Mathematician review: Sykes-Chris, "Guidelines to Bucking Coils" (v2.7, 26.02.16)

Reviewer role: PhD mathematician (Step 2). I check derivations, notation, units and hidden assumptions, and I weigh the Investigator's leads. Written for the reader profile (undergraduate calculus, ODEs, intro E&M). Every symbol is defined where it first appears.

Tags: [read] from the document or a saved file, with page; [read: page image p.N] from a rendered page; [web] with URL; [memory] with confidence. My own calculations are tagged [calc]; the numbers are reproducible with the short scripts described at the end.

Source note: there is one primary document, and it carries all the claims. The Step 1 deep read covers only pp.14-19. I read the full text layer (pp.1-49) myself and rendered pp.1, 8, 9 and 46 to check the equations.

---

0. Summary of verdicts

Overall: the experiment the document actually reports is a clean textbook L/R demonstration. I found no derivation anywhere in the document that supports a power gain. What the author calls cancelling the "magnetic component" is ordinary coupled-inductor behaviour, and ordinary coupled-inductor networks cannot deliver more average power than they absorb.

---

1. Background the reader needs (short)

Inductor and mutual inductance. A coil with current i stores magnetic energy W = ½ L i² (L is self-inductance, in henries). Its terminal voltage is v = L di/dt. Two coils share flux through mutual inductance M, with v₂ = M di₁/dt. The coupling coefficient is k = M/√(L₁L₂), and 0 ≤ |k| ≤ 1. The bound is not a choice: the stored energy must stay non-negative for every pair of currents. [memory, high]

Reciprocity. The M that couples coil 1 to coil 2 is the same M that couples coil 2 back to coil 1. This is why a load on a secondary always reflects into the primary. [memory, high]

Passivity and energy balance. For any network built only of resistors and coupled inductors, d/dt(stored energy) = (power in) − (power dissipated). In steady state, average input power = total power dissipated in all resistors, load included. Hence P_load ≤ P_in. No wiring pattern, winding direction or coupling arrangement can change this, because it follows from L's being positive-definite and R being positive. [memory, high]

Potentials. With B the magnetic field, E the electric field, A the vector potential and φ the scalar potential: B = ∇×A and E = −∇φ − ∂A/∂t. A is a potential from which B is derived; it is not a "component" of B. φ (lowercase) is the scalar electric potential. [memory, high] Feynman Lectures Vol. II ch. 15 is the chapter Sykes cites for this [read: p.21].

---

2. Claim-by-claim review

Claim 1. Partnered (bucking) output coils cancel the magnetic component, A adds, Lenz drag removed

a. Is the mathematics internally consistent?

The document gives no derivation for this claim. It gives words, pictures and equations for other things (emf = −dΦ/dt, emf = Bvl). I checked the pieces it does state.

1. Notation clash. On pp.4 and 6 "A, and Phi" are the vector potential and "its partner Phi", with "A, the Electric Component of the Magnetic Field and, Phi, the Magnetic Component" [read: p.4, 6]. On p.9 "dPhi" is "a quantity of the Electromagnetic Field" in the induction law emf = dΦ/dt [read: p.8-9, page image p.9]. On p.46 "Phi, the Magnetic Flux" [read: page image p.46]. So one symbol means three things: scalar potential (Feynman's φ), an undefined "magnetic component", and magnetic flux. Flux is the integral of B over a surface (units weber, Wb), and the scalar potential is in volts. They cannot both be the "magnetic component". The claim "A is the electric component of the magnetic field" is not a statement the mathematics can express: E contains −∂A/∂t, so A is a source for part of E, but A is not "of" B.
2. Units error in the flux-cutting law. The document defines "l = typically the Area of the Element" in emf = Bvl [read: p.8-9, page image p.9]. With B in tesla (T), v in m/s and l an area in m², Bvl has units T·m³/s, which is not volts. l must be a length. Corrected: emf = B v l, with l the conductor length perpendicular to v and B. Also the variable "B = Electromagnetic quantity" is undefined.
3. Hidden assumption about "net zero field in a loaded transformer". The document says that when the transformer is loaded "there is a NET ZERO Magnetic Field. This means the Ampere Turns on the primary are equal but opposite to the Ampere Turns on the secondary" [read: p.21]. This is correct textbook transformer behaviour: N₁I₁ ≈ N₂I₂ because the core flux is fixed by the primary voltage (V₁ ≈ N₁ dΦ/dt). Ampere-turns means number of turns times current, which is the magnetomotive force (mmf). So the "zero-sum" the author wants to exploit is already what a normal loaded transformer does, and the primary current rises with load because of that balance, not in spite of it. Sykes then labels this "Tom Bearden's Whittaker Wave/Anti Wave" [read: p.21], which does not change the algebra.
4. The central step is missing. The step from "the two output coils' fields cancel" to "the primary no longer has to supply extra current" is stated, not derived. See below for the derivation the claim would need.

