produced, not by EMF in the disk ... but in the external circuit linking the brushes". It adds that "so long as a current loop is used to measure induced EMFs from the motion of the disk and magnet it is not possible to tell if the magnetic field does or does not rotate with the magnet." For this it cites A. G. Kelly, Institution of Engineers of Ireland monographs 5 and 6 (1998) [web: https://en.wikipedia.org/wiki/Faraday_paradox, saved research/pages/wiki-faraday-paradox.txt].

The same article then says: "Several experiments have been proposed using electrostatic measurements or electron beams to resolve the issue, but apparently none have been successfully performed to date" (tagged "citation needed", Feb 2019). This conflicts with the record in this project: - Kennard 1912 and 1917 are electrostatic open-circuit experiments. - Barnett 1919 reports Kennard's and Pegram's 1917 results as confirmed [read: research/files/AIEE-1919-report-electromagnetic-induction.pdf]. - Chyba et al. 2025 call the result "a standard part of the electromagnetics literature" [read: research/files].

So the Wikipedia sentence is the article editors' statement. It is contradicted by the published record. It is noted here as a stale secondary claim, not as evidence against Kennard.

4. Practical homopolar machines

Source: [web: https://en.wikipedia.org/wiki/Homopolar_generator, saved research/pages/wiki-homopolar-generator.txt]. - One-piece rotor. "Long after the original Faraday disc had been abandoned as a practical generator, a modified version combining the magnet and disc in a single rotating part (the rotor) was developed. Sometimes the name homopolar generator is reserved for this configuration." - Early patents. Delafield (US 278516), Ferranti, Batchelor, Tesla (US 406968: two discs joined by a metal belt, "so that ... both electrical pickups [interface] with the shafts" and avoid the high-speed rim contact), Steinmetz, Thomson, Noeggerath and Eickemeyer. The Forbes dynamo was "in widespread use during the beginning of the 20th century." - Pulsed power. The ANU (Australian National University) machine built under Sir Mark Oliphant "stored up to 500 megajoules" and supplied "up to 2 megaamperes". It was used from 1962 until it was taken apart in 1986. Parker Kinetic Designs built machines for railguns and other weapons, and 10 MJ industrial units for welding. - Contacts. "The electrical contacts are usually made through a brush or slip ring, which results in large losses at the low voltages generated." Mercury, gallium or NaK (sodium-potassium alloy) are used as liquid brushes. - Naval motors. Search snippets only, papers not opened (F14-7): Walters 1998 says superconducting homopolar motors "are well suited to Navy propulsion systems due to their inherent low noise". A General Atomics program for the Office of Naval Research built "a 500 kW Test Stand Motor and a 3.7 MW Motor". Eriksson 1982 treats "the current collection problem of superconducting homopolar machines".

5. Where the reaction torque acts (relational reading)

Guala-Valverde, Mazzoni and Achilles, "The homopolar motor: A true relativistic engine", Am. J. Phys. 70 (2002) [read: research/files/Guala-Valverde-2002-AJP-homopolar-motor.pdf, p.1053]: - They split the torques into "Magnet-probe ... equal but opposite torque on the magnet" and "Magnet-closing wire. The magnet exerts a clockwise torque on the closing wire, and the wire an equal but opposite torque on the magnet." - "With the probe attached to the magnet ... rotation is forbidden. Conversely, with the closing wire mechanically decoupled from the magnet ... The torque exerted by the closing wire on the magnet is responsible for the observed rotation." - They list "Kennard, Bartlett, Panofsky, Müller, Wesley" as taking "absolutistic viewpoints". Against them they set "Weber, Assis, and Kelly", who "adopted a relativistic framework". That is their characterization; they are a party to the dispute. - Their summary of the closed-circuit case: "the disk plus magnet rotation at ω with the closing wire at rest in the lab is entirely equivalent to the closing-wire rotation at -ω with the disk plus magnet at rest."