Test the claim with a model. Take a primary coil 1 (self-inductance L₁, resistance R₁) driven by a sinusoidal voltage source at angular frequency ω (rad/s). The two output coils 2 and 3 are connected in series with a load R_L. Define M_x as the net mutual inductance between the primary and the series loop (the signed sum of the two primary-to-output mutuals), and M₂₃ as the mutual between the two output coils.

• Case A. Both output coils see the same flux and are wound oppositely. The two induced emfs cancel: M_x = M₁₂ + M₁₃ = 0. No current is driven. This is a humbucking pickup. [calc: 0 W out, and the primary current is the same as with no outputs]
• Case B. Output coils sit on halves of the core where the flux changes in opposite directions (the geometry Sykes describes on pp.8, 23, 36) and are wound oppositely. The emfs add: M_x = 2M. Now Kirchhoff's voltage law for the two meshes (1 = primary, 2 = output loop) reads
• (R₁ + jωL₁) I₁ + jωM_x I₂ = V
• jωM_x I₁ + (R_loop + R_L + jωL_loop) I₂ = 0 where L_loop = L₂ + L₃ + 2M₂₃ and R_loop is the coils' own resistance. Eliminating I₂ gives the input impedance Z_in = R₁ + jωL₁ + (ωM_x)² / (R_loop + R_L + jωL_loop). The second term is the reflected impedance. Its real part is (ωM_x)²(R_loop+R_L)/|Z_loop|², which is positive whenever load current flows. That positive reflected resistance is Lenz drag in circuit form. For it to vanish, either M_x = 0 (no output, Case A) or the loop current must be zero.

Numbers [calc, linear model, 400 Hz, 1 V drive]. L₁ = 1 mH, L₂ = L₃ = 10 mH, M₂₃ = −0.5 mH, M_x = 4 mH (k_eff = 0.92), R₁ = 0.5 Ω, each output coil 0.2 Ω.

The primary current rises from 0.39 A (no load) to 1.37 A (R_L = 2 Ω). The opposite-wound pair removes nothing. P_out/P_in stays below 1 in every case I swept (best 0.67).

b. Consistent with established experiments? The weak form is. Humbucking pickups, buck transformers, bifilar non-inductive resistors and CRT degaussing all rely on this arithmetic and work as the algebra predicts [memory, high]. The document's own p.45 concedes "total cancellation is also impossible" and that "there is still a time constant in the charging of these Partnered Output Coils" [read: p.45], which is an acknowledgement that the field is not removed.

c. Tested? By Sykes, in videos and on his forums (2010-11 first device, then 2015-2026). The Investigator found: the 2014 hyiq.org statement "Input is 1 Watt (5V 200ma)... rated at 12V 300ma (3.6 Watt). I am getting slightly more Output than this" [web: https://www.hyiq.org/ via research/2026-10-04-hyiq.org.md]; the 2019 aboveunity thread, where a replicator (dewetw, 30 Oct 2019) built the three-coil circuit on an N87 ferrite core, reproduced the sawtooth waveform and reported "never saw any result where I am getting more than about 0.82 COP" (one documented forum post with photos; Sykes's reply addressed the waveform and did not dispute the figure); patrick1 (23 Oct 2019): "successful of course, but thats because it isnt OU"; the Itsu replication (overunityresearch.com topic 3951, 2020), described by Sykes as "failed" (Itsu's own data are behind a login, so only Sykes's account of it was read). Measurement-basis concerns from Partzman (diodes reversed against Sykes's own scope pictures; RMS vs mean) and from MileHigh in 2015 ("alleged COP 1.7"; "Where is your own data?"). The 2014 "slightly more output" figure is one self-reported statement with a stated input of 1 W and no stated method; "COP 1.7" is attributed to Sykes by MileHigh and Sykes's own statement of it was not found. No neutral calorimetric or integrated-power test is on record. The "Measurements" thread shows RMS-based power of 59.5 W against a scope mean of 17.1 W for the same 0.25 Ω shunt measurement [web: research/files/aboveunity-measurements.txt], which shows how fragile the power readings are.

d. If true, what would have to change? The positive-definiteness of the mutual-inductance matrix (energy is a quadratic form), or the sign of the dissipation in the loop resistances, or conservation of energy. All three are used in the design of every transformer.

Verdict. Split. Weak form (bucking pair cancels the external field / mutual coupling): Established. Strong form (the bucking pair lets load power flow without reaction at the input, so P_out > P_in): Refuted by calculation. The A/Phi vocabulary: Internally inconsistent (one symbol, three meanings; a units error in emf = Bvl).

---

Claim 2. Lenz's law is the result of drawing current from already-separated charge, not its cause

a. Consistent? In the document's own wording, "Lenz's law is the Result of drawing Current from the already Separated Charge, not the cause!" [read: p.10]. Taken literally this is half right and half a restatement. Lenz's law (the induced current opposes the change that produced it) is a statement about the load current and its field. With an open secondary, no current flows and there is no reaction at the primary; this is textbook [memory, high]. The document's p.21 diagram says the same ("If Loop B... is not connected to our Load, then Lenz's Law is not in effect"). Where it goes wrong:

• "Induction is the Separation of Charge Carriers" [read: p.9], and "Mechanical Energy is NOT the Source of the Energy Generation!" [read: p.8]. The mathematics that governs the energy is this: the force on a charge q moving with velocity v in a field B is F = q v × B. In a generator, the conductor moves, so the electrons are carried along; the magnetic force on the drift current (the part of the electron motion along the wire) is a force on the conductor, opposing the motion. The work the mechanical source does against that force equals the electrical power delivered: P = e·i = τ_mech·ω, with e the emf, i the current, τ_mech the shaft torque and ω the angular speed. If the "drag" is removed (τ_mech → 0) then e·i → 0 as well. Charge separation without drag is charge separation without delivered power.
• The statement "Energy in the form of Mechanical Rotation simply is not the actual causality of the Energy on the Output Terminals" [read: p.8] is not supported by any equation. The author's thought experiment (rotate a shaft with no coils and get zero output) is correct but is not an argument: it shows the conductors and field are needed, not that the shaft work is unnecessary.

Worked number [calc]. A 10 kW generator at 1000 rpm: ω = 104.7 rad/s, so the shaft torque needed at 100% efficiency is 10 000 / 104.7 = 95.5 N·m. If this drag were absent and the output stayed 10 kW, the extra 10 kW would have come from nowhere.

b. Consistent with the standard generator and transformer experiments (every loaded generator needs more shaft torque). The p.45 rotor experiment, "No or next to no Lenz's Law and at the same time, Electricity is Generated" [read: p.45], is reported without numbers (no torque, speed, output or load stated), so it cannot be compared with an energy budget.

c. Tested? Only in the author's own informal rotor test (p.45) and the alternator test (p.48), neither with measured numbers. The Investigator's finding that Sykes (overunity.com reply #4, 16 Jan 2015) wrote "Lenz's Law is ONLY apparent when Current is drawn from the resulting Charge Separation... So Energy is not Generated at all! It was already there!" and, in reply #7, "Energy can not be created or destroyed. You and I agree with Science on this one!" shows the author himself holds the energy-conservation view. [web: Wayback copy of overunity.com/15395, research/2026-10-04-overunity.com-wayback.md]

d. The relation e·i = τ ω (a corollary of Faraday's law and energy conservation).

Verdict. Split. "No load current, no Lenz reaction": Established. "Charge separation without a reflected field, usable for output power": Refuted by calculation (energy balance).

---

Claim 3. The magnetic field slows charge flow (filar 75 ms vs bifilar 650 µs)

I verified the Step 1 deep read against the page text and re-derived its numbers.

a. Mathematics. For a coil of inductance L and total series resistance R switched onto a constant voltage V, the current is i(t) = (V/R)(1 − e^{−t/τ}), with time constant τ = L/R (time to reach 63% of final value; about 3τ for 95%). [memory, high]

[calc; inputs from read: pp.14-16 and page images pp.14-15 via the deep read]

Points of agreement with the deep read, and where I go further:

• Experiment 1 matches τ = L/R: the author's "75 ms" is about 3τ. Experiment 2's L/R prediction (27 µs) is about 24 times shorter than the quoted 650 µs; the deep read reads the trace rise as about 0.4 ms, so the "non-inductive" coil still shows an effective inductance of a few mH (V/(di/dt) ≈ 4 mH) or a driver/drive limit. The deep read's conclusion that 650 µs is a pulse width set by the generator, not a measured intrinsic time, is supported by the arithmetic: 1/650.4 Hz = 1.54 ms period, and a 50% duty pulse is 0.77 ms, comparable to 650 µs. [calc]
• Ratio test. Inductance ratio 231 mH / 0.24 mH ≈ 960; ratio of the claimed times 75 ms / 0.65 ms ≈ 115. If the "time" were L/R the ratios would be equal. They are not, so at least one of the numbers is not an L/R time.
• Extra check I add. With two identical windings of 55.6 mH each, series aiding is L_s = 2L(1 + k), and k cannot exceed 1, so L_s ≤ 4 × 55.6 = 222 mH. The quoted 0.231 H is about 4% above that ceiling, which gives the formal "k = 1.08" [calc]. This is a small inconsistency in the author's inductance measurements (instrument, test frequency and drive level not stated: [read: p.15]), or the two windings differ slightly, or the meter reads a different permeability from the coil at 1.6 A. It does not change the conclusion, but the deep read's remark that 231 mH is "close to" 222 mH should be read as "slightly above the physical maximum". For Experiment 2, L = 2L(1 − k) with 0.24 mH gives k = 0.998 [calc].
• Uncontrolled differences. The two runs use different supply voltage (15.4 V and 7.0 V), generator drive (9 V and 5 V) and frequency (≈ 6 Hz and ≈ 650 Hz). The same coil was reconnected, so resistance (7.8 Ω) is common. Only the winding connection is the intended variable; three other things moved at the same time.