6. The N-machine claim and an independent test

Robert Kincheloe (Professor of Electrical Engineering, Emeritus, Stanford), "Homopolar 'Free-Energy' Generator Test", Society for Scientific Exploration meeting, 21 June 1986, revised 1 February 1987 [web: https://rexresearch.com/kinchelo/kinche.htm, saved research/pages/rexresearch-kincheloe.txt]: - DePalma's claim, as Kincheloe states it (DePalma is the inventor and an interested party): "when the conducting disk is attached to a rotating magnet, the interaction of the primary magnetic field with that produced by the radial output current results in torque between the disk and the magnet structure which is not reflected back to the mechanical driving source. Lenz's law therefore does not apply." - Kincheloe's finding: "While it did not perform as claimed, repeatable data showed anomalous results that did not seem to conform to traditional theory." One measured quantity "seemed to be either about 13% or 20% of the maximum computed generated power, depending on interpretation." - Kincheloe on the machine as a design: it is "inefficient and not suitable for power generation". The field is "concentrated near the axis where the tangential velocity is low", it takes "approximately 4 kilowatts" to energize the magnet, and the brush area of over 30 square inches means "considerable friction loss". He adds that it "was not intended as a practical generator but as a means for testing the free energy principle."

Reasoning: compare and contrast

Same physics. In both cases a conductor turns about the axis of an axially symmetric magnetic field. The free electrons feel the force q v x B and are pushed radially. The potential difference between axis and radius r is

V = ωBr²/2 = Φ(r)·n,

where ω is the angular speed (rad/s), B the axial field, Φ(r) = πr²B the flux inside r, and n = ω/2π the revolutions per second. Kennard's rule "any point ... metallically connected to earth comes to a potential of Nn" (p.5) is the Faraday-disk formula. In both, spinning an axisymmetric magnet alone does nothing at a conductor that is at rest. Faraday reported this with galvanometer circuits; Kennard confirmed it with an electrometer (solenoid-alone rows).

Two worked numbers. - Kennard 1917: flux 55,400 maxwells at 33 rev/s gives V0 = 55,400 x 33 x 10^-8 = 0.0183 V (arbiter section 1). - A laboratory homopolar machine: B = 1 tesla, R = 0.5 m, 3000 rpm (ω = 314 rad/s) gives V = 314 x 1 x 0.25 / 2 = 39 V [calc].

Homopolar machines are always low-voltage, high-current machines. That is why their sliding contacts dominate the losses.

Difference 1: what is measured. - Faraday measured currents in closed circuits. A closed loop measures only the total EMF around it. That total depends only on the relative rotation between the rotating part (disk, or disk plus magnet) and the stationary return wire. Faraday found no current when wires and magnet turned together. - So in Faraday's experiments the two pictures can always be traded for each other: "lines rotate with the magnet and cut the stationary wire" or "lines stay put and the moving metal cuts them". Kennard said exactly this in 1912, and the Faraday-paradox literature (Kelly) says it still. - Kennard measured charge on an open conductor with an electrometer. That tells you where the EMF arises. - Faraday's 1851 statement that the lines do not rotate was therefore a well-argued conclusion. It was not an experimental proof. Kennard and Pegram, together with Barnett's 1912 solenoid-alone null, made it one.

Difference 2: what can rotate. - Faraday's 1832 experiment, and Kennard's 1912 one, spun only the iron magnet, while the magnetizing coil (Kennard 1912) or the surrounding circuit stayed fixed. Barnett's objection was that this cannot show what fraction of the "lines" would move. - In 1917 Kennard made the source itself, a current-carrying solenoid, rotatable, together with or apart from the condenser. The "both together" case (condenser charges with no relative motion between it and the magnet) is the one Faraday's closed-circuit setups could never isolate.

Difference 3: the role of the measuring circuit. - In a Faraday disk the external circuit is part of the effect: no relative rotation means no current. - In Kennard's experiment the electrometer leads touch near the axis. Kennard argues from Faraday's own closed-circuit law (p.11) that rotating the leads would add nothing. The charging therefore belongs to the condenser's own rotation.