The interpretive step. The author writes: "This is an expected behaviour for the Current in the Coil when the Inductance Changes... This proves that the Magnetic Field Slows the Charge Flow Rates dramatically" [read: p.16]. The first sentence is the standard account. The second is a re-description of the same fact: inductance is the coil's magnetic-field opposition to a change of current. An experiment can prove something only if a rival theory predicts a different outcome; here both predict identical curves. Also, the quantity measured is the rate of rise of the current, di/dt, not "charge speed". Drift speed of conduction electrons depends on current density; current in a series loop is the same everywhere at any instant.

Sweet's "duty factor" quotation [read: p.17] ("a conductor that formerly had a temperature rise... would now operate at a temperature of 1.0. Thus the same gauge wire would carry 10 times more current at the same temperature"). The plateau currents in both runs equal V/R with the same 7.8 Ω. So the data show no change in steady resistance or heating. The claim is linked to the experiment by assertion only. The Investigator located the source: Sweet's "Letter to Mark from Sparky" (one page, PDF author "Patrik", 2011) on hyiq.org, with the wording "at the same time opening the gates wider" where the Guidelines version has "and at the same time" [web: https://www.hyiq.org/Downloads/Letter%20to%20Mark%20from%20Sparky.pdf]. It describes a 6-pole machine of 7.5-10 kW with a feedback winding "spirally traversing the total length of the coil conductor". The deep read had found the p.17 quote uncited; the letter supplies the source. Sweet's text is a claim about steady-state heating (I²R); the experiments measure only the transient.

b. The result matches every textbook RL circuit [memory, high]. Sykes's own site agrees: "A Vector Potential" on hyiq.org states "time constant t = L/R... 5 time constants to charge an inductor" [web: research/files/hyiq-Research-A-Vector-Potential.txt].

c. Tested? The reported experiments are the author's. No replication. The L/R law itself is replicated in every laboratory course.

d. Nothing; unless the claim is that DC resistance or I²R heat drops, which would require a new measurement.

Verdict. Split. Measured current curves and τ = L/R: Established. "Magnetic field slows charge flow" as physics beyond inductance: Not testable as stated (no distinguishing prediction). The 650 µs figure as an intrinsic charge-flow time: not supported by the trace (Tested, not supported, by the author's own data and the L/R calculation). Sweet's duty-factor and 10x current-at-same-temperature statement: Never properly tested (no heating or resistance measurement in the document).

---

Claim 4. Loose coupling, Bloch wall between coils, diode and square-wave drive at the BH-curve knee, enough load; input power falls as output rises; alternator test

a. Mathematics. No derivation is given for any of these requirements. They are experiential rules ("I have never seen an experiment that has been done several times show incorrect information" [read: p.34]; "For every one device that works, expect 10 devices that don't" [read: p.34, 45]).

1. Input falls as output rises (p.46). The text says the output coils' current "feeds back via confined Magnetic Fields... the Input Coils Impedance is increased dramatically. This brings back the input current and as a result the input power is dramatically lowered." It then explains by analogy: "Our Load just so happens to be another Power Source... the input power is reduced simply because we have a source of power returning back to our Transformer" [read: p.46, page image p.46]. The closing "simple case of applying Ohms Law, V²/R" is a fragment: the equation is not applied to anything.

My check [calc]: reuse the model of Claim 1 and tune the output loop to series resonance (add a capacitor C = 1/(ω²L_loop), loop reactance zero). Then Z_in = R₁ + jωL₁ + (ωM_x)²/(R_loop + R_L). Results at 400 Hz:

So in a purely linear, passive, textbook circuit, input current falls to a small fraction of its no-load value when a resonant output loop is loaded (the loop's reflected resistance, (ωM_x)²/R_loop, is large when R_loop is small). This reproduces the qualitative effect Sykes describes, with efficiency below 1 (best 0.91 across a sweep of R_L). I did not find a case where P_out rose while P_in fell: they move together. What the model shows is that falling input current is not evidence of gain, because current is not power and P_in is bounded below by P_out. 2. Mean versus RMS. Sykes's rule "Any COP > 1 claims must be accompanied by Measurements" and his statement that mean (not RMS) must be used [web: aboveunity Measurements thread, research/files/aboveunity-measurements.txt] are right in spirit: average power is the time average of the product v(t)·i(t), not the product of RMS values (RMS = root mean square, the DC-equivalent heating value of a waveform). The two are equal only for resistive loads with in-phase sinusoids; for a pulsed or reactive load they can differ by large factors, in either direction. The posted 59.5 W vs 17.1 W figure for the same shunt is exactly this effect. 3. Bloch wall. In ferromagnets a Bloch wall is the thin transition layer between two magnetic domains in which the magnetization rotates; it is typically tens to a few hundred nanometres wide [web: https://www.sciencedirect.com/topics/materials-science/domain-wall; summary of searches found "5-30 nm" generally and about 140-200 nm measured in iron]. Sykes uses "Bloch Wall" for a macroscopic region "between the bucking coils" that "grows" and "moves" with the field, with arrows "up and down" in drawings [read: p.4-5, 23, 36]. These two usages are not the same object. The macroscopic region he describes is not defined by any equation, so statements about it cannot be checked. His lamination "jumps over the Bloch wall with extreme reluctance" [read: p.7] is an informal observation of the nonuniform field of a magnetic core (a magnet-attracted lamination experiences the field gradient), not a domain-wall measurement. 4. Diode / knee. "Without the diode the input coil is 'choked off'" [read: p.41]. This is a circuit statement: with a series diode, current can flow only in one direction. Whether it is correct depends on the circuit, which is drawn but not analysed. Partzman's critique (Investigator) that the posted schematic's diodes point opposite to what the scope traces show would, if right, mean the drawn circuit is not the one measured [web: Wayback of overunity.com/18464, research/2026-10-04-overunity.com-wayback.md]; Sykes's reply was "My Diagram is correct, change it, it wont work! Period!". I could not check this without the original schematic images and the scope data. 5. Alternator test [read: p.48]. "By connecting the Field Coil on the Alternator... in series with the output Coils, enough Feed Back is produced to get an Electrical Output... Not Free Energy". This is self-excitation of a generator through residual magnetism, standard engineering [memory, high]. The mechanical input remains, and the author says "ONLY because of Lenz Law, Resistive Mechanical Force is applied on the Shaft". So the statement actually uses Lenz drag as the mechanism. 6. Loose coupling: from the model, k_eff < 1 only changes how much of the primary's flux reaches the loop. It reduces reflected resistance (ωM_x)²/R_loop in proportion to k²; this is a trade between transferred power and reflection, not a way around either.

b. Consistent with experiment? The effects that can be modelled (current falling in a resonant loaded transformer, gapped cores storing energy in the gap, diode-switched collapse of a field as in Bedini-type circuits) are consistent with ordinary circuit behaviour. Sykes's strongest numerical statements are COP-type (1.7 alleged, "slightly more output" 2014, 2019 three-coil circuit with 12 V 400 mA input and "some data off screen" output). The independent replicator on his own forum reported COP ≤ 0.82.

c. Tested? By the author (not neutral); by dewetw (COP ≤ 0.82; one forum post with photos; not disputed in Sykes's reply); by patrick1 ("successful... isn't OU"); by Itsu (called failed by Sykes, data not read); the Investigator also recorded that the forum thread of 514+ pages (overunity.com, topic 15395, 1.14 million reads by Jan 2017) is now a domain-for-sale page and the mirror archive is paywalled, so the record of the 2015-2021 replication debate is going dark; competing explanations are commercial paywalling, a lapsed domain or deliberate removal, and the evidence so far points only to the commercial one (the archive says "Now Open for Paid Subscribed Members Only"). Wayback copies of the first page of the thread and of the 2021 builders-group thread survive; a full CDX listing would show how much. I note this as an absence-of-evidence item to be followed up, not as a conclusion. Sykes's own account on aboveunity.com that "We have been DEPLATFORMED by everyone" and "PayPal De-Platformed me!" is an interested party's account; no second source found. Sykes's stated benchmark method (2020): "making sure you have around 80% efficiency at least. Then anything above the 80% mark you can verify, has entered the System" and "Input can become Negative!" is a testable protocol but depends on how power is measured; its logic is only sound if the 80% benchmark and the test use identical metering.

d. Allowing P_out > P_in would require a negative resistance somewhere in the loop (an active element supplying power), or a non-conservative energy term; neither occurs in a core-and-diode circuit unless the diode circuit's stored energy is mis-accounted.

Verdict. Split. (i) Input current falls under resonant loading: Plausible, untested as stated (my model reproduces the effect with P_out < P_in). (ii) P_out > P_in: Refuted by calculation (passivity; no negative resistance). (iii) Moving macroscopic "Bloch wall" and the "knee" rule: Not testable as stated. (iv) Gap, diode, square wave, "needs minimum load": Never properly tested (no controlled comparison in the document; one independent replication at COP ≤ 0.82 on a forum). (v) Alternator self-excitation: Established, and with mechanical input, as the author says.