Implications for theory (competing readings kept)

1. Standard reading. B is a field defined at points of an inertial frame. For a steady, axisymmetric source, spinning the source does not change B anywhere, so "do the lines rotate?" has no physical content. The v in v x B is the conductor's velocity in that frame. Faraday's results, Kennard's and the flux rule for closed loops all follow. Rotation is "absolute" only in the sense that a gyroscope detects it. Chyba et al. 2025 present this as settled [read]. On this reading Kennard's "stationary aether" and "inertial frame" predict identical charges, so the experiments cannot choose between them (arbiter, claim 6).
2. Kennard's reading (1917). The radial force on electrons in a body with no relative motion of parts requires motion relative to "something which is not matter", the Lorentz aether (p.11). Faraday's lines-in-a-medium view, which Kennard says was "adopted by Lorentz" (p.1), points the same way. Barnett 1919 allowed it as one of two options.
3. Relational reading (Weber, Assis, Kelly, Guala-Valverde). Only relative motion between parts of the system matters, and the torque reaction acts between the magnet and the closing wire. Guala-Valverde et al. classify Kennard as "absolutistic". This reading describes closed-circuit homopolar machines naturally. The open-circuit "both rotating" charge is what it has to explain; Weber-electrodynamics papers that cite Kennard 1917 (sites.md, related work) take it on directly.
4. What is ruled out. - The "moving-line" theory of Barnett 1912 (the moving lines act on electrons) is ruled out experimentally. - Hertz's fully dragged aether, which predicts no charge on a spinning condenser, is also ruled out. - None of this contradicts Faraday's flux rule for closed circuits, which Kennard himself uses. - The disk is a known case where the simple "rate of change of flux through the loop" needs care (Feynman's "exception to the flux rule"; McDonald argues the loop must be chosen correctly).

Implications for practical designs (inferred from the sources above)

1. The magnet may ride on the rotor. The field does not co-rotate, so a magnet fixed to (or part of) the rotating disk or drum gives the same EMF as a stationary magnet. This is the "one-piece" homopolar generator and Faraday's 1832 rotating magnet. Designers can put the field coil or magnet wherever suits the mechanics.
2. A stationary return path, and so sliding contacts, cannot be avoided. - Output requires relative rotation between the rotor and the external circuit. If the external circuit turns with the rotor, there is no current (Faraday Art. 3097, via McDonald). - Therefore every homopolar machine needs brushes, slip rings or liquid-metal contacts at low voltage and very high current. This is the main engineering cost: friction and contact loss, wear, and liquid NaK or gallium handling in naval superconducting motors. - Tesla's two-disk belt patent is one attempt to move the contacts to low-speed shafts.
3. The reaction torque has to go somewhere. - In the standard view, the current in the stationary return conductor and brushes, interacting with the field, carries the reaction. Guala-Valverde et al. measured that torque on the closing wire. - Supports and leads must be designed for it. - DePalma's N-machine claim that the torque stays inside a one-piece rotor and does not load the drive was tested by Kincheloe (1986). The machine "did not perform as claimed", but he also reported "repeatable data" with "anomalous results". Both statements are his.
4. Where the EMF arises matters even when the total does not. Kennard's one practical point (p.11): rotating a dynamo's armature is not the same as rotating its field magnets the other way. "The total E.M.F. is the same, but in the first case it is developed almost entirely in the longitudinal parts of the winding, while in the second case a large fraction of it is developed in the radial parts, and the distribution of electrification on the armature will be different." This bears on where insulation is stressed and where static charge collects. Kennard himself judged "the practical bearing of these experiments is small" (p.11).
5. Planet-scale version. The same principle (a conductor moving through an axisymmetric field that does not co-rotate with its source) underlies Chyba et al.'s proposed generator driven by the Earth's rotation. It is contested: there is a null result by Veltkamp and Wijngaarden and a 2026 Comment (see sites.md).
6. Pulsed power and propulsion. - The flywheel rotor of a homopolar machine stores energy and dumps it at megaampere currents (ANU 500 MJ, 2 MA; railgun supplies). - Its direct current needs no commutator, which is the basis of the low-noise naval motor programs.

Open

• Pegram 1917 (Phys. Rev. 10, 591) is still not obtained. It is the independent "both rotating" replication.
• Miller's chapter "The problem of whether magnetic lines of force rotate: 1831-1901" (Frontiers of Physics: 1900-1911, Springer) was not opened. It would give the history of the question from Faraday to Kennard.
• Faraday's Arts. 3090-3118 (1851) and 217-227 (1832) were read only through McDonald's summary, not in Faraday's text.
• Kincheloe's "anomalous" data: no follow-up test or reanalysis was found in this run.
• The Guala-Valverde torque experiments: Montgomery 2007 (Apeiron) comments on them; not read.