---

Claim 5. Sweet's VTA, Moray, Meyer, Jensen, Koenig, Frolov, Melnichenko and others work on this one principle

a. Mathematics. Three items I could check:

1. Sweet's quoted algebra. The Investigator located "Nothing is Something" (Sweet, 24 Jun 1988) with E = (B×V) + (−B×−V) = 2(B×V) [web: https://www.hyiq.org/Downloads/Nothing%20is%20Something.pdf; also in aboveunity thread quoting it]. Here B is the magnetic field, V a velocity, × the vector cross product. Since (−B)×(−V) = B×V (two sign flips in a cross product cancel), the identity is trivially true: two systems with both B and V reversed give the same motional E, and their sum is twice one. This is ordinary vector algebra and is also what happens when two generators in series add their emfs. It says nothing about vacuum energy or about cancelling a "magnetic component".
2. The Sweet passage Sykes quotes on p.36: "If the directions of the two signals are such that opposite H-fields cancel and E-fields add, an apparently steady E-Field will be created. The energy density of the fields remain as calculated above, but the value of the E-field will double from E/2 to E" [read: p.36; page images not checked]. For two equal waves moving in opposite directions, the sum is a standing wave: E₁ + E₂ = 2E₀ cos(kx) cos(ωt) for the case where H cancels at that plane and instant. The field is not steady in time (it oscillates as cos ωt), and the energy density is not reduced; energy sloshes between electric and magnetic form at quarter-wave spacing. The "apparently steady" phrase holds only for special pulse timing. This is standard wave physics [memory, high]. Bearden's own quote reproduced on p.18, "would certainly cancel (spatially) their amplitudes... but would not cancel their energies" [read: p.18, via deep read], says the same: cancelling amplitudes does not cancel energy.
3. Sweet's 1986 lab notes (hyiq.org transcription of scans from an "anonymous source"; interested-party transcription; the scans themselves not found). Two lines are checkable [web: https://www.hyiq.org/Downloads/Lab%20Notes%20Transcribed%20-%20Floyd%20Sweet.pdf, via research/2026-10-04-hyiq.org.md]: - "4.6A × 24.2V = 111.32 Watts": 4.6 × 24.2 = 111.32. Correct. - "19.1544w / 0.023mw = 83,260,000% Gains": 19.1544 W / 0.023 mW = 832 800, i.e. 83 280 000%. The transcription shows 83,260,000%. A small arithmetic discrepancy of 0.02%, consistent with transcription or rounding of the quotient. [calc] A third item: the aboveunity forum analysis by "Anor" (Sep 2026, a third-party forum post) lists the input as 7 V × 143 µA, which is about 1 mW, not 0.023 mW. The two input numbers differ by a factor of about 43. [calc: 7 × 143e-6 = 1.0 mW]. Both are far below the output, but the discrepancy shows that the "gain" is sensitive to which figure the input is. A more important point: for a coil drive of "hundreds of thousands of ohms" impedance at several hundred hertz, V × I read with a digital multimeter is apparent power (volt-amperes, VA), not real power; real power is V·I·cos φ, where φ is the phase angle between V and I. For a mostly reactive drive coil real power can be far smaller than VA, which would make the stated gain larger, not smaller; but then output real power of 111 W with an input of 0.02-1 mW means a power gain of 10⁵ or more with no calorimetry. The notes name a Beckman multimeter "purchased 1hr prior" and a Keithley 160, neither of which measures real power of a 400 Hz non-sinusoidal waveform.
4. Source-energy check [calc; memory medium → web]. Ceramic (ferrite) grade 8 magnets have a remanence of about 0.38 T [web: https://www.magnet4sale.com/c8-block-magnet-3x2x1-4-ceramic-magnets-hard-ferrite-magnets/, and the grade tables returned by the search]. The lab notes use two 4" × 6" × 1" ceramic magnets. The magnetic field energy density is u = B²/(2μ₀), with μ₀ = 4π × 10⁻⁷ H/m. Using B = 0.05 T (an upper estimate of the field outside the magnet) over the magnet volume of 3.9 × 10⁻⁴ m³ gives about 0.4 J per magnet; even using the internal field B ≈ 0.38 T gives about 22 J per magnet. A 111 W output sustained for one second needs 111 J, and 'minutes of running' needs thousands of joules. So the magnets are not a consumable energy store at this scale; Sweet's claim is that the vacuum is the source, which no calculation from the circuit can confirm or exclude; a calorimetric or total-energy test would decide.
5. Flux-swing check [calc]. For a sinusoidal emf with RMS value E_rms, N turns and frequency f, E_rms = 4.44 f N Φ_pk (Φ_pk is peak flux in webers). With E = 24.2 V, f ≈ 400 Hz and N = 12 (the notes list a 12-turn output coil of #12 wire, 1.5" diameter): Φ_pk = 24.2 / (4.44 × 400 × 12) = 1.1 × 10⁻³ Wb. The area of a 1.5" circle is 1.14 × 10⁻³ m², so a flux density of about 1 T would have to swing through the coil at 400 Hz, if the flux is confined to the coil's cross-section. That exceeds the remanence of the ferrite magnets (0.38 T) and is close to iron saturation. This independently supports a point made by "Anor" in the Investigator's find ("12-turn flux-swing problem unresolved"). It does not prove anything, because the lab notes give no core geometry, and the flux may be concentrated in a smaller iron section; but it makes the figures hard to reconcile with what is written.
6. Other devices. Earl Koenig's US4806834 (filed 1987-04-16, granted 1989-02-21; title "Electrical circuit for inductance conductors, transformers and motors") does describe "two coils... coiled about a bar of magnetizable material such that they are disposed in mirror image symmetrical relation" [web: https://patents.google.com/patent/US4806834A/en via research/2026-10-04-google-patents.md]. That supports the geometric statement that mirror-image coils exist in others' designs; the patent's own performance claims were not read, so "same principle" in the sense of delivering gain remains untested. The document has no circuit analysis for Moray, Meyer, Jensen, Frolov, Melnichenko or others; the author's own p.29 remark that in Melnichenko's device "the direction of turns is not reversed but results are still possible" weakens any single "winding direction" rule.

b. Consistent with experiment? Sweet's VTA performance is reported by Sweet himself, Bearden (who co-authored the 1991 IECEC paper "Utilizing Scalar Electromagnetics to Tap Vacuum Energy", Boston, 4-9 Aug 1991, and who promoted the device) and Rosenthal. Bearden's "6 watts of electrical power from the vacuum" [read: p.19-20] differs by a factor of about 800 from the "5000 W" shoe-box statement [read: p.13]; Sykes attributes the latter to the "Generation Six" VTA without a source. The rexresearch page records a dispute about whether Bearden's accounts are "hearsay... via telephone conversations with Sweet" and Bearden's reply that he "did lots of experiments with Sweet, in Sweet's lab, on personal visits to California" [web: https://www.rexresearch.com/sweet/1nothing.htm]; this is one dispute on one page and I treat it as evidence of what was said, not as a ruling.

c. Tested? For the VTA: a weight-reduction test with a load of 100 W lamps (Bearden's account), the IECEC 1991 paper (co-author an interested party), and Sweet's own lab notes (transcribed by an interested party). No independent laboratory test with calorimetry or instrument calibration is on record in the material I read. For Sykes's own devices: see Claim 4. Sykes himself now rejects Bearden's magnet "conditioning" as "nonsense" (aboveunity, Apr 2025), and in the same thread says "we have not succeeded, in general" while also writing that "Marcel [FringeIdeas] has succeeded", a self-reported contradiction left as is. The "Tinman" quote (2015, quoted by Sykes in 2020: "a bucking coil arrangement dose indeed add energy to a system... all my devices... are mechanical") is second-hand; I did not find the original post.

d. A real source of energy outside the circuit. Mainstream physics has the vacuum's zero-point energy but no known way to extract net power from it with magnets and coils; a demonstration would require a closed-energy-budget measurement.

Verdict. Never properly tested. There is no neutral, controlled, instrumented test of any of the cited devices on record in the material read, and it was not dismissed without any test: Sweet's own records and Bearden's weight test exist but are from interested parties. The algebra in the quoted Sweet passages does not support the gain claim (trivial sign algebra; standing-wave behaviour).

---

Claim 6. Generators lose 98-99% of their loss budget to Lenz drag; removing drag is the route to efficiency; "not a new energy form" but also "open system connected to the Aether" and "Free Energy Future is here NOW"

a. Mathematics. The text is: "the World's most efficient Generators are 98-99% efficient; this means that 100% of Rotational Shaft Energy, with a 1-2% loss in friction/heating and other losses, and the other 98-99% are the effects of Lenz Law creating Drag on the moving Rotor to generate Electrical Energy" [read: p.23]. Efficiency is η = P_out / P_in. At η = 0.98, 98% of the shaft power appears as electrical output and 2% is lost. The author is correct that the 98% corresponds to the Lenz drag torque, because drag torque × speed is the electrical power produced (Claim 2). He is wrong to treat it as a loss: it is the useful conversion. Step 1's claims.md paraphrase ("lose 98-99% of their loss budget to Lenz drag") misstates the source; the document does not say 98-99% of the loss budget, it says 98-99% of the shaft energy. For a 10 kW output at 98% efficiency the losses are 10 000/0.98 − 10 000 = 204 W [calc]. Removing the "drag" would remove the 10 kW, not the 204 W.

Internal inconsistency: p.47 says "It took me a long time to see that there was no new form of Energy. That it is an adaptation of the already known energy", p.48 says "Not Free Energy" about the alternator, p.22 says "if Lenz's Law could be by passed, then every Generator on the planet could be a perpetual motion machine of the first kind", and p.47 says that any system with fields is an "Open System... directly connected to the Aether", with p.49 closing "The Free Energy Future is here NOW!" [read: p.22, 47-49]. A system either has an energy source outside the circuit or it does not; the document asserts both.

b. Generator efficiency data: large generators are 98-99% efficient [memory, high], consistent with the first half of the author's sentence.

c. Tested? No test of "removing drag gives more efficiency" with a measured shaft torque and speed is reported. The p.45 rotor experiment gives no numbers.

d. If drag could be removed with output retained, electromagnetic power would exceed mechanical power, so conservation of energy (or the Lorentz force law) would need an exception.

Verdict. Split. The arithmetic as written (treating useful output as loss): Internally inconsistent. Removing drag to raise efficiency: Refuted by calculation. "Open system connected to the Aether" and "Free Energy Future": Not testable as stated (no measurable difference stated), and the author's own "not a new form of energy" contradicts it.

---

3. Investigator's findings weighed (kept in the record)

All of these are reported with their sources as leads. None is dismissed.

---

4. Corrections to Step 1

1. claims.md states "Read via digest only; no deep read was on file." A deep read now exists for pp.14-19 (deep/Guidelines-to-Bucking-Coils-5ecaec.md). facts.json how_read should say "digest + deep read pp.14-19 + page images". I read the full text layer and checked pp.1, 8, 9, 46 as page images.
2. claims.md claim 6 ("lose 98-99% of their loss budget to Lenz drag and only 1-2% to friction") misstates p.23. The document says 98-99% is the efficiency, i.e. the share of shaft energy converted to electricity, and calls that conversion "Lenz drag". It is not a loss budget (see Claim 6). Suggested wording: "Generators are 98-99% efficient; the author labels the 98-99% converted energy as Lenz drag and argues that removing the drag is the route to higher efficiency."
3. facts.json "98-99% ... [read: p.23-24]": the text is on p.23 (p.24 continues with other material).
4. Deep read, "231 mH close to 222 mH". The quoted 0.231 H is slightly above the physical maximum 4 × 55.6 = 222 mH for two identical windings (k ≤ 1), a 4% excess (k = 1.08). The conclusion "tightly coupled windings" stands; the exceedance points to measurement or winding-difference error, with the instrument and test conditions unstated.
5. facts.json p.46 summary ("input power falls as the output coils work correctly [read: p.46-48]") omits that the author's mechanism there is that the load "just so happens to be another Power Source" returning power. The claim as the author argues it is about a power source in the output side; this is what the energy check turns on.
6. facts.json key claim "Floyd Sweet's VTA and the devices of... operate on this same bucking-coil principle" is the author's classification, not a result; Koenig's patent shows a mirror-image two-coil geometry, but the performance claims were not read.
7. Step 1 relevance says the document gives "no tabulated power measurements for the author's own devices beyond the coil time-constant comparison". I confirm: the document contains no power-in/power-out figures for the author's devices. (The figures in the Investigator's leads are from later web posts.)
8. mainstream_conflict says A and φ are not "electric and magnetic components of B"; I agree, and add that the document uses "Phi" in three senses (scalar potential, "magnetic component", flux) and has a units error in emf = Bvl ("l = the Area").

No errors found in the deep read's reading of the scope traces and circuit; its L/R numbers reproduce exactly (26 ms, 27 µs, 79 ms; I get 26.2 ms and 27.3 µs at R = 8.8 Ω).

---

5. Web checks and honesty notes

• Checked [web]: ferrite grade 8 remanence about 0.38 T (https://www.magnet4sale.com/c8-block-magnet-3x2x1-4-ceramic-magnets-hard-ferrite-magnets/ and the other grade listings in the search); Bloch-wall widths tens to a few hundred nm (https://www.sciencedirect.com/topics/materials-science/domain-wall and arXiv listings from the same search). One search served both; I did not open the pages individually, so these are search-result summaries.
• Textbook facts used from memory at high confidence: τ = L/R, coupled-inductor energy positivity and |k| ≤ 1, reciprocity, transformer mmf balance, E = −∇φ − ∂A/∂t, standing-wave behaviour, e·i = τ·ω. My verdicts depend on these only.
• Not checked: any claim of replication outside the Investigator's notes; the contents of video demonstrations; Itsu's data; the Sweet lab-note scans; Koenig's patent claims; Jensen, Melnichenko, Frolov documents. I state no replication or failure without the sources given above.
• Source gap: Sykes's own device schematics and scope data for the three-coil circuit were not read (the 2019 thread says "some data off screen"). The conclusions about Claim 4 therefore rest on an energy-balance argument and on the model, not on his data.
• The reader can reproduce my numbers: the RL table is arithmetic (τ = L/R); the coupled-loop table uses the two mesh equations in Claim 1 with the stated parameter values; torque is 10 000 W / (1000 × 2π/60 rad/s); the flux swing uses E_rms = 4.44 f N Φ_pk.